�����    �   huggingface�{"info": {"features": {"id": {"dtype": "string", "_type": "Value"}, "text": {"dtype": "string", "_type": "Value"}, "dataset_id": {"dtype": "string", "_type": "Value"}}}}p4���� dataset_id��������text���� id����  (��   �� ���H&H&��5u����������>��� #)/5;AGMSY_djpv|����������������������� #(.39>CIOU[`ekqw}���������������������� $*06<BHNTZ`flrx~���������������������� %+17;AGMSY_ekpsy���������������������%+06<AGMSY_ekqw}����������������������$*05:@FLRX^djpv{���������������������� "(.4:?EKPUZ`flrx~���������������������� !'-38>DJPV\agmsy~���������������������� !'-27=CIOU[afkqw}����������������������     & , 2 8 > D I O U [ a g m s y  � � � � � � � � � � � � � � � � � � � � �     # ) / 4 : ? E K Q U [ a g m s y  � � � � � � � � � � � � � � � � � � � � � �     & , 1 7 = C H N T Z ` f l r x ~ � � � � � � � � � � � � � � � � � � � � � �      % + 1 7 = C I O U [ a f l r x } � � � � � � � � � � � � � � � � � � � � � �     # ) / 5 : ? E J P V \ a g l r x } � � � � � � � � � � � � � � � � � � � � � �  "(.4:?EKQW]ciou{���������������������� #).4:@EKQW]ciou{���������������������� "(.4:@DIOU[agmsx~���������������������� !'-28=CIOU[aflrx~���������������������� "'-39?EKQW]chntz����������������������� #).39?DJPV[_ejpv{����������������������$*/5:@FLRW]bhntz���������������������� "(.39>DJPV[aflrx}���������������������� #).37=BHNTY_ekpv{�����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�P�C��e�y���lw&D(&p'|)PŽ3�"�������+���$�(���p!�"v#(�@����-�������lN C�_$g�{֠#���7s�b��u�%�о������Qu�����B�I�O�QsS�U0Y3d�,�n�7��p8C�N�P8RTCU�UZW[�����%�/�� ���o(;�"�wRgaxg��������Z�� ��z�M����=�a�[]po�\���f�}�������}s~ű�Y�K�����$���� �s"}T��8�Q�����-X/�0�ZԷ��g��� � � � i" �, A. �E"iH"�V"��"\�"��"d�" �"�":�"��"�"��"Y�"��"��"|�"�#�#�g#�n#�|#(�#F�#��#%�#ܘ#��$��$@�$ �$�&��&��&�&��&��&E�&]�&��&��&-�&8G'�H'�Q'�['�_'�e'}n'�q'�'��'��'��',�'n�'�'�=(�=( A(�K(@r(1u(�z(M�(=�(��(��(�(��)��)��) **�1*u8* ;*V?*aC*�J*gS*>�*L�*�*�+2�+��+`�+��+��+��+B�+�,O,1q-�t-v�-��-�-��-]�-<�-ګ-̵-��-�-z�-�.@.�@/FB/q`/�/�/u 0�!0NH0QK0�Q0R�0��0D�0O�0��0c�0�0��0��0[�0d�0��01�11�1�51CJ1�O1��16�1��1v�1��1��1 �1��1; 3 03O13�13 33Y43`;3,S3�n3Kx3�|3��3C�3��3 4W=4�@4V4�Y4�Z40�4D�4��4,�4K5-O5�X5%}5"~5^�5z�5c�5*�5�5 �5�6t�6�&7 )7')7,7�V7��7�7A�7�<8��8��8�9�0949{L9 Z9�[9bg9�g9Dp9Q[:�\:^:�x: }:��:<�:��:��:��:�;;W;�;�;�!;t#;F:;pS;jV;OX;<�;�; �;��;�;��;��;��;��;��;��;-�;_�;M�;�<��<0=�"= $=�l=��=*�=��=b->k4>d�>2�>��>��>��>��>/i?�o?z?'�?�?��?��?��?E�?��?��?��?61@|1@�3@f5@T<@h?@s�@k�@��@1~AHAŀA �A��AʎA��Ar�A��A#�Ah�A_ B6B�B�Bg@BNBB�CB�NB|YB�pB��BWC�C� C�C�CpCi"C�"C�@C�C�C��C�E�$E|aE1�EJ�E�E��E��E �E��E5�E��E3DG<�G@�GJ}I�~I��Ig�Is�I��I`�I��I��I��I��I�J~#J (JzIJ�XJ�]J�bJ�hJ�jJӞJS�L��L��LSN�SN*ZN\NY]N[yN�NԙN��NK�N��N0AO_BOSZO�mO�pO�qO��O�P�P�P."P�'P��PY�P��P��P9�P�P��P]�P��PsQ�7Q=?Qp�QۧQ��QK�QR_'R�'R�(R,R_�S�{UV�U��UV� V^ V�"VJ(VـV��V{�V��VОV�V�V}�Vp�V�V�*W-W50W�4W.;W�;W@?WxBW%cW zW�W�WБW��W|�W��W��X�X0�XF:Yx=Y�>YEYPyZ�yZ�Z`�Z��Zi�Z��ZT�ZQ�ZY�Z��ZG�Z��Z��ZT [V[0[�[�[�[ک[�[Ѳ[<�[$�[S�[� \\& \�"\Bf�Dfxwf�zfU{f�fp�f@�f�-gig�ng�rg�sgBwg�yg�gu�gj�gZ�g��gh�h!�h��h��h��h��h�h�h�h�h2�h��h�iwi~i�~i3�i�iʠio�jF�jDkG�kO�ks�k&Ml �l<�l��lQ�l��l�li�l��l��l' m� m�m�mIm7m� m /mI7m�}+}�,} A}4L}4M}7b}�g}�p}_r}��}��V�dy�����f�g� %��ŀ?΀�Ɓ́���̙�/�����桄V(��:��p��x�耇„�������Т���������������cÇ�LJm���� �� �o?��?�3����K�Q�TQ�V�Z��\��d��r��r��u��}�b������͊�͊�2��6��6��:��:�]M�mP��X��Y�d[��_�bl�������8���ŋ�ƋwNj̋*Ӌ�ԋ��\��l��i%��'��,��b��g����g��5Ȍ�͌�ҌM����ÍNӍ׍~��A�z�S�t �I)�}-�B=��A��Q��]�m^�#ΎJ_��e�,g��k������r��������I��[��0�<�����"�I@��A��Z��a��b�g��o�p�璒h�����կ����㸓���~����ɕ�̕\͕�֕������ ��#�t+�O� R��Y��[�ga�f��y�q��n��^������0��9��@�%p����̗�ח����f�q�����f��8��$˘.И�Ә�֘٘CۘR���g�,N�tN��_��j�r�u�ZimpleX/starter-hugo-academiccontent/publication/shaDow-GNN/cite.bib0 @inproceedings{ zeng2021decoupling, title={Decoupling the Depth and Scope of Graph Neural Networks}, author={ and and and and and and and and }, booktitle={Advances in Neural Information Processing Systems}, editor={ and and and }, year={2021}, url={https://openreview.net/forum?id=_IY3_4psXuf} } A4-computational-geometry.tex1-10 \documentclass{beamer} \usepackage{algpseudocode, color, colortbl, listings, MnSymbol} \usepackage{hyperref} \hypersetup{ colorlinks=true, urlcolor=blue, } \usepackage{tikz, xcolor} \usetheme{Montpellier} \usecolortheme{rose} % page numbers, from % https://tex.stackexchange.com/questions/137022/how-to-insert-page-number-in-beamer-navigation-symbols \expandafter\def\expandafter\insertshorttitle\expandafter{% \insertshorttitle\hfill% \insertframenumber\,/\,\inserttotalframenumber} \definecolor{Gray}{gray}{0.8} \newcolumntype{g}{>{\columncolor{Gray}}c} \newcommand{\stanza}{ \\~\ } \title{14. Computational Geometry Introduction} \subtitle{CPSC 535} \author{} \institute{ \includegraphics[height=2cm]{csuf-logo-cmyk} } \date{\includegraphics[height=14pt]{by} \\ {\tiny This work is licensed under a \href{http://creativecommons.org/licenses/by/4.0/}{Creative Commons Attribution 4.0 International License}. }} \begin{document} \begin{frame} \titlepage \end{frame} \begin{frame} \frametitle{Big Idea: Output Sensitive Algorithm} \begin{itemize} \item \textbf{input sensitive}: time efficiency is a function of the input e.g. size $n$, \# edges $m$ \item \textbf{output sensitive}: efficiency is also a function of the \emph{output} size e.g. \# items returned \item most relevant when the size of the output could be the bottleneck \end{itemize} \end{frame} \begin{frame} \frametitle{Computational Geometry} \textbf{computational $X$}: interdisciplinary study of computer science with $X$ \stanza (computational finance, epidemiology, physics, finance, etc.) \stanza \emph{computational geometry (CG):} algorithms, data structures, asymptotic analysis, of geometric objects: points, lines, circles, triangle meshes, etc. \end{frame} \begin{frame} \frametitle{Computational Geometry Applications} Applications of CG: \begin{itemize} \item 3D computer graphics \item graphical user interfaces (GUIs) \item geographic information systems (GIS), geographic databases \item scene reconstruction, self-driving cars (e.g. LIDAR) \item business operations research (e.g. linear programming, aircraft control) \item manufacturing (e.g. feasibility of assembly, castings) \end{itemize} \end{frame} \begin{frame} \frametitle{Putting the Geo in CG} Some general algorithms can actually solve geometric problems efficiently, without any awareness of geometry. \stanza \begin{columns} \begin{column}{0.7 \textwidth} \emph{bounding box problem} \\ \textbf{input}: set of 2D points $P=\{p_1, p_2, \ldots, p_n\}$ \\ \textbf{output}: points $tl=(x_l, y_t)$ and $rb=(x_r, y_b)$ such that the rectangle with top-left corner $tl$ and bottom-right corner $rb$ contains $P$ \stanza \end{column} \begin{column}{0.3 \textwidth} \begin{center} \begin{tikzpicture} \draw [color=red] (0, 0) -- (3, 0) -- (3, 2) -- (0, 2) -- (0, 0); \foreach \Point in { % definining the bounding box (0, 0), (3, 2), % 10 random points (2.0951240918974277, 1.7968031777699034), (1.6893802109325138, 0.3988643181954554), (0.4478056694086142, 1.434611748573361), (1.21650131559343, 0.3708718939225579), (0.3622805800271429, 0.03376103132220609), (0.48201024301923834, 0.47617070155843133), (1.8378918317441353, 0.6021102667795148), (0.7819121241895028, 0.6833500815401432), (1.155185084745803, 0.05634877501390312), (1.8516975398072102, 1.5517971681158893) } { \draw [fill=black] \Point circle [radius=2pt]; } \end{tikzpicture} \end{center} \end{column} \end{columns} Na\"{i}ve, optimal algorithm: $x_l = \min x, y_t = \max y, x_r = \max x, y_b = \min y$; $\Theta(n)$ time \stanza Computational geometers are more interested when geometric properties matter. \end{frame} \begin{frame} \frametitle{Line Segment Predicates} We can use arithmetic to answer any of the following predicates (questions) about points $p_0, p_1, p_2, p_3$ in $\Theta(1)$ time: \stanza \begin{enumerate} \item Is line segment $\overline{p_0 p_1}$ clockwise from $\overline{p_0 p_2}$ around the common endpoint $p_0$? \item If we follow $\overline{p_0 p_1}$ and then $\overline{p_1 p_2}$, do we turn right or left? \item Do line segments $\overline{p_0 p_1}$ and $\overline{p_2 p_3}$ intersect? \stanza \end{enumerate} $\implies$ We may use any of these in pseudocode. \end{frame} \begin{frame} \frametitle{Line Segment Predicates} \begin{center} \includegraphics[width=4cm]{segment_clockwise.jpg} \hspace{2cm} \includegraphics[width=4cm]{segment_lr.jpg} \\ \vspace{2cm} \includegraphics[width=6cm]{segment_intersect.jpg} \end{center} \end{frame} \begin{frame} \frametitle{Degeneracy and Non-Degeneracy Assumptions} \textbf{degenerate} object: has the proper shape/type, but the values are a special case that betrays the spirit of the definition \stanza \emph{Example:} triangle $\equiv$ three points $(p_1, p_2, p_3)$ \\ degenerate triangle: $p_1=p_2=p_3$, or all points colinear \stanza \begin{center} \includegraphics[width=6cm]{degenerate_triangle.jpg} \end{center} \end{frame} \begin{frame} \frametitle{Non-Degeneracy Assumptions} \textbf{non-degeneracy assumption}: \begin{itemize} \item constraint that input to a CG algorithm is not degenerate in specific ways \item simplifies algorithm design \item assume that in practice, some combination of \begin{itemize} \item degeneracies do not occur \item input can be preprocessed to remove degeneracies \item implementer can modify algorithm to handle degeneracies \end{itemize} \end{itemize} \end{frame} \begin{frame} \frametitle{Sweep Algorithms} A pattern in CG algorithms: \begin{itemize} \item \emph{line sweep:} envision a line ``sweeping'' through the input \item e.g. a vertical line sweeping left-to-right \item helps us visualize a 2D situation as a 1D situation that changes over time \item like duality, doesn't actually change the problem, but might help us problem-solve \item generalizes to higher dimensions e.g. plane sweep in 3D, hyperplane sweep in any dimension \end{itemize} \end{frame} \begin{frame} \frametitle{Sweep Algorithms} \begin{center} \includegraphics[width=6cm]{line_sweep.jpg} \end{center} \end{frame} \end{document} \subsection{Random Forests} \label{random_forests} \noindent{\bf Description} \smallskip Random forest is one of the most successful machine learning methods for classification and regression. It is an ensemble learning method that creates a model composed of a set of tree models. This implementation is well-suited to handle large-scale data and builds a random forest model for classification in parallel.\\ \smallskip \noindent{\bf Usage} \smallskip {\hangindent=\parindent\noindent\it% {\tt{}-f }path/\/{\tt{}random-forest.dml} {\tt{} -nvargs} {\tt{} X=}path/file {\tt{} Y=}path/file {\tt{} R=}path/file {\tt{} bins=}integer {\tt{} depth=}integer {\tt{} num\_leaf=}integer {\tt{} num\_samples=}integer {\tt{} num\_trees=}integer {\tt{} subsamp\_rate=}double {\tt{} feature\_subset=}double {\tt{} impurity=}Gini$\mid$entropy {\tt{} M=}path/file {\tt{} C=}path/file {\tt{} S\_map=}path/file {\tt{} C\_map=}path/file {\tt{} fmt=}format } \smallskip \noindent{\bf Usage: Prediction} \smallskip {\hangindent=\parindent\noindent\it% {\tt{}-f }path/\/{\tt{}random-forest-predict.dml} {\tt{} -nvargs} {\tt{} X=}path/file {\tt{} Y=}path/file {\tt{} R=}path/file {\tt{} M=}path/file {\tt{} C=}path/file {\tt{} P=}path/file {\tt{} A=}path/file {\tt{} OOB=}path/file {\tt{} CM=}path/file {\tt{} fmt=}format }\smallskip \noindent{\bf Arguments} \begin{Description} \item[{\tt X}:] Location (on HDFS) to read the matrix of feature vectors; each row constitutes one feature vector. Note that categorical features in $X$ need to be both recoded and dummy coded. \item[{\tt Y}:] Location (on HDFS) to read the matrix of (categorical) labels that correspond to feature vectors in $X$. Note that classes are assumed to be both recoded and dummy coded. This argument is optional for prediction. \item[{\tt R}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to read matrix $R$ which for each feature in $X$ contains column-ids (first column), start indices (second column), and end indices (third column). If $R$ is not provided by default all features are assumed to be continuous-valued. \item[{\tt bins}:] (default:\mbox{ }{\tt 20}) Number of thresholds to choose for each continuous-valued feature (determined by equi-height binning). \item[{\tt depth}:] (default:\mbox{ }{\tt 25}) Maximum depth of the learned trees in the random forest model \item[{\tt num\_leaf}:] (default:\mbox{ }{\tt 10}) Parameter that controls pruning. The tree is not expanded if a node receives less than {\tt num\_leaf} training examples. \item[{\tt num\_samples}:] (default:\mbox{ }{\tt 3000}) Parameter that decides when to switch to in-memory building of the subtrees in each tree of the random forest model. If a node $v$ receives less than {\tt num\_samples} training examples then this implementation switches to an in-memory subtree building procedure to build the subtree under $v$ in its entirety. \item[{\tt num\_trees}:] (default:\mbox{ }{\tt 10}) Number of trees to be learned in the random forest model \item[{\tt subsamp\_rate}:] (default:\mbox{ }{\tt 1.0}) Parameter controlling the size of each tree in the random forest model; samples are selected from a Poisson distribution with parameter {\tt subsamp\_rate}. \item[{\tt feature\_subset}:] (default:\mbox{ }{\tt 0.5}) Parameter that controls the number of feature used as candidates for splitting at each tree node as a power of the number of features in the data, i.e., assuming the training set has $D$ features $D^{\tt feature\_subset}$ are used at each tree node. \item[{\tt impurity}:] (default:\mbox{ }{\tt "Gini"}) Impurity measure used at internal nodes of the trees in the random forest model for selecting which features to split on. Possible value are entropy or Gini. \item[{\tt M}:] Location (on HDFS) to write matrix $M$ containing the learned random forest (see Section~\ref{sec:decision_trees} and below for the schema) \item[{\tt C}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to store the number of counts (generated according to a Poisson distribution with parameter {\tt subsamp\_rate}) for each feature vector. Note that this argument is optional. If Out-Of-Bag (OOB) error estimate needs to be computed this parameter is passed as input to {\tt random-forest-predict.dml}. \item[{\tt A}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to store the testing accuracy (\%) from a held-out test set during prediction. Note that this argument is optional. \item[{\tt OOB}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to store the Out-Of-Bag (OOB) error estimate of the training set. Note that the matrix of sample counts (stored at {\tt C}) needs to be provided for computing OOB error estimate. Note that this argument is optional. \item[{\tt P}:] Location (on HDFS) to store predictions for a held-out test set \item[{\tt CM}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to store the confusion matrix computed using a held-out test set. Note that this argument is optional. \item[{\tt S\_map}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to write the mappings from the continuous-valued feature-ids to the global feature-ids in $X$ (see below for details). Note that this argument is optional. \item[{\tt C\_map}:] (default:\mbox{ }{\tt " "}) Location (on HDFS) to write the mappings from the categorical feature-ids to the global feature-ids in $X$ (see below for details). Note that this argument is optional. \item[{\tt fmt}:] (default:\mbox{ }{\tt "text"}) Matrix file output format, such as {\tt text}, {\tt mm}, or {\tt csv}; see read/write functions in SystemML Language Reference for details. \end{Description} \noindent{\bf Details} \smallskip Random forests~\cite{Breiman01:rforest} are learning algorithms for ensembles of decision trees. The main idea is to build a number of decision trees on bootstrapped training samples, i.e., by taking repeatedly samples from a (single) training set. Moreover, instead of considering all the features when building the trees only a random subset of the features---typically $\approx \sqrt{D}$, where $D$ is the number of features---is chosen each time a split test at a tree node is performed. This procedure {\it decorrelates} the trees and makes it less prone to overfitting. To build decision trees we utilize the techniques discussed in Section~\ref{sec:decision_trees} proposed in~\cite{PandaHBB09:dtree}; the implementation details are similar to those of the decision trees script. Below we review some features of our implementation which differ from {\tt decision-tree.dml}. \textbf{Bootstrapped sampling.} Each decision tree is fitted to a bootstrapped training set sampled with replacement (WR). To improve efficiency, we generate $N$ sample counts according to a Poisson distribution with parameter {\tt subsamp\_rate}, where $N$ denotes the total number of training points. These sample counts approximate WR sampling when $N$ is large enough and are generated upfront for each decision tree. \textbf{Bagging.} Decision trees suffer from {\it high variance} resulting in different models whenever trained on a random subsets of the data points. {\it Bagging} is a general-purpose method to reduce the variance of a statistical learning method like decision trees. In the context of decision trees (for classification), for a given test feature vector the prediction is computed by taking a {\it majority vote}: the overall prediction is the most commonly occurring class among all the tree predictions. \textbf{Out-Of-Bag error estimation.} Note that each bagged tree in a random forest model is trained on a subset (around $\frac{2}{3}$) of the observations (i.e., feature vectors). The remaining ($\frac{1}{3}$ of the) observations not used for training is called the {\it Out-Of-Bag} (OOB) observations. This gives us a straightforward way to estimate the test error: to predict the class label of each test observation $i$ we use the trees in which $i$ was OOB. Our {\tt random-forest-predict.dml} script provides the OOB error estimate for a given training set if requested. \textbf{Description of the model.} Similar to decision trees, the learned random forest model is presented in a matrix $M$ with at least 7 rows. The information stored in the model is similar to that of decision trees with the difference that the tree-ids are stored in the second row and rows $2,3,\ldots$ from the decision tree model are shifted by one. See Section~\ref{sec:decision_trees} for a description of the model. \smallskip \noindent{\bf Returns} \smallskip The matrix corresponding to the learned model is written to a file in the format specified. See Section~\ref{sec:decision_trees} where the details about the structure of the model matrix is described. Similar to {\tt decision-tree.dml}, $X$ is split into $X_\text{cont}$ and $X_\text{cat}$. If requested, the mappings of the continuous feature-ids in $X_\text{cont}$ (stored at {\tt S\_map}) as well as the categorical feature-ids in $X_\text{cat}$ (stored at {\tt C\_map}) to the global feature-ids in $X$ will be provided. The {\tt random-forest-predict.dml} script may compute one or more of predictions, accuracy, confusion matrix, and OOB error estimate in the requested output format depending on the input arguments used. \smallskip \noindent{\bf Examples} \smallskip {\hangindent=\parindent\noindent\tt \hml -f random-forest.dml -nvargs X=/user/biadmin/X.mtx Y=/user/biadmin/Y.mtx R=/user/biadmin/R.csv M=/user/biadmin/model.csv bins=20 depth=25 num\_leaf=10 num\_samples=3000 num\_trees=10 impurity=Gini fmt=csv }\smallskip \noindent To compute predictions: {\hangindent=\parindent\noindent\tt \hml -f random-forest-predict.dml -nvargs X=/user/biadmin/X.mtx Y=/user/biadmin/Y.mtx R=/user/biadmin/R.csv M=/user/biadmin/model.csv P=/user/biadmin/predictions.csv A=/user/biadmin/accuracy.csv CM=/user/biadmin/confusion.csv fmt=csv }\smallskip %\noindent{\bf References} % %\begin{itemize} %\item , , , and . \newblock{PLANET: massively parallel learning of tree ensembles with MapReduce}. In Proceedings of the VLDB Endowment, 2009. %\item . \newblock{Random Forests}. Machine Learning, 45(1), 5--32, 2001. %\end{itemize} \hypertarget{class_gr_model}{ \section{GrModel Class Reference} \label{class_gr_model}\index{GrModel@{GrModel}} } {\tt \#include $<$GrModel.h$>$} Inheritance diagram for GrModel::\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2cm]{class_gr_model} \end{center} \end{figure} \subsection*{Classes} \begin{CompactItemize} \item struct \textbf{SLightAttachment} \end{CompactItemize} \subsection*{Public Types} \begin{CompactItemize} \item typedef \hyperlink{class_u_fast_array}{UFastArray}$<$ \hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_model}{GrModel} $>$ $>$ \hyperlink{class_gr_model_1bc87aacfe8601c9096280a5257679cd}{ModelArray} \end{CompactItemize} \subsection*{Public Member Functions} \begin{CompactItemize} \item \hyperlink{class_gr_model_4def572ffb7a4575939f3286d566ac4a}{GrModel} (const \hyperlink{common__afx_8h_816fa58fd77499b0edb2c69ebe803d5c}{tstring} \&\hyperlink{glext__bak_8h_bb62efe59ccdd153ce42e1a418352209}{name}) \item \hyperlink{class_gr_model_15d52f8774a7781ad079f8bf3604226f}{GrModel} (const \hyperlink{class_u_path}{UPath} \&instName, \hyperlink{class_gr_polygon_node}{GrPolygonNode} $\ast$root) \item \hyperlink{class_gr_model_2cf8487be82f62f575f549e0081435ae}{GrModel} (const \hyperlink{class_u_path}{UPath} \&fileName, const \hyperlink{class_u_path}{UPath} \&instName, \hyperlink{class_u_reader}{UReader} \&reader) \item \hyperlink{class_gr_model_e48b47449f463e0b0a7b42943db1310c}{$\sim$GrModel} () \item const \hyperlink{class_u_path}{UPath} \& \hyperlink{class_gr_model_390da9636c9c396d1fa3241b72ff6166}{GetFileName} () const \item const \hyperlink{class_u_path}{UPath} \& \hyperlink{class_gr_model_e8a075163c88d6bf6623db37d65cd17a}{GetInstName} () const \item void \hyperlink{class_gr_model_155e00b717c10f201168f4dfc1f2c141}{SetPickable} (bool pickable) \item bool \hyperlink{class_gr_model_fa97c9215a2c2dcee2df132561f03698}{GetPickable} () const \item \hyperlink{class_gr_model}{GrModel} $\ast$ \hyperlink{class_gr_model_29d55ee007063d294ee9097c56d9f453}{Clone} (const \hyperlink{class_u_path}{UPath} \&instName, bool deep=false, bool lights=false) const \item \hyperlink{class_gr_model}{GrModel} $\ast$ \hyperlink{class_gr_model_fe7a05f880c1d67b55ef8130971836d6}{CloneFile} (const \hyperlink{class_u_path}{UPath} \&fileName, const \hyperlink{class_u_path}{UPath} \&instName) const \item \hyperlink{class_gr_model}{GrModel} $\ast$ \hyperlink{class_gr_model_ad017a6434f66a1051a57920b0561098}{GetParent} () const \item \hyperlink{class_gr_model_node}{GrModelNode} $\ast$ \hyperlink{class_gr_model_3497944f601c7427a96f66cafc3a151e}{GetParentNode} () const \item \hyperlink{class_gr_anim_player}{GrAnimPlayer} $\ast$ \hyperlink{class_gr_model_1d20cf8034ee3a0bd2fe1506fe81f73d}{GetAnimPlayer} (bool create=false) \item \hyperlink{class_gr_anim_player}{GrAnimPlayer} $\ast$ \hyperlink{class_gr_model_1759e9f0d11efd39892411dc80fd3312}{GetAnimPlayer} () const \item void \hyperlink{class_gr_model_8ae2497533c112953c1b4d4cc0e72cad}{SetDefaultAnim} (\hyperlink{class_gr_k_f_anim}{GrKFAnim} $\ast$anim) \item \hyperlink{class_gr_k_f_anim}{GrKFAnim} $\ast$ \hyperlink{class_gr_model_a9b644ea4efa86b73299934940c8b155}{GetDefaultAnim} () const \item void \hyperlink{class_gr_model_e99405dcb83200626b7c715755f70836}{SetLocal} (const \hyperlink{class_m_mat4x4}{MMat4x4} \&\hyperlink{glext__bak_8h_07993c0d92c1aeeb357ba0495c8b5325}{transform}) \item const \hyperlink{class_m_mat4x4}{MMat4x4} \& \hyperlink{class_gr_model_539780a5145b366b29e97a480846f67d}{GetLocal} () const \item void \hyperlink{class_gr_model_a3f1827d34de5f0c67d20103cf0d97bc}{SetWorld} (const \hyperlink{class_m_mat4x4}{MMat4x4} \&\hyperlink{glext__bak_8h_07993c0d92c1aeeb357ba0495c8b5325}{transform}) \item \hyperlink{class_m_mat4x4}{MMat4x4} \hyperlink{class_gr_model_624ce6372c1bdd5a077e51c16adb9242}{GetWorld} () \item const \hyperlink{class_m_a_a_box}{MAABox} \& \hyperlink{class_gr_model_3667043cfbdca6e413d4de9d0449aa5f}{GetBounds} () \item \hyperlink{class_gr_model_node}{GrModelNode} $\ast$ \hyperlink{class_gr_model_c5dbd78174e81a8cd4c64782d9b21bc9}{GetRootNode} () const \item void \hyperlink{class_gr_model_c232d8ab1c8c5e6349acd53eebc9db61}{AddChild} (\hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_model}{GrModel} $>$ model, \hyperlink{class_gr_model_node}{GrModelNode} $\ast$attachTo=0) \item unsigned int \hyperlink{class_gr_model_0d67f658382989a2dc3744dc88b28c22}{GetChildCount} () const \item \hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_model}{GrModel} $>$ \hyperlink{class_gr_model_462d3723b7fe4909bc9a92b3fcb0028c}{GetChild} (unsigned int idx) \item bool \hyperlink{class_gr_model_9148b455251ccd009790f95c126ec936}{RemoveChild} (\hyperlink{class_gr_model}{GrModel} $\ast$model) \item void \hyperlink{class_gr_model_79e3bfacd56b97bf810b971f064d4a91}{RemoveChildren} () \item void \hyperlink{class_gr_model_4d7fbb62dd1022da5a5fd25c6cb0e1fc}{AddLight} (\hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_light}{GrLight} $>$ light, \hyperlink{class_gr_model_node}{GrModelNode} $\ast$attachTo=0) \item unsigned int \hyperlink{class_gr_model_fbd44f70a2b06c82884eac31c40baeab}{GetLightCount} () const \item \hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_light}{GrLight} $>$ \hyperlink{class_gr_model_6f9177f0b2648997f0cb8dd4da0a481b}{GetLight} (unsigned int idx) \item \hyperlink{class_gr_model_node}{GrModelNode} $\ast$ \hyperlink{class_gr_model_7ff4ed21e0a840da8ce5570b31b68a15}{GetLightParent} (unsigned int idx) \item bool \hyperlink{class_gr_model_931e8f832b0858244d75b1dda2d5a70d}{RemoveLight} (\hyperlink{class_gr_light}{GrLight} $\ast$light) \item void \hyperlink{class_gr_model_95abd2705120c297a44287fe1d173224}{RemoveLights} () \item \hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_model}{GrModel} $>$ \hyperlink{class_gr_model_edeff0285708b660e0c21e2cb65da360}{FindModelByInstName} (const \hyperlink{class_u_path}{UPath} \&instName) \item void \hyperlink{class_gr_model_51f1bc0e7546b45969e3d40f600f7690}{FindModelsByFileName} (\hyperlink{class_u_fast_array}{UFastArray}$<$ \hyperlink{class_u_ref}{URef}$<$ \hyperlink{class_gr_model}{GrModel} $>$ $>$ \&models, const \hyperlink{class_u_path}{UPath} \&fileName) \item void \hyperlink{class_gr_model_0b9f56be2def3c655a5a37ed425539c4}{GetMeshInsts} (\hyperlink{class_u_fast_array}{UFastArray}$<$ \hyperlink{class_gr_mesh_inst}{GrMeshInst} $\ast$ $>$ \&meshInsts) \item void \hyperlink{class_gr_model_2c9c76c4d7715ab154fb9f89dd62d4d0}{GetAllMeshInsts} (\hyperlink{class_u_fast_array}{UFastArray}$<$ \hyperlink{class_gr_mesh_inst}{GrMeshInst} $\ast$ $>$ \&meshInsts) \item bool \hyperlink{class_gr_model_dbb1580e4409efde8590f9a6feacd0c9}{Update} (bool forceXFormUpdate=false) \item void \hyperlink{class_gr_model_fbc38dc0bd373a8b199535d55a06fc84}{UpdateXForms} () \item void \hyperlink{class_gr_model_12002759a1e3c55bf2ec71d21f197a0d}{EnsureUpdated} () \item bool \hyperlink{class_gr_model_fbbc5426863525467f35ff2df3cca3ce}{Pick} (float \&dist, const \hyperlink{class_m_ray}{MRay} \&ray) \item bool \hyperlink{class_gr_model_525915234ad6fe9e03ee62e699d27fad}{Pick} (\hyperlink{class_gr_model}{GrModel} $\ast$\&model, \hyperlink{class_gr_mesh_inst}{GrMeshInst} $\ast$\&meshInst, \hyperlink{struct_gr_mesh_1_1_s_tri_collision}{GrMesh::STriCollision} \&hit, const \hyperlink{class_m_ray}{MRay} \&ray) \item void \hyperlink{class_gr_model_89501566d6de6f63cda4527fb7cae760}{Traverse} (\hyperlink{class_gr_scene_traverser}{GrSceneTraverser} $\ast$traverser) \item void \hyperlink{class_gr_model_a125d73b384a3353dfbac65a93588282}{DbgDrawHierarchy} () \item void \hyperlink{class_gr_model_d3a2e4287aebf5351e971b9b64dbbe11}{Cache} () \item void \hyperlink{class_gr_model_2b4b150bcaf7982b1e70cb80549bc704}{Evict} () \item void \hyperlink{class_gr_model_c6e775a1ef9fd0321916e99a2de5db2d}{GetLitObjectsAndMarkVisible} (\hyperlink{class_gr_render_list}{GrRenderList} \&lightReceivers, const \hyperlink{class_gr_frustum}{GrFrustum} \&frustum, unsigned int frameId) const \item void \hyperlink{class_gr_model_4dbbe93eef470aa2339ae8921535780a}{GetLitObjects} (\hyperlink{class_gr_render_list}{GrRenderList} \&lightReceivers, \hyperlink{class_gr_render_list}{GrRenderList} $\ast$shadowCasters, const \hyperlink{class_m_sphere}{MSphere} \&sphere, unsigned int frameId) const \item void \hyperlink{class_gr_model_6533ef543b8238ce8434256f02be5f07}{GetLitObjects} (\hyperlink{class_gr_render_list}{GrRenderList} \&lightReceivers, \hyperlink{class_gr_render_list}{GrRenderList} $\ast$shadowCasters, const \hyperlink{class_gr_frustum}{GrFrustum} \&frustum, unsigned int frameId) const \item void \hyperlink{class_gr_model_d152158878fd8ee7d0f4f07915c54d01}{GetObjectsInFrustum} (\hyperlink{class_gr_render_list}{GrRenderList} \&objects, const \hyperlink{class_gr_frustum}{GrFrustum} \&frustum) const \item void \hyperlink{class_gr_model_94605ad2262faa9e3e858cd2f9b5d9d1}{Save} (\hyperlink{class_u_writer}{UWriter} \&writer) const \item void \hyperlink{class_gr_model_b29ae276eab8a6460d09a0f727765c40}{MarkAsDirty} () \item void \hyperlink{class_gr_model_905a993701b32fe30d1370deb32e35d2}{SetSceneModel} (bool sceneModel) \item bool \hyperlink{class_gr_model_2c82df629316012566ad43ae802e1ed9}{IsInScene} () const \item void \hyperlink{class_gr_model_7eb54cf62bdaf0cf4e32ef5628b54bc8}{SetUserPtr} (void $\ast$\hyperlink{glext__bak_8h_6a4f8a1a444e9080b297963b3db29fe0}{value}) \item void $\ast$ \hyperlink{class_gr_model_32d18da9318a7541316adaa52e8e2b99}{GetUserPtr} () const \end{CompactItemize} \subsection{Detailed Description} Definition at line 46 of file GrModel.h. \subsection{Member Typedef Documentation} \hypertarget{class_gr_model_1bc87aacfe8601c9096280a5257679cd}{ \index{GrModel@{GrModel}!ModelArray@{ModelArray}} \index{ModelArray@{ModelArray}!GrModel@{GrModel}} \subsubsection[{ModelArray}]{\setlength{\rightskip}{0pt plus 5cm}typedef {\bf UFastArray}$<$ {\bf URef}$<$ {\bf GrModel} $>$ $>$ {\bf GrModel::ModelArray}}} \label{class_gr_model_1bc87aacfe8601c9096280a5257679cd} Definition at line 49 of file GrModel.h. \subsection{Constructor \& Destructor Documentation} \hypertarget{class_gr_model_4def572ffb7a4575939f3286d566ac4a}{ \index{GrModel@{GrModel}!GrModel@{GrModel}} \index{GrModel@{GrModel}!GrModel@{GrModel}} \subsubsection[{GrModel}]{\setlength{\rightskip}{0pt plus 5cm}GrModel::GrModel (const {\bf tstring} \& {\em name})}} \label{class_gr_model_4def572ffb7a4575939f3286d566ac4a} Definition at line 51 of file GrModel.cpp.\hypertarget{class_gr_model_15d52f8774a7781ad079f8bf3604226f}{ \index{GrModel@{GrModel}!GrModel@{GrModel}} \index{GrModel@{GrModel}!GrModel@{GrModel}} \subsubsection[{GrModel}]{\setlength{\rightskip}{0pt plus 5cm}GrModel::GrModel (const {\bf UPath} \& {\em instName}, \/ {\bf GrPolygonNode} $\ast$ {\em root})}} \label{class_gr_model_15d52f8774a7781ad079f8bf3604226f} Definition at line 78 of file GrModel.cpp.\hypertarget{class_gr_model_2cf8487be82f62f575f549e0081435ae}{ \index{GrModel@{GrModel}!GrModel@{GrModel}} \index{GrModel@{GrModel}!GrModel@{GrModel}} \subsubsection[{GrModel}]{\setlength{\rightskip}{0pt plus 5cm}GrModel::GrModel (const {\bf UPath} \& {\em fileName}, \/ const {\bf UPath} \& {\em instName}, \/ {\bf UReader} \& {\em reader})}} \label{class_gr_model_2cf8487be82f62f575f549e0081435ae} Definition at line 122 of file GrModel.cpp.\hypertarget{class_gr_model_e48b47449f463e0b0a7b42943db1310c}{ \index{GrModel@{GrModel}!$\sim$GrModel@{$\sim$GrModel}} \index{$\sim$GrModel@{$\sim$GrModel}!GrModel@{GrModel}} \subsubsection[{$\sim$GrModel}]{\setlength{\rightskip}{0pt plus 5cm}GrModel::$\sim$GrModel ()}} \label{class_gr_model_e48b47449f463e0b0a7b42943db1310c} Definition at line 142 of file GrModel.cpp. \subsection{Member Function Documentation} \hypertarget{class_gr_model_c232d8ab1c8c5e6349acd53eebc9db61}{ \index{GrModel@{GrModel}!AddChild@{AddChild}} \index{AddChild@{AddChild}!GrModel@{GrModel}} \subsubsection[{AddChild}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::AddChild ({\bf URef}$<$ {\bf GrModel} $>$ {\em model}, \/ {\bf GrModelNode} $\ast$ {\em attachTo} = {\tt 0})}} \label{class_gr_model_c232d8ab1c8c5e6349acd53eebc9db61} Definition at line 322 of file GrModel.cpp.\hypertarget{class_gr_model_4d7fbb62dd1022da5a5fd25c6cb0e1fc}{ \index{GrModel@{GrModel}!AddLight@{AddLight}} \index{AddLight@{AddLight}!GrModel@{GrModel}} \subsubsection[{AddLight}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::AddLight ({\bf URef}$<$ {\bf GrLight} $>$ {\em light}, \/ {\bf GrModelNode} $\ast$ {\em attachTo} = {\tt 0})}} \label{class_gr_model_4d7fbb62dd1022da5a5fd25c6cb0e1fc} Definition at line 419 of file GrModel.cpp.\hypertarget{class_gr_model_d3a2e4287aebf5351e971b9b64dbbe11}{ \index{GrModel@{GrModel}!Cache@{Cache}} \index{Cache@{Cache}!GrModel@{GrModel}} \subsubsection[{Cache}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::Cache ()}} \label{class_gr_model_d3a2e4287aebf5351e971b9b64dbbe11} Definition at line 734 of file GrModel.cpp.\hypertarget{class_gr_model_29d55ee007063d294ee9097c56d9f453}{ \index{GrModel@{GrModel}!Clone@{Clone}} \index{Clone@{Clone}!GrModel@{GrModel}} \subsubsection[{Clone}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModel} $\ast$ GrModel::Clone (const {\bf UPath} \& {\em instName}, \/ bool {\em deep} = {\tt false}, \/ bool {\em lights} = {\tt false}) const}} \label{class_gr_model_29d55ee007063d294ee9097c56d9f453} Definition at line 157 of file GrModel.cpp.\hypertarget{class_gr_model_fe7a05f880c1d67b55ef8130971836d6}{ \index{GrModel@{GrModel}!CloneFile@{CloneFile}} \index{CloneFile@{CloneFile}!GrModel@{GrModel}} \subsubsection[{CloneFile}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModel} $\ast$ GrModel::CloneFile (const {\bf UPath} \& {\em fileName}, \/ const {\bf UPath} \& {\em instName}) const}} \label{class_gr_model_fe7a05f880c1d67b55ef8130971836d6} Definition at line 222 of file GrModel.cpp.\hypertarget{class_gr_model_a125d73b384a3353dfbac65a93588282}{ \index{GrModel@{GrModel}!DbgDrawHierarchy@{DbgDrawHierarchy}} \index{DbgDrawHierarchy@{DbgDrawHierarchy}!GrModel@{GrModel}} \subsubsection[{DbgDrawHierarchy}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::DbgDrawHierarchy ()}} \label{class_gr_model_a125d73b384a3353dfbac65a93588282} Definition at line 721 of file GrModel.cpp.\hypertarget{class_gr_model_12002759a1e3c55bf2ec71d21f197a0d}{ \index{GrModel@{GrModel}!EnsureUpdated@{EnsureUpdated}} \index{EnsureUpdated@{EnsureUpdated}!GrModel@{GrModel}} \subsubsection[{EnsureUpdated}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::EnsureUpdated ()}} \label{class_gr_model_12002759a1e3c55bf2ec71d21f197a0d} Definition at line 625 of file GrModel.cpp.\hypertarget{class_gr_model_2b4b150bcaf7982b1e70cb80549bc704}{ \index{GrModel@{GrModel}!Evict@{Evict}} \index{Evict@{Evict}!GrModel@{GrModel}} \subsubsection[{Evict}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::Evict ()}} \label{class_gr_model_2b4b150bcaf7982b1e70cb80549bc704} Definition at line 741 of file GrModel.cpp.\hypertarget{class_gr_model_edeff0285708b660e0c21e2cb65da360}{ \index{GrModel@{GrModel}!FindModelByInstName@{FindModelByInstName}} \index{FindModelByInstName@{FindModelByInstName}!GrModel@{GrModel}} \subsubsection[{FindModelByInstName}]{\setlength{\rightskip}{0pt plus 5cm}{\bf URef}$<$ {\bf GrModel} $>$ GrModel::FindModelByInstName (const {\bf UPath} \& {\em instName})}} \label{class_gr_model_edeff0285708b660e0c21e2cb65da360} Definition at line 499 of file GrModel.cpp.\hypertarget{class_gr_model_51f1bc0e7546b45969e3d40f600f7690}{ \index{GrModel@{GrModel}!FindModelsByFileName@{FindModelsByFileName}} \index{FindModelsByFileName@{FindModelsByFileName}!GrModel@{GrModel}} \subsubsection[{FindModelsByFileName}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::FindModelsByFileName ({\bf UFastArray}$<$ {\bf URef}$<$ {\bf GrModel} $>$ $>$ \& {\em models}, \/ const {\bf UPath} \& {\em fileName})}} \label{class_gr_model_51f1bc0e7546b45969e3d40f600f7690} Definition at line 519 of file GrModel.cpp.\hypertarget{class_gr_model_2c9c76c4d7715ab154fb9f89dd62d4d0}{ \index{GrModel@{GrModel}!GetAllMeshInsts@{GetAllMeshInsts}} \index{GetAllMeshInsts@{GetAllMeshInsts}!GrModel@{GrModel}} \subsubsection[{GetAllMeshInsts}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetAllMeshInsts ({\bf UFastArray}$<$ {\bf GrMeshInst} $\ast$ $>$ \& {\em meshInsts})}} \label{class_gr_model_2c9c76c4d7715ab154fb9f89dd62d4d0} Definition at line 543 of file GrModel.cpp.\hypertarget{class_gr_model_1759e9f0d11efd39892411dc80fd3312}{ \index{GrModel@{GrModel}!GetAnimPlayer@{GetAnimPlayer}} \index{GetAnimPlayer@{GetAnimPlayer}!GrModel@{GrModel}} \subsubsection[{GetAnimPlayer}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrAnimPlayer} $\ast$ GrModel::GetAnimPlayer () const}} \label{class_gr_model_1759e9f0d11efd39892411dc80fd3312} Definition at line 264 of file GrModel.cpp.\hypertarget{class_gr_model_1d20cf8034ee3a0bd2fe1506fe81f73d}{ \index{GrModel@{GrModel}!GetAnimPlayer@{GetAnimPlayer}} \index{GetAnimPlayer@{GetAnimPlayer}!GrModel@{GrModel}} \subsubsection[{GetAnimPlayer}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrAnimPlayer} $\ast$ GrModel::GetAnimPlayer (bool {\em create} = {\tt false})}} \label{class_gr_model_1d20cf8034ee3a0bd2fe1506fe81f73d} Definition at line 255 of file GrModel.cpp.\hypertarget{class_gr_model_3667043cfbdca6e413d4de9d0449aa5f}{ \index{GrModel@{GrModel}!GetBounds@{GetBounds}} \index{GetBounds@{GetBounds}!GrModel@{GrModel}} \subsubsection[{GetBounds}]{\setlength{\rightskip}{0pt plus 5cm}const {\bf MAABox}\& GrModel::GetBounds ()\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_3667043cfbdca6e413d4de9d0449aa5f} Definition at line 93 of file GrModel.h.\hypertarget{class_gr_model_462d3723b7fe4909bc9a92b3fcb0028c}{ \index{GrModel@{GrModel}!GetChild@{GetChild}} \index{GetChild@{GetChild}!GrModel@{GrModel}} \subsubsection[{GetChild}]{\setlength{\rightskip}{0pt plus 5cm}{\bf URef}$<$ {\bf GrModel} $>$ GrModel::GetChild (unsigned int {\em idx})}} \label{class_gr_model_462d3723b7fe4909bc9a92b3fcb0028c} Definition at line 355 of file GrModel.cpp.\hypertarget{class_gr_model_0d67f658382989a2dc3744dc88b28c22}{ \index{GrModel@{GrModel}!GetChildCount@{GetChildCount}} \index{GetChildCount@{GetChildCount}!GrModel@{GrModel}} \subsubsection[{GetChildCount}]{\setlength{\rightskip}{0pt plus 5cm}unsigned int GrModel::GetChildCount () const}} \label{class_gr_model_0d67f658382989a2dc3744dc88b28c22} Definition at line 348 of file GrModel.cpp.\hypertarget{class_gr_model_a9b644ea4efa86b73299934940c8b155}{ \index{GrModel@{GrModel}!GetDefaultAnim@{GetDefaultAnim}} \index{GetDefaultAnim@{GetDefaultAnim}!GrModel@{GrModel}} \subsubsection[{GetDefaultAnim}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrKFAnim}$\ast$ GrModel::GetDefaultAnim () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_a9b644ea4efa86b73299934940c8b155} Definition at line 81 of file GrModel.h.\hypertarget{class_gr_model_390da9636c9c396d1fa3241b72ff6166}{ \index{GrModel@{GrModel}!GetFileName@{GetFileName}} \index{GetFileName@{GetFileName}!GrModel@{GrModel}} \subsubsection[{GetFileName}]{\setlength{\rightskip}{0pt plus 5cm}const {\bf UPath}\& GrModel::GetFileName () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_390da9636c9c396d1fa3241b72ff6166} Definition at line 60 of file GrModel.h.\hypertarget{class_gr_model_e8a075163c88d6bf6623db37d65cd17a}{ \index{GrModel@{GrModel}!GetInstName@{GetInstName}} \index{GetInstName@{GetInstName}!GrModel@{GrModel}} \subsubsection[{GetInstName}]{\setlength{\rightskip}{0pt plus 5cm}const {\bf UPath}\& GrModel::GetInstName () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_e8a075163c88d6bf6623db37d65cd17a} Definition at line 61 of file GrModel.h.\hypertarget{class_gr_model_6f9177f0b2648997f0cb8dd4da0a481b}{ \index{GrModel@{GrModel}!GetLight@{GetLight}} \index{GetLight@{GetLight}!GrModel@{GrModel}} \subsubsection[{GetLight}]{\setlength{\rightskip}{0pt plus 5cm}{\bf URef}$<$ {\bf GrLight} $>$ GrModel::GetLight (unsigned int {\em idx})}} \label{class_gr_model_6f9177f0b2648997f0cb8dd4da0a481b} Definition at line 447 of file GrModel.cpp.\hypertarget{class_gr_model_fbd44f70a2b06c82884eac31c40baeab}{ \index{GrModel@{GrModel}!GetLightCount@{GetLightCount}} \index{GetLightCount@{GetLightCount}!GrModel@{GrModel}} \subsubsection[{GetLightCount}]{\setlength{\rightskip}{0pt plus 5cm}unsigned int GrModel::GetLightCount () const}} \label{class_gr_model_fbd44f70a2b06c82884eac31c40baeab} Definition at line 440 of file GrModel.cpp.\hypertarget{class_gr_model_7ff4ed21e0a840da8ce5570b31b68a15}{ \index{GrModel@{GrModel}!GetLightParent@{GetLightParent}} \index{GetLightParent@{GetLightParent}!GrModel@{GrModel}} \subsubsection[{GetLightParent}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModelNode} $\ast$ GrModel::GetLightParent (unsigned int {\em idx})}} \label{class_gr_model_7ff4ed21e0a840da8ce5570b31b68a15} Definition at line 454 of file GrModel.cpp.\hypertarget{class_gr_model_6533ef543b8238ce8434256f02be5f07}{ \index{GrModel@{GrModel}!GetLitObjects@{GetLitObjects}} \index{GetLitObjects@{GetLitObjects}!GrModel@{GrModel}} \subsubsection[{GetLitObjects}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetLitObjects ({\bf GrRenderList} \& {\em lightReceivers}, \/ {\bf GrRenderList} $\ast$ {\em shadowCasters}, \/ const {\bf GrFrustum} \& {\em frustum}, \/ unsigned int {\em frameId}) const}} \label{class_gr_model_6533ef543b8238ce8434256f02be5f07} Definition at line 792 of file GrModel.cpp.\hypertarget{class_gr_model_4dbbe93eef470aa2339ae8921535780a}{ \index{GrModel@{GrModel}!GetLitObjects@{GetLitObjects}} \index{GetLitObjects@{GetLitObjects}!GrModel@{GrModel}} \subsubsection[{GetLitObjects}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetLitObjects ({\bf GrRenderList} \& {\em lightReceivers}, \/ {\bf GrRenderList} $\ast$ {\em shadowCasters}, \/ const {\bf MSphere} \& {\em sphere}, \/ unsigned int {\em frameId}) const}} \label{class_gr_model_4dbbe93eef470aa2339ae8921535780a} Definition at line 769 of file GrModel.cpp.\hypertarget{class_gr_model_c6e775a1ef9fd0321916e99a2de5db2d}{ \index{GrModel@{GrModel}!GetLitObjectsAndMarkVisible@{GetLitObjectsAndMarkVisible}} \index{GetLitObjectsAndMarkVisible@{GetLitObjectsAndMarkVisible}!GrModel@{GrModel}} \subsubsection[{GetLitObjectsAndMarkVisible}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetLitObjectsAndMarkVisible ({\bf GrRenderList} \& {\em lightReceivers}, \/ const {\bf GrFrustum} \& {\em frustum}, \/ unsigned int {\em frameId}) const}} \label{class_gr_model_c6e775a1ef9fd0321916e99a2de5db2d} Definition at line 748 of file GrModel.cpp.\hypertarget{class_gr_model_539780a5145b366b29e97a480846f67d}{ \index{GrModel@{GrModel}!GetLocal@{GetLocal}} \index{GetLocal@{GetLocal}!GrModel@{GrModel}} \subsubsection[{GetLocal}]{\setlength{\rightskip}{0pt plus 5cm}const {\bf MMat4x4}\& GrModel::GetLocal () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_539780a5145b366b29e97a480846f67d} Definition at line 85 of file GrModel.h.\hypertarget{class_gr_model_0b9f56be2def3c655a5a37ed425539c4}{ \index{GrModel@{GrModel}!GetMeshInsts@{GetMeshInsts}} \index{GetMeshInsts@{GetMeshInsts}!GrModel@{GrModel}} \subsubsection[{GetMeshInsts}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetMeshInsts ({\bf UFastArray}$<$ {\bf GrMeshInst} $\ast$ $>$ \& {\em meshInsts})}} \label{class_gr_model_0b9f56be2def3c655a5a37ed425539c4} Definition at line 536 of file GrModel.cpp.\hypertarget{class_gr_model_d152158878fd8ee7d0f4f07915c54d01}{ \index{GrModel@{GrModel}!GetObjectsInFrustum@{GetObjectsInFrustum}} \index{GetObjectsInFrustum@{GetObjectsInFrustum}!GrModel@{GrModel}} \subsubsection[{GetObjectsInFrustum}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::GetObjectsInFrustum ({\bf GrRenderList} \& {\em objects}, \/ const {\bf GrFrustum} \& {\em frustum}) const}} \label{class_gr_model_d152158878fd8ee7d0f4f07915c54d01} Definition at line 816 of file GrModel.cpp.\hypertarget{class_gr_model_ad017a6434f66a1051a57920b0561098}{ \index{GrModel@{GrModel}!GetParent@{GetParent}} \index{GetParent@{GetParent}!GrModel@{GrModel}} \subsubsection[{GetParent}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModel}$\ast$ GrModel::GetParent () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_ad017a6434f66a1051a57920b0561098} Definition at line 71 of file GrModel.h.\hypertarget{class_gr_model_3497944f601c7427a96f66cafc3a151e}{ \index{GrModel@{GrModel}!GetParentNode@{GetParentNode}} \index{GetParentNode@{GetParentNode}!GrModel@{GrModel}} \subsubsection[{GetParentNode}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModelNode}$\ast$ GrModel::GetParentNode () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_3497944f601c7427a96f66cafc3a151e} Definition at line 72 of file GrModel.h.\hypertarget{class_gr_model_fa97c9215a2c2dcee2df132561f03698}{ \index{GrModel@{GrModel}!GetPickable@{GetPickable}} \index{GetPickable@{GetPickable}!GrModel@{GrModel}} \subsubsection[{GetPickable}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::GetPickable () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_fa97c9215a2c2dcee2df132561f03698} Definition at line 64 of file GrModel.h.\hypertarget{class_gr_model_c5dbd78174e81a8cd4c64782d9b21bc9}{ \index{GrModel@{GrModel}!GetRootNode@{GetRootNode}} \index{GetRootNode@{GetRootNode}!GrModel@{GrModel}} \subsubsection[{GetRootNode}]{\setlength{\rightskip}{0pt plus 5cm}{\bf GrModelNode}$\ast$ GrModel::GetRootNode () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_c5dbd78174e81a8cd4c64782d9b21bc9} Definition at line 96 of file GrModel.h.\hypertarget{class_gr_model_32d18da9318a7541316adaa52e8e2b99}{ \index{GrModel@{GrModel}!GetUserPtr@{GetUserPtr}} \index{GetUserPtr@{GetUserPtr}!GrModel@{GrModel}} \subsubsection[{GetUserPtr}]{\setlength{\rightskip}{0pt plus 5cm}void$\ast$ GrModel::GetUserPtr () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_32d18da9318a7541316adaa52e8e2b99} Definition at line 173 of file GrModel.h.\hypertarget{class_gr_model_624ce6372c1bdd5a077e51c16adb9242}{ \index{GrModel@{GrModel}!GetWorld@{GetWorld}} \index{GetWorld@{GetWorld}!GrModel@{GrModel}} \subsubsection[{GetWorld}]{\setlength{\rightskip}{0pt plus 5cm}{\bf MMat4x4} GrModel::GetWorld ()}} \label{class_gr_model_624ce6372c1bdd5a077e51c16adb9242} Definition at line 302 of file GrModel.cpp.\hypertarget{class_gr_model_2c82df629316012566ad43ae802e1ed9}{ \index{GrModel@{GrModel}!IsInScene@{IsInScene}} \index{IsInScene@{IsInScene}!GrModel@{GrModel}} \subsubsection[{IsInScene}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::IsInScene () const\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_2c82df629316012566ad43ae802e1ed9} Definition at line 168 of file GrModel.h.\hypertarget{class_gr_model_b29ae276eab8a6460d09a0f727765c40}{ \index{GrModel@{GrModel}!MarkAsDirty@{MarkAsDirty}} \index{MarkAsDirty@{MarkAsDirty}!GrModel@{GrModel}} \subsubsection[{MarkAsDirty}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::MarkAsDirty ()}} \label{class_gr_model_b29ae276eab8a6460d09a0f727765c40} Definition at line 906 of file GrModel.cpp.\hypertarget{class_gr_model_525915234ad6fe9e03ee62e699d27fad}{ \index{GrModel@{GrModel}!Pick@{Pick}} \index{Pick@{Pick}!GrModel@{GrModel}} \subsubsection[{Pick}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::Pick ({\bf GrModel} $\ast$\& {\em model}, \/ {\bf GrMeshInst} $\ast$\& {\em meshInst}, \/ {\bf GrMesh::STriCollision} \& {\em hit}, \/ const {\bf MRay} \& {\em ray})}} \label{class_gr_model_525915234ad6fe9e03ee62e699d27fad} Definition at line 654 of file GrModel.cpp.\hypertarget{class_gr_model_fbbc5426863525467f35ff2df3cca3ce}{ \index{GrModel@{GrModel}!Pick@{Pick}} \index{Pick@{Pick}!GrModel@{GrModel}} \subsubsection[{Pick}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::Pick (float \& {\em dist}, \/ const {\bf MRay} \& {\em ray})}} \label{class_gr_model_fbbc5426863525467f35ff2df3cca3ce} Definition at line 636 of file GrModel.cpp.\hypertarget{class_gr_model_9148b455251ccd009790f95c126ec936}{ \index{GrModel@{GrModel}!RemoveChild@{RemoveChild}} \index{RemoveChild@{RemoveChild}!GrModel@{GrModel}} \subsubsection[{RemoveChild}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::RemoveChild ({\bf GrModel} $\ast$ {\em model})}} \label{class_gr_model_9148b455251ccd009790f95c126ec936} Definition at line 363 of file GrModel.cpp.\hypertarget{class_gr_model_79e3bfacd56b97bf810b971f064d4a91}{ \index{GrModel@{GrModel}!RemoveChildren@{RemoveChildren}} \index{RemoveChildren@{RemoveChildren}!GrModel@{GrModel}} \subsubsection[{RemoveChildren}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::RemoveChildren ()}} \label{class_gr_model_79e3bfacd56b97bf810b971f064d4a91} Definition at line 396 of file GrModel.cpp.\hypertarget{class_gr_model_931e8f832b0858244d75b1dda2d5a70d}{ \index{GrModel@{GrModel}!RemoveLight@{RemoveLight}} \index{RemoveLight@{RemoveLight}!GrModel@{GrModel}} \subsubsection[{RemoveLight}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::RemoveLight ({\bf GrLight} $\ast$ {\em light})}} \label{class_gr_model_931e8f832b0858244d75b1dda2d5a70d} Definition at line 461 of file GrModel.cpp.\hypertarget{class_gr_model_95abd2705120c297a44287fe1d173224}{ \index{GrModel@{GrModel}!RemoveLights@{RemoveLights}} \index{RemoveLights@{RemoveLights}!GrModel@{GrModel}} \subsubsection[{RemoveLights}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::RemoveLights ()}} \label{class_gr_model_95abd2705120c297a44287fe1d173224} Definition at line 485 of file GrModel.cpp.\hypertarget{class_gr_model_94605ad2262faa9e3e858cd2f9b5d9d1}{ \index{GrModel@{GrModel}!Save@{Save}} \index{Save@{Save}!GrModel@{GrModel}} \subsubsection[{Save}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::Save ({\bf UWriter} \& {\em writer}) const}} \label{class_gr_model_94605ad2262faa9e3e858cd2f9b5d9d1} Definition at line 841 of file GrModel.cpp.\hypertarget{class_gr_model_8ae2497533c112953c1b4d4cc0e72cad}{ \index{GrModel@{GrModel}!SetDefaultAnim@{SetDefaultAnim}} \index{SetDefaultAnim@{SetDefaultAnim}!GrModel@{GrModel}} \subsubsection[{SetDefaultAnim}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetDefaultAnim ({\bf GrKFAnim} $\ast$ {\em anim})\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_8ae2497533c112953c1b4d4cc0e72cad} Definition at line 80 of file GrModel.h.\hypertarget{class_gr_model_e99405dcb83200626b7c715755f70836}{ \index{GrModel@{GrModel}!SetLocal@{SetLocal}} \index{SetLocal@{SetLocal}!GrModel@{GrModel}} \subsubsection[{SetLocal}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetLocal (const {\bf MMat4x4} \& {\em transform})}} \label{class_gr_model_e99405dcb83200626b7c715755f70836} Definition at line 271 of file GrModel.cpp.\hypertarget{class_gr_model_155e00b717c10f201168f4dfc1f2c141}{ \index{GrModel@{GrModel}!SetPickable@{SetPickable}} \index{SetPickable@{SetPickable}!GrModel@{GrModel}} \subsubsection[{SetPickable}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetPickable (bool {\em pickable})\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_155e00b717c10f201168f4dfc1f2c141} Definition at line 63 of file GrModel.h.\hypertarget{class_gr_model_905a993701b32fe30d1370deb32e35d2}{ \index{GrModel@{GrModel}!SetSceneModel@{SetSceneModel}} \index{SetSceneModel@{SetSceneModel}!GrModel@{GrModel}} \subsubsection[{SetSceneModel}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetSceneModel (bool {\em sceneModel})}} \label{class_gr_model_905a993701b32fe30d1370deb32e35d2} \hypertarget{class_gr_model_7eb54cf62bdaf0cf4e32ef5628b54bc8}{ \index{GrModel@{GrModel}!SetUserPtr@{SetUserPtr}} \index{SetUserPtr@{SetUserPtr}!GrModel@{GrModel}} \subsubsection[{SetUserPtr}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetUserPtr (void $\ast$ {\em value})\hspace{0.3cm}{\tt \mbox{[}inline\mbox{]}}}} \label{class_gr_model_7eb54cf62bdaf0cf4e32ef5628b54bc8} Definition at line 172 of file GrModel.h.\hypertarget{class_gr_model_a3f1827d34de5f0c67d20103cf0d97bc}{ \index{GrModel@{GrModel}!SetWorld@{SetWorld}} \index{SetWorld@{SetWorld}!GrModel@{GrModel}} \subsubsection[{SetWorld}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::SetWorld (const {\bf MMat4x4} \& {\em transform})}} \label{class_gr_model_a3f1827d34de5f0c67d20103cf0d97bc} Definition at line 279 of file GrModel.cpp.\hypertarget{class_gr_model_89501566d6de6f63cda4527fb7cae760}{ \index{GrModel@{GrModel}!Traverse@{Traverse}} \index{Traverse@{Traverse}!GrModel@{GrModel}} \subsubsection[{Traverse}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::Traverse ({\bf GrSceneTraverser} $\ast$ {\em traverser})}} \label{class_gr_model_89501566d6de6f63cda4527fb7cae760} Definition at line 704 of file GrModel.cpp.\hypertarget{class_gr_model_dbb1580e4409efde8590f9a6feacd0c9}{ \index{GrModel@{GrModel}!Update@{Update}} \index{Update@{Update}!GrModel@{GrModel}} \subsubsection[{Update}]{\setlength{\rightskip}{0pt plus 5cm}bool GrModel::Update (bool {\em forceXFormUpdate} = {\tt false})}} \label{class_gr_model_dbb1580e4409efde8590f9a6feacd0c9} Definition at line 559 of file GrModel.cpp.\hypertarget{class_gr_model_fbc38dc0bd373a8b199535d55a06fc84}{ \index{GrModel@{GrModel}!UpdateXForms@{UpdateXForms}} \index{UpdateXForms@{UpdateXForms}!GrModel@{GrModel}} \subsubsection[{UpdateXForms}]{\setlength{\rightskip}{0pt plus 5cm}void GrModel::UpdateXForms ()}} \label{class_gr_model_fbc38dc0bd373a8b199535d55a06fc84} Definition at line 611 of file GrModel.cpp. The documentation for this class was generated from the following files:\begin{CompactItemize} \item C:/Bootstrap/ProjectBX/Engine/Graphics/\hyperlink{_gr_model_8h}{GrModel.h}\item C:/Bootstrap/ProjectBX/Engine/Graphics/\hyperlink{_gr_model_8cpp}{GrModel.cpp}\end{CompactItemize} \hypertarget{_spark_button_handler_8cpp}{}\doxysection{Spark\+Button\+Handler.\+cpp File Reference} \label{_spark_button_handler_8cpp}\index{SparkButtonHandler.cpp@{SparkButtonHandler.cpp}} {\ttfamily \#include \char`\"{}Spark\+Button\+Handler.\+h\char`\"{}}\newline Include dependency graph for Spark\+Button\+Handler.\+cpp\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{_spark_button_handler_8cpp__incl} \end{center} \end{figure} doc/manuals/ocean/library/library.tex \documentstyle[11pt,a4]{article} \newcommand{\smc}[1]{{\sc #1}} \newcommand{\ii}{\"{\i}} % vertalen naar ps-file met dvips -t legal filenaam \input{psfig.sty} % path to the directory of this manual \newcommand{\topdir}{.} % Leave some vertical space between paragraphs; No paragraph indent: \setlength{\parskip}{0.6\baselineskip} \setlength{\parindent}{0cm} % this is how the computer displays command lines: \newcommand{\type}[1]{\begin{quote}% {\tt [op5/op5u9] ~ #1}% \end{quote}} % ...and this is how the computer displays two command lines: \newcommand{\typeb}[2]{\begin{quote}% {\tt [op5/op5u9] ~ #1% \newline% [op5/op5u9] ~ #2}% \end{quote}} % this is how to print mask names: \newcommand{\layer}[1]{{\sf #1}} \newcommand{\mask}[1]{{\sf #1}} % this is how to print names of nelsis tools: \newcommand{\tool}[1]{{\sl #1\/}} % this is how to print name of a button in an X window: \newcommand{\button}[1]{{\fbox{\small\rule[-0.5ex]{0mm}{2ex}$\!\!\!\!$\sf ~ #1}}} % this is how to print file names and cell names: \newcommand{\fname}[1]{{\tt #1}} % this is for reading figures from directory pictures \newcommand{\callpsfig}[2]{\psfig{figure=\topdir/plaatjes/#1,#2}} %voorbeeld:\centerline{\callpsfig{display.ps}{width=1.0\textwidth}} % To print an attention message: % macro for unimplemented figure %\newcommand{\futurecallpsfig}[2]{\centerline{\psfig{figure=futurepic.eps,#2}}} \newcommand{\attention}[2]{\begin{description}\item[{\bf #1}] #2 \end{description}} % to print a warning \newcommand{\warning}[1]{\begin{description}\item[{\bf Warning}:] #1 \end{description}} \sloppy \begin{document} \title{The 'fishbone' cell library} \author{, , ,\\ and \vspace*{1cm}\\ Delft University of Technology\\ faculty of Electrical Engineering\\ Delft, the Netherlands\\ e-mail: } \date{january 1993} \maketitle \input \topdir/intro.tex \input \topdir/iv110.tex \input \topdir/no210.tex \input \topdir/no310.tex \input \topdir/na210.tex \input \topdir/na310.tex \input \topdir/ex210.tex \input \topdir/buf20.tex \input \topdir/mu111.tex \input \topdir/mu210.tex \input \topdir/de211.tex \input \topdir/dfn10.tex \input \topdir/dfr11.tex \input \topdir/anabib.tex \input \topdir/osc10.tex \newpage \noindent\makebox[\textwidth]{\hrulefill} \tableofcontents \vspace*{3ex} \noindent\makebox[\textwidth]{\hrulefill} \newpage \end{document} andrazdeluisa/project-inareport/biblio.bib @article{bommarito-katz, title = {On the Stability of Community Detection Algorithms on Longitudinal Citation Data}, journal = {Procedia - Social and Behavioral Sciences}, volume = {4}, pages = {26 - 37}, year = {2010}, note = {Applications of Social Network Analysis}, issn = {1877-0428}, doi = {https://doi.org/10.1016/j.sbspro.2010.07.480}, url = {http://www.sciencedirect.com/science/article/pii/S1877042810018549}, author = { and and } } @article{karrer-levina, title = {Robustness of community structure in networks}, volume = {77}, issn = {1550-2376}, url = {http://dx.doi.org/10.1103/PhysRevE.77.046119}, doi = {10.1103/physreve.77.046119}, number = {4}, journal = {Physical Review E}, publisher = {American Physical Society (APS)}, author = { and .}, year = {2008}, month = {Apr} } @article{Newman_2006, title = {Finding community structure in networks using the eigenvectors of matrices}, volume = {74}, issn = {1550-2376}, url = {http://dx.doi.org/10.1103/PhysRevE.74.036104}, doi = {10.1103/physreve.74.036104}, number = {3}, journal = {Physical Review E}, publisher = {American Physical Society (APS)}, author = {.}, year = {2006}, month = {Sep} } @inproceedings{aadithya, author = {Aadithya, . and and Michalak, . and Jennings, .}, editor = {}, title = {Efficient Computation of the Shapley Value for Centrality in Networks}, booktitle = {Internet and Network Economics}, year = {2010}, publisher = {Springer Berlin Heidelberg}, address = {Berlin, Heidelberg}, pages = {1--13}, isbn = {978-3-642-17572-5} } @article{fenn-porter, author = { and and and and and and }, title = {Dynamical clustering of exchange rates}, journal = {Quantitative Finance}, volume = {12}, number = {10}, pages = {1493-1520}, year = {2012}, publisher = {Routledge}, doi = {10.1080/14697688.2012.668288}, url = {https://doi.org/10.1080/14697688.2012.668288}, eprint = {https://doi.org/10.1080/14697688.2012.668288} } @article{michalak, author = {}, year = {2014}, month = {02}, pages = {}, title = {Efficient Computation of the Shapley Value for Game-Theoretic Network Centrality}, volume = {46}, journal = {Journal of Artificial Intelligence Research}, doi = {10.1613/jair.3806} } @article{kannan-vempala, author = { }, title = {On Clusterings: Good, Bad and Spectral}, year = {2004}, issue_date = {May 2004}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, volume = {51}, number = {3}, issn = {0004-5411}, url = {https://doi.org/10.1145/990308.990313}, doi = {10.1145/990308.990313}, journal = {J. ACM}, month = may, pages = {497–515}, numpages = {19}, keywords = {graph algorithms, Clustering, spectral methods} } @article{narahari, author = {}, year = {2011}, month = {02}, pages = {130 - 147}, title = {A Shapley Value-Based Approach to Discover Influential Nodes in Social Networks}, volume = {8}, journal = {Automation Science and Engineering, IEEE Transactions on}, doi = {10.1109/TASE.2010.2052042} } @inproceedings{yang-leskovec, author = {J. {Yang} and J. {Leskovec}}, booktitle = {2012 IEEE 12th International Conference on Data Mining}, title = {Community-Affiliation Graph Model for Overlapping Network Community Detection}, year = {2012}, volume = {}, number = {}, pages = {1170-1175} } @article{girvan-newman, author = {. and Newman, .}, title = {Community structure in social and biological networks}, volume = {99}, number = {12}, pages = {7821--7826}, year = {2002}, doi = {10.1073/pnas.122653799}, publisher = {National Academy of Sciences}, issn = {0027-8424}, url = {https://www.pnas.org/content/99/12/7821}, journal = {Proceedings of the National Academy of Sciences} } @article{lfr, title = {Benchmark graphs for testing community detection algorithms}, author = {Lancichinetti, Andrea and and }, journal = {Phys. Rev. E}, volume = {78}, issue = {4}, pages = {046110}, numpages = {5}, year = {2008}, month = {Oct}, publisher = {American Physical Society}, doi = {10.1103/PhysRevE.78.046110}, url = {https://link.aps.org/doi/10.1103/PhysRevE.78.046110} } @techreport{pagerank, number = {1999-66}, month = {November}, author = { and and and }, note = {Previous number = SIDL-WP-1999-0120}, title = {The PageRank Citation Ranking: Bringing Order to the Web.}, type = {Technical Report}, publisher = {Stanford InfoLab}, year = {1999}, institution = {Stanford InfoLab}, url = {http://ilpubs.stanford.edu:8090/422/}, } @article{Jensen_2015, title = {Detecting global bridges in networks}, volume = {4}, issn = {2051-1329}, url = {http://dx.doi.org/10.1093/comnet/cnv022}, doi = {10.1093/comnet/cnv022}, number = {3}, journal = {Journal of Complex Networks}, publisher = {Oxford University Press (OUP)}, author = { Morini, Karsai, Venturini, Tommaso and Vespignani, Alessandro and Jacomy, , and }, year = {2015}, month = {Oct}, pages = {319–329} } @article{freeman, issn = {00380431}, url = {http://www.jstor.org/stable/3033543}, abstract = {A Family of new measures of point and graph centrality based on early intuitions of Bavelas (1948) is introduced. These measures define centrality in terms of the degree to which a point falls on the shortest path between others and therefore has a potential for control of communication. They may be used to index centrality in any large or small network of symmetrical relations, whether connected or unconnected.}, author = {}, journal = {Sociometry}, number = {1}, pages = {35--41}, publisher = {[American Sociological Association, Sage Publications, Inc.]}, title = {A Set of Measures of Centrality Based on Betweenness}, volume = {40}, year = {1977} } @article{coscia-rossetti, author = { }, title = {Uncovering Hierarchical and Overlapping Communities with a Local-First Approach}, year = {2014}, issue_date = {October 2014}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, volume = {9}, number = {1}, issn = {1556-4681}, url = {https://doi.org/10.1145/2629511}, doi = {10.1145/2629511}, journal = {ACM Trans. Knowl. Discov. Data}, month = aug, articleno = {6}, numpages = {27}, keywords = {Complex networks, community discovery, data mining} } @inproceedings{demon, author = { }, title = {DEMON: A Local-First Discovery Method for Overlapping Communities}, year = {2012}, isbn = {9781450314626}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/2339530.2339630}, doi = {10.1145/2339530.2339630}, booktitle = {Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining}, pages = {615–623}, numpages = {9}, keywords = {community discovery, complex networks, data mining}, location = {Beijing, China}, series = {KDD ’12} } @article{palla-derenyi, author = { }, year = {2005}, month = {07}, pages = {814-818}, title = {Uncovering the overlapping community structure of complex networks in nature and society}, volume = {435}, journal = {Nature} } @misc{leskovec-lang-mahoney, title = {Empirical Comparison of Algorithms for Network Community Detection}, author = { and and }, year = {2010}, eprint = {1004.3539}, archiveprefix = {arXiv}, primaryclass = {cs.DS} } @misc{github-network, title = {Multi-scale Attributed Node Embedding}, author = { and and }, year = {2019}, eprint = {1909.13021}, archiveprefix = {arXiv}, primaryclass = {cs.LG} } @misc{carissimo, title = {Validation of community robustness}, journal = {Computational Statistics & Data Analysis}, volume = {120}, pages = {1-24}, year = {2018}, doi = {https://doi.org/10.1016/j.csda.2017.10.006}, url = {http://www.sciencedirect.com/science/article/pii/S0167947317302347}, author = { and and } } @article{wang-yang-fan, author = { and and }, year = {2011}, month = {11}, pages = {e27418}, title = {Identifying and Characterizing Nodes Important to Community Structure Using the Spectrum of the Graph}, volume = {6}, journal = {PloS one}, doi = {10.1371/journal.pone.0027418} } @article{Raghavan-2007, title = {Near linear time algorithm to detect community structures in large-scale networks}, volume = {76}, issn = {1550-2376}, url = {http://dx.doi.org/10.1103/PhysRevE.76.036106}, doi = {10.1103/physreve.76.036106}, number = {3}, journal = {Physical Review E}, publisher = {American Physical Society (APS)}, author = { and and }, year = {2007}, month = {Sep} } @misc{github-graph, title={Multi-scale Attributed Node Embedding}, author={ and and }, year={2019}, eprint={1909.13021}, archivePrefix={arXiv}, primaryClass={cs.LG} } @article{louvain, doi = {10.1088/1742-5468/2008/10/p10008}, url = {https://doi.org/10.1088%2F1742-5468%2F2008%2F10%2Fp10008}, year = 2008, month = {oct}, publisher = {{IOP} Publishing}, volume = {2008}, number = {10}, pages = {P10008}, author = { and and and }, title = {Fast unfolding of communities in large networks}, journal = {Journal of Statistical Mechanics: Theory and Experiment} } @misc{infomap, author = { and and }, title = {The MapEquation software package}, url = {http://www.mapequation.org} } \documentclass[letterpaper, twoside, 12pt]{book} \usepackage{packet} \parskip=1em \begin{document} {\Large Notes on MATH 2630 Final Exam (Spring 2015)} The final exam will take place during one of the following periods: \begin{itemize} \item For MWF 10am class: Wed 2015-05-06 0800-1030 \item For MWF 12noon class: Tues 2015-05-05 1200-1430 \end{itemize} If you prefer, you may take the final exam during the other section's timeslot. Unless many students in one section take the exam during the other section, these exams will be held in the usual classroom. You must take your exam during one of these times. The structure of the exam follows: \begin{itemize} \item 6 questions based on Tests 1-4 \item 3 questions based on Problems in Packets 1-4 \item choose 1 of 2 new types of questions based on Packets 1-4 \end{itemize} Each question is worth 10 points, for a total of 100 points. This total will be scaled to match the weight of your final exam, with rounding up to the next integer (see syllabus for details on final exam weighting). The final exam is run exactly like the Individual Tests, including the open notes/computers policy, and with no collaboration allowed between classmates and no communication allowed with others outside the classroom. You will of course have much more time to complete this exam (150 minutes for only twice the questions asked on a 40 minute test). \textbf{Suggestions for studying:} Focus first on the questions asked on Tests. Then go over as many Problems from the Packets as possible. It's probably not useful to attempt to cram for the new types of questions, but you will choose which of the two questions you want to respond to. \end{document}@inproceedings{sahay-etal-2019-modeling, title = "Modeling Intent, Dialog Policies and Response Adaptation for Goal-Oriented Interactions", author = " and . and and and ", booktitle = "Proceedings of the 23rd Workshop on the Semantics and Pragmatics of Dialogue - Full Papers", month = sep, year = "2019", address = "London, United Kingdom", publisher = "SEMDIAL", url = "http://semdial.org/anthology/Z19-Sahay_semdial_0019.pdf", }samples/thesis/sample/Appendices/AppendixA.tex % Appendix A \chapter{TEST CASES} \section{TEST CASES FOR ERROR LEVEL ANALYSIS} In the data pre-processing module, error level analysis shows the difference in the quality level of the image when it is resaved and also the error potential. \subsection{TESTCASES} \subsubsection{Authentic image} Authentic image with a minimum difference of 4 \begin{figure}[htp] \centering \includegraphics[width=10cm]{Figures/2.jpg} \caption{Authentic Input image} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=15cm]{Figures/as.PNG} \caption{Error displayed} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=10cm,height=6cm]{Figures/2_ela.png} \caption{Output ELA image} \label{fig:lion} \end{figure} \subsubsection{Fake image} The fake image with a maximum difference of 29 and its ELA image is shown. \begin{figure}[h!] \centering \includegraphics[width=10cm,height=6cm]{Figures/tiger_ela.png} \caption{ELA image} \label{fig:lion} \end{figure} \begin{figure}[h!] \centering \includegraphics[width=16cm]{Figures/ela-cmd.PNG} \caption{Error} \label{fig:lion} \end{figure} \newpage \section{OVERALL TESTCASES FOR THE SYSTEM} \subsection{TESTCASE 1} \subsubsection{Authentic image} \begin{figure}[htp] \centering \includegraphics[width=13cm]{Figures/Au_ani_00021.jpg} \caption{Authentic input image} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=15cm,height=4cm]{Figures/auth.PNG} \caption{Output} \label{fig:lion} \end{figure} \newpage \subsection{TESTCASE 2} \subsubsection{Spliced image} \begin{figure}[htp] \centering \includegraphics[width=10cm,height=7cm]{Figures/s1_resaved.jpg} \caption{Spliced input image} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=15cm,height=5cm]{Figures/splicecmd.png} \caption{Output} \label{fig:lion} \end{figure} \newpage \subsection{TESTCASE 3} \subsubsection{Copy move} \begin{figure}[htp] \centering \includegraphics[width=7cm,height=6cm]{Figures/c1.png} \caption{Copy move input image} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=15cm]{Figures/copycomd.PNG} \caption{Output} \label{fig:lion} \end{figure} \begin{figure}[htp] \centering \includegraphics[width=7cm,height=6cm]{Figures/c1_analyzed.jpg} \caption{Resaved image} \label{fig:lion} \end{figure} nasfarley88/mutliplication-tables0 \documentclass{article} \usepackage{unicode-math} \usepackage{multiplicationtables} \usepackage[margin=1.5cm]{geometry} \setmathfont[Scale=1.3]{Free Sans} \setmainfont[Scale=1.3]{Free Sans} \begin{document} \centering \noindent\mtableblank{2,7,6,4,0,12,5,3,11,9}{4,4,4,4,4,4,4,4,4,4,3,4}\vfill \mtable{2,7,6,4,0,12,5,3,11,9}{4,4,4,4,4,4,4,4,4,4}\vfill \mtablelarge{2,7,6,4,0,12,5,3,11,100}{4,4,4,4,4,4,4,4,4,100} % \mtabletikz{2,7,6,4,0,12,5,3,11,9}{4,4,4,4,4,4,4,4,4,4} \end{document} % Note that the text in the [] brackets is the one that will % appear in the table of contents, whilst the text in the {} % brackets will appear in the main thesis. %% CHAPTER HEADER ///////////////////////////////////////////////////////////////////////////////////// \chapter[GW Propagation under Variable Temperature Conditions]{GW Propagation under Variable Temperature Conditions} \label{ch:tempEffects} %% CHAPTER INTRODUCTION /////////////////////////////////////////////////////////////////////////////// %% INCLUDE SECTIONS /////////////////////////////////////////////////////////////////////////////////// %\input{Chapters/Chapter7/sec:tempProperties} %\input{Chapters/Chapter7/sec:tempSetup} %\input{Chapters/Chapter7/sec:tempResults} %\input{Chapters/Chapter7/sec:conclusionsTemp}docs/latex/group___p_w_r_ex___i_s___p_w_r___definitions.tex \hypertarget{group___p_w_r_ex___i_s___p_w_r___definitions}{}\section{P\+W\+R\+Ex Private macros to check input parameters} \label{group___p_w_r_ex___i_s___p_w_r___definitions}\index{P\+W\+R\+Ex Private macros to check input parameters@{P\+W\+R\+Ex Private macros to check input parameters}} \subsection*{Macros} \begin{DoxyCompactItemize} \item \#define {\bfseries I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE}(R\+E\+G\+U\+L\+A\+T\+OR) \item \#define {\bfseries I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE}(V\+O\+L\+T\+A\+GE) \item \#define {\bfseries I\+S\+\_\+\+P\+W\+R\+\_\+\+W\+A\+K\+E\+U\+P\+\_\+\+P\+IN}(P\+IN)~(((P\+IN) == P\+W\+R\+\_\+\+W\+A\+K\+E\+U\+P\+\_\+\+P\+I\+N1) $\vert$$\vert$ ((P\+IN) == P\+W\+R\+\_\+\+W\+A\+K\+E\+U\+P\+\_\+\+P\+I\+N2))\hypertarget{group___p_w_r_ex___i_s___p_w_r___definitions_gac6fcc59d6ff95b8feda1b228517f9c3f}{}\label{group___p_w_r_ex___i_s___p_w_r___definitions_gac6fcc59d6ff95b8feda1b228517f9c3f} \end{DoxyCompactItemize} \subsection{Detailed Description} \subsection{Macro Definition Documentation} \index{P\+W\+R\+Ex Private macros to check input parameters@{P\+W\+R\+Ex Private macros to check input parameters}!I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE@{I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE}} \index{I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE@{I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE}!P\+W\+R\+Ex Private macros to check input parameters@{P\+W\+R\+Ex Private macros to check input parameters}} \subsubsection[{\texorpdfstring{I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE}{IS_PWR_REGULATOR_UNDERDRIVE}}]{\setlength{\rightskip}{0pt plus 5cm}\#define I\+S\+\_\+\+P\+W\+R\+\_\+\+R\+E\+G\+U\+L\+A\+T\+O\+R\+\_\+\+U\+N\+D\+E\+R\+D\+R\+I\+VE( \begin{DoxyParamCaption} \item[{}]{R\+E\+G\+U\+L\+A\+T\+OR} \end{DoxyParamCaption} )}\hypertarget{group___p_w_r_ex___i_s___p_w_r___definitions_ga33dc716af19621b8c3ad42d7d41abef5}{}\label{group___p_w_r_ex___i_s___p_w_r___definitions_ga33dc716af19621b8c3ad42d7d41abef5} {\bfseries Value\+:} \begin{DoxyCode} (((REGULATOR) == PWR\_MAINREGULATOR\_UNDERDRIVE\_ON) || \(\backslash\) ((REGULATOR) == PWR\_LOWPOWERREGULATOR\_UNDERDRIVE\_ON)) \end{DoxyCode} \index{P\+W\+R\+Ex Private macros to check input parameters@{P\+W\+R\+Ex Private macros to check input parameters}!I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE@{I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE}} \index{I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE@{I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE}!P\+W\+R\+Ex Private macros to check input parameters@{P\+W\+R\+Ex Private macros to check input parameters}} \subsubsection[{\texorpdfstring{I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE}{IS_PWR_VOLTAGE_SCALING_RANGE}}]{\setlength{\rightskip}{0pt plus 5cm}\#define I\+S\+\_\+\+P\+W\+R\+\_\+\+V\+O\+L\+T\+A\+G\+E\+\_\+\+S\+C\+A\+L\+I\+N\+G\+\_\+\+R\+A\+N\+GE( \begin{DoxyParamCaption} \item[{}]{V\+O\+L\+T\+A\+GE} \end{DoxyParamCaption} )}\hypertarget{group___p_w_r_ex___i_s___p_w_r___definitions_ga007e15203cc13b9f50824ffc103e0a5c}{}\label{group___p_w_r_ex___i_s___p_w_r___definitions_ga007e15203cc13b9f50824ffc103e0a5c} {\bfseries Value\+:} \begin{DoxyCode} (((VOLTAGE) == PWR\_REGULATOR\_VOLTAGE\_SCALE1) || \(\backslash\) ((VOLTAGE) == PWR\_REGULATOR\_VOLTAGE\_SCALE2) || \(\backslash\) ((VOLTAGE) == PWR\_REGULATOR\_VOLTAGE\_SCALE3)) \end{DoxyCode} @inproceedings{srinivasan2021learning, title={Learning over Families of Sets-Hypergraph Representation Learning for Higher Order Tasks}, author={ Zheng, }, booktitle={Proceedings of the 2021 SIAM International Conference on Data Mining (SDM)}, pages={756--764}, year={2021}, organization={SIAM} } deleeuw/jansweb @techreport{deleeuw_R_73a, author = {.}, bdsk-url-1 = {Url = http://deleeuwpdx.net/janspubs/1973/reports/deleeuw_R_73a.pdf}, bdsk-url-2 = {https://drive.google.com/open?id=1HNj-P1EDyQv6HXxcp22UzXXJdUDBRgP5}, date-modified = {2019-03-16 17:45:20 -0700}, institution = {Department of Data Theory FSW/RUL}, number = {001-73}, title = {Estimation in Latent Class Analysis}, type = {Research Note}, url = {http://deleeuwpdx.net/janspubs/1973/reports/deleeuw_R_73a.pdf}, year = {1973} } \documentclass[a4paper]{paper} \usepackage{amsmath} \usepackage{amsthm} \usepackage{amsfonts} \usepackage[alphabetic]{amsrefs} \usepackage[T1]{fontenc} \usepackage[utf8]{inputenc} \usepackage{lmodern} \usepackage[english]{babel} \usepackage{sagetex} \usepackage{csquotes} \usepackage{tikz} \usepackage{adjustbox} \usepackage{inconsolata} \newcommand{\id}{\operatorname{id}} \newcommand{\BP}{BP} \newcommand{\us}{_\ast} \newcommand{\os}{^\ast} \newcommand{\QQ}{\mathbb Q} \newcommand{\FF}{\mathbb F} \newcommand{\ZZ}{\mathbb Z} \newcommand{\NN}{\mathbb N} \newcommand{\stt}[1]{{\tt#1}} \newcommand{\scrb}{{\mathcal B}} \newcommand{\scrh}{{\mathcal H}} \lstdefinestyle{SageInput}{ style=DefaultSageInput, basicstyle={\small\linespread{0.8}\ttfamily}, } \lstdefinestyle{SageOutput}{ style=DefaultSageOutput, basicstyle={\small\linespread{0.8}\ttfamily}, } \raggedright \setlength{\parskip}{8pt plus 5pt minus 2pt} \title{A toy implementation of the $\BP$-bar complex in Sage} \author{} \date{\today} \begin{document} \maketitle \begin{sagesilent} load("bpbar.py") \end{sagesilent} We describe a small Sage library that implements enough of the Bar complex for $\BP$ to compute some low-lying Novikov-Ext groups. More specifically, we implement a class \stt{BPBar} that represents the infinite tensor power \begin{align*} \scrb &= \pi\us\left(\BP\land\BP\land\cdots\right) = \Gamma\otimes_A\Gamma\otimes_A\Gamma\otimes_A\cdots \intertext{and also a class \stt{HBPBar} which represents} \scrh &= H\us\left(\BP\land\BP\land\cdots\right) \subset \QQ\otimes {\mathcal B} \end{align*} Here $(A,\Gamma) = (\BP\us,\BP\us\BP)$ is the $\BP$-Hopf algebroid and we think of $\BP^{\land\infty}$ as the colimit of the $\BP^{\land k}$ where $\BP^{\land k}\rightarrow \BP^{\land(k+1)}$ is given by $x\mapsto x\land 1$. This code was an attempt to recreate some computations by in Sage, and we thank him heartily for sharing his code and for many inspiring communications during the Homotopy Theory Summer in Berlin in 2018. \begin{section}{Overview of some classes} The $\scrh$ and $\scrb$ come equipped with classes \stt{BasisEnumerator}, \stt{BarBasis} that allow to loop through a basis of a $\BP^{\land s}$ in a fixed dimension. Specifically, an instance of \stt{BasisEnumerator} knows how to \begin{enumerate} \item loop through the weighted partitions $P_w(n)$ of an integer $n$ where $w=(w_1,\ldots,w_k)$ is an arbitrary list of weights \item efficiently compute the reverse map $P_w(n)\rightarrow \NN_0$ (function \stt{seqno}) \end{enumerate} The \stt{BarBasis} translates this information from partitions to elements of $\scrb$ and $\scrh$. The structure maps of $\scrb$ and $\scrh$ are all derived from the map $\eta_R$ which on $\BP^{\land\infty}$ is given by $x\mapsto 1\land x$. For the generators $m_k$, $t_j$ of $\scrh$ this is given by $$\eta_R(m_n) = \sum_{i+j=n} m_i t_j^{p^i},\quad \eta_R(\underbrace{1\vert\cdots\vert1\vert t_k}_{\text{$p$ ones}}) = \underbrace{1\vert\cdots\vert1\vert t_k}_{\text{$p+1$ ones}}. $$ This is later used to compute the $\Delta(t_n)$ via $\Delta\eta_R(m_n) = (\eta_R\otimes\id)\eta_R(m_n)$. In Sage one can compute these as follows (here for $p=3$): \begin{sagecommandline} sage: M = HBPBar(3) ; m,t = M.gens() sage: e,D = M.etaR,M.Delta sage: m[1], e(m[1]) sage: m[2], e(m[2]) sage: m[3], e(m[3]) sage: # an iterated eta_R of t2 sage: t[2], e(t[2]), e(e(t[2])) sage: # two coproducts sage: D(t[1]) sage: D(t[2]) \end{sagecommandline} There is a class \stt{ArakiGens} that implements the isomorphisms $\scrb\otimes\QQ \leftrightarrow \scrh\otimes\QQ$. This is used to transport the structure formulas from \stt{HBPBar} to \stt{BPBar}: \begin{sagecommandline} sage: A = ArakiGens(3) sage: A.mapm2v() sage: A.mapm2v()(m[1]) sage: A.mapm2v()(m[2]) sage: A.mapv2m() sage: A.mapv2m()(A.mapm2v()(m[2])) \end{sagecommandline} For $\scrb$ at $p=3$ we then get, for example \begin{sagecommandline} sage: B = BPBar(3) ; v,t = B.gens() sage: B.etaR(v[1]) sage: B.etaR(v[2]) sage: B.Delta(t[1]) sage: B.Delta(t[2]) \end{sagecommandline} One can use a different base ring for \stt{BPBar} to compute with a reduction of $\scrb$ modulo some $p^N$: \begin{sagecommandline} sage: B = BPBar(3,IntegerModRing(3**2)) ; v,t = B.gens() sage: B.Delta(t[3]) \end{sagecommandline} For the bar complex we also need the \enquote{higher coproducts} $\Delta_k$ that are induced by the map $$a_1\land a_2\land\cdots \mapsto a_1\land\cdots a_k\land 1\land a_{k+1}\land\cdots$$ on spectra. These are available as \stt{Delta\_map(k)}: \begin{sagecommandline} sage: X = BPBar(2) ; v,t = X.gens() sage: e,D = X.etaR, X.Delta_map sage: D(3) sage: D(2)(t[1]) sage: D(2)(e(t[1])) sage: D(2)(e(e(t[1]))) sage: D(2)(e(e(e(t[1])))) \end{sagecommandline} Under the hood a lot of (hopefully well-positioned) caching is going on, to make sure that the basic structure formulas (including the $\eta_R(v_j)$ and $\Delta_k(t_j)$) are only computed once. This is particularly relevant when such a structure map is evaluated on a monomial: \begin{sagecommandline} sage: D(2)(e(t[1]**3*t[2])) \end{sagecommandline} \end{section} \begin{section}{Cohomology} The \stt{BPBar} supports a method \stt{matrix\_differential} that computes the matrix of the bar differential $\delta:\Gamma^{\otimes s}\rightarrow\Gamma^{\otimes (s+1)}$, i.~e.~ the alternating sum of the $\Delta_k$. \begin{sageexample} sage: BPBar(3).matrix_differential(2,8) \end{sageexample} Here the matrix is computed with respect to the following bases: \begin{sageexample} sage: list(BPBar(3).bar_basis(2,8)) sage: list(BPBar(3).bar_basis(3,8)) \end{sageexample} The method \stt{Ext\_smith} goes one step further and computes the Smith normal form of the differential; for example, in bidegree (2,16) we find one diagonal coefficient that is divisible by $p$: \begin{sagecommandline} sage: d,u,v = BPBar(3).matrix_differential(2,16).smith_form(integral=True) sage: d.diagonal() \end{sagecommandline} This coefficient corresponds to an $x\in \Gamma^{\otimes 2}$ such that $\delta(x)\equiv 0\mod 6$; we therefore find a non-zero cocycle $y=\frac16\delta(x)$. The \stt{Ext\_smith} routine uses the transformation matrix \stt{u} to compute this cocycle: \begin{sagecommandline} sage: L = list(BPBar(3).Ext_smith(3,16,withcocycle=True)) sage: # L is a list of tuples (cf,x,y) with delta(x) = cf*y sage: (c,x,y), = L \end{sagecommandline} Then $$c=\sage{c},\quad x=\sage{x}$$ and $$y=\sage{y}.$$ We conclude with some tables that illustrate the timing of such Ext computations and the size/growth of the bar resolution. \end{section} \begin{sagesilent} def normpower(p,n): Zp = IntegerModRing(p) ans = 1 while Zp(n) == 0: ans *= p n /= p return ans def cachedcomputation(c): import pickle def cf(X,s,d): p=X.prime fname = "bpbar.cache.%d.%d.%d" % (p,s,d) if not os.path.exists(fname): with open(fname, 'wb') as f: pickle.dump(c(X,s,d),f) print("stored cached data in %s" % fname) with open(fname, 'rb') as f: print("loaded cached data from %s" % fname) L = pickle.load(f) return L return cf @cachedcomputation def extcomp(X,s,d): p = X.prime g=globals() cmd = "globals()['L']=list(BPBar(%d).Ext_smith(%d,%d))" % (X.prime,s,d) if True: secs = timeit(cmd,seconds=True,number=1) L = g['L'] else: secs = 3.62123232 L = [(p,1,1),(7*p*p,1,1)] return [ secs, "+".join(str(normpower(p,c)) for (c,x,y) in L)] figcode="""\\begin{figure}[ht] \\begin{adjustbox}{addcode={\\begin{minipage}{0.8\\width}}{\\caption{%% %s }\\end{minipage}},rotate=270,center} \\begin{tikzpicture}[scale=1.4] %s\\end{tikzpicture} \\end{adjustbox} \\end{figure}""" def getfig(X,smax,dmax): p = X.prime caption="Novikov Ext for the prime $%d$" % p dims = [d for d in range(1,smax+dmax+1) if d % (2*(p-1)) == 0 ] fcode = "\\draw (0,1) grid (%d,%d);\n" % (len(dims),smax+1) for s in range(1,smax+1): fcode += "\\node[anchor=east] at (-0.2,%f) {\\small $%d$};\n" % (s+0.5,s) for (i,d) in enumerate(dims): fcode += "\\node[anchor=center] at (%f,0.8) {\\small $%d$};\n" % (i+0.5,d) for s in range(1,smax+1): skip = False for (i,d) in enumerate(dims): if skip: secs, bas = "skipped", "?" else: secs, bas = extcomp(X,s,d) print("Computed (%d,%d) for p=%d in %s" % (s,d,p,secs)) if secs>2000: skip=True secs = "%.1fs" % secs fcode += "\\node[anchor=center] at (%f,%f) {%s};\n" % (i+0.5,s+0.7,"\\small %s" % bas) fcode += "\\node[anchor=center] at (%f,%f) {%s};\n" % (i+0.5,s+0.2,"\\scriptsize %s" % secs) return figcode % (caption,fcode) \end{sagesilent} \sagestr{getfig(BPBar(2),6,18)} \sagestr{getfig(BPBar(3),6,40)} \sagestr{getfig(BPBar(5),6,100)} \begin{sagesilent} def dimtable(X,maxs,ndims): p = X.prime dims = list(2*(p-1)*_ for _ in range(1,ndims)) tab = "\\begin{figure}\\caption{Rank of $\\BP\\us\\BP^{\\land s}$ for the prime $"+str(X.prime)+"$}\n" tab += "\\medskip\\begin{center}\\begin{tabular}{c|r" + "r"*maxs + "}\n" tab += "dim&" + "&".join(str(_) for _ in range(maxs)) + "\\\\\n" tab += "\\hline\n" tab += "\\\\\n".join("%d&"%d + "&".join(str(len(X.bar_basis(sdeg,d))) for sdeg in range(maxs)) for d in dims) tab += "\n\\end{tabular}\\end{center}" tab += "\n\\end{figure}" return tab \end{sagesilent} \sagestr{dimtable(BPBar(2),7,30)} \sagestr{dimtable(BPBar(3),7,30)} \sagestr{dimtable(BPBar(5),7,30)} \end{document} chapter/03-sommerfeld.tex0 \chapter{Modèle des électrons libres} Le modèle des électrons libres a été proposé par en 1920, pour décrire le comportement des électrons dans un solide comme un gaz. On peut comprendre beaucoup des propritétés physiques des métaux, et pas seulement des métaux simples, avec un modèle d'électrons libres. En suivant ce modèle, les électrons de valence des atomes constituant le métal deviennent des électrons de conduction est bougent librement dans le volume du métal. Même dans les métaux pour lesquels le modèle des électrons libre marcherait le mieux, la distribution de charge des électrons de conduction reflète le potentiel électrostatique important des cœurs ioniques. L'utilité d'un modèle d'électrons libres est importante pour comprendre les propriétés qui dépendent essentiellement des propriétés cinétiques des électrons de conduction. L'interaction des électrons de conduction avec les ions du réseau est traitée dans le chapitre suivant sur le modèle des électrons presque libres. Les métaux les plus siples sont les métaux alcalins : le lithium, sodium, potassium, césium et rubidium. Dans un atome libre de sodium, l'électron de valence est dans un état 3s ; dans un métal, cet électron devient un électron de conduction dans la bande de conduction 3s. Un cristal monovalent qui contient N atomes aura N électrons de conduction et N cœurs ioniques positifs. Le cœur ionique \ch{Na+} contient 10 électrons qui occupent les couches 1s, 2s et 2p de l'ion libre, avec une distribution spatialle qui est essentielleement la même qu'il soit métallique ou sous forme d'ion libre. Les cœurs ioniques remplissent à peu près 15\% du volume dans un cristal de sodium. Le rayon de l'ion libre \ch{Na+} est de \SI{0.98}{\angstrom}, alors que la moitié de la distance aux plus proches voisins dans le métal est de \SI{1.83}{\angstrom}. L'interprétation des propriétés métalliques en terme de mouvement des électrons libres a été développée bien avant l'invention de la mécanique quantique. La théorie classique a eu quelques succès, notamment en démontrant une forme de la loi d'Ohm et la relation entre la conductivité électrique et thermique, que nous aborderons plus tard avec le modèle de Drude pour la conduction électronique. En revanche, la théorie classique a échoué à expliquer la capacité calorifique et la susceptibilité magnétique des élecrtons de conduction. En fait, ce ne sont pas des problèmes sur le modèle de l'électron libre, mais des modèles sur la distribution classique de Maxwell-Boltzman qui y est utilisée. Il y a une difficulté supplémentaire avec le modèle classique. Plusieurs expériences de plusieurs types donnent le résultat suivant : il est clair que les électrons de conduction dans un métal peuvent se déplacer avec un chemin direct sur plusieurs distances atomiques, sans être défvés par des collisions avec d'autres électrons de conduction ou par les collisions avec les noyaux atomiques. Dans un échantillon très pur à basse température, le libre parcours moyen est aussi long que \SI{e8}{} distances inter-atomiques (plus d'un centimètre !). Pourquoi est la matière condensée si transparente à la conduction des électrons ? La réponse à cette question peut être scindée en deux parties : premièrement, un électron de conduction n'est pas dévié par les cœurs ioniques arrangés sur un réseau \emph{périodique} parce que les ondes mécaniques peuvent se propager librement sur une structure périodique, comme conséquence des mathématiques traités dans ce chapitre. Deuxièmement, un électron de conduction est diffusé seulement rarement par d'autres électrons de conduction. Cette proriété est une conséquence du principe d'exclusion de Pauli. Quand on parle de gaz d'électrons libres de Fermi, on doit plutôt vouloir dire un gaz d'électrons libres sujets au principe de Pauli. \section{Effets du principe de Pauli} Considérons un électron libre dans un gaz 1D, en prenant en compte la théorie quantique et le principe de Pauli. Un électron qui a une masse m est confiné dans un puit de largeur L. La fonction d'onde $\psi_n(x)$ de l'électron est une solution de l'équation de Schrödinger $\mathcal{H}\psi = \epsilon \psi$, ce qui, en négligeant l'énergie potentielle, donne $\mathcal{H} = p^2/2m$, où $p$ est la quantité de mouvement. Dans la théorie quantique, $p$ peut représenter l'opérateur $-i\hbar d/dx$, de telle sorte que : \begin{equation} \mathcal{H}\psi_n = -\frac{\hbar^2}{2m} \frac{d^2\psi_n}{dx^2} = \epsilon_n\psi_n \end{equation} où $\epsilon_n$ est l'énergie de l'électron sur son orbitale. On utilise le terme orbitale pour dire une solution de l'équation d'onde pour un système d'un seul électron. Le terme permet de distinguer entre un état quantique exact de l'équation d'onde d'un système de N électrons interagissant et un état quantique approximé que l'on construit en associant aux N électrons N déffirentes orbitales, chacune d'entre elles étant une solution de l'équation d'onde pour un électron. Le modèle orbitalaire est exact s'il n'y a aucune interaction entre chacun des électrons. Les conditions aux limites sont : $\psi_n (0) = 0$, $\psi_n(L)=0$, comme imposé par le puit de potentiel infini. Ces conditions sont satisfaites si la fonction d'onde est sinusoidale, avec un nombre entier de demi-longeurs d'ondes entre 0 et L : \begin{equation} \psi_n = A \sin\left( \frac{2\pi}{\lambda_n} \right) ; \hfill \frac{1}{2}n\lambda_n = L \end{equation} où A est constante. On voit que c'est une solution de l'équation de Schrödinger, parce que : \begin{equation} \frac{d\psi_n}{dx} = A \left( \frac{n\pi}{L} \right) \cos\left( \frac{n\pi}{L}x \right) ; \hfill \frac{d^2\psi_n}{dx^2} = - A \left( \frac{n\pi}{L} \right)^2\sin\left( \frac{n\pi}{L}x \right) \end{equation} où l'énergie $\epsilon_n$ est donnée par : \begin{equation} \epsilon_n = \frac{\hbar^2}{2m} \left( \frac{n\pi}{L}\right)^2 \label{en} \end{equation} On veut pouvoir loger N électrons sur la ligne. Comme prévu par le principe d'exclusion de Pauli, deux électrons ne peuvent pas avoir le même nombre quantique. Ansi, chaque orbitale ne peut être occupée que par deux électrons au maximum. Ceci s'applique aux électrons dans les atomes, molécules, et solides. Dans un solide linéaire, les nombres quantiques d'une orbitale d'un électron de conduction sont $n$ et $m_s$, où $n$ est un nombre entier positif, et le nombre quantique magnétique $m_s = \pm \frac{1}{2}$, suivant l'orientation de spin. Une paire d'orbitales marquée par un nombre quantique $n$ peut recevoir deux électrons, un avec un spin vers le haut et un avec un spin vers le bas. S'il y a six électrons, alors l'état fondamental du système est rempli d'orbitales données par le tableau suivant : \begin{table}[ht] \begin{center} \begin{tabular}{rrr} \toprule n & $m_s$ & occupation par les électrons\\ \midrule 1 & $\uparrow$ & 1\\ 1 & $\downarrow$ & 1\\ 2 & $\uparrow$ & 1\\ 2 & $\downarrow$ & 1\\ 3 & $\uparrow$ & 1\\ 3 & $\downarrow$ & 1\\ 4 & $\uparrow$ & 0\\ 4 & $\downarrow$ & 0\\ \bottomrule \end{tabular} \end{center} \label{} \caption{remplissage des orbitales pour un système à 6 électrons} \end{table} Plus d'une orbitale peut voir la même énergie. Le nombre d'orbitales avec la même énergie est appelée \emph{dégénérescence}. Soit $n_F$ le niveau d'énergie rempli le plus élevé, où on commence à remplir les niveaux depuis le bas (n=1), et on continue à remplir les niveaux d'énergie supérieure, jusqu'à ce que tous les N électrons soient placés. Il est pratique de supposer que N est un nombre pair. la condition $2n_F = N$ détermine $n_F$, la valeur de n pour laquelle le niveau le plus élevé est rempli. L'énergie de Fermi $\epsilon_F$ est définie comme l'énergie du plus haut niveau rempli à l'état fondamental d'un système à N électrons. En utilisant la relation \ref{en}, avec $n=n_F$, on a, à une dimension : \begin{equation} \epsilon_F = \frac{\hbar^2}{2m}\left(\frac{n_F\pi}{L}\right)^2 = \frac{\hbar^2}{2m}\left( \frac{N\pi}{2L} \right)^2 \end{equation} \section{Statistique de Fermi-Dirac} L'état fondamental est l'état d'un système de N électrons au zéro absolu. Ce qui arrive quand on augmente la température ? C'est un problème standard dans la mécanique statistique élémentaire, et la solution es donnée par la distribution de Fermi-Dirac. \begin{figure} \TODO \caption{La distribution de Fermi Dirac} \label{fermidir} \end{figure} L'énergie cinétique du gas d'électrons augmente quand la température augmente : certains niveaux d'énergie sont occupés, d'autres restent vacants au zéro absolu, et certains niveuax sont vides alors qu'ils sont occupés au zéro absolu. La distribution de Fermi-Dirac donne la probabilité qu'une orbitale d'énergie $\epsilon$ soit occupée par un gas d'électrons idéal à l'équilibre thermique. \begin{equation} f(\epsilon) = \frac{1}{\exp\left( \frac{\epsilon - \mu}{k_BT} \right) + 1} \end{equation} La quantité $\mu$ est une fonction de la température. $\mu$ doit être choisie pour un problème particulier d'une telle façon que le nombre total de particules dans e système en dérive correctement, c'est à dire qu'on retrouve bien N à la fin. Au zéro absolu, $\mu = \epsilon_F$, parce que la limite $T\rightarrow 0$, la fonction $f(\epsilon)$ change de façon discontinue de la valeur 1 (rempli) à la valeur zéro (vide), à $\epsilon = \epsilon_F = \mu$. Pour toute température, $f(\epsilon)$ est égal à 1/2 lorsque $\epsilon = \mu$. $\mu$ est le potentiel chimique, et on doit voir qu'au zéro absolu, le potentiel chimique est égal à l'énergie de Fermi, définie comme l'énergie de la plus haute orbitale remplie au zéro absolu. La partie en queue de distribution, est cette parie pour laquelle $\epsilon - \mu \ll k_BT$. Ici, le terme exponentiel est dominant dans le dénominateur, et ainsi $f(\epsilon) \cong \exp(\frac{\mu - \epsilon}{k_BT})$. Cette limite est appelée la distribution de Maxwell-Boltzmann. \section{Généralisation à trois dimensions} Les électrons sont des particules quantiques. Ils peuvent donc être décrits par une fonction d'onde $\psi(\mathbf{r},t)$, qui est la solution de l'équation de Schödinger suivante : \section{conditions de } \section{Sphère de Fermi et espace des états} Dans l'état fondamental d'un système à N électrons libres, les orbitales occupées peuvent représenter des points dans une sphère de l'espace réciproque. L'énergie à la surface de la sphère est l'énergie de Fermi. Les vecteurs d'onde à la surface de Fermi ont une intensité $k_F$ telle que : \begin{equation} \epsilon_F = \frac{\hbar^2}{2m}k_F^2 \label{eef} \end{equation} On peut alors voir qu'il n'y a qu'un seul vecteur d'onde permis, qui est distinct du triplet de nombre quantiques $k_x, k_y, k_z$, pour l'élément de volume $(2\pi/L)^3$ de l'espace réciproque. Ainsi, dans la sphère de volume $\frac{4\pi k_F^3}{3}$, le nombre total d'orbitales est : \begin{equation} 2\cdot \frac{\frac{4\pi k_F^3}{3}}{\left( \frac{2\pi}{L}\right)^3} = \frac{V}{3\pi^2} k_F^3 = N \end{equation} où le facteur 2 à gauche vient des deux valeurs permises par le nombre quantique de spin pour chaque valeur de $\mathbf{k}$. Ainsi, on obtient : \begin{equation} k_F = \left( \frac{3\pi^2N}{V}\right)^{1/3} \label{kkf} \end{equation} qui ne dépend que de la concentration en particules. \begin{marginfigure} \TODO \caption{Sphère de Fermi} \label{spherefermi} \end{marginfigure} En combinant \ref{eef} et \ref{kkf}, on obtient : \begin{equation} \epsilon_F = \frac{\hbar^2}{2m} \left(\frac{3\pi^2N}{V}\right)^{2/3} \label{ef} \end{equation} Ceci se rapporte à l'énergie de Fermi pour la concentation en électrons N/V. La vitesse des électrons $v_F$ sur la surface de Fermi est : \begin{equation} v_F = \left( \frac{\hbar k_F}{m}\right) = \left( \frac{\hbar}{m} \right)\left(\frac{3\pi^2 N}{V} \right)^{1/3} \end{equation} Les valeurs calculées de $k_F$, $v_F$ et $\epsilon_F$ sont données dans le tableau suivant pour certains métaux. Le paramètre de rayon est un nombre sans dimension $r_n = r_0/a_H$ où $a_H$ est le rayon de Bohr et $r_0$ le rayon de la sphère qui contient un électron. \begin{table*}[ht] \footnotesize \begin{center} \begin{tabularx}{\textwidth}{rlRRRRRR} \toprule Valence & Métal & concentration en électrons ($10^{22} cm^{-3}$) & Paramètre de rayon $r_n$ & vecteur d'onde de Fermi ($10^8 cm^{-1}$) & vitesse de Fermi ($10^8 cm\cdot s^{-1}$) & énergie de Fermi (eV) & température de Fermi ($T_F =\epsilon_F / k_B$) ($10^4$ K)\\ \midrule 1 & Li & 4.70 & 3.25 & 1.11 & 1.29 & 4.72 & 5.48 \\ & Na (78K) & 2.65 & 3.93 & 0.92 & 1.07 & 3.23 & 3.75\\ & K (78K) & 1.40 & 4.86 & 0.75 & 0.86 & 2.12 & 2.46\\ & Rb (78K) & 1.15 & 5.20 & 0.70 & 0.81 & 1.85 & 2.15\\ & Cs (78K) & 0.91 & 5.63 & 0.64 & 0.75 & 1.58 & 1.83\\ & Cu & 8.45 & 2.67 & 1.36 & 1.57 & 7.00 & 8.12\\ & Ag & 5.85 & 3.02 & 1.20 & 1.39 & 5.48 & 6.36\\ & Au & 5.90 & 3.01 & 1.20 & 1.39 & 5.51 & 6.39\\ 2 & Be & 24.02 & 1.88 & 1.93 & 2.23 & 14.14 & 16.41\\ & Mg & 8.60 & 2.65 & 1.37 & 1.58 & 7.13 & 8.27\\ & Ca & 4.60 & 3.27 & 1.11 & 1.28 & 4.68 & 5.43\\ & Sr & 3.56 & 3.56 & 1.02 & 1.18 & 3.95 & 4.58\\ & Ba & 3.20 & 3.69 & 0.98 & 1.13 & 3.65 & 4.24\\ & Zn & 13.10 & 2.31 & 1.57 & 1.82 & 9.39 & 10.90\\ & Cd & 9.28 & 2.52 & 1.40 & 1.62 & 7.46 & 8.66\\ 3 & Al & 18.06 & 2.07 & 1.75 & 2.02 & 11.63 & 13.49\\ & Ga & 15.30 & 2.19 & 1.65 & 1.91 & 10.35 & 12.01\\ & In & 22.49 & 2.41 & 1.50 & 1.74 & 8.60 & 9.98\\ 4 & Pb & 13.20 & 2.30 & 1.57 & 1.82 & 9.37 & 10.87\\ & Sn(w) & 14.48 & 2.23 & 1.62 & 1.88 & 10.03 & 11.64\\ \bottomrule \end{tabularx} \end{center} \caption{Paramètres de la surface de Fermi pour le modèle de l'élecrton libre pour quelques métaux à température ambiante} \label{} \end{table*} Maintenant, on trouve une expression du nombre d'orbitales par gamme d'énergie unitaire., $D(\epsilon)$, appelée densité d'états\footnote{si on est rigoureux, $D(\epsilon)$ représente plutôt la densité d'états à une particules, ou la densité des orbitales}. On utilise \ref{ef} pour obtenir le nombre total d'orbitales d'énergie inférieure ou égale à $\epsilon$ : \begin{equation} N = \frac{V}{3\pi^2} \left(\frac{2m\epsilon}{\hbar^2}\right)^{3/2} \end{equation} Ainsi, la densité d'états est : \begin{equation} D(\epsilon) \equiv \frac{dN}{d\epsilon} = \frac{V}{2\pi^2} \cdot \left(\frac{2m}{\hbar^2}\right)^{3/2}\cdot\epsilon^{1/2} \end{equation} Ce résultat peut être exprimé plus simplement en comparant les deux formules, pour obtenir, à $\epsilon$ : \begin{equation} D(\epsilon) \equiv \frac{dN}{d\epsilon} = \frac{3N}{2\epsilon} \end{equation} Avec un facteur de l'ordre de l'unité, le nombre d'orbitales par unité d'énergie à l'energie de Fermi est le nombre total d'électrons de conduction divisé par l'énergie de Fermi, exactement ce qu'on espérait. \section{Énergie et chaleur spécifique du gaz d'électrons libres} \section{Densité d'états électroniques} \section{conductivité thermique} Constante de Wiedemann-Franz \section{Effet Pelletier} \begin{center} \textbf{ABSTRACT} \end{center} $\!$\\ % O resumo em inglês deve ser organizado em apenas um parágrafo mesmo. This work aims to create a Python library for two-dimensional laminar particle flow simulations that is developed using object oriented programming and analyze the results of cases applied to turbomachines. The Finite Element Method was used to solve the system of equations derived from the stream function-vorticity formulation of the Navier-Stokes equation for a incompressible and newtonian fluid. For the particle behavior modeling, the Finite Difference Method was applied to solve the Basset–Boussinesq–Oseen equation, considering one-way interactions, without taking to account the effect of the particles in the flow. \vspace{1cm} \hspace{-1.3cm}Keywords: Finite Element Method, Stream Function-Vorticity, Multiphase Flow, Particle Flow.\begin{align*} \tag{Entrada del sistema} u(t) & = e(t) \\ \tag{Sailda del sistema} y(t) & = i_1 \end{align*} Se utilizan las mismas ecuaciones de mallas y relaciones del problema 1.1. Se despeja $i_2$ de Malla 1 y se sustituye en Malla 2, igualmente se sustituyen las relaciones, y se resuelve para la RES: \begin{align*} e(t) = R_1 (i_1 + i_2) + v_L \Rightarrow & i_2 = \frac{u(t) - R_1 y(t) + v_L}{R_1}\\ v_L = v_c + R_2 i_2 \Rightarrow & v_L = \frac{1}{C}\int_0^t i_2 {dt} + R_2 \frac{u(t) - R_1 y(t) + L\frac{di}{dt}}{R_1} \\ & v_L = \frac{1}{C}\int_0^t \left( \frac{u(t) - R_1 y(t) + v_L}{R_1} \right) {dt} + R_2 \frac{u(t) - R_1 y(t) + L\frac{dy(t)}{dt}}{R_1} \\ & L\frac{dy(t)}{dt} = \frac{1}{R_1C}\int_0^t \left( u(t) - R_1 y(t) + L\frac{dy(t)}{dt} \right) {dt} + \frac{R_2}{R_1} \left(u(t) - R_1 y(t) + L\frac{dy(t)}{dt}\right) \\ & L\dot{y}(t) = \frac{1}{R_1C}\int_0^t \left( u(t) - R_1 y(t) \right) {dt} + \frac{L}{R_1C}y(t) + \frac{R_2}{R_1} \left(u(t) - R_1 y(t) + L\dot{y}(t)\right) \\ \end{align*} \begin{align*} S(\overline{u}, \overline{y}) &= \frac{1}{R_1C} \int_0^t u(t) {dt} - \frac{1}{C} \int_0^t y(t) dt + \left( \frac{L}{R_1C} - R_2 \right) y(t) + \left( \frac{R_2}{R_1} - 1 \right) L\dot{y}(t) = 0 \\ \tag{RES} S(\overline{u}, \overline{y}) &= \frac{1}{R_1C} u(t) - \frac{1}{C} y(t) + \left( \frac{L}{R_1C} - R_2 \right) \dot{y}(t) + \left( \frac{R_2}{R_1} - 1 \right) L\ddot{y}(t) = 0 \end{align*} Se aplica la transformada de Laplace y se resuleve: \begin{align*} \frac{1}{R_1C} U(s) - \frac{1}{C} Y(s) + \left( \frac{L}{R_1C} - R_2 \right) s Y(s) + \left( \frac{R_2}{R_1} - 1 \right) L s^2 Y(s) = 0 \\ \frac{1}{R_1C} U(s) + \left[ \left( \frac{L}{R_1C} - R_2 \right) s - \frac{1}{C} + \left( \frac{R_2}{R_1} - 1 \right) L s^2 \right] Y(s) = 0 \\ \frac{1}{R_1C} U(s) = \left[ \frac{1}{C} - \left( \frac{L}{R_1C} - R_2 \right) s - \left( \frac{R_2}{R_1} - 1 \right) L s^2 \right] Y(s) \\ \end{align*} \begin{align*} \Delta &= \left( \frac{L}{R_1C} - R_2 \right)^2 + 4 \left( \frac{R_2}{R_1} - 1 \right) L \frac{1}{C} \\ \Delta &= \left( \frac{L}{R_1C} \right)^2 - 2 \frac{R_2L}{R_1C} + R_2^2 + \frac{4R_2L}{R_1C} - \frac{4L}{C} \\ \Delta &= \left( \frac{L}{R_1C} \right)^2 + 2 \frac{R_2L}{R_1C} + R_2^2 - \frac{4L}{C} \\ \Delta &= \left( \frac{L}{R_1C} + R_2 \right)^2 - \frac{4L}{C} \end{align*} \begin{align*} s &= \frac{\left( \frac{L}{R_1C} - R_2 \right) \pm \sqrt{\left( \frac{L}{R_1C} + R_2 \right)^2 - \frac{4L}{C}}}{-2 \left( \frac{R_2}{R_1} - 1 \right) L } \\ s_0 &= \frac{\left( R_2 - \frac{L}{R_1C} \right) - \sqrt{\left( \frac{L}{R_1C} + R_2 \right)^2 - \frac{4L}{C}}}{2 L \left( \frac{R_2}{R_1} - 1 \right) } \\ s_1 &= \frac{\left( R_2 - \frac{L}{R_1C} \right) + \sqrt{\left( \frac{L}{R_1C} + R_2 \right)^2 - \frac{4L}{C}}}{2 L \left( \frac{R_2}{R_1} - 1 \right) } \end{align*} \begin{align*} R_1C (s-s_0)(s-s_1) Y(s) = U(s) \\ \frac{Y(s)}{U(s)} = \frac{1}{R_1C} \frac{1}{(s-s_0)(s-s_1)} = \frac{A}{s-s_0} + \frac{B}{s-s_1} \\ \frac{1}{R_1C} = A(s-s_1) + B(s-s_0) \\ \text{Cuando } s = s_0 \Rightarrow \frac{1}{R_1C} = A(s_0-s_1) \Rightarrow A = \frac{1}{R_1C} \frac{1}{s_0-s_1} \\ \text{Cuando } s = s_1 \Rightarrow \frac{1}{R_1C} = B(s_1-s_0) \Rightarrow B = \frac{1}{R_1C} \frac{1}{s_1-s_0} \\ \frac{Y(s)}{U(s)} = \frac{1}{R_1C} \frac{1}{(s_0-s_1)(s-s_0)} + \frac{1}{R_1C} \frac{1}{(s_1-s_0)(s-s_1)} \\ \end{align*} Se aplica la transformada inversa de Laplace para obtener la RESE: \begin{align*} \frac{y(t)}{u(t)} = \frac{1}{R_1C} \frac{1}{(s_0-s_1)}e^{s_0t} + \frac{1}{R_1C} \frac{1}{(s_1-s_0)}e^{s_1t} \\ \tag{RESE} y(t) = \frac{1}{R_1C} u(t) \left[\frac{1}{(s_0-s_1)}e^{s_0t} + \frac{1}{(s_1-s_0)}e^{s_1t}\right] \end{align*} %File: formatting-instruction.tex \documentclass[letterpaper]{article} \usepackage{aaai} \usepackage{times} \usepackage{helvet} \usepackage{courier} \usepackage{latexsym} \usepackage{amssymb} \usepackage{amsmath} \usepackage{graphics} \usepackage{natbib} \bibliographystyle{named} \newtheorem{definition}{Definition}[section] \newtheorem{lemma}[definition]{Lemma} \begin{document} % The file aaai.sty is the style file for AAAI Press % proceedings, working notes, and technical reports. % \title{Planning Sudoku for SAT} \author{\\ \\ Center for Computational Logic (ICCL), Technische Universitet Dresden } \maketitle \begin{abstract} Sudoku is a very simple and well-known puzzle. Many encoding schemes such as Minimal, Efficient, Extended, and Pre-Processed are known for this puzzle. Any Sudoku puzzle is formulated as a CNF formula $\phi$ which is satisfiable if the Sudoku has a solution. We restrict our attention to consider only the instance of Sudoku which has only one solution and can only be solved using reasoning (i.e.with no search). Pre-Processed encoding suffices for this characterization, where other encoding schemes adds redundant clauses to minimal satisfying formulae; We observe that updating Pre-Processed encoding regarding pre-assigned cells configurations facilitate a higher degree of pre-processing and unit propagation can reduce the number of final clauses generated to be checked for a model. \end{abstract} \section{Introduction} In recent years, there has been a considerable renewed interest in the satisfiability problem (SAT) and many efficient SAT solvers are available. Traditionally planning has been formalized as deduction~\citet{McCarthy1969}, but we'd formulated our model on satisfiability rather than on deduction. This approach not only provides a more flexible framework for stating different kinds of constraints on plans but also more accurately reflects the theory behind modern constraint-based planning systems~\citet{Henry2006}. \section{Preliminaries} \subsection{Sudoku} Standard Sudoku puzzle usually appear in Newspapers is a $9\ast9$ grid instance of a very easy problem; \begin{definition} A Sudoku puzzle is represented by a $\mathbb{N}\ast\mathbb{N}$ grid, which comprises of an $\sqrt{\mathbb{N}} \ast \sqrt{\mathbb{N}}$ sub-grids (also called boxes). Some of the entries in the grid are filled with numbers from $1$ to $\mathbb{N}$, whereas other entries are left blank~\citet{Lynce2006};\end{definition} Puzzles are often assigned a difficulty level, which usually depends on the number of initial non-blank entries provided. This number may be as few as $17$ to test expert players for $9\ast9$ instance~\citet{Lynce2006}. But it could also be shown that with an automated planning system it can be reduced further to only one assignment and the planner still able to solve the problem correctly. \begin{definition} A Sudoku puzzle is solved by assigning numbers from $1$ to $\mathbb{N}$ (size of puzzle) to the blank entries such that every row, every column, and every $\sqrt{\mathbb{N}} \ast \sqrt{\mathbb{N}}$ block (sub-grid) contains each of the any $\mathbb{N}$ possible numbers;\end{definition} Propositional logic formalism is used to encode Sudoku puzzle in our encoding and other well known encoding schemes~\citet{Weber2005}~\citet{Lynce2006}~\citet{Gihwon2006}. \section{Planning as Satisfiability} The complementary problem to deducibility is satisfiability that is finding a model for set of axioms. The formalism of planning as satisfiability turns out to have a number of attractive properties like~\citet{Stefik1981}~\citet{chapman1987}: \begin{itemize} \item It is easy to state arbitrary facts about any state of the world not just the initial and goal states; \item It is likewise easy to state arbitrary constraints on the plan e.g. spcifying an action performed at a any specifal time interval; \item Finally the approach provides a more accurate formal model of modern constraint based planners. \end{itemize} \subsection{Sudoku Axiomatization} Many Sudoku solvers are already available~\citet{Pete2005}~\citet{DeadMan2005} and since there are more than $6.10^{21}$ possible grids for a simple looking $9\ast9$ puzzle instance~\citet{Bertram2005}; hence, a naive backtracking algorithm would be infeasible and the only feasible approach in this scenario is to combine backtracking with some form of constraint propagation. But we utilize a SAT-based approach in which a puzzle is translated into propositional formula $\Phi$ then we use any standard SAT solver to obtain the model (satisfying assignment) of $\Phi$; this model can then be readily translated into the solution of puzzle. \subsubsection{Sudoku Encoding} Since puzzles are regarded as a propositional SAT problem, we'll provide an overview of various existing encoding schemes. A SAT problem is represented as a propositional formula $\Phi$ where each variable $P_i$ is assigned $0$ ($\mathbb{F}$) or $1$ ($\mathbb{T}$) where i $\in$ $(1,\cdot\cdot\cdot,n)$. % ($|n|$ is the length of formula). In an $\mathbb{N}\ast\mathbb{N}$ puzzle each cell can contains a number from $1$ to $n$. Each tuple $(r,c,v)$ denotes a variable which is true iff the cell in row $r$ and column $c$ is assigned a number $v$; $[r, c] = v$. The resulting set of formulas turn out to be $V$ = \{$(r,c,v)$ $|$ 1 $\leq$ $r,c,v$ $\leq$ n\}. Exactly one number for each cell, row, column and block can be viewed as a \textit{definedness} and \textit{Uniquness constraints}. Following is the formulae: \begin{itemize} \item There is at exactly one number in each cell\\ $\phi_{cell.ex}$ := $\phi_{cell.def}$ $\wedge$ $\phi_{cell.uniq}$\\ There is at least one number for each cell \\ $\phi_{cell.def}$ := $\bigwedge_{r=1}^n$ $\bigwedge_{c=1}^n$ $\bigvee_{v=1}^n$ $(r,c,v)$\\ Each number appears at most one in each cell\\ $\phi_{cell.uniq}$ := $\bigwedge_{r=1}^n$ $\bigwedge_{c=1}^n$ $\bigwedge_{v_{i}=1}^{(n-1)}$ $\bigwedge_{v_{j}=v_{i} + 1}^n$$\neg(r,c,v_i)$ $\vee$ $\neg(r,c,v_j)$ \item There is at exactly one number in each row\\ $\phi_{row.ex}$ := $\phi_{row.def}$ $\wedge$ $\phi_{row.uniq}$\\ There is at least one number for each row \\ $\phi_{row.def}$ := $\bigwedge_{r=1}^n$ $\bigwedge_{v=1}^n$ $\bigvee_{c=1}^n$ $(r,c,v)$\\ Each number appears at most one in each row\\ $\phi_{row.uniq}$ := $\bigwedge_{r=1}^n$ $\bigwedge_{v=1}^n$ $\bigwedge_{c_{i}=1}^{(n-1)}$ $\bigwedge_{c_{j}=c_{i} + 1}^n$$\neg(r,c_i,v)$ $\vee$ $\neg(r,c_j,v)$ \item There is at exactly one number in each column\\ $\phi_{col.ex}$ := $\phi_{col.def}$ $\wedge$ $\phi_{col.uniq}$\\ There is at least one number for each column \\ $\phi_{col.def}$ = $\bigwedge_{c=1}^n$ $\bigwedge_{v=1}^n$ $\bigvee_{r=1}^n$ $(r,c,v)$\\ Each number appears at most one in each column\\ $\phi_{col.uniq}$ := $\bigwedge_{c=1}^n$ $\bigwedge_{v=1}^n$ $\bigwedge_{r_{i}=1}^{(n-1)}$ $\bigwedge_{r_{j}=r_{i} + 1}^n$$\neg(r_i,c,v)$ $\vee$ $\neg(r_j,c,v)$ \item There is at exactly one number in each block\\ $\phi_{blk.ex}$ = $\phi_{blk.def}$ $\wedge$ $\phi_{blk.uniq}$\\ There is at least one number for each block \\ $\phi_{blk.def}$ = $\bigwedge_{rof=1}^{m}$ $\bigwedge_{cof=1}^{m}$ $\bigwedge_{v=1}^n$ $\bigvee_{r=1}^{m}$ $\bigvee_{c=1}^{m}$ $(rof \ast m + r,cof \ast m + c,v)$ where m := $\sqrt{n}$;\\ Each number appears at most one in each block\\ $\phi_{blk.uniq}$ = $\bigwedge_{rof=1}^{m}$ $\bigwedge_{cof=1}^{m}$ $\bigwedge_{v=1}^n$ $\bigwedge_{r=1}^{n}$ $\bigwedge_{c = r +1}^{n}$ \\ $\neg(rof \ast m + (r$ $mod$ $m),cof \ast m + (r$ mod $m),v)$ $\vee$ \\ $\neg(rof \ast m + (c$ mod $m),cof \ast m + (c$ mod $m),v)$ where m := $\sqrt{n}$; \end{itemize} \noindent The fixed cells are naturally represented as unit clauses in encoding of puzzle: $V^{+}$ = $\{ (r,c,v) \in V | [r,c]$ is a fixed cell and has a pre-assigned value $v$\}. $\phi_{assign}$ = $\bigwedge_{i=1}^k$ $(r,c,v)$ where $(r,c,v)$ $\in$ $V^{+}$ \noindent Now encoding is nothing more than the following set of clauses: \begin{itemize} \item $\Phi_{extended}$ := $\phi_{cell.ex}$ $\wedge$ $\phi_{row.ex}$ $\wedge$ $\phi_{col.ex}$ $\wedge$ $\phi_{blk.ex}$ $\wedge$ $\phi_{assign}$. \item $\Phi_{efficient}$ := $\phi_{cell.ex}$ $\wedge$ $\phi_{row.uniq}$ $\wedge$ $\phi_{col.uniq}$ $\wedge$ $\phi_{blk.uniq}$ $\wedge$ $\phi_{assign}$. \item $\Phi_{minimal}$ := $\phi_{cell.def}$ $\wedge$ $\phi_{row.uniq}$ $\wedge$ $\phi_{col.uniq}$ $\wedge$ $\phi_{blk.uniq}$ $\wedge$ $\phi_{assign}$. \end{itemize} %\begin{Minimal} \end{Minimal} %\begin{Efficient} \end{Efficient} $V^{\neg}$ = $\{ (r,c,v) \in V | \exists (r',c',v') \in V^{+} ((r = r') \wedge (c = c') \wedge (v \neq v')) \vee ((r = r') \wedge (c \neq c') \wedge (v = v')) \vee ((r \neq r') \wedge (c = c') \wedge (v = v')) \vee ((r \neq r') \wedge (c = c') \wedge (v = v')) \wedge \exists_{lRow, lCol}(lRow \leq r, r' \leq lRow + m) \wedge (lCol \leq c, c' \leq lCol + m)\}$\\ where:\\ \\ lRow = $\left\{\begin{array}{ll} $1$ & \mbox{$true$};\\ ((r-1)/m)\ast (m + 1)) & \mbox{$false$}. \end{array}\right.$ \\ lCol = $\left\{\begin{array}{ll} $1$ & \mbox{$true$};\\ ((c-1)/m)\ast (m + 1)) & \mbox{$false$}. \end{array}\right.$\\ \\ $m = \sqrt{n}$.\\ \noindent Reduction operators described in pre-assigned formulae is as follows:\\ $U := \{V^{+},V^{\neg}\}$ $\phi \Downarrow U = \{ C \in \phi | \neg\exists_{L \in C}.(L = x \wedge x \in U)\}$ \\ $\phi \downarrow V^{+} = \{ \forall C \in \phi. \forall {L \in C} | \neg(L = \neg x \Rightarrow x \in V^{+})\}$\\ $\phi \downarrow V^{\neg} = \{ \forall C \in \phi. \forall {L \in C} | \neg(L = x \Rightarrow x \in V^{\neg})\}$\\ \\ \noindent Giwhown~\citet{Gihwon2006} encoding is formulated as: \begin{itemize} \item $\Phi_{pre.process}$ := ($\phi_{cell.uniq}$ $\cup$ $\phi_{row.uniq}$ $\cup$ $\phi_{col.uniq}$ $\cup$ $\phi_{blk.uniq})\Downarrow V^{\neg}$ $\cup$ \\ ($\phi_{cell.def}$ $\cup$ $\phi_{row.def}$ $\cup$ $\phi_{col.def}$ $\cup$ $\phi_{blk.def})\downarrow V^{\neg}$ $\cup$ $\phi_{assign}\Downarrow V^{+}$. \end{itemize} \begin{lemma} \label{lemma:Satisfiablity} A satisfiability equivalence relation holds between $\Phi_{extended}$ and $\Phi_{pre.process}$ with respect to a satisfying assignment $\alpha$ ,i.e. $\alpha$ satisfies $\Phi_{extended}$ iff $\alpha$ satisfies $\Phi_{pre.process}$. \end{lemma} %\begin{proof}\textbf{(\ref{th:main})} %\end{proof} \section{Proposed Encoding} \noindent The phenomenon we'd used to further reduce redundancies is explained using an example for $4\ast4$ puzzle instance. Suppose, we have some pre-assigned cells $[1,1]=4$ , $[1,2]=3$, $[1,3]=2$: \begin{itemize} \item Literals in \textit{row definedness clause} are $(1,1,1)$, $(1,2,1)$, $(1,3,1)$, $(1,4,1)$; \item Literals $(1,1,1)$ is falsified by pre-assigned $[1,1]=4$ because of same cell constraint; likewise $(1,2,1),(1,3,1)$ are falsified because of same row constraint. \item $(1,4,1)$ is the only literal left where definedness constraint imposed it to be true thus assigning $[1,4]=4$. \end{itemize} The fact which we want to reveal from the above example is that by imposing all the above constraints consequence to obtaining new assigned values for the cells which are blank before and are not even in $\phi_{assign}$. So, rendering this process to include these newly assigned cells within our $\phi_{assign}$ set and perform the deletion and removal of clauses level and literals up to saturation result in further reductions and the following formulea confers it's detail as followed: $\varphi_{new}$ := ($\phi_{cell.def}$ $\cup$ $\phi_{row.def}$ $\cup$ $\phi_{col.def}$ $\cup$ $\phi_{blk.def})\downarrow V^{\neg}$\\ $V^{*}$ := $\{ \forall C \in \varphi_{new} . |C| = 1 \wedge C \not\in \phi_{assign} \Rightarrow C \cup V^{+} \}$\\ $\Phi_{new}$ := $\varphi_{new}\Downarrow V^{\neg}$ $\cup$ ($\phi_{cell.uniq}$ $\cup$ $\phi_{row.uniq}$ $\cup$ $\phi_{col.uniq}$ $\cup$ $\phi_{blk.uniq})\Downarrow V^{\neg}$\\ \begin{lemma} \label{lemma:Satisfiablity} A satisfiability equivalence relation holds between $\Phi_{pre.process}$ and $\Phi_{new}$ with respect to a satisfying assignment $\alpha$, i.e.$\alpha$ satisfies $\Phi_{pre.process}$ iff $\alpha$ satisfies $\Phi_{new}$. \end{lemma} \section{Conclusion} Our proposed encoding not only reduces a large amount of redundant clauses but also this process of redundant clause elimination coincides with the logic of finding newly assigned values for the blank cells. Thus not only eliminating a larger set of clauses for the solver but also filling in the blank cells in a correct fashion. \bibliography{ref1} \end{document} \begin{table} \begin{spacing}{1.2} \centering \caption{Wilcoxon test results for removals inclusion and exclusion configurations of the FLT task for OpenJPA v2.3.0 (\ctwo)} \label{table:versus-wilcox-openjpa-flt-removals} \begin{tabular}{ll|rr|rr} \toprule \multicolumn{2}{c|}{Configurations} & \multicolumn{2}{c|}{MRRs} & p-value & Effect size \\ \midrule $(A,R,C,M)$ & $(A,C,M)$ & $0.3137$ & $\bm{0.3687}$ & $0.0275$ & $0.2480$ \\ $(A,R,C)$ & $(A,C)$ & $\bm{0.2989}$ & $0.2869$ & $0.2604$ & $0.1273$ \\ $(A,R,M)$ & $(A,M)$ & $0.2933$ & $\bm{0.3281}$ & $0.1028$ & $0.1794$ \\ $(A,R)$ & $(A)$ & $\bm{0.2999}$ & $0.2990$ & $0.1323$ & $0.1654$ \\ $(R,C,M)$ & $(C,M)$ & $\bm{0.2996}$ & $0.2502$ & $0.6492$ & $0.0501$ \\ $(R,C)$ & $(C)$ & $0.2314$ & $\bm{0.2809}$ & $0.2256$ & $0.1292$ \\ $(R,M)$ & $(M)$ & $0.2207$ & $\bm{0.2928}$ & $p<0.01$ & $0.3409$ \\ \bottomrule \end{tabular} \end{spacing} \end{table} \begin{figure}[h!]\centering \includegraphics[scale=0.25]{embodied_cost_model/images/system_boundary2.png} \caption[System Boundary Diagram]{System Boundary for 1-kW of Provisioned Data Center Capacity.} \label{system_boundary} \end{figure}agalbachicar/swing_for_the_fences0 \section{Results} \label{sec:results}\hypertarget{structcr__mat4}{}\doxysection{cr\+\_\+mat4 Struct Reference} \label{structcr__mat4}\index{cr\_mat4@{cr\_mat4}} Struct for a 4x4 matrix. {\ttfamily \#include $<$C\+Ray.\+h$>$} \doxysubsection*{Public Attributes} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a18de954e61bae64b09c3cb409134dbf3}{x0}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_aa448ace6709fa61ed053c2e8648188cf}{y0}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a2eb1c49e0732782f62dacc7b79446bb4}{z0}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a14cca8968d75c50d0affa3a9d3a80fc1}{w0}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a3c05dae15eb33520cd153d1a2b6244a7}{x1}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_ace7d676c9423e28d98c447c82a6d1040}{y1}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a92f3f8fe7a2a2691aef8594bd8b79011}{z1}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_af1e0a83256fd0cccc45eeef0b7e5d19c}{w1}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_aebf18437997bfae1571da0526aafb6ef}{x2}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a3bca767f170fbdced1a20f1b0daba4e6}{y2}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_ae4e714764e9d02a2c1986944f2e5791e}{z2}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_ae9103574b82172dd809be204cf795c37}{w2}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a4f21d21c1ba812a86c011924da16962b}{x3}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a27047478fc3dce659f12c795fb27ff93}{y3}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_a6ea965db3224c0e6239afdbead85c101}{z3}} \item \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} \mbox{\hyperlink{structcr__mat4_ae88ea7d36c60d35f02afde032cfd5b5d}{w3}} \end{DoxyCompactItemize} \doxysubsection{Detailed Description} Struct for a 4x4 matrix. \doxysubsection{Member Data Documentation} \mbox{\Hypertarget{structcr__mat4_a14cca8968d75c50d0affa3a9d3a80fc1}\label{structcr__mat4_a14cca8968d75c50d0affa3a9d3a80fc1}} \index{cr\_mat4@{cr\_mat4}!w0@{w0}} \index{w0@{w0}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{w0}{w0}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::w0} \mbox{\Hypertarget{structcr__mat4_af1e0a83256fd0cccc45eeef0b7e5d19c}\label{structcr__mat4_af1e0a83256fd0cccc45eeef0b7e5d19c}} \index{cr\_mat4@{cr\_mat4}!w1@{w1}} \index{w1@{w1}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{w1}{w1}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::w1} \mbox{\Hypertarget{structcr__mat4_ae9103574b82172dd809be204cf795c37}\label{structcr__mat4_ae9103574b82172dd809be204cf795c37}} \index{cr\_mat4@{cr\_mat4}!w2@{w2}} \index{w2@{w2}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{w2}{w2}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::w2} \mbox{\Hypertarget{structcr__mat4_ae88ea7d36c60d35f02afde032cfd5b5d}\label{structcr__mat4_ae88ea7d36c60d35f02afde032cfd5b5d}} \index{cr\_mat4@{cr\_mat4}!w3@{w3}} \index{w3@{w3}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{w3}{w3}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::w3} \mbox{\Hypertarget{structcr__mat4_a18de954e61bae64b09c3cb409134dbf3}\label{structcr__mat4_a18de954e61bae64b09c3cb409134dbf3}} \index{cr\_mat4@{cr\_mat4}!x0@{x0}} \index{x0@{x0}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{x0}{x0}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::x0} \mbox{\Hypertarget{structcr__mat4_a3c05dae15eb33520cd153d1a2b6244a7}\label{structcr__mat4_a3c05dae15eb33520cd153d1a2b6244a7}} \index{cr\_mat4@{cr\_mat4}!x1@{x1}} \index{x1@{x1}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{x1}{x1}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::x1} \mbox{\Hypertarget{structcr__mat4_aebf18437997bfae1571da0526aafb6ef}\label{structcr__mat4_aebf18437997bfae1571da0526aafb6ef}} \index{cr\_mat4@{cr\_mat4}!x2@{x2}} \index{x2@{x2}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{x2}{x2}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::x2} \mbox{\Hypertarget{structcr__mat4_a4f21d21c1ba812a86c011924da16962b}\label{structcr__mat4_a4f21d21c1ba812a86c011924da16962b}} \index{cr\_mat4@{cr\_mat4}!x3@{x3}} \index{x3@{x3}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{x3}{x3}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::x3} \mbox{\Hypertarget{structcr__mat4_aa448ace6709fa61ed053c2e8648188cf}\label{structcr__mat4_aa448ace6709fa61ed053c2e8648188cf}} \index{cr\_mat4@{cr\_mat4}!y0@{y0}} \index{y0@{y0}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{y0}{y0}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::y0} \mbox{\Hypertarget{structcr__mat4_ace7d676c9423e28d98c447c82a6d1040}\label{structcr__mat4_ace7d676c9423e28d98c447c82a6d1040}} \index{cr\_mat4@{cr\_mat4}!y1@{y1}} \index{y1@{y1}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{y1}{y1}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::y1} \mbox{\Hypertarget{structcr__mat4_a3bca767f170fbdced1a20f1b0daba4e6}\label{structcr__mat4_a3bca767f170fbdced1a20f1b0daba4e6}} \index{cr\_mat4@{cr\_mat4}!y2@{y2}} \index{y2@{y2}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{y2}{y2}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::y2} \mbox{\Hypertarget{structcr__mat4_a27047478fc3dce659f12c795fb27ff93}\label{structcr__mat4_a27047478fc3dce659f12c795fb27ff93}} \index{cr\_mat4@{cr\_mat4}!y3@{y3}} \index{y3@{y3}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{y3}{y3}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::y3} \mbox{\Hypertarget{structcr__mat4_a2eb1c49e0732782f62dacc7b79446bb4}\label{structcr__mat4_a2eb1c49e0732782f62dacc7b79446bb4}} \index{cr\_mat4@{cr\_mat4}!z0@{z0}} \index{z0@{z0}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{z0}{z0}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::z0} \mbox{\Hypertarget{structcr__mat4_a92f3f8fe7a2a2691aef8594bd8b79011}\label{structcr__mat4_a92f3f8fe7a2a2691aef8594bd8b79011}} \index{cr\_mat4@{cr\_mat4}!z1@{z1}} \index{z1@{z1}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{z1}{z1}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::z1} \mbox{\Hypertarget{structcr__mat4_ae4e714764e9d02a2c1986944f2e5791e}\label{structcr__mat4_ae4e714764e9d02a2c1986944f2e5791e}} \index{cr\_mat4@{cr\_mat4}!z2@{z2}} \index{z2@{z2}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{z2}{z2}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::z2} \mbox{\Hypertarget{structcr__mat4_a6ea965db3224c0e6239afdbead85c101}\label{structcr__mat4_a6ea965db3224c0e6239afdbead85c101}} \index{cr\_mat4@{cr\_mat4}!z3@{z3}} \index{z3@{z3}!cr\_mat4@{cr\_mat4}} \doxysubsubsection{\texorpdfstring{z3}{z3}} {\footnotesize\ttfamily \mbox{\hyperlink{group__base_gac679409efbbe825bcee7bc9437d10194}{cr\+\_\+float}} cr\+\_\+mat4\+::z3} The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item Projects/\+C\+Ray/\mbox{\hyperlink{_c_ray_8h}{C\+Ray.\+h}}\end{DoxyCompactItemize} % !TEX encoding = UTF-8 Unicode \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{okithm}[2021/11/11 v0.9 by I] % ======== Options ======== \RequirePackage{kvoptions} \DeclareBoolOption[true]{theorems} \DeclareComplementaryOption{notheorem}{theorems} \DeclareBoolOption[true]{algorithms} \DeclareComplementaryOption{noalgorithm}{algorithms} \DeclareStringOption[section]{numberwithin} \DeclareStringOption[long]{optstyle} \DeclareBoolOption[true]{enumitem} \DeclareComplementaryOption{noenumitem}{enumitem} \ProcessKeyvalOptions* % ======== Theorems ======== \ifokithm@theorems \RequirePackage{amsthm} \RequirePackage{thmtools} \RequirePackage{translations} \ProvideTranslation{English}{theorem}{Theorem} \ProvideTranslation{English}{definition}{Definition} \ProvideTranslation{English}{lemma}{Lemma} \ProvideTranslation{English}{proposition}{Proposition} \ProvideTranslation{English}{corollary}{Corollary} \ProvideTranslation{English}{example}{Example} \ProvideTranslation{English}{remark}{Remark} \ProvideTranslation{English}{proof}{Proof} \ProvideTranslation{Japanese}{theorem}{定理} \ProvideTranslation{Japanese}{definition}{定義} \ProvideTranslation{Japanese}{lemma}{補題} \ProvideTranslation{Japanese}{proposition}{命題} \ProvideTranslation{Japanese}{corollary}{系} \ProvideTranslation{Japanese}{example}{例} \ProvideTranslation{Japanese}{proof}{証明} \declaretheoremstyle{okithm@defstyle} \declaretheoremstyle[qed=\qedsymbol]{okicmd@exstyle} \ifcurrentlanguage{Japanese}{% \declaretheoremstyle{okicmd@thmstyle} \declaretheoremstyle[qed=\qedsymbol]{okicmd@proofstyle} }{% \declaretheoremstyle[bodyfont=\normalfont\itshape]{okicmd@thmstyle} \declaretheoremstyle[headfont=\normalfont\itshape,qed=\qedsymbol]{okicmd@proofstyle} } \declaretheorem[% name=\protect\GetTranslation{theorem},% style=okicmd@thmstyle,% numberwithin=\okithm@numberwithin% ]{theorem} \declaretheorem[% name=\protect\GetTranslation{definition},% style=okithm@defstyle,% sibling=theorem% ]{definition} \declaretheorem[% name=\protect\GetTranslation{lemma},% style=okicmd@thmstyle,% sibling=theorem% ]{lemma} \declaretheorem[% name=\protect\GetTranslation{proposition},% style=okicmd@thmstyle,% sibling=theorem% ]{proposition} \declaretheorem[% name=\protect\GetTranslation{corollary},% style=okicmd@thmstyle,% sibling=theorem% ]{corollary} \declaretheorem[% name=\protect\GetTranslation{example},% style=okicmd@exstyle,% sibling=theorem% ]{example} \declaretheorem[% name=\protect\GetTranslation{remark},% style=okithm@defstyle,% sibling=theorem% ]{remark} \let\proof\@undefined \let\endproof\@undefined \declaretheorem[% name=\protect\GetTranslation{proof},% style=okicmd@proofstyle,% numbered=no% ]{proof} \fi % ======== Algorithms ======== \ifokithm@algorithms \let\Return\relax \RequirePackage{algorithm} \RequirePackage[noend]{algpseudocode} \algnewcommand{\algorithmicinput}{\makebox[\widthof{\textbf{Output}}][l]{\textbf{Input}}\textbf{:}} \algnewcommand{\Input}{\item[\algorithmicinput]} \algnewcommand{\algorithmicoutput}{\textbf{Output:}} \algnewcommand{\Output}{\item[\algorithmicoutput]} \algnewcommand{\algorithmicto}{\textbf{to}} \algdef{SE}[FOR]{ForTo}{EndFor}[2]{\algorithmicfor\ #1 \algorithmicto\ #2 \algorithmicdo}{\algorithmicend\ \algorithmicfor}\algtext*{EndFor} \fi % ======== Optimization Problems ======== \RequirePackage{array} \RequirePackage{xkeyval} \define@cmdkeys{Minimizekey}[optprob@]{label,name,variable} \define@choicekey*{Minimizekey}{style}[\val]{short,long}{\def\optprob@style{\val}} % https://tex.stackexchange.com/questions/271062/labeling-a-text-and-referencing-it-later \MakeRobust{\ref}% avoid expanding it when in a textual label \newcommand{\labeltext}[2]{% \@bsphack \csname phantomsection\endcsname% in case hyperref is used \def\@currentlabel{#1}{\label{#2}}% \@esphack } \newcommand{\okithm@makeoptprob}[5]{% % #1 : options % #2 : minimize % #3 : object function % #4 : subject to % #5 : constraints \setkeys{Minimizekey}{#1}% \ifdefined\optprob@label \labeltext{\optprob@name}{\optprob@label}% \fi \begin{align}% \ifdefined\optprob@name% (\text{\optprob@name})\quad% \fi% \begin{array}{|cl}% \ifdefined\optprob@variable% \underset{\optprob@variable}{\mathrm{#2}}% \else% \text{#2}% \fi% &$\begin{array}[t]{l}\displaystyle#3\end{array}$ \\ \text{#4}% &$\begin{array}[t]{>{\displaystyle}l<{}>{\displaystyle}l<{}}#5\end{array}$% \end{array}% \end{align}% \let\optprob@label\undefined% \let\optprob@name\undefined% \let\optprob@variable\undefined% \let\optprob@style\undefined% } \newcommand{\okithm@DeclareOptProb}[5]{% % #1 : name of the command % #2 : minimize % #3 : min % #4 : subject to % #5 : s.t. \newcommand#1[3][]{% \setkeys{Minimizekey}{##1}% \ifdefined\optprob@style\else% \def\optprob@style{\okithm@optstyle}% \fi% \ifthenelse{\equal{\optprob@style}{long}}{% \okithm@makeoptprob{##1}{#2}{##2}{#4}{##3}% }{% \okithm@makeoptprob{##1}{#3}{##2}{#5}{##3}% }\ignorespaces% }% }% \okithm@DeclareOptProb{\Minimize}{minimize}{min}{subject to}{s.t.} \okithm@DeclareOptProb{\Maximize}{maximize}{max}{subject to}{s.t.} % ======== Enumeration ======== \ifokithm@enumitem \RequirePackage{enumitem} \setlist{font=\upshape,leftmargin=*} \setlist[1]{labelindent=\parindent} \setlist[enumerate,1]{label={(\arabic*)}} \setlist[enumerate,2]{label={(\arabic{enumi}-\arabic*)}} \setlist[enumerate,3]{label={(\arabic{enumi}-\arabic{enumii}-\arabic*)}} \fi content/publication/glueck-2013-mnv-2532129-2532133/cite.bib @inproceedings{Glueck:2013:MNV:2532129.2532133, acmid = {2532133}, address = {Toronto, Ont., Canada, Canada}, author = {Glueck, Wigdor, Daniel}, booktitle = {Proceedings of Graphics Interface 2013}, isbn = {978-1-4822-1680-6}, keywords = {navigation model, panning, scrolling, very large data view, zooming}, location = {Regina, Sascatchewan, Canada}, numpages = {8}, pages = {9--16}, publisher = {Canadian Information Processing Society}, series = {GI '13}, title = {A Model of Navigation for Very Large Data Views}, url = {http://dl.acm.org/citation.cfm?id=2532129.2532133}, year = {2013} } \begin{minipage}{.3\textwidth} \begin{flushleft} \color{lightblue}\small\textbf{FirstName LastName}\\ ~ \end{flushleft} \end{minipage} \hfill \begin{minipage}{.7\textwidth} \begin{flushright} \small\color{lightblue}Street Number, PostalCode City\\ +49 123 4567890 | E-Mail \end{flushright} \end{minipage} {\large\textbf{Vitae}} Career \begin{table}[H] \begin{tabular}{!{\color{lightgray}\vrule} p{4cm} l} 10/2013 - 09/2016 & Text\\ \end{tabular} \end{table} School \begin{table}[H] \begin{tabular}{!{\color{lightgray}\vrule} p{4cm} l} 09/1995 - 08/2003 & Elementary School\\ \end{tabular} \end{table} Languages \begin{table}[H] \begin{tabular}{!{\color{lightgray}\vrule} l} English\\ \end{tabular} \end{table} Hobbies \begin{table}[H] \begin{tabular}{!{\color{lightgray}\vrule} l} :-) \end{tabular} \end{table}matcdac/IETF_RFCsRFC-cite-refs/bib/rfc2487.bib0 % Datatracker information for RFCs on the Legacy Stream is unfortunately often % incorrect. Please correct the bibtex below based on the information in the % actual RFC at https://rfc-editor.org/rfc/rfc2487.txt @misc{rfc2487, series = {Request for Comments}, number = 2487, howpublished = {RFC 2487}, publisher = {RFC Editor}, doi = {10.17487/RFC2487}, url = {https://rfc-editor.org/rfc/rfc2487.txt}, author = {}, title = {{SMTP Service Extension for Secure SMTP over TLS}}, pagetotal = 8, year = 1999, month = jan, abstract = {This document describes an extension to the SMTP service that allows an SMTP server and client to use transport-layer security to provide private, authenticated communication over the Internet. This gives SMTP agents the ability to protect some or all of their communications from eavesdroppers and attackers. {[}STANDARDS-TRACK{]}}, } EtiCui/starter-hugo-research-group @article{ernzerhof_assessment_1999, author = {Ernzerhof, , .}, doi = {10.1063/1.478401}, issn = {0021-9606, 1089-7690}, journal = {The Journal of Chemical Physics}, language = {en}, month = {mar}, number = {11}, pages = {5029--5036}, title = {Assessment of the {Perdew}–{Burke}–{Ernzerhof} exchange-correlation functional}, url = {http://aip.scitation.org/doi/10.1063/1.478401}, urldate = {2021-11-07}, volume = {110}, year = {1999} } 1-10 @article{DBLP:journals/tog/DoerschSGSE12, author = { and and and and }, title = {What makes Paris look like Paris?}, journal = {ACM Trans. Graph.}, volume = {31}, number = {4}, year = {2012}, pages = {101}, ee = {http://doi.acm.org/10.1145/2185520.2185597}, bibsource = {DBLP, http://dblp.uni-trier.de} } harshkakashaniya/Traffic_Lane_Detection_for_Autonomous_Vehicles \section{Class List} Here are the classes, structs, unions and interfaces with brief descriptions\+:\begin{DoxyCompactList} \item\contentsline{section}{\hyperlink{classLane}{Lane} }{\pageref{classLane}}{} \item\contentsline{section}{\hyperlink{classLaneDetectionModule}{Lane\+Detection\+Module} }{\pageref{classLaneDetectionModule}}{} \end{DoxyCompactList} 0 Define Architiype Define Walkers Describe their interaction and roles \section{Walkers and Graphs as First Order Citizens}\index{walkers in Jac}\index{graphs in Jac}\input{tex/sec/sec-jacwalkers} \section{Architypes and Actions}\index{architypes in Jac}\index{actions in Jac}\input{tex/sec/sec-jacarch}\chapter*{Acknowledgements} \addcontentsline{toc}{section}{Acknowledgements} \chaptermark{Acknowledgements} First and foremost, I would like to than ... \par % ! TODO: Delete this line \lipsum[1-2] alu0101221960/ExamenLHD0 \hypertarget{classcom_1_1thealgorithms_1_1datastructures_1_1stacks_1_1_reverse_stack}{}\doxysection{com.\+thealgorithms.\+datastructures.\+stacks.\+Reverse\+Stack Class Reference} \label{classcom_1_1thealgorithms_1_1datastructures_1_1stacks_1_1_reverse_stack}\index{com.thealgorithms.datastructures.stacks.ReverseStack@{com.thealgorithms.datastructures.stacks.ReverseStack}} \doxysubsection*{Static Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{classcom_1_1thealgorithms_1_1datastructures_1_1stacks_1_1_reverse_stack_a7adbc88bd9080960ef107e53c1c46732}\label{classcom_1_1thealgorithms_1_1datastructures_1_1stacks_1_1_reverse_stack_a7adbc88bd9080960ef107e53c1c46732}} static void {\bfseries main} (String args\mbox{[}$\,$\mbox{]}) \end{DoxyCompactItemize} \doxysubsection{Detailed Description} Reversal of a stack using recursion. \begin{DoxyAuthor}{Author} , 2021 \end{DoxyAuthor} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item src/main/java/com/thealgorithms/datastructures/stacks/Reverse\+Stack.\+java\end{DoxyCompactItemize} __latexindent_temp.tex % Template for documenting your Arduino projects % Author: % / % http://opensourcelab.salazarserrano.com %%% Template based in the template created by () \documentclass[a4paper,11pt]{article} \usepackage[T1]{fontenc} \usepackage[utf8]{inputenc} \usepackage{graphicx} \usepackage{xcolor} \renewcommand\familydefault{\sfdefault} \usepackage{tgheros} \usepackage[defaultmono]{droidmono} \usepackage{amsmath,amssymb,amsthm,textcomp} \usepackage{enumerate} \usepackage{multicol} \usepackage{tikz} \usepackage{courier} \usepackage{geometry} \geometry{total={210mm,297mm}, left=25mm,right=25mm,% bindingoffset=0mm, top=20mm,bottom=20mm} \linespread{1.3} \newcommand{\linia}{\rule{\linewidth}{0.5pt}} % my own titles \makeatletter \renewcommand{\maketitle}{ \begin{center} \vspace{2ex} {\huge \textsc{\@title}} \vspace{1ex} \\ \linia\\ \@author \hfill \@date \vspace{4ex} \end{center} } \makeatother %%% % custom footers and headers \usepackage{fancyhdr} \pagestyle{fancy} \lhead{} \chead{} \rhead{} \lfoot{Arduino Blink Example} \cfoot{} \rfoot{Page \thepage} \renewcommand{\headrulewidth}{0pt} \renewcommand{\footrulewidth}{0pt} % % code listing settings \usepackage{listings} \lstset{% language = Octave, backgroundcolor=\color{white}, basicstyle=\footnotesize\ttfamily, breakatwhitespace=false, breaklines=true, captionpos=b, commentstyle=\color{gray}, deletekeywords={...}, escapeinside={\%*}{*)}, extendedchars=true, frame=single, keepspaces=true, keywordstyle=\color{orange}, morekeywords={*,...}, numbers=left, numbersep=5pt, numberstyle=\footnotesize\color{gray}, rulecolor=\color{black}, rulesepcolor=\color{blue}, showspaces=false, showstringspaces=false, showtabs=false, stepnumber=2, stringstyle=\color{orange}, tabsize=2, title=\lstname, emphstyle=\bfseries\color{blue}% style for emph={} } %% language specific settings: \lstdefinestyle{Arduino}{% language = Octave, keywords={void, int boolean},% define keywords morecomment=[l]{//},% treat // as comments morecomment=[s]{/*}{*/},% define /* ... */ comments emph={HIGH, OUTPUT, LOW}% keywords to emphasize } %%%----------%%%----------%%%----------%%%----------%%% \begin{document} \title{Documenting Your Arduino Code} \author{ } \date{11/11/2019} \maketitle \section*{Circuit} Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. \begin{figure}[h] \centering \includegraphics[width=8cm]{Arduino.jpg} \caption{Circuit description.} \end{figure} Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. \section*{Code Description by Line} Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. \section*{Arduino Code} Following Listing~\ref{list:first}\ldots{} Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. \subsection*{Write the code in lstlisting} \begin{lstlisting}[label={list:first}, style=Arduino, caption=Blink.ino] /* Blink Turns on an LED on for one second, then off for one second, repeatedly. Most Arduinos have an on-board LED you can control. On the Uno and Leonardo, it is attached to digital pin 13. If you're unsure what pin the on-board LED is connected to on your Arduino model, check the documentation at http://arduino.cc This example code is in the public domain. modified 8 May 2014 by */ // the setup function runs once when you press reset or power the board void setup() { // initialize digital pin 13 as an output. pinMode(13, OUTPUT); } // the loop function runs over and over again forever void loop() { digitalWrite(13, HIGH); // turn the LED on (HIGH is the voltage level) delay(1000); // wait for a second digitalWrite(13, LOW); // turn the LED off by making the voltage LOW delay(1000); // wait for a second } \end{lstlisting} Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. \subsection*{Take code directly from *.ino file} Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. % \lstinputlisting[style=Arduino]{Blink.ino} Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. \end{document} sa-914/Energy-Intermittency-Paper \documentclass[aspectratio=169]{beamer} \usepackage[utf8]{inputenc} \usepackage[T1]{fontenc} \usepackage{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{graphicx} \usepackage{tikz} \usepackage{booktabs} \usepackage{color} \usepackage{multirow} \usepackage[percent]{overpic} \usetheme[ outer/progressbar=foot, outer/numbering=none]{metropolis} % theme settings \metroset{titleformat=smallcaps} % highlighting \newcommand{\highlight}[2][yellow]{\mathchoice% {\colorbox{#1}{$\displaystyle#2$}}% {\colorbox{#1}{$\textstyle#2$}}% {\colorbox{#1}{$\scriptstyle#2$}}% {\colorbox{#1}{$\scriptscriptstyle#2$}}}% % images \graphicspath{ {D:/Users/saketh/Documents/GitHub/Energy-Intermittency-Paper/code/matlab/simulation_april/}} % sig stars \def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi} \newcommand{\xoverbrace}[2][\vphantom{\dfrac{A}{A}}]{\overbrace{#1#2}} \newcommand{\xunderbrace}[2][\vphantom{\dfrac{A}{A}}]{\underbrace{#1#2}} \newcommand\blfootnote[1]{% \begingroup \renewcommand\thefootnote{}\footnote{#1}% \addtocounter{footnote}{-1}% \endgroup } \begin{document} \author{ \& } \title{Efficient Pollution Abatement in Electricity Markets with Intermittent Renewable Energy} %\subtitle{} %\logo{} \institute{Rutgers University} %\date{February 22} %\subject{} %\setbeamercovered{transparent} %\setbeamertemplate{navigation symbols}{} \begin{frame}[plain] \maketitle \end{frame} %%%%% To do %% Line up stuff between empirical methodology equations \begin{frame} \frametitle{Motivation} \vspace{1em} \begin{block}{\centering What is intermittency?} \vspace{1em} \begin{overpic}[width=1\textwidth,tics=10]{../figures/ercot_load.pdf} \put (42,13) {\only<2>{$\Bigg\}$ Underproduction }} \put (71,37) {\only<2>{$\Big\}$ Overproduction }} \end{overpic} \end{block} \end{frame} \begin{frame} \frametitle{Motivation} \begin{block}{\centering How should we model intermittency?} \vspace{2em} ``\textit{\dots the economics of all generating technologies, both intermittent and dispatchable, can be evaluated based on the expected market value of the electricity that they will supply, their total life-cycle costs, and their associated expected profitability. Such an analysis would reflect the actual expected production profiles of dispatchable and intermittent technologies, the value of electricity supplied at different times, and other costs of intermittency associated with reliable network integration.}'' (Joskow, 2011) \end{block} \end{frame} \begin{frame} \frametitle{Model -- Overview} \begin{figure} \includegraphics[width=0.8\textwidth]{../figures/model_diagram.png} \end{figure} \vspace{-1em} \begin{itemize} \setlength\itemsep{0.2em} \item Multi-period model with electricity consumers and producers \item Parametrize the model empirically \item Numerically evaluate the model to study its implications \end{itemize} \end{frame} \begin{frame} \frametitle{Model -- Preview of the Results} \begin{figure} \includegraphics[width=0.8\textwidth]{../figures/model_diagram.png} \end{figure} \vspace{-0.8em} \begin{itemize} \setlength\itemsep{0.2em} \item[$\implies$] Elasticity of substitution between renewables and fossil fuels is non-constant \item[$\implies$] The importance of intermittency (in terms of welfare) increases with the intermittency of present generation \item[$\implies$] Batteries can drastically increase the substitutability of renewables and fossil fuels \end{itemize} \end{frame} \begin{frame} \frametitle{Model -- Consumers} %\begin{block}{Consumer assumptions} % \begin{itemize} % \setlength\itemsep{0.25em} % \item Representative consumer who purchases and uses electricity in each period % \item Consumers prefer to use different amounts of electricity at different times % \item Consumers will substitute electricity consumption across time periods based on relative prices\footnote{ This has been empirically studied by Schwarz et al. (2002), Herriges et al. (1993), King \& Shatrawka (2011).} % \end{itemize} %\end{block} \begin{columns}[T]%beamer \column{0.4\textwidth} \begin{figure} \includegraphics[width=1\textwidth]{../figures/model_diagram_cons.png} \end{figure} \column{0.6\textwidth} \begin{block}{Consumer assumptions} \begin{itemize} \setlength\itemsep{0.25em} \item Representative consumer who purchases and uses electricity in each period \item Consumers prefer to use different amounts of electricity at different times \item Consumers will substitute electricity consumption across time periods based on relative prices%\footnote{ This has been empirically studied by Schwarz et al. (2002), Herriges et al. (1993), King \& Shatrawka (2011).} \end{itemize} \end{block} \end{columns} \end{frame} \begin{frame} \frametitle{Model -- Consumers} \metroset{block=fill} \vspace{1em} \begin{block}{\centering{Consumer's Optimization Problem}} \small \begin{align*} \\[-3em] \text{\textbf{Utility}: } &U = \left( \sum_t \alpha_t Z_t^\phi \right)^{1/\phi} \\[1em] \text{\textbf{Budget Constraint}: } &I = \sum_t p_t Z_t \\[-2.5em] \end{align*} \end{block} \begin{itemize} \setlength\itemsep{0.5em} \small \item $\alpha_t$ captures ``time'' preference \item $\sigma = 1/(1-\phi)$ is the intertemporal elasticity of substitution (IES) \item Total cost of electricity equals the cost of electricity consumed in each period \end{itemize} \end{frame} \begin{frame} \frametitle{Model -- Firms} \begin{columns}[T]%beamer \column{0.4\textwidth} \begin{figure} \includegraphics[width=1\textwidth]{../figures/model_diagram_prod.png} \end{figure} \column{0.6\textwidth} \begin{block}{Producer assumptions} \begin{itemize} \setlength\itemsep{0.25em} \item Representative firm that maximizes profit \item Firms build a fixed amount of capacity in different generation technologies \item Different technologies can access different fractions of their capacity in each period \item Output is not stochastic \end{itemize} \end{block} \end{columns} \end{frame} \begin{frame} \frametitle{Model -- Firms} \metroset{block=fill} \vspace{1em} \begin{block}{\centering{Firm's Optimization Problem}} \small \begin{align*} \\[-3em] \text{\textbf{Profit}: } &\Pi = p^T Z - c^T X \\[1em] &= p^T \xi X - c^T X \\[-2.5em] \end{align*} \end{block} \begin{itemize} \setlength\itemsep{0.25em} \item Firms choose fixed quantities $X_i$ of each generation technology $i$ \item Technology $i$ produces $\xi_{i,t}$ electricity per unit in period $t$ \item Cost of building $X_i$ units of generation technology $i$ is $c_i X_i$ \item Reminiscent of Joskow (2011) $\implies$ $X$ is chosen ``\textit{\dots based on the expected market value of the electricity that they will supply, their total life-cycle costs, and their associated expected profitability\dots}'' \end{itemize} \end{frame} \begin{frame} \frametitle{Model -- Solution} \metroset{block=fill} \begin{block}{\centering Equilibrium} \begin{align*} \\[-3em] \text{\textbf{Utility Max}: } Z_t &= \left(\frac{\alpha_t}{p_t} \right)^\sigma \frac{I}{\sum_t \alpha_t^\sigma p_t^{1-\sigma}} \\[1em] \text{\textbf{Profit Max}: \hspace{0.1em} } p &= \xi^{-1} c \\[-2.5em] \end{align*} \end{block} \vspace{1em} \begin{itemize} \setlength\itemsep{0.25em} \item Consumer's optimum is the standard CES solution \item Analytically tractable only when $\sigma = 1$ and number of generation technologies equals the number of periods \item Can study the solution numerically by parametrizing $\xi, c, \alpha$, and $\sigma$. \end{itemize} \end{frame} %\begin{frame} %\frametitle{Empirical Methodology} % %\vspace{2em} %\begin{block}{\centering{Goal}} % \centering \vspace{0.2em} % Estimate the intertemporal elasticity of \\ % substitution for electricity consumption $\sigma$ %\end{block} %\vspace{1em} % %\begin{itemize} % \setlength\itemsep{0.75em} % \item <1->CES First order condition: $\dfrac{Z_t}{Z_s} = \left(\dfrac{\alpha_t \, p_s}{\alpha_s \, p_t} \right)^\sigma $ % \item <1->Cet par, an IES of $\sigma$ implies a 1\% increase in the relative price of electricity $(p_{t}/p_{t+1})$ decreases relative electricity consumption $(Z_{t}/Z_{t+1})$ by $\sigma\%$ % \item <1->Directly affects the economic importance of intermittency %\end{itemize} %\vspace{2.75em} % %\end{frame} %\begin{frame} %\frametitle{Empirical Methodology -- Theory} % % % %$$\xoverbrace{\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i})}^{\text{CES First Order Condition}}$$ %\vspace{-1.75em} %$$+ \sigma \ln (\alpha_{t,i} / \alpha_{s,i}) + u_i$$ %\vspace{1em} %\begin{itemize} % \setlength\itemsep{0.25em} % \item $\sigma$ = Intertemporal elasticity of substitution for electricity consumption % \item $\ln (Z_{t, i} / Z_{s, i})$ = Log Difference in electricity consumption % \item $\ln (P_{t,i} / P_{s,i})$ = Log Difference in electricity prices % \item $\alpha_{t,i},\, \alpha_{s,i}$ = Demand shifters % \item Subscripts $t$ and $s$ refer to time periods while $i$ refers to observations %\end{itemize} %\vspace{2.75em} % %\end{frame} % % % %\begin{frame} %\frametitle{Empirical Methodology -- Theory} % % % % $$\xoverbrace{\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i})}^{\text{CES First Order Condition}}$$ % \vspace{-1.5em} % $$+ \xunderbrace{\,\gamma_{t,i} CDD_{t,i} + \omega_{t,i} HDD_{t,i} + \gamma_{s,i} CDD_{s,i} + \omega_{s,i} HDD_{s,i} + \eta \Delta_{t,s}}_{\text{Demand Controls}} + u_i$$ % % \begin{itemize} % \setlength\itemsep{0.25em} % \item $\sigma$ = Intertemporal elasticity of substitution for electricity consumption % \item $\ln (Z_{t, i} / Z_{s, i})$ = Log difference in electricity consumption % \item $\ln (P_{t,i} / P_{s,i})$ = Log difference in electricity prices % \item $CDD_{t,i}$ = Cooling Degree Days % \item $HDD_{t,i}$ = Heating Degree Days % \item $\Delta_{t,s}$ = Difference in months between periods $t$ and $s$ % \end{itemize} % %\end{frame} % % %\begin{frame} %\frametitle{Empirical Methodology -- Theory} % %\vspace{2em} %\begin{block}{\centering What if there was intertemporal substitution on the supply side?} % % \vspace{2em} % \begin{itemize} % \item During the 1973 oil crisis, refineries increased gasoline stocks expecting higher future prices % \item Previous literature has not accounted for endogeneity % \item Might bias $\sigma$ downwards % \end{itemize} % %\end{block} % % % %\end{frame} % $$\xoverbrace{\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i})}^{\text{CES First Order Condition}}$$ % \vspace{-1.5em} % $$+ \xunderbrace{\,\gamma_{t,i} CDD_{t,i} + \omega_{t,i} HDD_{t,i} + \gamma_{s,i} CDD_{s,i} + \omega_{s,i} HDD_{s,i} + \eta \Delta_{t,s}}_{\text{Demand Controls}} + u_i$$ \begin{frame} \frametitle{Empirical Methodology -- Theory} \vspace{1em} %\begin{table} % \begin{tabular}{@{\extracolsep{2em}}lc} % \multirow{2}{*}{\textbf{Demand:}}\quad & $\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i}) + \,\gamma_{t,i} CDD_{t,i} $\\ % & $+ \,\omega_{t,i} HDD_{t,i} + \gamma_{s,i} CDD_{s,i} + \omega_{s,i} HDD_{s,i} + \eta \Delta_{t,s} + u_i$ \\[1em] % \end{tabular} %\end{table} \begin{table} \begin{tabular}{@{\extracolsep{2em}}lc} \multirow{2}{*}{\textbf{Demand:}}\quad & $\xoverbrace{\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i})}^{\text{CES First Order Condition}}$\\ & $+ \xunderbrace{\,\gamma_{t,i} CDD_{t,i} + \omega_{t,i} HDD_{t,i} + \gamma_{s,i} CDD_{s,i} + \omega_{s,i} HDD_{s,i} + \eta \Delta_{t,s}}_{\text{Demand Controls}} + u_i$ \\[1em] \end{tabular} \end{table} \vspace{1em} \begin{itemize} \item $CDD_{t,i}$ = Cooling Degree Days \item $HDD_{t,i}$ = Heating Degree Days \item $\Delta_{t,s}$ = Difference in months between periods $t$ and $s$ \end{itemize} \end{frame} \begin{frame} \frametitle{Empirical Methodology -- Theory} \vspace{1em} \begin{table} \begin{tabular}{@{\extracolsep{2em}}lc} \multirow{2}{*}{\textbf{Demand:}}\quad & $\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i}) + \,\gamma_{t,i} CDD_{t,i} $\\ & $+ \,\omega_{t,i} HDD_{t,i} + \gamma_{s,i} CDD_{s,i} + \omega_{s,i} HDD_{s,i} + \eta \Delta_{t,s} + u_i$ \\[1em] \multirow{2}{*}{\textbf{Supply:}}\quad & $\ln (Z_{ t, i} / Z_{ s, i}) = \beta \ln (P_{t,i} / P_{s,i})$\\ & $+ \,\psi$ \hspace{-0.25em}\colorbox{yellow}{$\ln (C_{t,i} / C_{s,i})$}$ + v_{i}$ \end{tabular} \end{table} \vspace{1em} \begin{itemize} \item $\ln (C_{t,i} / C_{s,i})$ = Log difference in coal prices \item Coal prices affect the supply of electricity but not its demand \item Can use IV to handle endogeneity \end{itemize} \end{frame} \begin{frame} \frametitle{Empirical Methodology -- Approach} \begin{itemize} \setlength\itemsep{1em} \item Approach the problem using OLS, IV (2SLS), and a semiparametric regression \item \textbf{Semiparametric regression} - Partially Linear IV \begin{itemize} \item Places controls (degree days, time) and instrument (coal prices) in smooth, unknown functions estimated using Kernel regressions \item Error term assumed to be mean zero but may be non-Gaussian \item Robust to misspecified functional forms on controls and instrument \end{itemize} \end{itemize} \begin{table} \begin{tabular}{@{\extracolsep{2em}}lc} \multirow{2}{*}{\textbf{Demand:}}\quad & $\ln (Z_{ t, i} / Z_{ s, i}) = -\sigma \ln (P_{t,i} / P_{s,i}) $ \\ & $+ $ \colorbox{yellow}{$f(CDD_{t,i}, HDD_{t,i}, CDD_{s,i} , HDD_{s,i}, \Delta_{t,s} )$}$ + u_i$ \\[1em] \multirow{2}{*}{\textbf{Supply:}}\quad & $\ln (Z_{ t, i} / Z_{ s, i}) = \beta \ln (P_{t,i} / P_{s,i})$\\ & $+ $\colorbox{yellow}{$g( C_{t,i} / C_{s,i})$}$ + v_{i}$ \end{tabular} \end{table} \end{frame} \begin{frame} \frametitle{Data} \begin{itemize} \setlength\itemsep{1em} \item Panel consists of monthly data for each state in the US from 2011 to 2018 \item Monthly retail electricity price and consumption data is obtained from the EIA \item Coal prices for the same period are also obtained from the EIA \item Degree day data is collected from the NOAA \item Each variable is trimmed by 1\% \end{itemize} \end{frame} \begin{frame} \frametitle{Results} \begin{table} \caption{Partially Linear IV Regression Results} \label{table:3} \small \begin{tabular}{@{\extracolsep{4em}}lccc} \\[-5ex]\hline \hline \\[-1.8ex] & \multicolumn{3}{c}{\textit{Instrument: $ \ln (C_{t,i} / C_{s,i})$}} \\ \cline{2-4} %\\[-1.8ex] & ln\_load\_rel \textasciitilde ln\_price\_rel & ln\_load\_rel \textasciitilde (CDD\_1) & ln\_load\_rel \textasciitilde time\_diff + (CDD\_1) \\ \\[-1.8ex] & (1) & (2) & (3)\\ [0.5ex] \hline \\[-1.8ex] $\hat{\sigma} $ & 2.9976$^{***}$ & 1.2123$^{***}$ & \colorbox{yellow}{0.8847}$^{***}$ \\ & (0.169) & (0.052) & (0.044) \\ [0.9ex] \hline \\[-1.8ex] Time Control & & & Yes \\ Degree Day Controls & & Yes & Yes \\ Observations & 6817 & 6817 & 6817 \\ \hline \hline \\[-2.5ex] \multicolumn{4}{@{}p{37em}@{}}{\scriptsize \textit{Note: } The log difference in coal price between period $t$ and $s$, $ \ln (C_{t,i} / C_{s,i})$, is used as an instrument in these regressions. The sample covers all 50 US states from 2011 to 2018; outliers are removed by trimming 1\% of each variable except $\Delta_{t,s}$. The unit of observation is a set $(t,s,i)$ where $t \neq s$ are months and $i$ is a state. Robust standard errors are reported in parentheses. *p$\textless$0.05, **p$\textless$0.01, ***p$\textless$0.001} \\ \end{tabular} \end{table} \end{frame} \begin{frame} \frametitle{Discussion} \vspace{2em} \begin{block}{\centering What are the practical implications of IES $\boldsymbol{\sigma}$ = 0.88?} \end{block} \vspace{1em} \begin{itemize} \setlength\itemsep{0.5em} \item Consider our model in a setting with two periods (peak and off-peak) and two technologies (solar and coal) \item We parametrize our model using data on cost and efficiency from the EIA \item We then numerically evaluate the model \end{itemize} \end{frame} \begin{frame} \frametitle{Discussion} \vspace{2em} \begin{block}{Does intermittency motivate a CES relationship between renewables and fossil fuel technologies? } \begin{itemize} \setlength\itemsep{0.5em} \item<1-> Many papers assume electricity should be modeled as a CES function of renewable and fossil fuel capacity or capital \item<1-> Often something like: $Electricity = (\alpha \cdot {Renewables}^\phi + \beta \cdot {Fossils}^\phi)^{1/\phi}$ \item<2-> Does this relationship arise in our model? Is the elasticity of substitution ($e$) between these technologies constant? \item<2-> We numerically estimate the elasticity of substitution $e$ between solar and coal in our model \end{itemize} \end{block} \vspace{1em} \end{frame} \begin{frame} \frametitle{Discussion} \begin{block}{\centering The elasticity of substitution between technologies is not constant} \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_elasticity_workshop.png} \put (24,11.5) {\colorbox{yellow}{$\leftarrow$ More reliance on solar, lower $e$}} \put (46,36) {\colorbox{yellow}{More reliance on coal, higher $e$ $\rightarrow$ }} \end{overpic} \end{frame} \begin{frame} \frametitle{Discussion} \begin{block}{\centering The elasticity of substitution between technologies is not constant} \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_elasticity_workshop.png} \put (20,33) {\colorbox{yellow}{\parbox{22em}{As the electricity supply becomes more intermittent, baseload technologies become more valuable}}} \end{overpic} \end{frame} \begin{frame} \frametitle{Discussion} \begin{block}{\centering Demand for base load capacity becomes more inelastic with its price } \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_coal_elas_workshop.png} \put (35,22) {\colorbox{yellow}{\parbox{22em}{As the electricity supply becomes more intermittent, baseload technologies become more valuable}}} \end{overpic} \end{frame} %\begin{frame} %\frametitle{Discussion} % %\begin{block}{What does a non-constant elasticity of substitution between intermittent and base load technologies imply?} % \begin{itemize} % \setlength\itemsep{0.5em} % \item Models assuming a constant elasticity of substitution may not be realistic % \item Taxes/subsidies need to be increasingly larger to shift the same fraction of capacity towards renewables and away from fossil fuels % \item Over time, we may expect the share of renewable energy to increase $\implies$ elasticity of substitution between renewables and fossil fuels will decrease over time % \end{itemize} %\end{block} %\end{frame} \begin{frame} \frametitle{Discussion} \begin{block}{How should we handle pollution abatement when generation is intermittent?} \begin{itemize} \setlength\itemsep{0.5em} \item Carbon taxes and renewable subsidies still work \item Should also account for its distributional effects \item Welfare effects depend on access to non-intermittent renewables \item Keep equity-efficiency trade-off in mind \end{itemize} \end{block} \end{frame} \begin{frame} \frametitle{Discussion} \begin{block}{\centering Distribution of Hydrothermal and Geothermal Plants (EIA, 2019)} \end{block} \begin{figure} \vspace{-1em} \includegraphics[width=0.8\textwidth]{../figures/rel_renew_map.pdf} \end{figure} \end{frame} \begin{frame} \frametitle{Discussion -- Batteries} \begin{block}{Researching battery technology can make a significant difference} \begin{itemize} \setlength\itemsep{0.5em} \item Reducing the intermittency of renewables greatly increases their substitutability \item Mitigates the distributional side effects of intermittency + policy \item Can roughly approximate the effects of batteries in our model by shifting solar output to the off-peak period when it under-produces \end{itemize} \end{block} \end{frame} \begin{frame} \frametitle{Discussion -- Batteries} \begin{block}{\centering Using batteries to shift solar output greatly increases its substitutability} \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_batteries_workshop.png} \put (31,30) {\colorbox{yellow}{\parbox{15em}{Shifting 10\% of solar output doubles the elasticity when the majority of generation is intermittent}}} \linethickness{2pt} \put(31,27){\color{black}\vector(-1,-1){6.5}} \end{overpic} \end{frame} \begin{frame} \frametitle{Discussion -- Batteries} \begin{block}{\centering Initially, increasing returns to batteries and substitutability} \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_batteries_workshop.png} \put (32,35) {\colorbox{yellow}{\parbox{16em}{Elasticity increases by $\approx 25\%$ for a 5\% shift but by $\approx 70\%$ for a 10\% shift}}} \linethickness{2pt} \put(70,31.5){\color{black}\vector(4,-3){8}} \put(73,31.5){\color{black}\vector(3,-2){3}} \end{overpic} \end{frame} \begin{frame} \frametitle{Conclusion} \begin{itemize} \setlength\itemsep{1em} \item Welfare effects of carbon taxes and renewable subsidies depend on the intermittency of renewables \item Geographic heterogeneity in intermittency can create a trade-off between efficiently and equitably preventing climate change \item Subsidizing battery research can complement other policies by increasing the substitutability of renewable and fossil energy while mitigating their unintentional distributional consequences %\item The VES function can do a reasonable job of approximating the substitutability between renewables and fossil fuels \end{itemize} \end{frame} \begin{frame} \begin{block}{\centering \large Questions?} \end{block} \blfootnote{ (), ()} \end{frame} %%%%%%% References \begin{frame}[allowframebreaks] \frametitle{References} \tiny \begin{itemize} \item ., ., ., \& . (2012). The Environment and Directed Technical Change. American Economic Review, 102(1), 131–166. https://doi.org/10.1257/aer.102.1.131 \item . (1995). The genie out of the bottle: World oil since 1970. MIT Press. \item ., \& . (2012). Electricity provision with intermittent sources of energy. Resource and Energy Economics, 34(3), 319–336. https://doi.org/10.1016/j.reseneeco.2012.01.001 \item ., ., ., \& . (1995). Real-time pricing of electricity for residential customers: Econometric analysis of an experiment. Journal Of Applied Econometrics, 10, S171–S191. \item . (2012). The Private and Public Economics of Renewable Electricity Generation. Journal of Economic Perspectives, 26(1), 67–92. https://doi.org/10.1257/jep.26.1.67 \item . (2011). Efficient pricing and investment in electricity markets with intermittent resources. Energy Policy, 39(7), 3945–3953. https://doi.org/10.1016/j.enpol.2011.01.010 \item Climate Prediction Center (2019). Degree Days Statistics. National Oceanic and Atmospheric Administration. https://www.cpc.ncep.noaa.gov/products/analysis\_monitoring/cdus/degree\_days/ \item ., ., ., \& . (2010). Applying portfolio theory to the electricity sector: Energy versus power. Energy Economics, 33(1), 12–23. https://doi.org/10.1016/j.eneco.2010.05.003 \item ., ., \& . (2017). The Long-Run Dynamics of Electricity Demand: Evidence from Municipal Aggregation. NBER Working Paper Series. https://doi.org/10.3386/w23483 \item Electricity Information Administration (2019a). Net generation, United States, all sectors, annual. Electricity Data Browser. https://www.eia.gov/electricity/data/browser \item Electricity Information Administration (2019b). Levelized Cost and Levelized Avoided Cost of New Generation. Annual Energy Outlook 2019. https://www.eia.gov/outlooks/aeo/pdf/electricity\_generation.pdf \item Electricity Information Administration (2019c). U.S. renewable electricity generation has doubled since 2008. Today in Energy. https://www.eia.gov/todayinenergy/detail.php?id=38752 \item ., \& . (2011). The price elasticity of electricity demand in South Australia. Energy Policy, 39(6), 3709–3719. https://doi.org/10.1016/j.enpol.2011.03.080 \item ., ., ., \& . (2012). Current methods and advances in forecasting of wind power generation. Renewable Energy, 37(1), 1–8. https://doi.org/10.1016/j.renene.2011.05.033 \item . (2000). Models of technology diffusion. Research Policy. Elsevier B.V. https://doi.org/10.1016/S0048-7333(99)00092-X \item ., \& . (2019). On the efficient market diffusion of intermittent renewable energies. Energy Economics, 80, 812–830. https://doi.org/10.1016/j.eneco.2019.01.017 \item ., ., ., \& . (1993). The Response of Industrial Customers to Electric Rates Based Upon Dynamic Marginal Costs. The Review of Economics and Statistics, 75(3), 446. https://doi.org/10.2307/2109458 \item . (2011). Comparing the Costs of Intermittent and Dispatchable Electricity Generating Technologies. American Economic Review, 101(3), 238–241. https://doi.org/10.1257/aer.101.3.238 \item , and (1994). "Customer Response to a Permanent Time- Varying Pricing Program in the United Kingdom." Laurits R. Christensen Associates, Madison, Wisconsin. \item ., ., ., ., \& . (2016). A novel economic model for price-based demand response. Electric Power Systems Research, 135, 1–9. https://doi.org/10.1016/j.epsr.2016.03.026 \item ., \& . (2006). Modelling Dynamic Constraints in Electricity Markets and the Costs of Uncertain Wind Output. Faculty of Economics. \item ., . \& . (2007). Implications of intermittency and transmission constraints for renewables deployment. Cambridge Working Papers in Economics 0711, Faculty of Economics, University of Cambridge. \item . (1990). Efficient Instrumental Variables Estimation of Non-linear Models. Econometrica, 58(4), 809–837. https://doi.org/10.2307/2938351 \item ORNL (2004). Measurement practices for reliability and power quality. A toolkit of reliability measurement practices. U.S. Department of Energy. https://info.ornl.gov/sites/publications/Files/Pub57467.pdf \item ., ., \& . (2017). Substitution between clean and dirty energy inputs: a macroeconomic perspective.(Author abstract). Review of Economics and Statistics, 99(2), 201–212. https://doi.org/10.1162/REST\_a\_00623 \item ., \& . (2005). Household Electricity Demand, Revisited. The Review of Economic Studies, 72(252), 853. Retrieved from http://search.proquest.com/docview/204335314/ \item . (1971). A Class of Variable Elasticity of Substitution Production Functions. Econometrica, 39(1), 61–71. https://doi.org/10.2307/1909140 \item ., ., ., \& . (2002). Industrial response to electricity real-time prices: short run and long run. Economic Inquiry, 40(4), 597–610. https://doi.org/10.1093/ei/40.4.597 \item ., \& . (2018). The capacity value of optimal wind and solar portfolios. Energy, 148, 992–1005. https://doi.org/10.1016/j.energy.2017.12.121 \item . (1986). Density estimation for statistics and data analysis. London; Chapman and Hall. \item . \& ., (2013). Estimation of urban residential electricity demand in China using household survey data.(Statistical data). Energy Policy, 61, 394–402. \item ., \& . (1997). Instrumental Variables Regression with Weak Instruments. Econometrica, 65(3), 557–586. https://doi.org/10.2307/2171753 \item U.S. Bureau of Economic Analysis (2019). Personal consumption expenditures: chain-type price index. FRED, Federal Reserve Bank of St. Louis. https://fred.stlouisfed.org/series/PCEPI, April 27, 2019. \item ., \& . (2001). The Impact of Market Rules and Market Structure on the Price Determination Process in the England and Wales Electricity Market. NBER Working Paper Series. https://doi.org/10.3386/w8248 \item ., ., \& . (1996). Optional real-time pricing of electricity for industrial firms. Pacific Economic Review, 1(1), 79–92. https://doi.org/10.1111/j.1468-0106.1996.tb00175.x \item . (1990). Customer Responsiveness to Real-Time Pricing of Electricity. The Energy Journal, 11(4), 99. https://doi.org/10.5547/ISSN0195-6574-EJ-Vol11-No4-6 \end{itemize} \end{frame} \begin{frame} \frametitle{Appendix -- VES} \begin{block}{How can future models of energy generation substitutability be improved?} \begin{itemize} \setlength\itemsep{0.5em} \item <1-> Assuming a constant elasticity of substitution between renewables and fossil fuels does not accurately capture intermittency \item <1-> Empirical estimates of the elasticity of substitution may be incorrect since the functional form (CES) does not seem to be appropriate \item <1-> Directly implementing our model may not allow for analytical tractability \item <2-> Can try the variable elasticity of substitution (VES) form instead\footnote{The VES function was first defined in Revankar (1971).} \end{itemize} \end{block} \end{frame} \begin{frame} \frametitle{Appendix -- VES} \begin{block}{\centering VES approximation of the relationship between solar and coal} \end{block} \hspace*{-1.5em} \begin{overpic}[width=1.1\textwidth,tics=10]{../figures/fig_ves_approx_workshop.png} \end{overpic} \end{frame} \begin{frame} \frametitle{Appendix -- VES} \begin{block}{What exactly is the VES function?} \begin{itemize} \setlength\itemsep{0.5em} \item Assumes that the elasticity of substitution between factors varies as a linear function of their quantities \item Can be empirically estimated \item Simpler to implement in a theoretical or numerical framework than our model \end{itemize} \begin{align*} &Z = \gamma X_1^{\omega(1-\delta \rho)} \left( X_2 + (\rho - 1) X_1 \right)^{\omega \delta \rho} \\ &\highlight{e = 1 + \beta (X_1 / X_2)} \\ &\,\beta = (\rho - 1) / ( 1- \delta \rho ) \end{align*} \vspace{-4ex} $$\gamma > 0, \quad \omega > 0, \quad0 < \delta < 1, \quad 0 \leq \delta \rho \leq 1 , \quad (X_2/X_1) > -\beta $$ \end{block} \end{frame} %%%%%%remove \begin{frame} \frametitle{Appendix -- Literature} \begin{itemize} \setlength\itemsep{0.2em} \item Some literature approach the problem by numerically optimizing the capacity of intermittent renewables given reliability constraints \footnote{(Musgens \& Neuhoff, 2006), (Neuhoff et al., 2007)} \item Other papers study how intermittent technologies affect the market itself \footnote{(Ambec \& Crampes, 2012), (Chao, 2011), (Borenstein, 2012)} \item A common top-down approach is to model intermittency through a CES function between different energy technologies \footnote{See Papageorgiou et al. (2017) for a survey of the literature taking this approach} %\item Consumers prefer to smooth their electricity consumption\footnote{(Schwarz et al., 2002), (Herriges et al., 1993), (King \& Shatrawka, 2011).} \item Our model is closest to Helm and Mier (2019) \end{itemize} \end{frame} \begin{frame} \frametitle{Appendix -- OLS Results} \begin{table}[t] \centering %\label{table:1} \tiny \begin{tabular}{@{\extracolsep{5pt}}lcccccc} \\[-3ex]\hline \hline \\[-1.8ex] & \multicolumn{6}{c}{\textit{Dependent variable:} $\ln (Z_{ t, i} / Z_{ s, i})$} \\ [0.5ex] \cline{2-7} \\[-1.8ex] & (1) & (2) & (3) & (4) & (5) & (6)\\ [0.5ex] \hline \\[-1.8ex] $-\ln (P_{t,i} / P_{s,i})$ & 0.937$^{***}$ & 1.075$^{***}$ & 1.098$^{***}$ & 1.074$^{***}$ & 1.263$^{***}$ & 1.305$^{***}$ \\ & (0.030) & (0.026) & (0.026) & (0.224) & (0.172) & (0.169) \\ %& & & & & & \\ $\Delta_{t,s}$ & & & 0.0005$^{***}$ & & & 0.0065$^{*}$ \\ & & & (0.0001) & & & (0.0003) \\ %& & & & & & \\ CDD$_t$ & & 1.156$^{***}$ & 1.163$^{***}$ & & 1.164$^{***}$ & 1.174$^{***}$ \\ $\quad(\times 1000^{-1})$ & & (0.017) & (0.017) & & (0.072) & (0.075) \\ %& & & & & & \\ CDD$_s$ & & $-$1.143$^{***}$ & $-$1.158$^{***}$ & & $-$1.200$^{***}$ & $-$1.224$^{***}$ \\ $\quad(\times 1000^{-1})$& & (0.025) & (0.026) & & (0.096) & (0.098) \\ %& & & & & & \\ HDD$_t$ & & 0.246$^{***}$ & 0.245$^{***}$ & & 0.237$^{***}$ & 0.236$^{***}$ \\ $\quad(\times 1000^{-1})$ & & (0.007) & (0.007) & & (0.031) & (0.031) \\ %& & & & & & \\ HDD$_s$ & & $-$0.267$^{***}$ & $-$0.265$^{***}$ & & $-$0.268$^{***}$ & $-$0.263$^{***}$ \\ $\quad(\times 1000^{-1})$ & & (0.008) & (0.008) & & (0.038) & (0.038) \\ %& & & & & & \\ Intercept & 0.028$^{***}$ & 0.012$^{*}$ & 0.026$^{***}$ & & & \\ & (0.004) & (0.006) & (0.007) & & & \\ [0.9ex] \hline \\[-1.8ex] State FEs & & & & Yes & Yes & Yes \\ Observations & 6,817 & 6,817 & 6,817 & 6,817 & 6,817 & 6,817 \\ Adjusted R$^{2}$ & 0.079 & 0.506 & 0.508 & 0.085 & 0.518 & 0.520 \\ F Statistic & 582$^{***}$ & 1399$^{***}$ & 1172$^{***}$ & 685$^{***}$ & 1474$^{***}$ & 1241$^{***}$ \\ [0.5ex] \hline \hline \\[-1.8ex] %\multicolumn{7}{@{}p{41.4em}@{}}{\textit{Note: } The sample covers all 50 US states from 2011 to 2018; outliers are removed by trimming 1\% of each variable except $\Delta_{t,s}$. The unit of observation is a set $(t,s,i)$ where $t \neq s$ are months and $i$ is a state. The coefficient on $-\ln (P_{t,i} / P_{s,i})$ is the estimate of $\sigma$. The variable $\Delta_{t,s}$ is the difference in months between periods $t$ and $s$. CDD$_t$ and HDD$_t$ refer to the total number of heating and cooling degree days in month $t$. We scale the coefficients on degree days for clarity. Robust standard errors are reported in parentheses. *p$\textless$0.05, **p$\textless$0.01, ***p$\textless$0.001} \\ \end{tabular} \end{table} \end{frame} \begin{frame} \frametitle{Appendix -- IV Results} \begin{table}[h] \centering %\caption{IV (2SLS) Regression Results} \label{table:2} \tiny \begin{tabular}{@{\extracolsep{4pt}}lcccccc} \\[-4ex]\hline \hline \\[-1.6ex] & \multicolumn{3}{c}{First-Stage} & \multicolumn{3}{c}{Second-Stage} \\ [0.5ex] & \multicolumn{3}{c}{\textit{Dep. Variable:} $\ln (P_{t,i} / P_{s,i})$ } & \multicolumn{3}{c}{\textit{Dep. Variable:} $\ln (Z_{ t, i} / Z_{ s, i})$}\\ [0.5ex] \cmidrule(lr){2-4} \cmidrule(lr){5-7}\\[-2.9ex] & (A.1) & (B.1) & (C.1) & (A.2) & (B.2) & (C.2)\\ [0.5ex] \hline \\[-1.8ex] $ \ln (C_{t,i} / C_{s,i})$ & $-$0.042$^{***}$ & $-$0.018$^{***}$ & $-$0.018$^{***}$ & & & \\ & (0.002) & (0.002) & (0.002) & & & \\ & & & & & & \\[-1ex] $-\ln (P_{t,i} / P_{s,i})$ & & & & 2.978$^{***}$ & 5.896$^{***}$ & 5.818$^{***}$ \\ & & & & (0.180) & (0.548) & (0.524) \\ $\Delta_{t,s}$ & & & 0.001$^{***}$ & & & 0.003$^{***}$ \\ & & & (0.00004) & & & (0.0004) \\ %& & & & & & \\ CDD$_t$ & & 0.100$^{***}$ & 0.105$^{***}$ & & 1.637$^{***}$ & 1.657$^{***}$ \\ $\quad(\times 1000^{-1})$ & & (0.006) & (0.006) & & (0.068) & (0.067) \\ %& & & & & & \\ CDD$_s$ & & $-$0.096$^{***}$ & $-$0.114$^{***}$ & & $-$1.688$^{***}$ & $-$1.783$^{***}$ \\ $\quad(\times 1000^{-1})$ & & (0.009) & (0.009) & & (0.079) & (0.084) \\ %& & & & & & \\ HDD$_t$ & & $-$0.048$^{***}$ & $-$0.048$^{***}$ & & 0.001 & 0.007 \\ $\quad(\times 1000^{-1})$ & & (0.003) & (0.003) & & (0.031) & (0.030) \\ %& & & & & & \\ HDD$_s$ & & 0.053$^{***}$ & 0.055$^{***}$ & & 0.0001 & 0.007 \\ $\quad(\times 1000^{-1})$ & & (0.003) & (0.003) & & (0.035) & (0.034) \\ [0.9ex] \hline \\[-1.8ex] State FEs & Yes & Yes & Yes & Yes & Yes & Yes \\ Observations & 6817 & 6817 & 6817 & 6817 & 6817 & 6817 \\ Adjusted R$^{2}$ & 0.061 & 0.264 & 0.293 & & & \\ F Statistic & 443$^{***}$ & 489$^{***}$ & 472$^{***}$ & & & \\ \hline \hline \\[-1.8ex] %\multicolumn{7}{@{}p{41.5em}@{}}{\textit{Note: } The log difference in coal price between period $t$ and $s$, $ \ln (C_{t,i} / C_{s,i})$, is used as an instrument in these regressions. The sample covers all 50 US states from 2011 to 2018; outliers are removed by trimming 1\% of each variable except $\Delta_{t,s}$. The unit of observation is a set $(t,s,i)$ where $t \neq s$ are months and $i$ is a state. The coefficient on $\ln (P_{t,i} / P_{s,i})$ is an estimate of $-\sigma$. The variable $\Delta_{t,s}$ is the difference in months between periods $t$ and $s$. CDD$_t$ and HDD$_t$ refer to the total number of heating and cooling degree days in month $t$. We scale the coefficients on degree days for clarity. Robust standard errors are reported in parentheses. *p$\textless$0.05, **p$\textless$0.01, ***p$\textless$0.001} \\ \end{tabular} \end{table} \end{frame} \begin{frame} \frametitle{Appendix -- Model Parameters} $$\underbrace{\alpha_t = 0.6, \,\alpha_s = 0.4}_{Demand\,\,parameters}, \,\underbrace{\xi_1 = (1, \, 1), \,\xi_2 = (1, \, 0.1)}_{Output\,\,parameters}, \,\underbrace{c_1 = 104.3, \,c_2 = 60}_{Cost \,\, parameters}$$ \begin{itemize} \setlength\itemsep{0.5em} \item We normalize quantity units in terms of capacity, so cost is given in \$/MWh and output is given as the fraction of capacity available in each period \item EIA LCOE estimates for 2023 for ``Coal with 30\% CCS'' and ``Solar PV'' are 104.3\$/MWh and 60\$/MWh $\implies$ $c_1 = 104.3$, $c_2 = 60$ \item Based on the graph of ERCOT loads from earlier, consumers prefer that $\approx$ 60\% of their energy arrive between 900-2400 $\implies$ $(\alpha_t, \alpha_s) = (0.6, 0.4)$ \item For same periods, solar output seems to be $\approx 10:1 \implies \xi_2 = (1, \, 0.1)$ \item Coal output is assumed to be constant over both periods $\implies \xi_1 = (1, \, 1)$ \end{itemize} \end{frame} \end{document}src/results_and_discussion/assets/spectrograph_configuration.tex \begin{tikzpicture}[font=\sffamily, >=Latex] \tikzset{ mirror element/.style={color=black, line width=2*\pgflinewidth}, axis/.style={color=black, dash pattern=on 3pt off 3pt}, ray arrow/.tip={Latex[length=3mm]}, angle arrow/.tip={Latex[length=2mm]}, incident ray/.style={->,color=black, line width=2*\pgflinewidth,% >=ray arrow}, difracted ray/.style={->,color=cyan, line width=2*\pgflinewidth,% >=angle arrow}, directed angle/.style={->,color=black, >=angle arrow}, }; \clip (1,-.1) rectangle (11,4.1); \pgfdeclarelayer{bg} % declare background layer \pgfsetlayers{bg,main} % set the order of the layers (main is the standard layer) \newcommand*{\gratingStepA}{0.1} \newcommand*{\gratingStepB}{0.3} \newcommand*{\gratingStepY}{0.1} \newcommand*{\halfGrooveRepetition}{15} \newcommand*{\canvasHeight}{4} \newcommand*{\angleRadiusA}{1.0} \newcommand*{\angleRadiusB}{1.7} \newcommand*{\angleRadiusC}{1.8} \newcommand*{\angleRadiusD}{2.5} % draw the grating \draw[mirror element] (0,0) % first half of the grating \foreach \x in {1,...,\halfGrooveRepetition} { -- ++(\gratingStepA,\gratingStepY) -- ++(\gratingStepB,-\gratingStepY)} coordinate (gratingMiddle) % second half of the grating \foreach \x in {1,...,\halfGrooveRepetition} { -- ++(\gratingStepA,\gratingStepY) -- ++(\gratingStepB,-\gratingStepY)}; % draw axis perpendicular to the grating offseted from the bottom of the groove \draw[axis] ($ (gratingMiddle) + (-\gratingStepB/2,\gratingStepY/2) $) coordinate (zero) -- ++(0,10); % draw the rays in background for them not to overlay the grating \begin{pgfonlayer}{bg} % select the background layer \draw[difracted ray] (zero) -- ($ (zero) + (5,\canvasHeight) $) coordinate (difractedEnd); \draw[incident ray] ($ (zero) + (-1.7,\canvasHeight) $) coordinate (incidentStart) -- (zero); \end{pgfonlayer} % calculate incident and difracted light angles \mypgfextractangle{\incidentAngle}{zero}{incidentStart} \mypgfextractangle{\diffractedAngle}{zero}{difractedEnd} % calculate middle of Ebert angle \varphi \pgfmathparse{\incidentAngle/2 + \diffractedAngle/2} \let\diffractionAxisAngle\pgfmathresult % draw axis in the middle of Ebert angle \varphi \draw[axis] (zero) -- ++(\diffractionAxisAngle:10); % draw angles \draw[directed angle] ($ (zero) + (\incidentAngle:\angleRadiusA) $) arc (\incidentAngle:\diffractedAngle:\angleRadiusA) node[above,pos=0.85] {$\varphi$}; \draw[directed angle] ($ (zero) + (90:\angleRadiusB) $) arc (90:\incidentAngle:\angleRadiusB) node[above,pos=0.5] {$\alpha_i$}; \draw[directed angle] ($ (zero) + (90:\angleRadiusC) $) arc (90:\diffractedAngle:\angleRadiusC) node[above,pos=0.7] {$\alpha_m$}; \draw[directed angle] ($ (zero) + (90:\angleRadiusD) $) arc (90:\diffractionAxisAngle:\angleRadiusD) node[above,pos=0.5] {$\vartheta_m$}; \draw[directed angle] ($ (zero) + (\diffractionAxisAngle:\angleRadiusD) $) arc (\diffractionAxisAngle:\diffractedAngle:\angleRadiusD) node[above,pos=0.5] {$\frac{\varphi}{2}$}; \end{tikzpicture} % Authors: % License: Creative Commons Attribution 4.0 International % License link: creativecommons.org/licenses/by/4.0/legalcode % Contact: % \documentclass[12pt]{article} \usepackage{amsmath} \usepackage{verbatim} \usepackage{mathtools} \setlength{\parindent}{0pt} \setlength{\parskip}{2ex plus 0.5ex minus 0.2ex} \begin{document} \begin{comment} \begin{equation} R\left(x\right) = \frac{\Sigma_a}{2}\left(\int_0^x S\left(x_k\right) e^{-\Sigma_a\left(x-x_k\right)} dx_k + \int_x^L S\left(x_k\right) e^{-\Sigma_a\left(x_k-x\right)} dx_k\right) \label{eq:photo} \end{equation} \begin{equation} R\left(x\right) = \frac{\Sigma_a}{2}\left(\int_0^L S\left(x_k\right) e^{-\Sigma_a\lvert x-x_k\rvert} dx_k\right) \label{eq:simplifiedPhoto} \end{equation} \begin{equation} S\left(x\right) = \alpha\ n_e\left(x\right) \lvert W_e\left(x\right)\rvert\ exp\left(\frac{-E_c}{\lvert E\left(x\right)\rvert}\right) \label{eq:source} \end{equation} \begin{equation} \frac{\partial}{\partial u_j}\left(\frac{\left(\vec{a}\cdot\nabla u\right)^2}{\nabla u\cdot\nabla u}\left(\vec{c}\cdot\nabla u\right)\right) = \frac{\partial}{\partial u}\left(\frac{\left(\vec{a}\cdot\nabla u\right)^2}{\nabla u\cdot\nabla u}\left(\vec{c}\cdot\nabla u\right)\right)\phi_j \label{eq:Jacobians} \end{equation} \begin{equation} \frac{\partial}{\partial u}\nabla u = \nabla \label{eq:simple} \end{equation} \begin{equation} \frac{\partial}{\partial t}n_e+\nabla\cdot\Gamma_e=R_e \end{equation} \begin{equation} \frac{\partial}{\partial t}n_i+\nabla\cdot\Gamma_i=R_i \end{equation} \begin{equation} \frac{\partial}{\partial t}n_{\epsilon}+\nabla\cdot\Gamma_{\epsilon}+\vec{E}\cdot\Gamma_e=R_{\epsilon} \end{equation} \begin{equation} -\nabla\cdot\nabla V = \frac{e}{\epsilon_0}(n_i-n_e) \end{equation} \begin{align} \Gamma_e &= -n_e(\mu_e\cdot\vec{E}-D_e\cdot\nabla n_e) \\ \Gamma_{\epsilon} &= -n_{\epsilon}(\mu_{\epsilon}\cdot\vec{E}-D_{\epsilon}\cdot\nabla n_{\epsilon}) \\ n_{\epsilon} &= \frac{3}{2}n_eT_e \\ \vec{E} &= -\nabla V \\ \end{align} Example of a reaction in R$_{e,i,\epsilon}$: \begin{align} R_1 &= k_1n_en_{i} \\ k_1 &= c_1T_e^{0.5}exp(-E_a/T_e) \end{align} DGFunctionDiffusionDirichletBC \begin{equation} { \nabla u \cdot n_e} [v] + \epsilon { \nabla v \cdot n_e } [u] + (\frac{\sigma}{|e|} \cdot [u][v]) \end{equation} DGDiffusion \begin{equation*} \begin{multlined} {\nabla u * n_e} [v] + \epsilon { \nabla v * n_e } [u] + (\sigma / |e| * [u][v]) \\ [a] = [ a_1 - a_2 ] \\ {a} = 0.5 * (a_1 + a_2) \end{multlined} \end{equation*} \begin{gather*} \int \vec{\Gamma}\cdot\vec{n}\,ds\\ \int \vec{\Gamma}\cdot\vec{n}\,2\pi r\,ds \end{gather*} \end{comment} Hi all, I'm looking for suggestions. I'm running a 2D simulation with two variables: electrical potential (V) and some positively charged ions (u). At my cathode boundary, I have the following condition for the potential: \begin{equation} 0 = V_{src} - V + R \int_{x_1}^{x_2}\left(u\nabla V\cdot\vec{n} - \nabla u\cdot\vec{n}\right)dx \end{equation} where $V_{src}$ is a constant potential generated by a battery in an external circuit and R is a resistance in the external circuit. Does anyone have a suggestion for the best way to impose this condition that will maximize the efficiency of my solution? At heart it's a dirichlet condition on the potential, but it includes a non-local flux integral containing both of my dependent variables, u \& V. Currently I'm using a postprocessor to calculate the integrated flux on either timestep\_begin or timestep\_end and then feeding it to a Dirichlet condition on the potential. This method converges with PJFNK but not with NEWTON, presumably because I don't have a Jacobian for my condition. I would like to be able to use NEWTON if I can (I've been very meticulous about forming Jacobians for all my other kernels and boundary conditions). I was thinking about using something like a PenaltyDirichletBC and then my Jacobian elements might look something like: \begin{equation} \frac{\partial R_{v,i}}{\partial u_j} = R \int_{x_1}^{x_2}\left(\phi_j\nabla V\cdot\vec{n} -\nabla\phi_j\cdot\vec{n}\right)dx \end{equation} where obviously $\phi_j$ would only be non-zero over two adjacent boundary elements, but I am not quite sure how I would implement this, e.g. selectively integrate with respect to a certain shape function. Postprocessors I believe don't have access to $\phi$. So does anyone have any good ideas or have I just vomited up a bunch of craziness? \textbf{Graves, Moisan $H_{\phi}$ equation} \begin{equation} \frac{\partial}{\partial r}\left( \epsilon_r^{-1} \frac{1}{r}\frac{rH_{\phi}}{\partial r}\right) + \frac{\partial}{\partial z}\left(\epsilon_z^{-1} \frac{\partial H_{\phi}}{\partial z}\right) + k_0H_{\phi} = 0 \end{equation} \end{document} 0 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \iffalse %%%% % % % Copyright (c) 2017 - (www.mhelvens.net) % % % % https://github.com/mhelvens/latex-lt3graph % % % % This work may be distributed and/or modified under the conditions % % of the LaTeX Project Public License, either version 1.3 of this % % license or (at your option) any later version. The latest version % % of this license is in http://www.latex-project.org/lppl.txt % % and version 1.3 or later is part of all distributions of LaTeX % % version 2005/12/01 or later. % % % % This work has the LPPL maintenance status 'maintained'. % % % % The Current Maintainer of this work is . % % % % This work consists of the files lt3graph.tex and lt3graph.sty. % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \fi %%%% % \CheckSum{0} % % \CharacterTable % {Upper-case \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z % Lower-case \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z % Digits \0\1\2\3\4\5\6\7\8\9 % Exclamation \! Double quote \" Hash (number) \# % Dollar \$ Percent \% Ampersand \& % Acute accent \' Left paren \( Right paren \) % Asterisk \* Plus \+ Comma \, % Minus \- Point \. Solidus \/ % Colon \: Semicolon \; Less than \< % Equals \= Greater than \> Question mark \? % Commercial at \@ Left bracket \[ Backslash \\ % Right bracket \] Circumflex \^ Underscore \_ % Grave accent \` Left brace \{ Vertical bar \| % Right brace \} Tilde \~} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Package Info} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % \begin{macrocode} \NeedsTeXFormat{LaTeX2e} \RequirePackage{expl3} \ProvidesExplPackage{lt3graph}{2017/01/05}{0.1.8} {a LaTeX3 datastructure for representing directed graphs with data} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Required Packages} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % These are the packages we'll need: % % \begin{macrocode} \RequirePackage{l3keys2e} \RequirePackage{xparse} \RequirePackage{withargs} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Additions to \LaTeX3 Fundamentals} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % These are three macros for working with `set literals' % in an expandable context. They use internal macros from % |l3prop|... Something I'm really not supposed to do. % % \begin{macrocode} \prg_new_conditional:Npnn \__graph_set_if_in:nn #1#2 { p } { \__prop_if_in:nwwn {#2} #1 \s_obj_end \__prop_pair:wn #2 \s__prop { } \q_recursion_tail \__prg_break_point: } \cs_set_eq:NN \__graph_empty_set \s__prop \cs_new:Nn \__graph_set_cons:nn { #1 \__prop_pair:wn #2 \s__prop {} } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Data Access} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % These functions generate the multi-part csnames % under which all graph data is stored: % % \begin{macrocode} \cs_new:Nn \__graph_tl:n { g__graph_data (#1) _tl } \cs_new:Nn \__graph_tl:nn { g__graph_data (#1) (#2) _tl } \cs_new:Nn \__graph_tl:nnn { g__graph_data (#1) (#2) (#3) _tl } \cs_new:Nn \__graph_tl:nnnn { g__graph_data (#1) (#2) (#3) (#4) _tl } \cs_new:Nn \__graph_tl:nnnnn { g__graph_data (#1) (#2) (#3) (#4) (#5) _tl } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % The following functions generate multi-part keys to % use in property maps: % % \begin{macrocode} \cs_new:Nn \__graph_key:n { key (#1) } \cs_new:Nn \__graph_key:nn { key (#1) (#2) } \cs_new:Nn \__graph_key:nnn { key (#1) (#2) (#3) } \cs_new:Nn \__graph_key:nnnn { key (#1) (#2) (#3) (#4) } \cs_new:Nn \__graph_key:nnnnn { key (#1) (#2) (#3) (#4) (#5) } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % A quick way to iterate through property maps holding % graph data: % % \begin{macrocode} \cs_new_protected:Nn \__graph_for_each_prop_datatype:n { \seq_map_inline:Nn \g__graph_prop_data_types_seq {#1} } \seq_new:N \g__graph_prop_data_types_seq \seq_set_from_clist:Nn \g__graph_prop_data_types_seq {vertices, edge-values, edge-froms, edge-tos, edge-triples, indegree, outdegree} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Storing data through pointers} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % The following function embodies a \LaTeX3 design pattern % for representing non-null pointers. This allows data to % be 'protected' behind a macro redirection. Any number of % expandable operations can be applied to the pointer % indiscriminately without altering the data, even when % using |:x|, |:o| or |:f| expansion. Expansion using |:v| % dereferences the pointer and returns the data exactly % as it was passed through |#2|. Expansion using |:c| % returns a control sequence through which the data can % be modified. % % \begin{macrocode} \cs_new_protected:Nn \__graph_ptr_new:Nn { \withargs [\uniquecsname] { \tl_set:Nn #1 {##1} \tl_new:c {##1} \tl_set:cn {##1} {#2} } } \cs_new_protected:Nn \__graph_ptr_gnew:Nn { \withargs [\uniquecsname] { \tl_gset:Nn #1 {##1} \tl_new:c {##1} \tl_gset:cn {##1} {#2} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Creating and initializing graphs} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Globally create a new graph: % % \begin{macrocode} \cs_new_protected:Nn \graph_new:N { \graph_if_exist:NTF #1 { % TODO: error }{ \tl_new:N #1 \tl_set:Nf #1 { \tl_trim_spaces:f {\str_tail:n{#1}} } \int_new:c {\__graph_tl:nnn{graph}{#1}{vertex-count}} \__graph_for_each_prop_datatype:n { \prop_new:c {\__graph_tl:nnn{graph}{#1}{##1}} } } } \cs_generate_variant:Nn \tl_trim_spaces:n {f} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Remove all data from a graph: % % \begin{macrocode} \cs_new_protected:Nn \graph_clear:N {\__graph_clear:Nn #1 { } } \cs_new_protected:Nn \graph_gclear:N {\__graph_clear:Nn #1 {g} } \cs_new_protected:Nn \__graph_clear:Nn { \__graph_for_each_prop_datatype:n { \use:c{prop_#2clear:c} {\__graph_tl:nnn{graph}{#1}{##1}} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Create a new graph if it doesn't already exist, % then remove all data from it: % % \begin{macrocode} \cs_new_protected:Nn \graph_clear_new:N { \__graph_clear_new:Nn #1 { } } \cs_new_protected:Nn \graph_gclear_new:N { \__graph_clear_new:Nn #1 {g} } \cs_new_protected:Nn \__graph_clear_new:Nn { \graph_if_exists:NF #1 { \graph_new:N #1 } \use:c{graph_#2clear:N} #1 } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Set all data in graph |#1| equal to that in graph |#2|: % % \begin{macrocode} \cs_new_protected:Nn \graph_set_eq:NN { \__graph_set_eq:NNn #1 #2 { } } \cs_new_protected:Nn \graph_gset_eq:NN { \__graph_set_eq:NNn #1 #2 {g} } \cs_new_protected:Nn \__graph_set_eq:NNn { \use:c{graph_#3clear:N} #1 \__graph_for_each_prop_datatype:n { \use:c{prop_#3set_eq:cc} {\__graph_tl:nnn{graph}{#1}{##1}} {\__graph_tl:nnn{graph}{#2}{##1}} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % An expandable test of whether a graph exists. It does not % actually test whether the command sequence contains a % graph and is essentially the same as |\cs_if_exist:N(TF)|: % % \begin{macrocode} \cs_set_eq:NN \graph_if_exist:Np \cs_if_exist:Np \cs_set_eq:NN \graph_if_exist:NT \cs_if_exist:NT \cs_set_eq:NN \graph_if_exist:NF \cs_if_exist:NF \cs_set_eq:NN \graph_if_exist:NTF \cs_if_exist:NTF % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Manipulating graphs} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Put a new vertex inside a graph: % % \begin{macrocode} \cs_new_protected:Nn \graph_put_vertex:Nn { \__graph_put_vertex:Nnnn #1 {#2} {} { } } \cs_new_protected:Nn \graph_gput_vertex:Nn { \__graph_put_vertex:Nnnn #1 {#2} {} {g} } \cs_new_protected:Nn \graph_put_vertex:Nnn { \__graph_put_vertex:Nnnn #1 {#2} {#3} { } } \cs_new_protected:Nn \graph_gput_vertex:Nnn { \__graph_put_vertex:Nnnn #1 {#2} {#3} {g} } \cs_new_protected:Nn \__graph_put_vertex:Nnnn { %%% create pointer to value % \use:c{__graph_ptr_#4new:Nn} \l__graph_vertex_data_tl {#3} %%% add the vertex % \use:c{prop_#4put:cnV} {\__graph_tl:nnn{graph}{#1}{vertices}} {#2} \l__graph_vertex_data_tl %%% increment the vertex counter % \use:c{int_#4incr:c} {\__graph_tl:nnn{graph}{#1}{vertex-count}} \graph_get_vertex:NnNT #1 {#2} \l_tmpa_tl { %%% initialize degree to 0 % \use:c{prop_#4put:cnn} {\__graph_tl:nnn{graph}{#1}{indegree}} {#2} {0} \use:c{prop_#4put:cnn} {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} {0} } } \tl_new:N \l__graph_vertex_data_tl % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Put a new edge inside a graph: % % \begin{macrocode} \cs_new_protected:Nn \graph_put_edge:Nnn { \__graph_put_edge:Nnnnn #1 {#2} {#3} {} { } } \cs_new_protected:Nn \graph_gput_edge:Nnn { \__graph_put_edge:Nnnnn #1 {#2} {#3} {} {g} } \cs_new_protected:Nn \graph_put_edge:Nnnn { \__graph_put_edge:Nnnnn #1 {#2} {#3} {#4} { } } \cs_new_protected:Nn \graph_gput_edge:Nnnn { \__graph_put_edge:Nnnnn #1 {#2} {#3} {#4} {g} } \cs_new_protected:Nn \__graph_put_edge:Nnnnn { \graph_get_vertex:NnNTF #1 {#2} \l__graph_from_value_tl { \graph_get_vertex:NnNTF #1 {#3} \l__graph_to_value_tl { \graph_get_edge:NnnNF #1 {#2} {#3} \l_tmpa_tl { %%% increment outgoing degree of vertex #2 % \use:c{prop_#5put:cnf} {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} {\int_eval:n { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} + 1 }} %%% increment incoming degree of vertex #3 % \use:c{prop_#5put:cnf} {\__graph_tl:nnn{graph}{#1}{indegree}} {#3} {\int_eval:n { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{indegree}} {#3} + 1 }} } %%% actually add the edge % \withargs:VVn \l__graph_from_value_tl \l__graph_to_value_tl { \use:c{prop_#5put:cox} { \__graph_tl:nnn{graph}{#1}{edge-froms} } { \__graph_key:nn{#2}{#3} } { \tl_to_str:n{#2} } \use:c{prop_#5put:cox} { \__graph_tl:nnn{graph}{#1}{edge-tos} } { \__graph_key:nn{#2}{#3} } { \tl_to_str:n{#3} } \use:c{__graph_ptr_#5new:Nn} \l__graph_edge_data_tl {#4} \use:c{prop_#5put:coV} { \__graph_tl:nnn{graph}{#1}{edge-values} } { \__graph_key:nn{#2}{#3} } \l__graph_edge_data_tl \use:c{prop_#5put:cox} { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_key:nn{#2}{#3} } { {\tl_to_str:n{#2}} {\tl_to_str:n{#3}} {\l__graph_edge_data_tl} } } }{ % TODO: Error ('to' vertex doesn't exist) } }{ % TODO: Error ('from' vertex doesn't exist) } } \cs_generate_variant:Nn \prop_gput:Nnn {cox, coV, cnf} \cs_generate_variant:Nn \prop_put:Nnn {cox, coV, cnf} \cs_generate_variant:Nn \withargs:nnn {VVn} \tl_new:N \l__graph_edge_data_tl \tl_new:N \l__graph_from_value_tl \tl_new:N \l__graph_to_value_tl % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Remove a vertex from a graph, automatically removing % any connected edges: % % \begin{macrocode} \cs_new_protected:Nn \graph_remove_vertex:Nn { \__graph_remove_vertex:Nnn #1 {#2} { } } \cs_new_protected:Nn \graph_gremove_vertex:Nn { \__graph_remove_vertex:Nnn #1 {#2} {g} } \cs_new_protected:Nn \__graph_remove_vertex:Nnn { \graph_get_vertex:NnNT #1 {#2} \l__graph_vertex_data_tl { %%% remove outgoing edges % \graph_map_outgoing_edges_inline:Nnn #1 {#2} { \use:c{graph_#3remove_edge:Nnn} #1 {##1} {##2} } %%% remove incoming edges % \graph_map_incoming_edges_inline:Nnn #1 {#2} { \use:c{graph_#3remove_edge:Nnn} #1 {##1} {##2} } %%% remove the vertex % \use:c{prop_#3remove:cn} {\__graph_tl:nnn{graph}{#1}{vertices}} {#2} \use:c{prop_#3remove:cn} {\__graph_tl:nnn{graph}{#1}{indegree}} {#2} \use:c{prop_#3remove:cn} {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} %%% decrement the vertex counter % \use:c{int_#3decr:c} {\__graph_tl:nnn{graph}{#1}{vertex-count}} } } \cs_generate_variant:Nn \prop_put:Nnn {cnV} % \tl_new:N \l__graph_vertex_data_tl % reusing from other function % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Remove an edge from the graph: % % \begin{macrocode} \cs_new_protected:Nn \graph_remove_edge:Nnn { \__graph_remove_edge:Nnnn #1 {#2} {#3} { } } \cs_new_protected:Nn \graph_gremove_edge:Nnn { \__graph_remove_edge:Nnnn #1 {#2} {#3} {g} } \cs_new_protected:Nn \__graph_remove_edge:Nnnn { \graph_get_edge:NnnNT #1 {#2} {#3} \l__graph_edge_data_tl { %%% decrement outdegree of vertex #2 % \use:c{prop_#4put:cnf} {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} {\int_eval:n { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} - 1 }} %%% decrement indegree of vertex #3 % \use:c{prop_#4put:cnf} {\__graph_tl:nnn{graph}{#1}{indegree}} {#3} {\int_eval:n { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{indegree}} {#3} - 1 }} %%% actually remove edge % \use:c{prop_#4remove:co} { \__graph_tl:nnn{graph}{#1}{edge-froms} } { \__graph_key:nn{#2}{#3} } \use:c{prop_#4remove:co} { \__graph_tl:nnn{graph}{#1}{edge-tos} } { \__graph_key:nn{#2}{#3} } \use:c{prop_#4remove:co} { \__graph_tl:nnn{graph}{#1}{edge-values} } { \__graph_key:nn{#2}{#3} } \use:c{prop_#4remove:co} { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_key:nn{#2}{#3} } } } \cs_generate_variant:Nn \prop_remove:Nn {co} \cs_generate_variant:Nn \prop_gremove:Nn {co} \cs_generate_variant:Nn \prop_put:Nnn {cnf} \cs_generate_variant:Nn \prop_gput:Nnn {cnf} %\tl_new:N \l__graph_edge_data_tl % reusing from other function % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Add all edges from graph |#2| to graph |#1|, but only % between nodes already present in |#1|: % % \begin{macrocode} \cs_new_protected:Nn \graph_put_edges_from:NN { \__graph_gput_edges_from:NNn #1 #2 { } } \cs_new_protected:Nn \graph_gput_edges_from:NN { \__graph_gput_edges_from:NNn #1 #2 {g} } \cs_new_protected:Nn \__graph_gput_edges_from:NNn { \graph_map_edges_inline:Nn #2 { \graph_if_vertex_exist:NnT #1 {##1} { \graph_if_vertex_exist:NnT #1 {##2} { \use:c{graph_#3put_edge:Nnnn} #1 {##1} {##2} {##3} } } } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Recovering values from graphs with branching} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Test whether a vertex |#2| exists. If so, its value is % stored in |#3| and |T| is left in the input stream. If % it doesn't, |F| is left in the input stream. % % \begin{macrocode} \prg_new_protected_conditional:Nnn \graph_get_vertex:NnN {T, F, TF} { \prop_get:cnNTF { \__graph_tl:nnn {graph} {#1} {vertices} } {#2} #3 { \tl_set:Nv #3 {#3} \prg_return_true: } { \prg_return_false: } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Test whether an edge |#2|--|#3| exists. If so, its value is % stored in |#4| and |T| is left in the input stream. If it % doesn't, |F| is left in the input stream. % % \begin{macrocode} \prg_new_protected_conditional:Nnn \graph_get_edge:NnnN {T, F, TF} { \prop_get:coNTF { \__graph_tl:nnn{graph}{#1}{edge-values} } { \__graph_key:nn{#2}{#3} } #4 { \tl_set:Nv #4 {#4} \prg_return_true: } { \prg_return_false: } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Graph Conditionals} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % An expandable test for the existence of a vertex: % % \begin{macrocode} \prg_new_conditional:Nnn \graph_if_vertex_exist:Nn {p, T, F, TF} { \prop_if_in:cnTF { \__graph_tl:nnn {graph} {#1} {vertices} } { #2 } { \prg_return_true: } { \prg_return_false: } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % An expandable test for the existence of an edge: % % \begin{macrocode} \prg_new_conditional:Nnn \graph_if_edge_exist:Nnn {p, T, F, TF} { \prop_if_in:coTF { \__graph_tl:nnn {graph} {#1} {edge-values} } { \__graph_key:nn{#2}{#3} } { \prg_return_true: } { \prg_return_false: } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Test whether graph |#1| contains a cycle reachable from % vertex |#2|: % % \begin{macrocode} \cs_new:Npn \graph_if_vertex_can_reach_cycle_p:Nn #1#2 { \__graph_if_vertex_can_reach_cycle_p:Nnn #1 {#2} {\__graph_empty_set} } \cs_new:Npn \graph_if_vertex_can_reach_cycle:NnTF #1#2 { \__graph_if_vertex_can_reach_cycle:NnnTF #1 {#2} {\__graph_empty_set} } \cs_new:Npn \graph_if_vertex_can_reach_cycle:NnT #1#2 { \__graph_if_vertex_can_reach_cycle:NnnT #1 {#2} {\__graph_empty_set} } \cs_new:Npn \graph_if_vertex_can_reach_cycle:NnF #1#2 { \__graph_if_vertex_can_reach_cycle:NnnF #1 {#2} {\__graph_empty_set} } \prg_new_conditional:Nnn \__graph_if_vertex_can_reach_cycle:Nnn {p, T, F, TF} % #1: graph id % #2: vertex id % #3: visited vertices in 'prop literal' format (internal l3prop) { \graph_map_outgoing_edges_tokens:Nnn #1 {#2} { \__graph_if_vertex_can_reach_cycle:Nnnnn #1 {#3} } \prg_return_false: } \cs_new:Nn \__graph_if_vertex_can_reach_cycle:Nnnnn % #1: graph id % #2: visited vertices in 'prop literal' format (internal l3prop) % #3: start vertex (not used) % #4: current vertex % #5: edge value (behind ptr, not used) { \bool_if:nT { \__graph_set_if_in_p:nn {#2} {#4} || \__graph_if_vertex_can_reach_cycle_p:Nno #1 {#4} { \__graph_set_cons:nn {#2} {#4} } } { \prop_map_break:n {\use_i:nn \prg_return_true:} } } \cs_generate_variant:Nn \__graph_if_vertex_can_reach_cycle_p:Nnn {Nno} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Test whether graph |#1| contains any cycles: % % \begin{macrocode} \prg_new_conditional:Nnn \graph_if_cyclic:N {p, T, F, TF} % #1: graph id { \graph_map_vertices_tokens:Nn #1 { \__graph_if_cyclic:Nnn #1 } \prg_return_false: } \cs_new:Nn \__graph_if_cyclic:Nnn % #1: graph id % #2: vertex id % #3: vertex value (not used) { \bool_if:nT { \graph_if_vertex_can_reach_cycle_p:Nn #1 {#2} } { \prop_map_break:n {\use_i:nn \prg_return_true:} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % % Test whether graph |#1| contains any cycles: % % % % \begin{macrocode} % \prg_new_protected_conditional:Nnn \graph_get_cycle:NN % {T, F, TF} % % #1: graph id % % #2: l3seq variable to put the cycle description in % { % \seq_clear:N #2 % \__graph_get_cycle:NNTF #1 #2 % {\prg_return_true: } % {\prg_return_false:} % } % % \prg_new_protected_conditional:Nnn \__graph_get_cycle:NN % {T, F, TF} % % #1: graph id % % #2: l3seq variable % { % \graph_map_successors_inline:Nnn #1 {} { % \seq_if_in:NnTF #2 {##1} { % % TODO % }{ % % TODO % } % } % } % % \end{macrocode} % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Assume that graph |#1| is acyclic and test % whether a path exists from |#2| to |#3|: % % \begin{macrocode} \prg_new_conditional:Nnn \graph_acyclic_if_path_exist:Nnn {p, T, F, TF} % #1: graph id % #2: start vertex % #3: end vertex { \graph_map_outgoing_edges_tokens:Nnn #1 {#2} { \__graph_acyclic_if_path_exist:Nnnnn #1 {#3} } \prg_return_false: } \cs_new:Nn \__graph_acyclic_if_path_exist:Nnnnn % #1: graph id % #2: end vertex % #3: start vertex (not used) % #4: possible end vertex % #5: edge value (behind ptr, do not use) { \bool_if:nT { \str_if_eq_p:nn {#4} {#2} || \graph_acyclic_if_path_exist_p:Nnn #1 {#4} {#2} } { \prop_map_break:n {\use_i:nn \prg_return_true:} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Querying Information} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Get the number of vertices in the graph: % % \begin{macrocode} \cs_new:Nn \graph_vertex_count:N { \int_use:c {\__graph_tl:nnn{graph}{#1}{vertex-count}} } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Get the number of edges leading out of vertex |#2|: % % \begin{macrocode} \cs_new:Nn \graph_get_outdegree:Nn { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{outdegree}} {#2} } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Get the number of edges leading into vertex |#2|: % % \begin{macrocode} \cs_new:Nn \graph_get_indegree:Nn { \prop_item:cn {\__graph_tl:nnn{graph}{#1}{indegree}} {#2} } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Get the number of edges connected to vertex |#2|: % % \begin{macrocode} \cs_new:Nn \graph_get_degree:Nn { \int_eval:n{ \graph_get_outdegree:Nn #1 {#2} + \graph_get_indegree:Nn #1 {#2} } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Mapping Graphs} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#2| to all vertex name/value pairs in % the graph. The tokens are supplied with two arguments as % trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_vertices_tokens:Nn { \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{vertices} } { \__graph_map_vertices_tokens_aux:nnv {#2} } } \cs_new:Nn \__graph_map_vertices_tokens_aux:nnn { #1 {#2} {#3} } \cs_generate_variant:Nn \__graph_map_vertices_tokens_aux:nnn {nnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#2| to all vertex name/value pairs in % the graph. The function is supplied with two arguments as % trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_vertices_function:NN { \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{vertices} } { \exp_args:Nnv #2 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the inline function |#2| to all vertex name/value % pairs in the graph. The inline function is supplied with % two arguments: `|#1|' for the name, `|#2|' for the value. % % \begin{macrocode} \cs_new_protected:Nn \graph_map_vertices_inline:Nn { \withargs (c) [\uniquecsname] [#2] { \cs_set:Npn ##1 ####1####2 {##2} \graph_map_vertices_function:NN #1 ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#2| to all edge from/to/value triples % in the graph. The tokens are supplied with three % arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_edges_tokens:Nn { \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_edges_tokens_aux:nnn {#2} } } \cs_new:Nn \__graph_map_edges_tokens_aux:nnn { \__graph_map_edges_tokens_aux:nnnv {#1} #3 } \cs_new:Nn \__graph_map_edges_tokens_aux:nnnn { #1 {#2} {#3} {#4} } \cs_generate_variant:Nn \__graph_map_edges_tokens_aux:nnnn {nnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#2| to all edge from/to/value triples % in the graph. The function is supplied with three % arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_edges_function:NN { \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_edges_function_aux:Nnn #2 } } \cs_new:Nn \__graph_map_edges_function_aux:Nnn { \__graph_map_edges_function_aux:Nnnv #1 #3 } \cs_new:Nn \__graph_map_edges_function_aux:Nnnn { #1 {#2} {#3} {#4} } \cs_generate_variant:Nn \__graph_map_edges_function_aux:Nnnn {Nnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#2| to all edge from/to/value triples % in the graph. The tokens are supplied with three % arguments: `|#1|' for the `from' vertex, `|#2|' for the % `to' vertex and `|#3|' for the edge value. % % \begin{macrocode} \cs_new_protected:Nn \graph_map_edges_inline:Nn { \withargs (c) [\uniquecsname] [#2] { \cs_set:Npn ##1 ####1####2####3 {##2} \graph_map_edges_function:NN #1 ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#3| to the from/to/value triples % for the edges going `to' vertex |#2|. The tokens are % supplied with three arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_incoming_edges_tokens:Nnn { % #1: graph % #2: base vertex % #3: tokens to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_incoming_edges_tokens_aux:nnnn {#2} {#3} } } \cs_new:Nn \__graph_map_incoming_edges_tokens_aux:nnnn % #1: base vertex % #2: tokens to execute % #3: edge key % #4: edge-triple {from}{to}{value} { \__graph_map_incoming_edges_tokens_aux:nnnnv {#1} {#2} #4 } \cs_new:Nn \__graph_map_incoming_edges_tokens_aux:nnnnn % #1: base vertex % #2: tokens to execute % #3: edge 'from' vertex % #4: edge 'to' vertex % #5: edge value { \str_if_eq:nnT {#1} {#4} { #2 {#3} {#4} {#5} } } \cs_generate_variant:Nn \__graph_map_incoming_edges_tokens_aux:nnnnn {nnnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#3| to the from/to/value triples % for the edges going `to' vertex |#2|. The function is % supplied with three arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_incoming_edges_function:NnN { % #1: graph % #2: base vertex % #3: function to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_incoming_edges_function_aux:nNnn {#2} #3 } } \cs_new:Nn \__graph_map_incoming_edges_function_aux:nNnn % #1: base vertex % #2: function to execute % #3: edge key % #4: edge-triple {from}{to}{value} { \__graph_map_incoming_edges_function_aux:nNnnv {#1} #2 #4 } \cs_new:Nn \__graph_map_incoming_edges_function_aux:nNnnn % #1: base vertex % #2: function to execute % #3: edge 'from' vertex % #4: edge 'to' vertex % #5: edge value { \str_if_eq:nnT {#1} {#4} { #2 {#3} {#4} {#5} } } \cs_generate_variant:Nn \__graph_map_incoming_edges_function_aux:nNnnn {nNnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the inline function |#3| to the from/to/value triples % for the edges going `to' vertex |#2|. The inline function is % supplied with three arguments: `|#1|' for the `from' vertex, % `|#2|' is equal to the |#2| supplied to this function and % `|#3|' contains the edge value. % % \begin{macrocode} \cs_new_protected:Nn \graph_map_incoming_edges_inline:Nnn { % #1: graph % #2: base vertex % #3: body to execute \withargs (c) [\uniquecsname] [#2] [#3] { \cs_set:Npn ##1 ####1####2####3 {##3} \graph_map_incoming_edges_function:NnN #1 {##2} ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#3| to the from/to/value triples % for the edges going `from' vertex |#2|. The tokens are % supplied with three arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_outgoing_edges_tokens:Nnn { % #1: graph % #2: base vertex % #3: tokens to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_outgoing_edges_tokens_aux:nnnn {#2} {#3} } } \cs_new:Nn \__graph_map_outgoing_edges_tokens_aux:nnnn % #1: base vertex % #2: tokens to execute % #3: edge key (not used) % #4: edge-triple {from}{to}{value} { \__graph_map_outgoing_edges_tokens_aux:nnnnv {#1} {#2} #4 } \cs_new:Nn \__graph_map_outgoing_edges_tokens_aux:nnnnn % #1: base vertex % #2: tokens to execute % #3: edge 'from' vertex % #4: edge 'to' vertex % #5: edge value { \str_if_eq:nnT {#1} {#3} { #2 {#3} {#4} {#5} } } \cs_generate_variant:Nn \__graph_map_outgoing_edges_tokens_aux:nnnnn {nnnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#3| to the from/to/value triples % for the edges going `from' vertex |#2|. The function is % supplied with three arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_outgoing_edges_function:NnN { % #1: graph % #2: base vertex % #3: function to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_outgoing_edges_function_aux:nNnn {#2} #3 } } \cs_new:Nn \__graph_map_outgoing_edges_function_aux:nNnn % #1: base vertex % #2: function to execute % #3: edge key % #4: edge-triple {from}{to}{value} { \__graph_map_outgoing_edges_function_aux:nNnnv {#1} #2 #4 } \cs_new:Nn \__graph_map_outgoing_edges_function_aux:nNnnn % #1: base vertex % #2: function to execute % #3: edge 'from' vertex % #4: edge 'to' vertex % #5: edge value { \str_if_eq:nnT {#1} {#3} { #2 {#3} {#4} {#5} } } \cs_generate_variant:Nn \__graph_map_outgoing_edges_function_aux:nNnnn {nNnnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the inline function |#3| to the from/to/value triples % for the edges going `from' vertex |#2|. The inline function is % supplied with three arguments: `|#1|' is equal to the |#2| % supplied to this function, `|#2|' contains the `to' vertex and % `|#3|' contains the edge value. % % \begin{macrocode} \cs_new_protected:Nn \graph_map_outgoing_edges_inline:Nnn { % #1: graph % #2: base vertex % #3: body to execute \withargs (c) [\uniquecsname] [#2] [#3] { \cs_set:Npn ##1 ####1####2####3 {##3} \graph_map_outgoing_edges_function:NnN #1 {##2} ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#3| to the key/value pairs % of the vertices reachable from vertex |#2| in one step. % The tokens are % supplied with two arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_successors_tokens:Nnn { % #1: graph % #2: base vertex % #3: tokens to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_successors_tokens_aux:Nnnnn #1 {#2} {#3} } } \cs_new:Nn \__graph_map_successors_tokens_aux:Nnnnn { % #1: the graph % #2: base vertex % #3: tokens to execute % #4: edge key (not used) % #5: edge-triple {from}{to}{value} \__graph_map_successors_tokens_aux:Nnnnnn #1 {#2} {#3} #5 } \cs_new:Nn \__graph_map_successors_tokens_aux:Nnnnnn { % #1: the graph % #2: base vertex % #3: tokens to execute % #4: edge 'from' vertex % #5: edge 'to' vertex % #6: ptr to edge value (not used) \str_if_eq:nnT {#2} {#4} { \__graph_map_successors_tokens_aux:nnv {#3} {#5} {\prop_get:cn{\__graph_tl:nnn{graph}{#1}{vertices}}{#5}} } } \cs_new:Nn \__graph_map_successors_tokens_aux:nnn { % #1: tokens to execute % #2: successor key % #3: successor value #1 {#2} {#3} } \cs_generate_variant:Nn \__graph_map_successors_tokens_aux:nnn {nnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#3| to the key/value pairs % of the vertices reachable from vertex |#2| in one step. % The function is % supplied with two arguments as trailing brace groups. % % \begin{macrocode} \cs_new:Nn \graph_map_successors_function:NnN { % #1: graph % #2: base vertex % #3: function to execute \prop_map_tokens:cn { \__graph_tl:nnn{graph}{#1}{edge-triples} } { \__graph_map_successors_function_aux:NnNnn #1 {#2} #3 } } \cs_new:Nn \__graph_map_successors_function_aux:NnNnn { % #1: the graph % #2: base vertex % #3: function to execute % #4: edge key (not used) % #5: edge-triple {from}{to}{value} \__graph_map_successors_function_aux:NnNnnn #1 {#2} #3 #5 } \cs_new:Nn \__graph_map_successors_function_aux:NnNnnn { % #1: the graph % #2: base vertex % #3: function to execute % #4: edge 'from' vertex % #5: edge 'to' vertex % #6: ptr to edge value (not used) \str_if_eq:nnT {#2} {#4} { \__graph_map_successors_function_aux:Nnv #3 {#5} {\prop_get:cn{\__graph_tl:nnn{graph}{#1}{vertices}}{#5}} } } \cs_new:Nn \__graph_map_successors_function_aux:Nnn { % #1: function to execute % #2: successor key % #3: successor value #1 {#2} {#3} } \cs_generate_variant:Nn \__graph_map_successors_function_aux:Nnn {Nnv} % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the inline function |#3| to the key/value pairs % of the vertices reachable from vertex |#2| in one step. % The inline function is % supplied with two arguments: `|#1|' is the key, and `|#2|' % is the value of the successor vertex. % % \begin{macrocode} \cs_new_protected:Nn \graph_map_successors_inline:Nnn { % #1: graph % #2: base vertex % #3: body to execute \withargs (c) [\uniquecsname] [#2] [#3] { \cs_set:Npn ##1 ####1####2####3 {##3} \graph_map_successors_function:NnN #1 {##2} ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the tokens |#2| to all vertex name/value pairs in % topological order. The tokens are supplied with two % arguments as trailing brace groups. % Assumes that the graph is acyclic (for now). % % \begin{macrocode} \cs_new_protected:Nn \graph_map_topological_order_tokens:Nn { %%% Fill \l__graph_source_vertices with source-nodes and count indegrees % \prop_gclear_new:c {l__graph_source_vertices_(\int_use:N\g__graph_nesting_depth_int)_prop} \prop_gclear_new:c {l__graph_tmp_indeg_(\int_use:N\g__graph_nesting_depth_int)_prop} \graph_map_vertices_inline:Nn #1 { \prop_put:cnf {l__graph_tmp_indeg_(\int_use:N\g__graph_nesting_depth_int)_prop} {##1} { \graph_get_indegree:Nn #1 {##1} } \int_compare:nT {\graph_get_indegree:Nn #1 {##1} = 0} { \prop_put:cnn {l__graph_source_vertices_(\int_use:N\g__graph_nesting_depth_int)_prop} {##1} {} } } %%% Main loop % \bool_until_do:nn {\prop_if_empty_p:c {l__graph_source_vertices_(\int_use:N\g__graph_nesting_depth_int)_prop}} { %%% Choose any vertex (\l__graph_topo_key_tl, \l__graph_topo_value_tl) % \__graph_prop_any_key_pop:cN {l__graph_source_vertices_(\int_use:N\g__graph_nesting_depth_int)_prop} \l__graph_topo_key_tl \graph_get_vertex:NVNT #1 \l__graph_topo_key_tl \l__graph_topo_val_tl { %%% Deduct one from the counter of all affected nodes %%% and add all now-empty vertices to source_vertices % \graph_map_outgoing_edges_inline:NVn #1 \l__graph_topo_key_tl { \prop_put:cnf {l__graph_tmp_indeg_(\int_use:N\g__graph_nesting_depth_int)_prop} {##2} {\int_eval:n {\prop_item:cn {l__graph_tmp_indeg_(\int_use:N\g__graph_nesting_depth_int)_prop} {##2} - 1}} \int_compare:nT {\prop_item:cn {l__graph_tmp_indeg_(\int_use:N\g__graph_nesting_depth_int)_prop} {##2} = 0} { \prop_put:cnn {l__graph_source_vertices_(\int_use:N\g__graph_nesting_depth_int)_prop} {##2} {} } } %%% Run the mapping funtion on the key and value from that vertex %%% and manage the nesting depth counter % \int_gincr:N \g__graph_nesting_depth_int \withargs:VVn \l__graph_topo_key_tl \l__graph_topo_val_tl { #2 {##1} {##2} } \int_gdecr:N \g__graph_nesting_depth_int } } } \cs_new_protected:Nn \__graph_prop_any_key_pop:NN { \prop_map_inline:Nn #1 { \tl_set:Nn #2 {##1} \prop_remove:Nn #1 {##1} \prop_map_break:n {\use_none:nnn} } \tl_set:Nn #2 {\q_no_value} % TODO: test } \cs_generate_variant:Nn \__graph_prop_any_key_pop:NN {cN} \cs_generate_variant:Nn \withargs:nnn {VVn} \cs_generate_variant:Nn \graph_map_outgoing_edges_inline:Nnn {NVn} \cs_generate_variant:Nn \prop_put:Nnn {cnf} \cs_generate_variant:Nn \graph_get_vertex:NnNT {NVNT} \tl_new:N \l__graph_topo_key_tl \tl_new:N \l__graph_topo_val_tl \int_new:N \g__graph_nesting_depth_int % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the function |#2| to all vertex name/value pairs in % topological order. The function is supplied with two % arguments as trailing brace groups. % Assumes that the graph is acyclic (for now). % % \begin{macrocode} \cs_new:Nn \graph_map_topological_order_function:NN { \graph_map_topological_order_tokens:Nn #1 {#2} } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Applies the inline function |#2| to all vertex name/value % pairs in topological order. The inline function is supplied % with two arguments: `|#1|' for the name and `|#2|' for the value. % Assumes that the graph is acyclic (for now). % % \begin{macrocode} \cs_new_protected:Nn \graph_map_topological_order_inline:Nn { \withargs (c) [\uniquecsname] [#2] { \cs_set:Npn ##1 ####1####2 {##2} \graph_map_topological_order_function:NN #1 ##1 } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Transforming Graphs} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Set graph |#1| to the transitive closure of graph |#2|. % % \begin{macrocode} \cs_new_protected:Nn \graph_set_transitive_closure:NN { \__graph_set_transitive_closure:NNNnn #1 #2 \use_none:nnn {} { } } \cs_new_protected:Nn \graph_gset_transitive_closure:NN { \__graph_set_transitive_closure:NNNnn #1 #2 \use_none:nnn {} {g} } \cs_new_protected:Nn \graph_set_transitive_closure:NNNn { \__graph_set_transitive_closure:NNNnn #1 #2 #3 {#4} { } } \cs_new_protected:Nn \graph_gset_transitive_closure:NNNn { \__graph_set_transitive_closure:NNNnn #1 #2 #3 {#4} {g} } \cs_new_protected:Nn \__graph_set_transitive_closure:NNNnn { % #1: target % #2: source % #3: combination function with argspec :nnn % #4: default 'old' value \use:c{graph_#5set_eq:NN} #1 #2 \cs_set:Nn \__graph_edge_combinator:nnn { \exp_not:n { #3 {##1} {##2} {##3} } } \cs_generate_variant:Nn \__graph_edge_combinator:nnn {VVV} \graph_map_vertices_inline:Nn #2 { \graph_map_vertices_inline:Nn #2 { \graph_get_edge:NnnNT #2 {##1} {####1} \l__graph_edge_value_i_tl { \graph_map_vertices_inline:Nn #2 { \graph_get_edge:NnnNT #2 {####1} {########1} \l__graph_edge_value_ii_tl { \graph_get_edge:NnnNF #1 {##1} {########1} \l__graph_edge_value_old_tl { \tl_set:Nn \l__graph_edge_value_old_tl {#4} } \exp_args:NNx \tl_set:No \l__graph_edge_value_new_tl { \__graph_edge_combinator:VVV \l__graph_edge_value_i_tl \l__graph_edge_value_ii_tl \l__graph_edge_value_old_tl } \use:c{graph_#5put_edge:NnnV} #1 {##1} {########1} \l__graph_edge_value_new_tl } } } } } } \cs_generate_variant:Nn \graph_put_edge:Nnnn {NnnV} \cs_generate_variant:Nn \graph_gput_edge:Nnnn {NnnV} \cs_generate_variant:Nn \tl_to_str:n {o} \tl_new:N \l__graph_edge_value_i_tl \tl_new:N \l__graph_edge_value_ii_tl \tl_new:N \l__graph_edge_value_old_tl \tl_new:N \l__graph_edge_value_new_tl % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Assume that graph |#2| contains no cycles, and % set graph |#1| to its transitive reduction. % % \begin{macrocode} \cs_new_protected:Nn \graph_set_transitive_reduction:NN { \__graph_set_transitive_reduction:NNn #1 #2 { } } \cs_new_protected:Nn \graph_gset_transitive_reduction:NN { \__graph_set_transitive_reduction:NNn #1 #2 {g} } \cs_new_protected:Nn \__graph_set_transitive_reduction:NNn { % #1: target % #2: source \use:c{graph_#3set_eq:NN} #1 #2 \graph_map_vertices_inline:Nn #2 { \graph_map_vertices_inline:Nn #2 { \graph_get_edge:NnnNT #2 {##1} {####1} \l_tmpa_tl { \graph_map_vertices_inline:Nn #2 { \graph_get_edge:NnnNT #2 {####1} {########1} \l_tmpa_tl { \use:c{graph_#3remove_edge:Nnn} #1 {##1} {########1} } } } } } } % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \subsection{Displaying Graphs} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % We define some additional % functions that can display the graph in table-form. % This is the option-less version, which delegates % to the full version: % % \begin{macrocode} \cs_new_protected:Nn \graph_display_table:N { \graph_display_table:Nn #1 {} } % \end{macrocode} % % The full version has a second argument accepting options % that determine table formatting. We first define those options. % Please note that with the standard options, the |xcolor| package % is required with the |table| option, because of our use of the % |\cellcolor| command. % % \begin{macrocode} \keys_define:nn {lt3graph-display} { row_keys .bool_set:N = \l__graph_display_row_keys_bool, row_keys .initial:n = {true}, row_keys .default:n = {true}, vertex_vals .bool_set:N = \l__graph_display_vertex_vals_bool, vertex_vals .initial:n = {true}, vertex_vals .default:n = {true}, row_keys_format .tl_set:N = \l__graph_format_row_keys_tl, row_keys_format .initial:n = \textbf, row_keys_format .value_required:n = true, col_keys_format .tl_set:N = \l__graph_format_col_keys_tl, col_keys_format .initial:n = \textbf, col_keys_format .value_required:n = true, vertex_vals_format .tl_set:N = \l__graph_format_vertex_vals_tl, vertex_vals_format .initial:n = \use:n, vertex_vals_format .value_required:n = true, edge_vals_format .tl_set:N = \l__graph_format_edge_vals_tl, edge_vals_format .initial:n = \use:n, edge_vals_format .value_required:n = true, edge_diagonal_format .tl_set:N = \l__graph_format_edge_diagonal_tl, edge_diagonal_format .initial:n = \cellcolor{black!30!white}, edge_diagonal_format .value_required:n = true, edge_direct_format .tl_set:N = \l__graph_format_edge_direct_tl, edge_direct_format .initial:n = \cellcolor{green}, edge_direct_format .value_required:n = true, edge_transitive_format .tl_set:N = \l__graph_format_edge_transitive_tl, edge_transitive_format .initial:n = \cellcolor{green!40!yellow}\tiny(tr), edge_transitive_format .value_required:n = true, edge_none_format .tl_set:N = \l__graph_format_edge_none_tl, edge_none_format .initial:n = {}, edge_none_format .value_required:n = true } % \end{macrocode} % % Now we define the function itself. % It displays a table showing the structure and content % of graph |#1|. If argument |#2| is passed, its options % are applied to format the output. % % \begin{macrocode} \cs_new_protected:Nn \graph_display_table:Nn { \group_begin: % \end{macrocode} % % We process those options passed with |#2|: % % \begin{macrocode} \keys_set:nn {lt3graph-display} {#2} % \end{macrocode} % % We populate the top row of the table: % % \begin{macrocode} \tl_put_right:Nn \l__graph_table_content_tl {\hline} \seq_clear:N \l__graph_row_seq \bool_if:NT \l__graph_display_row_keys_bool { \seq_put_right:Nn \l__graph_row_seq {} \tl_put_right:Nn \l__graph_table_colspec_tl {|r|} } \bool_if:NT \l__graph_display_vertex_vals_bool { \seq_put_right:Nn \l__graph_row_seq {} \tl_put_right:Nn \l__graph_table_colspec_tl {|c|} } \graph_map_vertices_inline:Nn #1 { \tl_put_right:Nn \l__graph_table_colspec_tl {|c} \seq_put_right:Nn \l__graph_row_seq { { \l__graph_format_col_keys_tl {##1} } } } \tl_put_right:Nn \l__graph_table_colspec_tl {|} \tl_put_right:Nx \l__graph_table_content_tl { \seq_use:Nn \l__graph_row_seq {&} } \tl_put_right:Nn \l__graph_table_content_tl { \\\hline\hline } % \end{macrocode} % % We populate the remaining rows: % % \begin{macrocode} \graph_map_vertices_inline:Nn #1 { \seq_clear:N \l__graph_row_seq \bool_if:NT \l__graph_display_row_keys_bool { \seq_put_right:Nn \l__graph_row_seq { { \l__graph_format_row_keys_tl {##1} } } } \bool_if:NT \l__graph_display_vertex_vals_bool { \seq_put_right:Nn \l__graph_row_seq { { \l__graph_format_vertex_vals_tl {##2} } } } \graph_map_vertices_inline:Nn #1 { % \end{macrocode} % % We start building the vertex cell value. First we distinguish % between a direct connection, a transitive connection, % and no connection, and format accordingly: % % \begin{macrocode} \graph_get_edge:NnnNTF #1 {##1} {####1} \l_tmpa_tl { \quark_if_no_value:VF \l_tmpa_tl { \tl_set_eq:NN \l__graph_cell_content_tl \l_tmpa_tl \tl_set:Nf \l__graph_cell_content_tl { \exp_args:NV \l__graph_format_edge_direct_tl \l__graph_cell_content_tl } } }{\graph_acyclic_if_path_exist:NnnTF #1 {##1} {####1} { \tl_set_eq:NN \l__graph_cell_content_tl \l__graph_format_edge_transitive_tl }{ \tl_set_eq:NN \l__graph_cell_content_tl \l__graph_format_edge_none_tl }} % \end{macrocode} % % Secondary formatting comes from cells on the diagonal, % i.e., a key compared to itself: % % \begin{macrocode} \str_if_eq:nnT {##1} {####1} { \tl_set:Nf \l__graph_cell_content_tl { \exp_args:NV \l__graph_format_edge_diagonal_tl \l__graph_cell_content_tl } } % \end{macrocode} % % Tertiary formatting is applied to all vertex value cells: % % \begin{macrocode} \tl_set:Nf \l__graph_cell_content_tl { \exp_args:NV \l__graph_format_edge_vals_tl \l__graph_cell_content_tl } % \end{macrocode} % % We can now add the cell to the row sequence: % % \begin{macrocode} \seq_put_right:NV \l__graph_row_seq \l__graph_cell_content_tl % \end{macrocode} % \uninteresting\begin{macrocode} } % \end{macrocode} % % We are finished with this row; go on to the next iteration: % % \begin{macrocode} \tl_put_right:Nx \l__graph_table_content_tl { \seq_use:Nn \l__graph_row_seq {&} } \tl_put_right:Nn \l__graph_table_content_tl {\\\hline} % \end{macrocode} % \uninteresting\begin{macrocode} } % \end{macrocode} % % Finally, we print the table itself: % % \begin{macrocode} \withargs:VVn \l__graph_table_colspec_tl \l__graph_table_content_tl { \begin{tabular}{##1}##2\end{tabular} } % \end{macrocode} % \uninteresting\begin{macrocode} \group_end: } % \end{macrocode} % % Now follow the local variants and variables % used in the function: % % \begin{macrocode} \cs_generate_variant:Nn \quark_if_no_value:nF {VF} \cs_generate_variant:Nn \withargs:nnn {VVn} \tl_new:N \l__graph_table_colspec_tl \tl_new:N \l__graph_table_content_tl \tl_new:N \l__graph_cell_content_tl \bool_new:N \l__graph_table_skipfirst_bool \seq_new:N \l__graph_row_seq % \end{macrocode} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Selection Method} As per the approach that did not incorporate modularization, tournament selection \cite{miller1995genetic} was used as the selection method for all genetic operators. To expand \emph{T} individuals are selected entirely at random from the existing population. Then, from this set of individuals (ie. the \emph{tournament}) the fittest individual is selected. The size, \emph{T}, of the tournament was 48. If genetic operators required more than one individual from the existing population, the method was reapplied to select each individual.1-10 \usepackage{charter} %Charter fuer englische Texte %\linespread{1.05} % Durchschuss für Charter leicht erhöhen \usepackage[utf8]{inputenc} \usepackage{pmboxdraw} % utf8x is incompatible with biblatex, the following trick is a remedy % source: https://tex.stackexchange.com/a/213177 \input{binhex} \makeatletter \def\uc@dclc#1#2#3{% \ifnum\pdfstrcmp{#2}{mathletters}=\z@ \begingroup\edef\x{\endgroup \noexpand\DeclareUnicodeCharacter{\hex{#1}}}\x{#3}% \fi } \input{uni-3.def} \def\uc@dclc#1#2#3{% \ifnum\pdfstrcmp{#2}{default}=\z@ \begingroup\edef\x{\endgroup \noexpand\DeclareUnicodeCharacter{\hex{#1}}}\x{#3}% \fi } \input{uni-34.def} \makeatother \usepackage[tbtags,sumlimits,intlimits,namelimits]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bbm} \usepackage{ulem} \usepackage{tikz} \usepackage{pgf} \usepackage{ifpdf} \usepackage{color} \usepackage{esint} \usepackage{framed} \usepackage{wasysym} % \ocircle \usepackage{makecell} % \makecell{...\\...} to break in table cell \usepackage{textcomp} \usepackage{fancyvrb} % improved verbatim %\usepackage[colorlinks=true,linkcolor=black,citecolor=black,urlcolor=black]{hyperref} % print %\usepackage[colorlinks=true,linkcolor=blue,citecolor=blue,pdfpagelabels,plainpages=false]{hyperref} % web %\usepackage{hyperref} % plain \usepackage[top=2.3cm, bottom=3.45cm, left=2.3cm, right=2.3cm]{geometry} %\numberwithin{equation}{section} \usepackage{chngcntr} \counterwithin*{section}{part} %\graphicspath{{images/png/}{images/}} % Pfad, in dem sich Grafikdateien befinden %\usepackage{subfigure} % Unterbilder, deprecated %\usepackage(subfig} %\usepackage{cite} % Literatur \usepackage{graphicx} % Bilder in Tabellen \usepackage{float} % eigene Float-Umgebungen, H-Option, um Bilder an der aktuellen Stelle anzuzeigen \usepackage{caption} \usepackage{subcaption,array} %\usepackage{array} \restylefloat{figure} % Bilder an der Stelle, wo sie eingebunden werden \usepackage{multirow} \usepackage{listings} % Darstellung von Source-Code \usepackage{framed} % Rahmen um Text \usepackage{arydshln} % gestrichelte Linie in Tabelle mit \hdashline \usepackage{longtable} % Tabellen mit automatischen Seitenumbruch \usepackage{layouts} % textwidth in cm: \printinunitsof{cm}\prntlen{\textwidth} %\usepackage{ziffer} % Komma in Dezimalzahlen \usepackage{url} % Links %% end old \definecolor{darkblue}{rgb}{0,0,.5} \definecolor{black}{rgb}{0,0,0} \usepackage[framemethod=TikZ]{mdframed} % Rahmen um Text und Gleichungen %\usepackage{arydshln} % gestrichelte Linie in Tabelle mit \hdashline \usepackage{dirtytalk} % \say{...} erzeugt (deutsche) Anführungszeichen \usepackage{tipa} \usepackage{transparent} % needed for inkscape generated pdf_tex files \usepackage{multicol} % multiple columns \usepackage{moreverb} % verbatimwrite \usepackage{verbatimbox} % \begin{verbbox} \usepackage{booktabs} \usepackage{morefloats} % Increase the number of simultaneous LaTeX floats \usepackage{enumitem} % more control oover itemize %\usepackage{algorithm2e} %\usepackage{algorithmic} %\usepackage{algpseudocode} \newsavebox\lstbox \mdfdefinestyle{MyFrame}{% innertopmargin=0pt, innerbottommargin=10pt, innerrightmargin=20pt, innerleftmargin=20pt} \definecolor{darkgreen}{HTML}{009900} % settings for algorithm \lstset{literate=% {Ö}{{\"O}}1 {Ä}{{\"A}}1 {Ü}{{\"U}}1 {ß}{{\ss}}1 {ü}{{\"u}}1 {ä}{{\"a}}1 {ö}{{\"o}}1 {⇐}{{$\leftarrow$}}1 {>=}{{$\geq$}}1 {~}{{\textasciitilde}}1 {`}{\textquotedbl}1, language=C++, numbers=none, numberstyle=\tiny, xleftmargin=2.0ex, %basicstyle=\small, % print whole listing small basicstyle=\small\ttfamily, morekeywords={elif,do,end,then,proc,local,Eingabe,Ausgabe,alignof,loop,each}, deletekeywords={new}, columns=flexible, % alignment tabsize=2, % size of tabs keepspaces, gobble=2, % remove 2 characters at begin of each line mathescape % wandle $$ in latex um } % Versuche stärker, Abbildungen dort einzubinden, wo sie definiert wurden \renewcommand{\topfraction}{.85} % Anteil, den floats auf einer Seite von oben her einnehmen dürfen \renewcommand{\bottomfraction}{.7} % Anteil, den floats auf einer Seite von unten her einnehmen dürfen \renewcommand{\textfraction}{.15} % Anteil der Seite, der mind. für Text zur Verfügung steht \renewcommand{\floatpagefraction}{.66} % Anteil der Seite, der belegt sein muss, bevor eine weitere Seite angelegt wird \setcounter{topnumber}{9} % maximale Anzahl floats, die im oberen Bereich der Seite sein dürfen \setcounter{bottomnumber}{9} % maximale Anzahl floats, die im unteren Bereich der Seite sein dürfen % language-specific things (hyphenation, ...) \usepackage[ngerman,main=american]{babel} % biblatex package wants to have csquotes (otherwise: warning) \usepackage{csquotes} % lower-case page header \usepackage[markcase=upper]{scrlayer-scrpage} % dummy texts \usepackage[math]{blindtext} % amsmath with improvements \usepackage{mathtools} % math theorems %\usepackage{amsthm} % format math theorems \usepackage{thmtools} % for setting size of text area \usepackage{geometry} % for \luadirect command %\usepackage{luacode} % check mode %\iftoggle{checkMode}{ % show non-whitelisted hyphenations % \usepackage[mark]{lua-check-hyphen} %}{} % debug mode %\iftoggle{debugMode}{ % % show boxes, glues, and kerning % \usepackage{lua-visual-debug} %}{} % use GoMono as typewriter font \usepackage[ % otherwise, bold is not available ("undefined font shape" warnings) type1, % scale to match main font size scale=0.88, ]{GoMono} % use TeX Gyre Heros as sans-serif font \usepackage{tgheros} % need T1 font encoding for Charter, % otherwise there will be "undefined font shape" warnings \usepackage[T1]{fontenc} % use Bitstream Charter as main font \usepackage[bitstream-charter]{mathdesign} % do not use mathcal of mathdesign, but replace by standard mathcal \DeclareSymbolFont{usualmathcal}{OMS}{cmsy}{m}{n} \DeclareSymbolFontAlphabet{\mathcal}{usualmathcal} % micro-typographic adjustments \usepackage[ % don't deactivate extensions due to document class draft option final, % increase letter spacing in small-caps tracking=smallcaps, % font expansion: don't stretch/shrink lines by more than 1%, % default of 2% looks a bit weird stretch=10, shrink=10, ]{microtype} % reference last page of document \usepackage{lastpage} % extract number from page reference \usepackage{refcount} % graphics \usepackage{graphicx} % colors, load colortbl for coloring table rows %\usepackage[table]{xcolor} % contours around text \usepackage{contour} % additional space in table cells \usepackage{cellspace} % custom float captions, subfigures \usepackage{subcaption} % allows to put float caption beside image \usepackage{sidecap} % L, C, R column types for tables \usepackage{array} % X column type for tables (filling rest of line) \usepackage{tabularx} % more beautiful tables \usepackage{booktabs} % table cells spanning multiple rows \usepackage{multirow} % for controlling space in enumerate and itemize \usepackage{enumitem} % disable "Column type for cellspace package moved to 'C'." warning % when loading siunitx % (this is due to the cellspace package loaded above) \usepackage{expl3} \ExplSyntaxOn \msg_redirect_name:nnn{siunitx}{moved-cellspace-column}{none} \ExplSyntaxOff % consistent style of units and numbers \usepackage[binary-units=true]{siunitx} % watermarks \usepackage{everypage} % rules left of theorems %\usepackage[framemethod=tikz]{mdframed} % drawings \usepackage{tikz} % styling of table of contents \usepackage{tocbasic} % local table of contents (mini TOC) \usepackage{etoc} % wrap text around boxes \usepackage{wrapfig} % initials/dropped captial letters at beginning of chapters \usepackage{lettrine} % inclusion of external PDFs \usepackage{pdfpages} % reference current chapter, section, subsection, ... \usepackage{nameref} % line spacing (one and a half lines) \usepackage[onehalfspacing]{setspace} % filter specific warnings %\usepackage{silence} % patch hard-coded stuff in packages \usepackage{regexpatch} % allow breaking after hyphens in long URLs %\usepackage[hyphens]{url} % PDF links \usepackage{hyperref} \usepackage[all]{hypcap} % \cref verbessert \usepackage[ % capitalize names automatically capitalise, % mark all of "Lemma 1" as a link, not only "1" nameinlink, % pass language used in document to enable cleveref in otherlanguage env ngerman,english, ]{cleveref} % generate Eqs. instead of Equations \crefmultiformat{equation}{#2Eqs.~(#1)#3}{ and~#2(#1)#3}{, #2(#1)#3}{ and~#2(#1)#3} %\crefrangemultiformat{equation}{#2Eqs.~(#1)#3}{ and~#2(#1)#3}{, #2(#1)#3}{ and~#2(#1)#3} %\crefrangemultiformat{equation} % {Eqs. #3(#1)#4 to #5(#2)#6} % { and #3(#1)#4 to #5(#2)#6} % {, #3(#1)#4 to #5(#2)#6} % { and #3(#1)#4 to #5(#2)#6} % algorithms \usepackage{algorithm} \usepackage{algorithmicx} % pseudo-code \usepackage[noend]{algpseudocode} % create custom PDF bookmarks \usepackage{bookmark} % old formatting using the titlesec package, which is now obsolete %\usepackage{titlesec} % space before chapter head %\titleformat{\chapter}[display]{\normalfont\huge\bfseries}{\chaptertitlename\ \thechapter}{20pt}{\Huge} \renewcommand*{\chapterformat}{\normalfont\large\bfseries Chapter~\thechapter\autodot\enskip} %Chapter~\fontsize{60}{68}\selectfont\color{gray} \thechapter\autodot\enskip} \renewcommand\chapterlinesformat[3]{\ifstr{#2}{}{}{#2\vspace*{5mm}\\*}{\Huge #3}} % analogous to \cleardoublepage but for left page \newcommand*\cleartoleftpage{% \clearpage \ifodd\value{page}\hbox{}\newpage\fi } % BibLaTeX \usepackage[ % abbreviate author names giveninits=true, % only show years in dates date=year, % use alphabetic style instead of numeric style=alphabetic, % only use first author's name for the style maxalphanames=1, ]{biblatex} %\usepackage{scrhack} % named references %\usepackage[ % % capitalize names automatically % capitalise, % % mark all of "Lemma 1" as a link, not only "1" % nameinlink, % % pass language used in document to enable cleveref in otherlanguage env % ngerman,english, %]{cleveref} @inproceedings{Harris2014:TFRs-SemDial, Address = {Edinburgh, Scotland}, Author = {}, Booktitle = {The Proceedings of the 18th Workshop on the Semantics and Pragmatics of Dialogue}, Editor = { and }, Keywords = {commitments, perspective}, Pages = {71--79}, Title = {Signaling Non-Speaker commitment in Transparent Free Relatives: A paired Speaker-Hearer judgment study.}, Year = {2014}} slliac/Ljus4Food \hypertarget{vendor_2mikey179_2vfs_stream_2readme_8md}{}\section{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/mikey179/vfs\+Stream/readme.md File Reference} \label{vendor_2mikey179_2vfs_stream_2readme_8md}\index{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/mikey179/vfs\+Stream/readme.\+md@{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/mikey179/vfs\+Stream/readme.\+md}} 10-100 @Article{BanksSchulpenBrockington:2020, Author = { and and }, Journal = {Development in Practice}, Number = {6}, Pages = {695--705}, Title = {New Sectoral Perspectives on International {NGO}s: Scale, Dynamics, and Influences}, Volume = {30}, Year = {2020}, Doi = {10.1080/09614524.2020.1801595} } @Article{bekkers2003trust, Author = {{\'e}}, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {12}, Number = {4}, Pages = {596--615}, Title = {Trust, Accreditation, and Philanthropy in the {Netherlands}}, Volume = {32}, Year = {2003}, Doi = {10.1177/0899764003258102} } @Article{BekkersWiepking:2011, Author = {} and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Number = {5}, Pages = {924--973}, Title = {A Literature Review of Empirical Studies of Philanthropy: Eight Mechanisms That Drive Charitable Giving}, Volume = {40}, Year = {2011}, Doi = {10.1177/0899764010380927} } @Article{BrownFerris2007, Author = { .}, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {3}, Number = {1}, Pages = {85--99}, Title = {Social Capital and Philanthropy: An Analysis of the Impact of Social Capital on Individual Giving and Volunteering}, Volume = {36}, Year = {2007}, Doi = {10.1177/0899764006293178} } @Article{CasaleBaumann:2015, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Number = {1}, Pages = {98--122}, Title = {Who Gives to International Causes? {A} Sociodemographic Analysis of {US} Donors}, Volume = {44}, Year = {2015}, Doi = {10.1177/0899764013507141} } @Article{ChaudhryHeiss:2021, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Title = {Dynamics of International Giving: How Heuristics Shape Individual Donor Preferences}, Year = {2021}, Doi = {10.1177/0899764020971045} } @Article{chong2007framing, Author = { and .}, Journal = {American Political Science Review}, Month = {11}, Number = {4}, Pages = {637--655}, Title = {Framing Public Opinion in Competitive Democracies}, Volume = {101}, Year = {2007}, Doi = {10.1017/s0003055407070554} } @Article{Cooley:2002, Author = { and }, Journal = {International Security}, Number = {1}, Pages = {5--39}, Title = {The {NGO} Scramble: Organizational Insecurity and the Political Economy of Transnational Action}, Volume = {27}, Year = {2002}, Doi = {10.1162/016228802320231217}, Issue = {Summer} } @Article{coombs2007, Author = {}, Journal = {Corporate Reputation Review}, Month = {9}, Number = {3}, Pages = {163--176}, Title = {Protecting Organization Reputations During a Crisis: The Development and Application of Situational Crisis Communication Theory}, Volume = {10}, Year = {2007}, Doi = {10.1057/palgrave.crr.1550049} } @Article{desaikharas, Author = {. and }, Journal = {International Studies Quarterly}, Month = {9}, Number = {3}, Pages = {505--519}, Title = {What Motivates Private Foreign Aid? Evidence from Internet-Based Microlending}, Volume = {62}, Year = {2018}, Doi = {10.1093/isq/sqy023} } @Article{Dietrich2013, Author = {}, Journal = {International Studies Quarterly}, Month = {12}, Number = {4}, Pages = {698--712}, Title = {Bypass or Engage? Explaining Donor Delivery Tactics in Foreign Aid Allocation}, Volume = {57}, Year = {2013}, Doi = {10.1111/isqu.12041} } @Article{DPR:2015, Author = { and and }, Journal = {Review of International Political Economy}, Number = {2}, Pages = {419--56}, Title = {Who Survived? {Ethiopia}'s Regulatory Crackdown on Foreign-Funded {NGO}s}, Volume = {22}, Year = {2015}, Doi = {10.1080/09692290.2014.903854} } @Article{DPR:2016, Author = { and and }, Journal = {World Development}, Month = {8}, Pages = {299--311}, Title = {Hands Off My Regime! Governments' Restrictions on Foreign Aid to Non-Governmental Organizations in Poor and Middle-Income Countries}, Volume = {84}, Year = {2016}, Doi = {10.1016/j.worlddev.2016.02.001} } @Article{DupuyPrakash:2018, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {2}, Number = {1}, Pages = {89--106}, Title = {Do Donors Reduce Bilateral Aid to Countries with Restrictive {NGO} Laws? {A} Panel Study, 1993--2012}, Volume = {47}, Year = {2018}, Doi = {10.1177/0899764017737384} } @Article{dwyer1987, Author = { and . and }, Journal = {Journal of Marketing}, Number = {2}, Pages = {11--27}, Title = {Developing Buyer-Seller Relationships}, Volume = {51}, Year = {1987}, Doi = {10.2307/1251126} } @Article{FaulknerTruongRomaniuk:2015, Author = { and and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {10}, Number = {5}, Pages = {1007--1025}, Title = {Barriers to Increasing Donor Support: Evidence on the Incidence and Nature of Brand Rejection}, Volume = {44}, Year = {2015}, Doi = {10.1177/0899764014555985} } @Article{givingusa:2019, Author = {{Giving USA}}, Title = {Giving USA 2019: Americans gave \$427.71 billion to charity in 2018 amid complex year for charitable giving}, Url = {https://givingusa.org/giving-usa-2019-americans-gave-427-71-billion-to-charity-in-2018-amid-complex-year-for-charitable-giving/}, Date = {2019-06-18} } @Article{greenhill2013, Author = { and and }, Journal = {ODI Centre for Aid \& Public Expenditure}, Title = {How are Developing Countries Managing the New Aid Landscape}, Year = {2013} } @Article{hossainlamb2012, Author = { and }, Journal = {Economic Papers: A Journal of Applied Economics and Policy}, Month = {6}, Number = {2}, Pages = {265--273}, Title = {The Dynamics of Environmental Giving in {Canada}: Evidence of Rising Demand for Environmental Quality?}, Volume = {31}, Year = {2012}, Doi = {10.1111/j.1759-3441.2012.00177.x} } @Article{hossainlamb2017, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {10}, Number = {5}, Pages = {963--983}, Title = {Associational Capital and Adult Charitable Giving: A {Canadian} Examination}, Volume = {46}, Year = {2017}, Doi = {10.1177/0899764017703707} } @Article{KeatingThrandardottir:2017, Author = { and }, Journal = {The British Journal of Politics and International Relations}, Month = {2}, Number = {1}, Pages = {134--151}, Title = {{NGO}s, Trust, and the Accountability Agenda}, Volume = {19}, Year = {2017}, Doi = {10.1177/1369148116682655} } @Article{KnowlesSullivan:2017, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {4}, Number = {5}, Pages = {944--62}, Title = {Does Charity Begin at Home or Overseas?}, Volume = {46}, Year = {2017}, Doi = {10.1177/0899764017703710} } @Article{KnudsenJohannesson:2019, Author = { and }, Journal = {Political Communication}, Number = {2}, Pages = {259--71}, Title = {Beyond the Limits of Survey Experiments: How Conjoint Designs Advance Causal Inference in Political Communication Research}, Volume = {36}, Year = {2019}, Doi = {10.1080/10584609.2018.1493009} } @Article{McGill:2018, Author = {}, Journal = {PND by Candid}, Title = {Current Trends in Philanthropy: International Giving by {U.S.} Foundations}, Year = {2018}, Url = {https://pndblog.typepad.com/pndblog/2018/11/current-trends-in-philanthropy-international-giving-by-us-foundations.html} } @Article{Micklewright:2009, Author = { and }, Journal = {Journal of Social Policy}, Month = {4}, Number = {2}, Pages = {317-41}, Title = {Who Gives for Overseas Development?}, Volume = {38}, Year = {2009}, Doi = {10.1017/s0047279408002869} } @Article{Milner:2017, Author = {}, Journal = {Alliance Magazine}, Title = {The State of {European} Philanthropy}, Year = {2017}, Url = {https://www.alliancemagazine.org/opinion/state-european-philanthropy/} } @Article{Phillips:2019, Author = {}, Journal = {Nonprofit Policy Forum}, Month = {11}, Number = {4}, Pages = {1--11}, Title = {Putting {Humpty} Together Again: How Reputation Regluation Failes the Charitable Sector}, Volume = {10}, Year = {2019}, Doi = {10.1515/npf-2019-0032} } @Article{prakash2010trust, Author = { }, Journal = {Regulation \& Governance}, Month = {2}, Number = {1}, Pages = {22--47}, Title = {Trust but Verify? Voluntary Regulation Programs in the Nonprofit Sector}, Volume = {4}, Year = {2010}, Doi = {10.1111/j.1748-5991.2009.01067.x} } @Article{Rajan:2009, Author = { and and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {6}, Number = {3}, Pages = {413--40}, Title = {Sociodemographic and Personality Characteristics of {Canadian} Donors Contributing to International Charity}, Volume = {38}, Year = {2009}, Doi = {10.1177/0899764008316056} } @Article{RobsonHart:2020, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Pages = {1--25}, Title = {Understanding the Correlates of Donor Intention: A Comparison of Local, National, and International Charity Destinations}, Year = {2020}, Doi = {10.1177/0899764020927097} } @Article{Roth:2016, Author = {}, Journal = {Foreign Policy}, Title = {The Great Cicil Society Choke-out}, Url = {https://foreignpolicy.com/2016/01/27/the-great-civil-society-choke-out-human-rights-democracy-india-russia-china-kenya/}, Date = {2016-01-27}, Entrysubtype = {magazine} } @Article{sargeant2008, Author = { and }, Journal = {International Journal of Nonprofit and Voluntary Sector Marketing}, Month = {2}, Number = {1}, Pages = {89--101}, Title = {Donor Retention: An Exploratory Study of Door-to-Door Recruits}, Volume = {13}, Year = {2008}, Doi = {10.1002/nvsm.301} } @Article{SaxtonWang:2014, Author = { and }, Journal = {Nonprofit and Voluntary Sector Quarterly}, Month = {10}, Number = {5}, Pages = {850--868}, Title = {The Social Network Effect: The Determinants of Giving Through Social Media}, Volume = {43}, Year = {2014}, Doi = {10.1177/0899764013485159} } @Article{ScurlockDolsakPrakash:2020, Author = { and and }, Journal = {Voluntas: International Journal of Voluntary and Nonprofit Organizations}, Month = {2}, Pages = {94--110}, Title = {Recovering from Scandals: {Twitter} Coverage of {Oxfam} and {Save the Children} Scandals}, Volume = {31}, Year = {2020}, Doi = {10.1007/s11266-019-00148-x} } @Article{SzperPrakash:2011, Author = { and }, Journal = {Voluntas: International Journal of Voluntary and Nonprofit Organizations}, Month = {3}, Number = {1}, Pages = {112--41}, Title = {Charity Watchdogs and the Limits of Information-Based Regulation}, Volume = {22}, Year = {2011}, Doi = {10.1007/s11266-010-9156-2} } @Article{Tarrow:2001, Author = {}, Journal = {Annual Review of Political Science}, Pages = {1--20}, Title = {Transnational Politics: Contention and Institutions in International Politics}, Volume = {4}, Year = {2001} } @Article{TBPrakash:2017, Author = {Tremblay-Boire, Prakash, Aseem}, Journal = {Voluntas: International Journal of Voluntary and Nonprofit Organizations}, Month = {4}, Number = {2}, Pages = {621--47}, Title = {Will You Trust Me?: How Individual {American} Donors Respond to Informational Signals Regarding Local and Global Humanitarian Charities}, Volume = {28}, Year = {2017}, Doi = {10.1007/s11266-016-9782-4} } @Article{Wiepking:2010, Author = {}, Journal = {Social Science Research}, Month = {11}, Number = {6}, Pages = {1073--87}, Title = {Democrats Support International Relief and the Upper Class Donates to Art? {H}ow Opportunity, Incentives and Confidence Affect Donations to Different Types of Charitable Organizations}, Volume = {39}, Year = {2010}, Doi = {10.1016/j.ssresearch.2010.06.005} } @Article{WiepkingBekkers:2010, Author = { and {\'e}}, Journal = {Voluntas: International Journal of Voluntary and Nonprofit Organizations}, Number = {2}, Pages = {240--263}, Title = {Does Who Decides Really Matter? Causes and Consequences of Personal Financial Management in the Case of Larger and Structural Charitable Donations}, Volume = {21}, Year = {2010}, Doi = {10.1007/s11266-010-9124-x} } @Article{wiepkingmaas, Author = { and }, Journal = {Social Forces}, Month = {6}, Number = {4}, Pages = {1973--1995}, Title = {Resources That Make You Generous: Effects of Social and Human Resources on Charitable Giving}, Volume = {87}, Year = {2009}, Doi = {10.1353/sof.0.0191} } @Article{youde2019, Author = {}, Journal = {Review of International Studies}, Month = {1}, Number = {1}, Pages = {39--56}, Title = {The Role of Philanthropy in International Relations}, Volume = {45}, Year = {2019}, Doi = {10.1017/s0260210518000220} } @Book{Brechenmacher:2017, Address = {Washington, D. C.}, Author = {}, Publisher = {Carnegie Endowment for International Peace}, Title = {Civil Society Under Assault: Repression and Response in {Russia}, {Egypt}, and {Ethiopia}}, Year = {2017}, Url = {http://carnegieendowment.org/2017/05/18/civil-society-under-assault-repression-and-responses-in-russia-egypt-and-ethiopia-pub-69953} } @Book{Bush:2015, Address = {Cambridge, UK}, Author = {}, Publisher = {Cambridge University Press}, Title = {The Taming of Democracy Assistance: Why Democracy Promotion Does Not Confront Dictators}, Year = {2015}, Doi = {10.1017/cbo9781107706934} } @Book{CarothersBrechenmacher:2014, Address = {Washington, D. C.}, Author = { and }, Publisher = {Carnegie Endowment for International Peace}, Title = {Closing Space: Democracy and Human Rights Support Under Fire}, Year = {2014} } @Book{coleman1990, Address = {Cambridge, MA}, Author = {.}, Publisher = {Harvard University Press}, Title = {Foundations of Social Theory}, Year = {1990} } @Book{Stroup:2012, Address = {Ithaca, New York}, Author = {}, Publisher = {Cornell University Press}, Title = {Borders among Activists: International {NGOs} in the {United States}, {Britain}, and {France}}, Year = {2012}, Doi = {10.7591/9780801464256} } @Book{USTrust2014, Address = {Indianapolis, IN}, Author = {{US Trust}}, Publisher = {US Trust and the Lilly Family School of Philanthropy}, Title = {High Net Worth Philanthropy}, Year = {2014} } @Incollection{Castle:2004, Address = {Santa Barbara, CA}, Author = {}, Booktitle = {Philanthropy in {America}: A Comprehensive Historical Encyclopedia}, Editor = {}, Pages = {100--103}, Publisher = {ABC-CLIO}, Title = {Contemporary Philanthropy}, Volume = {1}, Year = {2004} } @Report{CIVICUS:2017, Author = {{CIVICUS}}, Title = {People Power Under Attack: Findings from the {CIVICUS} {Monitor}}, Url = {http://www.civicus.org/images/People_Power_Under_Attack_Findings_from_the_CIVICUS_Monitor.pdf}, Date = {2017-04} } @Techreport{havens2007, Author = { and }, Institution = {The Boston Foundation}, Title = {Geography and Giving: The culture of Philanthropy in {New England} and the Nation}, Type = {Report}, Year = {2007} } @Techreport{NeedlesWolcheckSato:2018, Author = { and and and and and and }, Institution = {{Foundation Center} and {The Council on Foundations}}, Title = {The State of Global Giving by {U.S.} Foundations, 2011--2015}, Type = {Report}, Year = {2018}, Url = {https://www.issuelab.org/resources/31306/31306.pdf} } @Unpublished{ChaudhryHeiss:2018, Author = { and }, Note = {Working paper}, Title = {Are Funders Really Responding? {A}nalyzing the Impact of Global Restrictions on {NGO}s}, Year = {2018} } @Unpublished{icnl:2021, Author = {{International Center for Not-For-Profit Law}}, Title = {The Civic Space Initiative}, Year = {2021}, Url = {https://www.icnl.org/our-work/global-programs/the-civic-space-initiative} } % Created 2021-02-04 jue 10:07 % Intended LaTeX compiler: pdflatex \documentclass[presentation,aspectratio=169]{beamer} \usepackage[utf8]{inputenc} \usepackage[T1]{fontenc} \usepackage{graphicx} \usepackage{grffile} \usepackage{longtable} \usepackage{wrapfig} \usepackage{rotating} \usepackage[normalem]{ulem} \usepackage{amsmath} \usepackage{textcomp} \usepackage{amssymb} \usepackage{capt-of} \usepackage{hyperref} \usepackage{khpreamble} \usepackage{amssymb} \usepgfplotslibrary{groupplots} \newcommand*{\shift}{\operatorname{q}} \usetheme{default} \author{Kjartan} \date{2021-02-08} \title{Análisis de elementos de la mecatrónica} \hypersetup{ pdfauthor={Kjartan}, pdftitle={Análisis de elementos de la mecatrónica}, pdfkeywords={}, pdfsubject={}, pdfcreator={Emacs 26.3 (Org mode 9.4.4)}, pdflang={English}} \begin{document} \maketitle \section{Intro} \label{sec:orge79708b} \begin{frame}[label={sec:org583e729}]{Objetivos, contenido, evaluación} \end{frame} \section{Simscape} \label{sec:org417d912} \begin{frame}[label={sec:org004495d}]{Intuición para sistemas mecanicas} \begin{center} \begin{tikzpicture} \begin{scope}[scale=0.3, xscale=-1, xshift=-10cm] \shade[top color=red, bottom color=white, shading angle={135}] [draw=black,fill=red!20,rounded corners=1.2ex,very thick] (1.5,.5) -- ++(0,1) -- ++(1,0.3) -- ++(3,0) -- ++(1,0) -- ++(0,-1.3) -- (1.5,.5) -- cycle; \draw[very thick, rounded corners=0.5ex,fill=black!20!blue!20!white,thick] (2.5,1.8) -- ++(1,0.7) -- ++(1.6,0) -- ++(0.6,-0.7) -- (2.5,1.8); \draw[thick] (4.2,1.8) -- (4.2,2.5); \draw[draw=black,fill=gray!50,thick] (2.75,.5) circle (.5); \draw[draw=black,fill=gray!50,thick] (5.5,.5) circle (.5); \draw[draw=black,fill=gray!80,semithick] (2.75,.5) circle (.4); \draw[draw=black,fill=gray!80,semithick] (5.5,.5) circle (.4); \draw[thin, ] (7,1) -- (8,1); \draw[thin, ] (6.8,1.5) -- (7.8,1.5); \draw[thin, ] (6,2) -- (7,2); \end{scope} \draw[->,semithick] (-.5,0) -- (8,0); \draw (8.5,0) node {$x(t)$}; \end{tikzpicture} \end{center} Un coche va a velocidad constante en una autopista horizontal. En la instante $t=t_1$, el conductor empuje el clutch, desconectando el motor y las ruedas. Cuál de las siguientes graficas describe mejor la velocidad $v(t)=\dot{x}(t)$ del coche? \begin{center} \begin{tikzpicture} \small \begin{axis}[ width=7cm, height=2.5cm, xlabel={$t$}, ylabel={$v(t)$}, xmin=-3.5, xmax=10.5, ytick = {0}, xtick = {0}, xticklabels = {$t_1$}, ] \addplot+[black, no marks, domain=-4:10, samples=400,variable=k] { (k < 0) + (k>0)*(1+exp(-4))/(1+exp(4*(0.5*k-1)))}; \node[black!40!red] at (axis cs: 5, 0.5) {\huge 1}; \end{axis} \begin{axis}[ xshift=7cm, width=7cm, height=2.5cm, xlabel={$t$}, ylabel={$v(t)$}, xmin=-3.5, xmax=10.5, ytick = {0}, xtick = {0}, xticklabels = {$t_1$}, ] \addplot+[black, no marks, domain=-4:10, samples=400,variable=k] { (k<0) + ((k>=0) - (k>4))*(1/4*(4-k)) }; \node[black!40!red] at (axis cs: 5, 0.5) {\huge 2}; \end{axis} \begin{axis}[ xshift=0cm, yshift=-2.5cm, width=7cm, height=2.5cm, xlabel={$t$}, ylabel={$v(t)$}, xmin=-3.5, xmax=10.5, ytick = {0}, xtick = {0}, xticklabels = {$t_1$}, ] \addplot+[black, no marks, domain=-4:10, samples=400,variable=k] { (k<0) + (k>0)*exp(-0.9*k)}; \node[black!40!red] at (axis cs: 5, 0.5) {\huge 3}; \end{axis} \begin{axis}[ xshift=7cm, yshift=-2.5cm, width=7cm, height=2.5cm, xlabel={$t$}, ylabel={$v(t)$}, xmin=-3.5, xmax=10.5, ytick = {0}, xtick = {0}, xticklabels = {$t_1$}, ] \addplot+[black, no marks, domain=-4:10, samples=400,variable=k] { (k<0) + ((k>=0) - (k>4))*(1-1/16*pow(-k,2)) }; \node[black!40!red] at (axis cs: 5, 0.5) {\huge 4}; \end{axis} \end{tikzpicture} \end{center} \end{frame} \begin{frame}[label={sec:org2a28d69}]{Intuicón para sistemas mecanicas - Simución} \begin{center} \begin{tikzpicture} \tikzstyle{damper}=[thick,decoration={markings, mark connection node=dmp, mark=at position 0.5 with { \node (dmp) [thick,inner sep=0pt,transform shape,rotate=-90,minimum width=15pt,minimum height=3pt,draw=none] {}; \draw [thick] ($(dmp.north east)+(2pt,0)$) -- (dmp.south east) -- (dmp.south west) -- ($(dmp.north west)+(2pt,0)$); \draw [thick] ($(dmp.north)+(0,-5pt)$) -- ($(dmp.north)+(0,5pt)$); } }, decorate] \tikzstyle{ground}=[fill,pattern=north east lines,draw=none,minimum width=0.75cm,minimum height=0.3cm] \begin{scope}[scale=0.3, xscale=-1, xshift=-10cm] \shade[top color=red, bottom color=white, shading angle={135}] [draw=black,fill=red!20,rounded corners=1.2ex,very thick] (1.5,.5) -- ++(0,1) -- ++(1,0.3) -- ++(3,0) -- ++(1,0) -- ++(0,-1.3) -- (1.5,.5) -- cycle; \draw[very thick, rounded corners=0.5ex,fill=black!20!blue!20!white,thick] (2.5,1.8) -- ++(1,0.7) -- ++(1.6,0) -- ++(0.6,-0.7) -- (2.5,1.8); \draw[thick] (4.2,1.8) -- (4.2,2.5); \draw[draw=black,fill=gray!50,thick] (2.75,.5) circle (.5); \draw[draw=black,fill=gray!50,thick] (5.5,.5) circle (.5); \draw[draw=black,fill=gray!80,semithick] (2.75,.5) circle (.4); \draw[draw=black,fill=gray!80,semithick] (5.5,.5) circle (.4); \draw[thin, ] (7,1) -- (8,1); \draw[thin, ] (6.8,1.5) -- (7.8,1.5); \draw[thin, ] (6,2) -- (7,2); \node[coordinate] (fender) at (6.5, 1.5) {}; \end{scope} \draw[semithick] (-0.5,0) -- (-0.5, 1); \draw[damper] (-0.5, 0.5 |- fender) -- (fender); \node[ground, rotate=90, anchor=south] at (-0.5, 0.5) {}; \draw[->,semithick] (-.5,0) -- (8,0); \draw (8.5,0) node {$x(t)$}; \end{tikzpicture} \end{center} mass \(m = \unit{1000}{\kilo\gram}\), friction coefficient \(f=\unit{20}{\newton\per(\meter\per\second)}\) \end{frame} \begin{frame}[label={sec:org0c3a87c}]{Intuición para sistemas electricas} \begin{columns} \begin{column}{0.4\columnwidth} \begin{center} \includegraphics[height=0.8\textheight]{../../figures/RC-circuit} \end{center} \#+begin\textsubscript{export} latex \begin{center} \begin{tikzpicture} \begin{axis}[ xshift=0cm, yshift=-2.5cm, width=7cm, height=2.5cm, xlabel={$t$}, ylabel={$v(t)$}, xmin=-3.5, xmax=10.5, ytick = {0}, xtick = {0}, xticklabels = {$t_1$}, ] \addplot+[black, no marks, domain=-4:10, samples=400,variable=k] { (k<0) + (k>0)*exp(-0.9*k)}; \node[black!40!red] at (axis cs: 5, 0.5) {\huge 3}; \end{axis} \end{tikzpicture} \end{center} \#+end\textsubscript{export} latex \end{column} \begin{column}{0.6\columnwidth} \alert{Actividad individual} Al principio (\(t=0\)) el circuito está abierto y no hay carga en el capacidor. En el instante \(t=0\) el interruptor S cierre el circuito y lo mantiene cerrado. Grafica el voltage sobre el capacidor como función de tiempo. El constante de tiempo del sistem es \(\tau=RC\), y determine el comportamiento del sistema. Indica en tú gráfica como se puede identificar \(tau\). Tomo fotó y mandamelo por \alert{Remind}. \end{column} \end{columns} \end{frame} \begin{frame}[label={sec:orgb0433e8}]{Intuition for electrical circuits - Simulation} Let \(R=\unit{1}{\kilo}\Omega\) and \(C = \unit{100}{\micro\farad}\). \[ \tau = RC = (1\times 10^{3})(100\times 10^{-6}) = \unit{10^{-1}}{\second} \] \end{frame} \end{document}%File: ~/OOP/modelbuilder/ModelBuilder.tex %What: "@(#) ModelBuilder.tex, revA" \noindent {\bf Files} \\ \indent \#include $<\tilde{ }$/modelbuilder/ModelBuilder.h$>$ \\ \noindent {\bf Class Declaration} \\ \indent class ModelBuilder; \\ \noindent {\bf Class Hierarchy} \\ \indent {\bf ModelBuilder} \\ \noindent {\bf Description} \\ \indent The ModelBuilder class is an abstract base class. A subclass of ModelBuilder is a class which creates the finite element discretization of a structure: that is it discretizes the structure to be modeled into Elements, Nodes, Constraints, etc. and adds these components to the Domain. \\ \noindent {\bf Class Interface} \\ \indent // Constructor \\ \indent {\em ModelBuilder(theDomain \&theDomain);}\\ \\ \indent // Destructor \\ \indent {\em virtual $\tilde{ }$ModelBuilder();}\\ \\ \indent // Public Methods \\ \indent {\em virtual buildFE\_Model(void) = 0;} \\ \\ \indent // Protected Methods \\ \indent {\em Domain *getDomainPtr(void) const;} \\ \noindent {\bf Constructor} \\ \indent {\em ModelBuilder(theDomain \&theDomain);}\\ All models are associated with a single domain, this constructor sets up the link between the model and the domain, setting its link to the Domain object {\em theDomain}. \\ \noindent {\bf Destructor} \\ \indent {\em virtual~ $\tilde{}$ModelBuilder();}\\ Does nothing. \\ \noindent {\bf Public Methods} \\ \indent {\em virtual buildFE\_Model(void) = 0;} \\ The ModelBuilder will construct the Element, Node, Load and Constraint objects and add them to the Domain object associated with the ModelBuilder. \\ \noindent {\bf Protected Methods} \\ \indent {\em Domain *getDomainPtr(void) const;} \\ Returns a pointer to the Domain object passed in the constructor. This method can be used in the subclasses to get a pointer the Domain object to which to add the domain components. \\ \section{Eigenvectors, Eigenvalues and Diagonal Matrices} This is the first step into the wonderful land of diagonalisation of endomorphisms. Consider a vector space $V$ over $F$ with $\dim V=n<\infty$ and let $\alpha:V\to V$ be an endomorphism. The general problem is whether we can find a basis $B$ of $V$ such that $[\alpha]_B$ is in a nice enough form. In other words, by our change-of-basis formula, we want to know when can a matrix be conjugate to another matrix in a nice form. \begin{definition} 1. $\alpha\in L(V)=L(V,V)$ is diagonalisable if there exists a basis $B$ of $V$ such that $[\alpha]_B$ is diagonal, i.e. $([\alpha]_B)_{ij}=0$ for $i\neq j$.\\ 2. $\alpha\in L(V)$ is triangulable if there exists a basis $B$ of $V$ such that $[\alpha]_B$ is (upper) triangular. \end{definition} \begin{remark} A matrix is diagonalisable (resp. triangulable) iff it is conjugate to a diagonal (resp. triangular) matrix. \end{remark} \begin{definition} 1. $\lambda\in F$ is an eigenvalue of $\alpha$ if $\alpha(v)=\lambda v$ for some $v\neq 0$.\\ 2. $v\in V$ is an eigenvector of $\alpha$ if $v\neq 0$ and there exists some $\lambda\in F$ such that $\alpha(v)=\lambda v$.\\ 3. $V_\lambda=\{v\in V:\alpha(v)=\lambda v\}\le V$ is called the eigenspace of $\alpha$ associated to $\lambda$. \end{definition} \begin{lemma} If $\alpha\in L(V)$ and $\lambda\in F$, then $\lambda$ is an eigenvalue iff $\det(\alpha-\lambda\operatorname{id}_V)=0$. \end{lemma} \begin{proof} Follows from the fact that matrices with nonzero determinant have zero kernel. \end{proof} \begin{remark} If $\alpha(v_j)=\lambda v_j$ for $v_j\neq 0$, then completing it into a basis $B=\{v_1,\ldots,v_j,\ldots,v_n\}$ of $V$ gives $([\alpha]_B)_{ij}=\lambda\delta_{ij}$. \end{remark} Recall that for a field $F$, a polynomail in $F$ is $f(t)=a_nt^n+\cdots+a_0\in F[t]$ with $a_i\in F$. Let $\deg f$ be the largest $m$ such that $a_m\neq 0$, then we know that $\deg(f+g)\le\max\{\deg f,\deg g\}$ and $\deg(fg)=\deg(f)+\deg(g)$. We say $\lambda$ is a root of $f$ iff $f(\lambda)=0$, and $g(t)$ divides $f(t)$ if there is some $q(t)\in F[t]$ such that $f(t)=g(t)q(t)$. \begin{lemma} If $\lambda$ is a root of $f$, then $x-\lambda$ divides $f$. \end{lemma} \begin{proof} Write $f(t)=f(t)-f(\lambda)$ and factrorise. \end{proof} \begin{remark} We say $\lambda$ is a root of multiplicity $k$ if $(t-\lambda)^k$ divides $f$ but $(t-\lambda)^{k+1}$ does not. \end{remark} \begin{example} $f(t)=(t-1)^2(t-2)^3$ has roots $1$ with multiplicity $2$ and $2$ with multiplicity $3$. \end{example} \begin{corollary} A polynomial of degree $n$ has at most $n$ roots, counted with multiplicity. \end{corollary} \begin{proof} Induction. \end{proof} \begin{corollary} For polynomials $f_1,f_2$ of degree less than $n$ with $f_1(t_i)=f_2(t_i)$ for distinct $t_1,\ldots,t_n$, then $f_1=f_2$. \end{corollary} \begin{proof} $\deg(f_1-f_2)1$, there is $\lambda$ such that $\chi_\alpha(\lambda)=0$ by assumption. Let $\{v_1,\ldots,v_k\}$ be a basis of $U=V_\lambda$ and extend it to a basis $B=\{v_1,\ldots,v_n\}$ of $V$. We then have $$[\alpha]_B=\left( \begin{array}{c|c} I_k&\ast\\ \hline 0&C \end{array} \right)$$ So the induced endomorphism $\bar\alpha:V/U\to V/U$ has matrix $C$ under the basis $\{v_{k+1}+U,\ldots,v_n+U\}$. Then by the induction hypothesis, we can choose another set of basis $\{\tilde{v}_{k+1}+U,\ldots,\tilde{v}_n+U\}$ so that $C$ is triangular. Hence $\alpha$ is triangular under the basis $\{v_1,\ldots,v_k,\tilde{v}_{k+1},\ldots,\tilde{v}_n\}$. This completes the proof. \end{proof} \begin{lemma} Suppose $V$ is a vector space over $F=\mathbb R$ or $\mathbb C$ such that $\dim V=n<\infty$ and suppose $\alpha\in L(V)$ is an endomorphism with matrix $A$. Say $\chi_\alpha(t)=(-1)^nt^n+c_{n-1}t^{n-1}+\cdots+c_0$, then $c_0=\det A$, $c_{n-1}=(-1)^{n-1}\operatorname{tr}A$. \end{lemma} \begin{proof} $\det A=\chi_A(0)=c_0$. For $c_{n-1}$, note that the statement is true for triangular $A$, so we are done by the preceding theorem. \end{proof} docu/Monty_Mumup_Glossar.tex \newglossaryentry{raspberrypi}{ name={Raspberry Pi}, description={ Eine sehr bekannte Reihe von \gls{open-source} \glspl{sbc} welche durch die Raspberry Pi Foundation entwickelt werden} } \newglossaryentry{sbc}{ name={SBC}, description={ Ein Computer, bei welchem alle für den Betrieb benötigten Komponenten auf einer Platinen zusammengefasst sind, wird als \glsentrylong{sbc} (\glsentryname{sbc}, Einplatinencomputer) bezeichnet}, plural={SBC}, long={Single-board Computer}, first={\glsentrylong{sbc}} (\glsentryname{sbc}), firstplural={Single-board Computern (\glsentryname{sbc})} } \newglossaryentry{iot}{ name={IoT}, description={ Eingebettete Computer in alltäglichen Gegenständen können Daten sammeln und sich über das Internet vernetzen, um den Nutzer über die gewohnte Funktionalität hinaus zu unterstützen, dabei jedoch nicht aufzufallen }, long={Internet of Things}, first={\glsentrylong{iot} (\glsentryname{iot})}, } \newglossaryentry{uuid}{ name={UUID}, description={ Ein \glsentrylong{uuid} ist eine 128-Bit-Zahl die dazu dient, Computer und Daten weltweit eindeutig zu identifizieren. Durch den verwendeten Prozess bei der Erstellung einer \glsentryname{uuid} soll mit annähernd absoluter Sicherheit garantiert werden, dass es unter UUIDs keine Duplikate gibt}, long={Universally Unique Identifier}, first={\glsentrylong{uuid} (\glsentryname{uuid})}, } \newglossaryentry{xor}{ name={XOR}, description={ Eine XOR-Verknüpfung bezeichnet eine bitweise Operation bei der eine exklusive Disjunktion angewendet wird. Bitweise werden beide Eingaben verglichen. Sind die Eingaben an dieser Stelle gleich, so ist das Resultat an der Stelle 0. Unterscheiden sich die Eingaben, ist das Resultat an der Stelle 1. Beide Eingaben produzieren gemeinsam eine Ausgabe. Wendet man XOR auf eine der Eingaben und die vorherige Ausgabe an, so ist die andere Ausgabe das Resultat}, } \newglossaryentry{hash}{ name={Hash}, description={ Eine Hashfunktion bildet eine große Datenmenge auf eine kleinere ab. Es ist damit möglich, bis zu einem gewissen Grad an Sicherheit, eine Eingabe eindeutig zu identifizieren, ohne die Eingabe vollständig speichern oder übertragen zu müssen} } \newglossaryentry{dac}{ name={DAC}, description={ Digital-Analog-Wandler (engl. Digital-Analog-Converter) sind Module, welche digitale zu analogen Signalen umsetzen. Sie werden hauptsächlich eingesetzt um in elektronischen Geräten analoge Signale für Lautsprecher zu erzeugen, welche als Schallwellen wiedergegeben werden}, short={DAC}, long={Digital-Analog-Wandler}, first={\glsentrylong{dac}} (\glsentryname{dac}), } \newglossaryentry{i2s}{ name={I$^2$S}, description={\glsentryname{i2s} steht für \glsentrylong{i2s} und ist ein Standard für die digitale Übertragung von Tönen zwischen Prozessoren}, long={Inter-IC Sound}, short={I$^2$S}, first={\glsentrylong{i2s}} (\glsentryname{i2s}) } \newglossaryentry{distribution}{ name={Distribution}, description={ Eine Distribution ist eine Bündelung von Software, welche einander benötigt, um sinnvoll genutzt werden zu können. Meist wird der Begriff benutzt um verschiedene Varianten von Linux-Systemen zu beschreiben, die neben dem Linux-Kernel noch weitere Softwarepakete enthalten um als vollständiges Betriebssystem nutzbar zu sein}, plural={Distributionen} } \newglossaryentry{headless}{ name={headless}, description={ Computer, welche über keine grafische Oberfläche verfügen werden als \enquote{\gls{headless}} bezeichnet} } \newglossaryentry{open-source}{ name={Open-Source}, description={ Software deren Quellcodes oder Hardware deren technische Spezifikationen und Layouts von jedem eingesehen, verbreitet, modifiziert und genutzt werden können wird als \gls{open-source} bezeichnet } } \newglossaryentry{python}{ name={Python}, description={Python ist eine höhere Interpretersprache. Sie unterstützt funktionales, objektorientiertes und aspektorientiertes Programmieren } } \newglossaryentry{api}{ name={API}, description={ Ein Application Programming Interface (API) ist eine Menge von Regeln und Spezifikationen, die ein Programm befolgen kann, um auf bestimmte Dienste und Ressourcen zugreifen zu können, die von einem anderen Programm bereit gestellt werden, welches die API implementiert}, long={Application Programming Interface}, first={\glsentrylong{api}} (\glsentryname{api}), firstplural={\glsentrylong{api}\glspluralsuffix\ (\glsentryname{api}\glspluralsuffix )} } \newglossaryentry{cli}{ name={CLI}, description={ Ein Command Line Interface (CLI) ist eine Schnittstelle zu zu einem Programm, welche über die Kommandozeile aufgerufen und genutzt werden kann}, long={Command Line Interface}, first={\glsentrylong{cli}} (\glsentryname{cli}), firstplural={\glsentrylong{cli}\glspluralsuffix\ (\glsentryname{cli}\glspluralsuffix )} } \newglossaryentry{html}{ name={HTML}, description={ Hypertext Markup Language (HTML) ist eine \gls{xml}-basierte Auszeichnungssprache. Sie die Grundlage für Seiten im \gls{www} und wird von \glspl{browser} interpretiert}, long={Hypertext Markup Language}, first={\glsentrylong{html}} (\glsentryname{html}) } \newglossaryentry{js}{ name={JavaScript}, description={ JavaScript (JS) ist eine von \glspl{browser} interpretierte Programmiersprache, welche das Generieren von dynamischen \glspl{website} ermöglicht} } \newglossaryentry{json}{ name={JSON}, description={ \glsentrylong{json} (\glsentryname{json}) ist das Standarddatenformat von \gls{js} und zeichnet sich durch kompakte, klar lesbare und einfach interpretierbare notation von Objekten aus}, long={JavaScript Object Notation}, first={\glsentrylong{json} (\glsentryname{json})} } \newglossaryentry{dom}{ name={DOM}, description={ Das \glsentrylong{dom} (\glsentryname{dom}) ist eine in \gls{js} enthaltene \gls{api} um den \gls{html}-Code der aufrufenden Seite in Form von Objekten zu laden und dynamisch zu verändern}, long={Document Object Model}, first={\glsentrylong{dom} (\glsentryname{dom})} } \newglossaryentry{ajax}{ name={AJAX}, description={ \glsentrylong{ajax} (\glsentryname{ajax}) bezeichnet asynchrone Datenübertragungen zwischen \gls{js}-basierten Webanwendungen und einem Server }, long={Asynchronous JavaScript and XML}, first={\glsentrylong{ajax} (\glsentryname{ajax})} } \newglossaryentry{sse}{ name={SSE}, description={ \glsentrylong{sse} (\glsentryname{sse}) bezeichnet eine in \gls{html}5 enthaltene Möglichkeit, über einen zuvor vom Client geöffneten Stream, Daten vom Server an den Client zu senden }, long={Server-Send Events}, first={\glsentrylong{sse} (\glsentryname{sse})}, firstplural={\glsentrylong{sse}\glspluralsuffix\ (\glsentryname{sse}\glspluralsuffix )} } \newglossaryentry{websocket}{ name={Websocket}, description={ Websockets sind eine Erweiterung des klassischen \gls{http}-Protokolls, die eine bidirektionale Kommunikation zwischen Client und Server ermöglichen}, plural={Websockets} } \newglossaryentry{http}{ name={HTTP}, description={ Das \glsentrylong{http} (\glsentryname{http}) ist das Übertragungsprotokoll, welches genutzt wird, um im \gls{www} \glspl{website} von einem Server zu laden}, long={Hypertext Transfer Protocol}, first={\glsentrylong{http} (\glsentryname{http})} } \newglossaryentry{css}{ name={CSS}, description={ \glsentrylong{css} (\glsentryname{css}) ist eine Sprache zur Definition von Styles für \glossary{html}-Elemente}, long={Cascading Style Sheets}, first={\glsentrylong{css} (\glsentryname{css})} } \newglossaryentry{xml}{ name={XML}, description={ \glsentrylong{xml} (engl. Erweiterbare Auszeichnungssprache)}, long={Extensible Markup Language}, first={\glsentrylong{xml} (\glsentryname{xml})} } \newglossaryentry{www}{ name={WWW}, description={ Das World Wide Web (WWW) besteht aus \glspl{website}, die per Hyperlinks miteinander verknüpft sind und über das Internet abgerufen werden können}, long={World Wide Web}, first={\glsentrylong{www} (\glsentryname{www})} } \newglossaryentry{website}{ name={Webseite}, description={ Über \glspl{url} erreichbares Dokument, das mit \gls{html} formatiert wurde und von \glspl{browser} aufgerufen werden kann}, plural={Webseiten} } \newglossaryentry{webapp}{ name={Web-App}, description={ In \gls{html} und \gls{js} umgesetzte Anwendungen, welche ohne Installation von einem \gls{browser} geladen und ausgeführt werden kann}, plural={Web-Apps} } \newglossaryentry{browser}{ name={Webbrowser}, description={ Programm zur Darstellung von \glspl{website}}, plural={Webbrowser} } \newglossaryentry{url}{ name={URL}, description={ \glsentrylong{url}\glspluralsuffix\ (\glsentryname{url}\glspluralsuffix) (engl. einheitlicher Ressourcenzeiger) dienen im \gls{www} der eindeutigen Definition und referenzierung von Ressourcen}, plural={URLs}, long={Uniform Resource Locator}, first={\glsentrylong{url}} (\glsentryname{url}), firstplural={\glsentrylong{url}\glspluralsuffix\ (\glsentryname{url}\glspluralsuffix )} } \newglossaryentry{koroutine}{ name={Koroutine}, description={Prozeduren, deren Ablauf unterbrochen und bei Bedarf wieder aufgenommen werden kann}, plural={Koroutinen} } priyankaanehra/regolith \include{/tmp/regolith_fake/hello.txt}\newcommand{\ignore}[1]{} %\subsubsection{\stid{3.07} Factorization Based Sparse Solvers and Preconditioners for Exascale} \label{subsubsect:strumpack} \subsubsection{\stid{3.07} STRUMPACK-SuperLU} \label{subsubsect:strumpack} \paragraph{Overview} % \textit{Provide an overview of your project. You might find that the introductory text from your Fall 2017 Project Summary \url{https://confluence.exascaleproject.org/display/1ST/Fall+2017+ECP+ST+Project+Summaries} useful as a starting draft.} In this project we will deliver factorization based sparse solvers encompassing the two widely used algorithm variants: supernodal (SuperLU library: \url{https://portal.nersc.gov/project/sparse/superlu}) and multifrontal (STRUMPACK library: \url{http://portal.nersc.gov/project/sparse/strumpacK}). STRUMPACK is further enhanced with scalable preconditioning functionality using hierarchical matrix algebra. Both libraries are purely algebraic, applicable to a large variety of application domains. We will address several challenges that arise in Exascale computing, with the following focus areas: (1) Develop novel approximation algorithms that have lower arithmetic and communication complexity with respect to the size of the input matrix; (2) Develop new parallelization strategies that reduce inter-process communication and expose task parallelism and vectorization for irregular computations involving sparse data structures to better use on-node resources; (3) Integrate our software into the higher level algebraic solvers such as hypre, PETSc, Trilinos, and collaborate with ECP application teams for application-specific and hardware-specific tuning of the parameters space to achieve optimal efficiency of our solvers. Our solver technology is essential for ECP, because many DOE simulation and data analysis codes expected to run on the Exascale machines need solutions of sparse algebraic systems, and many high fidelity simulations involve large-scale multiphysics and multiscale modeling problems that generate highly ill-conditioned and indefinite algebraic equations, for which pure iterative methods such as Krylov and multigrid, albeit readily parallelizable on large machines, cannot converge to the solution. The factorization based algorithms being developed herein represent an important class of methods that are indispensable building blocks for solving those numerically challenging problems. Our software can often be used as a reliable standalone solver, or as a preconditioner for Krylov iterative methods, or as a coarse grid solver in multigrid methods, just to name a few. \paragraph{Key Challenges} %\textit{Describe what is hard to do, why it is challenging.} At Exascale we need to address several major challenges: decreasing amount of memory per core, larger impact of communication cost and load imbalance, more heterogeneous architecture. Our new design of algorithms and codes need to focus on reducing communication and synchronization and task scheduling instead of floating point operation throughput. In sparse factorization methods, we expect new bottlenecks in parts of the code that previously received little attention. For example, the preprocessing step involves numerical pivoting for selecting stable pivots and symbolic factorization, which do not yet parallelize well on manycore architectures with fine-grained parallelism. At Exascale, direct solvers are more likely to be used in a preconditioning strategy, for example, in block Jacobi preconditioning, in domain decomposition methods or as coarse-grid solvers in algebraic multigrid, which requires repeated triangular solves. The challenge here is to mitigate the low arithmetic intensity and high degree of data dependency. Compared to iterative methods, the primary bottleneck of direct solvers is the asymptotically higher growth in memory need and floating point operations, especially for problems from three-dimensional geometry. It is imperative to develop novel factorization methods which require much less memory footprint and data movement. \paragraph{Solution Strategy} %\textit{Describe your basic strategy for addressing the challenges.} We will address these challenges in several thrust areas. The new techniques will be implemented in the two software packages SuperLU and STRUMPACK. The former is a widely used sparse direct solver based on supernodal factorization and the latter is a newer direct solver/preconditioner package based on multifrontal factorization and hierarchical low-rank matrix structures. % Parallel pre-pivoting for both packages. The improvements for SuperLU will be mainly in two areas: (1) develop the communication-avoiding 3D factorization and triangular solve algorithms and codes that have provably lower communication complexity; (2) develop a synchronization-avoiding triangular solve code to enable more overlap of communications of different processes at different substitution steps; (3) develop new multi-GPU codes for both symbolic preprocessing step and numerical factorization and solve steps. In addition to exploiting structural sparsity as SuperLU does, STRUMPACK also exploits data sparseness in the dense blocks of sparse factors using low-rank representations, which leads to linear scaling $O(n)$ or $O(n \log n)$ memory and arithmetic complexity for PDEs with smooth kernels. The developments for STRUMPACK will focus on several areas: (1) develop robust stopping criteria --- both absolute and relative --- for adaptive (incremental) randomized sampling schemes to reveal numerical ranks in the low-rank compression routine. The goal is to use enough samples for stability, but not too many for efficiency; (2) add OpenMP support for both HSS compression and ULV factorization routines, especially use OpenMP task construct to support irregular parallelism. (3) reduce MPI communication in all stages of the code, including HSS construction, ULV factorization and triangular solve; (4) in addition to HSS, develop codes to support other simpler low-rank formats, such as HOLDR and BLR. The HSS format has asymptotically lower complexity than HOLDR and BLR, but has a larger prefactor constant. We expect HSS to be more useful for large-scale problems while HOLDR and BLR are more useful for mid-range problems; (5) work with the ECP application teams to examine their specific problem characteristics and develop the best clustering/ordering methods to reveal low-rank structures. \paragraph{Recent Progress} %\textit{Describe what you have done recently. It would be good to have some kind of figure or diagram in this section.} In the past six months, we focused on adding more support for GPU and improving on-node threading performance. To this end, we participated in the 3.5 days ECP OpenMP Hackathon at NERSC in August 2019. During hackathon, we worked closely with the other two ECP ST teams: HPCToolkit and SOLLVE, as well as the OpenMP/vectorization experts from Intel. In particular, using HPCToolkit helps reveal several performance bottlenecks. We rewrote the code to remove a few inefficiencies, and identified solution strategies for the other bottlenecks. We also worked with two ECP applications: MFEM electromagnetic diffusion problems governed by high frequency indefinite Maxwell equations (CEED Co-Design Center) and ExaSGD Miniapp2 -- sparse ACOPF matrices: \url{https://github.com/LLNL/hiop/tree/sandbox/matrices/data/acopf_matrices} The algorithmic changes and the results are detailed below. \paragraph\ \underline{STRUMPACK} \begin{itemize} \item Algorithmic change: changed recursive tree traversal to a level-by-level traversal, which has less task scheduling overhead and allows to use batched dense linear algebra routines. \item There are many small dense linear algebra operations on lower levels of the supernodal elimination tree. Current solution strategy is to perform small DLA operations on CPU, for larger ones use cuBLAS with CUDA streams. The future solution strategy is to use variable sized batched operations from MAGMA and/or SLATE. \item On Summit, for the MFEM matrix, 8CPU + 1GPU factorization achieved 8x speedup over 8CPU cores, and 19x speedup over 1CPU core. \item Added the new Butterfly and low rank compression schemes. For the MFEM matrix, this obtained much smaller numerical rank and better compression rate compared to the HSS low rank format. \end{itemize} \underline{SuperLU} \begin{itemize} \item Developed the first GPU code to perform supernodal sparse triangular solve. On Summit 1 GPU, it achieved 2x speedup over 8 CPU cores for ExaSGD matrix. It is 5.5x faster than Nvidia’s cuSPARSE. \item Factorization: on-node OpenMP performance: currently a single ``guided'' scheduling strategy is used in various OMP for-loops. HPCToolkit points out serious load imbalance for a couple of for-loops. We are examining some hybrid scheduling strategies. \end{itemize} \vspace{-.3in} \begin{figure}[htb] \begin{minipage}[t]{0.48\columnwidth} \centering \includegraphics[scale=0.7]{projects/2.3.3-MathLibs/2.3.3.07-STRUMPACK-SuperLU/strumpack-Summit.pdf} \caption{STRUMPACK factorization on Summit GPU.} \label{fig:strumpack-parmetis-scaling} \end{minipage} \begin{minipage}[t]{0.48\columnwidth} \centering \includegraphics[scale=0.8]{projects/2.3.3-MathLibs/2.3.3.07-STRUMPACK-SuperLU/superlu-solve-Summit.pdf} \caption{SuperLU solve on Summit GPU.} \label{fig:strumpack-metis-scaling} \end{minipage} \end{figure} %%-------------------------------------- \ignore{ %%%%%%%% from previous period In the past six months, we have made good progress in several areas. For the latest releases of both packages, we integrated a parallel pre-ordering algorithm called Approximate-Weight Perfect Matching (AWPM) for pivot selection~\cite{AWPM2018}. The other improvements are: \begin{enumerate} \item We released a new version v6.1.0 of SuperLU\_DIST, which contains the following features: 1) Improvement on strong scaling of the triangular solve -- up to 4.4x faster than Version 5.x on 4000+ cores~\cite{LiuTriSolve2018}; 2) On-node threading optimization leading up to 3x speedup on a Cori-KNL node; \item We released a new version v3.1.0 of STRUMPACK, which contains the following new features: 1) Changes to the build system for xSDK compliance; 2) Improvements on the scalability of the HSS algorithms -- dense matrix HSS compression is up to 4.7x faster on 8 nodes (256 cores) of Cori-Haswell and 2.4x faster on Cori-NKL, the HSS-embedded sparse factorization is up to 2.2x faster on 8 Cori-KNL nodes; 3) Improvement in HSS ULV solve with reduced communication and more OpenMP support, leading up to 7x faster in matrix redistribution and 1.4x faster in the entire solve. \item For both STRUMPACK and SuperLU, we performed initial bottleneck studies for two ECP applications: CEED (MFEM indefinite Maxwell simulation) and ExaSGD (Optimizing stochastic grid dynamics). We identified performance bottlenecks of the solvers, proposed remedies, and documented the findings in the milestone memo: {\url{https://jira.exascaleproject.org/secure/attachment/15207/MS-ECP-App-Bottlenecks-study-Oct-2018.pdf}}. The plots below show strong scaling of STRUMPACK up to 8192 cores of Cori-Haswell. Numerical factorization scales well, but ParMETIS ordering becomes a serious bottleneck; it is even slower than serial METIS when using 1000+ cores. But for large problems, serial METIS cannot be run on one node. The symbolic analysis phase also needs improvement at large scale. \end{enumerate} \vspace{-.3in} \begin{figure}[htb] \begin{minipage}[b]{0.48\columnwidth} \centering \includegraphics[scale=0.7]{projects/2.3.3-MathLibs/2.3.3.07-STRUMPACK-SuperLU/periodic-cube-scaling-strumpack.pdf} \caption{STRUMPACK scaling with ParMETIS.} \label{fig:strumpack-parmetis-scaling} \end{minipage} \begin{minipage}[b]{0.48\columnwidth} \centering \includegraphics[scale=0.7]{projects/2.3.3-MathLibs/2.3.3.07-STRUMPACK-SuperLU/periodic-cube-scaling-strumpack_metis.pdf} \caption{STRUMPACK scaling with METIS.} \label{fig:strumpack-metis-scaling} \end{minipage} \end{figure} } %%%%%---- ignored \ignore{ %%%%%%%%%%%% from last period .... \begin{enumerate} \item We developed and evaluated a fully algebraic sparse preconditioner in STRUMPACK. On top of the baseline multifrontal direct solver, we use low-rank compression in dense frontal matrices to obtain approximate factorization. We showed that our MF+HSS preconditioner is more robust for numerically hard problems than many alternatives. Our code strong scales to over 6000 cores~\cite{ghysels2017-ipdps} (Fig.~\ref{fig:strumpack-scaling}). \item We developed several strategies to enhance scalability of triangular solve in SuperLU\_DIST. One is an asynchronous tree-based broadcast/reduction scheme which reduces latency and improves communication load balance. Another is efficient threading implementation and BLAS operations. The new code is 4.4x and 6.1x faster on 4096 cores with one and 50 right-hand sides, respectively~\cite{LiuTriSolve2018} (Fig.~\ref{fig:superlu-trisolve}). \item We developed a new communication-avoiding 3D sparse LU factorization (IPDPS) algorithm that has provably asymptotic lower communication complexity in both latency and volume. The prototype implementation in SuperLU\_DIST achieves up to 27x improvement over the baseline 2D algorithm when run on 24,000 cores of Edison at NERSC~\cite{sao2018}. \item In collaboration with ExaGraph ECP project, we evaluated the performance of a parallel pre-ordering algorithm called Approximate-Weight Perfect Matching (AWPM) for pivot selection in SuperLU\_DIST. For most practical problems (e.g.,DOE apps, and SuiteSparse) the weights of the perfect matchings generated by AWPM often within 99\% of the optimum. The MPI+OpenMP implementation on Cori at NERSC scales up to 256 nodes –-- 2500x faster than serial MC64, and up to 114x speedup on 256 nodes (17,408 cores) Cori-KNL~\cite{AWPM2018}. The interface to AWPM are already implemented in both STRUMPACK and SuperLU and are released. \end{enumerate} } %%%% ignored from last period .... %%-------------------------------------- \paragraph{Next Steps} Our future efforts will focus on the following areas: %\textit{Describe what you are working on next.} \begin{itemize} \item For STRUMPACK, Gather/Scatter operations are still on CPU. The future solution strategy is to use CUDA kernel, or OpenMP 4.5+ target off-load feature. \item For SuperLU, we will develop a more scalable multi-GPU symbolic factorization code, and develop a communication strategy to combine threading with one-sided communications. \item We will build detailed performance models and performance specific code optimizations for the ECP applications that use our solvers. \end{itemize} %----------- \ignore{ %%%% from last period .... \begin{itemize} \item For STRUMPACK, we will improve the performance of the HSS solve routine, add OpenMP and reduce communication. We will implement the HOLDR low-rank format. \item For both STRUMPACK and SuperLU, we will build detailed performance models and performance specific code optimizations for the ECP applications that use our solvers. \end{itemize} } %%%%------ ignored from last period .... %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%% FFTX sub-project %%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsubsection{\stid{3.07} Sub-project: FFTX} \label{subsubsect:fftx} \noindent \paragraph{Overview} The use of FFTs span a broad range of DOE science applications, including ones represented in the Exascale applications space. Most applications use the API from FFTW, an open-source library developed in the 1990's. FFTW is still used extensively on DOE HPC platforms, and the FFTW API has become the de-facto standard FFT interface: vendors that provide FFT libraries implement (at least a subset) of that interface. Thus, FFTW both defines the standard FFT library interface, as well as being a key software component for applications. In the FFTX project (\url{https://github.com/spiralgen/fftx}), we are developing a new package for supporting FFT applications on exascale architectures. Our approach based on two ideas. The first is develop a backwards-compatible approach that builds on the FFTW interface but extends it to enable extraction of high performance on exascale machines. The second idea is to provide a toolchain that enables the specialization of the FFT calculation and its surrounding use case calculations (scaling, data layout transformation, marshalling/unmarshalling for communication), using code generation, symbolic analysis, automatic performance tuning, and applications-specific code generation. We will use SPIRAL, an open-source toolchain for FFT developed at CMU, as the basis for developing FFTX, and will use specific ECP applications and target ECP exascale platforms to provide a focus for this work. \paragraph{Key Challenges} The traditional approach of of applications using FFTs is to build up high-performance implementations out of calls to (usually 1D) FFT libraries (either FFTW or vendor libraries), interleaved with user-implemented code for use-case-specific operations. This approach may break down on the emerging ECP platforms, for two reasons. The first is that the node architectures have become more complex. Multiple cores and accelerators lead to multiple levels of parallelism, including threading and SIMD/SIMT. In addition, there are on-node complex memory hierarchies that are to varying extents user-controlled, and across which it is expensive to move data. This leads to more complex interleaving of the other components of multidimensional FFT-based applications with the core library FFTs in order to maximize the effective use of the floating point capabilities and minimize data movement across the memory hierarchy. Some of these are simply not expressible in the current FFTW interface; others can be expressed, but with great effort on the part of the applications programmer, and often with an outcome of not yielding the theoretically-predicted performance due to unexpected and opaque behavior of the FFT library software. A second problem is that the open-source FFTW libraries are no longer supported. The original developers have gone on to other things, and the support of FFTW, such as it is, is performed by volunteer labor. As a result, the extensions to support new node architectures are more brittle and provide less coverage. Expanding the feature set of FFTW to enable the more effective use of the new node architectures is not feasible, since it would entail significant modification and use of the back-end software system, which no one is supporting, and on which the expertise is no longer available. \paragraph{Solution Strategies} There are three components to our approach to providing a new software stack for FFT applications. \begin{trivlist} \item (1) We will design an extension of the FFTW interface that both meets the needs of the FFT use cases arising in ECP applications, and exposes the oppportunities for obtaining high performance on current and future architectures. The FFTX interface will be backwards compatible with FFTW so that legacy code using FFTW runs unmodified and gains substantially on hardware to which FFTX has been ported. To express the additional opportunities for obtaining improved performance, we will add a small number of new features beyond the FFTW interface to express algorithmic features such as futures/delayed execution, offloading, data placement, callback kernels, and sparsity of inputs or outputs. Such changes will have the potential to extract much higher performance than standard FFTW calls, since higher level operations and new hardware features can be addressed. This interface will be designed as an embedded DSL, for which we will provide a standard C/C++ reference library implementation that enables rapid assessment of the interface by applications developers. \item (2) We will develop a new code-generation back end. FFT-based application kernels implemented using the extended FFTW interface described above will be treated as specifications. This allows the extraction of the algorithm semantics from source code and known library semantics, thus providing whole-kernel semantics and whole-kernel specifications. This enables build-time source-to-source translation and advanced performance optimizations, such as cross-library optimization, targeting of accelerators through off-loading, and inlining of user-provided kernels. Our approach allows for fine control over resource expenditure during optimization. Applications can control compile-time, initialization-time, invocation time optimization resources if they need to. \item (3) We will develop higher-level FFT-based applications driven primarily by the requirements of ECP applications projects, with the development of interfaces between FFTX and full ECP applications part of the co-design process. The strategy of having a library implementation of the FFTX interface will enable us to use the requirements of ECP applications for the design of the expanded FFTW interface and of the SPIRAL-based toolchain; in addition, the insights provided by opening up the design / tuning space for the constituent FFTs will lead to new ways of designing the applications solvers in order to obtain high performance. We will release the resulting integrated FFT-based packages as a library, called {\em SpectralPack}. \end{trivlist} The core code generation, symbolic analysis, and autotuning software for this project will be based on the open-source SPIRAL software stack, building on 20 years of research by the SPIRAL team at CMU. SPIRAL automatically maps computational kernels across a wide range of computing platforms to highly efficient code, and proves the correctness of the synthesized code. The SPIRAL approach has many of the same structural components as FFTW -- a high-level DSL, symbolic analysis, code generation, and autotuning. However, the approach used by SPIRAL integrates these ideas more closely into the user code, generating new source code for both the FFT calls and the glue code (e.g. pointwise operations, data motion) in an FFT-based application. \paragraph{Recent Progress} In the past six months, we did implementation of FFT use cases relevant to accelerator modeling and materials science using FFTX v1.0. We performed initial code generation for V100 GPU-based systems and baseline performance measurements on Summit. We released reference implementation of FFTX v1.0 for CPUs. \paragraph{Next Steps} Our future efforts will focus on the following areas: \begin{itemize} \item Deliver high performance on a single node GPU for ECP AD use cases. \item Use ``Plan of plans'' approach to integrate domain-specific operations with FFTs. \item Design detailed control of data placement / data motion. \item High-level C++ API to improve productivity, broaden adoption. \end{itemize} \section{Mathematics \label{SEC:support:math}} The mathematics support library is among the largest inside \phypp. It contains many functions, as well as a handful of useful constants. Some functionalities of this library are only available if you have installed the \texttt{fftw} and \texttt{gsl} libraries. If not, the specific functions that depend on these libraries will not be available, or will be slow, but the rest of the library will function properly. This library provides the following global constants: \begin{itemize} \item \cppinline{fnan}\itt{fnan} and \cppinline{dnan}\itt{dnan}. These are the \cppinline{float} and \cppinline{double} representation of the ``not-a-number'' (NaN) special value. This value is returned by some operations that are mathematically undefined in the real domain. For example, dividing zero by zero, or taking the square root of a negative number. NaN has some very peculiar properties that can surprise the newcomer. In particular, it propagates extremely fast, since any operation involving at least a NaN value will always return NaN, e.g., \cppinline{2.0 + fnan == fnan}). More troubling, any comparison operation involving a NaN will return \cppfalse, e.g., \cppinline{(10.0 < fnan) == false} and \cppinline{(10.0 >= fnan) == false} too. The only notable exception to this rule is that \cppinline{(fnan != fnan) == true}. Knowing this, NaN is a very useful return value to indicate that giving an actual value would not make sense. For example, in a galaxy catalog, some galaxies may have been observed at a certain wavelength, but not all of them. For those that are not observed, we do not know their flux. In this case, astronomers typically assign them a special, weird value, such as \cppinline{-99}. Using NaN in this case is clearer. \item \cppinline{fpi}\itt{fpi} and \cppinline{dpi}\itt{dpi}. This is the \cppinline{float} and \cppinline{double} closest representation of the number $\pi = 3.14159...$. \item \cppinline{finf}\itt{finf} and \cppinline{dinf}\itt{dinf}. This is the \cppinline{float} and \cppinline{double} representation of the positive infinity. The positive infinity is larger than any other finite value. \end{itemize} We now present the functions provided by this support library. One of the responsibilities of this library is to bring vectorized versions of standard mathematical functions that only work for scalar values. Since these functions are fairly common and well known, we will not describe their signature and behavior, and instead just list them here: \begin{itemize} \item exponentiation: \cppinline{sqrt}\itt{sqrt}, \cppinline{pow}\itt{pow}, \item trigonometry: \cppinline{cos}\itt{cos}, \cppinline{sin}\itt{sin}, \cppinline{tan}\itt{tan}, \cppinline{acos}\itt{acos}, \cppinline{asin}\itt{asin}, \cppinline{atan}\itt{atan}, \cppinline{atan2}\itt{atan2}, \cppinline{cosh}\itt{cosh}, \cppinline{sinh}\itt{sinh}, \cppinline{tanh}\itt{tanh}, \cppinline{acosh}\itt{acosh}, \cppinline{asinh}\itt{asinh}, \cppinline{atanh}\itt{atanh}, \item exponentials and logarithms: \cppinline{exp}\itt{exp}, \cppinline{log}\itt{log}, \cppinline{log2}\itt{log2}, \cppinline{log10}\itt{log10}, \item special functions: \cppinline{erf}\itt{erf}, \cppinline{erfc}\itt{erfc}, \cppinline{tgamma}\itt{tgamma}, \item rounding: \cppinline{ceil}\itt{ceil}, \cppinline{floor}\itt{floor}, \cppinline{round}\itt{round}, \item absolute value: \cppinline{abs}\itt{abs}. \end{itemize} We also introduce the functions \cppinline{bessel_j0}\itt{bessel_j0}, \cppinline{bessel_j1}\itt{bessel_j1}, \cppinline{bessel_y0}\itt{bessel_y0}, \cppinline{bessel_y1}\itt{bessel_y1}, \cppinline{bessel_i0}\itt{bessel_i0}, \cppinline{bessel_i1}\itt{bessel_i1}, \cppinline{bessel_k0}\itt{bessel_k0}, \cppinline{bessel_k1}\itt{bessel_k1}. The scalar version of the first four are provided by the C++ standard, while the last four are provided by the \texttt{gsl}. We now list the other, less common functions provided in this library. These are grouped by sections. \subsection{Low level mathematics \label{SEC:support:math:lowlevel}} \loadfunctions{functions_support_math_lowlevel.tex} \subsection{Sequences and bins \label{SEC:support:math:sequence}} \loadfunctions{functions_support_math_sequence.tex} \subsection{Randomization \label{SEC:support:math:random}} \loadfunctions{functions_support_math_random.tex} \subsection{Reduction \label{SEC:support:math:reduce}} \loadfunctions{functions_support_math_reduce.tex} \subsection{Interpolation \label{SEC:support:math:interp}} \loadfunctions{functions_support_math_interpol.tex} \subsection{Calculus \label{SEC:support:math:calculus}} \loadfunctions{functions_support_math_calculus.tex} \subsection{Algebra \label{SEC:support:math:algebra}} \loadfunctions{functions_support_math_algebra.tex} \subsection{Fitting \label{SEC:support:math:fit}} \loadfunctions{functions_support_math_fit.tex} \subsection{Geometry \label{SEC:support:math:geometry}} \loadfunctions{functions_support_math_geometry.tex} \subsection{Debug functions \label{SEC:support:math:debug}} \begin{itemize} \item \cppinline|void data_info(vec)| \itt{data_info} \item \cppinline|void mprint(vec<2,T>)| \itt{mprint} \end{itemize} \input{titlepage.tex} \section{GEN3D} GEN3D transforms a 2D finite element geometry database into a 3D database. SPONSOR: , 1562 (505) 844-2701 KEYWORDS: preprocessor, mesh\_generation \subsection{Documentation } ``GEN3D - A GENESIS Database 2D to 3D Transformation Program'' and , SAND89-0485, March 1989. ``Updates to the mesh generation program GEN3D,'' , 1521, dated April 11, 1990. \subsection{Unix} \begin{verbatim} gen3d [-help] [-options option] exo_in exo_out Options: -executable=exe Specifies alternate executable -help Prints short usage summary -MANUAL Prints extended usage summary -VMS Reads/Writes files in VMS format (Unicos only) \end{verbatim} \begin{verbatim} Path: The standard location for gen3d on Unix systems is: /usr/local/eng_sci/struct/etc/gen3d See the entry 'seacas usage' for an up-to-date list of SEACAS locations for all supported systems. \end{verbatim} \subsection{VMS} To execute GEN3D on VMS, type: \begin{verbatim} GEN3D 2d_database 3d_database [input_file] \end{verbatim} ``2d\_database'' is the filename of the input 2D GENESIS database. ``3d\_database'' is the filename of the output 3D GENESIS database. ``input\_file'' is the optional input file containing commands. Commands are read from the terminal keyboard if ``input\_file'' is not specified. \subsection{Commands} Mesh Transformation \begin{verbatim} TRANSLATE ROTATE WARP_P WARP_A SPLINE PROJECT SPLINE PROJECT TWIST INTERVAL SCALE CHANGE \end{verbatim} Mesh Orientation \begin{verbatim} REVOLVE REVCEN OFFSET MIRROR ZERO SHIFT \end{verbatim} Element Block Types \begin{verbatim} BLOCK TUNNEL CENTER NSETS SSETS CHANGE \end{verbatim} Information and Processing \begin{verbatim} SHOW LIST HELP END QUIT \end{verbatim} Undocumented and experimental \begin{verbatim} TRANSPLINE \end{verbatim} \subsubsection{TRANSLATE} \begin{verbatim} TRANSLATE {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} TRANSLATE causes the 2D mesh to be translated to create the 3D mesh. The number of levels is \{ntran\}, which is also the number of 3D elements derived from each input 2D element. The total range of the Z coordinate is \{tottran\} with a gradient of \{grad\}. The translation is always in the negative Z direction. This command supersedes previous transformation commands. The gradient affects the spacing of the levels. The displacement or thickness of level i is zi where: \begin{verbatim} z1 = tottran X grad - 1)/(grad^{ntran} - 1) if grad ne 1; tottran / ntran if grad = 1 zi = z1 X grad^{i-1} \end{verbatim} Multiple translation increments can be specified with a single translate command by repeating the \{ntran\}, \{tottran\}, and \{grad\} parameters on the command line. For example, the following command specifies two translation increments of thickness 1.0 for a total translation of 2.0: \begin{verbatim} TRANSLATE 5 1.0 0.5, 5 1.0 2.0 \end{verbatim} All increments must be specified with a single TRANSLATE command. \subsubsection{ROTATE} \begin{verbatim} ROTATE {nrot} <1>, {totdeg} <360.0>, {grad} <1.0>, {cenrot} <0.0> \end{verbatim} ROTATE causes the 2D mesh to be rotated to create the 3D mesh. The number of rotation levels is \{nrot\}, which is also the number of 3D elements derived from each input 2D element (with the exception of those affected by the CENTER command). The mesh is rotated a total of \{nrot\} rotations through a total arc of \{totdeg\} degrees. The angle of each rotation is equal to \{grad\} times the previous rotation. The center of rotation \{cenrot\} and the gradient \{grad\} are only meaningful if no center element blocks are defined (see the CENTER command). \paragraph{Gradient} The gradient affects the rotation of the levels. The angular rotation of level i is thetai where: \begin{verbatim} theta1 = totdeg X (grad - 1)/(grad^{nrot} - 1) if grad ne 1; totdeg / nrot if grad = 1 thetai = theta1 X grad^{i-1} \end{verbatim} Rotation is always counterclockwise. This command supersedes previous transformation commands. \subsubsection{WARP\_P} \begin{verbatim} WARP POINT {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, {radius} , {edge_type} \end{verbatim} WARP POINT causes the 2D mesh to be mapped onto a spherical surface to create the 3D mesh. The spherical surface has a radius of curvature equal to \{radius\}. The center of curvature is located on the z-axis, and it is a distance of \{radius\} above the X-Y plane. The number of levels is \{ntran\}, which is also the number of 3D elements derived from each input 2D element. The total thickness (measured radially) is \{tottran\} with a gradient of \{grad\}. Note that \{radius\} must be greater than the maximum distance from the z-axis to the boundary of the 2D mesh. This command supersedes previous transformation commands. The \{edge\_type\}, which can be either VERTICAL or RADIAL, determines how the created nodes are generated. If VERTICAL is selected, the x and y coordinates of the generated nodes are equal to the x and y coordinates of the original 2D node. If RADIAL is selected, the x and y coordinates of the generated nodes are calculated to lie on a line from the center of curvature (0.0, 0.0, \{radius\}) to the coordinates of the warped node (xw, yw, zw) where xw and yw are the coordinates of the original 2D node, and zw is determined such that the distance from the warped node to the center of curvature is equal to \{radius\}. Figure 2.1 in manual illustrates the VERTICAL edge type, and Figure 2.2 in manual illustrates the RADIAL edge type. The mesh transformation is performed in two parts. First, the warped nodal positions (xw, yw, zw) are calculated by mapping the original 2D mesh onto a spherical surface with a radius of curvature equal to \{radius\}. The original x and y coordinates of the 2D mesh remain at the same values; the z coordinate is calculated such that the distance to the center of curvature is equal to \{radius\}. \begin{verbatim} xw = x0 yw = y0 zw = radius - (radius^2 - x0^2 - y0^2)^{1/2} \end{verbatim} The warped nodal positions are projections parallel to the z-axis onto a spherical surface of radius \{radius\}; Figure 2.1 in manual illustrates this process. Then, the generated nodal positions are determined by translating either vertically or radially from the warped nodal position. A total of \{ntran\} translations are performed through a distance of \{tottran\} with a gradient of \{grad\}. Note that the thickness is measured radially for either \{edge\_type\}. The gradient affects the spacing of the levels. The thickness or length of level i is zi where: \begin{verbatim} z1 = tottran X grad - 1)/(grad^{ntran} - 1) if grad ne 1; tottran / ntran if grad = 1 zi = z1 X grad^{i-1} \end{verbatim} \subsubsection{WARP\_A} \begin{verbatim} WARP {axis} , {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, {radius} , {edge_type} \end{verbatim} This second form of the WARP command maps the 2D mesh to a cylindrical surface centered on the \{axis\}-axis to create the 3D mesh. The \{axis\} parameter must be either X or Y. The cylindrical surface has a radius of curvature equal to \{radius\}. The center of curvature is located a distance of \{radius\} above the X-Y plane. The number of levels is \{ntran\}, which is also the number of 3D elements derived from each input 2D element. The total thickness (measured radially) is \{tottran\} with a gradient of \{grad\}. This command supersedes previous transformation commands. The \{edge\_type\}, which can be either VERTICAL or RADIAL, determines how the created nodes are generated. If VERTICAL is selected, the x and y coordinates of the generated nodes are equal to the x and y coordinates of the projected 2D node. If RADIAL is selected, the x and y coordinates of the generated nodes are calculated to lie on a line from the center of curvature to the coordinates of the warped node (xw, yw, zw) where xw, yw, and zw are the coordinates of the mapped 2D node. The mesh transformation is performed in two parts. First, the warped nodal positions (xw, yw, zw) are calculated by mapping the original 2D mesh onto a cylinder about the \{axis\}-axis with a radius of curvature equal to \{radius\}. If \{axis\} is X, then the original X-coordinate remains at the same value. The generated Y and z coordinates are calculated such that the distance from the generated node to the X-Z plane measured along the cylindrical surface is equal to the X coordinate of the node in the 2D mesh. This is illustrated in Figure 2.2 in manual. If \{axis\} is Y, the X's and Y's are switched in the above discussion. Then, the generated nodal positions are determined by translating either vertically or radially from the warped nodal position. A total of \{ntran\} translations are performed through a distance of \{tottran\} with a gradient of \{grad\}. Note that the distance is measured radially for either \{edge\_type\}. The gradient affects the spacing of the levels. The thickness or length of level i is zi where: \begin{verbatim} z1 = tottran X grad - 1)/(grad^{ntran} - 1) if grad ne 1; tottran / ntran if grad = 1 zi = z1 X grad^{i-1} \end{verbatim} The resulting 3D mesh will have an cylindrical angle of (x\{max\}/radius) radians if warped about the Y axis, or (y\{max\}/radius) radians if warped about the X axis, where x\{max\} and y\{max\} are the maximum X and Y coordinates in the 2D mesh. \subsubsection{SPLINE} \begin{verbatim} SPLINE {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} where \{ntran\} is the number of levels, \{tottran\} is the total transformation distance (thickness), and \{grad\} is the gradient which affects the spacing of the levels (see page 21 in Reference GEN3D for an explanation of the gradient). Multiple transformation increments can be specified with a single SPLINE command by repeating the \{ntran\}, \{tottran\}, \{grad\} parameters on a single line. Note that the actual thickness of the generated mesh is determined by the input front and back spline surfaces; therefore, the distances entered as \{tottran\} are the proportional distance of the segments. For example, if the following command was entered: \begin{verbatim} SPLINE 2 1.0 0.5 4 2.0 1.0 2 1.0 2.0 \end{verbatim} then, segments 1 and 3 would each be 25\% of the total thickness, and segment 2 would be 50\% of the total thickness. Following the SPLINE command line, GEN3D enters the spline input mode in which the various spline options described below can be entered. \paragraph{LINEAR} \begin{verbatim} LINEAR \end{verbatim} the spline data are input as Radius-Z data pairs, and the slopes at the end of the curves are linear slopes. \paragraph{ANGULAR} \begin{verbatim} ANGULAR \end{verbatim} the spline data are input as Theta(degrees)-Distance data pairs, where Theta is the angle of the line between the origin (X = Y = Z = 0) and the defined point and the Distance is the length of this curve. The slopes at the end of the curves are relative to this curve. \paragraph{ANGULAR} \begin{verbatim} FRONT ! FRONT \end{verbatim} the curve data and slope specifications up to the next BACK, END, or EXIT command will the FRONT spline. The front surface Z values are greater (more positive) than the back surface Z values. \paragraph{BACK} \begin{verbatim} BACK \end{verbatim} the curve data and slope specifications up to the next FRONT, END, or EXIT command will the BACK spline. The front surface Z values are greater (more positive) than the back surface Z values. \paragraph{LEFT} \begin{verbatim} LEFT {slope} \end{verbatim} the parameter \{slope\} specifies the slope of the spline curve at the LEFT end of the curve. The slope is measured in the same units specified in the ANGULAR or LINEAR command. If the slope is not specified, the end conditions of the curve will be set such that the second derivative is equal to zero which is the so-called natural cubic spline. \paragraph{RIGHT} \begin{verbatim} RIGHT {slope} \end{verbatim} the parameter \{slope\} specifies the slope of the spline curve at the RIGHT end of the curve. The slope is measured in the same units specified in the ANGULAR or LINEAR command. If the slope is not specified, the end conditions of the curve will be set such that the second derivative is equal to zero which is the so-called em natural cubic spline. \paragraph{EXIT} \begin{verbatim} EXIT or END \end{verbatim} terminate spline input mode and return to general GEN3D command processing. \paragraph{See Also} \begin{verbatim} Memo: "Updates to the mesh generation program GEN3D," , 1521, dated April 11, 1990. See the author for copies. \end{verbatim} \subsubsection{PROJECT} \begin{verbatim} PROJECT {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} where \{ntran\} is the number of levels, \{tottran\} is the total transformation distance (thickness), and \{grad\} is the gradient which affects the spacing of the levels (see page 21 in Reference GEN3D for an explanation of the gradient). Multiple transformation increments can be specified with a single PROJECT command by repeating the \{ntran\}, \{tottran\}, \{grad\} parameters on a single line. \paragraph{NORMAL} \begin{verbatim} NORMAL {x_normal} <0.0>, {y_normal} <0.0>, {z_normal} <0.0> PLANE {x_normal} <0.0>, {y_normal} <0.0>, {z_normal} <0.0> \end{verbatim} NORMAL or PLANE the normal vector to the back surface. The \{x\_normal\}, \{y\_normal\}, and \{z\_normal\} parameters are vector components of the normal vector. The front surface will be projected onto the plane with the specified normal vector. Because of the way GEN3D generates the ddd mesh, the \{z\_normal\} component of the vector must be negative; if a positive value of \{z\_normal\} is entered, all of the components will be multiplied by negative one.footnoteThe test version of GEN3D did not do this reversal correctly; instead it always made \{z\_normal\} negative which made it confusing to determine the surface orientation. This bug has been fixed; however, old input files will now give an incorrect orientation. To use your old input files, either correct the normal vector, or put the keyword DOOLDWAY after the \{z\_normal\} component. NORMAL or PLANE supersede previous WARP commands. \paragraph{WARP} \begin{verbatim} WARP {distance} {CONVEX|CONCAVE} \end{verbatim} WARP projects the front surface (original dd mesh) onto a spherical surface with a radius of \{distance\}. If CONVEX is specified, the generated mesh will have a spherical or bulbous surface; if CONCAVE is specified, the generated mesh will have a dimpled surface. The distance \{tottran\} specified in the PROJECT command is measured to the center of the spherical surface. WARP supersedes previous NORMAL or PLANE commands. Note that the SPLINE transformation can be used if a non-spherical surface is required. \paragraph{SCALE} \begin{verbatim} SCALE {x_scale} <1.0>, {y_scale} <1.0> \end{verbatim} SCALE multiplies the X and Y coordinates of the projected surface by the respective scale factors. See the SCLCEN command for the equations used to transform the coordinates. \paragraph{SCLCEN} \begin{verbatim} SCLCEN {x_cen} <0.0>, {y_cen} <0.0> \end{verbatim} SCLCEN specifies the origin of the coordinate system for scaling. This is best illustrated by examining the equations used in the transformation: \begin{verbatim} x_new = (x_old - x_cen) X x_scale + x_cen y_new = (y_old - y_cen) X y_scale + y_cen \end{verbatim} \paragraph{OFFSET} \begin{verbatim} OFFSET {x_offset} <0.0>, {y_offset} <0.0> \end{verbatim} OFFSET offsets the nodal coordinates of the projected surface by the specified \{x\_offset\} and \{y\_offset\} which shifts the back surface with respect to the front surface. The offsets are performed after the surface has been projected onto the plane or spherical surface. \paragraph{RESET} \begin{verbatim} RESET \end{verbatim} RESET resets all parameters to their default values. \paragraph{EXIT} \begin{verbatim} EXIT or END \end{verbatim} EXIT or END terminates the PROJECT input mode and returns to normal GEN3D command processing. \paragraph{See Also} \begin{verbatim} Memo: "Updates to the mesh generation program GEN3D," , 1521, dated April 11, 1990. See the author for copies. \end{verbatim} \subsubsection{TWIST} \begin{verbatim} TWIST {twangl} <0.0>, {twxcen} <0.0>, {twycen} <0.0>, {TRANSLATE}, {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} \begin{verbatim} TWIST {twangl} <0.0>, {twxcen} <0.0>, {twycen} <0.0>, {Rotate}, {cenrot} <0.0>, {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} where \{twangl\} is the rotational offset in degrees of the front surface with respect to the back surface, \{twxcen\} and \{twxcen\} specify the center of the twist rotation, \{ntran\} is the number of levels, \{tottran\} is the total transformation distance (distance for the TRANSLATE option, or degrees for the ROTATE option), and \{grad\} is the gradient which affects the spacing of the levels (see page 21 in Reference GEN3D for an explanation of the gradient). Multiple transformation increments can be specified by repeating the \{ntran\}, \{tottran\}, \{grad\} parameters on a single line. The twist angle per level is determined by dividing \{twangl\} by the total number of translation increments \{ntran\}, not by the total translation/rotation distance \{tottran\}. Note that this means that the incremental twist per level depends on the gradient. \subsubsection{INTERVAL} \begin{verbatim} INTERVAL {ntran} <1>, {tottran} <1.0>, {grad} <1.0>, ... \end{verbatim} where \{ntran\} is the number of levels, \{tottran\} is the total transformation distance in units applicable to the currently active transformation option, and \{grad\} is the gradient which affects the spacing of the levels (see page 21 in Reference GEN3D for an explanation of the gradient). Multiple transformation increments can be specified with a single INTERVAL command by repeating the \{ntran\}, \{tottran\}, \{grad\} parameters on a single line. \subsubsection{SCALE} \begin{verbatim} SCALE {axis_1}, {scale_1}, {...} SCALE ALL {scale} SCALE RESET \end{verbatim} where \{axis\_i\} is either X, Y, or Z; and \{scale\_i\} is the factor by which the specified axis coordinates are multiplied. If ALL is specified, all three coordinates are multiplied by \{scale\}. Scaling can be reset by the command SCALE RESET. \subsubsection{CHANGE} \begin{verbatim} CHANGE {Material|Nodeset|Sideset} {old_id} {new_id} \end{verbatim} where \{old\_id\} is an existing ID in the original mesh, and \{new\_id\} is an ID that does not exist in the original mesh. Note that you cannot combine IDs using this command; the ID \{new\_id\} must not exist in the original mesh, and it must not match an ID created using the NODESET$|$SIDESET FRONT$|$BACK commands or the TUNNEL commands. \subsubsection{REVOLVE} \begin{verbatim} REVOLVE {axis_{1}}, {ndeg_{1}}, {axis_{2}}, {ndeg_{2}}, ... REVOLVE RESET \end{verbatim} REVOLVE causes the transformed 3D mesh to be rotated. Each (\{axis\}, \{ndeg\}) parameter pair specifies an axis (X or Y or Z) and the number of degrees to rotate. The axis refers to the ``viewing'' axis, not to the object axis. The rotations are according to right-hand rule. The center of the rotation is specified by the REVCEN command. Revolutions are cumulative; however, only one center of revolution may be specified. The REVOLVE RESET command resets to no rotation. \subsubsection{REVCEN} \begin{verbatim} REVCEN {xcen} <2D minimum X coordinate>, {ycen} <2D minimum Y coordinate>, {zcen} <0.0> \end{verbatim} REVCEN sets the center of revolution for the REVOLVE command to the point (\{xcen\},\{ycen\},\{zcen\}). \subsubsection{SHIFT} \begin{verbatim} SHIFT [ADD] {axis_{1}}, {offset_{1}}, {axis_{2}}, {offset_{2}}, ... SHIFT ALL {offset} <0.0> SHIFT RESET SHIFT {xoff} <0.0>, {yoff} <0.0>, {zoff} <0.0> \end{verbatim} (NOTE: SHIFT is a synonym for OFFSET) SHIFT specifies offsets to be added to the coordinates. If a REVOLVE command has been issued, the mesh is rotated before it is offset. The last form of the offset command is included to maintain compatibility with the offset command in GEN3D. SHIFT ALL offsets all of the coordinates by the specified offset, and SHIFT RESET resets the offsets to zero. Offsets are not cumulative unless ADD is specified, that is, if SHIFT X 0.5 X 1.0 is entered, the X coordinates will be offset by 1.0. if SHIFT ADD X 0.5 X 1.0 is entered, the X coordinates will be offset by 1.5. \subsubsection{OFFSET} \begin{verbatim} OFFSET [ADD] {axis_{1}}, {offset_{1}}, {axis_{2}}, {offset_{2}}, ... OFFSET ALL {offset} <0.0> OFFSET RESET OFFSET {xoff} <0.0>, {yoff} <0.0>, {zoff} <0.0> \end{verbatim} (NOTE: OFFSET is a synonym for SHIFT) OFFSET specifies offsets to be added to the coordinates. If a REVOLVE command has been issued, the mesh is rotated before it is offset. The last form of the offset command is included to maintain compatibility with the offset command in GEN3D. OFFSET ALL offsets all of the coordinates by the specified offset, and OFFSET RESET resets the offsets to zero. Offsets are not cumulative unless ADD is specified, that is, if OFFSET X 0.5 X 1.0 is entered, the X coordinates will be offset by 1.0. if OFFSET ADD X 0.5 X 1.0 is entered, the X coordinates will be offset by 1.5. \subsubsection{MIRROR} \begin{verbatim} MIRROR {axis_{1}}, {axis_{2}}, ... MIRROR RESET \end{verbatim} MIRROR causes the transformed 3D mesh to be reflected about a coordinate plane. Each \{axis\} parameter specifies an axis (X or Y or Z) which is the normal to the reflection plane. Reflections are performed after the mesh has been repositioned by the REVOLVE and OFFSET commands. The MIRROR RESET command resets to no reflection. Reflections are not cumulative, that is, if MIRROR X Y X is entered, only one reflection about the X axis will be performed. If an odd number of reflections are performed, the element connectivity and the sideset face numberings will be correctly reordered. \subsubsection{ZERO} \begin{verbatim} ZERO {axis_1}, {min_1}, {axis_2}, {min_2}, ... ZERO RESET \end{verbatim} ZERO sets all \{axis\_i\} coordinates with an absolute value less than \{min\_i\} equal to zero. The ZERO RESET command resets to no automatic zeroing. This command is used to zero nodal coordinates that should be equal to zero, but due to roundoff errors they have slightly nonzero values. \subsubsection{BLOCK} \begin{verbatim} BLOCK {block_id_{1}}, {block_id_{2}}, ... \end{verbatim} BLOCK defines the specified element blocks as normal blocks. This command supersedes any previous TUNNEL or CENTER commands. \subsubsection{TUNNEL} \begin{verbatim} TUNNEL {block_id} , {start} <1>, {end} , {inc} <1> \end{verbatim} TUNNEL defines the specified element block as a tunnel block. A TRANSLATE command must be in effect before this command is issued. If a ROTATE command is issued, all tunnel blocks are changed to normal blocks. For each tunnel block, a separate 3D element block is created starting at level \{start\}, with each block having \{inc\} levels. Any levels after level \{end\} are put in a single block. For example, the commands \begin{verbatim} TRANSLATE 15, 15.0 TUNNEL 999, 5, 9, 2 \end{verbatim} create five blocks consisting of the following 3D elements (derived from the 2D elements of element block 999): \begin{verbatim} 1) the elements in levels 1, 2, 3, and 4, 2) the elements in levels 5 and 6, 3) the elements in levels 7 and 8, 4) the elements in level 9, 5) the elements in levels 10, 11, 12, 13, 14, and 15. \end{verbatim} The block identifier of the first block is always \{block\_id\}. The new blocks are assigned consecutive identifiers greater than the maximum existing (and new) identifier. \subsubsection{CENTER} \begin{verbatim} CENTER {block_id_{1}}, {block_id_{2}}, ... \end{verbatim} CENTER defines the specified element blocks as center blocks. A ROTATE command must be in effect before this command is issued. The mesh must be rotated a complete quadrant (90, 180, 270 or 360 degrees) and the number of rotation levels must be a multiple of 2 for each 90 degrees of rotation. If em nrot is the number of rotations, there must be at least em nrot/2 elements along the X axis in the center block. If a TRANSLATE command is issued, all center blocks are changed to normal blocks. If center blocks are defined, the center of rotation defined by the ROTATE command is ignored. The center of rotation is the minimum coordinate of all elements in the center blocks. \subsubsection{NSETS} \begin{verbatim} NSETS FRONT or BACK , {set_id_{1}}, {set_id_{2}}, ... \end{verbatim} NSETS defines front or back node sets with the given identifiers. The identifiers must be unique from existing node set identifiers and previously defined front and back node set identifiers. Back sets cannot be defined on a 360-degree rotation. \subsubsection{SSETS} \begin{verbatim} SSETS FRONT or BACK , {set_id_{1}}, {set_id_{2}}, ... \end{verbatim} SSETS is equivalent to the NSETS command except that it defines side sets. \subsubsection{SHOW} \begin{verbatim} SHOW {command} \end{verbatim} SHOW displays the settings of parameters relevant to the \{command\}. For example, the command SHOW BLOCK displays information about all the element blocks. \subsubsection{LIST} \begin{verbatim} LIST VARS \end{verbatim} LIST VARS displays a summary of the input database. The summary includes the database title; the number of nodes, elements, and element blocks; and the number of node sets and side sets. \subsubsection{HELP} \begin{verbatim} HELP {command} \end{verbatim} HELP displays information about the program command given as the parameter. If no parameter is given, all the command verbs are displayed. This command is system-dependent and may not be available on some systems. \subsubsection{END} \begin{verbatim} END \end{verbatim} END ends the command input and starts the database transformation. \subsubsection{QUIT} QUIT ends the command input and exits the program immediately without writing an output database. \subsection{Problems/Bugs} \begin{verbatim} CENTER BLOCKS -- Incorrect mesh will be generated if center blocks are not contiguous (ie. contain slidelines). Problem will appear at junction of pieces connected by slideline. The nodes on the of the piece will not be generated correctly. \end{verbatim} \begin{verbatim} FIX: Generate each piece separately and join together with GJOIN Problem involves changing several routines and has only been reported once in two years. Sorry, \end{verbatim} \subsection{ChangeLog} \begin{verbatim} DATA (QAINFO(I), I=1,3) / * 'GEN3D ', * '07/16/90', * 'X1.02.03' / C Version : Date Time Modifier C ------- ---- ---- -------- C X1.01.05: --/--/-- --:--:-- C : Fixed orientation of NORMAL vector in GETPRO. Added option C : DOOLDWAY to maintain compatibility with previous bug version. C X1.01.05: --/--/-- --:--:-- 1 C : Initial compilation of TRANSLATE SPLINE routine C X1.01.06: --/--/-- --:--:-- C : Modify SPTIN2 - total Z distance given by spline coordinates C X1.01.06: --/--/-- --:--:-- C : Recompile with no debug C X1.01.06: --/--/-- --:--:-- 1 C : Fix problem in RDSPLN - increased MAXFLD to 4 C X1.01.07: 03/21/90 16:23:19 GREG SJAARDEMA1 C : Reorganized SPLINE transformation, read from sys$input C X1.01.08: 03/22/90 07:31:23 GREG SJAARDEMA1 C : Added A to COMAND call list to fix bug in SPLINE transformation C X1.01.09: 03/22/90 08:39:58 GREG SJAARDEMA1 C : Fixed SPLXYZ - Added INIGRD lines to set ZCORD C X1.01.10: 03/22/90 09:01:10 GREG SJAARDEMA1 C : Modify SPLINE input routine (GETSPL) to use Amy's FREFLD routines C X1.01.11: 03/27/90 08:44:46 GREG SJAARDEMA1 C : TRANSPLINE now reads from standard input, not file C X1.01.12: 03/27/90 12:26:46 GREG SJAARDEMA1 C : Installed as XGEN3D in XACCESS. Recompile only C X1.01.13: 03/27/90 13:05:47 GREG SJAARDEMA1 C : Reordered SPTXYZ for efficiency C X1.01.14: 03/29/90 15:49:40 GREG SJAARDEMA1 C : Fixed problem with 360 ROTATE - ndegr not set, X version only C X1.02.00: 04/12/90 07:43:50 GREG SJAARDEMA1 C : Install SPLINE version in XACCESS. No other changes C X1.02.01: 07/06/90 14:08:02 2 C : Problem with noncontiguous material blocks in center C : blocks. (ie. Slidelines separating materials.) C : Problem not fixed, error checking enabled. C : Problem is in setting node numbers in NPCEN (sub: FNPCEN) C X1.02.02: 07/09/90 14:32:11 GREG SJAARDEMA2 C : Fixed Typo in FNPCEN -- Second contiguous block warning C : was IBOT, now ITOP C X1.02.03: 07/16/90 15:44:30 GREG SJAARDEMA4 C : Added SHIFT as synonym for OFFSET, C : Added OFFSET/SHIFT RESET/ADD/ALL/X/Y/Z -- now same as GREPOS \end{verbatim} \subsection{QA} \begin{verbatim} VERSION X1.02.03 \end{verbatim} Code Goal/Status: (See HELP 1520\_CODES QA for more information) \begin{verbatim} Category Goal Status Comment ------------- ------ ------ ------- Documentation: Formal Formal See Sponsor: X X Traceability: X X Retrievability: X X Verification: X X Archived: X Maintainability: X X See Help Entry: X X Executable: X X \end{verbatim} \begin{verbatim} Source Code Location: $CVSROOT/ACCESS/prepost/gen3d Archive Location: --- Not yet Archived --- \end{verbatim} \subsubsection{Requirements} GEN3D Requires the following Utilities/Files: \begin{verbatim} o COMMON.OLB Library of common routines for reading/writing EXODUS files and other support routines. o SUPES.OLB Memory management and Free-field input routines o CPUIFC.OLB Interrupt processing (Control-C). Pseudo, non- functional routine provided for non-VMS systems \end{verbatim} \subsection{See\_Also} \begin{verbatim} o FASTQ: Generate two-dimensional GENESIS mesh file o GJOIN: Merge two or more mesh files into single file o GREPOS: Reposition/Modify mesh file o NUMBERS: Calculate several parameters of a mesh file o APREPRO: Algebraic preprocessor for input files. o BLOT: Plot mesh files. \end{verbatim} \end{document} % % Machine Learning Course Homework1 % Only available in ZJU % \documentclass[12pt,twoside]{article} \input{macros} \usepackage{amsmath} \usepackage{url} \usepackage{mdwlist} \usepackage{graphicx} \usepackage{clrscode3e} \newcommand{\isnotequal}{\mathrel{\scalebox{0.8}[1]{!}\hspace*{1pt}\scalebox{0.8}[1]{=}}} \usepackage{listings} \usepackage{tikz} \usetikzlibrary{arrows} \usetikzlibrary{matrix} \usetikzlibrary{positioning} \usetikzlibrary{shapes.geometric} \usetikzlibrary{shapes.misc} \usetikzlibrary{trees} \newcommand{\answer}{ \par\medskip \textbf{Answer:} } \newcommand{\collaborators}{ \textbf{Collaborators:} %%% COLLABORATORS START %%% \tabT Name: \tabT Student ID: 3160104340 %%% COLLABORATORS END %%% } \newcommand{\answerIa}{ \answer %%% PROBLEM 1(a) ANSWER START %%% \begin{enumerate} \item Given the size of testing set 1000:\\ When the size of training set is 10, training error rate: 0, testing error rate: 10.9\%;\\ When the size of training set is 100, training error rate: 0.146\%, testing error rate： 1.38\%. \item Given the size of testing set 1000:\\ When the size of training set is 10, average number of iterations: 52;\\ When the size of training set is 100, average number of iterations: 2241. \item Given the size of testing set 1000:\\ The Perceptron Machine will iterate infinitely if the maximum number of iterations is not defined.\\ The training error cannot reach zero, and the testing error is much higher than that in linearly separatable cases. \end{enumerate} %%% PROBLEM 1(\item d) ANSWER END %%% } \newcommand{\answerIb}{ \answer %%% PROBLEM 1(b) ANSWER START %%% \begin{enumerate} \item Given the size of testing set 1000:\\ The training error rate: 3.89\%, the testing error rate: 4.68\%. \item Given the size of testing set 1000:\\ The training error rate: 13.1\%, the testing error rate: 14.3\%. \item The training error rate: 49.0\%, the testing error rate: 55.0\%. \item The training error rate: 5.00\%, the testing error rate: 6.60\%. \end{enumerate} %%% PROBLEM 1(b) ANSWER END %%% } \newcommand{\answerIc}{ \answer %%% PROBLEM 1(c) ANSWER START %%% Here is the breif derivation of the logistic gradient: As the question suggests, the likelihood function can be presented as: $$P(y|x, \omega) = \hat{y}^y*(1-\hat{y})^{1-y}$$ where $\hat{y} = \frac{1}{1+\exp{[-\omega^Tx]}}$, so the total-log-likelihood function is: \begin{displaymath} \begin{aligned} L &= \sum\limits_{i=1}^{m}l_i\\ &= \sum\limits_{i=1}^{m}y^{(i)}\log{\hat{y^{(i)}}}+(1-y^{(i)})\log{(1-\hat{y^{(i)}})} \end{aligned} \end{displaymath} where $m$ is the number of samples. Thus we can define the loss function as the minus log-likelihood function, so the gradient is: \begin{displaymath} \begin{aligned} \nabla_\omega J &= -\frac{\partial}{\partial\omega}L\\ &= \sum\limits_{i=1}^{m}\frac{\hat{y^{(i)}} - y^{(i)}}{\hat{y^{(i)}}(1-\hat{y^{(i)}})}*\frac{\partial\hat{y^{(i)}}}{\partial \omega}\\ &= \sum\limits_{i=1}^{m}x^{(i)}*(\hat{y^{(i)}} - y^{(i)})\\ &= [\hat{y} - y]X^T \end{aligned} \end{displaymath} Based on this: \begin{enumerate} \item Given the size of testing set 1000:\\ The training error rate: 0.29\%, the testing error rate: 1.09\%. \item Given the size of testing set 10000:\\ The training error rate: 11.8\%, the testing error rate: 12.8\%. \end{enumerate} %%% PROBLEM 1(c) ANSWER END %%% } \newcommand{\answerId}{ \answer %%% PROBLEM 1(d) ANSWER START %%% \begin{enumerate} \item Given the size of testing set 1000:\\ The training error rate: 0, the testing error rate: 3.36\%. \item Given the size of testing set 1000:\\ The training error rate: 0, the testing error rate: 1.03\%. \item Average number of support vectors: 3.5. \end{enumerate} %%% PROBLEM 1(d) ANSWER END %%% } \newcommand{\answerIIa}{ \answer %%% PROBLEM 2(a) ANSWER START %%% \begin{enumerate} \item According to LOOCV, the optimal lambda is 10. \item When $\lambda=0$, $\sum\limits_{i=0}^{P}\omega_i^2=14.6$; when $\lambda=10$, $\sum\limits_{i=0}^{P}\omega_i^2=1.23$. \item When $\lambda=0$, the training error is 0 and the testing error is 10.7\%; When $\lambda=10$, the training error is 0 and the testing error is 7.23\%. \end{enumerate} %%% PROBLEM 2(a) ANSWER END %%% } \newcommand{\answerIIb}{ \answer %%% PROBLEM 2(b) ANSWER START %%% According to LOOCV, the optimal lambda is 0.001.\\ When $\lambda=0$, the training error is 0 and the testing error is 6.73\%; When $\lambda=0.001$, the training error is 0 and the testing error is 6.58\%. %%% PROBLEM 2(b) ANSWER END %%% } \newcommand{\answerIIIa}{ \answer %%% PROBLEM 3(a) ANSWER START %%% \begin{enumerate} \item False. Once the model is defined, the bias is defined, too. It doesn't work no matter how many samples are added if the model is over-simple. \item False. We don't like neither high-bias and high-variance. Although models with high variance can fit the training data perfectly, they don't generalize to testing data very well. However, high-variance problems can be tackled by adding more training samples, which is likely to get further improvement compared with high-bias cases. \item True. When the parameters get more, there's a greater chance that training data are not enough, thus creating too complicated models. \item False. Regularization is used to mitigate over-fitting, where testing error will be reduced but training error will not necessarily. On the contrary, the training error will probably increase a little bit because some mistakes may be allowed. \item False. By using very large $\lambda$, the over-fitting penalty gets extremely huge, but the high-bias penalty gets relatively small. Therefore, models with large $\lambda$ usually suffer from high-bias problem. \end{enumerate} %%% PROBLEM 3(a) ANSWER END %%% } \setlength{\oddsidemargin}{0pt} \setlength{\evensidemargin}{0pt} \setlength{\textwidth}{6.5in} \setlength{\topmargin}{0in} \setlength{\textheight}{8.5in} % Fill these in! \newcommand{\theproblemsetnum}{2} \newcommand{\releasedate}{May 16, 2019} \newcommand{\partaduedate}{May 30, 2019} \newcommand{\tabUnit}{3ex} \newcommand{\tabT}{\hspace*{\tabUnit}} \begin{document} \handout{Homework \theproblemsetnum}{\releasedate} \collaborators % Please download the .zip archive for this problem set, and refer to the % hw2.pdf file for instructions on preparing your solutions. % \medskip \hrulefill \begin{problems} \problem \textbf{A Walk Through Linear Models} \begin{problemparts} \problempart Perceptron \answerIa \problempart Linear Regression \answerIb \problempart Logistic Regression \answerIc \problempart Support Vector Machine \answerId \end{problemparts} \newpage \problem \textbf{Regularization and Cross-Validation} \begin{problemparts} \problempart Implement Ridge Regression and use LOOCV to tune... \answerIIa \problempart Implement Logistic Regression and use LOOCV to tune... \answerIIb \end{problemparts} \newpage \problem \textbf{Bias Variance Trade-off} \begin {problemparts} \problempart True or False: \answerIIIa \end{problemparts} \end{problems} \end{document} 0 \subsection{Binary functions} \subsubsection{Properties of binary functions} Binary functions can be written as: \(f(a,b)=a\oplus b\) A function is commutative if: \(x\oplus y = y\oplus x\) A function is associative if: \((x\oplus y)\oplus z = x\oplus (y\oplus z)\) A function \(\otimes \) is left distributive over \(\oplus \) if: \(x\otimes (y\oplus z)=(x\otimes y) \oplus (x\otimes z)\) Alternatively, function \(\otimes \) is right distributive over \(\oplus \) if: \((x\oplus y)\otimes z=(x\otimes z) \oplus (y\oplus z)\) A function is distributive over another function if it both left and right distributive over it. code_source/latex/SDL__arduino_8c.tex \hypertarget{SDL__arduino_8c}{}\section{S\+D\+L\+\_\+arduino.\+c File Reference} \label{SDL__arduino_8c}\index{SDL\_arduino.c@{SDL\_arduino.c}} {\ttfamily \#include \char`\"{}S\+D\+L\+\_\+arduino.\+h\char`\"{}}\newline Include dependency graph for S\+D\+L\+\_\+arduino.\+c\+: % FIG 0 \subsection*{Functions} \begin{DoxyCompactItemize} \item int \mbox{\hyperlink{SDL__arduino_8c_abefc6906065c0c43ba99477e403e07d9}{event\+Handler\+Arduino}} (\mbox{\hyperlink{structhero}{hero}} $\ast$player, S\+D\+L\+\_\+\+Surface $\ast$calque\+\_\+game) \item void \mbox{\hyperlink{SDL__arduino_8c_af921e7b2ce75aadfb16b2ce49424a589}{turn\+On\+Off}} (int e1, int e2) \item char \mbox{\hyperlink{SDL__arduino_8c_aca0b6081e0acbb8178c9eaba3049358f}{read\+From\+Ard}} () \item int \mbox{\hyperlink{SDL__arduino_8c_a803a9741486484988283cd8a17ca8a8c}{write\+To\+Ard}} (int x) \end{DoxyCompactItemize} \subsection{Function Documentation} \mbox{\Hypertarget{SDL__arduino_8c_abefc6906065c0c43ba99477e403e07d9}\label{SDL__arduino_8c_abefc6906065c0c43ba99477e403e07d9}} \index{SDL\_arduino.c@{SDL\_arduino.c}!eventHandlerArduino@{eventHandlerArduino}} \index{eventHandlerArduino@{eventHandlerArduino}!SDL\_arduino.c@{SDL\_arduino.c}} \subsubsection{\texorpdfstring{eventHandlerArduino()}{eventHandlerArduino()}} {\footnotesize\ttfamily int event\+Handler\+Arduino (\begin{DoxyParamCaption}\item[{\mbox{\hyperlink{structhero}{hero}} $\ast$}]{player, }\item[{S\+D\+L\+\_\+\+Surface $\ast$}]{calque\+\_\+game }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 1 \mbox{\Hypertarget{SDL__arduino_8c_aca0b6081e0acbb8178c9eaba3049358f}\label{SDL__arduino_8c_aca0b6081e0acbb8178c9eaba3049358f}} \index{SDL\_arduino.c@{SDL\_arduino.c}!readFromArd@{readFromArd}} \index{readFromArd@{readFromArd}!SDL\_arduino.c@{SDL\_arduino.c}} \subsubsection{\texorpdfstring{readFromArd()}{readFromArd()}} {\footnotesize\ttfamily char read\+From\+Ard (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{SDL__arduino_8c_af921e7b2ce75aadfb16b2ce49424a589}\label{SDL__arduino_8c_af921e7b2ce75aadfb16b2ce49424a589}} \index{SDL\_arduino.c@{SDL\_arduino.c}!turnOnOff@{turnOnOff}} \index{turnOnOff@{turnOnOff}!SDL\_arduino.c@{SDL\_arduino.c}} \subsubsection{\texorpdfstring{turnOnOff()}{turnOnOff()}} {\footnotesize\ttfamily void turn\+On\+Off (\begin{DoxyParamCaption}\item[{int}]{e1, }\item[{int}]{e2 }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 2 \mbox{\Hypertarget{SDL__arduino_8c_a803a9741486484988283cd8a17ca8a8c}\label{SDL__arduino_8c_a803a9741486484988283cd8a17ca8a8c}} \index{SDL\_arduino.c@{SDL\_arduino.c}!writeToArd@{writeToArd}} \index{writeToArd@{writeToArd}!SDL\_arduino.c@{SDL\_arduino.c}} \subsubsection{\texorpdfstring{writeToArd()}{writeToArd()}} {\footnotesize\ttfamily int write\+To\+Ard (\begin{DoxyParamCaption}\item[{int}]{x }\end{DoxyParamCaption})} 0 \begin{tiny}(Ege12)\end{tiny} Un repère étant fixé, trouver l'équation du plan symétrique du plan d'équation $z=0$ par rapport au plan d'équation $x+y-z=1$. 1-10 % ------------- MY COMMANDS ------------- % % Update list item colour % http://tex.stackexchange.com/questions/110101/how-to-change-the-color-of-bullets-in-a-list \newcommand{\itemcolor}[1]{ \renewcommand{\makelabel}[1]{\color{#1}\hfil ##1}} % Use this to embed contents in the middle of slides where some slides are greyed out \newcommand{\B}{{\bf B}} \newcommand{\J}{{\bf J}} \newcommand{\E}{{\bf E}} \newcommand{\F}{{\bf F}} \newcommand{\D}{{\bf D}} \renewcommand{\r}{{\bf r}} \renewcommand{\t}{{\bf t}} \renewcommand{\H}{{\bf H}} \newcommand{\courtesy}[1]{\begin{center}{\tiny [#1]}\end{center}} \newcommand{\showsection}{ALL} \newcommand{\tocitem}[2]{% \ifthenelse{\equal{\showsection}{ALL}} {\item #2} {\ifthenelse{\equal{\showsection}{#1}} {\item #2} {\itemcolor{gray} \item {\color{gray} #2} \itemcolor{title}} } } \newcommand{\subtoc}[2]{\ifthenelse{\equal{\showsection}{#1}}{#2}{}} % to show something only on extended slides \newcommand{\extended}[1]{#1} %\newcommand{\extended}[1]{} \mdfdefinestyle{taustyle}{% rightline=true, innerleftmargin=10pt, innerrightmargin=10pt, outerlinewidth=2pt, outerlinecolor=title, backgroundcolor=hlbg, topline=false, rightline=false, bottomline=false, skipabove=0, skipbelow=0 } \newcommand{\tauBox}[1]{ \vspace*{2mm} \begin{mdframed}[style=taustyle]#1\end{mdframed} \vspace*{2mm} } \newcommand{\tauEmph}[1]{% {\bf\color{title}{% #1% }}% } thesis_cover.tex0 \documentclass[master,nowatermark]{NTHUthesis} \usepackage[utf8]{inputenc} %%% Necessary %%% \input{thesis_info} \begin{document} \makecover \end{document}0 % An optional dedication: you can thank whomever you wish (your supervisor, % consultant, a person who lent the software, etc.) \def\Dedication{% I wish to thank my supervisor, Mgr. , Ph.D., for his guidance, help, and mostly patience during my work on this thesis. I am thankful for his friendly attitude and valuable advice. I would also want to express my sincere gratitude to my family and friends, whose support was paramount for successful accomplishment of this thesis. Computational resources were supplied by the project "e-Infrastruktura CZ" (e-INFRA LM2018140) provided within the program Projects of Large Research, Development and Innovations Infrastructures. I would like to thank CESNET for allowing me to use the computational power of MetaCentrum. I could not get results presented in this thesis without their support. }\documentclass[../main.tex]{subfiles} \begin{document} \chapter{Benchmark Comparison} \label{benchmark_comparison} We now have a data-driven methodology for deriving learned sector universes (addressing RG-1), and have developed a objective criteria-driven ranking methodology to compare the learned sector universes against each other, addressing RG-2. The final step is to evaluate our objectively-identified risk-adjusted return optimal learned sector universe (see Section~\ref{optimal_sector_universe:risk_adj_return_optimal}) against the benchmark classification. Thus, this section addresses the third and final research goal, RG-3 (see Section~\ref{research_goals:specific_research_goals}). \begin{table}[h!] \centering \begin{tabular}{| c | c |} \hline & \\ RG-3 & Evaluate our risk-adjusted return optimal sector universe against the benchmark. \\ & \\ \hline \end{tabular} \end{table} \section{Comparison Overview} To preserve the impartial basis for comparison developed and maintained throughout this report, we isolated sector assignments for our benchmark sector universe, the \textit{GICS S\&P 500 Classification}. Unfortunately, as mentioned previously, we were only able to isolate transverse sector assignments for the year 2019, and were unable to access historical sector assignments, thus making a truly longitudinal comparison of our learned sector to the benchmark impossible. To mitigate this issue, we compared the latest learned sector as implied by our clustering algorithm to the benchmark classification. To maintain the consistency of our analysis, we utilized reIndexer (see Section~\ref{candidate_universe_ranking:reindexer}) to model SETFs of the benchmark, and to perform a backtest using the same configuration as was used for the candidate learned sector ranking (see Section~\ref{candidate_universe_ranking:backtest_config}). Similarly, we utilized the same performance metrics as were used to compare the candidate learned sector universes (see Section~\ref{candidate_universe_ranking:eval_metrics}) to compare the risk-adjusted return optimal learned sector universe to the benchmark. \section{Performance Metric Comparison} Figure~\ref{fig:benchmark_comparison:performance_metrics} contains four panels, (a) through (d), with each displaying one of the four performance metrics outlined in Section~\ref{candidate_universe_ranking:eval_metrics}. To best decompose the results of the comparison with the \textit{GICS S\&P 500 Classification} benchmark, we will analyze each of the performance metric comparisons in turn. \subsection{Cumulative Turnover Comparison} Panels (a) and (b) in Figure~\ref{fig:benchmark_comparison:performance_metrics} plot the cumulative turnover of SETF restructuring, and portfolio rebalancing, respectively. Recall from the previous section that for these turnover metrics, lower is better. As the red line represents the benchmark, it is apparent that the risk-adjusted return optimal learned sector did not outperform the benchmark with respect to both the SETF restructuring turnover, and the portfolio rebalancing turnover. \pagebreak This phenomenon may be explained through analysis of the Sankey Diagram of the risk-adjusted return learned sector universe in Figure~\ref{fig:optimal_sector_universe:max_sharpe}. It indicates a significant proliferation of component assets in 2 large sectors, with a large number of smaller sectors. Due to the fact that larger sectors have a higher notional value, and thus imply higher turnover when bought or sold, it is not surprising that the portfolio rebalancing turnover is higher for the risk-adjusted return optimal learned sector universe, compared to the more uniformly distributed benchmark universe. Additionally, the larger individual sectors \textit{Alpha} and \textit{Golf} would also imply a higher rate of turnover during SETF restructuring. As a higher number of assets implies a more volatile total value, the extremely large sectors unique to the risk-adjusted return optimal learned sector universe would command a higher level of turnover during SETF restructuring, when compared to the more modestly sized sectors of the benchmark universe. % See: https://tex.stackexchange.com/questions/238923/refer-to-a-table-as-figure \begin{table}[!h] \centering % \fbox{ \begin{tabular}{|c|c|} \hline & \\ \includegraphics[width=.475\linewidth]{images/etf_restrut_compare.png} & \includegraphics[width=.475\linewidth]{images/port_rebal_compare.png} \\ \textit{(a) SETF Restructuring Turnover} & \textit{(b) Portfolio Rebalancing Turnover} \\ & \\ \hline & \\ \includegraphics[width=.475\linewidth]{images/value_compare.png} & \includegraphics[width=.475\linewidth]{images/sharpe_compare.png} \\ \textit{(c) Absolute Portfolio Value} & \textit{(d) Risk-adjusted Return} \\ & \\ \hline \end{tabular} % } \captionof{figure}{A comparison of sector universe performance metrics, with the \textcolor{red}{benchmark universe in red}, and the \textcolor{blue}{risk-adjusted return optimal learned sector universe in blue}.} \label{fig:benchmark_comparison:performance_metrics} \end{table} \pagebreak \subsection{Absolute Portfolio Value Comparison} Panel (c) in Figure~\ref{fig:benchmark_comparison:performance_metrics} is a comparison of the absolute portfolio value of both the risk-adjusted return optimal learned sector universe, and the benchmark universe. As indicated by the graph, the learned sector universe provides a significantly higher value at the terminus of the backtest, beating out the benchmark by nearly \$15,000,000,000 on a starting capital base of \$10,000,000,000 each, which translates to an outperformance of nearly 150\%. The progression of the portfolio over time for both the learned sectors universe and the benchmark universe indicate that the portfolio returns of the two sector universes are lightly correlated. This is to be expected, as the underlying base of investable assets is identical (by design) between the two sector universes. However, there does appear to be significantly less historical volatility in the returns of the learned sector universe portfolio compared to the benchmark portfolio. This is particularly evident in the 750 - 1250 day interval in panel (c). This period shows that the portfolio value of the benchmark increased rapidly, at a significantly greater rate than its learned sector universe counterpart. However, at approximately the 1150 day mark, the benchmark suffers a extremely severe drop, losing nearly all of its gains of the preceding period. Interestingly, the learned sector universe portfolio does not appear to fluctuate in value significantly (relative to the benchmark) during this period. This observation, coupled with the commensurate final rally in both sector universe portfolios near the end of the backtesting period suggests that the diversification profile of the learned sectors portfolio is \textit{significantly} superior to that of the benchmark, resulting in not only a higher terminal portfolio value, but also significantly less volatility in reaching that value. \subsection{Risk-Adjusted Return Comparison} Figure~\ref{fig:benchmark_comparison:performance_metrics} (d) is a comparison of the rolling risk-adjusted return (i.e. Sharpe Ratio) of the benchmark sector universe and the risk-adjusted return optimal learned sector universe. Given the results of the analysis of the absolute portfolio value comparison above, the outperformance of the learned sector universe relative to the benchmark universe is not a surprising result. Continuing on the same line of analysis as the previous section, the Sharpe ratio of the benchmark sector universe performs extremely poorly during the interval of 750 - 1250 days discussed above. The negative effect of the increased volatility, despite a rally in the underlying portfolio is better reflected in the Sharpe ratio plot compared to the portfolio value plot of panel (c). In fact, the Sharpe ratio graph in panel (d) indicates that it was nearly more beneficial to own and hold the risk-free asset than the benchmark sector universe portfolio at approximately the 900 day mark, as the rolling Sharpe ratio of the portfolio briefly approaches 0. Despite rallying significantly during the interval, the Sharpe ratio of the benchmark sector universe never recovers, and does not approach the significantly higher value of the learned sector universe portfolio. Additionally, despite rallying toward the end of the backtest, the Sharpe ratio plots indicate that the trend of the rolling risk-adjusted return for both sector universes was negative, with a much more smooth slope on the learned sector universe portfolio line. This indicates a lower \textit{vol of vol} for the learned sector universe compared to the benchmark universe, which is further indication that the learned sector algorithm provides significantly better diversification benefits compared to the benchmark sector universe. \section{Qualitative Comparison} In this section, we attempt to conduct a more qualitatively-driven comparison and contrast of the risk-adjusted return optimal learned sector universe against the benchmark sector universe. Given the drastic difference in performance between the benchmark sector universe portfolio and the optimal learned sector universe portfolio, we believe that there is significant insight to be had by analyzing the composition of each of the sectors in the universe. Analyzing each of the learned sectors in turn (from Figure~\ref{fig:optimal_sector_universe:max_sharpe} and Figure~\ref{fig:benchmark_comparison:ls_optimal_assets}), it is clear that beyond the large sectors \textit{Alpha} and \textit{Golf}, a large majority of the remaining sectors are extremely small with respect to their numbers of component assets. Despite this however, the two large (major) sectors - as well as a selection of the smaller (i.e. minor) sectors - have an extremely high dispersion rate relative to the benchmark. That is, there doesn't seem to be a high level of congruence between the old and new sector assignments. This lack of agreement between the benchmark and learned sector universes is particularly apparent in the apparent lack of any direct transitional sector mappings in Figure~\ref{fig:benchmark_comparison:ls_optimal_assets}. \pagebreak Both major learned sectors \textit{Alpha} and \textit{Golf} comprise a large number of assets as their components. Particularly, it can be observed that learned sector \textit{Alpha} contains a majority of the benchmark \textit{Financials} sector, and the benchmark \textit{Utilities} sector. Given that these sector assignments are derived from fundamentals data, this is a particularly interesting result, as both \textit{Financials} and \textit{Utilities} have become extremely risk-averse businesses over the last decade; \textit{Financials} due to the Great Recession of 2008, and \textit{Utilities} due to extensive capital damage incurred by the increased severity and number of Natural Disasters. This grouping indicates that the capital structure of these businesses are also becoming increasingly similar. \begin{figure}[!h] \centering \fbox{ \includegraphics[width=.7\linewidth]{images/complete_17_benchmark_diff.png} } \caption{Sector assignment transitions between the benchmark sector classification universe and the risk-adjusted return optimal learned sector universe.} \label{fig:benchmark_comparison:ls_optimal_assets} \end{figure} Considering learned sector \textit{Golf}, it seems to be a mini-index within the original sector universe. From a component count perspective, it ingests a large amount of the benchmark \textit{Consumer Discretionary} and \textit{Consumer Staples} industries, as well as large swaths of industries that form the backbone of the US Economy as a whole; namely, the \textit{Information Technology}, \textit{Industrials}, and \textit{Real Estate} sectors. Appendix~\ref{appendix:portfolio_weights} contains stacked bar charts representing the level of investment in each of the sector SETFs for both the benchmark sector universe (see Figure~\ref{fig:appendix_weights:benchmark}), and the risk-adjusted return optimal learned sector universe (see Figure~\ref{fig:appendix_weights:sharpe_optimal}). Analyzing these graphs, in conjunction with the transition profile of sector assignments between the benchmark and optimal learned sector universe, it is clear that the learned sector \textit{Golf} was not a strong performer. The benefit of containing a large cross-section of companies from myriad traditional sectors, combined with their poor performance during the years of 2014 and 2015 appears to be a key factor in the outperformance of the benchmark sector universe. \end{document} % \begin{thm} % \label{thm:good_covers_theorem} % If $ \U$, $\U'$ are \textbf{good covers} of $ X$ with associated Cech complexes $ \L^\bullet(\U)$ and $ \L^\bullet(\U')$ then % \begin{align} % H^i(\L(\U)) \cong H^i(\L(\U')) % \end{align} % for all $ i \in \Z$ i.e. the cohomology of the Cech complex is independent of the good cover and hence is an invariant of the underlying space. Hence, we can define the Cech cohomology of $ X$ as % \begin{align} % \check H^i(X; \F) := H^i(\L(\U)) % \end{align} % for any good cover $ \U$ of $ X$. % \end{thm} \subsection{Locally Constant Functions} \begin{definition} For a topological space $ X$, let $ \L(X)$ denote the space of set maps $ f: X \rightarrow \F$ which are constant on each connected component of $ X$. Such functions are called \textbf{locally constant functions}.\footnote{$\L$ is an example of a \textbf{locally constant sheaf}, more on this later. } \end{definition} \begin{ques} Prove that if $ X$ has $ k$ connected components $ X_1, \dots, X_k$ then as an $ \F$ vector space \begin{align*} \L(X) \cong \F^k \end{align*} Further, $ \L(X)$ has a \textbf{canonical basis} given by functions $ f_1, \dots, f_k$ defined as \begin{align*} f_i(X_j) & \equiv \begin{cases} 1 & \mbox{ if } i=j \\ 0 & \mbox{ otherwise } \end{cases} \end{align*} \end{ques} \begin{ques} What is $ \L(X)$ if $ X$ is the empty set? \end{ques} \begin{ques} Show that if we think of $ \F$ as a disjoint union of two points then $ \L(X)$ is precisely the space of continuous functions $ X \rightarrow \F$. (We'll see later that this is the reason why $\L$ is a sheaf.) \end{ques} % \begin{definition} % For a topological space $ X$, let $ \L(X)$ denote the space of functions $ f: X \rightarrow \F$ which are constant on each connected component of $ X$. Such functions are called \textbf{locally constant functions}. % \end{definition} % % \begin{ques} % Show that if we think of $ \F$ as a disjoint union of two points then $ \L(X)$ is precisely the space of continuous functions $ X \rightarrow \F$. % \end{ques} % % \begin{ques} % Prove that if $ X$ has $ k$ connected components $ X_1, \dots, X_k$ then as an $ \F$ vector space % \begin{align} % \L(X) \cong \F^k % \end{align} % Further, $ \L(X)$ has a \textbf{canonical basis} given by functions $ f_1, \dots, f_k$ defined as % \begin{align} % f_i(X_j) &\equiv % \begin{cases} % 1 & \mbox{ if } i=j \\ 0 & \mbox{ otherwise } % \end{cases} % \end{align} % \end{ques} % % \begin{ques} % What is $ \L(X)$ if $ X$ is the empty set? % \end{ques} \begin{ques} \label{q:gluing_diagram} We can construct a 2 holed torus $M^2$ by gluing the sides of an octagon as in Figure \ref{fig:genus2}. Use the gluing diagram to construct a good cover of $M^2$. Find the cohomology using this good cover. \end{ques} \begin{figure}[h] \centering \includegraphics[width=11cm]{GluingDiagramHatcher} \caption{Constructing a $g$ holed torus $M^g$ using a $4g$-gon. (Image from Hatcher.)} \label{fig:genus2} \end{figure} \begin{ques}* More generally, it is possible to construct a $g$ holed torus by gluing polygons of $4g$ sides. Repeat Question \ref{q:gluing_diagram} for this surface. \end{ques} \begin{figure}[H] \centering \includegraphics[width=11cm]{Genus2} \caption{Gluing diagram for constructing a 2 holed torus $M^2$ using an octagon. (Googled image.)} \label{fig:genus2} \end{figure} \begin{ques} By gluing the sides of a square in funky ways we can create the Klein Bottle and the Real Projective Plane. Find their Cech Cohomologies. \begin{figure}[H] \centering \begin{subfigure}[t]{0.4\textwidth} \centering \includegraphics[height=3cm]{KleinBottle} \caption{Klein Bottle} \end{subfigure} \begin{subfigure}[t]{0.59\textwidth} \centering \includegraphics[height=3cm]{ProjectivePlane} \caption{Projective Plane} \end{subfigure} \end{figure} \end{ques} doxygen/latex/d3/daf/variables__0_8js_source.tex \subsection{variables\+\_\+0.\+js} \label{variables__0_8js_source}\index{/home/erik/prefix/default/src/soapysdr/build/docs/html/search/variables\+\_\+0.\+js@{/home/erik/prefix/default/src/soapysdr/build/docs/html/search/variables\+\_\+0.\+js}} \begin{DoxyCode} 00001 var searchData= 00002 [ 00003 [\textcolor{stringliteral}{'description'},[\textcolor{stringliteral}{'description'},[\textcolor{stringliteral}{'../structSoapySDRArgInfo.html#afca2eb0168826d1bbf40f3f8316ce47f'},1,\textcolor{stringliteral}{' SoapySDRArgInfo::description()'}],[\textcolor{stringliteral}{'../classSoapySDR\_1\_1ArgInfo.html#a6d999c94c0da4d201a13664ebb9f91fd'},1,\textcolor{stringliteral}{' SoapySDR::ArgInfo::description()'}]]] 00004 ]; \end{DoxyCode} thlautenschlaeger/sds % This file was created by tikzplotlib v0.9.1. \begin{tikzpicture} \definecolor{color0}{rgb}{0.12156862745098,0.466666666666667,0.705882352941177} \begin{axis}[ tick pos=both, x grid style={white!69.0196078431373!black}, xmajorgrids, xmin=1, xmax=25, xtick style={color=black}, y grid style={white!69.0196078431373!black}, ymajorgrids, ymin=0.762504885067262, ymax=1.00517165482061, ytick style={color=black} ] \path [draw=color0] (axis cs:1,0.972651459698655) --(axis cs:1,0.990885384470965); \path [draw=color0] (axis cs:5,0.934204509726188) --(axis cs:5,0.989399972451008); \path [draw=color0] (axis cs:10,0.917438758164983) --(axis cs:10,0.992139923171174); \path [draw=color0] (axis cs:15,0.8814828749611) --(axis cs:15,0.996462738021126); \path [draw=color0] (axis cs:20,0.830504455383245) --(axis cs:20,0.992113399952084); \path [draw=color0] (axis cs:25,0.762504885067262) --(axis cs:25,1.00517165482061); \addplot [color0, mark=-, mark size=7, mark options={solid}, only marks] table {% 1 0.972651459698655 5 0.934204509726188 10 0.917438758164983 15 0.8814828749611 20 0.830504455383245 25 0.762504885067262 }; \addplot [color0, mark=-, mark size=7, mark options={solid}, only marks] table {% 1 0.990885384470965 5 0.989399972451008 10 0.992139923171174 15 0.996462738021126 20 0.992113399952084 25 1.00517165482061 }; \addplot [color0, mark=*, mark size=2.5, mark options={solid}] table {% 1 0.98176842208481 5 0.961802241088598 10 0.954789340668079 15 0.938972806491113 20 0.911308927667665 25 0.883838269943938 }; \end{axis} \end{tikzpicture} \subsection{Formal Definition of Contexts and Context Sets} Definitions~\ref{def:context} formally describe contexts and context sets. \begin{definition}[context]\index{Contexts} \label{def:action} A \emph{context} is a representation of data that can be expressed as a $2$-tuple $(k,v)$, where \begin{enumerate} \item $k$ is a unique identifier of the context \item $v$ is an encoding of the data represented by the context \end{enumerate} \end{definition} Amrit-Gill/awesome-cv \cvsection{Experience} \begin{cventries} \cventry {title} {company} {address} {2000-01-23 - 2000-01-23} { \begin{cvitems} \item {value} \item {value} \end{cvitems} } \cventry {title} {company} {address} {2000-01-23 - 2000-01-23} { \begin{cvitems} \item {value} \item {value} \end{cvitems} } \end{cventries}goodcucumber/x40paraguide @Book{xie2015, title = {Dynamic Documents with {R} and knitr}, author = {}, publisher = {Chapman and Hall/CRC}, address = {Boca Raton, Florida}, year = {2015}, edition = {2nd}, note = {ISBN 978-1498716963}, url = {http://yihui.name/knitr/}, } @Book{csapp, title = {深入理解计算机系统}, author = {, }, translator = {龚奕利, 雷迎春}, publisher = {机械工业出版社}, year = {2010}, edition = {2nd}, note = {ISBN 978-7-111-32133-0}, } @Misc{cuda_doc, title = {CUDA Toolkit Documentation v11.0.3}, author = {nVidia}, year = {2020}, url = {https://docs.nvidia.com/cuda/archive/11.0/index.html}, } igorbotian/msc-thesis \Citation{Вдохновение приходит только во время работы}{} % \Paragraph{Что содержится в данном разделе} % В данном разделе содержится общая информация о проекте \IT{PeerHood}. % Рассматривается концепция, цели и ключевые требования к разрабатываемому проекту. % Подробно описывается архитектура \IT{PeerHood}. % Ставится вопрос о безопасности как пользовательских данных, так и безопасности ПО в текущей реализации проекта.diego9627/mathOrgConvert \documentclass[11pt]{article} \input{preamble} \begin{document} {\bf \noindent MOP 2008; Josh \\ Red Inequalities} \be \ii Show that if $a_i$ are reals with $\sum_{i=1}^n a_i = 1$, then $\sum_{i=1}^n a_i^2 \geq 1/n.$ \ii If $a,b,c>0$, prove that $(a^2b+b^2c+c^2a)(ab^2+bc^2+ca^2)\geq 9a^2b^2c^2.$ \ii Let $(a_i)_{i=1}^n$ be a sequence of positive reals and let $(b_i)_{i=1}^n$ be a permutation of $(a_i)$. Prove that $\sum_{i=1}^n a_i/b_i \geq n$. \ii Let $x,y,z>0$ with $xyz=1$. Prove that $x+y+z\leq x^2+y^2+z^2$. \ii (Romanian selection test) Let $a,b,x,y,z>0$. Prove that \[\frac x{ay+bz}+\frac y{az+bx} + \frac z{ax+by} \geq \frac 3{a+b}.\] \ii (IMO 95/2) Let $a, b, c$ be positive real numbers such that $abc = 1$. Prove that \[ \frac{1}{a^3(b+c)} + \frac{1}{b^3(c+a)} + \frac{1}{c^3(a+b)} \geq\frac{3}{2}. \] \ii Let $a,b,c,d>0$. Prove that \[\frac{1}{a}+\frac{1}{b}+\frac{4}{c} + \frac{16}{d}\geq\frac{64}{a+b+c+d}.\] \ii (Part of USAMO 79/3) Let $x,y,z\geq 0$ with $x+y+z=1$. Prove that \[x^3+y^3+z^3+6xyz\geq \frac14.\] \ii (IMO 74/5) Determine all possible values of \[ S= \frac{a}{a+b+d} +\frac{b}{a + b+c} + \frac{c}{b+c+d} +\frac{d}{a+c+d}\] where $a,b,c,d$ are arbitrary positive numbers. \ii (USAMO 77/5) If $a,b,c,d,e$ are positive numbers bounded by $p$ and $q$, $0 %% Email: %% Date: 2017-12-10 %% License: Latex Project Public License %% %% A package for drawing spectral sequences %% % TODO: % Try catch blocks % Make an argument type for the \d page argument. % deal with xmin, xmax, etc (was there actually a problem we were trying to fix?) % Make xmirror not mirror axes labels % % Redo sseqerrortest and set up regression test script as part of build (damn I didn't realize we'd lost anything imporant with that find -d disaster) % Maybe we should add some other regression tests too % % Lower priority: % speed up off page edges (uses 10% of the draw time for page 0 of tmfass) % improve \DoUntilOutOfBounds progress check % error messages that should be warnings by default? % should we specify our favorite conditionals library? % \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{spectralsequences}[2017/12/10 v1.2.0] \RequirePackage{tikz} \RequirePackage{etoolbox} \RequirePackage{xparse} \RequirePackage{verbatim} \usetikzlibrary{quotes} \usetikzlibrary{fit} \usetikzlibrary{positioning} \usetikzlibrary{intersections} \usetikzlibrary{backgrounds} \usepgflibrary{arrows.meta} \usetikzlibrary{shapes} %\usetikzlibrary{patterns} %\usetikzlibrary{profiler} \newif\ifsseq@draftmode \newif\ifsseq@tooltip \DeclareOption{draft}{\sseq@draftmodetrue} \DeclareOption{tooltips}{\sseq@tooltiptrue} \ProcessOptions\relax \ifsseq@tooltip \RequirePackage{pdfcomment} \fi \def\sseq@authorname{} \def\sseq@authoremail{} % Commands we are going to expose just inside of environments \def\sseq@macrolist{% \xcoord\ycoord\page %\xmin\xmax\ymin\ymax % these just get protected % Defined in sseqmessages: \quiet\endquiet % These are defined in sseqmain: \class\classoptions\replaceclass\replacesource\replacetarget \d\doptions\kill\structline\structlineoptions\circleclasses % The following are defined in sseqparsers: \pgfmathparse\isalive\lastx\lasty\lastclass\savestack\restorestack\pushstack\nameclass\tagclass \parsecoordinate\parsedifferential\getdtarget\gettag \IfOutOfBoundsTF\IfOutOfBoundsT\IfOutOfBoundsF\IfInBoundsTF\IfInBoundsT\IfInBoundsF \IfExistsTF\IfExistsT\IfExistsF\IfAliveTF\IfAliveT\IfAliveF \IfValidDifferentialTF\IfValidDifferentialT\IfValidDifferentialF \DrawIfValidDifferentialTF\DrawIfValidDifferentialT\DrawIfValidDifferentialF\DrawIfValidDifferential % sseqforeach \Do\DoUntilOutOfBounds\DoUntilOutOfBoundsThenNMore } % All the tikz commands. We replace these in our environment too. Replacements defined in sseqparsers. \def\sseq@tikzcommands{% \clip\coordinate\draw\fill\filldraw \graph\matrix\node\path\pattern \shade\shadedraw\useasboundingbox } %%%%%% %%%%%% %% %% %% Declarations and preliminaries %% %% %% %%%%%% %%%%%% \newif\ifsseq@inprogress \newif\ifsseq@hasname \newif\ifsseq@updateexisting \newif\ifsseq@ispageenv \newif\ifsseq@keepchanges \newif\ifsseq@keepglobaloptions \newif\ifsseq@globaldetone \newif\ifsseq@needstikz \newif\ifsseq@thispage \newif\ifsseq@outofrange \newif\ifsseq@classlabel \newif\ifsseq@draworphanedges \newif\ifsseq@draw \newif\ifsseq@drawedge \newif\ifsseq@tikzprims@integershift \newif\ifsseq@anchor \sseq@drawedgetrue \sseq@tikzprims@integershifttrue \newif\ifsseq@patchforeach \newif\ifsseq@patchfit \newif\ifsseq@patchxparseU \newif\ifsseq@tempif \newif\ifsseq@gtempif \newif\ifsseq@error \newtoks\sseq@temptoks \newtoks\sseq@temptoksii \newtoks\sseq@scope@toks \newcount\sseq@thepagecount \newcount\sseq@anonsseqcount \newcount\sseq@x \newcount\sseq@y \newcount\sseq@tempcount \newcount\sseq@tempcountb \newcount\sseq@tempx \newcount\sseq@tempy \newcount\sseq@tempxb \newcount\sseq@tempyb \newcount\sseq@xoffset % We add these to everything to avoid overflow errors as much as possible \newcount\sseq@yoffset \newcount\sseq@stackdepth \newdimen\sseq@tempdimen \newdimen\sseq@tempxdimen \newdimen\sseq@tempydimen \newdimen\sseq@xscalecm \newdimen\sseq@yscalecm \newdimen\sseq@clip@xcenter \newdimen\sseq@clip@ycenter \newdimen\sseq@tooltip@height \newdimen\sseq@tooltip@width \newdimen\sseq@gridstrokethickness \sseq@gridstrokethickness=.1pt % Ensure \@xp and \@nx have the correct values in case that amsmath isn't loaded \let\@xp\expandafter \let\@nx\noexpand \def\@xptwo{\@xp\@xp\@xp} \def\@xpthree{\@xp\@xp\@xp\@xp\@xp\@xp\@xp} \def\sseq@nil{\sseq@thisshouldnthappen@nil unique expansion} % This expansion text should be unique so that \ifx\sseq@nil\othercommand is false. \def\sseq@infinity{10000} % Larger than any coordinate anyone will ever use. \newcount\sseq@infinitycount \sseq@infinitycount=\sseq@infinity\relax \def\sseq@macroname{\@xp\@gobble\string} \def\sseq@gobble@to@nil#1\sseq@nil{} \def\sseq@macrogobble#1->{} % These are only used by defertikzcommand now \def\sseq@callas#1{\def\sseq@callcmd{#1}} \def\sseq@call#1{\bgroup\@xp\let\sseq@callcmd#1\@xptwo\egroup\sseq@callcmd} \def\sseq@getfirstchar#1#2\sseq@nil{#1} % used in shift/checkshift transform \def\sseq@smuggle@macro#1#2\egroup{\@xp\egroup\@xp\def\@xp#1\@xp{#1}} \def\sseq@protected@edef{\let\sseq@store@slsl\\\def\\{\protect\\}\let\@@protect \protect \let \protect \@unexpandable@protect \afterassignment \sseq@restore@protect \edef} \def\sseq@protected@xdef{\let\sseq@store@slsl\\\def\\{\protect\\}\let\@@protect \protect \let \protect \@unexpandable@protect \afterassignment \sseq@restore@protect \xdef} \def\sseq@restore@protect{\let\protect\@@protect\let\\\sseq@store@slsl} \def\sseq@eval#1{\bgroup\edef\sseq@temp{#1}\@xp\egroup\sseq@temp}% I got this from sseq.sty \def\sseq@protectedeval#1{\bgroup\sseq@protected@edef\sseq@temp{#1}\@xp\egroup\sseq@temp} \def\sseq@eval@show#1{\bgroup\edef\sseq@temp{#1}\show\sseq@temp\@xp\egroup\sseq@temp} \def\sseq@profilenew#1#2{\pgfprofilenew{#1}\pretocmd#2{\pgfprofilestart{#1}}{}{\error}\apptocmd#2{\pgfprofileend{#1}}{}{\error}} %%%% add to macro commands \def\sseq@d@addto@macro#1#2{\@xp\def\@xp#1\@xp{#1#2}} \def\sseq@e@addto@macro#1#2{\edef#1{\unexpanded\@xp{#1}#2}} % let #2 be expanded \def\sseq@eo@addto@macro#1#2{\edef#1{\unexpanded\@xp{#1}\unexpanded\@xp{#2}}} % let #2 be expanded once \def\sseq@g@addto@macro#1#2{\@xp\gdef\@xp#1\@xp{#1#2}} \def\sseq@x@addto@macro#1#2{\xdef#1{\unexpanded\@xp{#1}#2}} % let #2 be expanded \def\sseq@xprotected@addto@macro#1#2{\sseq@protected@xdef#1{\unexpanded\@xp{#1}#2}} \def\sseq@d@addto@toks#1#2{#1\@xp{\the#1#2}} \def\sseq@e@addto@toks#1#2{\sseq@eval{#1{\the#1#2}}} \def\sseq@d@addto@temptoks{\sseq@d@addto@toks\sseq@temptoks} \def\sseq@e@addto@temptoks{\sseq@e@addto@toks\sseq@temptoks} % Used in sseqkeys to delete spaces from style commands. % Set the catcode of space to ignore, and then reparse the characters in #2. \def\sseq@setmacronospaces#1#2{% \bgroup\catcode`\ =9\relax \makeatletter \scantokens{\expandafter\egroup\expandafter\def\expandafter#1\expandafter{\csname #2\endcsname}}% } \def\sseq@setnospaces#1#2{% \bgroup\catcode`\ =9\relax \makeatletter \scantokens{\expandafter\egroup\expandafter\def\expandafter#1\expandafter{\@firstofone{#2}}}% } \def\sseq@removeparens{\@xp\sseq@removeparens@} \def\sseq@removeparens@(#1){#1} % Stolen from trimspaces.sty \bgroup \catcode`\Q=3 \gdef\sseq@trimspaces#1{% \romannumeral-`\q\sseq@trim@trim@\noexpand#1Q Q% } \long\gdef\sseq@trim@trim@#1 Q{\sseq@trim@trim@@#1Q} \long\gdef\sseq@trim@trim@@#1Q#2{#1} \egroup \def\sseq@trimleadingspaces{\romannumeral-`q} \def\sseq@ifempty#1{% \@xp\ifx\@xp\sseq@nil\detokenize{#1}\sseq@nil \@xp\@firstoftwo \else \@xp\@secondoftwo \fi } \def\sseq@ifnil#1{ \ifx\sseq@nil#1 \@xp\@firstoftwo \else \@xp\@secondoftwo \fi } \let\sseq@breakpoint\relax \let\sseq@breakpointfinally\@gobble \def\sseq@break#1\sseq@breakpoint{} \def\sseq@break@finally#1\sseq@breakpoint#2{#2} \def\sseq@breakfi{\fi\sseq@break} \def\sseq@breakfifi{\fi\fi\sseq@break} \def\sseq@breakfififi{\fi\fi\fi\sseq@break} \long\def\sseq@breakdataenv#1\end#2{ \def\sseq@tempa{sseqdata}\def\sseq@tempb{#2}\ifx\sseqtempa\sseqtempb \@xp\sseq@breakdataenv@ \else \@xp\sseq@breakdataenv \fi } \def\sseq@breakdataenv@{\let\endsseqdata\sseq@breakendsseqdata\end{sseqdata}} \long\def\sseq@breakpageenv#1\end#2{ \def\sseq@tempa{sseqpage}\def\sseq@tempb{#2}\ifx\sseqtempa\sseqtempb \@xp\sseq@breakpageenv@ \else \@xp\sseq@breakpageenv \fi } \def\sseq@breakpageenv@{\let\endsseqpage\sseq@breakendsseqpage\end{sseqpage}} \def\sseq@breakendsseqdata{} \def\sseq@breakendsseqpage{} \def\sseq@seteverythingtonoops{% \let\sseqdata\comment \let\sseqpage\comment \let\sseqkeys\@gobble \let\sseqnewgroup\@gobblethree } \input sseqmessages.code.tex % Exposes directly: \sseqerrortowarning \input sseqcheckdefinitions.code.tex \ifsseq@tempif\else % Set to false in checkdefinitions if it failed to patch the key-value system. \sseq@seteverythingtonoops \sseq@pgfkeyspatchfailed \fi \input sseqloadstore.code.tex % Responsible for installing environment-only macros \input sseqmacromakers.code.tex % Exposes directly: \DeclareSseqCommand, \NewSseqCommand, \DeclareSseqGroup, \NewSseqGroup \input sseqparsers.code.tex % Responsible for making tikz modifications, exposes directly \sseqnormalizemonomial, \sseqparseint \ifsseq@patchforeach \input sseqforeach.code.tex \else \def\sseq@patchfor{} \fi \input sseqkeys.code.tex % Exposes directly: \sseqset, \sseqnewfamily \input sseqmain.code.tex % Defines all the main commands. Exposes directly: the environments, \xmin, \xmax, etc, \SseqCopyPage \input sseqdrawing.code.tex % Give standard definitions for savedpaths wrappers \def\sseq@beginscope@object{\begin{scope}} \def\sseq@endscope@object{\end{scope}} \let\sseq@scope@object\@firstofone \let\sseq@style@object\@firstofone \let\sseq@class@object\sseq@class@draw@ifpage \let\sseq@differential@object\sseq@differential@draw@ifpage \let\sseq@structline@object\sseq@structline@draw@ifpage \let\sseq@circleclass@object\sseq@circleclass@draw@ifpage \let\sseq@tikzpath@object\@firstofone %%% Some default key settings \sseqset{ edge labels={auto=right}, classes={draw,circle,inner sep=0pt,minimum size=0.35em}, circle classes={newellipse, ellipse ratio=1.2,draw, inner sep=2pt}, edges=draw, math nodes, differentials=->, pins=help lines } \ifsseq@tooltip \let\sseqtooltip\sseq@tooltip@wrapper \fi % Extra commands to expose: \let\sseqifempty\sseq@ifempty \let\SseqParseInt\sseqparseint \let\SseqNewFamily\sseqnewfamily \def\sseqpower#1#2{\@xp\sseqtypesetpower@\@xp{\the\numexpr#2}{#1}{1}} \def\sseqpowerempty#1#2{\@xp\sseqtypesetpower@\@xp{\the\numexpr#2}{#1}{}} \def\sseqtypesetpower@#1#2#3{\ifnum#1=\z@#3\else\ifnum#1=\@ne#2\else#2^{#1}\fi\fi} 1-10 \section*{Aufgabe 2} \subsection*{Teilaufgabe a} \textbf{Aufgabe} Formulieren Sie einen Algorithmus in Pseudocode zum Lösen des Gleichungssystems \[Ly = b,\] wobei $L$ eine invertierbare, untere Dreiecksmatrix ist. Geben Sie die Formel zur Berechnung von $y_i$ an. \textbf{Lösung:} \[y_i = \frac{b_i - \sum_{k=1}^{i-1} l_{ik} \cdot y_k}{l_{ii}}\] \begin{algorithm}[H] \begin{algorithmic} \Require Lower, invertable, triangular Matrix $L \in \mathbb{R}^{n \times n}$, Vektor $b$ \Procedure{solve}{$L$, $b$} \For{$i \in \Set{1, \dots n}$} \State $y_i \gets b_i$ \For{$k \in \Set{1, \dots, i-1}$} \State $y_i \gets y_i - l_{ik} \cdot y_k$ \EndFor \State $y_i \gets \frac{y_i}{l_{ii}}$ \EndFor \EndProcedure \end{algorithmic} \caption{Calculate $y$ in $Ly = b$} \end{algorithm} \subsection*{Teilaufgabe b} \[Ax = b \Leftrightarrow PAx = Pb \Leftrightarrow LRx = Pb \] \begin{algorithm}[H] \begin{algorithmic} \Require Matrix $A$, Vektor $b$ \Procedure{LoeseLGS}{$A$, $b$} \State $P, L, R \gets \Call{LRZer}{A}$ \State $b^* \gets P \cdot b$ \State $c \gets \Call{VorSub}{L, b^*}$ \State $x \gets \Call{RueckSub}{R, c}$ \State \Return $x$ \EndProcedure \end{algorithmic} \caption{Löse ein LGS $Ax = b$} \end{algorithm} \subsection*{Teilaufgabe c} Der Gesamtaufwand ist: \begin{itemize} \item LR-Zerlegung, $\frac{1}{3}n^3 - \frac{1}{3} n^2$ \item Vektormultiplikation, $2n$ \item Vorwärtssubstitution, $\frac{1}{2} n^2$ \item Rückwärtssubstitution, $\frac{1}{2} n^2$ \end{itemize} 0 @techreport{rfc2104hmac, author = { and and }, title = {HMAC: Keyed-Hashing for Message Authentication}, howpublished = {Internet Requests for Comments}, type = {RFC}, number = {2104}, year = {1997}, month = {feb}, issn = {2070-1721}, publisher = {RFC Editor}, institution = {RFC Editor}, url = {http://www.rfc-editor.org/rfc/rfc2104.txt} }@article{morrison2017differential, author = {Morrison, and Clowers, }, doi = {10.1007/s13361-017-1621-3}, journal = {Journal of The American Society for Mass Spectrometry}, number = {6}, pages = {1236--1241}, publisher = {Springer US New York}, title = {Differential fragmentation of mobility-selected glycans via ultraviolet photodissociation and ion mobility-mass spectrometry}, volume = {28}, year = {2017} } matcdac/IETF_RFCs @misc{rfc1977, series = {Request for Comments}, number = 1977, howpublished = {RFC 1977}, publisher = {RFC Editor}, doi = {10.17487/RFC1977}, url = {https://rfc-editor.org/rfc/rfc1977.txt}, author = {}, title = {{PPP BSD Compression Protocol}}, pagetotal = 25, year = 1996, month = aug, abstract = {This document describes the use of the Unix Compress compression protocol for compressing PPP encapsulated packets. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind.}, } REAL IRAV,IR,LAMDAS,LAMDAC,LATUDE,Longitude REAL WATACT,WATRAT,WATPOT,RNLU Integer AutoIrrigateF Common /Weath/ MSW1,MSW2,MSW3,MSW4,MSW5,MSW6, & MSW7,BSOLAR,ETCORR, & BTEMP,ATEMP,ERAIN,BWIND,BIR,WINDA,IRAV,JDAY, & NCD,JDLAST, CLDFAC,DEL(24),RINT(24),RNS, & RNC,RAIN,IR,WIND,CO2,TDUSK,TDUSKY, & CPREC(NumSD),TAIR(24),VPD(24), $ ROUGH, & RADINT(24),WATTSM(24),DIFINT(24), & ROWINC(24),CLOUD,SHADOW(24),DIFWAT(24), & DIRINT(24),WATACT,WATRAT,WATPOT,RNLU, & NumF(40),NumFP,hFur(40),QF,IFUR,GAIR(NumGD),PG, & LATUDE,Longitude, Altitude, RI,PAR(24), & PARINT(24),daylng,AutoIrrigAmt, & AutoIrrigateF content/publication/wolfers-2019/cite.bib1-10 @article{Wolfers2019, abstract = {Pattern classification and stratification approaches have increasingly been used in research on Autism Spectrum Disorder (ASD) over the last ten years with the goal of translation towards clinical applicability. Here, we present an extensive scoping literature review on those two approaches. We screened a total of 635 studies, of which 57 pattern classification and 19 stratification studies were included. We observed large variance across pattern classification studies in terms of predictive performance from about 60% to 98% accuracy, which is among other factors likely linked to sampling bias, different validation procedures across studies, the heterogeneity of ASD and differences in data quality. Stratification studies were less prevalent with only two studies reporting replications and just a few showing external validation. While some identified strata based on cognition and intelligence reappear across studies, biology as a stratification marker is clearly underexplored. In summary, mapping biological differences at the level of the individual with ASD is a major challenge for the field now. Conceptualizing those mappings and individual trajectories that lead to the diagnosis of ASD, will become a major challenge in the near future.}, author = {Wolfers, , Dinga, , Daan and Isakoglou, , and Zabihi, Mariam and Llera, Alberto and Chowdanayaka, Rajanikanth and Kumar, . and Peng, Han and Laidi, Charles and Batalle, Dafnis and Dimitrova, Ralica and Charman, Tony and Loth, Eva and Lai, Meng-Chuan and Jones, Emily and Baumeister, Sarah and Moessnang, Carolin and Banaschewski, Tobias and Ecker, Christine and Dumas, Guillaume and O'Muircheartaigh, Jonathan and Murphy, Declan and Buitelaar, . and Marquand, . and Beckmann, .}, doi = {10.1016/j.neubiorev.2019.07.010}, file = {:C\:/Users/taylo/AppData/Local/Mendeley Ltd./Mendeley Desktop/Downloaded/Wolfers et al. - 2019 - From pattern classification to stratification towards conceptualizing the heterogeneity of Autism Spectrum Disor.pdf:pdf}, issn = {01497634}, journal = {Neuroscience & Biobehavioral Reviews}, keywords = {Autism spectrum disorder,Biotypes,Classification,Clustering,Machine learning,Pattern recognition,Precision medicine,Stratification}, mendeley-groups = {Website - Publications}, month = {sep}, pages = {240--254}, pmid = {31330196}, publisher = {Elsevier Ltd}, title = {From pattern classification to stratification: towards conceptualizing the heterogeneity of Autism Spectrum Disorder}, url = {https://doi.org/10.1016/j.neubiorev.2019.07.010 https://linkinghub.elsevier.com/retrieve/pii/S0149763419303197}, volume = {104}, year = {2019} } \hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic}{}\doxysection{Класс crutchesbicycles.\+studyhelper.\+domain.\+Traffic} \label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic}\index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} Граф наследования\+:crutchesbicycles.\+studyhelper.\+domain.\+Traffic\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic} \end{center} \end{figure} \doxysubsection*{Открытые члены} \begin{DoxyCompactItemize} \item long \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_aec310155019933195fdd010e4c5d910a}{get\+Id\+Traffic}} () \item void \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6f8b7a23956a16e08dd13db139602662}{set\+Id\+Traffic}} (long \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a435fc4ac2914436e70546d03892c54a1}{id\+Traffic}}) \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a7ec4704ba89123075901ddc64ba83d25}{get\+Student}} () \item void \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_ab7c798b79bf0ad886ba01accc5daaea9}{set\+Student}} (\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_add6b5d98651e1fcebfd735b3f4d2f500}{student}}) \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a621df7b217cf551bf0ab76b14a52a056}{get\+Schedule\+Record}} () \item void \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a9209324bc00021d7e98447ab8a04c121}{set\+Schedule\+Record}} (\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6ec39aebc5fd2fe542608103981e9e74}{schedule\+Record}}) \item Date \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8e35f561c67c41b22970e44329f27848}{get\+Date}} () \item void \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a2b798ccbb641f3532dad9ccf4036b663}{set\+Date}} (Date \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_af45d27b1c2df80da6400ca0e790146e0}{date}}) \item boolean \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a09ee19393d769dc45a829f77326a5d12}{is\+Attend}} () \item void \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a0e9f3ec11f2a02cf88d2bc2e527369d4}{set\+Attend}} (boolean attend) \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a933fc124817c004ff6cf176f01414eba}{Traffic}} () \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a5a8c5560b9bf24d962ea1fccea3c0c24}{Traffic}} (\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_add6b5d98651e1fcebfd735b3f4d2f500}{student}}, \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6ec39aebc5fd2fe542608103981e9e74}{schedule\+Record}}, Date \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_af45d27b1c2df80da6400ca0e790146e0}{date}}, boolean \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8583c888c831f7f386355a3247d8cafc}{is\+Attend}}) \end{DoxyCompactItemize} \doxysubsection*{Закрытые данные} \begin{DoxyCompactItemize} \item long \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a435fc4ac2914436e70546d03892c54a1}{id\+Traffic}} \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_add6b5d98651e1fcebfd735b3f4d2f500}{student}} \item \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6ec39aebc5fd2fe542608103981e9e74}{schedule\+Record}} \item Date \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_af45d27b1c2df80da6400ca0e790146e0}{date}} \item boolean \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8583c888c831f7f386355a3247d8cafc}{is\+Attend}} \end{DoxyCompactItemize} \doxysubsection{Подробное описание} Сущность \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic}{Traffic}}. Посещение студентов.~\newline Состоит из полей\+: \begin{DoxyParams}{Аргументы} {\em id\+Student} & \\ \hline {\em student} & -- Студент \\ \hline {\em schedule\+Record} & -- Запись в расписании \\ \hline {\em date} & -- Дата пары \\ \hline {\em is\+Attend} & -- отметка посещения (true/false) \\ \hline \end{DoxyParams} \begin{DoxyAuthor}{Автор} vgtstptlk / \end{DoxyAuthor} \begin{DoxyVersion}{Версия} 1.\+0.\+0 \end{DoxyVersion} \doxysubsection{Конструктор(ы)} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a933fc124817c004ff6cf176f01414eba}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a933fc124817c004ff6cf176f01414eba}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!Traffic@{Traffic}} \index{Traffic@{Traffic}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{Traffic()}{Traffic()}\hspace{0.1cm}{\footnotesize\ttfamily [1/2]}} {\footnotesize\ttfamily crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+Traffic (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a5a8c5560b9bf24d962ea1fccea3c0c24}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a5a8c5560b9bf24d962ea1fccea3c0c24}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!Traffic@{Traffic}} \index{Traffic@{Traffic}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{Traffic()}{Traffic()}\hspace{0.1cm}{\footnotesize\ttfamily [2/2]}} {\footnotesize\ttfamily crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+Traffic (\begin{DoxyParamCaption}\item[{\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}}}]{student, }\item[{\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}}}]{schedule\+Record, }\item[{Date}]{date, }\item[{boolean}]{is\+Attend }\end{DoxyParamCaption})} \doxysubsection{Методы} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8e35f561c67c41b22970e44329f27848}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8e35f561c67c41b22970e44329f27848}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!getDate@{getDate}} \index{getDate@{getDate}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{getDate()}{getDate()}} {\footnotesize\ttfamily Date crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+get\+Date (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_aec310155019933195fdd010e4c5d910a}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_aec310155019933195fdd010e4c5d910a}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!getIdTraffic@{getIdTraffic}} \index{getIdTraffic@{getIdTraffic}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{getIdTraffic()}{getIdTraffic()}} {\footnotesize\ttfamily long crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+get\+Id\+Traffic (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a621df7b217cf551bf0ab76b14a52a056}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a621df7b217cf551bf0ab76b14a52a056}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!getScheduleRecord@{getScheduleRecord}} \index{getScheduleRecord@{getScheduleRecord}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{getScheduleRecord()}{getScheduleRecord()}} {\footnotesize\ttfamily \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+get\+Schedule\+Record (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a7ec4704ba89123075901ddc64ba83d25}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a7ec4704ba89123075901ddc64ba83d25}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!getStudent@{getStudent}} \index{getStudent@{getStudent}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{getStudent()}{getStudent()}} {\footnotesize\ttfamily \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+get\+Student (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a09ee19393d769dc45a829f77326a5d12}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a09ee19393d769dc45a829f77326a5d12}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!isAttend@{isAttend}} \index{isAttend@{isAttend}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{isAttend()}{isAttend()}} {\footnotesize\ttfamily boolean crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+is\+Attend (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a0e9f3ec11f2a02cf88d2bc2e527369d4}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a0e9f3ec11f2a02cf88d2bc2e527369d4}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!setAttend@{setAttend}} \index{setAttend@{setAttend}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{setAttend()}{setAttend()}} {\footnotesize\ttfamily void crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+set\+Attend (\begin{DoxyParamCaption}\item[{boolean}]{attend }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a2b798ccbb641f3532dad9ccf4036b663}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a2b798ccbb641f3532dad9ccf4036b663}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!setDate@{setDate}} \index{setDate@{setDate}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{setDate()}{setDate()}} {\footnotesize\ttfamily void crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+set\+Date (\begin{DoxyParamCaption}\item[{Date}]{date }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6f8b7a23956a16e08dd13db139602662}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6f8b7a23956a16e08dd13db139602662}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!setIdTraffic@{setIdTraffic}} \index{setIdTraffic@{setIdTraffic}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{setIdTraffic()}{setIdTraffic()}} {\footnotesize\ttfamily void crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+set\+Id\+Traffic (\begin{DoxyParamCaption}\item[{long}]{id\+Traffic }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a9209324bc00021d7e98447ab8a04c121}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a9209324bc00021d7e98447ab8a04c121}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!setScheduleRecord@{setScheduleRecord}} \index{setScheduleRecord@{setScheduleRecord}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{setScheduleRecord()}{setScheduleRecord()}} {\footnotesize\ttfamily void crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+set\+Schedule\+Record (\begin{DoxyParamCaption}\item[{\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}}}]{schedule\+Record }\end{DoxyParamCaption})} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_ab7c798b79bf0ad886ba01accc5daaea9}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_ab7c798b79bf0ad886ba01accc5daaea9}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!setStudent@{setStudent}} \index{setStudent@{setStudent}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{setStudent()}{setStudent()}} {\footnotesize\ttfamily void crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+set\+Student (\begin{DoxyParamCaption}\item[{\mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}}}]{student }\end{DoxyParamCaption})} \doxysubsection{Данные класса} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_af45d27b1c2df80da6400ca0e790146e0}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_af45d27b1c2df80da6400ca0e790146e0}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!date@{date}} \index{date@{date}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{date}{date}} {\footnotesize\ttfamily Date crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+date\hspace{0.3cm}{\ttfamily [private]}} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a435fc4ac2914436e70546d03892c54a1}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a435fc4ac2914436e70546d03892c54a1}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!idTraffic@{idTraffic}} \index{idTraffic@{idTraffic}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{idTraffic}{idTraffic}} {\footnotesize\ttfamily long crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+id\+Traffic\hspace{0.3cm}{\ttfamily [private]}} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8583c888c831f7f386355a3247d8cafc}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a8583c888c831f7f386355a3247d8cafc}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!isAttend@{isAttend}} \index{isAttend@{isAttend}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{isAttend}{isAttend}} {\footnotesize\ttfamily boolean crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+is\+Attend\hspace{0.3cm}{\ttfamily [private]}} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6ec39aebc5fd2fe542608103981e9e74}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_a6ec39aebc5fd2fe542608103981e9e74}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!scheduleRecord@{scheduleRecord}} \index{scheduleRecord@{scheduleRecord}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{scheduleRecord}{scheduleRecord}} {\footnotesize\ttfamily \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_schedule_record}{Schedule\+Record}} crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+schedule\+Record\hspace{0.3cm}{\ttfamily [private]}} \mbox{\Hypertarget{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_add6b5d98651e1fcebfd735b3f4d2f500}\label{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_traffic_add6b5d98651e1fcebfd735b3f4d2f500}} \index{crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}!student@{student}} \index{student@{student}!crutchesbicycles.studyhelper.domain.Traffic@{crutchesbicycles.studyhelper.domain.Traffic}} \doxysubsubsection{\texorpdfstring{student}{student}} {\footnotesize\ttfamily \mbox{\hyperlink{classcrutchesbicycles_1_1studyhelper_1_1domain_1_1_student}{Student}} crutchesbicycles.\+studyhelper.\+domain.\+Traffic.\+student\hspace{0.3cm}{\ttfamily [private]}} Объявления и описания членов класса находятся в файле\+:\begin{DoxyCompactItemize} \item /\+Users/vgtstptlk/\+Desktop/МИРЭА/\+Study\+Helper/src/main/java/crutchesbicycles/studyhelper/domain/\mbox{\hyperlink{_traffic_8java}{Traffic.\+java}}\end{DoxyCompactItemize} 1-10 \chapter{Cycles in the Collatz Graph} \label{ch:cycles} \section{A remark about cycles} \label{sec:cycles} In graph theory, a path of length $n\geq 1$ that starts and ends at the same vertex is called a circuit. A circuit, in which no vertex is repeated with the sole exception that the initial vertex is the terminal vertex, is called a cycle. A cycle of length $n$ is referred to as an $n$-cycle. For these definitions, we rely on \cite[p.~599]{Ref_Rosen}, \cite[p.~35]{Ref_Benjamin_Chartrand_Zhang} and \cite[p.~445]{Ref_Chartrand_Zhang}. Furthermore, we call a cycle originating from the root a trivial cycle. \begin{remark} In order for the cycles to become graphically visible, we now require that in a graph $H$ two vertices $v_1$ and $v_2$ are one and the same if the label of both nodes are identical: $l_{V(H)}(v_1)=l_{V(H)}(v_2)\rightarrow v_1=v_2$. As a consequence, there is no guarantee that the graph precisely refers to the algebraic structure of a free monoid anymore. A free monoid requires that each of its elements can be written in one and only one way. \end{remark} When different nodes collapse on one, the graph is no longer necessarily a tree. Let us point to the monoid $S^\ast$, which we introduced in section \ref{sec:groups_graphs}. Take for example four of its elements, the empty string $e$, the strings $qqr$, $qqrqqr$, and $qqrqqrqqr$. These elements lie as well within the subset $U\subset T\subset S^\ast$, and they are represented by nodes of the tree $H_U$ that all have the same label $1=ev_{S^\ast}(qqr,1)=ev_{S^\ast}(qqrqqr,1)=ev_{S^\ast}(qqrqqrqqr,1)$. These nodes are one and the same, the root of $H_U$. Visually, then in $H_U$ a directed edge goes from the vertex labeled with $4$ back to the root node. Analogically, in $H_{C,3}$ a loop connects the root to itself, since due to the path contraction even labeled nodes do not exist in $H_{C,3}$. The aforementioned example reflects the trivial cycle of the Collatz sequence. Figure~\ref{fig:5} depicts a section of $H_{C,5}$, which includes the $3$-cycle $43,17,27$. Because of the two non-trivial cycles $(43,17,27)$ and $(83,33,13)$, in $H_{C,5}$ there does not exist a path between the root and the vertex $43$ and between the root and the vertex $83$. Hence, $H_{C,5}$ is said to be a disconnected graph. Generally, a graph is called a disconnected graph if it is impossible to walk (along its edges) from any vertex to any other \cite[pp.~46-47]{Ref_Benjamin_Chartrand_Zhang}. \begin{figure} \includegraphics[width=1.00\textwidth]{figures/h_c5a.png} \caption{Section of $H_{C,5}$ including the $3$-cycle $43,17,27$} \label{fig:5} \end{figure} The following considerations focus on non-trivial cycles, and therefore on cycles that do not originate from the root, but cause the graph to be a disconnected graph. Utilizing the example of the graph $H_{C,5}$ we are able to deduct from the cycle $(43,17,27)$ the simple and self-evident equality $\textit{left-child}^3(43)=43$: \begin{equation*} \begin{array}{l} \textit{left-child}(43)=\frac{1}{5}*\left(43*2^1-1\right)=17 \\[\medskipamount] \textit{left-child}(17)=\frac{1}{5}*\left(17*2^3-1\right)=27 \\[\medskipamount] \textit{left-child}(27)=\frac{1}{5}*\left(27*2^3-1\right)=43 \end{array} \end{equation*} Obviously, the authors note, it would be interesting to find out what circumstances enable a graph to have non-trivial cycles, whether it be the $5x+1$ variant, the $7x+1$ variant of $H_C$ or any variant $H_{C,k}$ with $k\geq 1$. \section{\texorpdfstring{Which variants of $H_C$ have non-trivial cycles?}{Which variants of HC have non-trivial cycles?}} \label{sec:non_trivial_cycles} The generalization of the relationship between successive nodes, given by equation~\ref{eq:generalized_reachability} leads to the condition for an existence of an $n$-cycle in any $kx+1$ variant of $H_C$, which looks analogous to the condition given by equation~\ref{eq:func_cycle} that specifies $H_{C,3}$ has a cycle: \begin{equation} \label{eq:generalized_cycle} 2^\alpha=\prod_{i=1}^{n}\left(k+\frac{1}{v_i}\right) \end{equation} The natural number $\alpha$ is the sum of edges that have been contracted between the vertices $v_i$ forming the cycle, in other words $\alpha$ is the number of divisions by $2$ within the sequence. The natural number $n$ is the cycle length and $k$ obviously specifies the variant of $H_C$. Since between each vertex at least one edge has been contracted (at least one division by $2$ took place), we know that our exponent alpha is greater than or equal to the sequence length: \begin{equation} \label{eq:n_alpha} \alpha\ge n \end{equation} In their 2020 publication Koch et al. \cite{Ref_Koch_2020} provide a list of cycles for different values of $k$, identified with a linear search performed by a Python script \cite{Ref_Koch_Github}. Table~\ref{table:known_cycles} lists all these discovered cycles (refer to \cite{Ref_Koch_2020} for details on the discovery procedure and search intervals). Note that the cycles in table~\ref{table:known_cycles} are written in reverse order, i.e. in the order which corresponds to the Collatz sequence. To obtain the cycles in terms of graph theory referring to the graph $H_C$, read them from right to left. \begin{table}[H] \centering \begin{tabular}{|L|R|R|C|} \hline \thead{\boldsymbol{k}} & \thead{\textbf{cycle}} & \thead{\boldsymbol{\alpha}} & \thead{\textbf{non-trivial}} \\ \hline 1 & 1 & 1 & \\ \hline 3 & 1 & 2 & \\ \hline 5 & 1,3 & 5 & \\ \hline 5 & 13,33,83 & 7 & \checkmark \\ \hline 5 & 27,17,43 & 7 & \checkmark \\ \hline 7 & 1 & 3 & \\ \hline 15 & 1 & 4 & \\ \hline 31 & 1 & 5 & \\ \hline 63 & 1 & 6 & \\ \hline 127 & 1 & 7 & \\ \hline 181 & 27,611 & 15 & \checkmark \\ \hline 181 & 35,99 & 15 & \checkmark \\ \hline 255 & 1 & 8 & \\ \hline 511 & 1 & 9 & \\ \hline \end{tabular} \caption{Known $n$-cycles in $kx+1$ variants of $H_C$ for $k\leq1000$, $n\leq 100$} \label{table:known_cycles} \end{table} Based on the results shown in table~\ref{table:known_cycles} we state the following theorem~\ref{theo:2} that renders more precisely the prerequisite for cycles that may occur in any variants of $H_C$. \begin{theorem} \label{theo:2} An $n$-cycle can only exist in a graph $H_{C,k}$, if the following equation holds: \begin{equation*} 2^{\bar\alpha}=2^{\lfloor n\log_2k\rfloor+1}=\prod_{i=1}^{n}\left(k+\frac{1}{v_i}\right) \end{equation*} \end{theorem} The statement behind theorem~\ref{theo:2} consists in the claim that, in order for an $n$-cycle to occur, the exponent $\alpha$ has to be $\bar\alpha=\lfloor n\log_2k\rfloor+1$. This statement is true if the following general condition for the validity of the cycle-alpha's upper limit always holds (see \cite{Ref_Koch_2020}): \begin{equation} \label{eq:condition_max} n\log_2k-\lfloor n\log_2k\rfloor<2-\log_2\left(\prod_{i=1}^{n}\left(1+\frac{1}{kv_{i}}\right)\right) \end{equation} A product $\prod(1+a_n)$ with positive terms $a_n$ is convergent if the series $\sum a_n$ converges, see Knopp \cite[p.~220]{Ref_Knopp}. A similar statement provides Murphy \cite{Ref_Murphy}, who write the factors in the form $c_n=1+a_n$ and explains that if $\prod c_n$ is convergent then $c_n\rightarrow1$ and therefore if $\prod (1+a_n)$ is convergent then $a_n\rightarrow0$. Thus, to verify whether the product in condition~\ref{eq:condition_max} is converging towards a limiting value, it is sufficient to examine the following sum: \begin{equation*} \sum_{i=1}^{n}\frac{1}{kv_{i}} \end{equation*} The sum of reciprocal vertices depending only from $v_1$ is given in appendix~\ref{appx:sum_reciprocal_vertices}. \section{Cycles and the product in the condition for cycle-alpha's upper limit} Let us start with the following product equality, which will give us insights into the relationship between cycles and the product in the condition for alpha's upper limit. The variables $V_1,\ldots,V_m$ and $W_1,\ldots,W_n$ are all odd positive integers: \begin{equation} \label{eq:product_equality} (V_1+1)\cdots(V_m+1)\cdot W_1\cdots W_n=V_1\cdots V_m\cdot(W_1+1)\cdots(W_n+1) \end{equation} Every natural odd number $V$ can be expressed in the form of $V=v\cdot2^{\alpha}-1$ whereby $v$ is an positive odd integer and $\alpha>0$ is any natural number. This allows us to perform the following substitution (we use $\alpha_V$ for denoting the divisions by two between successive nodes $v_i$ and $\alpha_W$ for divisions by two between successive nodes $w_i$): \begin{equation} \label{eq:product_equality_substitution} \begin{array}{lll} V_1&=v_22^{\alpha_{V,1}}-1&=kv_1\\ V_2&=v_32^{\alpha_{V,2}}-1&=kv_2\\ \vdots&\vdots&\vdots\\ V_{m-1}&=v_m2^{\alpha_{V,m-1}}-1&=kv_{m-1}\\ V_m&=v_12^{\alpha_{V,m}}-1&=kv_m \end{array}\qquad \begin{array}{lll} W_1&=w_22^{\alpha_{W,1}}-1&=kw_1\\ W_2&=w_32^{\alpha_{W,2}}-1&=kw_2\\ \vdots&\vdots&\vdots\\ W_{n-1}&=w_n2^{\alpha_{W,n-1}}-1&=kw_{n-1}\\ W_n&=w_12^{\alpha_{W,n}}-1&=kw_n \end{array} \end{equation} The substitution rotating from $v_2=(kv_1+1)\cdot2^{-\alpha_{V,1}}$ to $v_m=(kv_{m-1}+1)\cdot2^{-\alpha_{V,m-1}}$ and finally back to $v_1=(kv_m+1)\cdot2^{-\alpha_{V,m}}$ describes a cycle. The result of these substitutions into equation~\ref{eq:product_equality} is the following equality: \begin{flalign*} v_22^{\alpha_{V,1}}\cdots v_m2^{\alpha_{V,m-1}}v_12^{\alpha_{V,m}}\cdot W_1\cdots W_n&=V_1\cdots V_m\cdot w_22^{\alpha_{W,1}}\cdots w_n2^{\alpha_{W,n-1}}w_12^{\alpha_{W,n}}\\ v_22^{\alpha_{V,1}}\cdots v_m2^{\alpha_{V,m-1}}v_12^{\alpha_{V,m}}\cdot kw_1\cdots kw_n&=kv_1\cdots kv_m\cdot w_22^{\alpha_{W,1}}\cdots w_n2^{\alpha_{W,n-1}}w_12^{\alpha_{W,n}} \end{flalign*} The trivial case where $n=m$ and the sum of exponents are equal $\sum_{i=1}^{m}\alpha_{V,i}=\sum_{i=1}^{n}\alpha_{W,i}$ simplifies the product equality as follows: \begin{flalign*} (V_1+1)\cdots(V_n+1)\cdot W_1\cdots W_n&=V_1\cdots V_n\cdot(W_1+1)\cdots(W_n+1)\\ v_1\cdots v_n\cdot\cancel{2^{\alpha_{V,1}+\ldots+\alpha_{V,n}}}\cdot W_1\cdots W_n&=V_1\cdots V_m\cdot w_1\cdots w_n\cdot\cancel{2^{\alpha_{W,1}+\ldots+\alpha_{W,n}}} \end{flalign*} This equality becomes immediatly true if $V_1\cdots V_n=W_1\cdots W_n$ which is the less spectacular case. The more interesting case arises from setting $V_i=kv_i$ and $W_i=kw_i$ as given by substitution~\ref{eq:product_equality_substitution} wich turns the product equality into an always true statement as well: \begin{flalign*} v_1\cdots v_n\cdot W_1\cdots W_n&=V_1\cdots V_m\cdot w_1\cdots w_n\\ v_1\cdots v_n\cdot k^n\cdot w_1\cdots w_n&=k^n\cdot v_1\cdots v_n\cdot w_1\cdots w_n \end{flalign*} \begin{example} The following exemplarly product equality fullfills equation~\ref{eq:product_equality}, whereby $V_1=65$, $V_2=165$, $V_3=415$ and $W_1=135$, $W_2=85$, $W_3=215$: \[ (65+1)(165+1)(415+1)\cdot135\cdot85\cdot215=65\cdot165\cdot415\cdot(135+1)(85+1)(215+1) \] We perform the following substitutions: \[ \arraycolsep=0.2em\begin{array}{ll} V_1=65&=v_22^{\alpha_{V,1}}-1=33\cdot2^1-1=5v_1\\ V_2=165&=v_32^{\alpha_{V,2}}-1=83\cdot2^1-1=5v_2\\ V_3=415&=v_12^{\alpha_{V,3}}-1=13\cdot2^5-1=5v_3 \end{array}\hspace{1em} \begin{array}{ll} W_1=135&=w_22^{\alpha_{W,1}}-1=17\cdot2^3-1=5w_1\\ W_2=85&=w_32^{\alpha_{W,2}}-1=43\cdot2^1-1=5w_2\\ W_3=215&=w_12^{\alpha_{W,3}}-1=27\cdot2^3-1=5w_3 \end{array} \] The result of these substitutions is: \[ 33\cdot\cancel{2^1}\cdot83\cdot\cancel{2^1}\cdot13\cdot\cancel{2^5}\cdot135\cdot85\cdot215=65\cdot165\cdot415\cdot17\cdot\cancel{2^3}\cdot43\cdot\cancel{2^1}\cdot27\cdot\cancel{2^3} \] Since the sum of exponents $\alpha_{V,i}$ and $\alpha_{W,i}$ are equal, we can cancel out all powers of two and obtain: \[ v_2v_3v_1W_1W_2W_3=33\cdot83\cdot13\cdot135\cdot85\cdot215=65\cdot165\cdot415\cdot17\cdot43\cdot27=V_1V_2V_3w_2w_3w_1 \] This product equality becomes true $v_2v_3v_1\cdot k^3\cdot w_1w_2w_3=k^3\cdot v_1v_2v_3\cdot w_2w_3w_1$ when we set $V_i=kv_i$ and $W_i=kw_i$ (for $i=1,2,3$) which inevitably leads to the two corresponding cycles for $k=5$ that are already presented by table~\ref{table:known_cycles}. \end{example} Let us define the difference $\Delta=(1+\nicefrac{1}{V_1})(1+\nicefrac{1}{V_2})\cdots(1+\nicefrac{1}{V_m})-(1+\nicefrac{1}{W_1})(1+\nicefrac{1}{W_2})\cdots(1+\nicefrac{1}{W_n})$. We know that if this difference is zero, then we have found two cycles, as for example $0=(1+\nicefrac{1}{65})(1+\nicefrac{1}{165})(1+\nicefrac{1}{415})-(1+\nicefrac{1}{135})(1+\nicefrac{1}{85})(1+\nicefrac{1}{215})$. Can you identify empirically some set pairs $\{V_1,V_2,\ldots,V_m\}$ and $\{W_1,W_2,\ldots,W_n\}$, where the difference is not necessarly zero but a whole number, e.g. $\Delta=1,2,3,\ldots$? DS_Ch7/doc/latex/a04548.tex \hypertarget{a04548}{}\doxysection{Queue Namespace Reference} \label{a04548}\index{Queue@{Queue}} 队列 \doxysubsection*{Classes} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{a05365}{Binary\+Heap}} \begin{DoxyCompactList}\small\item\em Priority queue. \end{DoxyCompactList}\item class \mbox{\hyperlink{a05425}{priority\+\_\+queue\+\_\+int}} \begin{DoxyCompactList}\small\item\em 基于最小堆的整型的优先级队列类 \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Detailed Description} 队列 jiuguangw/dayone2latex \pagestyle{empty} % Half title page { \centering ~ \vspace{24pt} {\scshape\Huge \booktitle \par} } \cleardoublepage % Title page \begin{titlepage} \centering ~ \vspace{24pt} {\scshape\Huge \booktitle\par} \vspace{6pt} {\scshape\large \subtitle\par} \vspace{\stretch{3.25}} {\itshape\large by\par} \vspace{6pt} {\itshape\Large \authorname\par} \vspace{\stretch{6}} {\large \publisher\par} \end{titlepage} %% complexity and convenience make diagnosing PaaS application performance hard %% -- PaaS is opaque to programmers %% -- PaaS services must operate at scale (asynchronous, distributed) %% new PaaS service for performance diagnotics %% -- can't require application instrumentation to be a PaaS service %% -- must be fully automated %% -- must be integrated with the runtime system %% this paper: anaomaly detection and root cause identification %% -- anaomaly: change in performance that results in SLA violation %% -- root cause: diagnosis of the reason for the anomaly as either a change %% in workload, increased latency an application component, or increased %% latency in a PaaS service and the identification of which one %% -- Roots is more general with respect to detectors and handlers but we %% investigate anomaly detection and root cause identification defined %% above in this paper %% diagnosis => after the fact %% -- can use asynchronous, but must be able to time correlate Cloud computing is a popular approach for deploying applications at scale~\cite{Antonopoulos:2010:CCP:1855007,Pinheiro:2014:ACC:2618168.2618188}. This widespread adoption of cloud computing, particularly for deploying web applications, is facilitated by ever-deepening software abstractions. These abstractions elide the complexity necessary to enable scale, while making application development easier and faster. But they also obscure the runtime details of cloud applications, making the diagnosis of performance problems challenging. Therefore, the rapid expansion of cloud technologies combined with their increasing opacity has intensified the need for new techniques to monitor applications deployed in cloud platforms~\cite{DaCunhaRodrigues:2016:MCC:2851613.2851619}. Application developers and cloud administrators generally wish to monitor application performance, detect anomalies, and identify bottlenecks. To obtain this level of operational insight into cloud-hosted applications, the cloud platforms must support data gathering and analysis capabilities that span the entire software stack of the cloud. However, most cloud technologies available today do not provide adequate application monitoring support. Cloud administrators must therefore trust the application developers to implement necessary instrumentation at the application level. This typically entails using third party, external monitoring software~\cite{newrelic,dynatrace,datadog}, which significantly increases the effort and financial cost of maintaining applications. Developers must also ensure that their instrumentation is both correct, and does not degrade application performance. Nevertheless, since the applications depend on extant cloud services (e.g. scalable database services, scalable in-memory caching services, etc.) that are performance opaque, it is often difficult, if not impossible to diagnose the ``root cause'' of a performance problem using such extrinsic forms of monitoring. Further compounding the performance diagnosis problem, today's cloud platforms are very large and complex~\cite{DaCunhaRodrigues:2016:MCC:2851613.2851619,Ibidunmoye:2015:PAD:2808687.2791120}. They are comprised of many layers, where each layer may consist of many interacting components. Therefore when a performance anomaly manifests in a user application, it is often challenging to determine the exact layer or the component of the cloud platform that may be responsible for it. Facilitating this level of comprehensive root cause analysis requires both data collection at different layers of the cloud, and mechanisms for correlating the events recorded at different layers. Moreover, performance monitoring for cloud applications must be customizable. Different applications have different monitoring requirements in terms of data gathering frequency (sampling rate), length of the history to consider when performing statistical analysis (sample size), and the performance SLOs (service level objectives~\cite{Keller:2003:WFS:635430.635442}) that govern the application. Cloud monitoring should be able to facilitate these diverse requirements on a per-application basis. Designing such customizable and extensible performance monitoring frameworks that are built into the cloud platforms is a novel and challenging undertaking. To address these challenges, we develop a full-stack, application performance monitor (APM) called \textit{Roots}~\cite{Jayathilaka:2017:PMR:3038912.3052649}, as a cloud Platform-as-a-service (PaaS) extension. PaaS clouds provide a set of managed services which developers compose into applications, via high-level interfaces (i.e., defined and exported via a software development kit (SDKs)). We design Roots as another PaaS service so that it can be managed automatically and directly capture events and performance data across the PaaS without requiring application code instrumentation. Prior work outlines several key requirements for cloud APMs~\cite{DaCunhaRodrigues:2016:MCC:2851613.2851619,Ibidunmoye:2015:PAD:2808687.2791120}, which we incorporate into Roots. They are: \begin{LaTeXdescription} \item[Scalability] Roots is lightweight, and does not cause any noticeable overhead in application performance. It puts strict upper bounds on the data kept in memory. The persistent data is accessed on demand, and can be removed after their usefulness has expired. \item[Multitenancy] Roots facilitates configuring monitoring policies at the granularity of individual applications. Users can employ different statistical analysis methods to process the monitoring data in ways that are most suitable for their applications. \item[Complex application architecture] Roots collects data from the entire cloud stack (load balancers, app servers, built-in PaaS services etc.). It correlates data gathered from different parts of the cloud platform, and performs systemwide bottleneck identification. \item[Dynamic resource management] Cloud platforms are dynamic in terms of their magnitude and topology. Roots captures performance events of applications by augmenting the key components of the cloud platform. When new processes/components become active in the cloud platform, they inherit the same augmentations, and start reporting to Roots automatically. \item[Autonomy] Roots detects performance anomalies online without manual intervention. When Roots detects a problem, it attempts to automatically identify the root cause by analyzing available workload and service invocation data. \vspace{-0.05in} \end{LaTeXdescription} Roots collects data from the logs and the interfaces of internal PaaS components. In addition to high-level metrics including request throughput and latency, Roots measures PaaS service invocations and their duration. It uses batch operations and asynchronous communication to minimize its overhead on request latency. When Roots detects a performance anomaly in an application, it attempts to identify its root cause by analyzing the workload data and the performance of the internal PaaS services on which the application depends. Roots first determines if the detected anomaly was most likely caused by a change in the application workload (e.g. a sudden spike in the number of client requests), or by an internal bottleneck in the cloud platform (e.g. a slow database query). For the latter, Roots employs a statistical bottleneck identification method that combines quantile analysis, change point detection, and linear regression to identify the root cause bottleneck (i.e. the PaaS component that most likely caused the performance degredation). We also devise a mechanism for Roots that distinguishes between different paths of execution in the application (control flows). Our approach does not require static analysis but instead uses the runtime data collected by Roots. This mechanism calculates the proportion of user requests processed by each path and uses it to characterize the workload of an application (e.g. read-heavy vs write-heavy workload in a data management application). Using this approach, Roots is able to detect when application workloads change. We prototype Roots as an extension to the AppScale, open source PaaS~\cite{6488671}. We evaluate the feasibility and the efficacy of Roots by conducting a series of empirical trials using our prototype. We show that Roots is able to detect manually injected faults within 5 minutes of their injection with very low overhead. We also show that Roots is able to scale to tens of thousands concurrent applications. adamdboult/nodeHomePage \subsection{Vector Autoregression (VAR)} We consider a vector of observables, not just one Autoregressive (AR) model for a vector. VAR(p) looks \(p\) back. The AR(\(p\)) model is: \(y_t=\alpha + \sum_{i=1}^p\beta y_{t-i}+\epsilon_t\) VAR(\(p\)) generalises this to where \(y_t\) is a vector. We define VAR(\(p\)) as: \(y_t\) \(y_t=c + \sum_{i=1}^pA_i y_{t-i}+\epsilon_t\) gecco/hyperneat.bib @article{Samuel_59, title={Some Studies in Machine Learning Using the Game of Checkers}, volume={3}, url={http://www.research.ibm.com/journal/rd/033/ibmrd0303B.pdf}, number={3}, journal={IBM Journal of Research and Development}, publisher={IBM}, author={ L}, year={1959}, pages={210--229}} @mastersthesis{naddaf10, AUTHOR = {}, TITLE = {Game-Independent AI Agents For Playing Atari 2600 Console Games}, School= {University of Alberta}, YEAR = {2010} } @article{verbancsics10, author = { Stanley, .}, title = {Evolving Static Representations for Task Transfer}, journal = {J. Mach. Learn. Res.}, issue_date = {3/1/2010}, volume = {11}, month = {August}, year = {2010}, issn = {1532-4435}, pages = {1737--1769}, numpages = {33}, url = {http://dl.acm.org/citation.cfm?id=1756006.1859909}, acmid = {1859909}, publisher = {JMLR.org}, } @inproceedings{gauci08, abstract = {{An important feature of many problem domains in machine learning is their geometry. For example, adjacency relationships, symmetries, and Cartesian coordinates are essential to any complete description of board games, visual recognition, or vehicle control. Yet many approaches to learning ignore such information in their representations, instead inputting flat parameter vectors with no indication of how those parameters are situated geometrically. This paper argues that such geometric information is critical to the ability of any machine learning approach to effectively generalize; even a small shift in the configuration of the task in space from what was experienced in training can go wholly unrecognized unless the algorithm is able to learn the regularities in decision-making across the problem geometry. To demonstrate the importance of learning from geometry, three variants of the same evolutionary learning algorithm (NeuroEvolution of Augmenting Topologies), whose representations vary in their capacity to encode geometry, are compared in checkers. The result is that the variant that can learn geometric regularities produces a significantly more general solution. The conclusion is that it is important to enable machine learning to detect and thereby learn from the geometry of its problems.}}, author = { Stanley, .}, booktitle = {Proceedings of the 23rd National Conference on Artificial Intelligence (AAAI)}, citeulike-article-id = {9744476}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1620169}, keywords = {evolution, machine\_learning}, posted-at = {2011-09-06 17:36:17}, priority = {2}, title = {{A case study on the critical role of geometric regularity in machine learning}}, url = {http://portal.acm.org/citation.cfm?id=1620169}, year = {2008} } @inproceedings{ambrosio08, abstract = {{This paper argues that multiagent learning is a potential "killer application" for generative and developmental systems (GDS) because key challenges in learning to coordinate a team of agents are naturally addressed through indirect encodings and information reuse. For example, a significant problem for multiagent learning is that policies learned separately for different agent roles may nevertheless need to share a basic skill set, forcing the learning algorithm to reinvent the wheel for each agent. GDS is a good match for this kind of problem because it specializes in ways to encode patterns of related yet varying motifs. In this paper, to establish the promise of this capability, the Hypercube-based NeuroEvolution of Augmenting Topologies (HyperNEAT) generative approach to evolving neurocontrollers learns a set of coordinated policies encoded by a single genome representing a team of predator agents that work together to capture prey. Experimental results show that it is not only possible, but beneficial to encode a heterogeneous team of agents with an indirect encoding. The main contribution is thus to open up a significant new application domain for GDS.}}, address = {New York, NY, USA}, author = {D'Ambrosio, . and .}, booktitle = {GECCO '08: Proceedings of the 10th annual conference on Genetic and evolutionary computation}, citeulike-article-id = {6235855}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1389095.1389256}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1389095.1389256}, doi = {10.1145/1389095.1389256}, isbn = {978-1-60558-130-9}, keywords = {alife10, behavior, evolution, indirect}, location = {Atlanta, GA, USA}, pages = {819--826}, posted-at = {2010-01-27 10:51:11}, priority = {2}, publisher = {ACM}, title = {{Generative encoding for multiagent learning}}, url = {http://dx.doi.org/10.1145/1389095.1389256}, year = {2008} } @inproceedings{clune09, author = { Beckmann, . and Ofria, Charles and Pennock, .}, title = {Evolving coordinated quadruped gaits with the HyperNEAT generative encoding}, booktitle = {Proceedings of the Eleventh conference on Congress on Evolutionary Computation}, series = {CEC'09}, year = {2009}, isbn = {978-1-4244-2958-5}, location = {Trondheim, Norway}, pages = {2764--2771}, numpages = {8}, url = {http://dl.acm.org/citation.cfm?id=1689599.1689966}, acmid = {1689966}, publisher = {IEEE Press}, address = {Piscataway, NJ, USA}, } @Article{stanley02, title={Evolving Neural Networks Through Augmenting Topologies}, author={ and }, volume={10}, journal={Evolutionary Computation}, number={2}, pages={99-127}, url="http://nn.cs.utexas.edu/?stanley:ec02", year={2002} } @InProceedings(stone01, Author=" and ", Title="Scaling Reinforcement Learning toward {R}obo{C}up Soccer", BookTitle="Proceedings of the Eighteenth International Conference on Machine Learning", publisher = "Morgan Kaufmann, San Francisco, CA", pages = "537--544", year = "2001", abstract={ RoboCup simulated soccer presents many challenges to reinforcement learning methods, including a large state space, hidden and uncertain state, multiple agents, and long and variable delays in the effects of actions. We describe our application of episodic SMDP Sarsa(lambda)with linear tile-coding function approximation and variable lambda to learning higher-level decisions in a keepaway subtask of RoboCup soccer. In keepaway, one team, ``the keepers,'' tries to keep control of the ball for as long as possible despite the efforts of ``the takers.'' The keepers learn individually when to hold the ball and when to pass to a teammate, while the takers learn when to charge the ball-holder and when to cover possible passing lanes. Our agents learned policies that significantly out-performed a range of benchmark policies. We demonstrate the generality of our approach by applying it to a number of task variations including different field sizes and different numbers of players on each team. }, wwwnote={ICML-2001
Some simulations of keepaway referenced in the paper.}, ) @ARTICLE{genesereth05, author = { and }, title = {General game playing: Overview of the AAAI competition}, journal = {AI Magazine}, year = {2005}, volume = {26}, pages = {62--72} } @inproceedings{duik08, abstract = {{Rich representations in reinforcement learning have been studied for the purpose of enabling generalization and making learning feasible in large state spaces. We introduce Object-Oriented MDPs (OO-MDPs), a representation based on objects and their interactions, which is a natural way of modeling environments and offers important generalization opportunities. We introduce a learning algorithm for deterministic OO-MDPs and prove a polynomial bound on its sample complexity. We illustrate the performance gains of our representation and algorithm in the wellknown Taxi domain, plus a real-life videogame.}}, author = {, Littman, .}, booktitle = {Proceedings of 25th International Conference on Machine Learning (ICML)}, citeulike-article-id = {7358711}, citeulike-linkout-0 = {http://paul.rutgers.edu/\~{}cdiuk/papers/OORL.pdf}, citeulike-linkout-1 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.149.7056}, keywords = {learning, phd, planning, relational-rl}, pages = {240--247}, posted-at = {2010-06-25 10:25:50}, priority = {0}, title = {{An object-oriented representation for efficient reinforcement learning}}, url = {http://paul.rutgers.edu/\~{}cdiuk/papers/OORL.pdf}, year = {2008} } @misc{atarihist, Author = {}, Title = {Atari and the deep history of video games}, howpublished = {\url{http://www.boston.com/bostonglobe/ideas/articles/2009/03/08/a_talk_with_nick_montfort/}} } @misc{pacmancompetition, Author = {}, Title = {Ms Pac-Man Competition (screen capture mode)}, howpublished = {\url{http://dces.essex.ac.uk/staff/sml/pacman/CIG2011Results.html}} } @ARTICLE{sigevolution2007, author = {}, title = {Ms Pac-Man Competition}, journal = {SIGEVOlution}, year = {2007}, volume = {2}, number = {4}, pages = {37--38} } @InProceedings{Urieli+MKBS:2010, author = " MacAlpine, Bentor, ", title = "On Optimizing Interdependent Skills: A Case Study in Simulated 3D Humanoid Robot Soccer", booktitle = "Proceedings of the Tenth International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2011)", year = "2011", ISBN = "978-0-9826571-5-7", editor = " ", volume = "2", publisher = "IFAAMAS", pages = "769--776", } @article{thain2005distributed, title={Distributed computing in practice: The Condor experience}, author={. and . and .}, journal={Concurrency and Computation: Practice and Experience}, volume={17}, number={2-4}, pages={323--356}, year={2005}, publisher={Wiley Online Library} } @article{tesauro_94, author = {}, title = {TD-Gammon, a self-teaching backgammon program, achieves master-level play}, journal = {Neural Comput.}, volume = {6}, issue = {2}, month = {March}, year = {1994}, issn = {0899-7667}, pages = {215--219}, numpages = {5}, url = {http://dl.acm.org/citation.cfm?id=188104.188107}, doi = {10.1162/neco.1994.6.2.215}, acmid = {188107}, publisher = {MIT Press}, address = {Cambridge, MA, USA}, } @Article{stone05, Author=" and and ", Title="Reinforcement Learning for {R}obo{C}up-Soccer Keepaway", journal="Adaptive Behavior", volume="13",number="3", year = "2005", pages="165--188", abstract={ RoboCup simulated soccer presents many challenges to reinforcement learning methods, including a large state space, hidden and uncertain state, multiple independent agents learning simultaneously, and long and variable delays in the effects of actions. We describe our application of episodic SMDP Sarsa(lambda) with linear tile-coding function approximation and variable lambda to learning higher-level decisions in a keepaway subtask of RoboCup soccer. In keepaway, one team, ``the keepers,'' tries to keep control of the ball for as long as possible despite the efforts of ``the takers.'' The keepers learn individually when to hold the ball and when to pass to a teammate. Our agents learned policies that significantly outperform a range of benchmark policies. We demonstrate the generality of our approach by applying it to a number of task variations including different field sizes and different numbers of players on each team. }, wwwnote={Contains material that was previously published in an ICML-2001 paper and a RoboCup 2003 Symposium paper.
Some simulations of keepaway referenced in the paper and keepaway software.}, } @article{parker09, author = { and }, title = {Backpropagation without Human Supervision for Visual Control in Quake II}, journal = {Proceedings of the 2009 IEEE Symposium on Computational Intelligence and Games (CIG'09)}, year = {2009}, pages = {287-293} } @article{sutton98, author = { and }, title = {Reinforcement Learning: An Introduction}, journal = {IEEE Transactions on Neural Networks}, volume = {9}, number = {5}, year = {1998}, pages = {1054-1054}, ee = {http://doi.ieeecomputersociety.org/10.1109/TNN.1998.712192}, bibsource = {DBLP, http://dblp.uni-trier.de} } @article{campbell02, author = { and . and }, title = {Deep Blue}, journal = {Artif. Intell.}, volume = {134}, number = {1-2}, year = {2002}, pages = {57-83}, ee = {http://dx.doi.org/10.1016/S0004-3702(01)00129-1}, bibsource = {DBLP, http://dblp.uni-trier.de} } @article{ferrucci10, abstract = {{IBM Research undertook a challenge to build a computer system that could compete at the human champion level in real time on the American TV Quiz show, Jeopardy! The extent of the challenge includes fielding a real-time automatic contestant on the show, not merely a laboratory exercise. The Jeopardy! Challenge helped us address requirements that led to the design of the DeepQA architecture and the implementation of Watson. After 3 years of intense research and development by a core team of about 20 researches, Watson is performing at human expert-levels in terms of precision, confidence and speed at the Jeopardy! Quiz show. Our results strongly suggest that DeepQA is an effective and extensible architecture that may be used as a foundation for combining, deploying, evaluating and advancing a wide range of algorithmic techniques to rapidly advance the field of QA.}}, author = { Brown, Chu-Carroll, Fan, Gondek, Kalyanpur, . and Lally, , , }, citeulike-article-id = {8061574}, citeulike-linkout-0 = {http://www.aaai.org.proxy.lib.sfu.ca/ojs/index.php/aimagazine/article/view/2303}, journal = {AI Magazine}, keywords = {artificial\_intelligence, inference\_system, knowledge\_base, language\_comprehension, language\_processing, natural\_language}, number = {3}, posted-at = {2010-10-21 05:41:58}, priority = {0}, publisher = {Association for the Advancement of Artificial Intelligence}, title = {{Building Watson: An Overview of the DeepQA Project}}, url = {http://www.aaai.org.proxy.lib.sfu.ca/ojs/index.php/aimagazine/article/view/2303}, volume = {31}, year = {2010} } \chapter{How to Use CugThesis} For a thesis based on CugTehsis, we just focused on contents, e.g., text, figures, references, tables, equations and their cross reference. The other details about this template will be updated on \href{https://www.jianshu.com/p/c9bb775fe0f4}{简书文章: CugTheis使用说明} 。 \subsection{Incert Figure} \begin{figure} [htbp] \centering \includegraphics[width=\textwidth ]{SinglePassModel} \caption[Single-Pass Model]{Single Pass Model} \fnote{Diagrammatic sketch of single-pass model, which include recharge zone and discharge zone. The down flow fluid is heated up by deep magmatic heat soource and then phase seperation occurs.} \label{fig:singlepassmodel} \end{figure} \autoref{fig:singlepassmodel} is a single figure \footnote{This is a foot note：note that reference citation} \citep{andersen2017faulting}。Each figure has a labe, caption and an alternative short caption。 \begin{figure} [htbp] \centering% \subcaptionbox{Color CUG logo\label{fig:cug1} } {\includegraphics[height=0.3\textwidth]{CUG_Logo1} } \hspace{0.01\textwidth} \subcaptionbox{Wechat \label{fig:weixin} } {\includegraphics[height=0.3\textwidth]{weixingongzhong} } \hspace{0.01\textwidth} \subcaptionbox{White-Black CUG logo\label{fig:cug2} } {\includegraphics[height=0.3\textwidth]{CUG_Logo2} } \caption{Three-column Figure} \label{fig:figure_3col} \end{figure} \autoref{fig:figure_3col} is a three-column figure, it has a main caption and three sub-captions and sub-numbers。For instance, following wechat \ref{fig:weixin} to get much more academic resoources. \subsection{Table} Three-line table \begin{table}[htbp] \centering \caption{Symbols and Values} \label{tab:symbols_values} \begin{tabular}{cccc} \toprule Symbol & Definition & Value &Unit \\ \midrule $\vec{g}$ & Gravitational acceleration & 9.8 & $m/s^2$ \\ $\rho_f$ & Density of water & 1.0 & $kg/m^3$ \\ \bottomrule \end{tabular} \end{table} \autoref{tab:symbols_values} is a three-line table with label and caption。 \subsection{Equations} Inline equation $E=MC^2$，the following is a independent equation, \begin{equation} \left( {\phi {\rho _f}{c_{pf}} + \left( {1 - \phi } \right){\rho _r}{c_{pr}}} \right)\frac{{\partial T}}{{\partial t}} = \nabla \cdot \left( {{K_r}\nabla T} \right) - {\rho _f}{c_{pf}}\vec v \cdot \nabla T + \frac{{{\mu _f}}}{k}{\vec v^2} - {\left( {\frac{{\partial \;ln\rho }}{{\partial \;lnT}}} \right)_p}\frac{{Dp}}{{Dt}} \label{eq:EnergyConservation} \end{equation} \ref{eq:EnergyConservation} represents energy conservation of hydrothermal fluid flow。 \subsection{References} First cite format：\cite{vehling2018implementation} noted that phase seperation occurs in deep。 The second cite format：there are some research focus on 3-D simulation \citep{coumou2008structure,coumou2006dynamics}。 0 \plainFrame{% \centering \pause% \includegraphics[width=.5\pagewidth,height=.4\pageheight,keepaspectratio]{media/lsaqi1oct4aguj5dd7jm.png} \pause% \includegraphics[width=.5\pagewidth,height=.4\pageheight,keepaspectratio]{media/unnamed__1_.jpg} \pause% \includegraphics[width=.5\pagewidth,height=.4\pageheight,keepaspectratio]{media/screenshotGoals.png} \pause% \includegraphics[width=.5\pagewidth,height=.4\pageheight,keepaspectratio]{media/screenshotAspects.png} } content/publication/kolberg-iris-he-wifs-2019/cite.bib @inproceedings{Kolberg-IrisHE-WIFS-2019, author = { and and and and and }, booktitle = {Proc. Int. Workshop on Information Forensics and Security (WIFS)}, title = {Template Protection based on Homomorphic Encryption: Computational Efficient Application to Iris-Biometric Verification and Identification}, year = {2019} } src/volume-1/decisions.tex \section{Prosecution and Declination Decisions} \markboth{Prosecution and Declination Decisions}{Prosecution and Declination Decisions} \hyperlink{section.3.1}{The Appointment Order} authorized the Special Counsel's Office ``to prosecute federal crimes arising from [its] investigation'' of the matters assigned to it. In deciding whether to exercise this prosecutorial authority, the Office has been guided by the Principles of Federal Prosecution set forth in the Justice (formerly U.S. Attorney's) Manual. In particular, the Office has evaluated whether the conduct of the individuals considered for prosecution constituted a federal offense and whether admissible evidence would probably be sufficient to obtain and sustain a conviction for such an offense. Justice Manual \S~9-27.220~(2018). Where the answer to those questions was yes, the Office further considered whether the prosecution would serve a substantial federal interest, the individuals were subject to effective prosecution in another jurisdiction, and there existed an adequate non-criminal alternative to prosecution. \textit{Id.} As explained below, those considerations led the Office to seek charges against two sets of Russian nationals for their roles in perpetrating the active-measures social media campaign and computer-intrusion operations. \blackout{Harm to Ongoing Matter} The Office similarly determined that the contacts between Campaign officials and Russia-linked individuals either did not involve the commission of a federal crime or, in the case of campaign-finance offenses, that our evidence was not sufficient to obtain and sustain a criminal conviction. At the same time, the Office concluded that the Principles of Federal Prosecution supported charging certain individuals connected to the Campaign with making false statements or otherwise obstructing this investigation or parallel congressional investigations. \subsection{Russian ``Active Measures'' Social Media Campaign} On February~16, 2018, a federal grand jury in the District of Columbia returned an indictment charging 13 Russian nationals and three Russian entities---including the Internet Research Agency (IRA) and Concord Management and Consulting LLC (Concord)---with violating U.S. criminal laws in order to interfere with U.S. elections and political processes.%1276 \footnote{A more detailed explanation of the charging decision in this case is set forth in a separate memorandum provided to the Acting Attorney General before the indictment.} The indictment charges all of the defendants with conspiracy to defraud the United States (Count One), three defendants with conspiracy to commit wire fraud and bank fraud (Count Two), and five defendants with aggravated identity theft (Counts Three through Eight). \textit{Internet Research Agency} Indictment. Concord, which is one of the entities charged in the Count One conspiracy, entered an appearance through U.S. counsel and moved to dismiss the charge on multiple grounds. In orders and memorandum opinions issued on August~13 and November~15, 2018, the district court denied Concord's motions to dismiss. \textit{United States~v.\ Concord Management \& Consulting LLC}, 347~F. Supp.~3d 38 (D.D.C. 2018). \textit{United States~v.\ Concord Management \& Consulting LLC}, 317~F. Supp.~3d 598 (D.D.C. 2018). As of this writing, the prosecution of Concord remains ongoing before the U.S. District Court for the District of Columbia. The other defendants remain at large. Although members of the IRA had contact with individuals affiliated with the Trump Campaign, the indictment does not charge any Trump Campaign official or any other U.S. person with participating in the conspiracy. That is because the investigation did not identify evidence that any U.S. person who coordinated or communicated with the IRA knew that he or she was speaking with Russian nationals engaged in the criminal conspiracy. The Office therefore determined that such persons did not have the knowledge or criminal purpose required to charge them in the conspiracy to defraud the United States (Count One) or in the separate count alleging a wire- and bank-fraud conspiracy involving the IRA and two individual Russian nationals (Count Two). The Office did, however, charge one U.S. national for his role in supplying false or stolen bank account numbers that allowed the IRA conspirators to access U.S. online payment systems by circumventing those systems' security features. On February~12, 2018, pleaded guilty, pursuant to a single-count information, to identity fraud, in violation of 18~U.S.C. \S~1028(a)(7) and (b)(1)(D). Plea Agreement, \textit{United States~v.\ }, No.~1:18-cr-24 (D.D.C. Feb.~12, 2018), Doc.~10. The investigation did not establish that Pinedo was aware of the identity of the IRA members who purchased bank account numbers from him. Pinedo's sales of account numbers enabled the IRA members to anonymously access a financial network through which they transacted with U.S. persons and companies. \textit{See} Gov't Sent.~Mem.~at~3, \textit{United States~v.\ o}, No.~1:18-cr-24 (D.D.C. Sept.~26, 2018), Doc.~24. On October~10, 2018, Pinedo was sentenced to six months of imprisonment, to be followed by six months of home confinement, and was ordered to complete 100 hours of community service. \subsection{Russian Hacking and Dumping Operations} \subsubsection{Section 1030 Computer-Intrusion Conspiracy} \paragraph{Background} On July~13, 2018, a federal grand jury in the District of Columbia returned an indictment charging Russian military intelligence officers from the GRU with conspiring to hack into various U.S. computers used by the Clinton Campaign, DNC, DCCC, and other U.S. persons, in violation of 18~U.S.C. \S\S~1030 and~371 (Count One); committing identity theft and conspiring to commit money laundering in furtherance of that hacking conspiracy, in violation of 18~U.S.C. \S\S~1028A and~1956(h) (Counts Two through Ten); and a separate conspiracy to hack into the computers of U.S. persons and entities responsible for the administration of the 2016 U.S. election, in violation of 18~U.S.C. \S\S~1030 and~371 (Count Eleven). \textit{Netyksho} Indictment.%1277 \footnote{The Office provided a more detailed explanation of the charging decision in this case in meetings with the Office of the Acting Attorney General before the indictment.} As of this writing, all 12 defendants remain at large. The \textit{Netyksho} indictment alleges that the defendants conspired with one another and with others to hack into the computers of U.S. persons and entities involved in the 2016 U.S. presidential election, steal documents from those computers, and stage releases of the stolen documents to interfere in the election. \textit{Netyksho} Indictment~\P~2. The indictment also describes how, in staging the releases, the defendants used the Guccifer~2.0 persona to disseminate documents through WikiLeaks. On July~22, 2016, WikiLeaks released over 20,000 emails and other documents that the hacking conspirators had stolen from the DNC\null. \textit{Netyksho} Indictment~\P~48. In addition, on October~7, 2016, WikiLeaks began releasing emails that some conspirators had stolen from Clinton Campaign chairman John Podesta after a successful spearphishing operation. \textit{Netyksho} Indictment~\P~49. \blackout{Harm to Ongoing Matter} \blackout{Grand Jury} \blackout{Harm to Ongoing Matter} \paragraph{Charging Decision As to [\protect\censor{Harm to Ongoing Matter}]} \blackout{Harm to Ongoing Matter}%1278 \footnote{The Office also considered, but ruled out, charges on the theory that the post-hacking sharing and dissemination of emails could constitute trafficking in or receipt of stolen property under the National Stolen Property Act (NSPA), 18~U.S.C. \S\S~2314 and~2315. The statutes comprising the NSPA cover ``goods, wares, or merchandise,'' and lower courts have largely understood that phrase to be limited to tangible items since the Supreme Court's decision in \textit{Dowling~v.\ United States}, 473~U.S. 207~(1985). \textit{See United States~v.\ Yijia Zhang}, 995F, Supp.~2d 340, 344--48 (E.D. Pa.~2014) (collecting cases). One of those post-\textit{Dowling} decisions---\textit{United States~v.\ Brown}, 925~F.2d 1301 (10th~Cir.~1991)---specifically held that the NSPA does not reach ``a computer program in source code form,'' even though that code was stored in tangible items (\textit{i.e.}, a hard disk and in a three-ring notebook). \textit{Id.}~at~1302--03. Congress, in turn, cited the Brown opinion in explaining the need for amendments to 18~U.S.C. \S~1030(a)(2) that ``would ensure that the theft of intangible information by the unauthorized use of a computer is prohibited in the same way theft of physical items [is] protected.'' S. Rep.~104-357, at~7~(1996). That sequence of events would make it difficult to argue that hacked emails in electronic form, which are the relevant stolen items here, constitute ``goods, wares, or merchandise'' within the meaning of the NSPA.} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter}%1279 \footnote{\blackout{Harm to Ongoing Investigation}} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \subsubsection{Potential Section 1030 Violation By [\protect\censor{Personal Privacy}]} \blackout{Personal Privacy} \blackout{Personal Privacy} \textit{See United States~v.\ Willis}, 476~F.3d 1121, 1125 n.1 (10th~Cir.~2007) (explaining that the 1986 amendments to Section~1030 reflect Congress's desire to reach ``intentional acts of unauthorized access---rather than mistaken, inadvertent, or careless ones'') (quoting S.~Rep.~99-432, at~5~(1986)). In addition, the computer \blackout{Personal Privacy} likely qualifies as a ``protected'' one under the statute, which reaches ``effectively all computers with Internet access.'' \textit{United States~v.\ Nosal}, 676~F.3d 854, 859 (9th~Cir.~2012) (en banc). \blackout{Personal Privacy} Applying the Principles of Federal Prosecution, however, the Office determined that prosecution of this potential violation was not warranted. Those Principles instruct prosecutors to consider, among other things, the nature and seriousness of the offense, the person's culpability in connection with the offense, and the probable sentence to be imposed if the prosecution is successful. Justice Manual \S~9-27.230. \blackout{Personal Privacy} \subsection{Russian Government Outreach and Contacts} As explained in \hyperlink{section.1.4}{Section~IV} above, the Office's investigation uncovered evidence of numerous links (\textit{i.e.}, contacts) between Trump Campaign officials and individuals having or claiming to have ties to the Russian government. The Office evaluated the contacts under several sets of federal laws, including conspiracy laws and statutes governing foreign agents who operate in the United States. After considering the available evidence, the Office did not pursue charges under these statutes against any of the individuals discussed in \hyperlink{section.1.4}{Section~IV} above---with the exception of FARA charges against and based on their activities on behalf of Ukraine. One of the interactions between the Trump Campaign and Russian-affiliated individuals---the June~9, 2016 meeting between high-ranking campaign officials and Russians promising derogatory information on Hillary Clinton---implicates an additional body of law: campaign finance statutes. Schemes involving the solicitation or receipt of assistance from foreign sources raise difficult statutory and constitutional questions. As explained below, the Office evaluated those questions in connection with the June~9 meeting \blackout{Harm to Ongoing Matter}. The Office ultimately concluded that, even if the principal legal questions were resolved favorably to the government, a prosecution would encounter difficulties proving that Campaign officials or individuals connected to the Campaign willfully violated the law. Finally, although the evidence of contacts between Campaign officials and Russian-affiliated individuals may not have been sufficient to establish or sustain criminal charges, several U.S. persons connected to the Campaign made false statements about those contacts and took other steps to obstruct the Office's investigation and those of Congress. This Office has therefore charged some of those individuals with making false statements and obstructing justice. \subsubsection{Potential Coordination: Conspiracy and Collusion} As an initial matter, this Office evaluated potentially criminal conduct that involved the collective action of multiple individuals not under the rubric of ``collusion,'' but through the lens of conspiracy law. In so doing, the Office recognized that the word ``collud[e]'' appears in the Acting Attorney General's August~2, 2017 memorandum; it has frequently been invoked in public reporting; and it is sometimes referenced in antitrust law, \textit{see, e.g., Brooke Group~v.\ Brown \& Williamson Tobacco Corp.}, 509~U.S. 209, 227~(1993). But collusion is not a specific offense or theory of liability found in the U.S. Code; nor is it a term of art in federal criminal law. To the contrary, even as defined in legal dictionaries, collusion is largely synonymous with conspiracy as that crime is set forth in the general federal conspiracy statute, 18~U.S.C. \S~371. \textit{See Black's Law Dictionary} 321 (10th ed.~2014) (collusion is ``[a]n agreement to defraud another or to do or obtain something forbidden by law''); 1~, \textit{A Law Dictionary and Glossary} 311 (1871) (``An agreement between two or more persons to defraud another by the forms of law, or to employ such forms as means of accomplishing some unlawful object.''); 1 \textit{Bouvier's Law Dictionary} 352 (1897) (``An agreement between two or more persons to defraud a person of his rights by the forms of law, or to obtain an object forbidden by law.''). For that reason, this Office's focus in resolving the question of joint criminal liability was on conspiracy as defined in federal law, not the commonly discussed term ``collusion.'' The Office considered in particular whether contacts between Trump Campaign officials and Russia-linked individuals could trigger liability for the crime of conspiracy---either under statutes that have their own conspiracy language (\textit{e.g.}, 18~U.S.C. \S\S~1349, 1951(a)), or under the general conspiracy statute (18~U.S.C. \S~371). The investigation did not establish that the contacts described in \hyperlink{section.1.4}{Volume~I, Section~IV}, \textit{supra}, amounted to an agreement to commit any substantive violation of federal criminal law---including foreign-influence and campaign-finance laws, both of which are discussed further below. The Office therefore did not charge any individual associated with the Trump Campaign with conspiracy to commit a federal offense arising from Russia contacts, either under a specific statute or under Section~371's offenses clause. The Office also did not charge any campaign official or associate with a conspiracy under Section~371's defraud clause. That clause criminalizes participating in an agreement to obstruct a lawful function of the U.S. government or its agencies through deceitful or dishonest means. \textit{See Dennis~v.\ United States}, 384~U.S. 855, 861~(1966); \textit{Hammerschmidt~v.\ United States}, 2605~U.S. 182, 188~(1924); \textit{see also United States~v.\ Concord Mgmt.\ \& Consulting LLC}, 347~F. Supp.~3d 38, 46 (D.D.C. 2018). The investigation did not establish any agreement among Campaign officials---or between such officials and Russia-linked individuals---to interfere with or obstruct a lawful function of a government agency during the campaign or transition period. And, as discussed in \hyperlink{subsection.1.5.1}{Volume~I, Section V.A}, \textit{supra}, the investigation did not identify evidence that any Campaign official or associate knowingly and intentionally participated in the conspiracy to defraud that the Office charged, namely, the active-measures conspiracy described in \hyperlink{section.1.2}{Volume~I, Section~II}, \textit{supra}. Accordingly, the Office did not charge any Campaign associate or other U.S. person with conspiracy to defraud the United States based on the Russia-related contacts described in \hyperlink{section.1.4}{Section~IV} above. \subsubsection{Potential Coordination: Foreign Agent Statutes (FARA and 18~U.S.C. \S~951)} The Office next assessed the potential liability of Campaign-affiliated individuals under federal statutes regulating actions on behalf of, or work done for, a foreign government. \paragraph{Governing Law} Under 18~U.S.C. \S~951, it is generally illegal to act in the United States as an agent of a foreign government without providing notice to the Attorney General. Although the defendant must act on behalf of a foreign government (as opposed to other kinds of foreign entities), the acts need not involve espionage; rather, acts of any type suffice for liability. \textit{See United States~v.\ Duran}, 596~F.3d 1283, 1293--94 (11th~Cir.~2010); \textit{United States~v.\ Latchin}, 554~F.3d 709, 715 (7th~Cir.~2009); \textit{United States~v.\ Dumeisi}, 424~F.3d 566, 581 (7th~Cir.~2005). An ``agent of a foreign government'' is an ``individual'' who ``agrees to operate'' in the United States ``subject to the direction or control of a foreign government or official.'' 18~U.S.C. \S~951(d). The crime defined by Section~951 is complete upon knowingly acting in the United States as an unregistered foreign-government agent. 18~U.S.C. \S~951(a). The statute does not require willfulness, and knowledge of the notification requirement is not an element of the offense. \textit{United States~v.\ Campa}, 529~F.3d 980, 998--99 (11th~Cir.~2008); \textit{Duran}, 596~F.3d at~1291--94; \textit{Dumeisi}, 424~F.3d at~581. The Foreign Agents Registration Act (FARA) generally makes it illegal to act as an agent of a foreign principal by engaging in certain (largely political) activities in the United States without registering with the Attorney General. 22~U.S.C. \S\S~611--621. The triggering agency relationship must be with a foreign principal or ``a person any of whose activities are directly or indirectly supervised, directed, controlled, financed, or subsidized in whole or in major part by a foreign principal.'' 22~U.S.C. \S~611(c)(1). That includes a foreign government or political party and various foreign individuals and entities. 22~U.S.C. \S~611(b). A covered relationship exists if a person ``acts as an agent, representative, employee, or servant'' or ``in any other capacity at the order, request, or under the [foreign principal's] direction or control.'' 22~U.S.C. \S~611(c)(1). It is sufficient if the person ``agrees, consents, assumes or purports to act as, or who is or holds himself out to be, whether or not pursuant to contractual relationship, an agent of a foreign principal.'' 22~U.S.C. \S~611(c)(2). The triggering activity is that the agent ``directly or through any other person'' in the United States (1) engages in ``political activities for or in the interests of [the] foreign principal,'' which includes attempts to influence federal officials or the public; (2) acts as ``public relations counsel, publicity agent, information-service employee or political consultant for or in the interests of such foreign principal''; (3) ``solicits, collects, disburses, or dispenses contributions, loans, money, or other things of value for or in the interest of such foreign principal''; or (4) ``represents the interests of such foreign principal'' before any federal agency or official. 22~U.S.C. \S~611(c)(1). It is a crime to engage in a ``[w]illful violation of any provision of the Act or any regulation thereunder.'' 22~U.S.C. \S~618(a)(1). It is also a crime willfully to make false statements or omissions of material facts in FARA registration statements or supplements. 22~U.S.C. \S~618(a)(2). Most violations have a maximum penalty of five years of imprisonment and a \$10,000 fine. 22~U.S.C. \S~618. \paragraph{Application} The investigation uncovered extensive evidence that 's and 's pre-campaign work for the government of Ukraine violated FARA\null. Manafort and Gates were charged for that conduct and admitted to it when they pleaded guilty to superseding criminal informations in the District of Columbia prosecution.%1280 \footnote{\textit{Gates} Superseding Criminal Information; Waiver of Indictment, \textit{United States~v.\ . Gates~III}, 1:17-cr-201 (D.D.C. Feb.~23, 2018), Doc.~203; Waiver of Trial by Jury, \textit{United States~v.\ . Gates~III}, 1:17-cr-201 (D.D.C. Feb.~23, 2018), Doc.~204; \textit{Gates} Plea Agreement; Statement of Offense, \textit{United States~v.\ ~III}, 1:17-cr-201 (D.D.C. Feb.~23, 2018), Doc.~206; Plea Agreement, \textit{United States~v.\ , Jr.}, 1:17-cr-201 (D.D.C. Sept.~14, 2018), Doc.~422; Statement of Offense, \textit{United States~v.\ , Jr.}, 1:17-cr-201 (D.D.C. Sept.~14, 2018), Doc.~423.} The evidence underlying those charges is not addressed in this report because it was discussed in public court documents and in a separate prosecution memorandum submitted to the Acting Attorney General before the original indictment in that case. In addition, the investigation produced evidence of FARA violations involving . Those potential violations, however, concerned a country other than Russia (\textit{i.e.}, Turkey) and were resolved when Flynn admitted to the underlying facts in the Statement of Offense that accompanied his guilty plea to a false-statements charge. Statement of Offense, \textit{United States~v.\ }, No.~1:17-cr-232 (D.D.C. Dec.~1, 2017), Doc.~4 (``\textit{Flynn} Statement of Offense'').%1281 \footnote{\blackout{Harm to Ongoing Investigation}} The investigation did not, however, yield evidence sufficient to sustain any charge that any individual affiliated with the Trump Campaign acted as an agent of a foreign principal within the meaning of FARA or, in terms of Section~951, subject to the direction or control of the government of Russia, or any official thereof. In particular, the Office did not find evidence likely to prove beyond a reasonable doubt that Campaign officials such as , , and Carter Page acted as agents of the Russian government---or at its direction, control, or request---during the relevant time period.%1282 \footnote{On four occasions, the Foreign Intelligence Surveillance Court (FISC) issued warrants based on a finding of probable cause to believe that Page was an agent of a foreign power. 50~U.S.C. \S\S~1801(b), 1805(a)(2)(A). The FISC's probable-cause finding was based on a different (and lower) standard than the one governing the Office's decision whether to bring charges against Page, which is whether admissible evidence would likely be sufficient to prove beyond a reasonable doubt that Page acted as an agent of the Russian Federation during the period at issue. \textit{Cf.~United States~v.\ Cardoza}, 713~F.3d 656, 660 (D.C.~Cir.~2013) (explaining that probable cause requires only ``a fair probability,'' and not ``certainty, or proof beyond a reasonable doubt, or proof by a preponderance of the evidence'').} \blackout{Personal Privacy} As a result, the Office did not charge \blackout{PP} any other Trump Campaign official with violating FARA or Section~951, or attempting or conspiring to do so, based on contacts with the Russian government or a Russian principal. Finally, the Office investigated whether one of the above campaign advisors------acted as an agent of, or at the direction and control of, the government of Israel. While the investigation revealed significant ties between Papadopoulos and Israel (and search warrants were obtained in part on that basis), the Office ultimately determined that the evidence was not sufficient to obtain and sustain a conviction under FARA or Section~951. \subsubsection{Campaign Finance} Several areas of the Office's investigation involved efforts or offers by foreign nationals to provide negative information about candidate Clinton to the Trump Campaign or to distribute that information to the public, to the anticipated benefit of the Campaign. As explained below, the Office considered whether two of those efforts in particular---the June~9, 2016 meeting at Trump Tower \blackout{Harm to Ongoing Matter}---constituted prosecutable violations of the campaign-finance laws. The Office determined that the evidence was not sufficient to charge either incident as a criminal violation. \paragraph{Overview Of Governing Law} ``[T]he United States has a compelling interest \dots\ in limiting the participation of foreign citizens in activities of democratic self-government, and in thereby preventing foreign influence over the U.S. political process.'' \textit{Bluman~v.\ FEC}, 800~F. Supp.~2d 281, 288 (D.D.C. 2011) (Kavanaugh,~J., for three-judge court), \textit{aff'd}, 565~U.S. 1104~(2012). To that end, federal campaign finance law broadly prohibits foreign nationals from making contributions, donations, expenditures, or other disbursements in connection with federal, state, or local candidate elections, and prohibits anyone from soliciting, accepting, or receiving such contributions or donations. As relevant here, foreign nationals may not make---and no one may ``solicit, accept, or receive'' from them---``a contribution or donation of money or other thing of value'' or ``an express or implied promise to make a contribution or donation, in connection with a Federal, State, or local election.'' 52~U.S.C. \S~30121 (a)(1)(A), (a)(2).%1283 \footnote{Campaign-finance law also places financial limits on contributions, 52~U.S.C. \S~30116(a), and prohibits contributions from corporations, banks, and labor unions, 52~U.S.C. \S~30118(a); \textit{see Citizens United~v.\ FEC}, 558~U.S. 310, 320~(2010). Because the conduct that the Office investigated involved possible electoral activity by foreign nationals, the foreign-contributions ban is the most readily applicable provision.} The term ``contribution,'' which is used throughout the campaign-finance law, ``includes'' ``any gift, subscription, loan, advance, or deposit of money or anything of value made by any person for the purpose of influencing any election for Federal office.'' 52~U.S.C. \S~30101(8)(A)(i). It excludes, among other things, ``the value of [volunteer] services.'' 52~U.S.C. \S~30101(8)(B)G). Foreign nationals are also barred from making ``an expenditure, independent expenditure, or disbursement for an electioneering communication.'' 52~U.S.C. \S~30121(a)(1)(C). The term ``expenditure'' ``includes'' ``any purchase, payment, distribution, loan, advance, deposit, or gift of money or anything of value, made by any person for the purpose of influencing any election for Federal office.'' 52~U.S.C. \S~30101(9)(A)(i). It excludes, among other things, news stories and non-partisan get-out-the-vote activities. 52~U.S.C. \S~30101(9)(B)(i)--(ii). An ``independent expenditure'' is an expenditure ``expressly advocating the election or defeat of a clearly identified candidate'' and made independently of the campaign. 52~U.S.C. \S~30101(17). An ``electioneering communication'' is a broadcast communication that ``refers to a clearly identified candidate for Federal office'' and is made within specified time periods and targeted at the relevant electorate. The statute defines ``foreign national'' by reference to FARA and the Immigration and Nationality Act, with minor modification. 52~U.S.C. \S~30121(b) (cross-referencing 22~U.S.C. \S~611(b)(1)--(3) and 8~U.S.C. \S~1101(a)(20), (22)). That definition yields five, sometimes overlapping categories of foreign nationals, which include all of the individuals and entities relevant for present purposes---namely, foreign governments and political parties, individuals outside of the U.S. who are not legal permanent residents, and certain non-U.S., entities located outside of the~U.S. A ``knowing[] and willful[]'' violation involving an aggregate of \$25,000 or more in a calendar year is a felony. 52~U.S.C. \S~30109(d)(1)(A)(i); \textit{see Bluman}, 800~F. Supp.~2d at~292 (noting that a willful violation will require some ``proof of the defendant's knowledge of the law''); \textit{United States~v.\ Danielczyk}, 917~F. Supp.~2d 573, 577 (E.D. Va.~2013) (applying willfulness standard drawn from \textit{Bryan~v.\ United States}, 524~U.S. 184, 191--92~(1998)); \textit{see also Wagner~v.\ FEC}, 793~F.3d 1, 19 n.23 (D.C.~Cir.~2015) (en banc) (same). A ``knowing[] and willful[]'' violation involving an aggregate of \$2,000 or more in a calendar year, but less than \$25,000, is a misdemeanor. 52~U.S.C. \S~30109(d)(1)(A)(ii). \paragraph{Application to June~9 Trump Tower Meeting} The Office considered whether to charge Trump Campaign officials with crimes in connection with the June~9 meeting described in \hyperlink{subsubsection.1.4.1.5}{Volume~I, Section~IV.A.5}, \textit{supra}. The Office concluded that, in light of the government's substantial burden of proof on issues of intent (``knowing'' and ``willful''), and the difficulty of establishing the value of the offered information, criminal charges would not meet the Justice Manual standard that ``the admissible evidence will probably be sufficient to obtain and sustain a conviction.'' Justice Manual \S~9-27.220. In brief, the key facts are that, on June~3, 2016, emailed Donald Trump~Jr., to pass along from Emin and an ``offer'' from Russia's ``Crown prosecutor'' to ``the Trump campaign'' of ``official documents and information that would incriminate Hillary and her dealings with Russia and would be very useful to [Trump~Jr.'s] father.'' The email described this as ``very high level and sensitive information'' that is ``part of Russia and its government's support to Mr.~Trump---helped along by Aras and Emin.'' Trump~Jr.\ responded: ``if it's what you say I love it especially later in the summer.'' Trump~Jr.\ and had follow-up conversations and, within days, scheduled a meeting with Russian representatives that was attended by Trump~Jr., Manafort, and Kushner. The communications setting up the meeting and the attendance by high-level Campaign representatives support an inference that the Campaign anticipated receiving derogatory documents and information from official Russian sources that could assist candidate Trump's electoral prospects. This series of events could implicate the federal election-law ban on contributions and donations by foreign nationals, 52~U.S.C. \S~30121(a)(1)(A). Specifically, Goldstone passed along an offer purportedly from a Russian government official to provide ``official documents and information'' to the Trump Campaign for the purposes of influencing the presidential election. Trump~Jr.\ appears to have accepted that offer and to have arranged a meeting to receive those materials. Documentary evidence in the form of email chains supports the inference that Kushner and Manafort were aware of that purpose and attended the June~9 meeting anticipating the receipt of helpful information to the Campaign from Russian sources. The Office considered whether this evidence would establish a conspiracy to violate the foreign contributions ban, in violation of 18~U.S.C. \S~371; the solicitation of an illegal foreign source contribution; or the acceptance or receipt of ``an express or implied promise to make a [foreign-source] contribution,'' both in violation of 52~U.S.C. \S~30121(a)(1)(A), (a)(2). There are reasonable arguments that the offered information would constitute a ``thing of value'' within the meaning of these provisions, but the Office determined that the government would not be likely to obtain and sustain a conviction for two other reasons: first, the Office did not obtain admissible evidence likely to meet the government's burden to prove beyond a reasonable doubt that these individuals acted ``willfully,'' \textit{i.e.}, with general knowledge of the illegality of their conduct; and, second, the government would likely encounter difficulty in proving beyond a reasonable doubt that the value of the promised information exceeded the threshold for a criminal violation, \textit{see} 52~U.S.C. \S~30109(d)(1)(A)G). \subparagraph{Thing-of-Value Element} A threshold legal question is whether providing to a campaign ``documents and information'' of the type involved here would constitute a prohibited campaign contribution. The foreign contribution ban is not limited to contributions of money. It expressly prohibits ``a contribution or donation of money or \textit{other thing of value}.'' 52~U.S.C. \S~30121(a)(1)(A), (a)(2) (emphasis added). And the term ``contribution'' is defined throughout the campaign-finance laws to ``include[]'' ``any gift, subscription, loan, advance, or deposit of money or \textit{anything of value}.'' 52~U.S.C. \S~30101(8)(A)(i) (emphasis added). The phrases ``thing of value'' and ``anything of value'' are broad and inclusive enough to encompass at least some forms of valuable information. Throughout the United States Code, these phrases serve as ``term[s] of art'' that are construed ``broad[ly].'' \textit{United States~v.\ Nilsen}, 967~F.2d 539, 542 (11th~Cir.~1992) (per curiam) (``thing of value'' includes ``both tangibles and intangibles''); \textit{see also, e.g.}, 18~U.S.C. \S\S~201(b)(1), 666(a)(2) (bribery statutes); \textit{id.}~\S~641 (theft of government property). For example, the term ``thing of value'' encompasses law enforcement reports that would reveal the identity of informants, \textit{United States~v.\ Girard}, 601~F.2d 69, 71 (2d~Cir.~1979); classified materials, \textit{United States~v.\ Fowler}, 932~F.2d 306, 310 (4th~Cir.~1991); confidential information about a competitive bid, \textit{United States~v.\ Matzkin}, 14~F.3d 1014, 1020 (4th~Cir.~1994); secret grand jury information, \textit{United States~v.\ Jeter}, 775~F.2d 670, 680 (6th~Cir.~1985); and information about a witness's whereabouts, \textit{United States~v.\ Sheker}, 618~F.2d 607, 609 (9th~Cir.~1980) (per curiam). And in the public corruption context, ``\thinspace`thing of value' is defined broadly to include the value which the defendant subjectively attaches to the items received.'' \textit{United States~v.\ Renzi}, 769~F.3d 731, 744 (9th~Cir.~2014) (internal quotation marks omitted). Federal Election Commission (FEC) regulations recognize the value to a campaign of at least some forms of information, stating that the term ``anything of value'' includes ``the provision of any goods or services without charge,'' such as ``membership lists'' and ``mailing lists.'' 11~C.F.R. \S~100.52(d)(1). The FEC has concluded that the phrase includes a state-by-state list of activists. \textit{See Citizens for Responsibility and Ethics in Washington~v.\ FEC}, 475~F.3d 337, 338 (D.C.~Cir.~2007) (describing the FEC's findings). Likewise, polling data provided to a campaign constitutes a ``contribution.'' FEC Advisory Opinion 1990-12 (Strub), 1990~WL 153454 (citing 11~C.F.R. \S~106.4(b)). And in the specific context of the foreign-contributions ban, the FEC has concluded that ``election materials used in previous Canadian campaigns,'' including ``flyers, advertisements, door hangers, tri-folds, signs, and other printed material,'' constitute ``anything of value,'' even though ``the value of these materials may be nominal or difficult to ascertain.'' FEC Advisory Opinion 2007-22 (Hurysz), 2007~WL 5172375, at $\ast$5. These authorities would support the view that candidate-related opposition research given to a campaign for the purpose of influencing an election could constitute a contribution to which the foreign-source ban could apply. A campaign can be assisted not only by the provision of funds, but also by the provision of derogatory information about an opponent. Political campaigns frequently conduct and pay for opposition research. A foreign entity that engaged in such research and provided resulting information to a campaign could exert a greater effect on an election, and a greater tendency to ingratiate the donor to the candidate, than a gift of money or tangible things of value. At the same time, no judicial decision has treated the voluntary provision of uncompensated opposition research or similar information as a thing of value that could amount to a contribution under campaign-finance law. Such an interpretation could have implications beyond the foreign-source ban, \textit{see} 52~U.S.C. \S~30116(a) (imposing monetary limits on campaign contributions), and raise First Amendment questions. Those questions could be especially difficult where the information consisted simply of the recounting of historically accurate facts. It is uncertain how courts would resolve those issues. \subparagraph{Willfulness} Even assuming that the promised ``documents and information that would incriminate Hillary'' constitute a ``thing of value'' under campaign-finance law, the government would encounter other challenges in seeking to obtain and sustain a conviction. Most significantly, the government has not obtained admissible evidence that is likely to establish the scienter requirement beyond a reasonable doubt. To prove that a defendant acted ``knowingly and willfully,'' the government would have to show that the defendant had general knowledge that his conduct was unlawful. U.S. Department of Justice, \textit{Federal Prosecution of Election Offenses}~123 (8th ed.\ Dec.~2017) (``\textit{Election Offenses}''); \textit{see Bluman}, 800~F. Supp. 2d at~292 (noting that a willful violation requires ``proof of the defendant's knowledge of the law''); \textit{Danielczyk}, 917~F. Supp.~2d at~577 (``knowledge of general unlawfulness''). ``This standard creates an elevated scienter element requiring, at the very least, that application of the law to the facts in question be fairly clear. When there is substantial doubt concerning whether the law applies to the facts of a particular matter, the offender is more likely to have an intent defense.'' \textit{Election Offenses}~123. On the facts here, the government would unlikely be able to prove beyond a reasonable doubt that the June~9 meeting participants had general knowledge that their conduct was unlawful. The investigation has not developed evidence that the participants in the meeting were familiar with the foreign-contribution ban or the application of federal law to the relevant factual context. The government does not have strong evidence of surreptitious behavior or efforts at concealment at the time of the June~9 meeting. While the government has evidence of later efforts to prevent disclosure of the nature of the June~9 meeting that could circumstantially provide support for a showing of scienter, \textit{see} \hyperlink{subsection.2.2.7}{Volume~II, Section~II.G}, \textit{infra}, that concealment occurred more than a year later, involved individuals who did not attend the June~9 meeting, and may reflect an intention to avoid political consequences rather than any prior knowledge of illegality. Additionally, in light of the unresolved legal questions about whether giving ``documents and information'' of the sort offered here constitutes a campaign contribution, Trump~Jr.\ could mount a factual defense that he did not believe his response to the offer and the June~9 meeting itself violated the law. Given his less direct involvement in arranging the June~9 meeting, Kushner could likely mount a similar defense. And, while Manafort is experienced with political campaigns, the Office has not developed evidence showing that he had relevant knowledge of these legal issues. \subparagraph{Difficulties in Valuing Promised Information} The Office would also encounter difficulty proving beyond a reasonable doubt that the value of the promised documents and information exceeds the \$2,000 threshold for a criminal violation, as well as the \$25,000 threshold for felony punishment. \textit{See} 52~U.S.C. \S~30109(d)(1). The type of evidence commonly used to establish the value of non-monetary contributions---such as pricing the contribution on a commercial market or determining the upstream acquisition cost or the cost of distribution---would likely be unavailable or ineffective in this factual setting. Although damaging opposition research is surely valuable to a campaign, it appears that the information ultimately delivered in the meeting was not valuable. And while value in a conspiracy may well be measured by what the participants expected to receive at the time of the agreement, \textit{see, e.g., United States~v.\ Tombrello}, 666~F.2d 485, 489 (11th~Cir.~1982), Goldstone's description of the offered material here was quite general. His suggestion of the information's value---\textit{i.e.}, that it would ``incriminate Hillary'' and ``would be very useful to [Trump~Jr.'s] father''---was nonspecific and may have been understood as being of uncertain worth or reliability, given Goldstone's lack of direct access to the original source. The uncertainty over what would be delivered could be reflected in Trump~Jr.'s response (``\textit{if it's what you say} I love it'') (emphasis added). Accordingly, taking into account the high burden to establish a culpable mental state in a campaign-finance prosecution and the difficulty in establishing the required valuation, the Office decided not to pursue criminal campaign-finance charges against Trump~Jr.\ or other campaign officials for the events culminating in the June~9 meeting. \paragraph{Application to [\protect\censor{Harm to Ongoing Matter}]} \blackout{Harm to Ongoing Matter} \subparagraph{Questions Over [\protect\censor{Harm to Ongoing Matter}]} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \subparagraph{Willfulness} As discussed, to establish a criminal campaign-finance violation, the government must prove that the defendant acted ``knowingly and willfully.'' 52~U.S.C. \S~30109(d)(1)(A)(i). That standard requires proof that the defendant knew generally that his conduct was unlawful. \textit{Election Offenses}~123. Given the uncertainties noted above, the ``willfulness'' requirement would pose a substantial barrier to prosecution. \subparagraph{Constitutional Considerations} Finally, the First Amendment could pose constraints on a prosecution. \blackout{Harm to Ongoing Matter} \subparagraph{Analysis [\protect\censor{Harm to Ongoing Matter}]} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \subsubsection{False Statements and Obstruction of the Investigation} The Office determined that certain individuals associated with the Campaign lied to investigators about Campaign contacts with Russia and have taken other actions to interfere with the investigation. As explained below, the Office therefore charged some U.S. persons connected to the Campaign with false statements and obstruction offenses. \paragraph{Overview Of Governing Law} \textit{False Statements}. The principal federal statute criminalizing false statements to government investigators is 18~U.S.C. \S~1001. As relevant here, under Section 1001(a)(2), it is a crime to knowingly and willfully ``make[] any materially false, fictitious, or fraudulent statement or representation'' ``in any matter within the jurisdiction of the executive \dots\ branch of the Government.'' An FBI investigation is a matter within the Executive Branch's jurisdiction. \textit{United States~v.\ Rodgers}, 466~U.S. 475, 479~(1984). The statute also applies to a subset of legislative branch actions---\textit{viz.}, administrative matters and ``investigation[s] or review[s]'' conducted by a congressional committee or subcommittee. 18~U.S.C. \S~1001(c)(1) and~(2); \textit{see United States~v.\ Pickett}, 353~F.3d 62, 66 (D.C.~Cir.~2004). Whether the statement was made to law enforcement or congressional investigators, the government must prove beyond a reasonable doubt the same basic non-jurisdictional elements: the statement was false, fictitious, or fraudulent; the defendant knew both that it was false and that it was unlawful to make a false statement; and the false statement was material. \textit{See, e.g., United States~v.\ Smith}, 831~F.3d 1207, 1222 n.27 (9th~Cir.~2017) (listing elements); \textit{see also} Ninth Circuit Pattern Instruction 8.73 \& cmt.\ (explaining that the Section~1001 jury instruction was modified in light of the Department of Justice's position that the phrase ``knowingly and willfully'' in the statute requires the defendant's knowledge that his or her conduct was unlawful). In the D.C. Circuit, the government must prove that the statement was actually false; a statement that is misleading but ``literally true'' does not satisfy Section 1001(a)(2). \textit{See United States~v.\ Milton}, 8~F.3d 39, 45 (D.C.~Cir.~1993); \textit{United States~v.\ Dale}, 991~F.2d 819, 832--33 \& n.22 (D.C.~Cir.~1993). For that false statement to qualify as ``material,'' it must have a natural tendency to influence, or be capable of influencing, a discrete decision or any other function of the agency to which it is addressed. \textit{See United States~v.\ Gaudin}, 515~U.S. 506, 509~(1995); \textit{United States~v.\ Moore}, 612~F.3d 698, 701 (D.C.~Cir.~2010). \textit{Perjury}. Under the federal perjury statutes, it is a crime for a witness testifying under oath before a grand jury to knowingly make any false material declaration. \textit{See} 18~U.S.C. \S~1623. The government must prove four elements beyond a reasonable doubt to obtain a conviction under Section~1623(a): the defendant testified under oath before a federal grand jury; the defendant's testimony was false in one or more respects; the false testimony concerned matters that were material to the grand jury investigation; and the false testimony was knowingly given. \textit{United States~v.\ Bridges}, 717~F.2d 1444, 1449 n.30 (D.C.~Cir.~1983). The general perjury statute, 18~U.S.C. \S~1621, also applies to grand jury testimony and has similar elements, except that it requires that the witness have acted willfully and that the government satisfy ``strict common-law requirements for establishing falsity.'' \textit{See Dunn~v.\ United States}, 442~U.S. 100, 106 \& n.6~(1979) (explaining ``the two-witness rule'' and the corroboration that it demands). \textit{Obstruction of Justice}. Three basic elements are common to the obstruction statutes pertinent to this Office's charging decisions: an obstructive act; some form of nexus between the obstructive act and an official proceeding; and criminal (\textit{i.e.}, corrupt) intent. A detailed discussion of those elements, and the law governing obstruction of justice more generally, is included in \hyperref[chap:volume-2]{Volume~II} of the report. \paragraph{Application to Certain Individuals} \subparagraph{} Investigators approached Papadopoulos for an interview based on his role as a foreign policy advisor to the Trump Campaign and his suggestion to a foreign government representative that Russia had indicated that it could assist the Campaign through the anonymous release of information damaging to candidate Clinton. On January~27, 2017, Papadopoulos agreed to be interviewed by FBI agents, who informed him that the interview was part of the investigation into potential Russian government interference in the 2016 presidential election. During the interview, Papadopoulos lied about the timing, extent, and nature of his communications with , , and . With respect to timing, Papadopoulos acknowledged that he had met Mifsud and that Mifsud told him the Russians had ``dirt'' on Clinton in the form of ``thousands of emails.'' But Papadopoulos stated multiple times that those communications occurred before he joined the Trump Campaign and that it was a ``very strange coincidence'' to be told of the ``dirt'' before he started working for the Campaign. This account was false. Papadopoulos met Mifsud for the first time on approximately March~14, 2016, after Papadopoulos had already learned he would be a foreign policy advisor for the Campaign. Mifsud showed interest in Papadopoulos only after learning of his role on the Campaign. And Mifsud told Papadopoulos about the Russians possessing ``dirt'' on candidate Clinton in late April 2016, more than a month after Papadopoulos had joined the Campaign and been publicly announced by candidate Trump. Statement of Offense \P\P~25--26, \textit{United States~v.\ George Papadopoulos}, No.~1:17-cr-182 (D.D.C. Oct.~5, 2017), Doc.~19 (``\textit{Papadopoulos} Statement of Offense''). Papadopoulos also made false statements in an effort to minimize the extent and importance of his communications with Mifsud. For example, Papadopoulos stated that ``[Mifsud]'s a nothing,'' that he thought Mifsud was ``just a guy talk[ing] up connections or something,'' and that he believed Mifsud was ``BS'ing to be completely honest with you.'' In fact, however, Papadopoulos understood Mifsud to have substantial connections to high-level Russian government officials and that Mifsud spoke with some of those officials in Moscow before telling Papadopoulos about the ``dirt.'' Papadopoulos also engaged in extensive communications over a period of months with Mifsud about foreign policy issues for the Campaign, including efforts to arrange a ``history making'' meeting between the Campaign and Russian government officials. In addition, Papadopoulos failed to inform investigators that Mifsud had introduced him to Timofeev, the Russian national who Papadopoulos understood to be connected to the Russian Ministry of Foreign Affairs, despite being asked if he had met with Russian nationals or ``[a]nyone with a Russian accent'' during the campaign. \textit{Papadopoulos} Statement of Offense \P\P~27--29. Papadopoulos also falsely claimed that he met Polonskaya before he joined the Campaign, and falsely told the FBI that he had ``no'' relationship at all with her. He stated that the extent of their communications was her sending emails---``Just, `Hi, how are you?' That's it.'' In truth, however, Papadopoulos met Polonskaya on March~24, 2016, after he had joined the Campaign; he believed that she had connections to high-level Russian government officials and could help him arrange a potential foreign policy trip to Russia. During the campaign he emailed and spoke with her over Skype on numerous occasions about the potential foreign policy trip to Russia. \textit{Papadopoulos} Statement of Offense \P\P~30--31. Papadopoulos's false statements in January 2017 impeded the FBI's investigation into Russian interference in the 2016 presidential election. Most immediately, those statements hindered investigators' ability to effectively question Mifsud when he was interviewed in the lobby of a Washington, D.C. hotel on February~10, 2017. \textit{See} Gov't Sent.~Mem.\ at~6, \textit{United States~v.\ George Papadopoulos}, No.~1:17-cr-182 (D.D.C. Aug.~18, 2017), Doc.~44. During that interview, Mifsud admitted to knowing Papadopoulos and to having introduced him to Polonskaya and Timofeev. But Mifsud denied that he had advance knowledge that Russia was in possession of emails damaging to candidate Clinton, stating that he and Papadopoulos had discussed cybersecurity and hacking as a larger issue and that Papadopoulos must have misunderstood their conversation. Mifsud also falsely stated that he had not seen Papadopoulos since the meeting at which Mifsud introduced him to Polonskaya, even though emails, text messages, and other information show that Mifsud met with Papadopoulos on at least two other occasions---April~12 and April~26, 2016. In addition, Mifsud omitted that he had drafted (or edited) the follow-up message that Polonskaya sent to Papadopoulos following the initial meeting and that, as reflected in the language of that email chain (``Baby, thank you!''), Mifsud may have been involved in a personal relationship with Polonskaya at the time. The false information and omissions in Papadopoulos's January 2017 interview undermined investigators' ability to challenge Mifsud when he made these inaccurate statements. Given the seriousness of the lies and omissions and their effect on the FBI's investigation, the Office charged Papadopoulos with making false statements to the FBI, in violation of 18~U.S.C. \S~1001. Information, \textit{United States~v.\ George Papadopoulos}, No.~1:17-cr-182 (D.D.C. Oct.~3, 2017), Doc.~8. On October~7, 2017, Papadopoulos pleaded guilty to that charge pursuant to a plea agreement. On September~7, 2018, he was sentenced to 14 days of imprisonment, a \$9,500 fine, and 200 hours of community service. \subparagraph{[\protect\censor{Personal Privacy}]} \blackout{Grand Jury} \blackout{Grand Jury} \subparagraph{} agreed to be interviewed by the FBI on January~24, 2017, four days after he had officially assumed his duties as National Security Advisor to the President. During the interview, Flynn made several false statements pertaining to his communications with the Russian ambassador. First, Flynn made two false statements about his conversations with Russian Ambassador Kislyak in late December 2016, at a time when the United States had imposed sanctions on Russia for interfering with the 2016 presidential election and Russia was considering its response. \textit{See Flynn} Statement of Offense. Flynn told the agents that he did not ask Kislyak to refrain from escalating the situation in response to the United States's imposition of sanctions. That statement was false. On December~29, 2016, Flynn called Kislyak to request Russian restraint. Flynn made the call immediately after speaking to a senior Transition Team official (K.T. McFarland) about what to communicate to~Kislyak. Flynn then spoke with McFarland again after the Kislyak call to report on the substance of that conversation. Flynn also falsely told the FBI that he did not remember a follow-up conversation in which Kislyak stated that Russia had chosen to moderate its response to the U.S. sanctions as a result of Flynn's request. On December~31, 2016, Flynn in fact had such a conversation with Kislyak, and he again spoke with McFarland within hours of the call to relay the substance of his conversation with~Kislyak. \textit{See Flynn} Statement of Offense \P~3. Second, Flynn made false statements about calls he had previously made to representatives of Russia and other countries regarding a resolution submitted by Egypt to the United Nations Security Council on December~21, 2016. Specifically, Flynn stated that he only asked the countries' positions on how they would vote on the resolution and that he did not request that any of the countries take any particular action on the resolution. That statement was false. On December~22, 2016, Flynn called Kislyak, informed him of the incoming Trump Administration's opposition to the resolution, and requested that Russia vote against or delay the resolution. Flynn also falsely stated that Kislyak never described Russia's response to his December~22 request regarding the resolution. Kislyak in fact told Flynn in a conversation on December~23, 2016, that Russia would not vote against the resolution if it came to a vote. \textit{See Flynn} Statement of Offense \P~4. Flynn made these false statements to the FBI at a time when he was serving as National Security Advisor and when the FBI had an open investigation into Russian interference in the 2016 presidential election, including the nature of any links between the Trump Campaign and Russia. Flynn's false statements and omissions impeded and otherwise had a material impact on that ongoing investigation. \textit{Flynn} Statement of Offense \P\P~1--2. They also came shortly before Flynn made separate submissions to the Department of Justice, pursuant to FARA, that also contained materially false statements and omissions. \textit{Id.}~\P~5. Based on the totality of that conduct, the Office decided to charge Flynn with making false statements to the FBI, in violation of 18~U.S.C. \S~1001(a). On December~1, 2017, and pursuant to a plea agreement, Flynn pleaded guilty to that charge and also admitted his false statements to the Department in his FARA filing. \textit{See id.}; Plea Agreement, \textit{United States~v.\ }, No.~1:17-cr-232 (D.D.C. Dec.~1, 2017), Doc.~3. Flynn is awaiting sentencing. \subparagraph{} was the executive vice president and special counsel to the Trump Organization when Trump was president of the Trump Organization. Information \P~1, \textit{United States~v.\ Cohen}, No.~1:18-cr-850 (S.D.N.Y. Nov.~29, 2018), Doc.~2 (``\textit{Cohen} Information''). From the fall of 2015 through approximately June 2016, Cohen was involved in a project to build a Trump-branded tower and adjoining development in Moscow. The project was known as Trump Tower Moscow. In 2017, Cohen was called to testify before the House Permanent Select Committee on Intelligence (HPSCI) and the Senate Select Committee on Intelligence (SSCI), both of which were investigating Russian interference in the 2016 presidential election and possible links between Russia and the presidential campaigns. In late August 2017, in advance of his testimony, Cohen caused a two-page statement to be sent to SSCI and HPSCI addressing Trump Tower Moscow. \textit{Cohen} Information \P\P~2--3. The letter contained three representations relevant here. First, Cohen stated that the Trump Moscow project had ended in January 2016 and that he had briefed candidate Trump on the project only three times before making the unilateral decision to terminate it. Second, Cohen represented that he never agreed to travel to Russia in connection with the project and never considered asking Trump to travel for the project. Third, Cohen stated that he did not recall any Russian government contact about the project, including any response to an email that he had sent to a Russian government email account. \textit{Cohen} Information \P~4. Cohen later asked that his two-page statement be incorporated into his testimony's transcript before SSCI, and he ultimately gave testimony to SSCI that was consistent with that statement. \textit{Cohen} Information \P~5. Each of the foregoing representations in Cohen's two-page statement was false and misleading. Consideration of the project had extended through approximately June 2016 and included more than three progress reports from Cohen to Trump. Cohen had discussed with his own travel to Russia as part of the project, and he had inquired about the possibility of Trump traveling there---both with the candidate himself and with senior campaign official . Cohen did recall that he had received a response to the email that he sent to Russian government spokesman ---in particular, that he received an email reply and had a follow-up phone conversation with an English-speaking assistant to Peskov in mid-January~2016. \textit{Cohen} Information \P~7. Cohen knew the statements in the letter to be false at the time, and admitted that he made them in an effort (1) to minimize the links between the project and Trump (who by this time was President), and (2) to give the false impression that the project had ended before the first vote in the Republican Party primary process, in the hopes of limiting the ongoing Russia investigations. \textit{Id.} Given the nature of the false statements and the fact that he repeated them during his initial interview with the Office, we charged Cohen with violating Section~1001. On November~29, 2018, Cohen pleaded guilty pursuant to a plea agreement to a single-count information charging him with making false statements in a matter within the jurisdiction of the legislative branch, in violation of 18~U.S.C. \S~1001(a)(2) and (c). \textit{Cohen} Information. The case was transferred to the district judge presiding over the separate prosecution of Cohen pursued by the Southern District of New York (after a referral from our Office). On December~7, 2018, this Office submitted a letter to that judge recommending that Cohen's cooperation with our investigation be taken into account in sentencing Cohen on both the false-statements charge and the offenses in the Southern District prosecution. On December~12, 2018, the judge sentenced Cohen to two months of imprisonment on the false-statements count, to run concurrently with a 36-month sentence imposed on the other counts. \subparagraph{[\protect\censor{Harm to Ongoing Matter}]} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \blackout{Harm to Ongoing Matter} \subparagraph{Jeff Sessions} As set forth in \hyperlink{subsubsection.1.4.1.6}{Volume~I, Section~IV.A.6}, \textit{supra}, the investigation established that, while a U.S. Senator and a Trump Campaign advisor, former Attorney General interacted with Russian Ambassador Kislyak during the week of the Republican National Convention in July 2016 and again at a meeting in Sessions's Senate office in September 2016. The investigation also established that Sessions and Kislyak both attended a reception held before candidate Trump's foreign policy speech at the Mayflower Hotel in Washington, D.C., in April 2016, and that it is possible that they met briefly at that reception. The Office considered whether, in light of these interactions, Sessions committed perjury before, or made false statements to, Congress in connection with his confirmation as Attorney General. In January 2017 testimony during his confirmation hearing, Sessions stated in response to a question about Trump Campaign communications with the Russian government that he had ``been called a surrogate at a time or two in that campaign and I didn't have -- did not have communications with the Russians.'' In written responses submitted on January~17, 2017, Sessions answered ``[n]o'' to a question asking whether he had ``been in contact with anyone connected to any part of the Russian government about the 2016 election, either before or after election day.'' And, in a March 2017 supplement to his testimony, Sessions identified two of the campaign-period contacts with Ambassador Kislyak noted above, which had been reported in the media following the January 2017 confirmation hearing. Sessions stated in the supplemental response that he did ``not recall any discussions with the Russian Ambassador, or any other representatives of the Russian government, regarding the political campaign on these occasions or any other occasion.'' Although the investigation established that Sessions interacted with Kislyak on the occasions described above and that Kislyak mentioned the presidential campaign on at least one occasion, the evidence is not sufficient to prove that Sessions gave knowingly false answers to Russia-related questions in light of the wording and context of those questions. With respect to Sessions's statements that he did ``not recall any discussions with the Russian Ambassador \dots\ regarding the political campaign'' and he had not been in contact with any Russian official ``about the 2016 election,'' the evidence concerning the nature of Sessions's interactions with Kislyak makes it plausible that Sessions did not recall discussing the campaign with Kislyak at the time of his statements. Similarly, while Sessions stated in his January 2017 oral testimony that he ``did not have communications with Russians,'' he did so in response to a question that had linked such communications to an alleged ``continuing exchange of information'' between the Trump Campaign and Russian government intermediaries. Sessions later explained to the Senate and to the Office that he understood the question as narrowly calling for disclosure of interactions with Russians that involved the exchange of campaign information, as distinguished from more routine contacts with Russian nationals. Given the context in which the question was asked, that understanding is plausible. Accordingly, the Office concluded that the evidence was insufficient to prove that Sessions was willfully untruthful in his answers and thus insufficient to obtain or sustain a conviction for perjury or false statements. Consistent with the Principles of Federal Prosecution, the Office therefore determined not to pursue charges against Sessions and informed his counsel of that decision in March 2018. \subparagraph{Others Interviewed During the Investigation} The Office considered whether, during the course of the investigation, other individuals interviewed either omitted material information or provided information determined to be false. Applying the Principles of Federal Prosecution, the Office did not seek criminal charges against any individuals other than those listed above. In some instances, that decision was due to evidentiary hurdles to proving falsity. In others, the Office determined that the witness ultimately provided truthful information and that considerations of culpability, deterrence, and resource-preservation weighed against prosecution. \textit{See} Justice Manual \S\S~9-27.220, 9-27.230. \blackout{Personal Privacy} \blackout{Personal Privacy} \blackout{Personal Privacy} \blackout{Grand Jury} \blackout{Personal Privacy} \blackout{Grand Jury} \blackout{Personal Privacy} \blackout{Personal Privacy} \blackout{Grand Jury} \blackout{Personal Privacy} \blackout{Personal Privacy} \blackout{Personal Privacy} @inproceedings{DASC2016, author = {}, booktitle = {2016 IEEE/AIAA 35th Digital Avionics Systems Conference (DASC)}, doi = {10.1109/DASC.2016.7778089}, number = {}, pages = {1-8}, title = {A runtime reconfiguration design targeting avionics systems}, volume = {}, year = {2016} } documentation/njpipes/hash.tex \section{hash} \index{hash} \begin{shaded} \begin{alltt} /** hash Call hashr at the start of a pipe segment \end{alltt} \end{shaded} @incollection{kidwelldoherty_2015a, booktitle = {Theology and Economics: A Christian Vision of the Common Good}, date-added = {2020-05-03 20:07:00 +0100}, date-modified = {2020-05-03 20:07:00 +0100}, editor = { and }, publisher = {Palgrave MacMillan}, title = {Radical or Realist? The Ethics of Work in John Chrysostom}, year = {2015} } \begin{minipage}[c]{0.49\textwidth} \begin{flushleft} \begin{tikzpicture}[scale=1.5]%,cap=round,>=latex] \draw (0,1) rectangle (4,2); \filldraw[fill=white] (1,0) rectangle (3,4); \filldraw[fill=white] (0,2) rectangle (4,3); \end{tikzpicture} TOP \end{flushleft} \end{minipage} \vfill \begin{minipage}[c]{0.49\textwidth} \begin{flushleft} \begin{tikzpicture}[scale=1.5]%,cap=round,>=latex] \draw (0,0) rectangle (1,4); \draw (3,0) rectangle (4,4); \draw (1.5,0) rectangle (2.5,2); \draw (1,2) rectangle (3,3); \draw (0,4) rectangle (4,6); \end{tikzpicture} FRONT \end{flushleft} \end{minipage}% \begin{minipage}[c]{0.5\textwidth} \flushright \begin{center} \begin{tikzpicture}[scale=1.5]%,cap=round,>=latex] \draw (0,0) rectangle (4,2); \draw (0,2) rectangle (4,3); \filldraw[fill=white] (1,0) rectangle (3,4); \draw (2,4) rectangle (3,6); \end{tikzpicture} RIGHT \end{center} \end{minipage} \vfill docs/tex/fbmathlit.bib @PROCEEDINGS{2001a, title = {Error Estimation and Solution Adaptive Discretization in {CFD}}, year = {2001}, note = {NASA Ames Research Center, September 10-14, 2001}, } @PROCEEDINGS{BallHunt2000, editor = {. and .}, title = {ICIAM 99, Proceedings of the Fourth International Congress on Industrial and Applied Mathematics}, publisher = {Oxford University Press}, year = {2000}, } @PROCEEDINGS{GaldiRannacher2002, editor = {. and .}, title = {Topics in Mathematical Fluid Mechanics}, publisher = {Aracne}, year = {2002}, note = {Buch steht im Lehrstuhlhandapparat}, } @PROCEEDINGS{HaussmannJetterReimerStoeckler2002, editor = {. and \"{.}, title = {Modern Developments in Multivariate Approximation}, publisher = {Birkh\"{a}user}, year = {2002}, note = {Buch ist ein Geschenk; steht im Lehrstuhlhandapparat}, } @PROCEEDINGS{Schindler2004, editor = {.}, title = {2nd International Berlin Workshop {}--{} IBW2 on Transport Phenomena with Moving Boundaries}, year = {2004}, } @PROCEEDINGS{Flaherty2005, editor = {.}, title = {Selected papers from the 16th Chemnitz Finite Element Symposium}, publisher = {Elsevier}, year = {2005}, } @PROCEEDINGS{HuelsemannKowarschikRuede2005, editor = {. and . and R\"{u}.}, title = {Simulationstechnique 18th Symposium in Erlangen}, publisher = {SCS Publisching House e.V.}, year = {2005}, note = {ASIM 2005, September 2005}, } @TECHREPORT{BlumHarigMuellerTurek1992, author = {. and \"{ .}, title = {FEAT2D . Finite element analysis tools. User Manual. Release 1.3}, institution = {Universit\"{a}t Heidelberg}, year = {1992}, } @TECHREPORT{Turek1998, author = {.}, title = {FEATFLOW . Finite element software for the incompressible {Navier}--{Stokes} equations: User Manual, Release 1.1}, institution = {Universit\"{a}t Heidelberg}, year = {1998}, } @TECHREPORT{Turek1998a, author = {.}, title = {OMEGA2D: User Manual (deutsch)}, institution = {Universit\"{a}t Heidelberg}, year = {1998}, note = {with the //}, } @TECHREPORT{BeckerGoeddeke2002, author = { G\"{.}, title = {DeViSoR Grid}, institution = {Universit\"{a}t Dortmund}, year = {2002}, address = {\url{http:/ / www.featflow.de/ feast_hp/ devisormain.html}}, } @TECHREPORT{AVS1998, author = {AVS}, title = {Introduction to {AVS}}, institution = {AVS centre}, year = {1998}, address = {\url{http:/ / www.iavsc.org/ training/ express/ index.html#introxp}}, } @TECHREPORT{Various1996, author = {Various}, title = {Java allgemein}, institution = {Sun Microsystems}, year = {1996}, address = {\url{http:/ / java.sun.com/ docs/ books/ tutorial}}, } @TECHREPORT{Various1996a, author = {Various}, title = {Java3D}, institution = {Sun Microsystems}, year = {1996}, address = {\url{http:/ / java.sun.com/ products/ java-media/ 3D/ index.html}}, } @TECHREPORT{Martin2002, author = {.}, title = {Propagation of Singularities}, institution = {Deutsche Forschungsgesellschaft}, year = {2002}, month = jul, note = {ISBN 3-88722-533-3}, type = {Preprint Series DFG SPP 1114}, } @TECHREPORT{MuellerTimmer2002, author = {. and .}, title = {Fitting parameters in partial differential equations from partially observed noisy data}, institution = {Deutsche Forschungsgesellschaft}, year = {2002}, month = aug, note = {ISBN3-88722-536-8}, type = {Preprint Series DFG SPP 1114}, } @TECHREPORT{DahlkeSteidlTeschke2002, author = { .}, title = {Coorbit Spaces and Banach Frames on Homogeneous Spaces with Applications to the Sphere}, institution = {Deutsche Forschungsgesellschaft}, year = {2002}, month = aug, note = {ISBN 3-88722-537-6}, type = {Preprint Series DFG -SPP 1114}, } @TECHREPORT{KlingmuellerMuellerSandraSwameyeTimmer2002, author = { .}, title = {Modeling the non--linear dynamics of cellular signal transduction}, institution = {Deutsche Forschungsgesellschaft}, year = {2002}, month = sep, note = {ISBN3-88722-539-2}, type = {Preprint Series DFG-SPP 1114}, } @TECHREPORT{KurthsRomanoSchwarzThielTimmer2002, author = {. and .}, title = {Surrogate Based Hypothesis Test without Surrogates}, institution = {Deutsche Forschungsgesellschaft}, year = {2002}, month = sep, note = {ISBN3-88722-540-6}, type = {Preprint Series DFG-SPP 1114}, } @TECHREPORT{Otto2002, author = {.}, title = {Das DeViSoR--Framework}, institution = {Universit\"{a}t Dortmund}, year = {2002}, note = {\url{http:/ / ls7-www.cs.uni-dortmund.de/ pg/ devisor/ internal}}, } @TECHREPORT{TurekKilian2002, author = {. and .}, title = {The Virtual Album of Fluid Motion}, institution = {Universit\"{a}t Dortmund}, year = {2002}, note = {Springer (Multimedia DVD), ISBN 3-540-14900-7}, } @TECHREPORT{TurekBecker1999, author = {. and .}, title = {FEATFLOW {}--{} Finite element software for the incompressible {Navier}--{Stokes} equations}, institution = {Universit\"{a}t Dortmund}, year = {1999}, type = {User Manual}, } @TECHREPORT{KilianTurek2000, author = {. and .}, title = {Virtual Album of Fluid Motion}, institution = {Universit\"{a}t Dortmund}, year = {2000}, address = {\url{http:/ / www.featflow.de/ album/ index.html}}, } @TECHREPORT{RannacherTurekWehrse1994, author = {. and . and .}, title = {Numerical Methods for Multidimensional Radiative Transfer Problems}, institution = {Universit\"{a}t Heidelberg}, year = {1994}, note = {SFB 359 - Contributions to a workshop at the University of Heidelberg, 10.-12. December 1994}, } @TECHREPORT{BeckerTurek1998, author = {. and .}, title = {FEATFLOW Finite Element Software for the Incompressible {Navier}--{Stokes} equations {}--{} User Manual Release 1.1}, institution = {Universit\"{a}t Heidelberg}, year = {1998}, } @TECHREPORT{JohnMatthies2001, author = {. and .}, title = {Higher order finite element discretizations in a benchmark problem for incompressible flows}, institution = {Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2001}, address = {Institut f\"{u}r Analysis und Numerik, Postfach 4120, 39016 Magdeburg}, type = {Preprint}, } @TECHREPORT{AVS2003, author = {AVS}, title = {Advanced Visualization Systems}, institution = {Advanced Visualization Systems}, year = {2003}, note = {\url{http:/ / www.avs.com}}, } @TECHREPORT{NumericalAlgorithmsGroup2003, author = {Numerical Algorithms Group}, title = {IRIS Explorer}, institution = {IRIS Explorer}, year = {2003}, note = {www.nag.co.uk}, } @TECHREPORT{GNU1991, author = {GNU}, title = {Free Software Foundation: The GNU General Public Licence (GPL)}, institution = {Free Software Foundation: The GNU General Public Licence (GPL)}, year = {1991}, note = {\url{http:/ / www.gnu.org/ licenses/ Licenses.html#GPL} }, } @TECHREPORT{Borland2002, author = {Borland}, title = {TogetherJ 6.0}, institution = {TogetherJ 6.0}, year = {2002}, note = {www.togethersoft.com}, } @TECHREPORT{quicktime2003, author = {quicktime}, title = {Apple Computer Inc.: QuickTime Documentation}, institution = {Apple Computer Inc.: QuickTime Documentation}, year = {2003}, note = {\url{http:/ / developer.apple.com/ documentation/ QuickTime/ QuickTime.html}}, } @TECHREPORT{Swing1996, author = {Swing}, title = {Java allgemein}, institution = {Sun Microsysteme}, year = {1996}, note = {\url{http:/ / java.sun.com/ docs/ books/ tutorial/ uiswing/ index.html}}, } @TECHREPORT{Risch2003, author = {Risch}, title = {Superconvergence of a nonconforming low order finite element}, institution = {Institute for Analysis and Numerics, Otto-von-Guericke-University Magdeburg}, year = {2003}, month = nov, type = {Preprint}, number = {14}, } @TECHREPORT{SouadniaLesageLatifi2002, author = { and Latifi}, title = {Some computational aspects in the simulation of Hydrodynamics in a Trickle--Bed Reactor}, institution = {Laboratoire des Sciences du Genie Chimique}, year = {2002}, type = {Preprint}, } @TECHREPORT{LapinMuellerReuss2003, author = {Lapin and M\"{u}ller and Reuss}, title = {Dynamic Behaviour of Microbial Populations in Stirred Bioreactors Simulated with Euler--{Lagrange}--Methods {}--{} Traveling Along the Lifelines of Single Cells}, institution = {Institute of Biochemical Engeneering, University of Stuttgart}, year = {2003}, month = oct, type = {Preprint}, } @TECHREPORT{GelhardLubeOlshanskii2002, author = {. and .}, title = {Stabilized finite element schemes with LBB--stable elements for incompressible flows}, institution = {Math. Dep.,University of G\"{o}ttingen, D-37093, Germany}, year = {2002}, month = nov, type = {Preprint}, } @TECHREPORT{Lee2003, author = {Lee}, title = {Flow Structures and Pressure Drop Across Bell--Shaped Constrictions in a Tube}, institution = {Mechanical Engeneering Department, Natioinal University of Singapore}, year = {2003}, address = {Singapore 119260}, month = oct, type = {Technischer Report}, } @TECHREPORT{Blender2003, author = {Blender}, title = {Blender 3D}, institution = {Blender}, year = {2003}, note = {\url{http:/ / www.blender3d.org}}, } @TECHREPORT{AutodescInc.2003, author = {Autodesk.}, title = {DXF Reference 2004}, institution = {Autodesc Inc.}, year = {2003}, month = feb, note = {\url{http:/ / www.autodesc.com}}, type = {Manual}, } @TECHREPORT{AckerTurek1998, author = {. and .}, title = {Arbeiten mit {GMV} unter Featflow}, institution = {Universit\"{a}t Heidelberg}, year = {1998}, type = {Preprint}, number = {98--50}, } @TECHREPORT{KuzminTurek2002, author = {. and .}, title = {High--Performance--Computing f\"{u}r Mehrphasenstr\"{o}mungen}, institution = {Universit\"{a}t Dortmund}, year = {2002}, note = {Uni-Report 35}, } @TECHREPORT{Elman2003, author = {.}, title = {Preconditioning strategies for models of incompressible flow}, institution = {Department of computer science and institute for advanced computer studies, University of Maryland}, year = {2003}, address = {University of Maryland, College Park, MD 20742}, note = {CFD1 NLA3 NLA6}, number = {CS-TR \#4543 / UMIACS TR \#2003-111}, } @TECHREPORT{ElmanLoghinWathen2003, author = { .}, title = {Preconditioning techniques for Newton`s method for the incompressible {Navier}--{Stokes} equations}, institution = {Department of Computer Science, University of Maryland}, year = {2003}, address = {University of Maryland, College Park, MD 20742}, type = {Technical Report}, number = {CERFACS Technical Report TR/PA/03/31}, } @TECHREPORT{BurmanJacotPicasso2003, author = { .}, title = {Adaptive finite elements with high aspect ratio for the computation of coalescence in binary alloys}, institution = {Institut de Mathematiques, Faculte des Science de Base, Ecole Polytechnique Federale de Lausanne}, year = {2003}, address = {Ecole Polytechnique Federale de Lausanne, 1015 Lausanne, Switzerland}, month = jun, type = {Preprint}, } @TECHREPORT{HansboHermanssonSvedberg2003, author = {. and Hermansson and Svedberg}, title = {Nitsche`s method combined with space--time finite elements for ALE fluid--structure interaction problems}, institution = {Chalmers Finite Element Center}, year = {2003}, address = {Chalmers University Of Technology, G\"{o}teberg Sweden}, type = {Preprint}, number = {2003-14}, } @TECHREPORT{Risch2004, author = {Risch}, title = {Superconvergence of a nonconforming low order finite element}, institution = {Institute for Analysis and Numerics, Magdeburg, Germany}, year = {2004}, address = {Otto von Guericke University, 39016 Magdenburg, Germany}, month = feb, type = {Preprint}, number = {PF 4120}, } @TECHREPORT{HoffmanJohnson2004, author = {. and .}, title = {Computability and adaptivity in {CFD}}, institution = {Courant Institute, New York}, year = {2004}, address = {Courant institute, 2541 Mercer Street, New York, NY-10012, USA}, note = {Encyclopedia of Computational Mechanics}, } @TECHREPORT{Massjung2003, author = {Massjung}, title = {Discretization and Algorithms for Strong Coupling in Computational Aeroelasticy}, institution = {Institut fuer Geometrie und Praktische Mathematik, RWTH Aachen, Germany}, year = {2003}, address = {RWTH Aachen, Germany}, month = oct, type = {Preprint}, } @TECHREPORT{SnavelyChunCasanovavanderWijngaartFrumkin2003, author = { Frumkin}, title = {Benchmarks for Grid Computing: A Review of Ongoing Efforts and Future Directions}, institution = {University of California}, year = {2003}, type = {Technical Report}, } @TECHREPORT{LuDinda2003, author = {}, title = {GridG: Generating Realistic Computational Grids}, institution = {Department of Computer Science, Northwestern University}, year = {2003}, type = {Technical Report}, } @TECHREPORT{ZhangZhu1995, author = {. and .}, title = {Analysis of the Superconvergent Patch Recovery Technique and A Posteriori Error Estimator in the Finite Element Method I}, institution = {Texas Tech University, Lubbock, TX 79409}, year = {1994}, type = {Technical Report}, } @TECHREPORT{ZhangZhu1995a, author = {. and .}, title = {Analysis of the Superconvergent Patch Recovery Technique and A Posteriori Error Estimator in the Finite Element Method II}, institution = {Texas Tech University, Lubbock, TX 79409}, year = {1995}, type = {Technical Report}, } @TECHREPORT{Schulz2002, author = {.}, title = {Cache oriented implementastion for numerical codes}, institution = {IWRMM, University of Karlsruhe, Germany}, year = {2002}, address = {\url{http:/ / www.mathematik.uni-karlsruhe.de/ ~schulz/ Preprint}}, type = {Preprint}, } @TECHREPORT{BrandtsKrizek2000, author = {. and .}, title = {History and Future of Superconvergence in Three--Dimensional Finite Element Methods}, institution = {Mathematical institute, Utrecht University, Utrecht, Netherlands}, year = {2000}, address = {Utrecht University, Budapestlaan 6, 3584 CD, Utrecht, The Netherlands}, month = sep, } @TECHREPORT{KnoblauchTobiska2003, author = {. and .}, title = {On Korn`s first inequaluty for quadrilateral nonconforming Finite Elements of first order approximation properties}, institution = {Fakultaet fuer Mathematik, Otto-von-Guericke-Universitaet Magdeburg}, year = {2003}, type = {Preprint}, number = {Preprint Nr. 28}, } @TECHREPORT{CarstensenVerfuerth2003, author = {. and Verf\"{.}, title = {Edge residuals dominate a posteriori error estimates for linear finite element methods}, institution = {Math. Seminar, Christian-Albrechts-Universit\"{a}t Kiel, Germany}, year = {2003}, address = {Christian-Albrechts-Universit\"{a}t Kiel, D-24098 Kiel, Germany}, type = {Technical Report}, } @TECHREPORT{Zhang2003a, author = {.}, title = {Ultraconvergence of the Patch Recovery Technique II}, institution = {Department of Mathematics, Texas Tech University, Texas}, year = {2003}, address = {Texas Tech University, Lubbock, Texas 79409}, type = {Technical Report}, } @TECHREPORT{Bebendorf2003, author = {.}, title = {A Note on the Poincare Inequality for Convex Domains}, institution = {Max-Planck-Institut, Leipzig}, year = {2003}, month = mar, type = {Preprint}, number = {24}, } @TECHREPORT{Verfuerth2004, author = {.}, title = {Robust a posteriori error estimates for stationary convection--diffusion equations}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Ruhr-Universit\"{a}t Bochum, D-44780 Bochum, Germany}, year = {2004}, type = {Technical Report}, } @TECHREPORT{JohnKaya2003, author = {. and .}, title = {Finite Element error analysis and implementation of a variational multiscale method for the {Navier}--{Stokes} equations}, institution = {Faculty of Methematics, Otto-von-Guericke-University Magdeburg, Germany}, year = {2003}, type = {Technical Report}, } @TECHREPORT{GoedertSuttmeier2004, author = {. and Suttmeier}, title = {On computational glaciology: FE--simulation of ice sheet dynamics}, institution = {Universti\"{a}t Dortmund}, year = {2004}, address = {Vogelpothsweg 87, D-44221 Dortmund, Germany}, note = {to appear in: Computing and Visualization in Science}, type = {Preprint}, } @TECHREPORT{BeckerBraak2000, author = {. and .}, title = {Multigrid techniques for dinite elements on locally refined meshes}, institution = {Institute of Applied Mathematics}, year = {2000}, address = {University of Heidelberg, INF 294, D-69120 Heidelberg, Germany}, note = {to appear in: Numerical Linear Algebra with Applications}, type = {Preprint}, } @TECHREPORT{Hackbusch2003, author = {.}, title = {Multi--Grid Methods for {FEM} and BEM Applications}, institution = {Max-Planck-Institut, Leipzig}, year = {2003}, address = {Max-Planck-Institut, Leipzig, Germany}, type = {Preprint}, number = {72}, } @TECHREPORT{Babuska1996, author = {.}, title = {On Besov and Sobolev spaces of fractional order}, institution = {Texas Institute for Comp. and Appl. Mathematics}, year = {1996}, address = {University of Texas at Austin}, type = {Technical Report}, } @TECHREPORT{PalleDantzig1992, author = { .}, title = {An adaptive mesh refinement scheme for solidification problems}, institution = {Ford Research Laboratory}, year = {1992}, address = {Ford Motor Company, Dearborn, Michigan 47121, USA; Department of Mechanical and Industrial Engeneerung, University of Illinois, Urbana, Illinois 61801 USA}, note = {to appear in: Metallurgical and Material Transactions}, type = {Preprint}, } @TECHREPORT{HeLuo1997, author = {}, title = {Lattice Boltzmann Model for the Incompressible {Navier}--{Stokes} Equation}, institution = {Center for Nonlinear Studies, Los Alamos National Laboratory}, year = {1997}, month = feb, type = {Technical Report}, } @TECHREPORT{ZhangNaga2004, author = {. and .}, title = {Validation of the Quality of the A Posteriori Error Estimator Based on Polynomial Preserving Recovery for Linear Elements}, institution = {Department of Mathematics; Wayne State University}, year = {2004}, type = {Preprint}, } @TECHREPORT{LinZhang2004, author = {. and .}, title = {Natural Superconvergent Points of Triangular Finite Elements}, institution = {Department of Mathematics; Wayne State University}, year = {2004}, type = {Preprint}, } @TECHREPORT{SheuLinLin2003, author = {Sheu and Lin and Lin}, title = {Development of a three--level method for {Navier}--{Stokes} equations}, institution = {Department of Engineering Science and Ocean Engineering, National Taiwan University}, year = {2003}, address = {No. 1, Sec. 4, Roosevelt Road, Taipei, Taiwan, Republic of China}, month = nov, type = {Preprint}, } @TECHREPORT{SheuLinLin2003a, author = {Sheu and Lin and Lin}, title = {A computational efficient two--level method for the {Navier}--{Stokes} equations}, institution = {Department of Engineering Science and Ocean Engineering, National Taiwan University}, year = {2003}, address = {No. 1, Sec. 4, Roosevelt Road, Taipei, Taiwan, Republic of China}, month = nov, type = {Preprint}, } @TECHREPORT{KimmelSethian1997, author = {. and .}, title = {Computing Geodesic Paths on Manifolds}, institution = {Department of Mathematics and Lawrence Berkley Maboratory, University of Califormia}, year = {1997}, address = {Berkley, California, 94720}, month = jul, type = {Preprint}, } @TECHREPORT{ZhouShimada2000, author = {}, title = {An angle--based approach to two--dimensional mesh smoothing}, institution = {Canegie Mellon University, Pittsburgh, PA, USA}, year = {2000}, note = {\url{http:/ / citeseer.ist.psu.edu/ cis}}, type = {Preprint}, } @TECHREPORT{ParoliniQuarteroni2004, author = { .}, title = {Mathematical Models and Numerical Simulations for the America`s Cup}, institution = {CMCS, Institut d`Analyse et Calcul Scientifique, EPFL, Lausanne, Swizerland}, year = {2004}, month = mar, type = {Preprint}, } @TECHREPORT{MedicMohammadi1999, author = {. and .}, title = {NSIKE {}--{} an incompressible {Navier}--{Stokes} solver for unstructured meshes}, institution = {Institut national de recherche en informatique et en automatique}, year = {1999}, address = {Domaine de Voluceau, Rocquencourt, BP 105, 78153 LE CHENSAY Cedex, France}, month = mar, type = {Technical Report}, } @TECHREPORT{HermanssonHansbo2002, author = {Hermansson and .}, title = {A variable diffusion method for mesh smoothing}, institution = {Chalmers Finite Element Center, Chalmers University of Technology, G\"{o}teborg Sweden}, year = {2002}, type = {Preprint}, } @TECHREPORT{OlmosMunoz2002, author = { Munoz}, title = {A cellular automation model for the traffic flow in Bogota}, institution = {Dpto. de Fisica, Univ. Nacional de Columbia, Bogota D.C., Comumbia}, year = {2002}, type = {Preprint}, } @TECHREPORT{BlazyNazarovSpecovius-Neugebauer2004, author = { Nazarov and Specovius-Neugebauer}, title = {Artificial boundary conditions of pressure type for viscous flows in a system of pipes}, institution = {Paderborn Center for Parallel Computing, Universit\"{a}t Paderborn, F\"{u}rstenallee 11, 33102 Paderborn}, year = {2004}, month = may, type = {Preprint}, } @TECHREPORT{BrandtsKrizek2002, author = {. and .}, title = {Gradient Superconvergence of Uniform Simplificial Partitions of Polytopes}, institution = {Korteweg-de Vries Institure, Faculty of Science, University of Amsterdam, Plantage Muidergracht 24, 1018 TV Amsterdam, Netherlands}, year = {2002}, month = oct, type = {Preprint}, } @TECHREPORT{FairagAlmulla2004, author = {}, title = {Finite Element Technique for Solving the Stream Function Form of a Linearized {Navier}--{Stokes} Equations Using Argyris Element}, institution = {unknown}, year = {2004}, address = {arXiv:math.NA/0406070}, month = jun, type = {Technical Report}, } @TECHREPORT{Wang2000, author = {Wang}, title = {A superconvergence analysis for finite element solutions by the least--squares surface fitting on irregular meshes for smooth problems}, institution = {unknown}, year = {2000}, note = {NSF GRAND \#DMS-9706985}, type = {Technical Report}, } @TECHREPORT{EwingLiuWang2003, author = {. and }, title = {Superconvergence of mixed finite element approximations over quadrilaterals}, institution = {Institute for scientific computation, Texas A\&M university, College Station, TX 77843}, year = {2003}, type = {Technical Report}, } @TECHREPORT{Suttmeier2004h, author = {Suttmeier}, title = {Reliable approximation of weight factors entering residual--based error bounds for FE--discretizations}, institution = {Dpt. of mathematics, University of Dortmund, Germany}, year = {2004}, type = {Technical Report}, } @TECHREPORT{Cools2003, author = {Cools}, title = {An Encyclopaedia of Cubature Formulas}, institution = {Dept. of Computer Science, Katholieke Universiteit Leuven, Celestijnenlaan 200A, B-3001 Heverlee, Belgium}, year = {2003}, type = {Technical Report}, } @TECHREPORT{Ainsworth2000, author = {.}, title = {Essential Boundary Conditions and Multi--point Constraints in Finite Element Analysis}, institution = {Mathematics Department, Strathclyde University, Livingston Tower, 26 Richmond Street, Glasgow G1 1XH, Scotland}, year = {2000}, type = {Technical Report}, } @TECHREPORT{AinsworthKelly2001, author = { Kelly}, title = {A Posteriori Error Estimators and Adaptivity for Finite Element Approximation of the Non--homogeneous Dirichlet Problem}, institution = {Mathematics Department, Strathcycle University, 26 Richmond Street, Glasgow G1 1XH, Scotland}, year = {2001}, month = feb, type = {Technical Report}, } @TECHREPORT{Bebendorf2004, author = {.}, title = {Approximate Inverse Preconditioning of FE Systems for Elliptic Operators with non--smooth Coefficients}, institution = {Fakult\"{a}t f\"{u}r Mathematik und Informatik, Universit\"{a}t Leipzig, Augustaplatz 10/11, D-04109 Leipzig, Germany}, year = {2004}, month = feb, type = {Preprint}, number = {7}, } @TECHREPORT{AlexandrovSantosa2004, author = { Santosa}, title = {A Topology--Preserving Level--Set Method for Shape Optimization}, institution = {University of Minnesota, School of Mathematics}, year = {2004}, month = may, note = {arXiv:math.OC/0405142}, type = {Technical Report}, } @TECHREPORT{DietzHeineHeuvelineRichter2004, author = { }, title = {Test of a new numerical approach to the quantization of billiards}, institution = {Institut f\"{u}r Kernphysic, Technische Universit\"{a}t Darmstadt, D-64289 Darmstadt, Germany}, year = {2004}, month = mar, type = {Technical Report}, } @TECHREPORT{Davis2003, author = {.}, title = {A column pre--ordering strategy for the unsymmetric--pattern multifrontal method}, institution = {-}, year = {2003}, note = {Submitted to ACM Trans. Math. Software}, type = {Technical Report}, number = {TR-03-006}, } @TECHREPORT{Davis2004, author = {.}, title = {UMFPACK}, institution = {University of Florida}, year = {2004}, address = {\url{http:/ / www.cise.ufl.edu/ research/ sparse/ umfpack/ }}, type = {Mathematic Library}, } @TECHREPORT{HysomPothen2000, author = { .}, title = {Efficient Parallel Computation of ILU(k) Preconditioners}, institution = {NASA Langley Research Center Hampton}, year = {2000}, address = {Institute for Computer Applications in Science and Engineering Mail Stop 132C}, month = may, note = {\url{http:/ / www.icase.edu/ library/ reports/ rdp/ 2000/ 2000-23RDP.tex.refer.html}}, type = {Preprint}, number = {VA 23681-2199}, } @TECHREPORT{LinTobiskaZhou2001, author = {. and .}, title = {On the superconvergence of nonconforming elements applied to the Poisson problem}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2001}, month = aug, type = {Preprint}, number = {17}, } @TECHREPORT{MuellerSchirmTeschnerHeidelbergerGross2004, author = { .}, title = {Interaction of Fluids with Deformable Solids}, institution = {ETH Z\"{u}rich, Swizerland}, year = {2004}, type = {Technical Report}, } @TECHREPORT{WolffeOleszkiewiczCherbaQi2002, author = {. and .}, title = {Parallelizating Solid Particles in Lattice--Boltzmann Fluid Dynamics}, institution = {Dept. of Computer Science, Grand Valley State University}, year = {2002}, address = {Allendale, MI 49401}, note = {\url{http:/ / www.csis.gvsu.edu/ den/ papers/ Pdpta-02.pdf}}, } @TECHREPORT{ToelkeKrafczykRank2002, author = {.}, title = {A Multigrid--Solver for the Discrete Boltzmann--Equation}, institution = {Lst. Bauinformatik, Techn. Univ. M\"{u}nchen, Germany}, year = {2002}, note = {erschienen in: J. Stat. Phys., Vol. 107, Nos.1/2, pp. 573-591}, type = {Preprint}, } @TECHREPORT{PengXiDuncan1999, author = {. and . and . and .}, title = {Lattice Boltzmann method of irregular meshes}, institution = {Dpt. of Physics and Astronomy, Bowling Green State University}, year = {1999}, address = {Bowling Green OH 43403}, note = {erschienen in: Phys. Rev. Lett. 82, 5245-5248}, type = {Preprint}, } @TECHREPORT{LockardLuoSinger2000, author = {. and . and .}, title = {Evaluation of the Lattice--Boltzmann equation solver powerflow for aerodynamic applications}, institution = {NASA Langley Research Center, Hampton, Virginia}, year = {2000}, type = {Rechnical Report}, number = {NASA/CR-2000-210550}, } @TECHREPORT{Kehrwald2004, author = {.}, title = {Parallel lattice Boltzmann simulation of complex flows}, institution = {Fraunhofer Institut Techno- und Wirtschaftsmathematik}, year = {2004}, type = {Technical Report}, number = {Nr. 61}, } @TECHREPORT{Junk2001, author = {.}, title = {A Finite Difference Interpretation of the Lattice Boltzmann Method}, institution = {FB Mathematik, Universit\"{a}t Kaiserslautern}, year = {2001}, address = {67663 Kaiserslautern, Germany}, note = {erschienen in: Numer. Methods Partial Differ. Equations, Vol. 17, 383-402, 2001}, type = {Preprint}, } @TECHREPORT{Bourisli2003, author = {.}, title = {Cellular Automata Methods in Fluid Flow}, institution = {Decision Sciences and Engeneering Systems COmputational Intelligence}, year = {2003}, type = {Technical Report}, } @TECHREPORT{DazevedoForsythTang1992, author = {. and . and .}, title = {Towards a cost--effective ILU preconditioner with higher level fills}, institution = {Mathematical Science Section, Oak Ridge National Laboratory}, year = {1992}, address = {Oak Ridge, Tennessee 37831}, note = {BIT Volume 32 , Issue 3, pp. 442-463}, type = {Preprint}, } @TECHREPORT{Fefferman2000, author = {.}, title = {Existence \& Smoothness of the {Navier}--{Stokes} equation}, institution = {Princeton University, Dpt. of Mathematics}, year = {2000}, address = {Princeton, NJ 08544-1000}, month = may, type = {Technical Report}, } @TECHREPORT{HeuvelineSchieweck2004, author = { .}, title = {An interpolation operator for \$H\^{} 1\$ functions on general quadrilateral and hexahedral meshes with hanging nodes}, institution = {Institut f\"{u}r Angewandte Mathematik, Universit\"{a}t Heidelberg}, year = {2004}, address = {INF 293, D-68120 Herdelberg, Germany}, note = {\url{http:/ / www..gaia.iwr.uni-heidelberg.de}}, type = {Preprint}, } @TECHREPORT{HeimsundTaiWang2002, author = { . and .}, title = {Superconvergence for the gradient of finite element approximations by \$L\^{} 2\$--Projections}, institution = {-}, year = {2002}, note = {erschienen in: SIAM J. Numer. Anal., Vol. 40, No. 4, pp 1263-1280}, type = {Preprint}, } @TECHREPORT{WangYe2002, author = {. and .}, title = {Superconvergence of finite element approximations for the stokes problem by least squares surface fitting}, institution = {-}, year = {2002}, note = {erschienen in: Appl. Num. Math., Vol. 41, No. 4, pp 515-527}, type = {Preprint}, } @TECHREPORT{MaggiVillani2004, author = {. and .}, title = {Balls Have the Worst Best Sobolev Inequalities}, institution = {Max-Planck-Institut f\"{u}r Mathematik in den Naturwissenschaften, Leipzig}, year = {2004}, type = {Preprint}, number = {32}, } @TECHREPORT{ZhangZhu1995b, author = {. and .}, title = {Superconvergence of The Derivative Patch Recovery Technique and A Posteriori Error Estimation}, institution = {-}, year = {1995}, type = {Preprint}, } @TECHREPORT{BrandtsChen2004, author = {. and .}, title = {Superconvergence of Least--Squares Mixed Finite Elements}, institution = {-}, year = {2004}, type = {Preprint}, } @TECHREPORT{BarangerNajib1989, author = {. and .}, title = {Analyse Numerique Des Ecoulements Quasi--Newtoniens Dont La Viscosite Obeit A La Loi Puissance Ou La Loi De Carreau}, institution = {LAN - Bat. 101 - Universite Lyon 1}, year = {1989}, address = {69622 VILLEURBANNE Cedex}, type = {Technical Report}, number = {82}, } @TECHREPORT{BarnesHopkins1996, author = {. and .}, title = {A lightweight Fortran 90 interface to the basic linear algebra subroutines}, institution = {Computing Laboratory, University Canterbury}, year = {1996}, address = {Kent. CT2 7NF, United Kingdom}, type = {Preprint}, } @TECHREPORT{HysomPothen2002, author = {. and .}, title = {Level--based Incomplete LU Factorization: Graph Model and Algorithms}, institution = {U.S. Department of Energy}, year = {2002}, month = nov, note = {\url{http:/ / www.cs.odu.edu/ ~pothen/ papers.html;to} appear in: SIAM Journal On Matrix Analysis and Applications}, type = {Preprint}, number = {UCRL-JC-150789}, } @TECHREPORT{FleitasJiangLiao2004, author = {. and . and .}, title = {Adaptive Grid Generation Based on the Least--Squares Finite Element Method}, institution = {Dpt. of Math., University of Texas}, year = {2004}, address = {Arlington, Texas 76019}, type = {Technical Report}, } @TECHREPORT{BraessDahmen2002, author = {. and .}, title = {The Mortar Element Method Revisited {}--{} What are the Right Norms?}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Ruhr-Universit\"{a}t Bochum}, year = {2002}, address = {44780 Bochum, Germany}, note = {Erschienen in: Proceedings of the 13th International Conference on Domain Decomposition Methods in Lyon, France}, type = {Preprint}, } @TECHREPORT{LaytonSchieweckYotov2003a, author = {. and . and .}, title = {Coupling fluid flow with porous media flow}, institution = {Dpt. of Math., University of Pittsburgh}, year = {2003}, address = {Pittsburgh, PA 15260, USA}, note = {Erschienen in: SIAM J. Numer. Anal., Vol. 40, No. 6, pp. 2195-2218}, type = {Preprint}, } @TECHREPORT{Schieweck2001, author = {.}, title = {Upper and Lower A Posteriori Error Estimates with Post--Processing for Nonconforming Finite Elements}, institution = {Institut f\"{u}r Analysis und Numerik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2001}, address = {Postfach 4120, D-39016 Magdeburg, Germany}, type = {Preprint}, } @TECHREPORT{Schieweck2002, author = {.}, title = {A Multigrid {Stokes} Solver Using a Discretely Divergence Free Basis}, institution = {Institut f\"{u}r Analysis und Numerik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2002}, address = {Postfach 4120, D-39016 Magdeburg, Germany}, month = mar, type = {Preprint}, } @TECHREPORT{BischofBueckerLangRaschSlusanschi2002, author = {. and B\"{. and . and . and .}, title = {Efficient and accurate derivatives for a software process chain in airfoil shape optimization}, institution = {Institute for Scientific Computing, Aachen University of Technology}, year = {2002}, address = {D-52056 Aachen, Germany}, type = {Preprint}, } @TECHREPORT{BueckerRasch2002, author = {. and .}, title = {Amdahl`s Law in Automatic Differentiation: a Performance Model for Interface Contraction}, institution = {Institute for Scientific Computing RWTH-CS-SC-02-07, Aachen University of Technology}, year = {2002}, type = {Preprint}, } @TECHREPORT{Olshanskii2002, author = {.}, title = {Solving steady incompressible {Navier}--{Stokes} problem with coupled {Galerkin} finite element method}, institution = {Dpt. of Mechanics and Mathematics, Moscow State university}, year = {2002}, address = {Moscow 119899, Russia}, type = {Preprint}, } @TECHREPORT{GieringKaminskiSlawig2002, author = {. and .}, title = {Generating efficient derivative code with TAF: Adjoint and tangent linear Euler flow around an airfoil}, institution = {FastOpt}, year = {2002}, address = {Martinistr. 21, 20251 Hamburg, Germany}, month = oct, type = {Preprint}, } @TECHREPORT{BischofBueckerLangRasch2001, author = {. and B\"{. and . and .}, title = {Automated Gradient Calculation}, institution = {Institute f\"{u}r Scientific Computing, Aachen University of Technology}, year = {2001}, address = {56056 Aachen, Germany}, note = {eingereicht bei: Notes on Numerical Fluid Mechanics}, type = {Preprint}, } @TECHREPORT{BoehrnsenAntes2001, author = {. and .}, title = {Granular flow: Dynamic Effects in Silos}, institution = {Institute of Applien Mathematics, Technical University of Braunschweig}, year = {2001}, address = {Spielmannstra"se 11, D-38106 Braunschweig, Germany}, note = {\url{http:/ / www.infam.tu-bs.de/ infam2/ include/ Literatur/ veroeff/ Boehrnsen01-Granu_flow_Dynam_Effec_Silos.pdf}}, type = {Technical Report}, } @TECHREPORT{BambergerBaenschSiebert2001, author = {. and .}, title = {Experimental and Numerical Investigation of Edge Tones}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Stochastik}, year = {2001}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {681}, } @TECHREPORT{SvendsenReese2002, author = { .}, title = {On the modeling of internal variables as structure tensors in anisotropic, finite--deformation inelasticy}, institution = {Dpt. of Mechanical Engineering, University of Dortmund}, year = {2002}, address = {D-44227 Dortmund, Germany}, note = {Submitted to: Int. J. Plasticity}, type = {Preprint}, } @TECHREPORT{BaenschHausserLakkisLiVoigt2003, author = { .}, title = {Finite Element Method for Epitaxial Growth with Attachment--Detachment Kinetics}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2003}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {848}, } @TECHREPORT{BaenschBergOhlhoff2002, author = { .}, title = {Uniaxial extension flows in a liquid bridges}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2002}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {761}, } @TECHREPORT{BaenschDavisLangmach2003, author = {. and .}, title = {Two--{} and three--dimensional transient melt--flow simulation in a vapour--pressure--controlled Czochralski crystal growth}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2003}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {852}, } @TECHREPORT{AzeradBaensch2001, author = {. and B\"{a}.}, title = {Quasi--stability of the primary flow in a cone and plate viscometer}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2001}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {708}, } @TECHREPORT{BaenschMorinNochetto2002, author = {. and . and .}, title = {Surface Diffusion of Graphs: Variational Formulation, Error Analysis and Simulation}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2002}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {797}, } @TECHREPORT{BaenschMorinNochetto2001, author = {. and . and .}, title = {An Adaptive Uzawa {FEM} for Stokes: Convergence without the Inf--Sup}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2001}, address = {Mohrenstra"se 39, D-10117 Berlin, Germany}, type = {Preprint}, number = {662}, } @TECHREPORT{HauserMattoxLebauDietzHuang2003, author = { .}, title = {Code optimizations for complex microprocessors applied to {CFD} software}, institution = {Dpt. of Mechanical and Aerospace Engineering, Utah State University}, year = {2003}, address = {Logan, UT, 84322}, note = {erschienen in: SIAM Journal on Scientific Computing, Vol. 25, No. 4, pp pp. 1461-1477, 2004}, type = {Preprint}, } @TECHREPORT{LangtangenMardalWinther2002, author = {. and . and .}, title = {Numerical Methods for Incompressible Viscous Flow}, institution = {Dept. of Scientific Computing, Simula Research Laboratory, University of Oslo}, year = {2002}, note = {erschienen in: Advances in Water Resources, vol 25, pp. 1125-1146, Elsevier, 2002}, type = {Technical Report}, } @TECHREPORT{TiwariManservisi2000, author = {. and .}, title = {Modeling incompressible {Navier}--{Stokes} flows by LSQ--SPH}, institution = {Institut f\"{u}r Techno- und Wirtschaftsmathematilk}, year = {2000}, address = {Erwin-Schr\"{o}dinger-Strasse, Geb\"{a}ude 49, D-67663 Kaiserslautern, Germany}, type = {Preprint}, } @TECHREPORT{PozoHerouxRemington1997, author = { . and .}, title = {Sparse BLAS Library: Lite and Toolkit Level Specifications}, institution = {BLAS Technical Forum}, year = {1997}, type = {Manual}, } @TECHREPORT{BaenschMorinNochetto2003, author = { .}, title = {Finite Element Methods for Surface Diffusion}, institution = {Weierstra"s-Institut f\"{u}r Angewandte Analysis und Statistik}, year = {2003}, address = {Mohrenstra"se 39, D10117 Berlin}, type = {Preprint}, number = {805}, } @TECHREPORT{BraackRichter2004, author = {. and .}, title = {Solutions of 3D {Navier}--{Stokes} benchmark problems with adaptive finite elements}, institution = {Institute of Applied Mathematics, University of Heidelberg}, year = {2004}, address = {INF 294, 69120 Heidelberg, Germany}, month = apr, note = {Submitted to Elsevier Science}, type = {Preprint}, } @TECHREPORT{LuBeckermannKarma2002, author = {. and . and .}, title = {Convection Effects in Three--Dimensional Dendritic Growth}, institution = {Dpt. of Mechanical and Industrial Engineering, University of Iowa}, year = {2002}, address = {Iowa City, IA 52242-1527, USA}, note = {erschienen in: Mat. Res. Soc. Symp. Proc., Vol. 701, 2002, T2.2.1-T2.2.10}, type = {Preprint}, } @TECHREPORT{GajewskiGaertner2003, author = { .}, title = {On a nonlocal model of image segmentation}, institution = {Weierstrass Institute for Applied Analysis and Stochastics}, year = {2003}, address = {Berlin, Germany}, note = {\url{http:/ / www.wias-berlin.de/ publications/ preprints/ 762/ }}, type = {Technical Report}, number = {762}, } @TECHREPORT{BankSmith1997, author = {. and .}, title = {Mesh smoothing using a posteriori error estimates}, institution = {Dpt. of Mathematics, University of California at San Diego}, year = {1997}, address = {La Jolla, CA 92093}, note = {published in: SIAM J. Numerical Analysis 34(3):979-997, Jun 1997}, type = {Preprint}, } @TECHREPORT{AkrivisKarakatsani2002, author = {. and .}, title = {Modified implicit--explicit BDF methods for nonlinear parabolic equation}, institution = {Computer Science Department, University of Ioannina}, year = {2002}, address = {Ioannina GR-451 10, Greece}, month = dec, note = {published in: Bit Numerical Mathematics 43 (3): 467-483, September 2003 }, type = {Preprint}, } @TECHREPORT{VoigtMetzger2000, author = {. and .}, title = {Numerical Simulation and Control of Industrial Crystal Growth by the Czochralski and Vertical Gradient Freece Method}, institution = {research center caesar, nanotechnology - crystal growth}, year = {2000}, address = {Friedensplatz 16, 53111 Bonn, Germany}, note = {\url{http:/ / 192.168.3.11:81/ preprints/ db/ preprint/ pp_entry.php?myaction=ShowList,} available via http://www.caesar.de}, type = {Preprint}, number = {2}, } @TECHREPORT{HorbeltTimmerHenning2002, author = {. and . and .}, title = {Parameter Estimation in a Nonlinear Delayed Feedback Systems from Noisy Data}, institution = {Zentrum f\"{u}r Technimathematik, Fachbereich 3 - Mathematik und Informtik, Universit\"{a}t Bremen}, year = {2002}, address = {28334 Bremen}, month = may, type = {Preprint}, number = {1}, } @TECHREPORT{MarsikPrevorovskaStembera2003, author = {. and . and .}, title = {The Influence of Compliance and Resistance of Arteries on Korotkoff Sound Generation in Numerical Modelling}, institution = {Institute of \&Thermomechanics AS CR}, year = {2003}, address = {Dolejskova 5, 182 00 Praha 8, Czech Republic}, note = {published in: Cardiovascular Engineering: An International Journal, June 2004, vol. 4, no. 2, pp. 193-199(7) }, type = {Preprint}, } @TECHREPORT{AlbukrekUrbanDahmenRempferLumley2000, author = {. and .}, title = {Divergence--Free Wavelet Analysis of Turbulent Flows}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, Rheinisch-Westf\"{a}lische Technische Hochschule Aachen}, year = {2000}, address = {Germany}, month = nov, type = {Technical Report}, number = {198}, } @TECHREPORT{HansboLarson2001, author = {. and .}, title = {A Simple Nonconforming Bilinear Element for the Elasticity Problem}, institution = {Chalmers Finite Element Center}, year = {2001}, address = {Chalmers University of Technology, G\"{o}teborg Sweden}, type = {Preprint}, number = {2001-01}, } @TECHREPORT{DahmenKlintUrban2002, author = {. and . and .}, title = {On Frictitious Domain Formulations for Maxwell`s Equations}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, Rheinisch-Westf\"{a}lische Technische Hochschule Aachen}, year = {2002}, address = {Germany}, month = apr, type = {Technical Report}, number = {214}, } @TECHREPORT{BarinkaBarschUrbanVorloeper2001, author = {. and .}, title = {The Multilevel Library Software Tools for Multiscale Methods and Wavelets {}--{} Version 2.1, Documentation}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, Rheinisch-Westf\"{a}lische Technische Hochschule Aachen}, year = {2001}, address = {Germany}, month = jul, type = {Technical Report}, number = {205}, } @TECHREPORT{DahmenUrbanVorloeper2002, author = {. and . and .}, title = {Adaptive Wavelet Methods {}--{} Basic Concepts and Applications to the {Stokes} Problem}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, Rheinisch-Westf\"{a}lische Technische Hochschule Aachen}, year = {2002}, address = {Germany}, type = {Technical Report}, number = {215}, } @TECHREPORT{DahlkeDahmenUrban2001, author = {. and . and .}, title = {Adaptive Wavelet Methods for Saddle Point Problems {}--{} Optimal Convergence Rates}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, Rheinisch-Westf\"{a}lische Technische Hochschule Aachen}, year = {2001}, address = {Germany}, type = {Technical Report}, number = {204}, } @TECHREPORT{Hanke2004, author = {.}, title = {Benchmarking FEMLAB 3.0a: Laminar Flows in 2D}, institution = {Royal institute of Technology, Dpt. of Numerical Analysis and Computer Science}, year = {2004}, month = apr, type = {Technical Report}, number = {2004:01}, } @TECHREPORT{DieningProhlRuzicka2002, author = {. and . and .}, title = {On Time--Discretizations for Generalized Newtonian Fluids}, institution = {Seminar f\"{u}r Angewandte Mathematik, Eidgen\"{o}ssische Technische Hochschule Z\"{u}rich}, year = {2002}, address = {CH-8092 Z\"{u}rich, Swizerland}, note = {to appear in: Nonlinear Problems in Mathematical Physics and Relate Topics II, Kleuwer/Plenum Publishers, 2002}, type = {Technical Report}, number = {2003-03}, } @TECHREPORT{FengProhl2002, author = {. and .}, title = {Analysis of Total Variation Flow and Its Finite Element Approximations}, institution = {Department of Mathematics, The University of Tennessee}, year = {2002}, address = {Knoxville, 37996 Tennessee, U.S.A. }, note = {published in: M2AN, Vol. 37, No. 3, pp. 533-556, DOI 10.1051/m2an:2003041 }, type = {Preprint}, } @TECHREPORT{Blank2001, author = {.}, title = {Preconditioning via a Schur Complement Method {}--{} an Application in State Estimation}, institution = {Institut f\"{u}r Geometrie und Praktische Mathematik, RWTH Aachen}, year = {2001}, address = {Templergraben 55, D-52056 Aachen, Germany}, month = nov, type = {Technical Report}, number = {209}, } @TECHREPORT{Emmerich1999, author = {.}, title = {Front capturing versus front tracking for the Thin Film Growth}, institution = {Max Planck Institute for the Physics of Complex Systems}, year = {1999}, address = {N\"{o}thnitzer Str. 38, D-01187 Dresden}, type = {Technical Report}, } @TECHREPORT{Beckermann2000, author = {.}, title = {Modeling of macrosegregation: past, present and future}, institution = {Dpt. of Mechanical Engineering, University of Iowa}, year = {2000}, address = {Iowa City, IA 52242}, note = {Presented at the Flemings Symposium, Boston, MA, June 2000}, type = {Technical Report}, } @TECHREPORT{KimProvatasGoldenfeldDantzig1999, author = {. and . and . and .}, title = {Universal Dynamics of a Phase--Field Model for Dendritic Growth}, institution = {Dpt. of Physics, University of Illinois at Urbana-Champaign}, year = {1999}, address = {110 West Green Street, Urbana, IL, 61801}, month = mar, type = {Preprint}, } @TECHREPORT{Nestler2003, author = {.}, title = {Phasenfeldmodellierung}, institution = {Gie"serei-Institut, RWTH Aachen}, year = {2003}, type = {Technical Report}, } @TECHREPORT{ProvatasGoldenfeldDantzig1998, author = { .}, title = {Adaptive Mesh Refinement Computation of Solidification Microstructures using Dynamic Data Structures}, institution = {Dpt. of Physics, University of Illinois at Urbana-Champaign}, year = {1998}, address = {1110 West-Green Street, Urbana, IL, 61801}, month = dec, type = {Technical Report}, } @TECHREPORT{HardinLiuBeckernann2000, author = {. and . and .}, title = {Development of a model for transient simulation and control of a continuous steel slab caster}, institution = {Solidification Laboratory of Mech. Engineering, University of Iowa}, year = {2000}, address = {Iowa City, Iowa 52242}, type = {Preprint}, } @TECHREPORT{ProvatasGoldenfeldDantzig1998a, author = {. and .}, title = {Efficient Computation of Dendritic Microstructures using Adaptive Mesh Refinement}, institution = {Dpt. of Physics, University of Illinois at Urbana-Champaign}, year = {1998}, address = {1110 West-Green Street, Urbana, IL, 61801}, month = mar, type = {Technical Report}, } @TECHREPORT{GarckeNestler2000, author = {. and .}, title = {Multi--Phase--Field Model for the Motion of Multiple Interfaces}, institution = {Institut f\"{u}r Angewandte Mathematik}, year = {2000}, address = {Wegelstr. 6, D-53115 Bonn}, type = {Preprint}, } @TECHREPORT{ArnoldBoffiFalk2000, author = {. and . and .}, title = {Approximation by quadrilateral finite elements}, institution = {Institute for Mathematics and its Applications, University of Minnesota}, year = {2000}, address = {426 Lind Hall, 207 Church St. SE, Minneapolis, MN 55455}, type = {Preprint}, } @TECHREPORT{Gupta2001, author = {.}, title = {Recent Advances in Direct Methods for Solving Unsymmetric Sparse Systems on Linear Equations}, institution = {IBM Research Division, T. J. Watson Research Center}, year = {2001}, address = {P. O. Box 218, Yorktown Heights, NY 10598}, month = apr, type = {Technical Report}, } @TECHREPORT{GuptaJoshi2001, author = {. and .}, title = {WSMP: A High--Performance Sharted--{} and Distributed--Memory Parallel Sparse Linear Equation Solver}, institution = {IMB Research Division, T. J. Watson Research Center}, year = {2001}, address = {P. O. Box 218, Yorktown Heights, NY 10598}, month = apr, type = {Technical Report}, } @TECHREPORT{Callies1997, author = {.}, title = {Verbesserte Randapproximation zur Str\"{o}mungssimulation mit dem Praktikumscode}, institution = {Institut f\"{u}r Informatik, Technische Universit\"{a}t M\"{u}nchen}, year = {1997}, month = feb, type = {Technical Report}, } @TECHREPORT{Owen2000, author = {.}, title = {A Survey of Unstructured Mesh Generation Technology}, institution = {Dpt. of Civil and Environmental Engineering, Carnie Mellon University, Pittsburgh}, year = {2000}, note = {\url{http:/ / www.andrew.cmu.edu/ user/ sowen/ survey/ }}, type = {Technical Report}, } @TECHREPORT{SeppaenenVauhkonenVauhkonenSomersaloKaipio2001, author = {. and . and . and .}, title = {State estimation in three dimensional impedance imaging {}--{} Use of fluid dynamical evolution models}, institution = {Dpt. of Applied Physics, University of Kuopio}, year = {2001}, address = {P. O. Box 1627, FIN-70211 Kuopio, Finland}, month = apr, note = {2nd world congress on industrial process tomography, Hannover, Germany, August 29-31,2001}, type = {Preprint}, } @TECHREPORT{SeppaenenVauhkonenVauhkonenSomersaloKaipio2001a, author = {. and . and . and . and .}, title = {Fluid dynamical models and state estimation in process tomography: Effect due to inaccuracies in flow fields}, institution = {Dpt. of Applied Physics, University of Kuopio}, year = {2001}, address = {P. O. Box 1627, FIN-70211 Kuopio, Finland}, month = mar, type = {Preprint}, } @TECHREPORT{PierceGiles2003, author = {. and .}, title = {Adjoint and defect error bounding and correction for functional estimates}, institution = {Applied \& Conputational Mathematics, California Institute of Technology}, year = {2003}, type = {Preprint}, } @TECHREPORT{WeickertSchnoerr2000, author = {. and Schn\"{o}.}, title = {{PDE}--based Preprocessing of Medical Images}, institution = {Computer Vision, Graphics, and Pattern Recognition Group, Dpt. of Math. and Computer Science, University of Mannheim}, year = {2000}, address = {D-68131 Mannheim, Germany}, month = feb, type = {Technical Report}, number = {8/2000}, } @TECHREPORT{FengProhl2001, author = {. and .}, title = {Analysis of a fully discrete finite element method for the phase field model and approximation of its sharp interface limits}, institution = {Department of Mathematics, ETH Z\"{u}rich}, year = {2001}, address = {CH-8092 Z\"{u}rich, Switzerland}, month = nov, note = {published in: Math. Comp. 73 (2004), 541-567}, type = {Preprint}, } @TECHREPORT{Hinze2003, author = {.}, title = {A generalized discretization concept for optimal control problems with control constrains}, institution = {Technische Universit\"{a}t Dresden}, year = {2003}, month = apr, type = {Technical Report}, number = {MATH-NM-02-2003}, } @TECHREPORT{Strain1998, author = {.}, title = {Fast Tree--based Redistancing for Level Set Computations}, institution = {Dpt. of Mathematics, University of California}, year = {1998}, address = {970 Evans Hall \#3840, Berekley, California 94720-3840}, note = {published in: J. Comp. Phys., 152 (1999), 664-686}, type = {Preprint}, } @TECHREPORT{Necasova2001, author = {.}, title = {Some remarks on the steady fall of body in {Stokes} and Oseen flow}, institution = {Mathematical Institute, Academy of science of the Czech republic}, year = {2001}, type = {Preprint}, number = {143}, } @TECHREPORT{CaboussatMaronnierPicassoRappaz2002, author = {. and . and . and .}, title = {Numerical Simulation of Three Dimensional Free Surface Flows with Bubbles}, institution = {Institut de Mathematiques, Ecole Polytechnique Federale de Lausanne}, year = {2002}, address = {1015 Lausanne, Swizerland}, note = {published in: Challenges in Scientific Computing--Cisc 2002 : Proceedings of the Conference Challenges in Scientific Computing, Berlin, October 2-5, 2002}, type = {Preprint}, } @TECHREPORT{HaarioKuzmin2000, author = {. and .}, title = {A numerical study of mass transfer and chemical reaction phenomena in bubble columns}, institution = {Dpt. of Mathematics, University of Helsinki}, year = {2000}, address = {P. O. Box 4, FIN-00014, Helsinki, Finland}, type = {Technical Report}, } @TECHREPORT{Lukacova-Medvidova1996, author = {.}, title = {On the Error Estimate of a Combined Finite Element {}--{} Finite Volume Method}, institution = {Institut f\"{u}r Analysis und Numerik, Fakult\"{a}t f\"{u}r Mathematik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {1996}, address = {Postfach 4120, 39016 Magdeburg, Germany}, type = {Preprint}, number = {9}, } @TECHREPORT{Necasova2002, author = {.}, title = {Asymptotic properties of then steady fall of a body in viscous fluids}, institution = {Mathematical Institute, Academy of science of the Czech republic}, year = {2002}, type = {Preprint}, number = {149}, } @TECHREPORT{Thorne2003, author = {.}, title = {Cache aware Multigrid on AMR hierarchies}, institution = {Dpt. of Computer Science, University of Kentucky}, year = {2003}, address = {325 MvVel Hall, Lexington, KY 40506-0045, USA}, note = {published in: Proceedings of the Copper Mountain Conference on Multigrid Methods, MGNet Virtual Proceedings Series, Cos Cob, 2003}, type = {Preprint}, } @TECHREPORT{Bergen2003, author = {.}, title = {Hierarchical Hybrid Grids: A Framework for Efficient Multigrid on High Performance Architectures}, institution = {Lehrstuhl f\"{u}r Informatik, Institut f\"{u}r Informatik, Friedrich-Alexander-Universit\"{a}t Erlangen-N\"{u}rnberg}, year = {2003}, note = {Stdent Paper; \url{http:/ / www10.informatik.uni-erlangen.de/ Publications/ TechnicalReports/ TechRep03-5.pdf}}, type = {Technical Report}, } @TECHREPORT{John2001, author = {.}, title = {Higher order finite element methods and multigrid solvers in a benchmark problem for the 3D {Navier}--{Stokes} equations}, institution = {Institute of Analysis and Numerical Mathematics, Department of Mathematics, Otto-von-Guericke-University Magdeburg}, year = {2001}, address = {PF 4120, D-39106 Magdeburg, Germany}, note = {published in: Int. J. Num. Meth. Fluids 40, 775 - 798, 2002; \url{http:/ / www-ian.math.uni-magdeburg.de/ ~home/ john/ ELECTRONIC_PAPERS/ Joh02.IJNMF.pdf}}, type = {Preprint}, } @TECHREPORT{OlhanskiiReusken2003, author = {. and .}, title = {Grad--div stabilization for {Stokes} equations}, institution = {Dpt. Mechanics and Mathematics, Moscow State University}, year = {2003}, address = {Moscow 11989, Russia}, note = {published in: Math. Comp. 73 (2004), 1699-1718}, type = {Preprint}, } @TECHREPORT{JangJeongKimSheenParkKim2003, author = {. and . and . and . and . and .}, title = {Checkerboard--Free Topology Optimization Using Nonconforming Finite Elements}, institution = {School of Mechanical and Aerospace Engineering}, year = {2003}, address = {Seoul 151-742, Korea}, type = {Preprint}, } @TECHREPORT{Layton2001, author = {.}, title = {SOME REPORTS: Numerical Analysis of Large Eddy Simulation}, institution = {Dpt. of Mathematics, University of Pittsburgh}, year = {2001}, address = {Pittsburgh, PA 15260 , U.S.A.}, note = {\url{http:/ / www.math.pitt.edu/ ~wjl/ reports.html}}, type = {Report}, } @TECHREPORT{John2000, author = {.}, title = {On smoothers in parallel coupled multigrid methods for incompressible {Navier}--{Stokes} equations}, institution = {Institut f\"{u}r Analysis und Numerik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2000}, address = {Postfach 4120, 39016 Magdeburg}, note = {published in: Scientific Computing and Applications, Nova Science Publishers Inc. Huntington, 75 - 82, 2001 }, type = {Preprint}, } @TECHREPORT{BeckerBraackRannacher1998, author = { .}, title = {Adaptive Finite Element Methods for Flow Problems}, institution = {Institut f\"{u}r Angewandte Mathematik, Universit\"{a}t Heidelberg}, year = {1998}, address = {Im Neuheimer Feld 294, D-69120 Heidelberg, Germany}, note = {published in: Proc. Foundations of Computational Mathematics 99, (R. DeVore, A. Iserles \& E. Suli, eds), Cambridge University Press, Cambridge, 2001}, type = {Preprint}, } @TECHREPORT{Lohmann2000, author = {.}, title = {Realisierung direkter und iterativer L\"{o}ser f\"{u}r Systeme zweidimensionaler Integralgleichungen}, institution = {Universit\"{a}t Dortmund, Informatik IV, Rechnersysteme und Leistungsbewertung}, year = {2000}, month = mar, type = {Pr\"{a}sentation, Technical Report}, } @TECHREPORT{Brenner1999, author = {.}, title = {Convergence of the Multigrid {V}--Cycle Algorithm for Second Order Boundary Value Problems Without Full Elliptic Reguilatity}, institution = {Dpt. of Mathematics, University of South Carolina}, year = {1999}, note = {published in: Math. Comp. 71, No. 238, 507-525}, type = {Preprint}, number = {1999:07}, } @TECHREPORT{Mavriplis2000, author = {.}, title = {Parallel Performance Investigations of an Unstructured Mesh {Navier}--{Stokes} Solver}, institution = {Institute for Computer Applications in Science and Engineering, NASA Langley Research Center}, year = {2000}, address = {Hampton, Virginia 23681-2199}, month = mar, note = {NASA/CR-2000-210088}, type = {Technical Report}, number = {2000-13}, } @TECHREPORT{Brenner2001b, author = {.}, title = {Convergence of nonconforming {V}--Cycle and {F}--Cycle multigrid algorithms for second order elliptic boundary value problems}, institution = {Industrial Mathematics Institute, Dpt. of Mathematics, University of South Carolina}, year = {2001}, note = {published in: Math. Comp.73, No. 247, pp. 1041-1066}, type = {Preprint}, number = {2001:14}, } @TECHREPORT{ProhlRuzicka2000, author = {. and .}, title = {On fully implicit space--time discretization for motions of incompressible fluids with shear dependent viscosities: The case \$p/le 2\$.}, institution = {Math. Seminar, Christian-Albrechts-Universit\"{a}t Kiel}, year = {2000}, address = {Ludewig-Meyn-Str. 4, D-24098 Kiel, Germany}, month = apr, type = {Preprint}, } @TECHREPORT{KlarUnterreiter2001, author = {. and .}, title = {Uniform stability of a Finite Difference scheme for transport equations in diffusive regimes}, institution = {Fachbereich Mathematik, Techn. Universit\"{a}t Darmstadt}, year = {2001}, address = {D-64289 Darmstadt, Germany}, note = {published in: SIAM J. Numer. Anal. 40, No. 3, pp 891-913, 2003}, type = {Preprint}, } @TECHREPORT{CanutoTabaccoUrban1997, author = { . and .}, title = {The Wavelet Element method part I: Construction and analysis}, institution = {Dipartimento di Mathematica, Politecnico di Torino}, year = {1997}, address = {Corso Duca degli Abruzzi, 24 - 10129 Torino - Italia}, type = {Technical Report}, number = {13}, } @TECHREPORT{EhlersMarkert2000, author = {. and .}, title = {Modelling of Open and Closed Cellular Foams}, institution = {Fakult\"{a}t Bau- und Umweltingenieurwissenschaften, Institut f\"{u}r Mechanik (Bauwesen), Lehrstuhl II, Universit\"{a}t Stuttgart}, year = {2000}, address = {Pfaffenwaldring 7, 70569 Stuttgart (Deutschland)}, note = {published in: GAMM 2000, 2.-7. April, G\"{o}ttingen}, type = {Preprint}, } @TECHREPORT{KeshtibanBelblodiaWebster2004, author = {. and .}, title = {Compressible flow solvers for low mach number flows {}--{} a review}, institution = {Institute of Non-Newtonian Fluid Mechanics, Dpt. of Computer Science, University of Wales}, year = {2004}, address = {Swansea, SA2 8PP, UK}, note = {\url{http:/ / www.cs.swan.ac.uk/ reports/ 2004.html}}, type = {Preprint}, number = {CSR 2-2004 }, } @TECHREPORT{RadmoserScherzerWeickert1999, author = {. and . and .}, title = {Scale--Space Properties of Nonstationary Iterative Regularization Methods}, institution = {Cpmputer Vision, Graphics, and Pattern Recognition Group; Dpt. of Math. and Computer Science, University of Mannheim}, year = {1999}, address = {D-68131 Mannheim, Germany}, month = oct, type = {Technical Report}, number = {8/1999}, } @TECHREPORT{BirosGhattas1999, author = {. and .}, title = {Parallel Newton--Krylow methods for {PDE}--constrained optimization}, institution = {Comp. Mech. Maboratory, Carnegie Mellon University, Pittsburgh}, year = {1999}, address = {Pensylvania, 15213, USA}, type = {Preprint}, } @TECHREPORT{Tardos2002, author = {.}, title = {Granular Flows in the Intermediate Regime}, institution = {Dpt. of Chemical Engineering, The City College of The City University of New York}, year = {2002}, address = {Convent Ave. \& 140-th Street, New York, NY 10031}, month = jul, type = {Technical Report}, } @TECHREPORT{TardosMcNamara2000, author = {. and .}, title = {A fluid mechanistic approach to slow (colomb) and intermediate powder flows}, institution = {Dpt. of the Chemical Engineerung, The City College of the City University of New York}, year = {2000}, address = {Convent Ave. and 140-th Street, New York, NY 10031, USA}, type = {Preprint}, } @TECHREPORT{Hendy2002, author = {.}, title = {Instabilities in Granular flows}, institution = {IRL Applied Math. Industrial Research Ltd.}, year = {2002}, address = {PO Box 31-310, Lower Hutt, New Zealand}, month = apr, type = {Preprint}, } @TECHREPORT{GazzolaSecchi1997, author = {. and .}, title = {Some results about stationary {Navier}--{Stokes} equations with a pressure--dependent viscosity}, institution = {Dipartmento di Scienze T.A. - via Cavour 83}, year = {1997}, address = {15100 Alessandria (Italy)}, note = {\url{http:/ / www1.mate.polimi.it/ ~gazzola/ papers.html}}, type = {Preprint}, } @TECHREPORT{Rannacher2004, author = {.}, title = {Incompressible viscous flows}, institution = {Institute of Applied Mathematics, University of Heidelberg, Germany}, year = {2004}, type = {Preprint}, } @TECHREPORT{BijlCarpenter2000, author = { .}, title = {Iterative solution techniques for unsteady flow computations using higher order time integration schemes}, institution = {Faculty of Aerospace Engineering, Delft University of Technology, The Netherlands}, year = {2000}, note = {to appear in: Int. J. Numer. Meth. Fluids}, type = {Preprint}, } @TECHREPORT{SeegerHoffmann2004, author = {. and .}, title = {The cumulant method for the space--homogeneous Boltzmann equation}, institution = {Institut f\"{u}r Physik, Techn. Universit\"{a}t Chemnitz, Germany}, year = {2004}, address = {09107 Chemnitz, Germany}, note = {to appear in: Continuum Mechanics and Thermodynamics}, type = {Preprint}, } @TECHREPORT{MatthiesTobiska2004, author = {. and .}, title = {Inf--sup stable non--conforming finite elements of arbitrary order on triangles}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Otto-von-Guericke-Universit\"{a}t Magdeburg}, year = {2004}, address = {Postfach 4120, 39016 Magdeburg, Germany}, type = {Preprint}, number = {12}, } @TECHREPORT{SeegerHoffmann2004a, author = {. and .}, title = {Thermal Boundary Conditions for the Cumulant Method}, institution = {Institute of Physics, University of Technology Chemnitz}, year = {2004}, month = may, note = {Submitted to: Europhysics Letters}, type = {Preprint}, } @TECHREPORT{ApelBerzinsJimackKunertPlaksTsukermanWalkley2000, author = { .}, title = {Mesh shape and anisotropic elements: Theory and Practice}, institution = {Fak. Mathematik, TU Chemnitz}, year = {2000}, address = {D-09107 Chemnitz, Germany}, note = {Published in: (ed.): The Mathematics of Finite Elements and Applications X, Elsevier, Amsterdam, 2000, 367-376}, type = {Preprint}, } @TECHREPORT{QuarteroniSalaValli2004, author = {. and . and .}, title = {The Swiss--carpet domain decomposition preconditioner}, institution = {CMCS-IACS-SB, EPFL}, year = {2004}, address = {CH-1015 Lausanne, Swizerland}, type = {Preprint}, } @TECHREPORT{ApelGrosmanJimackMeyer2001, author = {. and .}, title = {A New Methodology for Anisotropic Mesh Refinement Based Upon Error Gradients}, institution = {Fak. f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, year = {2001}, address = {09107 Chemnitz, Germany}, type = {Preprint}, } @TECHREPORT{FormaggiaMichelettiPerotto2003, author = {. and .}, title = {Anisotropic Mesh Adaption in Computational Fluid Dynamics: Application to the Advection--Diffusion--Reaction and the {Stokes} Problems}, institution = {MOX, Modellistica e Calcolo Scientifico, Dipartimento di Mathematica ``F. Brioschi``, Politecnico di Milano}, year = {2003}, address = {via Bornadi 9, I-20133 Milano, Italy}, type = {Preprint}, } @TECHREPORT{CarstensenFunken2001, author = {. and .}, title = {Averaging technique for FE--A Posteriori error control in elasticity. Part I: Conforming {FEM}}, institution = {Universit\"{a}t Kiel}, year = {2001}, note = {published in: Comput. Meth. Appl. Mech. Engrg. 190 (1002), 18-19, 2483,2488}, type = {Preprint}, } @TECHREPORT{CarstensenFunken2001a, author = {. and .}, title = {A posteriori error control in low--order Finite Element discretisations of incompressible stationary flow problems}, institution = {Universit\"{a}t Kiel}, year = {2001}, note = {published in: Math. Comp. 70 (2001), 236, p. 1353-1381}, type = {Preprint}, } @TECHREPORT{CarstensenBartels2001, author = { .}, title = {Each averaging technique yields reliable A Posteriori error control in {FEM} on unstructured grids Part II: Low order conforming, nonconforming, and mixed {FEM}}, institution = {Universit\"{a}t Kiel}, year = {2001}, note = {published in: Math. Comp. 73 (2004), 1153-1165}, type = {Preprint}, } @TECHREPORT{BrandtsChen2003, author = {. and .}, title = {An alternative to the least--squares mixed finite element method for elliptic problems}, institution = {Korteweg-de Vries Institute for Mathematics, Faculty of Science, Universityx of Amsterdam}, year = {2003}, address = {Plantage Muidergracht 24, 1018 TV Amsterdam, Netherlands}, type = {Preprint}, } @TECHREPORT{CarstensenSchoeberl2000, author = {. and .}, title = {Residual--based a posteriori error estimate for a mixed Reissner--Mindelin plate finite element method}, institution = {Universit\"{a}t Linz }, year = {2000}, note = {submitted to: Numer. Math., 2000}, type = {Preprint}, } @TECHREPORT{GriebelPreusserRumpfSchweitzerTelea2004, author = { . and .}, title = {Flow Field Clustering via Algebraic Multigrid}, institution = {Institute for Numerical Simulation, University of Bonn}, year = {2004}, address = {Wegelerstr. 6, 53115 Bonn, Germany}, note = {SFB611}, type = {Preprint}, number = {153}, } @TECHREPORT{Hoffman2004, author = {.}, title = {Computation of the mean drag of a surface mounted cube using adaptive DNS/LES}, institution = {Courant Institute of Math. Science, New York University}, year = {2004}, address = {251 Mercer Street, New York, NY 10012-1185, USA}, note = {\url{http:/ / www.cims.nyu.edu/ ~hoffman/ archive/ papers/ sc_benchmark.pdf}}, type = {Preprint}, } @TECHREPORT{Suttmeier2004k, author = {.}, title = {Optimised mesh design for FE--{Galerkin} schemes using global norms}, institution = {Universit\"{a}t Dortmund, Fachbereich Mathematik, Lehrstuhl X}, year = {2004}, address = {Germany}, type = {Technical Report}, number = {267}, } @TECHREPORT{FischerKirkilionisLogashenkoWittum2004, author = {. and . and . and .}, title = {Fast Numerical Integration for Simulation of Disperse Systems}, institution = {Zentrum f\"{u}r Molekulare Biologie, Universit\"{a}t Heidelberg}, year = {2004}, address = {Im Neuheimer Feld 282, 69120 Heidelberg}, month = mar, type = {Preprint}, } @TECHREPORT{LogashenkoFischerMotzGillesWittum2004, author = {. and . and .}, title = {Simulation of a Crystal Growth and Attrition in a Stirred Tank}, institution = {Institut f\"{u}r Informatik, Universit\"{a}t Heidelberg}, year = {2004}, address = {Im Neuheimer Feld 282, 69120 Heidelberg}, month = mar, type = {Preprint}, } @TECHREPORT{BroserSchulteRothHelmchenLangWittumSakmann2003, author = {. and . and . and . and . and .}, title = {Nonlinear anisotropic diffusion filtering of three--dimensional image data from two--photon microscopy}, institution = {Zellphysionlogie, Max-Planck-Institut f\"{u}r Medizinische Forschung}, year = {2003}, address = {Jahnstr. 29, D-69120 Heidelberg, Germany}, month = oct, type = {Preprint}, } @TECHREPORT{TurekBuijssenGrajewskiWobker2004a, author = {. and . and .}, title = {High Performance {FEM} Simulation}, institution = {Universit\"{a}t Dortmund}, year = {2004}, address = {Leonhard-Euler-Str. 5, 44221 Dortmund}, month = sep, note = {p.52--55}, type = {Research Report, NRW Graduate School of Production Engineering and Logistics }, } @TECHREPORT{Guias2004, author = {.}, title = {A new stochastic method for approximation of diffusion equations and applications to spatially inhomogenous coagulation process}, institution = {Dpt. of Mathematics, University Dortmund}, year = {2004}, address = {Vogelpothsweg 87, 44221 Dortmund, Germany}, note = {Published in: Deutsche Mathematiker-Vereinigung, Jahrestagung 2004, 12.-17. September}, type = {Preprint}, } @TECHREPORT{SchmidKoestlerRuede2004, author = {. and K\"{. and R\"{u}.}, title = {Multigrid accelerated Poisson--Solver for ab initio Moleculardynamic Applications}, institution = {Lehrstuhlbericht 04-5, IMMD10}, year = {2004}, address = {FAU Erlangen}, } @TECHREPORT{WeissHellwagnerRuedeStals2002, author = {. and . and R\"{ .}, title = {Data Locality Optimizations to Improve the Efficiency of Multigrid Methods}, institution = {Lehrstuhlbericht 02-1, IMMD10}, year = {2002}, address = {FAU Erlangen}, } @TECHREPORT{KowarschikWeiss2002, author = {. and .}, title = {Cache Performance Tuning of Numerically Intensive Codes}, institution = {Lehrstuhlbericht 02-2, IMMD10}, year = {2002}, address = {FAU Erlangen}, } @TECHREPORT{BabuskaNistor2004, author = {. and .}, title = {Interior numerical approximation of boundary value problems with a distributional data}, institution = {Math. Dept. Penn St. Univ.}, year = {2004}, note = {\url{http:/ / www.math.psu.edu/ nistor/ ART/ gfemd.pdf}}, } @TECHREPORT{BacutaNistorZikatanov2004, author = {. and .}, title = {Regularity and well posedness for the laplace operator on polyhedral domains}, institution = {Math. Dept. Penn St. Univ.}, year = {2004}, note = {\url{http:/ / www.math.psu.edu/ nistor/ ART/ bonp_note.pdf}}, } @TECHREPORT{BangerthKanschat1999, author = {. and .}, title = {Concepts for Object--Oriented Finite Element Software {}--{} the deal.II Library}, institution = {IWR Heidelberg}, year = {1999}, number = {99-43}, } @TECHREPORT{WhaleyDongarra1997, author = {. and .}, title = {Automatically Tuned Linear Algebra Software}, institution = {University of Tennessee}, year = {1997}, month = dec, number = {UT-CS-97-366}, } @TECHREPORT{GargSharapov2001, author = {. and .}, title = {Techniques for Optimizing Applications: High Performance Computing}, institution = {Sun Microsystems Inc.}, year = {2001}, } @TECHREPORT{Hossfeld2001, author = {.}, title = {Perspektiven f\"{u}r Supercomputer--Architekturen}, institution = {FZ J\"{u}lich - Zentralinstitut f\"{u}r Angewandte Mathematik}, year = {2001}, } @TECHREPORT{LeiLiaodelaPenaAnderson2001, author = {. and . and .}, title = {A Moving Grid Algorithm Based on Deformation Method}, institution = {Department of Mathematics, University of Texas - Arlington}, year = {2001}, type = {Preprint}, } @TECHREPORT{ElmanHowleShuttleworthTuminaro2004, author = {. and .}, title = {Block Preconditioners Based on Approximate Commutators}, institution = {Sandia National Laboratories}, year = {2004}, address = {PO Box 969, MS 9159, Livermore, CA 94551 USA}, type = {Preprint}, } @TECHREPORT{SauterWarnke2004, author = {. and .}, title = {Composite Finite Elements for Elliptic Boundary Value Problems with Discontinuous Coefficients {}}, institution = {Institut f\"{u}r Mathematik, Universit\"{a}t Z\"{u}rich}, year = {2004}, address = {Winterthurerstr. 190, Ch-8057 Z\"{u}rich, Switzerland }, note = {\url{http:/ / www.math.unizh.ch/ fileadmin/ math/ preprints/ 12-04.pdf}}, type = {Preprint}, number = {12-2004}, } @TECHREPORT{Khattri2005, author = {.}, title = {Adaptive Mesh Refinement ( `Finite Volume Methods`)}, institution = {Department of Applied Mathematics, University of Bergen}, year = {2005}, note = {\url{http:/ / www.ddm.org/ DD16/ Talks/ kumar1.pdf}}, type = {Talk}, } @TECHREPORT{MohammadiMedic1996, author = {. and .}, title = {A Critical Evaluation of the Classical {} K--E {} Model and Wall--Laws for Unsteady Flows Over Bluff Bodies}, institution = {Institut National De Recherche En Informatique Et En Automatique}, year = {1996}, address = {Unite De Recherche INRIA Rocquencourt, Domaine De Voluceau, Rocquencourt, BP 105, 78153 LE CHESNAY Cedex (France)}, note = {ISSN 0249-6399}, type = {Technical Report}, number = {N 3056}, } @TECHREPORT{Junk2001a, author = {.}, title = {Do finite Volume Mehtods Need a Mesh?}, institution = {Universit\"{a}t Kaiserslautern}, year = {2001}, address = {Fachbereich Mathematik, Universit\"{a}t Kaiserslautern,67663 Kaiserslautern, Germany.}, note = {\url{http:/ / wissrech.iam.uni-bonn.de/ meshfree/ }}, type = {Talk}, } @TECHREPORT{HietelJunkKeckTeleaga2000, author = {. and . and . and .}, title = {The Finite--Volume--Particle Method for Conservation Laws.}, institution = {Fraunhofer-Institut f\"{u}r Techno- und Wirtschaftsmathematik}, year = {2000}, address = {Fraunhofer-Institut f\"{u}r Techno- und Wirtschaftsmathematik, Gottlieb-Daimler Stra"se, 67663 Kaiserslautern, Germany .}, type = {Technical Report}, } @TECHREPORT{CanicMikelicTambaca2005, author = {. and . and .}, title = {A Two--Dimensional Effective Model Describing Fluid--Structure Interaction in Blood Flow: Analysis, Simulation and Experimental Validation}, institution = {Department of Mathematics, UNiversity of houston}, year = {2005}, address = {Department of Mathematics, University of Housten, 4800 Calhoun Rd., Houston TX 77204-3476, United States}, note = {\url{http:/ / www.math.uh.edu/ ~canic/ papers/ hemodynamics/ crasRoland.pdf}}, type = {Technical Report}, } @TECHREPORT{MichelettiPerotto2004, author = {. and .}, title = {Reliability and Efficiency of an Anisotropic Zienkiewicz--Zhu Error Estimator}, institution = {MOX-Modellistica e Cacolo Scientifico}, year = {2004}, address = {MOX-Modellistica e Cacolo Scientifico, Dipartimento Di Mathematica `F.Brioschi` Politecnico Di Milano Via Bonardi 9, 20133 Milano, Italy}, note = {\url{http:/ / mox.polimi.it/ it/ progetti/ pubblicazioni/ quaderni/ mox37.pdf}}, type = {Technical Report}, } @TECHREPORT{Stratmann2002, author = {.}, title = {Numerische Str\"{o}mungsuntersuchung der Gasspaltstr\"{o}mung am Hauptrotor--Geh\"{a}usespalt in Schraubenmaschinen}, institution = {Universit\"{a}t Dortmund}, year = {2002}, address = {Raum 101, Emil-Figge-Str. 71 b, Universit\"{a}t Dortmund}, note = {ISSN 0945-1870}, type = {Technical Report}, } @TECHREPORT{KauderSachs2002, author = {. and .}, title = {Experimental Investigation of the Gas Flow at the Male Rotor Housing Gap of a Plane Screw--Type Machine Model}, institution = {Universit\"{a}t Dortmund}, year = {2002}, address = {Raum LE 5-360, Leonhard-Euler-Str.5 , Universit\"{a}t Dortmund}, note = { ISSN 0945-1870}, type = {Technical Rerport}, } @TECHREPORT{BristeauGlowinskiPeriauxViviand1987, author = {. and . and . and .}, title = {Presentations of Problems and Discussion of Results}, institution = {INRIA}, year = {1987}, address = {INRIA, B.P.105, 78153 Le Chesnay Cedex , France}, type = {Technical Rerport}, } @TECHREPORT{MeiselEhrhard2005, author = {. and .}, title = {Electrically--Excited (Electroosmotic) Flows in Microchannels for Mixing Applications {}}, institution = {Institute for Micro Process Engineering}, year = {2005}, address = {Forschungszentrum Karlsruhe, Institute for Micro Process Engineering, P.O.Box 3640, d-76021 Karlsruhe, Germany}, type = {Preprint}, } @TECHREPORT{HoteitAckererMoseErhelPhilippe2002, author = {. and . and . and . and .}, title = {New Two--Dimensional slope Limiters for Discontinuous {Galerkin} Methods on Arbitrary Meshes}, institution = {INRIA}, year = {2002}, address = {INRIA, Institut National De Recherche En Informatique Et En Automatique}, note = {ISSN 0249-6399}, type = {Technical Rerport}, number = {N 4491}, } @TECHREPORT{Bebendorf2005, author = {.}, title = {Why Approximate LU Decomposition of Finite Element Discretizations of Elliptic Operators {} can be Computed with Almost Linear Complexity {}}, institution = {Max-Planck-Institut f\"{u}r Mathematik- Leipzig}, year = {2005}, address = {Max-Planck-Institut f\"{u}r Mathematik in den naturwissenschaften Leipzig}, type = {Preprint}, number = {8}, } @TECHREPORT{BenziGolubLiessen2005, author = {. and . and .}, title = {Numerical solution of Saddle Point Problems {}}, institution = {Department of Mathematics and Computer Science , Emory University }, year = {2005}, address = {Department of Mathematics and Computer Science , Emory University , Atlanta, georgia 30322, USA}, note = {\url{http:/ / www.math.tu-berlin.de/ ~liesen/ / Publicat/ BenGolLie05.pdf}}, type = {Technical Rerport}, } @TECHREPORT{OlshanskiiReusken2004a, author = {. and .}, title = {Analysis of a {Stokes} Interface Problem}, institution = {Department of Mechanics and Mathematics , Moscow State University}, year = {2004}, address = {Department of Mechanics and Mathematics , Moscow State University, Moscow 119899, Russia.}, note = {ftp://ftp.igpm.rwth-aachen.de/pub/reports/ps/IGPM237.ps.gz}, type = {Technical Rerport}, number = {237}, } @TECHREPORT{MichelettiPerotto2004a, author = {. and .}, title = {{} Anisotropic Mesh Adaptivity via a Dual--Based a Posteriori Error Estimation for Semiconductors}, institution = {MOX-Modeling and Scientific Computing.}, year = {2004}, address = {MOX-Modeling and Scientific Computing , Dipartimento Di Matematica `F.Brioschi`, Politecnico Di Milano, Italy.}, note = {\url{http:/ / mox.polimi.it/ it/ progetti/ pubblicazioni/ quaderni/ mox47.pdf}}, type = {Technical Rerport}, } @TECHREPORT{ChuTai2005, author = {. and .}, title = {MoXi: Real--Time Ink Dispersion in Absorbent Paper}, institution = {The Hong Kong University of Science and Technology}, year = {2005}, address = {Vision and Graphics Group, Computer Science Department, The Hong Kong University of Science and Technology.}, note = {\url{http:/ / visgraph.cs.ust.hk/ MoXi/ moxi.pdf}}, type = {Technical Rerport}, } @TECHREPORT{MatsumotoZadehEhrhard2004, author = {. and . and .}, title = {Quantitaive Measurement of Depth Averaged Concentration Fields in Microchannels by Means of a Fluorescence Intensity Method}, institution = {Kansai University , Department of Mechanical Engineering Systems}, year = {2004}, address = {Kansai University , Department of Mechanical Engineering Systems}, type = {Technical Rerport}, } @TECHREPORT{PetersreicheltReusken2004, author = {. and . and .}, title = {Fast Iterative solvers for Discrete {Stokes} Equations {}}, institution = {Institut f\"{u}r geometrie und Praktische Mathematik, RWTH- Aachen}, year = {2004}, address = {Institut f\"{u}r geometrie und Praktische Mathematik, RWTH- Aachen, D-52056 Aachen, Germany.}, note = {\url{http:/ / www.igpm.rwth-aachen.de/ reichelt/ papers/ PaperStokes.pdf}}, type = {Technical Rerport}, number = {241}, } @TECHREPORT{Bruter2003, author = {.}, title = {The Poincare Surprises Claude}, institution = {Departement de Mathematiques ,Universite Paris .}, year = {2003}, address = {Universite Paris 12, 61 Avenue du General de Gaulle, 94000 Creteil France}, note = {\url{http:/ / arpam.free.fr/ the_poincare_surprises.htm}}, type = {Technical Rerport}, } @TECHREPORT{BucherMeyerGoerkeKreissig2004, author = {. and .}, title = {About A Nodal Based Transfer Algorithm and Error Estimators in Nonlinear Adaptive {FEM}}, institution = {Institute of Mechanics, Chemnitz University of Technology}, year = {2004}, address = {Institute of Mechanics, Chemnitz University of Technology, Str.d.Nationen 62, d-09111 Chemnitz, Germany}, note = {\url{http:/ / www-user.tu-chemnitz.de/ ~anbuc/ eccomas.ps}}, type = {Technical Rerport}, } @TECHREPORT{NagrathJansenLahey2003, author = {. and . and .}, title = {Three Dimensional Simulation of Incompressible Two--Phase Flows Using a Stabilized finite Method and a Level Set Approach}, institution = {Scientific Computation Research Center, rensselaer Plytechnic Institute , Troy, New York, USA.}, year = {2003}, address = {Research Center,CII-7013, Rensselaer Plytechnic Institute , Troy, New York 12180, USA.}, note = {\url{http:/ / www.scorec.rpi.edu/ REPORTS/ 2003-12.pdf}}, type = {Preprint}, } @TECHREPORT{RutkaWiegmann2002, author = {. and .}, title = {A Fast Finite Difference Method for Elliptic PDEs in Domains with Non--Grid aligned Boundaries with Application to 3D Linear Elasticity}, institution = {Fraunhofer ITWM Kaiserslautern, Germany}, year = {2002}, address = {Fraunhofer ITWM Kaiserslautern, Germany}, note = {\url{http:/ / www.itwm.fraunhofer.de/ sks/ employees/ wiegmann/ References/ ecmi.pdf}}, type = {Technical Rerport}, } @TECHREPORT{BottassomaisanoMichelettiPerotto2005, author = {. and . and . and .}, title = {New Recovery Based a Posteriori Error Estimators}, institution = {D.Guggenheim School of Aerospace Engineering, Georgia Institute of Technology}, year = {2005}, address = {D.Guggenheim School of Aerospace Engineering, Georgia Institute of Technology, 270 Ferst Dr.30332-0150 Atlanta GA, USA.}, note = {\url{http:/ / www.aero.polimi.it/ ~bottasso/ downloads/ GMaisanoSMichelettiSPerottoCLBottasso_2005.pdf}}, type = {Technical Rerport}, } @TECHREPORT{RepinSauterSmolianski2002, author = {. and . and .}, title = {A Posteriori Error Estimation for the Dirichlet Problem with Account of the Error in the approximation of Boundary Conditions}, institution = {St. Petersburg}, year = {2002}, address = {St. Petersburg}, note = {\url{http:/ / www.math.unizh.ch/ fileadmin/ math/ preprints/ 03-02.pdf}}, type = {Preprint}, } @TECHREPORT{CarstensenSauter2001, author = {. and .}, title = {A Posteriori Error Analysis for Elliptic PDEs on Domains with Complicated Structures}, institution = {Institut f\"{u}r Mathematik, Universit\"{a}t Z\"{u}rich}, year = {2001}, address = {Institut f\"{u}r Mathematik, Universit\"{a}t Z\"{u}rich,Winterthurerstrasse 190, CH-8057 Z\"{u}rich.}, note = {\url{http:/ / www.math.unizh.ch/ fileadmin/ math/ preprints/ 10-01.pdf}}, type = {Preprint}, } @TECHREPORT{RepinSauterSmolianski2003, author = {. and . and .}, title = {A Posteriori Error Estimation for the Poisson Equation with Mixed Dirichlet/Neumann Boundary Conditions}, institution = {V.A. Steklov Institute of Mathematics}, year = {2003}, address = {V.A. Steklov Institute of Mathematics, Russian Academy of Sciences, 191011 St. Petersburg, Russia.}, note = {\url{http:/ / www.math.unizh.ch/ fileadmin/ math/ preprints/ 02-03.pdf}}, type = {Preprint}, } @TECHREPORT{Uhlmann2003, author = {.}, title = {Simulation of Particulate Flows on Multi--Processor Machines with Distributed Memory}, institution = {Dept. Combutibles Fosiles, CIEMAT}, year = {2003}, address = {Dept. Combutibles Fosiles, CIEMAT, Avda.Complutense 22, 28040 Madrid (Spain)}, note = {\url{http:/ / www.ciemat.es/ sweb/ comfos/ personal/ uhlmann/ particle/ report/ report2.pdf}}, type = {Technical Report}, number = {1039}, } @TECHREPORT{Uhlmann2003a, author = {.}, title = {First Experiments with the Simulation of Particulate Flows}, institution = {Dept. Combutibles Fosiles, CIEMAT}, year = {2003}, address = {Dept. Combutibles Fosiles, CIEMAT, Avda.Complutense 22, 28040 Madrid (Spain)}, note = {\url{http:/ / www.ciemat.es/ sweb/ comfos/ personal/ uhlmann/ particle/ report/ report1.pdf}}, type = {Technical Report}, number = {1020}, } @TECHREPORT{CroceGriebelSchweitzer2004, author = {. and . and .}, title = {A Parallel Level--Set Approach for Two--Phase Flow Problems with Surface Tension in Three Space Dimensions}, institution = {Institute for Numerical Simulation}, year = {2004}, address = {Institute for Numerical Simulation, Division Numerical Simulation in the Natural and Engineering Sciences ,Wegelerstr. 6 ,53115 Bonn ,Germany}, note = {\url{http:/ / wissrech.ins.uni-bonn.de/ research/ pub/ schweitz/ lsm.pdf}}, type = {Preprint}, } @TECHREPORT{GrossReichteltReusken2004, author = { .}, title = {A Finite Element Based Level Set Method for Two--Phase Incompressible Flows}, institution = {Institut f\"{u}r Geometriw und Praktische Mathematik}, year = {2004}, address = {Institut f\"{u}r Geometriw und Praktische Mathematik}, note = {ftp://ftp.igpm.rwth-aachen.de/pub/reports/pdf/IGPM243.pdf}, } @TECHREPORT{POllulReusken2004, author = {. and .}, title = {Preconditioners for Linearized Discrete Compressible Euler Equations}, institution = {Institut f\"{u}r Geometriw und Praktische Mathematik}, year = {2004}, address = {Institut f\"{u}r Geometriw und Praktische Mathematik, RWTH \_Aachen, D-52056 Aachen. Germany}, note = {ftp://ftp.igpm.rwth-aachen.de/pub/reports/pdf/IGPM247.pdf}, type = {Preprint}, } @TECHREPORT{BraackRichter2005, author = {. and .}, title = {Solutions of 3D {Navier}--{Stokes} Benchmark Problems with Adaptive Finite Elements}, institution = {Institute of Applied Mathematics, University of Heidelberg.}, year = {2005}, address = {Institute of Applied Mathematics, University of Heidelberg, INF 294, 69120 Heidelberg, Germany}, type = {Preprint}, } @TECHREPORT{JunkYong2003, author = {. and .}, title = {Rigorous {Navier}--{Stokes} Limit of the Lattice Boltzmann Equation}, institution = {FB Mathematik, Universit\"{a}t Kaiserslautern}, year = {2003}, address = {FB Mathematik, Universit\"{a}t Kaiserslautern, 67663 Kaiserslautern, Germany}, note = {\url{http:/ / www.mathematik.uni-kl.de/ ~junk/ publications/ papers/ J+Y01.ps}}, type = {Technical Report}, } @TECHREPORT{LehoucqScott1996, author = {. and .}, title = {An Evaluation of {} Software for Computing Eigenvalues of Sparse Nonsymmetric Matrices}, institution = {Mathematics and Computer Science Division, Argonne National Laboratory}, year = {1996}, address = {Mathematics and Computer Science Division, Argonne National Laboratory, Argonne, IL 60439.}, type = {Technical Report}, } @TECHREPORT{LehoucqScott1997, author = {. and .}, title = {Implicitly Restarted Arnoldi Methods and Eigenvalues of the Discretized {Navier} {Stokes} Equations}, institution = {Sandia National Laboratories}, year = {1997}, address = {Sandia National Laboratories, MS 1110, P.O.Box 5800 Albuquerque, NM 87185-1110.}, type = {Technical Report}, } @TECHREPORT{Junk2000, author = {.}, title = {LBM--{} Discrete Dynamics and Finite Difference Method}, institution = {FB Mathematik, Universitaet Kaiserslautern.}, year = {2000}, address = {FB Mathematik, Universitaet Kaiserslautern, Erwin-Schroedingerstrasse, 67663 Kaiserslautern, Germany.}, note = {\url{http:/ / www.mathematik.uni-kl.de/ ~junk/ publications/ papers/ Jun00d.ps.gz}}, type = {Technical Report}, } @TECHREPORT{JunkYang2003, author = {. and .}, title = {Asymptotic Analysis of Finite Difference Methods}, institution = {FB Mathematik, Universit\"{a}t Kaiserslautern}, year = {2003}, address = {FB Mathematik, Universit\"{a}t Kaiserslautern, Erwin-Schr\"{o}dingerstra"se, 67663, Kaiserslautern, Germany}, note = {\url{http:/ / www.mathematik.uni-kl.de/ ~junk/ publications/ papers/ J+Y03a.ps}}, type = {Preprint}, } @TECHREPORT{LinTobiskaZhou2001a, author = {. and . and .}, title = {On the superconvergence of nonlinear low order finite elements applied to the Poisson equation}, institution = {Fakult\"{a}t f\"{u}r Mathematik}, year = {2001}, address = {Otto-von-Guericke-Universit\"{a}t Magdeburg}, type = {Preprint}, number = {17}, } @TECHREPORT{ProvatasGoldenfeldDantzigLaCombeLupulescuKossGlicksmanAlmgren1999, author = {. and . and . and . and . and .}, title = {Crossover Scaling in Dendritic Evolution at Low Undercooling}, institution = {Dpt. of Physics, University of Illinois at Urbana-Champaign}, year = {1999}, address = {1110 West-Green Street, Urbana, IL, 61801}, month = jun, type = {Technical Report}, } @TECHREPORT{UniversitaetBremen2003, author = {Universit\"{a}t Bremen}, title = {T\"{a}tigkeitsbericht 2001/2002}, institution = {Universit\"{a}t Bremen}, year = {2003}, type = {Technical Report}, } @TECHREPORT{ColellaDunningGroppKeyes2003, author = {. and . and . and .}, title = {A Science--based Case for Large--scale Simulation}, institution = {Science Case for Large-scale Simulation}, year = {2003}, address = {Office of Science of the U.S. Department of Energy}, note = {whitepaper commissioned by the U.S. DOE Office of Science, \url{http:/ / www.pnl.gov/ scales/ }}, } @TECHREPORT{MuellerSzilagyi2000, author = { .}, title = {Nonconforming Finite Element Approximation of Steady, Incompressible 3D Euler Equations in a Rotating Frame of Reference}, institution = {Technische Universit\"{a}t M\"{u}nchen, Lehrstuhl f\"{u}r Hydraulische Maschinen}, year = {2000}, address = {Boltzmannstrasse 15,85748 Garchingt}, note = {to appear in: Int. J. Numer. Meth. Fluids}, type = {Preprint}, } @TECHREPORT{ZijlemaSegalWesseling1994, author = {. and . and .}, title = {Finite Volume Computation of 2D incompressible Turbulent Flows in General Coordinates on Staggered Grids}, institution = {Faculty of Technical Mathematics and Informatics}, year = {1994}, address = {Faculty of Technical Mathematics and Informatics, Delft University of Technology, Netherlands}, note = {\url{http:/ / ta.twi.tudelft.nl/ isnas/ report94-24/ isrep2.html}}, type = {techreport}, } @TECHREPORT{AbbrahamBehrHeinkenschloss2004, author = {. and . and .}, title = {Shape Optimization in Seatdy Blood Flow: A Numerical Study of Non--Newtonian Effects}, institution = {Mechanical Enineering and Appliend Mathematics, University of Pennsylvania}, year = {2004}, address = {Phiadelphia, PA 19104, USA}, note = {\url{http:/ / www.caam.rice.edu/ ~heinken/ papers/ shape_opt_blood_steady.pdf}}, type = {Technical Report}, } @TECHREPORT{BartlettBoggsLongSalingervanBloemenWaanders2002, author = { .}, title = {Large Scale Non--Linear Programming for {PDE} Constrained Optimization.}, institution = {Sandia National Laboratories}, year = {2002}, address = {Albuquerque, New Mexico 87185 and Livermore, California 94550}, month = oct, note = {\url{http:/ / endo.sandia.gov/ DAKOTA/ papers/ pdeco_ldrd_finalreport.pdf}}, type = {Technical Report}, } @TECHREPORT{BuijssenWobkerTurek2005, author = {. and . and .}, title = {High Performance {FEM} Simulation in {CFD} and {CSM}}, institution = {Universit\"{a}t Dortmund}, year = {2005}, address = {Leonhard-Euler-Str. 5, 44221 Dortmund}, month = aug, note = {p.46--49}, type = {Research Report, NRW Graduate School of Production Engineering and Logistics}, } @TECHREPORT{LefohnWhitaker2002, author = {. and .}, title = {A {GPU}--Based Three--dimensional Level Set Solver with Curvature Flow}, institution = {University of Utah, School of Computing}, year = {2002}, address = {School of Computing, University of Utah, Salt Lake City, UT 84112 USA}, month = dec , note = {\url{http:/ / www.cs.utah.edu/ techreports/ 2002/ pdf/ UUCS-02-017.pdf}}, } @TECHREPORT{SanderTatarchukMitchell2004, author = {. and .}, title = {Explicit Early--Z Culling for Efficient Fluid Flow Simulation and Rendering}, institution = {ATI Reasearch}, year = {2004}, note = {\url{http:/ / www.ati.com/ developer/ techreports/ ATITechReport_EarlyZFlow.pdf}}, } @TECHREPORT{AckerRivkindTurek1998, author = {. and . and .}, title = {Software--Praktikum f\"{u}r Anf\"{a}nger}, institution = {Universit\"{a}t Heidelberg}, year = {1998}, note = {Wintersemester 1998/99 - Skriptum 99-01}, } @TECHREPORT{KarypisKumar1998, author = {. and .}, title = {{METIS} {}--{} A Software Package for Partitioning Unstructured Graphs, Partitioning Meshes, and Computing Fill--Reducing Orderings of Sparse Matrices}, institution = {University of Minnesota, Department of Computer Science}, year = {1998}, note = {\url{http:/ / www-users.cs.umn.edu/ ~karypis/ metis/ metis/ index.html}}, } @TECHREPORT{PreisDiekmann1996, author = {. and .}, title = {The PARTY Partitioning Library, User Guide {}--{}--{} Version 1.1}, institution = {Department of Mathematics and Computer Science, Universit\"{a}t Paderborn}, year = {1996}, month = sep, note = {\url{http:/ / wwwcs.uni-paderborn.de/ fachbereich/ AG/ monien/ RESEARCH/ PART/ party.html}}, } @PHDTHESIS{Becker1995, author = {.}, title = {An Adaptive Finite Element Method for the Incompressible {Navier}--{}--{Stokes} Equations on Time--Dependent Domains}, school = {Universit\"{a}t Heidelberg}, year = {1995}, } @PHDTHESIS{Kilian2001, author = {.}, title = {{Ein verallgemeinertes Gebietszerlegungs-/Mehrgitterkonzept auf Parallelrechnern}}, school = {Universit\"{a}t Dortmund}, year = {2001}, } @PHDTHESIS{Schreiber1996a, author = {.}, title = {A New Finite Element Solver for the Nonstationary Incompressible {Navier}--{}--{Stokes} Equations in Three Dimensions}, school = {Universit\"{a}t Heidelberg}, year = {1996}, } @PHDTHESIS{Turek1991, author = {.}, title = {{Ein robustes Mehrgitterverfahren zur L\"{o}sung der instation\"{a}ren, inkompressiblen 2D Navier-Stokes-Gleichungen mit diskret divergenzfreien Finiten Elementen}}, school = {Universit\"{a}t Heidelberg}, year = {1991}, } @PHDTHESIS{Schieweck1996, author = {.}, title = {Parallele {L}\"{o}sung der station\"{a}ren, inkompressiblen {N}avier--{}--{S}tokes {G}leichungen}, school = {Otto-von-Guericke Universit\"{a}t Magdeburg, Fakult\"{a}t f\"{u}r Mathematik}, year = {1996}, note = {Habilitation}, } @PHDTHESIS{Droll2002, author = {.}, title = {Numerical Simulation of Fluid Flow with Pseudospectral Methods}, school = {TU Darmstadt}, year = {2002}, } @PHDTHESIS{Turek1997c, author = {.}, title = {Multilevel Pressure Schur Complement Techniques for the Numerical Solution of the Incompressible {Navier}--{Stokes} Equations}, school = {Universit\"{a}t Heidelberg}, year = {1997}, type = {Habilitation Thesis}, } @PHDTHESIS{Oswald1998, author = {.}, title = {Parallel Multigrid Algorithms for Solving the Incompressible {Navier}--{Stokes} Equations with Nonconforming Finite Elements in Three Dimensions}, school = {Universit\"{a}t Heidelberg}, year = {1998}, note = {(mit R. Rannacher)}, } @PHDTHESIS{Becker2003, author = {.}, title = {FEAST {}--{} The Realisation of Finite Element Software for High--Performance Applications}, school = {Universit\"{a}t Dortmund}, year = {2004}, note = {to appear}, } @PHDTHESIS{Schmachtel2003, author = {.}, title = {{Robuste lineare und nichtlineare L\"{o}sungsverfahren f\"{u}r die inkompressiblen Navier--Stokes-Gleichungen}}, school = {Universit\"{a}t Dortmund}, year = {2003}, month = jun, } @PHDTHESIS{Park2002, author = {.}, title = {A Study on Locking Phenomena in Finite Element Methods}, school = {Seoul National University}, year = {2002}, month = feb, type = {phdthesis}, } @PHDTHESIS{Hormann2001, author = {Hormann}, title = {Theory and Applications of Parametrizing Triangulations}, school = {Universitaet Erlangen, Institut f\"{u}r Informatik}, year = {2001}, type = {PhD Thesis}, } @PHDTHESIS{Chang2001, author = {.}, title = {A Survey of Geometric Data Structures for Ray Tracing}, school = {Department of Computer and Information Science, Polytechnic University, Brooklyn, New York}, year = {2001}, type = {PhD Thesis}, } @PHDTHESIS{Kuzmin1999, author = {.}, title = {Numerical Simulation of Reactive Bubbly Flows}, school = {University of Jyv\"{a}skyl\"{a}}, year = {1999}, type = {Jyv\"{a}skyl\"{a} Studies in Computing 2}, } @PHDTHESIS{Kobbelt1994, author = {.}, title = {Iterative Erzeugung glatter Interpolanten}, school = {Universit\"{a}t Karlsruhe }, year = {1994}, address = {Verlag Shaker, Aachen, 1995}, month = dec, type = {Dissertation}, } @PHDTHESIS{Blecher1993, author = {.}, title = {Str\"{o}mungsvorg\"{a}nge in R\"{u}hrwerkm\"{u}hlen}, school = {Technische Universit\"{a}t Carolo-Wilhelminia zu Braunschweig}, year = {1993}, address = {Fakult\"{a}t f\"{u}r Maschinenbau und Elektrotechnik}, month = jun, type = {Dissertation}, } @PHDTHESIS{Neuenschwander1995, author = {.}, title = {Elastic Deformable Contour and Surface Models for 2--D and 3--D Image Segmentation}, school = {Eidgen\"{o}ssische Techn. Hochschule Z\"{u}rich}, year = {1995}, } @PHDTHESIS{Oswald2001, author = {.}, title = {Parallele {L}\"{o}sung der instation\"{a}ren inkompressiblen {N}avier--{}--{S}tokes--{G}leichungen}, school = {Universit\"{a}t Heidelberg}, year = {2001}, note = {(mit Rolf Rannacher)}, } @PHDTHESIS{Ouazzi2005, author = {.}, title = {Finite Element Simulation of Nonlinear Fluids with Application to Granular Material and Powder}, school = {Universit\"{a}t Dortmund}, year = {2005}, type = {PhD Thesis}, note = {to appear}, } @PHDTHESIS{Vugrin2005, author = {.}, title = {On the Effects of Noise on Parameter Identification Optimization Problems}, school = {Faculty of the Virginia Polytechnic Institute and State University}, year = {2005}, type = {PhD Thesis}, } @PHDTHESIS{Stappert2003, author = {.}, title = {From Low--Level to Model--Based and Constructive Worst--Case Execution Time Analysis}, school = {Universit\"{a}t Paderborn}, year = {2003}, month = dec, } @MASTERSTHESIS{Becker1996, author = {.}, title = {Peano--Stoppregeln zur Approximation linearer Funktionale}, school = {Universit\"{a}t Hildesheim}, year = {1996}, } @MASTERSTHESIS{Buijssen2002, author = {.}, title = {{Numerische Analyse eines parallelen 3-D-Navier-Stokes-L\"{o}sers}}, school = {Universit\"{a}t Heidelberg}, year = {2002}, month = oct, } @MASTERSTHESIS{Wallis1999a, author = {.}, title = {Efficient Multigrid Poisson Solvers in General Coordinates on Tensorproduct meshes}, school = {Universit\"{a}t Heidelberg}, year = {1999}, } @MASTERSTHESIS{Altieri2001, author = {.}, title = {Robuste und effiziente Mehrgitter--Verfahren auf verallgemeinerten Tensorprodukt--Gittern}, school = {Universit\"{a}t Dortmund}, year = {2001}, } @MASTERSTHESIS{Christadler2003, author = {.}, title = {Patch--adaptive Relaxation als Gl\"{a}tter im Mehrgitterverfahren}, school = {IMMD10, FAU Erlangen}, year = {2003}, type = {Studienarbeit}, } @MASTERSTHESIS{Koester2004, author = {.}, title = {{Robuste Mehrgitter-Krylowraum-Techniken f\"{u}r FEM-Verfahren}}, school = {Universit\"{a}t Dortmund}, year = {2004}, type = {Diplomarbeit}, } @MASTERSTHESIS{Goeddeke2004, author = {.}, title = {{Geometrische Projektionstechniken auf Oberfl\"{a}chentriangulierungen zur numerischen Str\"{o}mungssimulation mit hierarchischen Mehrgitterverfahren}}, school = {Universit\"{a}t Dortmund}, year = {2004}, type = {Diplomarbeit}, } @MASTERSTHESIS{Acker2003a, author = {.}, title = {{Effiziente Realisierung von hierarchischen Pre- und Postprocessingmethoden in der mathematischen Str\"{o}mungssimulation}}, school = {Universit\"{a}t Dortmund}, year = {2003}, month = may, } @MANUAL{Buijssen2003, author = {.}, title = {parpp3d++ User Guide}, organization = {Universit\"{a}t Dortmund}, year = {2003}, month = jun, edition = {Release 1.0}, } @MANUAL{Buijssen2001, author = {.}, title = {Batch--oriented MPEG generation with {GMV} in background mode}, organization = {University of Dortmund}, year = {2001}, month = sep, edition = {Release 2.0}, } @MANUAL{Cederquist2003, author = {Cederquist}, title = {Version Management with CVS}, organization = {Cederqvist et al}, year = {2003}, note = {\url{http:/ / www.cvshome.org/ docs/ manual/ cvs.html}}, } @MANUAL{Ortega2003a, author = {.}, title = {The General Mesh Viewer Version 3.2}, organization = {Los Alamos National Laboratory}, year = {2003}, note = {\url{http:/ / www-xdiv.lanl.gov/ XCM/ gmv/ GMVHome.html}}, } @MANUAL{RedHat2003, author = {Red Hat}, title = {Cygwin 1.3}, year = {2003}, note = {\url{http:/ / www.cygwin.com}}, } @MANUAL{YousefSaad1999, author = {.}, title = {SPARSEKIT}, organization = {University of Minnesota}, year = {1999}, address = {\url{http:/ / www-users.cs.umn.edu/ ~saad/ software/ SPARSKIT/ sparskit.html}}, } @MANUAL{SEMATECH1997, author = {SEMATECH}, title = {The national technology roadmap for semiconductors, 1997 edition}, organization = {Semiconductor Industry Association}, year = {1997}, address = {\url{http:/ / www.sematech.org}}, } @MANUAL{Buijssen2005, author = {Buijssen, .}, title = {Batch--{}--oriented MPEG generation with {GMV} in background mode}, organization = {University of Dortmund}, year = {2005}, month = oct, edition = {Release 3.0}, } @MANUAL{Buijssen2005a, author = {Buijssen, .}, title = {parpp3d++ User Guide}, organization = {Universit\"{a}t Dortmund}, year = {2005}, month = nov, edition = {Release 1.1}, } @INCOLLECTION{DouglasHuIskandaraniKowarschikRuedeWeiss1999, author = { . and R\"{u}deWeiss}, booktitle = {Multiphase Flows and Transport in Porous Media: State of the Art}, title = {Maximizing cache memory usage for multigrid algorithms}, publisher = {Springer, Berlin}, year = {1999}, pages = {124--137}, } @INCOLLECTION{Douglas1996a, author = {.}, editor = {.}, booktitle = {Applications on Advanced Architecture Computers}, title = {{} A review of numerous parallel multigrid methods}, publisher = {SIAM}, year = {1996}, address = {Philadelphia}, pages = {187--202}, } @INCOLLECTION{DouglasHuKarlKowarschikRuedeWeiss2000b, author = {. and . and . and . and R\"{u}. and .}, editor = {. and . and .}, booktitle = {Multigrid Methods VI}, title = {Fixed and adaptive cache aware algorithms for multigrid methods}, publisher = {Springer, Berlin}, year = {2000}, address = {Berlin}, pages = {87--93}, series = {Lecture Notes in Computational Sciences}, } @INCOLLECTION{KornhuberPeriauxWidlundHoppePironneauJinchao2005, author = {\`{e}.}, editor = {. and .}, booktitle = {Lecture Notes in Computational Science and Engineering}, title = {Domain Decomposition Methods in Science and Engineering}, publisher = {Springer, Berlin}, year = {2005}, note = {ISBN 3-540-22523-4}, } @INCOLLECTION{KuzminTurek2004, author = {. and .}, editor = {.}, booktitle = {Bubbly Flows {}--{} Analysis, Modelling and Calculation}, title = {Finite Element Discretization Tools for Gas--Liquid Flows}, publisher = {Springer, Berlin}, year = {2004}, pages = {191--201}, series = {Heat and Mass Transfer}, note = {ISBN: 3-540-40791-X}, } @INCOLLECTION{RannacherBeckerTurek2003, author = {. and . and .}, editor = {J\"{a}. and .}, booktitle = {Mathematics Key Technology for the Future}, title = {An Efficient {Navier}--{Stokes} Solver for Automotive Aerodynamics}, publisher = {Springer, Berlin}, year = {2003}, pages = {43--54}, } @INBOOK{Lane1997, author = {Lane}, editor = {. and \"{u}ller}, title = {Scientific Visualization}, publisher = {IEEE Computer Society Press}, year = {1997}, pages = {-}, chapter = {Scientific Visualizaiton of Large Scale Unsteady Fluid Flows}, } @INBOOK{SardajoenvanWalsumHinPost1997, author = { and }, editor = {Nielsen and . and M\"{u}ller}, title = {Scientific Visualization}, publisher = {IEEE Computer Society Press}, year = {1997}, pages = {-}, chapter = {Particle Tracing Algorithms for 3D Curvilinear Grids}, } @INBOOK{NeittaanmaekiKorotovMartikainen2004, author = { .}, editor = {. and .}, title = {Conjugate Gradient Algorithms and Finite Element Methods}, publisher = {Springer, Berlin}, year = {2004}, pages = {171--181}, chapter = {A Posteriori Error Estimation of `Quantities of Interest` on `Quantity-Adapted` Meshes}, } @INBOOK{Carey1997a, author = {.}, editor = {. and .}, title = {Computational Grids}, publisher = {Taylor und Francis}, year = {1997}, pages = {367--402}, chapter = {7, Mesh redistribution}, } @INBOOK{Pukhnachov1994b, author = {.}, editor = {. and .}, title = {Free Boundaries in Viscous Flows}, publisher = {Springer, Berlin}, year = {1994}, pages = {67--77}, chapter = {Long wave instability of viscous liquid free surface due to anomalous Marangoni effect}, } @INBOOK{LohnerMorganZienkiewicz1986, author = {. and . and .}, editor = {. and . and .}, title = {Accuracy estimates and adaptive refinements in finite element computations}, publisher = {John Wiley \& Sons Ltd.}, year = {1986}, pages = {281--297}, chapter = {Adaptive grid refinement for the compressible Euler equations.}, } @INBOOK{BazaraaSheraliShetty1993, author = {. and . and .}, editor = {. and . and .}, title = {Nonlinear Programming--Theory and Algorithms}, publisher = {J. Wesley \& Sons, New York}, year = {1993}, volume = {2nd ed.}, pages = {353--354}, chapter = {8}, } @ARTICLE{AchdouGuermond2000, author = { .}, title = {Convergence analysis of a finite element projection/Lagrange--{Galerkin} method for the incompressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {3}, pages = {799--826}, } @ARTICLE{Acosta2001, author = {.}, title = {{Lagrange} and average interpolation over 3D anisotropic elemants}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {135}, pages = {91--109}, } @ARTICLE{AinsworthDoerfler2001, author = {\"{o}rfler}, title = {Fundamental systems of numerical schemes for linear convection--diffusion equations and their relationship to accuracy}, journal = {Computing}, year = {2001}, volume = {66}, pages = {199--229}, } @ARTICLE{AllaireClercKokh2000, author = { . and Kokh}, title = {A five--equation model for the numerical simulation of interfaces in two--phase flows}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {331}, number = {1}, pages = {1017--1022}, } @ARTICLE{AlmgrenBellCrutchfield2000, author = {. and Crutchfield}, title = {Approximate projection methods: part I. Invicid analysis}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {4}, pages = {1139--1159}, } @ARTICLE{AlvarezDericheSanchezWeickert2000, author = { .}, title = {Dense Disparity Map Estimation Respecting Image Discontinuities: A {PDE} and Scale--Space Based Approach}, journal = {INRIA}, year = {2000}, volume = {3}, number = {3874}, } @ARTICLE{AlvesPinhoOliveira2000, author = { Oliveira}, title = {Effect of a high--resolution differencing scheme om finite--volume predictions of viscoelastic flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {93}, pages = {287--314}, } @ARTICLE{AlvesPinhoOliveira2001, author = {Alves and Oliveira}, title = {The flow of viscoelastic fluids past a cylinder: finite--volume high--resolution methods}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {97}, pages = {207--232}, } @ARTICLE{AlvesPinhoOliveira2001a, author = { Oliveira}, title = {Study of steady pipe and channel flows of a single--mode Phan--Thien--Tanner fluid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {101}, pages = {55--76}, } @ARTICLE{AndersonGalaktionovPetersvandeVosseMeijer2000, author = { }, title = {Mixing of non--Newtonian fluids in time--periodic cavity flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {93}, pages = {256--286}, } @ARTICLE{AnjaliDaviKandasamy2000, author = { Kandasamy}, title = {Effects of chemical reaction, heat and mass transfer on MHD flow past a semi infinite plate}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {10}, pages = {697--701}, } @ARTICLE{ApelNicaiseSchoeberl2001, author = {. and \"{.}, title = {Crouzeix--Raviart Type Finite Elements on Anisotropic Meshes}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {193--223}, } @ARTICLE{Arany2001, author = {Arany}, title = {Numerical experiences of solving elasticity systems by PCG methods}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {42}, pages = {1025--1033}, } @ARTICLE{ArbogastCowsarWheelerYotov2000, author = { .}, title = {Mixed finite element methods on nonmatching mutliblock grids}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {4}, pages = {1295--1315}, } @ARTICLE{Babuska1972, author = {.}, title = {The Finite Element Method with {L}agrangian Multipliers}, journal = {Numerische Mathematik}, year = {1973}, volume = {20}, pages = {179--192}, } @ARTICLE{BabuskaFeistauerSolin2001, author = {. and }, title = {On One Approach to a Posteriori Error Estimates for Evolution Problems Solved by the Method of Lines}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {225--256}, } @ARTICLE{BaiFreund2001, author = {}, title = {A symmetric band lanczos process based on coupled recurrences and some applications}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {542--562}, } @ARTICLE{BairWiner1992, author = {}, title = {The high pressure high shear stress rheology of liquid lubricants}, journal = {Journal of Tribology}, year = {1992}, volume = {114}, } @ARTICLE{BalRyzhik2000, author = {}, title = {Diffusion approximation of radiative transfer problems with interfaces}, journal = {SIAM Journal on Applied Mathematics}, year = {2000}, volume = {60}, number = {6}, pages = {1887--1912}, } @ARTICLE{Bank1988, author = {.}, title = {PLTMG Users' Guide {}--{} Edition 5.0}, journal = {0}, year = {1988}, } @ARTICLE{BankChanCoughranSmith0, author = {. and . and }, title = {The Alternate--Block--Factorization Procedure for Systems of Partial Differential Equations}, journal = {0}, year = {0}, } @ARTICLE{Baensch2001, author = {.}, title = {Finite element discretization of the {Navier}--{Stokes} equations with a free capillary surface}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {203--235}, note = {DOI 10.1007/s002110000225}, } @ARTICLE{BaenschHoehn2000, author = {B\"{a}. and H\"{o}hn}, title = {Numerical treatment of the {Navier}--{Stokes} equations with slip boundary condition}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2144--2162}, } @ARTICLE{BarrettBlowey2001, author = {. and .}, title = {An improved error bound for a finite element approximation of a model for phase separation of a multi--component alloy with a concentration dependent mobility matrix}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {255--297}, } @ARTICLE{BartelsBreuerWechslerDurst2002, author = {. and .}, title = {Computational fluid dynamics applications on parallel--vector computers: computations of stirred vessel flows}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {69--97}, } @ARTICLE{BeckersDelhezDeleersnijder2001, author = {. and . and .}, title = {Some properties of generalized age--distribution equations in fluid dynamics}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {5}, pages = {1526--1544}, } @ARTICLE{Belgacem2000, author = {.}, title = {The mixed mortar finite element method for the incompressible {Stokes} problem: convergence analysis}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {4}, pages = {1085--1100}, } @ARTICLE{Belgacem2000a, author = {.}, title = {Numerical simulation of some variational inequalities arisen from unilateral contact problems by the finite elemnt methods}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {4}, pages = {1198--1216}, } @ARTICLE{BenziCullumTuma2000, author = { .}, title = {Robust approximate inverse preconditioning for the conjugate gradient method}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {4}, pages = {1318--1332}, } @ARTICLE{BenziFrommerNabbenSzyld2001, author = {. and . and Nabben and Szyld}, title = {Algebraic theory of multiplicative Schwarz methods}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {605--639}, } @ARTICLE{BenziHawsTuma2000, author = { .}, title = {Preconditioning highly indefinite and nonsymmetric matrices}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {4}, pages = {1333--1353}, } @ARTICLE{BenziTuma2000, author = {. and .}, title = {Orderings for factorized sparse approximate inverse preconditioners}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1851--1868}, } @ARTICLE{BercovierEngelman1979, author = {Bercovier and Engelman}, title = {A Finite Element for the Numerical Solution of Viscous Incompressible Flows}, journal = {J. Comput. Phys.}, year = {1979}, pages = {181--201}, } @ARTICLE{BercovierHasbaniGilonBathe0, author = { .}, title = {On a Finite Element Procedure for Non--Linear Incompressible Elasticity}, journal = {0}, year = {0}, } @ARTICLE{BercovierPironneau0, author = { .}, title = {Error Estimates for Finite Element Method Solution of the {Stokes} Problem in the Primitive Variables}, journal = {0}, year = {0}, } @ARTICLE{BerestyckiBuscaFlorent2000, author = { Busca and Florent}, title = {An inverse parabolic problem arising in finance}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {331}, number = {1}, pages = {965--969}, } @ARTICLE{BermejoInfante2000, author = {Bermejo and Infante}, title = {A multigrid algorithm for the p--Laplacian}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1774--1789}, } @ARTICLE{BernardiFietierOwens2001, author = { .}, title = {An error indicator for mortar element solutions to the {Stokes} problem}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {857--886}, } @ARTICLE{BertoluzzaKunoth2000, author = {Bertoluzza and Kunoth}, title = {Wavelet stabilization and preconditioning for domain decomposition}, journal = {IMA Journal of Numerical Analysis}, year = {2000}, volume = {20}, pages = {533--559}, } @ARTICLE{BewleyTemamZiane2000, author = {. and Ziane}, title = {Existence and uniqueness of optimal control to the {Navier}--{Stokes} equations}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, number = {1}, pages = {1007--1011}, } @ARTICLE{BialeckiFairweather2001, author = {Bialecki and Fairweather}, title = {Orthogonal spline collocation methods for partial differential equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {55--82}, } @ARTICLE{BiroVelazquez2000, author = {Biro and Velazquez}, title = {Analysis of a free boundary problem arising in bubble dynamics}, journal = {SIAM Journal on Mathematical Analysis}, year = {2000}, volume = {32}, number = {1}, pages = {142--171}, } @ARTICLE{BiscariCerciganiSlemrod2000, author = { Cercigani and Slemrod}, title = {Time--derivates and frame--invariance beyond Newtonian fluids}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {328}, number = {2}, pages = {417--422}, } @ARTICLE{BischofCarleCorlissGriewankHovland1991, author = {. and Carle and Corliss and Hovland}, title = {ADIFOR {}--{} Generating derivative codes from Fortran programs}, journal = {0}, year = {1991}, } @ARTICLE{Blum1979, author = {.}, title = {nn}, journal = {0}, year = {1979}, } @ARTICLE{BlumLiskyRannacher1992, author = {. and . and .}, title = {A Domain Splitting Algorithm for Parabolic Problems}, journal = {0}, year = {1992}, } @ARTICLE{BlumRannacher1980, author = {. and .}, title = {On the Boundary Value Problem of the Biharmonic Operator on Domains with Angular Corners}, journal = {Mathematical Methods in the applied Sciences}, year = {1980}, pages = {556--581}, } @ARTICLE{BlumStuttmeier1999, author = {. and .}, title = {An Adaptive Finite Element Discretisation for a Simplified Signorini Promblem}, journal = {0}, year = {1999}, } @ARTICLE{Boffi2000, author = {.}, title = {Fortin operator and discrete compactness for edge elements}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {229--246}, } @ARTICLE{Boglaev2001, author = {.}, title = {Domain decomposition for a singularly pertubed parabolic problem with a convecton--dominated term}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {134}, pages = {283--299}, } @ARTICLE{BoehmeRubart1989, author = {}, title = {Non--Newtonian flow analysis by finite elements}, journal = {Fluid Dyn.}, year = {1989}, pages = {147--158}, } @ARTICLE{BoehmeRubart1991, author = {B\"{}, title = {Numerical Simulation of Shear--Thinning Flow Problems in Mixing Vessels}, journal = {Theoretical and Computational Fluid Dynamics}, year = {1991}, pages = {95--115}, } @ARTICLE{BoehmWechslerSchaefer1998, author = {B\"{ and Sch\"{a}.}, title = {A Parallel Moving Grid Multigrid Method For Flow Simulation In Rotor--Stator Configurations}, journal = {Int. J. Numer. Methods Eng.}, year = {1998}, volume = {42}, pages = {175--189}, } @ARTICLE{Bollhoefer2000, author = {Bollh\"{.}, title = {A robust ILU based on monitoring the growth of the inverse factors}, journal = {0}, year = {2000}, } @ARTICLE{BollhoeferMehrmann1998, author = {Bollh\"{. and .}, title = {Nested divide and conquer methods for the solution of large sparse linear systems}, journal = {0}, year = {1998}, month = nov, } @ARTICLE{BollhoeferMehrmann1999, author = {Bollh\"{. and .}, title = {Algebraic multilevel methods and sparse approximate inverses}, journal = {0}, year = {1999}, } @ARTICLE{BollhoeferSaad2000, author = {Bollh\"{. and .}, title = {ILUs and factorized approximate inverses are strongly related. Part I: overview of results}, journal = {0}, year = {2000}, month = mar, } @ARTICLE{BollhoeferSaad2000a, author = {. and .}, title = {ILUs and factorized approximate inverses are strongly related. Part II: applications to stabilization}, journal = {0}, year = {2000}, month = apr, } @ARTICLE{Boerm2001, author = {B\"{o}rm}, title = {Tensor product multigrid for Maxwell's equation with aligned anisotropy}, journal = {Computing}, year = {2001}, volume = {66}, pages = {321--342}, } @ARTICLE{Boyer2002, author = {Boyer}, title = {A theoretical and numerical model for the study of incompressible mixture flows}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {41--68}, } @ARTICLE{BraessBloemer1990, author = {. and Bl\"{o}mer}, title = {A Multigrid Method for a Parameter Dependent Problem in Solid Mechanics}, journal = {Numerische Mathematik}, year = {1990}, pages = {747--761}, } @ARTICLE{BraessDahmen1999, author = {. and .}, title = {A cascadic multigrid algorithm for the {Stokes} equations}, journal = {Numerische Mathematik}, year = {1999}, pages = {179--191}, } @ARTICLE{BraessPeisker1986, author = {. and Peisker}, title = {On the Numerical Solution of the Biharmonic Equation and the Role of Squaring Matrices for Preconditioning}, journal = {IMA Journal of Numerical Analysis}, year = {1986}, pages = {393--404}, } @ARTICLE{BraessPeisker1987, author = { Peisker}, title = {A conjugate gradient method and a multigrid algorithm for Moreley`s finite element approximation of the biharmonic equation}, journal = {0}, year = {1987}, } @ARTICLE{BraessSarazin1997, author = { Sarazin}, title = {An efficient smoother for the {Stokes} problem}, journal = {Applied Numerical Mathematics}, year = {1997}, pages = {19. Mrz}, } @ARTICLE{BraessVerfuerth0, author = {. and .}, title = {Multi--Grid Methods for Non--Conforming Finite Element Methods}, journal = {0}, year = {0}, } @ARTICLE{BrambleThomee1974, author = {. and Thomee}, title = {Interior Maximum Norm Estimates For Some Simple Finite Element Methods}, journal = {RAIRO}, year = {1974}, } @ARTICLE{BrambleZhang2000, author = {. and .}, title = {Uniform Convergence of the Multigrid {V}--Cycle for an Anisotropic Problem}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {234}, pages = {453--470}, } @ARTICLE{BrandtDendyRuppel0, author = {Brandt and Ruppel}, title = {The Multigrid Method for Semi--Implicit Hydrodynamics Codes}, journal = {0}, year = {0}, } @ARTICLE{Brenner1990, author = {Brenner}, title = {A Nonconforming Multigrid Method For The Stationary {Stokes} Equations}, journal = {Math. Comput.}, year = {1990}, volume = {55}, number = {192}, pages = {411--137}, } @ARTICLE{Brenner2000, author = {Brenner}, title = {Lower bounds for two--level additive schwarz preconditioners with small overlap}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1657--1669}, } @ARTICLE{Brenner1989, author = {Brenner}, title = {Multigrid Methods for Nonconforming Finite Elements}, journal = {0}, year = {1989}, } @ARTICLE{Brenner1996, author = {Brenner}, title = {Two--Level Additive Schwarz Preconditioners For Nonconforming Finite Element Methods}, journal = {0}, year = {1996}, } @ARTICLE{Brenner1992, author = {Brenner}, title = {A Multigrid Algorithm for the Lowest Order Raviart--Thomas Mixed Triangular Finite Element Method}, journal = {0}, year = {1992}, } @ARTICLE{Brenner1994, author = {Brenner}, title = {A Nonconforming Mixed Multigrid Method For The Pure Traction Problem In Planar Linear Elasticity}, journal = {0}, year = {1994}, } @ARTICLE{Brenner1996a, author = {Brenner}, title = {Multigrid Methods For Parameter Dependent Problems}, journal = {0}, year = {1996}, } @ARTICLE{Brenner1996b, author = {Brenner}, title = {Preconditioning Complicated FEM's By Simple FEM's}, journal = {0}, year = {1996}, } @ARTICLE{Brenner1993, author = {Brenner}, title = {A Nonconforming Multigrid Method For The Pure Displacement Problem In Planar Linear Elasticity}, journal = {0}, year = {1993}, } @ARTICLE{BrennerSung1992, author = { Sung}, title = {Linear Finite Element Methods For Planar Linear Elasticity}, journal = {0}, year = {1992}, } @ARTICLE{BrennerSung2000, author = {Brenner and Sung}, title = {Lower bounds for nonoverlapping domain decomposition preconditioners in two dimensions}, journal = {Mathematics of Computation}, year = {2000}, volume = {69}, number = {232}, pages = {1319--1339}, } @ARTICLE{Brezzi1974, author = {.}, title = {On the Existence, Uniqueness and Approximation of Saddle--Point Problems Arising from {L}agrangian Multipliers}, journal = {RAIRO}, year = {1974}, volume = {8}, number = {R-2}, pages = {129--151}, } @ARTICLE{BrezziArnold0, author = {. and .}, title = {Mixed and Nonconforming Finite Element Methods: Implementation, Postprocessing and Error Estimates}, journal = {0}, year = {0}, } @ARTICLE{BrezziFortin2001, author = {. and .}, title = {A Minimal Stabilisation Procedure for Mixed Finite Element Methods}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {457--491}, } @ARTICLE{BrezziMarini2000, author = {. and Marini}, title = {Error Estimates for the Three--Field Formulation with Bubble Stabilization}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {235}, pages = {911--934}, } @ARTICLE{BridsonTang2001, author = {}, title = {Multiresolution approximate inverse preconditioners}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {463--479}, } @ARTICLE{BrileyMcDonald2001, author = {}, title = {An overview and generalization of implicit {Navier}--{Stokes} algorithms and approximate factorization}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {807--828}, } @ARTICLE{BrownWoodward2001, author = {. and Woodward}, title = {Preconditioning strategies for fully implicit radiation diffusion with material--energy transfer}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {499--516}, } @ARTICLE{BuddPiggott2001, author = {}, title = {The geometric integration of scale--invariant ordinary and partial differential equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {399--422}, } @ARTICLE{BurrageBurrageMitsui2000, author = { Mitsui}, title = {Numerical solutions of stochastic differential equations {}--{} implementaton and stability issues}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {125}, pages = {171--182}, } @ARTICLE{BuzdinLogashenko2000, author = {. and .}, title = {Incomplete block triangular decompositions of order l}, journal = {Computing}, year = {2000}, volume = {64}, pages = {69--95}, } @ARTICLE{CaiLeeManteuffelMcCormick2000, author = {. and . and . and .}, title = {First--order system least squares for linear elasticity: numerical results}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1706--1727}, } @ARTICLE{CaiLeeManteuffelMcCormick2000a, author = { .}, title = {First--order system least squares for linear elasticity: further results}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1728--1739}, } @ARTICLE{CaldererLiu2000, author = {}, title = {Liquid crystal flow: dynamic and static configurations}, journal = {SIAM Journal on Applied Mathematics}, year = {2000}, volume = {60}, number = {6}, pages = {1925--1949}, } @ARTICLE{CanutoMasson2000, author = {. and .}, title = {Stabilized wavelet approximations of the {Stokes} problem}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {236}, pages = {1397--1416}, } @ARTICLE{CaolaJooArmstrongBrown2001, author = {. and .}, title = {Highly parallel time integration of viscoelastic flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {100}, pages = {191--216}, } @ARTICLE{CarstensenFunken2000, author = { .}, title = {A posteriori error control in low--order finite element discretizations of incompressible sationary flow problems}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {236}, pages = {1353--1381}, } @ARTICLE{CarstensenPlechac2000, author = {. and Plechac}, title = {Numrical analysis of compatible phase transitions in elastic solids}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {6}, pages = {2061--2081}, } @ARTICLE{CarstensenRoubicek2000, author = { Roubicek}, title = {Numerical approximation of young measures in non--convex variational problems}, journal = {Numerische Mathematik}, year = {2000}, volume = {84}, pages = {395--415}, } @ARTICLE{CarvalhoGiraudTallec2001, author = { . and .}, title = {Algebraic two--level preconditioners for the shur complement method}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {1987--2005}, } @ARTICLE{CasadeiHalleux1995, author = {}, title = {An algorithm for permanent fluid--structure interaction in explicit transient dynamics}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1995}, volume = {128}, pages = {231--289}, note = {Ispra (It)}, } @ARTICLE{Caughey2001, author = {Caughey}, title = {Implicit multigrid computation of unsteady flows past cylinders of square cross--section}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {939--960}, } @ARTICLE{CebralLoehner1997, author = { .}, title = {Conservative Load Projektion and Tracking for Fluid--Structure Problems}, journal = {IAIAA Journal}, year = {1997}, volume = {35}, number = {4}, pages = {687--692}, month = apr, } @ARTICLE{ChanWan2000, author = {. and Wan}, title = {Robust multigrid methods for nonsmooth coefficient elliptic linear systems}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {323--352}, } @ARTICLE{ChauviereOwens2000, author = { .}, title = {How accurate is your solution? Error indicators for viscoelastic flow calculations}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {1 bis 33}, } @ARTICLE{Chen2000, author = {.}, title = {Formulations and numerical methods of the black oil model in porous media}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {2}, pages = {489--514}, } @ARTICLE{ChenFriedman2000, author = {. and Friedman}, title = {A free boundary problem arising in a model of wound healing}, journal = {SIAM Journal on Mathematical Analysis}, year = {2000}, volume = {32}, number = {4}, pages = {778--800}, } @ARTICLE{Cheng2000, author = {.}, title = {On the nonlinear inexact Uzawa algorithm for saddle--point problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {6}, pages = {1930--1934}, } @ARTICLE{Cheng2000a, author = {.}, title = {A discontinuous quadratic finite element method for one--dimensional fiffusion problems}, journal = {Computers and Mathematics with Applications}, year = {2000}, volume = {40}, pages = {921--925}, } @ARTICLE{Chibi2001, author = {Chibi}, title = {Defect correction and domain decomposition for second--order boundary value problems}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {130}, pages = {41--51}, } @ARTICLE{ChizhonkovOlshanskii2000, author = { .}, title = {On the domain geometry dependence of the LBB condition}, journal = {Mathematical Modelling and Numerical Analysis}, year = {2000}, volume = {34}, number = {5}, pages = {935--951}, } @ARTICLE{ChouTang2000, author = {. and Tang}, title = {Conservative P1 conforming and nonconforming {Galerkin} FEMs: effective flux evaluation via a nonmixed method approach}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {2}, pages = {660--680}, } @ARTICLE{ChristodoulouLindblad2000, author = {.}, title = {On the motion of the free surface of a liquid}, journal = {Communications on Pure and Applied Mathematics}, year = {2000}, volume = {LIII}, number = {12}, } @ARTICLE{ChunKailathLev-Ari1987, author = { Lev-Ari}, title = {Fast parallel algorithms for QR and triangular factorization}, journal = {SIAM Journal on Scientific Computing}, year = {1987}, volume = {8}, number = {6}, } @ARTICLE{CiarletRaviart0, author = {. and .}, title = {General {Lagrange} and Hermite Interpolation in $R^n$ with Applications to Finite Element Methods}, journal = {0}, year = {0}, } @ARTICLE{ClearyFalgoutVanEmdenJonesManteuffelMcCormickMirandaRuge2000, author = {. and . and . and Miranda and .}, title = {Robustness and scalabitlity of algebraic multigrid}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1886--1908}, } @ARTICLE{Cockburn2001, author = {.}, title = {Devising discontinuous {Galerkin} methods for non--linear hyperbolic conservation laws}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {187--204}, } @ARTICLE{CodinaBlasco2000, author = {. and Blasco}, title = {Analysis of a pressure--stabilized finite element approximation of the stationery {Navier}--{Stokes} equations}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {59--81}, } @ARTICLE{CodinaCervera1996, author = {. and Cervera}, title = {Block--iterative algorithms for nonlinear coupled problems}, journal = {Advanced Computational Methods in Structural mechanics}, year = {1996}, } @ARTICLE{CohenDahmenDevore2000, author = {. and . and Devore}, title = {Adaptive wavelet methods for elliptic operator equations: convergence rates}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {233}, pages = {27--75}, } @ARTICLE{CrouzeixRaviart1974, author = {. and .}, title = {Conforming and Nonconforming Finite Element Methods for Solving the Stationary {Stokes} Equations}, journal = {0}, year = {1974}, } @ARTICLE{CummingsRubinstein2000, author = {Cummings and Rubinstein}, title = {A fluid and heat flow problem arising in the soup--canning industry}, journal = {Q. J. Mech. Appl. Math.}, year = {2000}, volume = {53}, pages = {583--608}, } @ARTICLE{DahmenKunoth2001, author = {. and Kunoth}, title = {Appending boundary conditions by {Lagrange} multipliers: analysis of the LBB condition}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {009--042}, } @ARTICLE{DalPassoGiacomelliShishkov2001, author = {. and . and .}, title = {The thin film equation with nonlinear diffusion}, journal = {Communications in Partial Differential Equations}, year = {2001}, volume = {26}, number = {9}, pages = {1509--1557}, } @ARTICLE{Dauge0, author = {Dauge}, title = {Stationary {Stokes} and {Navier}--{Stokes} systems on two--{} or three--dimensional domains with $corners^K$. Part1: Linearized equations}, journal = {0}, year = {0}, } @ARTICLE{Dazevedo2000, author = {.}, title = {Are bilinear quadrilaterials better than linear triangles?}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {198--217}, } @ARTICLE{DeakinRasmussen2001, author = { Rasmussen}, title = {Nonreflecting boundary condition for the Helmholtz equation}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {41}, pages = {307--318}, } @ARTICLE{DengPiquetVasseurVisonneau2001, author = {. and . and Visonneau}, title = {A new fully coupled method for computing turbulent flows}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {445--472}, } @ARTICLE{DesjardinsEsteban2000, author = { Esteban}, title = {On weak solutions for fluid--rigid structure interaction: compressible and incompressible models}, journal = {Communications in Partial Differential Equations}, year = {2000}, volume = {25}, number = {7}, pages = {1399--1413}, } @ARTICLE{DiazMacedo1989, author = {. and Macedo}, title = {Fully vectorizable block preconditionings with approximate inverses for non--symmetric systems of equations}, journal = {Int. J. Numer. Methods Eng.}, year = {1989}, volume = {27}, pages = {501--522}, } @ARTICLE{DiewaldPreusserRumpfStrzodka2000, author = { .}, title = {Diffusion Models and Their Accelerated Solution in Image and Surface Processing}, journal = {Acta Mathematica}, year = {2000}, volume = {LXX}, pages = {15--31}, } @ARTICLE{Dobrowolski0, author = {.}, title = {Stabilit\"{a}t und Konvergenz der Methode der finiten Elemente in der $L--\infty$--Norm f\"{u}r n--dimensionale Randwertprobleme}, journal = {0}, year = {0}, } @ARTICLE{DobrowolskiThomas0, author = {. and .}, title = {Über die Struktur diskret--divergenzfreier Finiter Elemente zur numerischen Approximation der {Navier}--{Stokes}--Gleichung}, journal = {0}, year = {0}, } @ARTICLE{Donea1983, author = {.}, title = {Arbitrary {L}agrangian--{E}ulerian Finite Element Methods}, journal = {Computational Methods for Transient Analysis}, year = {1983}, pages = {473--516}, } @ARTICLE{DongarraJohnsson1987, author = {. and Johnsson}, title = {Solving banded systems on a parallel processor}, journal = {Parallel Computing}, year = {1987}, pages = {219--246}, } @ARTICLE{DonggarraEijkhout2000, author = {. and Eijkhout}, title = {Numerical linear algebra algorithms and software}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {489--514}, } @ARTICLE{DoerflerWilderotter2000, author = {}, title = {An Adaptive Finite Element Method for a Linear Elliptic Equation with Variable Coefficients}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {7}, pages = {481--491}, } @ARTICLE{DouglasDupontWahlbin1974, author = {. and }, title = {The Stability in $L^q$ of the $L^2$--Projection into Finite Element Function Spaces}, journal = {Numerische Mathematik}, year = {1974}, } @ARTICLE{DouPhan-Thien1998, author = { .}, title = {Parallelization of an unstructured finite volume code with PVM: viscoelastic flow around a cylinder}, journal = {J. Non-Newtonian Fluid Mech.}, year = {1998}, volume = {77}, pages = {21--51}, } @ARTICLE{Duff2000, author = {.}, title = {The impact of high--performance computing in the solution of linear systems: trends end problems}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {515--530}, } @ARTICLE{DuranHervella-NietoLibermanRodriguezSolomin2000, author = {. and Solomin}, title = {Finite element analysis of the vibration problem of a plate coupled with a fluid}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, pages = {591--616}, } @ARTICLE{DuttoLepageHabashi2000, author = {Dutto and Habashi and Lepage}, title = {Effect of the storage format of sparse linear systems on parallel {CFD} computations}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2000}, volume = {188}, pages = {441--453}, } @ARTICLE{Eagles2001, author = {Eagles}, title = {On the stability of flow between a rotating circular cylinder and an outer non--circular cylinder}, journal = {Q. J. Mech. Appl. Math.}, year = {2001}, volume = {54}, number = {4}, pages = {641--653}, } @ARTICLE{EbertShaw2001, author = {. and .}, title = {Minimally Immersive Flow Visualization}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2001}, volume = {7}, number = {4}, } @ARTICLE{EbnerZenger1999, author = {. and .}, title = {A distributed functional framework for recursive finite element simulations}, journal = {Parallel Computing}, year = {1999}, pages = {813--826}, } @ARTICLE{EfendievHouWu2000, author = {Efendiev and . and Wu}, title = {Convergence of nonconforming multiscale finite element method}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {3}, pages = {888--910}, } @ARTICLE{ErvinLaytonMaubach2000, author = {. and .}, title = {Adaptive defect--correction methods for viscous incompressible flow problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {4}, pages = {1165--1185}, } @ARTICLE{EsselaouiNajibRamadaneZine2000, author = { .}, title = {Analyse numerique et approche decouplee d un probleme d ecoulements de fluides non newtoniens de type Phan--Thien--Tanner: Constraintes discontinues}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, pages = {931--936}, } @ARTICLE{EwingLinWang2000, author = {. and Lin and Wang}, title = {A numerical approximation of nonfickian flows with mixing length growth in porous media}, journal = {Acta Mathematica}, year = {2000}, volume = {LXX}, number = {1}, pages = {75--84}, } @ARTICLE{EwingWang2001, author = {. and Wang}, title = {A summary of numerical methods for time--dependent advection--dominated partial differential equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {423--445}, } @ARTICLE{FaragoKaratson2001, author = {.}, title = {The gradient--finite element method for elliptic problems}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {42}, pages = {1043--1053}, } @ARTICLE{FaragoTarvainen2001, author = {Farago and .}, title = {Qualitative analysis of matrix splitting methods}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {42}, pages = {1055--1067}, } @ARTICLE{FarhloulNicaisepaquet2000, author = {. and . and Paquet}, title = {Refined mixed finite element method for the Boussinesq equations in polygonal domaind}, journal = {IMA Journal of Numerical Analysis}, year = {2000}, volume = {331}, number = {I}, pages = {143--148}, } @ARTICLE{FarhoulNicaisepaquet2001, author = {Farhoul and . and Paquet}, title = {A refined mixed finite element method for the boussinesq equations in polygonal domains}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {525--551}, } @ARTICLE{Feng2000, author = {.}, title = {Analysis of finite element methods and domain decomposition algorithms for a fluid--solid interaction problem}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {4}, pages = {1312--1336}, } @ARTICLE{FlemmingOlmsteadDavis2000, author = { .}, title = {Shear localization with an Arrhenius flow law}, journal = {SIAM Journal on Applied Mathematics}, year = {2000}, volume = {60}, number = {6}, pages = {1867--1886}, } @ARTICLE{FoiasManleyRosaTemam2001, author = { .}, title = {Cascade of energy in turbulent flows}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2001}, volume = {332}, pages = {509--514}, } @ARTICLE{FontelosFriedman2000, author = { Friedman}, title = {The flow of a class of Oldroyd fluids around a re--entant corner}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {185--198}, } @ARTICLE{FormaggiaGerbeauNobileQuarteroni2000, author = {. and . and . and .}, title = {On the Coupling of 3D and 1D {Navier}--{Stokes} Equations for Flow Problems in Compliant Vessels}, journal = {0}, year = {2000}, } @ARTICLE{FrankVuik2001, author = {. and .}, title = {On the construction of deflation--based preconditioners}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {442--462}, } @ARTICLE{FriedrichHuettlManhartWagner2001, author = { }, title = {Direct numerical simulation of incompressible turbulent flows}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {555--579}, } @ARTICLE{FrommerSzyld2000, author = { Szyld}, title = {On asynchronous iterations}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {201--216}, } @ARTICLE{GaldiLayton0, author = { .}, title = {Approximation of the larger eddies in fluid motion II: A model for space--filtered flow}, journal = {0}, year = {0}, } @ARTICLE{GallagherIbrahimMajdoub2000, author = {. and Majdoub}, title = {Solutions axisymétriques des équations de {Navier}--{Stokes}}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, pages = {791--794}, } @ARTICLE{GanderNataf2000, author = {. and .}, title = {Ailu for Helmholtz problems: a new preconditioner based on an analytic factorization}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {331}, number = {I}, pages = {261--266}, } @ARTICLE{GarbeyKuznetsovVassilevski2000, author = { . and .}, title = {A parallel Schwarz method for a convection--diffusion problem}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {3}, pages = {891--916}, } @ARTICLE{GarckeGriebelThess2001, author = { . and .}, title = {Data Mining with Sparse Grids}, journal = {Computing}, year = {2001}, volume = {67}, pages = {225--253}, } @ARTICLE{GarckePreusserRumpfTeleaWeikardvanWijk2001, author = { . and WeikardvanWijk}, title = {A phase field model for continuous clustering on vector fields}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2001}, volume = {7}, number = {3}, } @ARTICLE{GaticaHeuer2001, author = {. and Heuer}, title = {An expanded mixed finite element approach via a dual--dual formulation and the minimum residual method}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {132}, pages = {371--385}, } @ARTICLE{GaugerLeinenYserentant2000, author = { .}, title = {The finite mass method}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {6}, pages = {1768--1799}, } @ARTICLE{Geib2000, author = {Geib}, title = {Thermodynamische und str\"{o}mungstechnische untersuchungen an gashochdruckleitungen}, journal = {Gas Erdgas}, year = {2000}, volume = {141}, } @ARTICLE{GeorgeHuangTangWu2000, author = {. and . and . and .}, title = {Numerical Simulation of Unsteady Incompressible Flow (Re$<$9500) on the Cuvilinear Half--Staggered Mesh}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2331--2351}, } @ARTICLE{GerritsmaPhillips2001, author = {Gerritsma and Phillips}, title = {On the use of characteristic variables in viscoelastic flow problems}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {66}, pages = {127--147}, } @ARTICLE{GiraultGlowinskiLopezVila2001, author = {. and . and . and Vila}, title = {A boundary multiplier/fictituous domain method for steady incompressible {Navier}--{Stokes} equations}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {75--103}, } @ARTICLE{GlimmGroveLiTan2000, author = {. and }, title = {Robust computational algorithms for dynamic interface tracking in three deimensions}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2240--2256}, } @ARTICLE{GlowinskiPanPeriaux1994, author = {. and Pan and .}, title = {A fictitious domain method for external incompressible viscous flow modeled by {Navier}--{Stokes} equations}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1994}, volume = {112}, pages = {133--148}, } @ARTICLE{GlowinskiPanPeriaux1997, author = { Pan and .}, title = {A {Lagrange} Multiplier/fictitiuos domain method for the numerical simulation of incompressible viscous flow around moving rigid bodies: (I) case where the rigid body motions are known a priori}, journal = {C. R., Math., Acad. Sci. Paris}, year = {1997}, pages = {361--369}, } @ARTICLE{GlowinskiPanPeriaux1998, author = {. and Pan and .}, title = {Distributed {Lagrange} multiplier methods for incompressible viscous flow around moving rigid bodies}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1998}, volume = {151}, pages = {181--194}, } @ARTICLE{GolubvanderVorst2000, author = {. and , .}, title = {Eigenvalue computation in the 20th century}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {35--65}, } @ARTICLE{GolubYe1999, author = {}, title = {Inexact Preconditioned Conjugate Gradient Method with Inner--Outer Iteration}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {21}, number = {4}, pages = {1305--1320}, } @ARTICLE{GopalakrishnanPasciak2000, author = { Pasciak}, title = {Multigrid for the Mortar finite element method}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {3}, pages = {1029--1052}, } @ARTICLE{GottliebHesthaven2001, author = { Hesthaven}, title = {Spectral methods for hyperbolic problems}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {83--131}, } @ARTICLE{GrilletShaqfehKhomami2000, author = { .}, title = {Observations of elastic instabilities in lid--driven cavity flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, pages = {15--35}, } @ARTICLE{GruenRumpf2000, author = { .}, title = {Nonnegativity preserving convergent schemes for the thin film equation}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {113--152}, } @ARTICLE{Guermond2001, author = {.}, title = {Subgrid stabilization of {Galerkin} approximations of linear monotone operators}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {165--197}, } @ARTICLE{GuermondLu2000, author = {. and Lu}, title = {A domain decomposition method for simulating advection dominated, external incompressible viscous flows}, journal = {Computers and Fluids}, year = {2000}, volume = {29}, pages = {525--546}, } @ARTICLE{GunzburgerLee2000, author = {. and Lee}, title = {An optimization--based domain decomposition method for the {Navier}--{Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {5}, pages = {1455--1480}, } @ARTICLE{GunzburgerManservisi2000, author = {. and .}, title = {Analysis and approximation of the velocity tracking problem for {Navier}--{Stokes} flows with distributed control}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {5}, pages = {1481--1512}, } @ARTICLE{GutknechtRozloznik2001, author = {}, title = {Residual smoothing techniques: do they improve the limiting accuracy of iterative solvers?}, journal = {BIT Numerical Mathematics}, year = {2001}, volume = {41}, number = {1}, pages = {086--114}, } @ARTICLE{HackbuschKhoromskij2000, author = { Khoromskij}, title = {A sparse H--matrix arithmetic. Part 2: Application to multi--dimensional problems}, journal = {Computing}, year = {2000}, volume = {64}, pages = {21--47}, } @ARTICLE{Hadjidimos2000, author = {Hadjidimos}, title = {Succesive overrelaxations (SOR) and related methods}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {177--199}, } @ARTICLE{Halpern2001, author = {Halpern}, title = {A spectral method for the {Stokes} problem in three--dimensional unbounded domains}, journal = {Mathematics of Computation}, year = {2001}, volume = {70}, number = {236}, pages = {1417--1436}, } @ARTICLE{HauserMrozBischiGroeller2001, author = {. and Gr\"{o}ller}, title = {Two--level volume rendering}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2001}, volume = {7}, number = {3}, } @ARTICLE{HayderIerotheouKeyes0, author = { .}, title = {Three parallel programming paradigms: comparisons on an archetypal {PDE} computation}, journal = {0}, year = {0}, } @ARTICLE{Heil1998, author = {Heil}, title = {{Stokes} Flow In An Elastic Tube {}--{} A Large--Displacemant Fluid--Structure Interaction Problem}, journal = {International Journal for Numerical Methods in Fluids}, year = {1998}, volume = {28}, pages = {243--265}, note = {Cambridge, Massachusetts}, } @ARTICLE{Helmy2000, author = {Helmy}, title = {On unsteady magnetohydrodynamic flow of viscous conducting fluid}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {10}, pages = {665--680}, } @ARTICLE{HemkerShishkinShishkina2001, author = {. and . and Shishkina}, title = {High--order time--accurate parallel schemes for parabolic singularly pertubed problems with convection}, journal = {Computing}, year = {2001}, volume = {66}, pages = {139--161}, } @ARTICLE{HendricksonKolda2000, author = { .}, title = {Partitioning rectangular and structurally unsymmetric sparse matrices for parallel processing}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2048--2072}, } @ARTICLE{HensonVassilevski2001, author = {. and .}, title = {Element--free AMGe: General algorithms for computing interpolation weights in AMG}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {629--650}, } @ARTICLE{HillSueli2000, author = { .}, title = {Approximation of the global attractor for the incompressible {Navier}--{Stokes} equations}, journal = {IMA Journal of Numerical Analysis}, year = {2000}, volume = {20}, pages = {633--667}, } @ARTICLE{HinterbergerScherzer2001, author = {Hinterberger and .}, title = {Models for image interpolation based on the optical flow}, journal = {Computing}, year = {2001}, volume = {66}, pages = {231--247}, } @ARTICLE{HirtNichols1981, author = {Hirt and Nichols}, title = {Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries}, journal = {J. Comput. Phys.}, year = {1981}, volume = {39}, pages = {201--225}, } @ARTICLE{Hochstenbach2001, author = {.}, title = {A {Jacobi}--Davidson type SVD method}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {606--628}, } @ARTICLE{HoffmannSchatzWahlbinWittum2001, author = {. and .}, title = {Asymtotically exact a posteriori estimators for the pointwise gradient error on each element in irregular meshes. Part 1: a smooth problem and globally quasi--uniform meshes}, journal = {Mathematics of Computation}, year = {2001}, volume = {70}, number = {235}, pages = {897--909}, } @ARTICLE{Holzbecher2001, author = {Holzbecher}, title = {On the relevance of oscillatory convection regimes in porous media {}--{} review and numerical experiments}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {189--209}, } @ARTICLE{HoustonSueli2000, author = {. and .}, title = {Adaptive {Lagrange}--{Galerkin} methods for unsteady convection--diffusion problems}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {233}, pages = {77--106}, } @ARTICLE{Hu1996, author = {Hu}, title = {Direct Simulation of Flows of Solid--Liquid Mixtures}, journal = {International Journal of Multiphase Flow}, year = {1996}, volume = {22}, number = {2}, pages = {335--352}, } @ARTICLE{HuangGreengard2000, author = { Greengard}, title = {A fast direct solver for elliptic partial differential equations on adaptively refined meshes}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1551--1566}, } @ARTICLE{HughesLiuZimmermann1981, author = {Hughes, . and Zimmermann}, title = {{L}agrangian--{E}ulerian Finite Element Formulation for Incompressible Viscous Flows}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1981}, volume = {29}, pages = {329--349}, } @ARTICLE{HughesMazzeiOberai2001, author = {Hughes, . and Mazzei and Oberai}, title = {The multiscale formulation of large eddy simulation: Decay of homogeneous isotropic turbulence}, journal = {Physics of Fluids}, year = {2001}, volume = {13}, number = {2}, pages = {505--512}, } @ARTICLE{HughesOberaiMazzei2001, author = {Hughes, . and Oberai and Mazzei}, title = {Large eddy simulation of turbulent channel flows by the variational multiscale method}, journal = {Physics of Fluids}, year = {2001}, volume = {13}, number = {6}, pages = {1784--1799}, } @ARTICLE{HughsonDodge2000, author = { Dodge}, title = {The effect of cavity wall irregularities on seepage exclusion from horizontal cylindrical underground openings}, journal = {Journal Hydrology}, year = {2000}, volume = {228}, pages = {206--214}, } @ARTICLE{HuhtanenNevalinna2000, author = {. and Nevalinna}, title = {Minimal decomositions and iterative methods}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, pages = {257--281}, } @ARTICLE{Hui0, author = {Hui}, title = {Solutions analytiques de quelques equations aux derivees pertielles en mecanique des fluides}, journal = {0}, year = {0}, } @ARTICLE{HuigolMena2000, author = {Huigol and Mena}, title = {On the time estimate for start--up of pipe flows in a Bingham fluid {}--{} a proof of the result due to Glowinski, Lions and Tremolieres}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, pages = {113--118}, } @ARTICLE{HuettlFriedrich2001, author = { .}, title = {Direct numerical simulation of turbulent flows in curved and helically coiled pipes}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {591--605}, } @ARTICLE{HysomPothen2001, author = { .}, title = {A scalable parallel algorithm for incomplete factor preconditioning}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2194--2215}, } @ARTICLE{IliescuJohnLaytonMatthiesTobiska2000, author = {Iliescu and . and . and . and .}, title = {An assessment of models in large eddy simulation}, journal = {0}, year = {2000}, } @ARTICLE{JaegerMikelicNeuss2001, author = {J\"{. and .}, title = {Asymptotic analysis of the laminarviscous flow over a porous bed}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2006--2028}, } @ARTICLE{Johannsen2000, author = {.}, title = {A robust smoother for convection--diffusion problems with {} closed characteristics}, journal = {Computing}, year = {2000}, volume = {65}, pages = {203--225}, } @ARTICLE{JohnLayton2002, author = {. and .}, title = {Analysis of numerical errors in large eddy simulation}, journal = {0}, year = {2002}, } @ARTICLE{JohnLaytonIliescuMatthiesTobiska2000, author = {. and . and Iliescu and . and .}, title = {An assessment of models in large Eddy simulation}, journal = {0}, year = {2000}, } @ARTICLE{JohnMatthiesMitkovaTobiskaVassilevski2000, author = { .}, title = {A Comparison of three solvers for the incompressible Navier.Stokes equations}, journal = {Notes on Numerical Fluid Mechanics}, year = {2000}, } @ARTICLE{JohnsonTezduyar1996, author = { .}, title = {Simulation of Multiple spheres falling in a liquid--filled tube}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1996}, volume = {134}, pages = {351--373}, } @ARTICLE{JoungPhan-ThienFan2001, author = { . and .}, title = {Direct simulation of flexible fibers}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {99}, pages = {001--036}, } @ARTICLE{JuarezScottMetcalfeBagheri2000, author = {Juarez and Scott and Metcalfe and Bagheri}, title = {Direct simulation of freely rotating cylinders in viscous flows by hing--odrer finite element methods}, journal = {Computers and Fluids}, year = {2000}, volume = {29}, pages = {547--582}, } @ARTICLE{JunkKlar2000, author = {. and .}, title = {Discretizations for the incompressible {Navier}--{Stokes} equations based on the latice Boltzmann method}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {001--019}, } @ARTICLE{JuvignyRyan2001, author = { .}, title = {Preconditioners for domain decomposition methods}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {42}, pages = {1143--1155}, } @ARTICLE{Karaa2000, author = {.}, title = {A block preconditioner for the convection--diffusion equation}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, pages = {851--856}, } @ARTICLE{KaerkkaeinenToivanen2001, author = {K\"{a}rkk\"{a}inen and Toivanen}, title = {Building blocks for odd--even multigrid with applications to reduced systems}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {131}, pages = {015--033}, } @ARTICLE{KaySilvester1999, author = {. and .}, title = {A Posteriori Error Estimation for Stabilized Mixed Approximations of the {Stokes} Equations}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {21}, number = {4}, pages = {1321--1336}, } @ARTICLE{KaySilvester2001, author = {. and .}, title = {The Reliability of Local Error Estimators for Convection--Diffusion Equations}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {107--122}, } @ARTICLE{KelloggLiu2000, author = {. and Liu}, title = {The analysis of an finite element method for the {Navier}--{Stokes} equations with compressibility}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {153--170}, } @ARTICLE{Khayat2000, author = {Khayat}, title = {Transient twodimensional coating flow of a viscoelastic fluid film on a substrate of arbitrary shape}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {199--233}, } @ARTICLE{KimChoJeonHogenauerKensey2000, author = { . and Kensey}, title = {A new method for blood viscosity measurement}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, number = { }, pages = {47--56}, } @ARTICLE{KimYun2000, author = {Kim and .}, title = {Block ILU factorization preconditioners for a block--tridiagonal H--matrix}, journal = {Linear Algebra and\_its Applications}, year = {2000}, volume = {317}, pages = {103--125}, } @ARTICLE{KlarSchmeiser2001, author = { Schmeiser}, title = {Numerical passage from radiative heat transfer to nonlinear diffusion models}, journal = {Mathematical Models and Methods in Applied Sciences}, year = {2001}, volume = {11}, number = {5}, pages = {749--767}, } @ARTICLE{KlawonnWidlund2001, author = { .}, title = {FETI and Neumann--Neumann iterative substructuring methods: connections and new results}, journal = {Communications on Pure and Applied Mathematics}, year = {2001}, volume = {LIV}, } @ARTICLE{Klein2000, author = {.}, title = {Asymptotic analyses for atmospheric flows and the constructon of asymptotically adaptive numerical methods}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {11}, pages = {765--777}, } @ARTICLE{KnollVanderheydenMousseauKothe2001, author = { .}, title = {On preconditioning Newton--Krylov methods in solidifying flow applications}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {381--397}, } @ARTICLE{KnuppRobidoux2000, author = {Knupp and Robidoux}, title = {A framework for variational grid generation: conditioning the Jacobian matrix with matrix norms}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2029--2047}, } @ARTICLE{Knyazev2001, author = {.}, title = {Toward the optimal preconditioned eigensolver: locally optimal block preconditioned conjugate gradient method}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {517--541}, } @ARTICLE{Koh2000, author = {Koh}, title = {On boundary conditions and numerical methods for the unsteady incompressible {Navier}--{Stokes} equations}, journal = {Computers and Fluids}, year = {2000}, volume = {29}, pages = {791--812}, } @ARTICLE{Kopteva2001, author = {Kopteva}, title = {Uniform pointwise convergence of difference schemes for convection--diffusion problems on layer--adapted meshes}, journal = {Computing}, year = {2001}, volume = {66}, pages = {179--197}, } @ARTICLE{KoetterMarkus0, author = {K\"{o}tter and Markus}, title = {Simplest implementation of double--diffusive convection}, journal = {0}, year = {0}, } @ARTICLE{KovenyaChernySharovKaramyshevLebedev2001, author = { Lebedev}, title = {On some approaches to solve {CFD} problems}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {903--916}, } @ARTICLE{KowarschikRuedeWeissKarl2000, author = {. and R\"{u} .}, title = {Cache--aware multigrid methods for solving poissons equation in two dimensions}, journal = {Computing}, year = {2000}, volume = {64}, pages = {381--399}, } @ARTICLE{KrafczykCerrolazaSchulzRank1998, author = { .}, title = {Analysis of 3D transient blood flow passing trough an artificial aortic valve by Lattice--Boltzmann methods}, journal = {Journal of Biomechanics}, year = {1998}, volume = {31}, pages = {453--462}, } @ARTICLE{KrausBrand2000, author = {}, title = {Condition numbers of approximate Schur complements in two--{} and threedimensional discretizations on hierarchically ordered grids}, journal = {Computing}, year = {2000}, volume = {65}, pages = {135--154}, } @ARTICLE{KrechelPlum1991, author = {}, title = {A parallel block preconditioned conjugate gradient method}, journal = {0}, year = {1991}, } @ARTICLE{KrechelPlumStueben1990, author = { and St\"{u}ben}, title = {Parallelization and vectorization aspects of the solution of tridiagonal linear systems}, journal = {Parallel Computing}, year = {1990}, pages = {31--49}, } @ARTICLE{Kunert2000, author = {.}, title = {An a posteriori residual error estimator for the finite element method on anisotropic tetrahedral meshes}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, number = { }, pages = {471--490}, } @ARTICLE{KunertVerfuerth2000, author = {. and Verf\"{u}.}, title = {Edge residuals dominate a posteriori error estimates for linear finite element methods on anisotropic triangular and tetrahedral meshes}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, pages = {283--303}, } @ARTICLE{Kweon2000, author = {.}, title = {An optimal order convergence for a weak formulation of the compressible {Stokes} system with inflow boundary condition}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, pages = {305--318}, } @ARTICLE{LacknerMenikoff2000, author = {. and Menikoff}, title = {Multiscale linear solvers for very large systems derived from PDEs}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1950--1968}, } @ARTICLE{LarsenThoemmesKlar2001, author = {. and Th\"{o}. and .}, title = {Simplified $P N$ Approximations to the equations of radiative heat transfer in glass I: modelling}, journal = {0}, year = {2001}, month = oct, } @ARTICLE{LarsenThoemmesKlar2001a, author = {. and .}, title = {Frequency averaged approximations to the equations of radiative heat transfer in glass}, journal = {0}, year = {2001}, month = oct, } @ARTICLE{Larson2000, author = {.}, title = {A posteriori and a priori error enalysis for finite element approximations of self--adjoint elliptic eigenvalue problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {2}, pages = {608--625}, } @ARTICLE{LaytonTobiska1998, author = {. and .}, title = {A two--level method with backtracking for the {Navier}--{Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {1998}, volume = {35}, number = {5}, pages = {2035--2054}, } @ARTICLE{LeeManteuffelMcCormickRuge2000, author = { . and . and .}, title = {First--order system least--squares for the Helmholtz equation}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1927--1949}, } @ARTICLE{Leray1933, author = {Leray}, title = {On the motion of a viscous liquid filling space}, journal = {Comptes Rendues Acad. Sciences}, year = {1933}, volume = {196}, } @ARTICLE{LericheLabrosse2000, author = {.}, title = {High--order direct {Stokes} solvers with or without temporal splitting: numerical investigations of their comparative properties}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {4}, pages = {1386--1410}, } @ARTICLE{LeRoux2000, author = {.}, title = {Numerical solution of nonlinear reacton diffusion processes}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {5}, pages = {1644--1656}, } @ARTICLE{LesoinneFarhat1996, author = { .}, title = {Geometric conservation laws for flow problems with moving boundaries and deformable meshes, and their impact on aeroelastic computations}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1996}, volume = {134}, pages = {71--90}, } @ARTICLE{LeTallec2000, author = {.}, title = {A hierarchy of hyperbolic models linking Boltzmann to {Navier} {Stokes} equations for polyatomic gases}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {11-12}, pages = {779--789}, } @ARTICLE{LeTallecMani2000, author = {. and .}, title = {Numerical analysis of a linearised fluid--structure interaction problem}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {317--354}, note = {Digital Object Identifier (DOI): 10.1007/s002110000183}, } @ARTICLE{Li2000, author = {Li}, title = {Convergence analysis of finite element methods for singularly pertubed problems}, journal = {Computers and Mathematics with Applications}, year = {2000}, volume = {40}, pages = {735--745}, } @ARTICLE{Li2001, author = {Li}, title = {CGNR is an error reducing algorithm}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2109--2112}, } @ARTICLE{LianXuHuangAdams2001, author = { .}, title = {On the squeeze flow of a power--law fluid between rigid spheres}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {100}, pages = {151--164}, } @ARTICLE{Liberman2000, author = {Liberman}, title = {A posteriori error estimator for mixed finite element method for Ressner--Mindlin plate}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {236}, pages = {1383--1396}, } @ARTICLE{LinSaigal2000, author = {}, title = {An incomplete Cholesky factorization for dens symmetric positive definite matrices}, journal = {BIT Numerical Mathematics}, year = {2000}, volume = {40}, number = {3}, pages = {536--558}, } @ARTICLE{Linss2000, author = {Lin"s}, title = {Analysis of a {Galerkin} finite element method on a Bahkov--Shishkin mesh for a linear convection--diffusion problem}, journal = {IMA Journal of Numerical Analysis}, year = {2000}, volume = {20}, pages = {621--632}, } @ARTICLE{Linss2001, author = {Lin"s}, title = {Uniform pointwise convergence of finite difference schemes using grid equidistribution}, journal = {Computing}, year = {2001}, volume = {66}, pages = {27--39}, } @ARTICLE{LinssStynes2000, author = {Lin"s and .}, title = {The SDFEM on Shishkin meshes for linear convection--diffusion problems}, journal = {Numerische Mathematik}, year = {2000}, } @ARTICLE{LionsPironneau2000, author = {. and .}, title = {Overlapping domain decomposition of evolution operators}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, pages = {937--942}, } @ARTICLE{LionsPironneau2000a, author = {. and .}, title = {Non--overlapping domain decomposition for evolution operators}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2000}, volume = {330}, pages = {943--950}, } @ARTICLE{LiPetzold2000, author = {.}, title = {Software and algorithms for sensitivity analysis of large--scale differntial algebraic systems}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {125}, pages = {131--145}, } @ARTICLE{Liu2000, author = {Liu}, title = {The analysis of a finite element method with streamline diffusion for the compressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {1}, pages = {001--016}, } @ARTICLE{LiuJiLiao1998, author = {. and . and .}, title = {An Adaptive Grid Method and its Application to Steady Euler Flow Calculations}, journal = {SIAM Journal on Scientific Computing}, year = {1998}, volume = {20}, number = {3}, pages = {811--825}, } @ARTICLE{LiuWalkington2000, author = { .}, title = {Approximation of liquid crystal flows}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {3}, pages = {725--741}, } @ARTICLE{LiuWeinan2000, author = {}, title = {Simple finite element method in vorticity formulation for incompressible flows}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {234}, pages = {579--593}, } @ARTICLE{LiuYan2001, author = {Liu and Yan}, title = {Quasi--norm a priori and a posteriori error estimates for the nonconforming approximation of p--Laplacian}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {341--378}, } @ARTICLE{LlorenteDiskinMelson2000, author = { Melson}, title = {Alternating plane smoothers for multiblock grids}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {218--241}, } @ARTICLE{LohHui2000, author = {}, title = {A New {L}agrangian Method for Time--Dependent Inviscid Flow Computation}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {330--350}, } @ARTICLE{Loehner1998, author = {.}, title = {Renumbering strategies for unstructured--grid solvers operating on shared--memory, cache--based parallel machines}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1998}, volume = {163}, pages = {95--109}, } @ARTICLE{LubeMuellerOtto2000, author = { M\"{ .}, title = {A non--overlapping domain decomposition method for the advection--diffusion problem}, journal = {Computing}, year = {2000}, volume = {64}, pages = {49--68}, } @ARTICLE{LubeMuellerOtto2001, author = { .}, title = {A nonoverlapping domain decomposition method for stabilized finite element approximations of the Oseen equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {132}, pages = {211--236}, } @ARTICLE{Lui2001, author = {.}, title = {On Schwarz alternating methods for the incompressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {1974--1986}, } @ARTICLE{Lui2001a, author = {.}, title = {On accelerated convergence of nonoverlapping Schwarz methods}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {130}, pages = {309--321}, } @ARTICLE{MacCormack2001, author = {.}, title = {Iterative modified approximate factorization}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {917--925}, } @ARTICLE{ManouziFarhoul2001, author = {}, title = {Mixed finite element analysis of a non--linear three--fields stokes model}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {143--164}, } @ARTICLE{MardalTaiWinther0, author = { .}, title = {A robust finite element method for Darcy--{Stokes} flow}, journal = {0}, year = {0}, } @ARTICLE{MariammaMajhi2000, author = {Mariamma and Majhi}, title = {Flow of a Newtonian fluid in a blood vessel with permeable wall {}--{} a theoretical model}, journal = {Computers and Mathematics with Applications}, year = {2000}, volume = {40}, pages = {1419--1432}, } @ARTICLE{Martinez2000, author = {.}, title = {Practical quasi--Newton methods for solving nonlinear systems}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {124}, pages = {97--121}, } @ARTICLE{Masmoudi2001, author = {Masmoudi}, title = {From Vlasov--poisson system to the incompressible euler system}, journal = {Communications in Partial Differential Equations}, year = {2001}, volume = {26}, number = {9}, pages = {1913--1928}, } @ARTICLE{MasudHughes1997, author = {, .}, title = {A space--time Galerkin/least--squares finite element formulation of the {Navier}--{Stokes} equations for moving domain problems}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1997}, volume = {146}, pages = {91--126}, } @ARTICLE{Maury1999, author = {Maury}, title = {Direct Simulations of 2D Fluid--Particle Flows in Biperiodic Domains}, journal = {J. Comput. Phys.}, year = {1999}, volume = {156}, pages = {325--351}, } @ARTICLE{McDavidDantzig1997, author = {McDavid and .}, title = {Design sensitivity and finite element analysis of free surface flows with application to optimal design of casting rigging systems}, journal = {0}, year = {1997}, } @ARTICLE{MelenkKirchnerSchwab2000, author = {. and .}, title = {Spectral {Galerkin} discretization for hydrodynamic stability problems}, journal = {Computing}, year = {2000}, volume = {65}, pages = {97--118}, } @ARTICLE{MeleshkoPukhnachev1999, author = {Meleshko and .}, title = {One class of partially invariant solutions of the {Navier}--{Stokes} equations}, journal = {Journal of Applied Mechanics and Technical Physics}, year = {1999}, volume = {40}, number = {2}, } @ARTICLE{MikulaRamarosy2001, author = {. and Ramarosy}, title = {Semi.implicit finite volume scheme for solving nonlinear diffusion equations in image processing}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {561--590}, } @ARTICLE{MikulaSevcovic2001, author = {. and .}, title = {Evolution of plane curves driven by a nonlinaer function of curvature and anisotropy}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {5}, } @ARTICLE{MinYooChoi2001, author = { Yoo, . and Choi}, title = {Effect of spatial discretization schemes on numerical solutions of viscoelastic fluid flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {100}, pages = {27--47}, } @ARTICLE{MitsoulisZisis2001, author = {Mitsoulis and Zisis}, title = {Flow of Bingham plastics in a lid--driven square cavity}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {101}, pages = {173--180}, } @ARTICLE{MittalAl-Kurdi2001, author = {. and .}, title = {Application of the Cramer rule in the solution of sparse systems of linear algebraic equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {136}, pages = {001--015}, } @ARTICLE{Monga-MadeNotay2000, author = {Monga-Made and Notay}, title = {Dynamically relaxed block incomplete factorizations for solving two--{} and three--dimensional problems}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {2008--2028}, } @ARTICLE{Moore2000, author = {.}, title = {An adaptive finite element method for parabolic differential systems: some algorithmic considerations in solving in three space dimensions}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1567--1586}, } @ARTICLE{MorinNochettoKunibertSiebert2000, author = {. and . and }, title = {Data oscillation and convergence of adaptive {FEM}}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {2}, pages = {466--488}, } @ARTICLE{MukherjeeMazumderCalderer2001, author = {Mukherjee and Calderer}, title = {Poiseuille flow of liquid crystals: highly oscilalatory regimes}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {99}, number = { }, pages = {37--55}, } @ARTICLE{MurphyGolubWathen2000, author = {. and . and .}, title = {A Note on Preconditioning for Indefinite Linear Systems}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {1969--1972}, } @ARTICLE{Nash2000, author = {.}, title = {A survey of truncated--Newton methods}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {124}, pages = {45--59}, } @ARTICLE{NestlerWheelerRatkeStoecker2000, author = { \"{o}cker}, title = {Phase--field model for solidification of a monotectic alloy with convection}, journal = {Physica D}, year = {2000}, volume = {141}, pages = {133--154}, } @ARTICLE{NeussJaegerWittum2001, author = { J\"{a} .}, title = {Homogenization and multigrid}, journal = {Computing}, year = {2001}, volume = {66}, pages = {001--026}, } @ARTICLE{NewmarkAsce1959, author = {}, title = {A method of computation for structural dynamics}, journal = {J. of the Engineering Mechanics Division}, year = {1959}, } @ARTICLE{NikolovDoghriPierardZealouk2000, author = {. and . and . and .}, title = {Multi--scale constitutive modeling of the small deformations of semi--crystalline polymers}, journal = {J. Mech. Phys. Solids}, year = {2000}, } @ARTICLE{NilssenTaiWinther2000, author = { .}, title = {A robust nonconforming $H^2$--element}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {234}, pages = {489--505}, } @ARTICLE{NoordmansHemker2000, author = { .}, title = {Application of an adaptive sparse--gris technique to a model singular pertubation problem}, journal = {Computing}, year = {2000}, volume = {65}, pages = {357--378}, } @ARTICLE{Notay2000, author = {Notay}, title = {Optimal order preconditioning of finite difference matrices}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {6}, pages = {1991--2007}, } @ARTICLE{Oliveira2000, author = {Oliveira}, title = {A traceless stress tensor formulation for viscoelastic fluid flow {}}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {55--65}, } @ARTICLE{Oliveira2001, author = {Oliveira}, title = {Method for time--dependent simulations of viscoelastic flows: vortex shedding behind cylinder}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {101}, pages = {113--137}, } @ARTICLE{OlshanskiiReusken2000, author = {. and .}, title = {On the convergence of a multigrid method for linear reaction--diffusion problems}, journal = {Computing}, year = {2000}, volume = {65}, pages = {193--202}, } @ARTICLE{OlshanskiiStaroverov2000, author = {. and Staroverov}, title = {On simulation of outflow boundary conditions in finite difference calculations for incompressible fluid}, journal = {International Journal for Numerical Methods in Fluids}, year = {2000}, volume = {33}, pages = {499--534}, } @ARTICLE{OttoLubeMueller2001, author = {. and }, title = {An iterative substructuring method for div--stable finite element approximations of the Oseen problem}, journal = {Computing}, year = {2001}, volume = {67}, pages = {91--117}, } @ARTICLE{Ould-SalihiCottetHamraoui2000, author = {Ould-Salihi and . and Hamraoui}, title = {Blending finite--difference and vortex methods for incompressible flow computations}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {5}, pages = {1655--1674}, } @ARTICLE{PalanisamyKawahara1993, author = {Palanisamy and Kawahara}, title = {A Fractional Step Arbitrary {L}agrangian--{E}ulerian Finite Element Method For Free Surface Sensity Flows}, journal = {International Journal of Computational Fluid Dynamics}, year = {1993}, volume = {1}, pages = {57--77}, } @ARTICLE{PareschiRusso2000, author = {. and .}, title = {Numerical solution of the Boltzmann equation I: spectrally accurate approximation of the collision operator}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {4}, pages = {1217--1245}, } @ARTICLE{ParkRyu2001, author = {. and .}, title = {Hopf Bifurcation in Thermal Convection of Viscoelastic Fluids Within Finite Domains}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {101}, pages = {001--019}, } @ARTICLE{ParkRyu2001a, author = {Park and .}, title = {Rayleigh--Benard convection of viscoelastic fluids in finite domains}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {98}, pages = {169--184}, } @ARTICLE{Parlett2000, author = {Parlett}, title = {Rof tridiagonals T replace T with \$LDL\^{} t\$}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {117--130}, } @ARTICLE{PatankarHu2001, author = {. and Hu}, title = {Rheology of a suspension of particles in viscoelastic fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {96}, pages = {427--443}, } @ARTICLE{PerniceTocci2001, author = {Pernice and Tocci}, title = {A multigrid--preconditioned Newton--Krylov method for the incompressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {398--418}, } @ARTICLE{PetrovaPukhnachev1999, author = { .}, title = {One--dimensional motion of an emulsion with solidification}, journal = {Journal of Applied Mechanics and Technical Physics}, year = {1999}, volume = {40}, number = {3}, } @ARTICLE{Pflaum2001, author = {Pflaum}, title = {Semi--unstructured grids}, journal = {Computing}, year = {2001}, volume = {67}, pages = {141--166}, } @ARTICLE{PhillipsWilliams2001, author = { Williams}, title = {A Semi--{L}agrangian Finite Volume Method for Newtonian Contraction Flows}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2152--2177}, } @ARTICLE{PinnauThoemmes2001, author = {. and .}, title = {Optimal boundary control of glass cooling processes}, journal = {0}, year = {2001}, month = oct, } @ARTICLE{PipernoFarhatLarrouturou1995, author = {. and . and Larrouturou}, title = {Partitioned procedures for the transient solution of coupled aroelastic problems. Part 1: Model problem, theory and two--dimensional application}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1995}, volume = {124}, pages = {79--112}, } @ARTICLE{Pozrikidis2001, author = {.}, title = {Three--dimensional oscillations of inviscid drops induced by surface tension}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {417--444}, } @ARTICLE{PrietoSantiagoEspadasLlorenteTirado2001, author = {}, title = {Parallel multigrid for anisotropic elliptic equations}, journal = {Journal of Parallel and Distributed Computing}, year = {2001}, volume = {61}, pages = {96--114}, } @ARTICLE{Pukhnachov0, author = {.}, title = {Solvability of initial boundary value problem in non--standard model of convection}, journal = {0}, year = {0}, } @ARTICLE{Pukhnachov1994, author = {.}, title = {Microconvection in a vertical layer}, journal = {Fluid Dyn.}, year = {1994}, volume = {29}, number = {5}, } @ARTICLE{Pukhnachov1995, author = {.}, title = {Exact multidimensional solutions of the nonlinear diffusion equation}, journal = {Journal of Applied Mechanics and Technical Physics}, year = {1995}, volume = {36}, number = {2}, } @ARTICLE{Pukhnachov1996, author = {.}, title = {Interaction between a distributed source and the free surface of a viscous fluid}, journal = {Fluid Dyn.}, year = {1996}, volume = {31}, number = {2}, } @ARTICLE{Pukhnachov1999, author = {.}, title = {On a problem of viscous strip deformation with a free boundary}, journal = {C. R., Math., Acad. Sci. Paris}, year = {1999}, volume = {328}, number = {1}, pages = {357--362}, } @ARTICLE{PukhnachovSemenova1999, author = {. and Semenova}, title = {Model problem of instantaneous motion of a three--phase contact line}, journal = {Journal of Applied Mechanics and Technical Physics}, year = {1999}, volume = {40}, number = {4}, } @ARTICLE{QuarteroniTuveriVeneziani0, author = {. and Tuveri and Veneziani}, title = {Computational vascular fluid dynamics: problems, models and methods}, journal = {0}, year = {0}, } @ARTICLE{QuarteroniVeneziani1997, author = {. and Veneziani}, title = {Modeling and Simulation of Blood Flow Problems}, journal = {Comput. Sciencs for the 21st Century}, year = {1997}, } @ARTICLE{Queck1989, author = {Queck}, title = {The convergence factor of preconditioned algotrithms of the arrow--Hurwicz type}, journal = {SIAM Journal on Numerical Analysis}, year = {1989}, volume = {26}, number = {4}, pages = {1016--1030}, } @ARTICLE{Ramaswamy1990, author = {Ramaswamy}, title = {Numerical Simulation of Viscous Free Surface Flow}, journal = {J. Comput. Phys.}, year = {1990}, volume = {90}, pages = {396--430}, } @ARTICLE{Rannacher2001, author = {.}, title = {Adaptive {Galerkin} Finite Element Methods for Partial Differential Equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {205--233}, } @ARTICLE{Rasmussen2000, author = {Rasmussen}, title = {Lagrangian Viscoelastic Flow Computations Using the Rivlin--Sawyers Constitutive Model}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {92}, pages = {227--243}, } @ARTICLE{RaviVenkatesh2000, author = {Ravi and Venkatesh}, title = {A new framework for recontruction of images from boundaries}, journal = {SIAM Journal on Applied Mathematics}, year = {2000}, volume = {60}, number = {6}, pages = {1841--1866}, } @ARTICLE{Renardy2000a, author = {.}, title = {Location of the continuous spectrum on complex flows of the UCM fluid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, pages = {75--85}, } @ARTICLE{Renardy2001, author = {.}, title = {Self--similar breakup of a Giesekus jet}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {97}, pages = {283--293}, } @ARTICLE{Rieder2001, author = {Rieder}, title = {On convergence rates of inexact Newton regularizations}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {347--356}, } @ARTICLE{RobertsSidilkoverTsynkov2002, author = { .}, title = {On the combined performance of nonlocal artificial boundary conditions with the new generation of advanced multigrid flow solvers}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {269--308}, } @ARTICLE{RomineOrtega1988, author = {Romine and .}, title = {Parallel solution of triangular systems of equations}, journal = {Parallel Computing}, year = {1988}, pages = {109--114}, } @ARTICLE{RoosLinss2001, author = {. and Lin"s}, title = {Gradient recovery for singulary pertubed boundary value problems I}, journal = {Computing}, year = {2001}, volume = {66}, pages = {163--178}, } @ARTICLE{RustenWinther1992, author = {.}, title = {A preconditioned iterative methods for saddlepoint problems}, journal = {SIAM Journal on Matrix Analysis and Applications}, year = {1992}, volume = {13}, number = {3}, pages = {887--904}, } @ARTICLE{SaadSosonkina1999, author = {. and Sosonkina}, title = {Distributed Schur complement techniques for general sparse linear systems}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {21}, number = {4}, pages = {1337--1356}, } @ARTICLE{SaadvanderVorst2000, author = {. and .}, title = {Iterative solution of linear systems in the 20th century}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {123}, pages = {001--033}, } @ARTICLE{SaadYeungErhelGuyomarch2000, author = {. and Yeung and . and Guyomarch}, title = {A deflated version of the conjugate gradient algorithm}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1909--1926}, } @ARTICLE{SaadZhang2001, author = {. and .}, title = {Enhanced Multi--level Block ILU Preconditioning Strategies for General Sparse Linear Systems}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {130}, pages = {99--118}, } @ARTICLE{SanderWanner2000, author = {. and Wanner}, title = {Unexpectedly linear behavior for the Chan--Hillard equation}, journal = {SIAM Journal on Applied Mathematics}, year = {2000}, volume = {60}, number = {6}, pages = {2182--2202}, } @ARTICLE{Sangalli2001, author = {.}, title = {A robust a posteriori estimator for the residual--free bubbles method applied to advection--diffusion problems}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {379--399}, } @ARTICLE{SaniGreshoLeeGriffiths1980, author = { Sani}, title = {The cause and cure of the spurious pressures generated by certain {FEM} solutions of the incompressible {Navier}--{Stokes} equations}, journal = {International Journal for Numerical Methods in Fluids}, year = {1980}, } @ARTICLE{SchmidtSiegbert2000, author = {Schmidt and Siegbert}, title = {Albert--software for scientific computations and applications}, journal = {Acta Mathematica}, year = {2000}, volume = {LXX}, number = {1}, pages = {105--122}, } @ARTICLE{SchreckSchaefer2000, author = { .}, title = {Numerical study of bifurcation in three--dimensional sudden channel expansions}, journal = {Computers and Fluids}, year = {2000}, volume = {29}, pages = {583--593}, } @ARTICLE{Schwandt1989, author = {Schwandt}, title = {Cyclic reduction for tridiagonal systems of equations with interval coefficients on vector computers}, journal = {SIAM Journal on Numerical Analysis}, year = {1989}, volume = {26}, number = {3}, pages = {661--680}, } @ARTICLE{Schweizer2000, author = {Schweizer}, title = {Homogenization of a fluid problem with a free boundary}, journal = {Communications on Pure and Applied Mathematics}, year = {2000}, volume = {LIII}, pages = {1118--1152}, } @ARTICLE{Schweizer2000a, author = {Schweizer}, title = {Uniform estimates in two periodic homogenization problems}, journal = {Communications on Pure and Applied Mathematics}, year = {2000}, volume = {LIII}, pages = {1153--1176}, } @ARTICLE{SeegerHoffmann2000, author = {. and .}, title = {The cumulant method for computaional kinetic theory}, journal = {Continuum Mechanics and Thermodynamics}, year = {2000}, volume = {12}, pages = {403--421}, } @ARTICLE{SeegerHoffmann2001, author = {. and .}, title = {The cumulant method applied to a mixture of Maxwell gases}, journal = {0}, year = {2001}, month = jun, } @ARTICLE{Sesterhenn2001, author = {Sesterhenn}, title = {A characteristic--type formulation of the {Navier}--{Stokes} equations for high order upwind schemes}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {37--67}, } @ARTICLE{SheuWangTsai2000, author = { Tsai}, title = {Element--by--element parallel computation of incompressible {Navier}--{Stokes} equations in three dimensions}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1387--1400}, } @ARTICLE{ShihElman2000, author = {. and .}, title = {Iterative methods for stabilized discrete convection--diffusion problems}, journal = {IMA Journal of Numerical Analysis}, year = {2000}, volume = {20}, pages = {333--358}, } @ARTICLE{SieberSchaeferLauschkeSchierholz1999, author = {Sieber and Sch\"{a}. and Schierholz}, title = {Str\"{o}mungssimulation in Wendel--{} und Dispersionsr\"{u}hrwerken}, journal = {Chemie Ingenieur Technik}, year = {1999}, volume = {71}, number = {10}, pages = {1159--1163}, month = oct, } @ARTICLE{SieberSchaeferWechslerDurst0, author = { Sch\"{ .}, title = {Numerical Prediction of Time--Dependent Turbulent Flow in a Hyperbolic Stirrer}, journal = {0}, year = {0}, note = {TH Darmstadt, Uni Erlangen}, } @ARTICLE{SimoensVandewalle2000, author = { Vandewalle}, title = {Waveform relaxation with fast direct methods as preconditioner}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {5}, pages = {1755--1773}, } @ARTICLE{SinghJoseph2000, author = {Singh and .}, title = {Sedimentation of a sphere near a vertical wall in an Oldroyd--B fluid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, pages = {179--203}, } @ARTICLE{SmileyChun2001, author = {Smiley and Chun}, title = {An algorithm for finding all solutions of a nonlinear system}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {137}, pages = {293--315}, } @ARTICLE{SmyrnaiosTsamopoulos2001, author = {Smyrnaios and .}, title = {Squeeze flow of Bingham plastics}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {100}, pages = {165--190}, } @ARTICLE{Starke2000, author = {Starke}, title = {{Gauss}--Newton multilevel methods for least--squares finite element computations of variably saturated subsurface flow}, journal = {Computing}, year = {2000}, volume = {64}, pages = {323--338}, } @ARTICLE{Stevenson1994, author = {.}, title = {Modified ILU as a smoother}, journal = {Numerische Mathematik}, year = {1994}, volume = {68}, pages = {295--309}, } @ARTICLE{Stuckmeier2000, author = {Stuckmeier}, title = {A steady--state particle method for the Boltzmann equation}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1435--1457}, } @ARTICLE{StynesTobiska2001, author = {. and .}, title = {The streamline--diffusion method for nonconforming Q1--rot elements on rectangular tensor--product meshes}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, volume = {21}, pages = {123--142}, } @ARTICLE{SumiyoshiEbisuzaki1998, author = {Sumiyoshi and Ebisuzaki}, title = {Performance of parallel solution of a block--tridiagonal linear system on Fujitsu VPP500}, journal = {Parallel Computing}, year = {1998}, pages = {287--304}, } @ARTICLE{SunwooParkLeeAhnLee2001, author = { }, title = {Numerical simulation of three--dimensional viscoelastic flow using the open boundary condition method in coextrusion process}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {99}, pages = {125--144}, } @ARTICLE{Sureshkumar2000, author = {Sureshkumar}, title = {Numerical observations on the continuous spectrum of the linearized viscoelastic operator in shear dominated complex flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {94}, pages = {205--211}, } @ARTICLE{Suri2001, author = {.}, title = {The p and hp finite element method for problems on thin domains}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {235--260}, } @ARTICLE{SzepessyTemponeZouraris2001, author = {}, title = {Adaptive weak approximation of stochastic differential equations}, journal = {Communications on Pure and Applied Mathematics}, year = {2001}, volume = {LIV}, pages = {1169--1214}, } @ARTICLE{SzyldVogel2001, author = {Szyld and .}, title = {FQMR: A flexible quasi--minimal residual method with enexact preconditioning}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {363--380}, } @ARTICLE{TabataTagami2000, author = {. and .}, title = {A finite element analysis of a linearized problem of the {Navier}--{Stokes} equations with surface tension}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {1}, pages = {40--57}, } @ARTICLE{TadmorTang2000, author = { Tang}, title = {Pointwise error estimates for relaxation approximations to conservation laws}, journal = {SIAM Journal on Mathematical Analysis}, year = {2000}, volume = {32}, number = {4}, pages = {870--886}, } @ARTICLE{TangYang2000, author = { Yang}, title = {A free moving boundary model and boundary iteration method for unsteady iscous flow in stenotic elastic tubes}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1370--1386}, } @ARTICLE{ThomasDiskinBrandt2001, author = { Brandt}, title = {Textbook multigrid efficiency for the incompressible {Navier}--{Stokes} equations: high Reynolds number wakes and boundary layers}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {853--874}, } @ARTICLE{ThomasLombard0, author = {. and Lombard}, title = {Geometric Conservation Law and Its Application to Flow on Moving Grids}, journal = {AIAA Journal}, year = {0}, volume = {17}, pages = {1030--1037}, note = {Lockheed Palo Alto Research Lab, Calif.}, } @ARTICLE{Thomee2001, author = {Thomee}, title = {From finite differences to finite elements. A short history of numerical analysis of partial differential equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {001--054}, } @ARTICLE{ThoemmesPinnauSeaidGoetzKlar0, author = {Th\"{. and . and \"{o}.}, title = {Numerical methods and optimal control for glass cooling processes}, journal = {0}, year = {0}, } @ARTICLE{ThoemmesSeaidGoetzKlar2001, author = {Th\"{. and G\"{o} .}, title = {Simplified $P N$ Approximations to the equations of radiative heat transfer in glass I: numerics}, journal = {0}, year = {2001}, month = oct, } @ARTICLE{Tiihonen2000, author = {.}, title = {Shape calculus and finite element method in smooth domains}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {233}, pages = {001--015}, } @ARTICLE{Tyrtyshnikov2000, author = {Tyrtyshnikov}, title = {Incomplete cross approximation in the mosaic--skeleton method}, journal = {Computing}, year = {2000}, volume = {64}, pages = {367--380}, } @ARTICLE{vanderVorst1987, author = {, .}, title = {Large tridiagonal and block tridiagonal linear systems on vector and parallel computers}, journal = {Parallel Computing}, year = {1987}, pages = {45--54}, } @ARTICLE{vanderVorst1987a, author = {, .}, title = {Analysis of a parallel solution method for tridiagonal linear systems}, journal = {Parallel Computing}, year = {1987}, pages = {303--311}, } @ARTICLE{VanekBrezinaMandel2001, author = {. and .}, title = {Convergence of algebraic multigrid based on smoothed aggregation}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {559--579}, } @ARTICLE{vanMeerveldWaters2001, author = { .}, title = {Numerical computation of the steady secondary flow in a tube with time--dependent curvature}, journal = {Q. J. Mech. Appl. Math.}, year = {2001}, volume = {54}, number = {4}, pages = {631--640}, } @ARTICLE{VespucciBroyden2001, author = {Vespucci and Broyden}, title = {Implementation of different computational variations of biconjugate residual methods}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {42}, pages = {1239--1253}, } @ARTICLE{Volkwein2001, author = {.}, title = {Optimal Control of a Phase--Field Model Using Proper Orthogonal Decomposition}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2001}, volume = {81}, number = {2}, pages = {83--97}, } @ARTICLE{WachsClermont2000, author = {Wachs and Clermont}, title = {Non--isothermal viscoelastic flow computations in an axisymmetric contraction at high Weissenberg numbers by a finite volume method}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {147--184}, } @ARTICLE{Wagner2000, author = {Wagner}, title = {On the algebraic construction of multilevel transfer operators}, journal = {Computing}, year = {2000}, volume = {65}, pages = {73--95}, } @ARTICLE{WagnerHuettlFriedrich2001, author = { .}, title = {Low--Reynolds--number effects derived from direct numerical simulations of turbulent pipe flow}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {581--590}, } @ARTICLE{WalshawCross2000, author = { Cross}, title = {Mesh partitioning: a multilevel balancing and refinement algorithm}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {63--80}, } @ARTICLE{WanChanSmith2000, author = { . and Smith}, title = {An energy--minimizing interpolation for robust multigrid methods}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1632--1649}, } @ARTICLE{Wang2001, author = {.}, title = {Special bilinear quadrilateral elements for locally refined finite element grids}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2029--2050}, } @ARTICLE{WangLiangewingLyonsQin2000, author = {. and . and . and . and .}, title = {An Approximation to Miscible Fluid Flows in Porous Media with Point Sources and Sinks by an Eulerian--Lagrangian Localized Adjoint Method and Mixed Finite Element Methods}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {2}, pages = {561--581}, } @ARTICLE{WangLiu2000, author = {. and .}, title = {Convergence of Gauge method for incompressible flow}, journal = {Mathematics of Computation}, year = {2000}, volume = {69}, number = {232}, pages = {1385--1407}, } @ARTICLE{WangXu1999, author = {. and .}, title = {A crosswind block iterative method for convection--dominated problems}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {21}, number = {2}, pages = {620--645}, } @ARTICLE{WapperomKeuningsLegat2000, author = {}, title = {The Backward--Tracking {L}agrangian Particle Method for Transient Viscoelastic Flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {91}, pages = {273--295}, } @ARTICLE{Weickert2000, author = {.}, title = {Applications of nonlinear diffusion in image processing and computer vision}, journal = {Acta Mathematica}, year = {2000}, volume = {LXX}, pages = {33--50}, } @ARTICLE{WesselingvanderHeul2001, author = {. and }, title = {Uniformly effective numerical methods for hyperbolic systems}, journal = {Computing}, year = {2001}, volume = {66}, pages = {249--267}, } @ARTICLE{WestermannJohnsonErtl2001, author = { .}, title = {Topology--preserving smoothing of vector fields}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2001}, volume = {7}, number = {3}, } @ARTICLE{Wetton2000, author = {Wetton}, title = {Error analysis of pressure increment schemes}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {1}, pages = {160--169}, } @ARTICLE{WetzelTucker2001, author = {. and Tucker}, title = {Microstructural evolution during complex laminar flow of liquid--liquid dispersions}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2001}, volume = {101}, pages = {21--41}, } @ARTICLE{White2000, author = {White}, title = {Domain decomposition splittings}, journal = {Linear Algebra and\_its Applications}, year = {2000}, volume = {316}, pages = {105--112}, } @ARTICLE{WienandsOosterlee2001, author = {. and .}, title = {On three--grid Fourier analysis for multigrid}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {2}, pages = {651--671}, } @ARTICLE{Wieners2000, author = {.}, title = {Robust multigrid methods for nearly incompressible elasticity}, journal = {Computing}, year = {2000}, volume = {64}, pages = {289--306}, } @ARTICLE{WoznickiJedrzejec2001, author = {Jedrzejec and Woznicki}, title = {A new class of modified line--SOR algorithms}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {131}, pages = {89--142}, } @ARTICLE{Xu2000, author = {Xu}, title = {On the accuracy of nonconforming quadrilateral Q1 element approximation for the {Navier}--{Stokes} problem}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {1}, pages = {17--39}, } @ARTICLE{XuChen2001, author = {Xu and .}, title = {Multigrid for the mortar element method for P1 nonconforming element}, journal = {Numerische Mathematik}, year = {2001}, volume = {88}, pages = {381--398}, } @ARTICLE{Xue2002, author = {Xue}, title = {Study of laminar flow in helical circular pipes with {Galerkin} method}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {113--129}, } @ARTICLE{XuZhou1999, author = {Xu and Zhou}, title = {Local and parallel finite element algorithms based on two--grid discretizations}, journal = {Mathematics of Computation}, year = {1999}, volume = {69}, number = {231}, pages = {881--909}, } @ARTICLE{XuZhou2000, author = { Zhou}, title = {A two--grid discretization scheme for eigenvalue problems}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {233}, pages = {17--25}, } @ARTICLE{YedderBilgen1997, author = {}, title = {Laminar natural convection in inclined enclosures by a solid wall}, journal = {Int. J. Heat Mass Transfer}, year = {1997}, volume = {32}, pages = {455--462}, } @ARTICLE{YedderBilgen1999, author = {}, title = {Turbulent natural convection in a square enclosure bounded by a solid wall with localized heating}, journal = {Int. J. Heat Mass Transfer}, year = {1999}, volume = {35}, pages = {401--408}, } @ARTICLE{Yun2000, author = {.}, title = {Block ILU preconditioners for a nonsymmetric block--tridiagonal M--matrix}, journal = {BIT Numerical Mathematics}, year = {2000}, volume = {40}, number = {3}, pages = {583--605}, } @ARTICLE{YurikoRenardy2000, author = {Yuriko and .}, title = {Shear--induced rupturing of a viscous drop in a Bingham liquid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {235--251}, } @ARTICLE{ZhangXu2001, author = {. and .}, title = {Properties and Numerical Performance of Quasi--Newton Methods with Modified Quasi--Newton Equations}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {137}, pages = {269--278}, } @ARTICLE{ZhangYu2000, author = {. and .}, title = {On a Singular Nonlinear Dirichlet Problem with a Convection Term}, journal = {SIAM Journal on Mathematical Analysis}, year = {2000}, volume = {32}, number = {4}, pages = {916--927}, } @ARTICLE{Zulehner2000, author = {.}, title = {A class of smoothers for saddle point problems}, journal = {Computing}, year = {2000}, volume = {65}, pages = {227--246}, } @ARTICLE{Grimmer1997, author = {.}, title = {A Comparative Study of Transport--Oriented Discretizations of Convection--dominated Problems with Application to the Incompressible {Navier}--{Stokes} Equations}, journal = {PhD Thesis}, year = {1997}, } @ARTICLE{Schreiber1996, author = {.}, title = {Eine nichtkonforme Finite--Elemente--Methode zur Loesung der inkompressiblen 3--D {Navier}--{Stokes} Gleichungen}, journal = {PhD Thesis}, year = {1996}, } @ARTICLE{Wallis1999, author = {.}, title = {Efficient Multigrid Poisson Solvers in General Coordinates on Tensorproduct Meshes}, journal = {PhD Thesis}, year = {1999}, } @ARTICLE{KotalikNishiyma2002, author = {. and .}, title = {An effect of magnetic field on arc plasma flow}, journal = {IEEE Transactions on Plasma Science}, year = {2002}, volume = {30}, number = {1}, pages = {160--161}, } @ARTICLE{ApelDobrowolski1992, author = {. and .}, title = {Anisotropic Interpolation with Applications to the Finite Element Method}, journal = {Computing}, year = {1992}, volume = {47}, pages = {277--293}, } @ARTICLE{Chien1982, author = {.}, title = {Predictions of channel and boundary layer flows with a low--Reynolds--number turbulence model}, journal = {AIAA Journal}, year = {1982}, number = {20}, pages = {33--38}, } @ARTICLE{Ciarlet1976, author = {.}, title = {The Finite Element Method for Elliptic Problems}, journal = {North-Holland}, year = {1976}, } @ARTICLE{CrouzeixRaviart1973, author = {. and .}, title = {Conforming and non--conforming finite element mehtods for solving the stationary {Stokes} equations}, journal = {RAIRO}, year = {1973}, pages = {77--104}, } @ARTICLE{DongarraDuCrozHammarlingHanson1988, author = {. and . and . and .}, title = {An extended set of FORTRAN Basic Linear Algebra Subprograms}, journal = {ACM Transactions on Mathematical Software}, year = {1988}, number = {14}, pages = {1--17}, } @ARTICLE{DupontScott1980, author = {. and .}, title = {Polynomial Approximation of Functions in Sobolev Spaces}, journal = {Math. Comput.}, year = {1980}, number = {34}, pages = {441--463}, } @ARTICLE{John1999, author = {.}, title = {A comparison of parallel solvers for the incompressible {Navier}--{Stokes} equations}, journal = {Computing and Visualization in Science}, year = {1999}, volume = {4}, pages = {193--200}, } @ARTICLE{JohnMatthiesSchieweckTobiska1998, author = {. and . and . and .}, title = {A streamline--diffusion method for nonconforming finite element approximations applied to convection diffusion problems}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1998}, number = {166}, pages = {85--97}, } @ARTICLE{JohnTobiska2000, author = {. and .}, title = {A Coupled Multigrid Method for Nonconforming Finite Element Discretizations of the 2D--{Stokes} Equation}, journal = {Computing}, year = {2000}, volume = {64}, number = {307-321}, pages = {307--321}, month = sep, } @ARTICLE{Johnson1987, author = {.}, title = {Numerical Solution of Partial Differential Equations by the Finite Element Method}, journal = {Cambridge University Press}, year = {1987}, } @ARTICLE{LawsonHansonKincaidKroghF.T.1979, author = {. and . and . and .}, title = {Basic Linear Algebra Subprograms for FORTRAN usage}, journal = {ACM Transactions on Mathematical Software}, year = {1979}, number = {5}, pages = {308--323}, } @ARTICLE{Lube1994, author = {.}, title = {Stabilized {Galerkin} finite element methods for convection dominated and incompressible flow problems}, journal = {Banach Center Publications}, year = {1994}, pages = {85--104}, note = {Polish Academy of Scinece, Warszawa, Inst. of Mathematics}, } @ARTICLE{LaunderSpalding1974, author = {. and .}, title = {The numerical Simulation of Turbulent Flows}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1974}, volume = {3}, pages = {269--289}, } @ARTICLE{MohammadiPironneau1994, author = {. and .}, title = {Analysis of the K--Epsilon TURBULENCE MODEL}, journal = {Recherches en Mathematiques Appliquees}, year = {1994}, } @ARTICLE{MatthiesTobiska2001, author = {. and .}, title = {The Streamline--Diffusion Method for Conforming and Nonconforming Finite Elements of Lowest Order Applied to Convection--Diffusion Problems}, journal = {Computing}, year = {2001}, number = {66}, pages = {343--364}, } @ARTICLE{OhmoriUshijima1984, author = {. and .}, title = {A Technique of Upstream Type Applied to a Linear Nonconforming Finite Element Approximation of Convective Diffusion Equations}, journal = {RAIRO}, year = {1984}, number = {18}, pages = {309--332}, } @ARTICLE{SchieweckTobiska.L.1989, author = {. and .}, title = {A nonconforming finite element method of upstream type applied to the stationary {Navier}--{Stokes} equation}, journal = {RAIRO}, year = {1989}, volume = {23}, number = {4}, pages = {627--647}, } @ARTICLE{StynesTobiska2001a, author = {. and .}, title = {The streamlinediffusion method for nonconforming Q\_1 elements on rectangular tensor--product meshes}, journal = {IMA Journal of Numerical Analysis}, year = {2001}, number = {21}, pages = {123--142}, } @ARTICLE{Suttmeier2001, author = {.}, title = {An Adaptive Displacement/Pressure Finite Element Scheme for Treating Incompressibility Effects in Elasto--Plastic Materials}, journal = {Numerical Methods for Partial Differential Equations}, year = {2001}, number = {17}, pages = {369--382}, } @ARTICLE{Temam1977, author = {.}, title = {Theory and Numerical Analysis of the {Navier}--{Stokes} Equations}, journal = {0}, year = {1977}, note = {North-Holland}, } @ARTICLE{Vanka1985, author = {.}, title = {Implicit Multigrid Solutions of {Navier}--{Stokes} Equations in Primitive Variables}, journal = {J. Comput. Phys.}, year = {1985}, number = {65}, pages = {138--158}, } @ARTICLE{Wilcox1993, author = {.}, title = {Turbulenc Modeling for {CFD}}, journal = {DCW Industries}, year = {1993}, note = {La Canada}, } @ARTICLE{Turek1997b, author = {.}, title = {On discrete projection methods for the incompressible {Navier}--{Stokes} equations: An algorithmical approach}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1997}, volume = {143}, pages = {271--288}, } @ARTICLE{TurekMalekHron2000, author = {. and . and .}, title = {A numerical investigation of flows of shear--thinning fluids with applications to blood flow}, journal = {International Journal for Numerical Methods in Fluids}, year = {2000}, volume = {32}, pages = {863--879}, } @ARTICLE{TurekKuzmin2002, author = {. and .}, title = {Flux correction tools for finite elements}, journal = {J. Comput. Phys.}, year = {2002}, volume = {175}, pages = {525--558}, } @ARTICLE{TurekSchmachtel2002, author = {. and .}, title = {Fully coupled and operator--splitting approaches for natural convection}, journal = {International Journal for Numerical Methods in Fluids}, year = {2002}, volume = {40}, pages = {1109--1119}, } @ARTICLE{TurekZietheSpohnBreuer2002, author = { .}, title = {Core formation: A new modelling approach}, journal = {Geophys. Res. Abstr.}, year = {2002}, volume = {4}, number = {PS1.01}, note = {on CD-ROM}, } @ARTICLE{TurekOuazziSchmachtel2002, author = {. and .}, title = {Multigrid Methods for Stabilized Nonconforming Finite Elements for Incompressible Flow involving the Deformation Tensor formulation}, journal = {J. Numer. Math.}, year = {2002}, volume = {10}, pages = {235--248}, } @ARTICLE{TurekRungeBecker2000, author = {. and . and .}, title = {Performance rating via the FEAST indices}, journal = {Computing}, year = {2000}, volume = {63}, pages = {283--297}, } @ARTICLE{TurekRungeBecker2001, author = {. and . and .}, title = {The FEAST Indices {}--{} Realistic evaluation of modern software components and processor technologies}, journal = {Computers and Mathematics with Applications}, year = {2001}, volume = {41}, pages = {1431--1464}, } @ARTICLE{TurekRannacher1992, author = { .}, title = {A Simple Nonconforming Quadrilateral {Stokes} Element}, journal = {Numerical Methods for Partial Differential Equations}, year = {1992}, volume = {8}, pages = {97--111}, } @ARTICLE{TurekMalek1993, author = {. and .}, title = {Non--{}--newtonian flow prediction by divergence free finite elements}, journal = {SAACM}, year = {1993}, volume = {3}, number = {3}, pages = {165--180}, } @ARTICLE{Turek1996c, author = {.}, title = {Numerik in der Str\"{o}mungsmechanik: Numerische Effizienz versus Gigaflops}, journal = {Phys. Bl.}, year = {1996}, volume = {52}, number = {11}, pages = {1137--1139}, } @ARTICLE{TurekHeywoodRannacher1996, author = {. and . and .}, title = {Artificial Boundaries and Flux and Pressure Conditions for the Incompressible {Navier}--{Stokes} Equations}, journal = {International Journal for Numerical Methods in Fluids}, year = {1996}, volume = {22}, pages = {325--352}, } @ARTICLE{Turek1994g, author = {.}, title = {Multigrid techniques for a divergence--{}--free finite element discretization {}}, journal = {East-West Journal of Numerical Mathematics}, year = {1994}, volume = {2}, number = {3}, pages = {229--255}, } @ARTICLE{TurekWehrse1994, author = {. and .}, title = {Spectral appearance of dust enshrouded stars: A combination of a strongly accelerated solutiontechnique with a finite element approach for 2D radiative transfer {}}, journal = {Astronomy and Astrophysics}, year = {1994}, } @ARTICLE{Turek1996d, author = {.}, title = {A comparative study of time--stepping techniques for the incompressible {Navier}--Stoke equations: From fully implicit non--linear schemes to semi--implicit projection methods}, journal = {International Journal for Numerical Methods in Fluids}, year = {1996}, volume = {22}, pages = {987--1011}, } @ARTICLE{Turek2000d, author = {.}, title = {{CFD} for incompressible flow: Numerical efficiency vs. Gigaflops}, journal = {Numerical Linear Algebra and Applications}, year = {2000}, volume = {7}, pages = {473--482}, } @ARTICLE{Turek1993b, author = {.}, title = {An efficient solution technique for the radiative transfer equation}, journal = {Impact of computing in science and engineering}, year = {1993}, volume = {5}, pages = {201--214}, } @ARTICLE{Turek1995b, author = {.}, title = {A generalized mean intensity approach for the numerical solution of the radiative transfer equation}, journal = {Computing}, year = {1995}, volume = {54}, number = {1}, pages = {27--38}, } @ARTICLE{Turek1994c, author = {.}, title = {Tools for Simulating Nonstationary Incompressible Flow via Discretely Divergence--{}--free Finite Element Models}, journal = {International Journal for Numerical Methods in Fluids}, year = {1994}, volume = {18}, pages = {71--105}, } @ARTICLE{Yang2002, author = {.}, title = {Iterative schemes for mixed finite element methods with applications to elasticity and compressible flow problems}, journal = {Numerische Mathematik}, year = {2002}, volume = {83}, number = {83}, pages = {177--200}, note = {DOI 10.1007/s002110100356}, } @ARTICLE{Zulehner2001, author = {.}, title = {Analysis of iterative methods for saddle point problems: A unified approach}, journal = {Mathematics of Computation}, year = {2001}, volume = {Volume 71}, number = {238}, pages = {479--505}, month = may, } @ARTICLE{GrajewskiHronTurek2003a, author = {. and . and .}, title = {Numerical Analysis and A Posteriori Error Control for a New Nonconformining Linear Finite Element on Quadrilaterals}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, note = {submitted}, } @ARTICLE{RivkindSolonnikov2000, author = {. and .}, title = {Jeffery--Hamel Asymptotics for Steady State {Navier}--{Stokes} Flow in {} Domains with Sector--like Outlets to Infinity}, journal = {Journal of Mathematical Fluid Mechanics}, year = {2000}, pages = {324--352}, } @ARTICLE{SolonnikovRivkind2000, author = {. and .}, title = {On nonsymmetric two--dimensional viscous flow through an aperture}, journal = {Port. Math.}, year = {2000}, volume = {Vol. 57}, number = {Fasc. 4}, note = {Presented by J. F.-Rodrigues}, } @ARTICLE{PanGlowinskiGaldi2002, author = {Pan and . and Galdi}, title = {Direct simulation of the motion of a settling ellipsoid in Newtonian fluid}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {71--82}, month = dec, } @ARTICLE{HagiwaraShinoda2002, author = {}, title = {The high precise/efficient pre/postprocessor}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {325--333}, month = mar, } @ARTICLE{KohnoTanahashi2002, author = {. and Tanahashi}, title = {Finite element simulation of single crystal growth process using GSMAC method}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {359--371}, month = mar, } @ARTICLE{KuznetsovNeittanmakiPironneau2003, author = {. and Neittaanm\"{.}, title = {On the convergence of operator--splitting methods}, journal = {Numerical Methods for Scientific Computing}, year = {2003}, } @ARTICLE{CaoLai2002, author = {. and .}, title = {Numerical Experiments of Some Krylow Subspace Methods for Black Oil Model}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {125--141}, month = jul, } @ARTICLE{BroekerGroteMayerReusken2000, author = { .}, title = {Robust Parallel Smoothing for Multigrid Via Sparse Approximate Inverses}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {23}, number = {4}, pages = {1396--1417}, month = nov, note = {Research Report No. 2000-13}, } @ARTICLE{KaushikKeyesSmith1998, author = {. and . and Smith}, title = {On the Interaction of Architecture and Algorithm in the Domain--based Parallelization of an Unstructured--grid Incompressible Flow Code}, journal = {Contemporary Mathematics}, year = {1998}, volume = {Vol. 218}, pages = {311--319}, } @ARTICLE{AlmeidaDerby2000, author = {}, title = {Construction of Solution curves for large two--dimensional problems of steady--state flows of incompressible fluids}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {1}, pages = {285--311}, month = jun, } @ARTICLE{vanKanVuikWesseling2000, author = {, . and .}, title = {Fast pressure calculation for 2D and 3D time dependent incompressible flow}, journal = {Numerical Linear Algebra and Applications}, year = {2000}, volume = {7}, pages = {429--447}, month = may, } @ARTICLE{KobelkovOlshanskii2000, author = { .}, title = {Effective preconditioning of Uzawa type schemes for a generalized {Stokes} problem}, journal = {Numerische Mathematik}, year = {2000}, volume = {86}, pages = {443--470}, month = may, } @ARTICLE{Matusu-NecasovaMedvidova-Lukacova2000, author = {. and Medvidova-Lukacova}, title = {Bipolar Barotropic non--newtonian compressible fluids}, journal = {Mathematical Modelling and Numerical Analysis}, year = {2000}, volume = {34}, number = {5}, pages = {923--934}, month = may, } @ARTICLE{Matusu-NecasovaMedvidova-Lukacova1994, author = {. and Medvidova-Lukacova}, title = {Bipolar barotropic nonnewtonian fluid}, journal = {Comment. Math. Univ. Carolinae}, year = {1994}, volume = {35}, number = {3}, pages = {467--483}, month = sep, } @ARTICLE{AgrawalMandalMarathe2001, author = {Agrawal and . and Marathe}, title = {Computations of laminar and turbulent mixed convection in a driven cavity using pseudo--compressibility approach}, journal = {Computers and Fluids}, year = {2001}, volume = {30}, pages = {607--620}, } @ARTICLE{MeinkeSchroederKrauseRister2002, author = { Schr\"{o}der}, title = {A comparison of second--{} and sixth--order methods for large--eddy simulations}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {695--718}, } @ARTICLE{Stevenson2002, author = {.}, title = {Nonconforming finite elements and the cascadic multi--grid method}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {351--387}, month = sep, } @ARTICLE{PapalexandrisLeonardDimotakis2002, author = {Papalexandris and Leonard and Dimotakis}, title = {Unsplit Algorithms for Multidimensional Systems of Hyperbolic Conservation Laws with Source Terms}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {24--49}, } @ARTICLE{JasakGosman2003, author = { Jasak}, title = {Element residual error estimate for the finite volume method}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {223--248}, } @ARTICLE{MunzRollerKleinGeratz2003, author = {Geratz and . and Munz and Roller}, title = {The extension of incompressible flow solvers to the weakly compressible regime}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {173--196}, } @ARTICLE{John2002, author = {.}, title = {Slip with friction and penetration with resistance boundary conditions for the {Navier}--{Stokes} equations / numerical tests and aspects of the implementation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {287--300}, } @ARTICLE{KoubogiannisAthanasiadisGiannakoglou2003, author = {Athanasiadis and Giannakoglou and Koubogiannis}, title = {One--{} and two--quation turbulence models for the prediction of complex cascade flow using unstructured grids}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {403--430}, } @ARTICLE{DaubeLeQuere2002, author = {Daube and }, title = {Numerical investigation of the first bifurcation for the flow in a rotor--stator cavity of radial aspect ratio 10}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {481--494}, } @ARTICLE{MarquillieEhrenstein2002, author = {Ehrenstein and Marquillie}, title = {Numerical simulation of separating boundary--layer flow}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {683--693}, } @ARTICLE{LyubimovLyubimovaSkuridinChenRoux2002, author = {. and , . and }, title = {Numerical investigation of meniscus deformation and flow in an isothrmal liquid bridge subject to high frequency vibrations under zero gravity conditions}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {663--682}, } @ARTICLE{RubinCelestinaSrinivasan2002, author = { Srinivasan}, title = {Adaptive and fixed segmented domain decomposition multigrid procedures for internal viscous flows}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {787--813}, } @ARTICLE{DoerflerNochetto2002, author = {D\"{ .}, title = {Small data oscillation implies the saturation assumption}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {1--12}, } @ARTICLE{HuangWangWarnecke2002, author = { .}, title = {Error bounds for the large time step Glimm scheme applied to scalar conservation laws}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {31--34}, } @ARTICLE{Leborgne2002, author = {Leborgne}, title = {An optimal consistent stabilization of the inf--sub condition}, journal = {Numerische Mathematik}, year = {2002}, volume = {31}, pages = {35--56}, } @ARTICLE{EfendievWu2002, author = { Wu}, title = {Multiscale finite element for problems with highly oscillatory coefficients}, journal = {Numerische Mathematik}, year = {2002}, volume = {90}, pages = {459--486}, } @ARTICLE{ChouKwak2002, author = {. and .}, title = {Multigrid algorithms for a vertex--centered covolume method for elliptic problems}, journal = {Numerische Mathematik}, year = {2002}, volume = {90}, pages = {441--458}, } @ARTICLE{Buffa2002, author = {.}, title = {Error estimate for a stabilized domain decomposition method with nonmatching grids}, journal = {Numerische Mathematik}, year = {2002}, volume = {90}, pages = {617--640}, } @ARTICLE{ElmanSilvesterWathen2002, author = {. and . and .}, title = {Performance and Analysis of Saddle Point Preconditioners for the Discrete Steady--State {Navier}--{Stokes} Equations}, journal = {Numerische Mathematik}, year = {2002}, volume = {90}, pages = {665--688}, } @ARTICLE{AlfeldSchumaker2002, author = { .}, title = {Smooth macro--elements based on Clough--Tocher triangle splits}, journal = {Numerische Mathematik}, year = {2002}, volume = {90}, pages = {597--616}, } @ARTICLE{CenicerosHou2001, author = { .}, title = {An Efficient Dynamically Adaptive Mesh for Potentially Singular Solutions}, journal = {J. Comput. Phys.}, year = {2001}, volume = {172}, pages = {609--639}, } @ARTICLE{ChenHusseinWan2002, author = {. and }, title = {On a class of new and practival preformance indexes for approximation of fold bifurcations of nonlinear power flow equations}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {119--141}, } @ARTICLE{DostalHaslingerKucera2002, author = { . and Kucera}, title = {Implementation of the fixed point method in contact problems with Coulomb friction based on a dual splitting technique}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {245--256}, } @ARTICLE{AhuesAlmeidaLargillierTitaudVasconcelos2002, author = { Largillier and }, title = {An L1 refined projection approximate solution of the radion transfer equation in stellar atmospheres}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {13--26}, } @ARTICLE{SakuraiSugiura2002, author = {. and Sugiura}, title = {Improvement of convergence of an iterative method for finding polynomial factors of an analytic function}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, number = {713-725}, } @ARTICLE{Swann2002, author = {Swann}, title = {On approximating the solution of the non--stationary {Stokes} equations using the cell discretiziation algorithm}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {727--750}, } @ARTICLE{TiesingaWubsVeldman2002, author = {Tiesinga and . and Veldman}, title = {Bifurcation analysis of incompressible frow in a driven cavity by the Newton--Picard method}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {751--772}, } @ARTICLE{FarhloulFortin2002, author = {. and .}, title = {Review and complements on mixed--hybrid finite element methods for fluid flows}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {140}, pages = {301--313}, } @ARTICLE{Knupp2001, author = {Knupp}, title = {Algebraic mesh quality metrics}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {193--218}, } @ARTICLE{LeRoux2001, author = {, .}, title = {A new triangular finite--element with optimum constraint ratio for compressible fluids}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {66--80}, } @ARTICLE{Chopp2001, author = {Chopp}, title = {Some improvements of the fast marching method}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {230--244}, } @ARTICLE{RapettiToselli2001, author = {. and .}, title = {A feti preconditioner for two dimensional edge element approximations of maxwell`s equations on nonmatching grids}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {92--108}, } @ARTICLE{Krzyzanowski2001, author = {Krzyzanowski}, title = {On block preconditioners for nonsymmetric saddle point problems}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {157--169}, } @ARTICLE{JonesVassilevski2001, author = { .}, title = {AMGe based on element agglomeration}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {109--133}, } @ARTICLE{MichelettiSaccoSaleri2001, author = {. and . and .}, title = {On some mixed finite element methods with numerical integration}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {245--270}, } @ARTICLE{Kupferman2001, author = {Kupferman}, title = {A central--difference scheme for a pure stream function formulation of incompressible viscous flow}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {1--18}, } @ARTICLE{Yu2001, author = {Yu}, title = {A local space--time adaptive scheme in solving two dimensional parabolic problems based on domain decomposition methods}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {1}, pages = {304--322}, } @ARTICLE{KeiterRombach2001, author = {. and .}, title = {Numerical Aspects of FE Simulations of Granular Flows in Silos}, journal = {Journal of Engineering Mechanics}, year = {2001}, volume = {127}, number = {10}, pages = {1044--1050}, month = oct, } @ARTICLE{OoiShe1997, author = {. and She}, title = {Finite element analysis of wall pressure in imperfect silos}, journal = {Int. J. Solids Struct.}, year = {1997}, volume = {34}, number = {16}, pages = {2061--2072}, } @ARTICLE{JofrietNegiLu1997, author = { Lu}, title = {A Numerical Model for Flow of Granular Materials in Silos. Part 3: Parametric Study}, journal = {Journal of Agricultural Engineering Research}, year = {1997}, volume = {68}, pages = {237--246}, } @ARTICLE{ZhangKassabNicholson1997, author = {. and . and .}, title = {A Boundary Element Solution of an Inverse Elasticity Problem and Applications to Determining Residual Stress and Contact Stress}, journal = {Int. J. Solids Struct.}, year = {1997}, volume = {34}, number = {16}, pages = {2073--2086}, } @ARTICLE{MalgouyresGuichard2001, author = {Malgouyres and Guichard}, title = {Edge direction preserving image zooming: A mathematical and numerical analysis}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {1}, pages = {1--37}, } @ARTICLE{Veeser2001, author = {Veeser}, title = {Efficient and reliable a posteriori error estimators for elliptic obstacle problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {1}, pages = {146--167}, } @ARTICLE{ProhlRuzicka2001, author = {. and .}, title = {On fully implicit space--time discretization for motions of incompressible fluids with shear--dependent viscosities: The case p $\leq$ 2}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {1}, pages = {214--249}, } @ARTICLE{CockburnKanschatPerugiaSchoetzau2001, author = {. and . and . and .}, title = {Superconvergence of the Local Discontinuous {Galerkin} Method for Elliptic Problems on Cartesian Grids}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {39}, number = {1}, pages = {364--285}, } @ARTICLE{WohlmuthKrause2001, author = {. and Krause}, title = {A multigrid method based on then unconstrained product space for mortar finite element discretizations}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {1}, pages = {192--213}, } @ARTICLE{LiNavonHussainiLeDimet2003, author = {. and . and .}, title = {Optimal control of cylinder wakes via suction and blowing}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {149--171}, } @ARTICLE{Benes2001, author = {Benes}, title = {Mathematical and computational aspects of solidification of pure substances}, journal = {Acta Mathematica Universitatis Comenianae}, year = {2001}, volume = {LXX}, number = {1}, pages = {123--151}, } @ARTICLE{JeongGoldenfeldDantzig2001, author = {. and .}, title = {Phase field model for three--dimensional dendritic growth with fluid flow}, journal = {Physical Review}, year = {2001}, volume = {64}, pages = {041602-1 - 041602-14}, } @ARTICLE{BoeschMueller-KrumbhaarShochet1995, author = {B\"{o}sch and M\"{u}ller-Krumbhaar and Shochet}, title = {Phase--field models for moving boundary problems: controlling metastability and anisotropy}, journal = {The European Physical Journal}, year = {1995}, volume = {97}, pages = {367--377}, } @ARTICLE{Almgren1999, author = {.}, title = {Second--order phase field asymptotics for unequal conductivities}, journal = {SIAM Journal on Applied Mathematics}, year = {1999}, volume = {59}, number = {6}, pages = {2086--2107}, } @ARTICLE{AndersonMcFaddenWheeler1998, author = {. and Wheeler}, title = {Diffuse--interface methods in fluid mechanics}, journal = {Annual Reviews of Fluid Mechanics}, year = {1998}, volume = {30}, pages = {139--165}, } @ARTICLE{KarmaRappel1996, author = {. and Rappel}, title = {Phase field method for computationally efficient modeling of solidification with arbitrary interface kinetics}, journal = {Physical Review}, year = {1996}, volume = {53}, number = {4}, pages = {3017--3020}, } @ARTICLE{Collins1985, author = {Collins}, title = {Diffuse interface model off diffusion--limited crystal growth}, journal = {Physical Review}, year = {1985}, volume = {31}, number = {9}, pages = {6119--6122}, month = may, } @ARTICLE{Nestler1999, author = {.}, title = {A multiphase--field model: sharp interface asymptotics and numerical simulations of moving phase boundaries and multijunctions}, journal = {Journal of Crystal Growth}, year = {1999}, volume = {204}, pages = {224--228}, } @ARTICLE{KimGoldenfeld2000, author = {Kim and .}, title = {Computation of dendritic microstructure using a level set method}, journal = {Physical Review}, year = {2000}, volume = {62}, number = {2}, pages = {2471--2474}, month = aug, } @ARTICLE{NestlerWheeler1998, author = {}, title = {Anisotropic multi--phase--field model}, journal = {Physical Review}, year = {1998}, volume = {57}, number = {3}, pages = {2602--2608}, month = mar, } @ARTICLE{KarmaRappel1998, author = {. and Rappel}, title = {Quantitative phase--field modeling of dendritic growth in two and three dimensions}, journal = {Physical Review}, year = {1998}, volume = {57}, number = {4}, pages = {4323--4349}, month = apr, } @ARTICLE{GarckeNestlerStoth1999, author = { Stoth}, title = {A multiphase field concept: Numerical simulations of moving ohase boundaries and multiple junctions}, journal = {SIAM Journal on Applied Mathematics}, year = {1999}, volume = {60}, number = {1}, pages = {295--315}, } @ARTICLE{CaginalpXie1993, author = {Caginalp and Xie}, title = {Phase--field and sharp--interface alloy models}, journal = {Physical Review}, year = {1993}, volume = {48}, number = {3}, pages = {1897--1909}, month = sep, } @ARTICLE{ToenhardtAmberg1998, author = { Amberg}, title = {Phase--field simulation of dendritic growth in a shear flow}, journal = {Journal of Crystal Growth}, year = {1998}, volume = {194}, pages = {406--425}, } @ARTICLE{VermolenVuik2000, author = { .}, title = {A mathmatical model for the dissolution of particles in multi--component alloys}, journal = {J. Comput. Appl. Math.}, year = {2000}, volume = {126}, pages = {233--254}, } @ARTICLE{NestlerWheeler2000, author = { Wheeler}, title = {A multi--phase field model of eutectic and peritectic alloys: numerical simulation of growth structures}, journal = {Physica D}, year = {2000}, volume = {138}, pages = {114--133}, } @ARTICLE{LallemandLuo2000, author = {}, title = {Theory of the lattice Boltzmann method: Dispersion, dissipation, isotropy, Galilean invariance, and stability}, journal = {Physical Review}, year = {2000}, volume = {61}, number = {6}, pages = {6546--6562}, month = jun, } @ARTICLE{GimbutasGreengardMinion2001, author = { Minion}, title = {Coulomb interactions on planar structures: Inverting the square root of the Laplacian}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {6}, pages = {2093--2108}, } @ARTICLE{HubersonVoutsinas2002, author = { Voutsinas}, title = {Particles and grid}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {607--625}, } @ARTICLE{RaspoHuguesSerreRandriamampianinaBontoux, author = { Hugues and Randriamampianina and Raspo and Serre}, title = {A spectral projection method for the simulation of complex three--dimensional rotating flows}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {745--767}, } @ARTICLE{HaldenwangPignol2002, author = { Pignol}, title = {Dynamically adepted mesh refinement for combustion front tracking}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {589--606}, } @ARTICLE{KirisKwak2002, author = {. and .}, title = {Aspects of unsteady incompressible flow simultaions}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {627--638}, } @ARTICLE{Canuto2002, author = {.}, title = {Multilevel stabilization device in {CFD}}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {421--436}, } @ARTICLE{PandolfiAmbrosio2002, author = {Ambrosio and Pandolfi}, title = {Performance of upwind methods in predicting shear--like flows}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {725--744}, } @ARTICLE{FerzigerKoseffMonismith2002, author = { }, title = {Numerical simulation of geophysical turbulence}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {557--568}, } @ARTICLE{SabbahForestierPasquetti2002, author = {Forestier and . and Sabbah}, title = {Filtering of the pressure spurious modes in Pn--Pn spectral approximations of the {Stokes} system}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {815--823}, } @ARTICLE{DeAngelisCasciolaPiva2002, author = {Casciola and Piva}, title = {DNS of wall turbulence: dulute polymers and self--sustaining mechanisms}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {495--507}, } @ARTICLE{GarckeNestlerStoth1998, author = {Garcke and . and Stoth}, title = {On anisotropic order parameter models for multi--phase systems and their sharp interface limits}, journal = {Physica D}, year = {1998}, volume = {115}, number = {87-108}, } @ARTICLE{GuBeckermann1999, author = { .}, title = {Simulation of Convection and Macrosegregation in a Large Steel Ingot}, journal = {Metallurgical and Materials Transactions}, year = {1999}, volume = {30A}, pages = {1357--1366}, month = may, } @ARTICLE{LiBeckermann1999, author = { .}, title = {Evolution of the sidebranch structure in free dendritic growth}, journal = {Acta Materialia}, year = {1999}, volume = {47}, number = {8}, pages = {2345--2356}, } @ARTICLE{LiBeckermann1998, author = {Li and .}, title = {Scaling behaviour of three--dimensional dendrites}, journal = {Physical Review}, year = {1998}, volume = {57}, number = {3}, pages = {3176--3188}, month = mar, } @ARTICLE{Vollerbeckermann1999, author = {Voller and beckermann}, title = {A Unified Model of Microsegregation and Coarsening}, journal = {Metallurgical and Materials Transactions}, year = {1999}, volume = {30A}, pages = {2183--2189}, month = aug, } @ARTICLE{GuBeckermannGiamei1997, author = {. and . and Giamei}, title = {Motion and Remelting of Dendrite Fragments during Directional Solidification of a Nickel--Base Superalloy}, journal = {Metallurgical and Materials Transactions}, year = {1997}, volume = {28A}, pages = {1533--1542}, month = jul, } @ARTICLE{SegalVorobevKarpovMakarovMokhovRammRammRownkovVodakovZhamakin1999, author = {. and Vorobev and Karpov and . and Mokhov and Ramm and Ramm and Rownkov and Vodakov and Zhamakin}, title = {Transport phenomena in sublimation growth of SiC bulk crystals}, journal = {Material Science and Engineering}, year = {1999}, volume = {61-62}, pages = {40--43}, } @ARTICLE{VorobevKommisarovSegalMakarovKarpovZhmakinRupp1999, author = { . and Rupp}, title = {Modeling analysis of gas phase nucleation in silicon carbite chemical vapor deposition}, journal = {Material Science and Engineering}, year = {1999}, volume = {61-62}, pages = {176--178}, } @ARTICLE{KarpovMakarovMokhovRammRammRoenkovTalalaevVodakov1997, author = { Makarov, . and Ramm and Ramm and Roenkov and Talalaev and Vodakov}, title = {Analysis of silicon carbite growth by sublimation sandwich method}, journal = {Journal of Crystal Growth}, year = {1997}, volume = {173}, pages = {408--416}, } @ARTICLE{VorobevKarpovZhmakinLovtsusMakarovKrishnan2000, author = { . and Krishnan}, title = {Effect of gas--phase nucleation on chemical vapor deposition of silicon carbite}, journal = {Journal of Crystal Growth}, year = {2000}, volume = {211}, pages = {343--346}, } @ARTICLE{FellerBeckermann1997, author = { .}, title = {Modeling of Solidification of Metal--Metrix Particulate Composites with Convection}, journal = {Metallurgical and Materials Transactions}, year = {1997}, volume = {28B}, pages = {1165--1183}, month = dec, } @ARTICLE{Scheichl2002, author = {Scheichl}, title = {Decoupling three--dimensional mixed problems using divergence--free finite elements}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {23}, number = {5}, pages = {1752--1776}, } @ARTICLE{Hager2002, author = {Hager}, title = {Minimizing the profile of a symmetric matrix}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {23}, number = {5}, pages = {1799--1816}, } @ARTICLE{CaramanaShashkov1998, author = { .}, title = {Elimination of Artificial Grid Distortion and Hourglass--Type Motion by means of {L}agrangian Subzonal Masses and Pressures}, journal = {J. Comput. Phys.}, year = {1998}, volume = {142}, pages = {521--561}, } @ARTICLE{Borcea2002, author = {.}, title = {Electrical Impedance Tomography}, journal = {Inverse Probl.}, year = {2002}, volume = {18}, pages = {R99-R136}, note = {Institute of Physics Publishing}, } @ARTICLE{FernandezHildViano2002, author = { Hild and Viano}, title = {Numerical Approximation of the Elastic--Viscoplastic Contact Problem with Non--matching Meshes}, journal = {Numerische Mathematik}, year = {2002}, volume = {Digital Object Identifier (DOI) 10.1007/s00211-002-0441-x}, month = oct, note = {Published Online}, } @ARTICLE{BrezinskiRedivo-ZagliaRodriguezSeatzu2002, author = { and Seatzu}, title = {Multi--parameter regularization techniques for ill--conditioned linear systems}, journal = {Numerische Mathematik}, year = {2002}, volume = {Digital Object Identifier (DOI) 10.1007/s00211-002-0435-8}, month = oct, note = {Published Online}, } @ARTICLE{Moore2002, author = {.}, title = {Applications of Lobatto polynomials to an adaptive finite element method: A posteriori error estimates for HP--Adaptivity and Grid--to--Grid Interpolation}, journal = {Numerische Mathematik}, year = {2002}, volume = {Digital Object Identifier (DOI) 10.1007/s00211-002-0424-y}, month = oct, note = {Published Online}, } @ARTICLE{KarbsteinSchubert1995, author = { Schubert}, title = {Development in the continuous mechanical production of oil--in--water macro--emulsions}, journal = {Chemical Engineering and Processing}, year = {1995}, volume = {34}, pages = {205--211}, } @ARTICLE{SchneiderBeckermannBoettingerKattner1997, author = { }, title = {Modelling of Micro--{} and Macrosegregation and Freckle Formation in Single--Crystal Nickel--Base Superalloy Directional Solidification}, journal = {Metallurgical and Materials Transactions}, year = {1997}, volume = {28A}, pages = {1517--1531}, } @ARTICLE{IhleMueller-Krumbhaar1994, author = {Ihle and M\"{u}ller-Krumbhaar}, title = {Fractal and compact growth in phase transactions with diffusion transport}, journal = {Physical Review}, year = {1994}, volume = {49}, number = {4}, pages = {2972--2990}, month = apr, } @ARTICLE{WheelerBoettingerMcFadden1993, author = {Wheeler and Boettinger and McFadden}, title = {Phase--field model of solute trapping during solidification}, journal = {Physical Review}, year = {1993}, volume = {47}, number = {3}, pages = {1893--1909}, month = mar, } @ARTICLE{WheelerBoettingerMcFadden1992, author = { Boettinger and McFadden}, title = {Phase--field model for isothermal phase transitions in binary alloys}, journal = {Physical Review}, year = {1992}, volume = {45}, number = {10}, pages = {7424--7438}, } @ARTICLE{AndersonMcFaddenWheeler2000, author = {. and }, title = {A phase--field model of solidification with convection}, journal = {Physica D}, year = {2000}, volume = {135}, pages = {175--194}, } @ARTICLE{BoettingerCoriellGreerKarmaKurzRappazTrivedi2000, author = { }, title = {Solidification microstructures: Recent developments, future directions}, journal = {Acta Materialia}, year = {2000}, volume = {48}, pages = {43--70}, } @ARTICLE{SteinbachBeckermannKauerhofLiGuo1999, author = {. and .}, title = {Three--dimensional modeling of equiaxed dendritic growth on a mesoscopic scale}, journal = {Acta Materialia}, year = {1999}, volume = {47}, number = {3}, pages = {971--982}, } @ARTICLE{TonogBeckermannKarma2000, author = { .}, title = {Velocity and shape selection of dendritic crystals in a forced flow}, journal = {Physical Review}, year = {2000}, volume = {61}, number = {1}, pages = {R49-R52}, } @ARTICLE{ScardovelliZaleski1999, author = {}, title = {Direct numerical simulation of free--surface and interfacial flow}, journal = {Annual Reviews of Fluid Mechanics}, year = {1999}, volume = {31}, pages = {567--603}, } @ARTICLE{SankaranarayananShanKevrekidisSundaresan2002, author = { Sundaresan}, title = {Analysis of drag and virtual mass forces in a bubble suspension using an implicit formulation of the lattice Boltzmann method}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {452}, pages = {61--96}, } @ARTICLE{SankaranarayananKevrekidisSundaresanLuTryggvason2003, author = { .}, title = {A comparative study of lattice Boltzmann and front--tracking finite--difference methods for bubble simulations}, journal = {International Journal of Multiphase Flow}, year = {2003}, volume = {29}, pages = {199--116}, } @ARTICLE{HolmgrenBrandenSterner2000, author = {Holmgren and Sterner}, title = {Convergence acceleration for the linearized {Navier}--{Stokes} equations using semicirculant approximations}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, number = {4}, pages = {1524--1550}, } @ARTICLE{UtkuCarey1982, author = {Utku and .}, title = {Boundary Penalty techniques}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1982}, volume = {30}, pages = {103--118}, } @ARTICLE{LaytonLeePeterson1998, author = {. and .}, title = {Numerical solution of the stationary {Navier}--{Stokes} equations using a multilevel Finite Element method}, journal = {SIAM Journal on Scientific Computing}, year = {1998}, volume = {20}, number = {1}, pages = {1--12}, } @ARTICLE{Xu2001, author = {Xu}, title = {The method of subspace corrections}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {335--362}, } @ARTICLE{Stueben2001, author = {St\"{u}ben}, title = {A review of algebraic multigrid}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {281--309}, } @ARTICLE{SilvesterElmanKayWathen2001, author = {. and . and . and .}, title = {Efficient preconditioning of the linearized {Navier}--{Stokes} equations for incompressible flow}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {261--279}, } @ARTICLE{WesselingOosterlee2001, author = {. and .}, title = {Geometric multigrid with applications to computational dynamics}, journal = {J. Comput. Appl. Math.}, year = {2001}, volume = {128}, pages = {311--334}, } @ARTICLE{SundaresanEatonKochOttino2003, author = {Sundaresan and Eaton and Koch and Ottino}, title = {Appendix 2: Report of study group on disperse flow}, journal = {International Journal of Multiphase Flow}, year = {2003}, volume = {29}, pages = {1069--1087}, } @ARTICLE{HanrattyTheofanousDelhayeEatoonMcLaughlinProsperettiSundaresanTryggvason2003, author = { .}, title = {Workshop Findings}, journal = {International Journal of Multiphase Flow}, year = {2003}, volume = {29}, pages = {1037--1059}, } @ARTICLE{LeRoux2000a, author = {.}, title = {Numerical Solution of nonlinear reaction diffusion processes}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {37}, number = {5}, pages = {1644--1656}, } @ARTICLE{LiuE2000, author = {}, title = {Simple finite element method in vorticity formulation for incompressible flows}, journal = {Mathematics of Computation}, year = {2000}, volume = {70}, number = {234}, pages = {597--593}, note = {DOI: S 0025-5718(00)01239-4}, } @ARTICLE{DeviKandasamy2000, author = {Devi and Kandasamy}, title = {Effects of Chemical Reaction, Heat and Mass Transfer onm MHD Flow past a Semi Infinite Plate}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2000}, volume = {80}, number = {10}, pages = {697--701}, } @ARTICLE{Kaminski2001, author = {.}, title = {Stochastic Problem of Viscous Incompressible Fluid Flow with Heat Transfer}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2001}, volume = {81}, number = {12}, pages = {827--837}, } @ARTICLE{ChenDai2001, author = {. and .}, title = {Adaptive galerkin method with error control for a dynamical Ginzburg--Landau model in superconductivity}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {38}, number = {6}, pages = {1961--1985}, } @ARTICLE{PearsonTardy2002, author = {Pearson and Tardy}, title = {Models for flow of non--Newtonian and complex fluids through porous media}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {102}, pages = {447--473}, } @ARTICLE{CoutandShkoller2001, author = {.}, title = {Well--posedness of the full Ericksen--Leslie model of nematic liquid crystals}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2001}, volume = {333}, number = {1}, pages = {919--924}, note = {french article}, } @ARTICLE{ELiu2001, author = {}, title = {Projection method III: Spatial discretization on the staggered grid}, journal = {Mathematics of Computation}, year = {2001}, volume = {71}, number = {237}, pages = {27--47}, } @ARTICLE{Reimer2003, author = {.}, title = {Approximation of Density Functions and Reconstruction of the Approximant}, journal = {International Series of Numerical Mathematics}, year = {2003}, volume = {145}, pages = {253--261}, } @ARTICLE{WallHolst2001, author = {}, title = {Estimation of parameters ini viscoplastic and creep material models}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {5}, pages = {2080--2103}, } @ARTICLE{BramblePasciakSteinbach2001, author = { .}, title = {On the stability of the L2 projection in H1(Omega)}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {237}, pages = {147--156}, month = may, } @ARTICLE{Carstensen2001, author = {.}, title = {Merging then Bramble--Pasciak--Steinbach and the Crouzeix--Thomee criterion for H1--stability of the L2--projection onto finite elements}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {237}, pages = {157--163}, month = may, } @ARTICLE{PavarinoWidlund2002, author = { .}, title = {Balancing Neumann--Neumann Methods for Incompressible {Stokes} Equations}, journal = {Communications on Pure and Applied Mathematics}, year = {2002}, volume = {LV}, pages = {302--335}, } @ARTICLE{HuilgolMenaPiau2002, author = {Huilgol and Piau}, title = {Finite stopping time problems and rheometry of Bingham fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {102}, pages = {97--107}, } @ARTICLE{DelilloIsakovValdiviaWang2001, author = {Delillo and Isakov and Valdivia and Wang}, title = {The detection of the source of acoustical noise in two dimensions}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {6}, pages = {2104--2121}, } @ARTICLE{GolseSaint-Raymond2001, author = { Saint-Raymond}, title = {The {Navier}--{Stokes} limit for the Boltzmann equation}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2001}, volume = {333}, number = {I}, pages = {897--902}, note = {french article}, } @ARTICLE{AinsworthCoggins2000, author = {. and .}, title = {The stability of mixed hp--finite element methods for stokes flow on high aspect ratio elements}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {5}, pages = {1721--1761}, } @ARTICLE{Sangalli2000, author = {.}, title = {Global and local error analysis for the residual--free bubbles method applied to advection--dominated problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {5}, pages = {1496--1522}, } @ARTICLE{CohenJolyRobertsTordjman2001, author = {. and . and Tordjman}, title = {Higher order triangurlar finite elements with mass lumping for the wave equation}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {38}, number = {6}, pages = {2047--2078}, } @ARTICLE{HeldDervieux2002, author = { Held}, title = {One--shot airfoil optimisation without adjoint}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {1015--1049}, } @ARTICLE{Mompean2002, author = {Mompean}, title = {On predicting abrupt contradiction flows with differential and algebraic viscoelastic models}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {935--956}, } @ARTICLE{SeppaenenVauhkonenVauhkonenSomersalolKaipio2001, author = { .}, title = {State estimation with fluid dynamical avolution models in process tomography {}--{} an application to impedance tomography}, journal = {Inverse Probl.}, year = {2001}, volume = {17}, pages = {467--483}, } @ARTICLE{KaipioKolehmainenSomersaloVauhkonen2000, author = {. and Kolehmainen and Vauhkonen}, title = {Statical inversion and Monte Carlo sampling methods in electrical impedance tomography}, journal = {Inverse Probl.}, year = {2000}, volume = {16}, pages = {1487--1522}, } @ARTICLE{BaroudiKaipioSomersalo1998, author = { .}, title = {Dynamical electric wire tomography: a time series approach}, journal = {Inverse Probl.}, year = {1998}, volume = {14}, pages = {799--813}, } @ARTICLE{Morgan2002, author = {.}, title = {GMRES with deflated restarting}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {1}, pages = {20--37}, } @ARTICLE{AdamsLi2002, author = {. and Li}, title = {The immersed interface/multigrid methods for interface problems}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {463--479}, } @ARTICLE{ItoLi2003, author = { Li}, title = {Solving a Nonlkinear Problem in Magneto--Rheological Fluids using the Immersed Interface Method}, journal = {Journal of Scientific Computing}, year = {2003}, volume = {19}, number = {1-3}, pages = {253--266}, } @ARTICLE{GlimmGroveZhang2002, author = {. and . and .}, title = {Interface Tracking for Axisymmetric Flows}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {1}, pages = {208--236}, } @ARTICLE{KandlikarJoshiTian2993, author = {. and Tian}, title = {Effect of Surface Roughness on Heat Transfer and Fluid Flow Characteristics at Low Reynolds Numbers in Small Diameter Tubes}, journal = {Heat Transfer Engineering}, year = {2993}, volume = {24}, number = {3}, pages = {4--16}, } @ARTICLE{KayLoghinWathen2002, author = {. and . and .}, title = {A Preconditioner for the Steady--State {Navier}--{}--{Stokes} Equations}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {1}, pages = {237--256}, } @ARTICLE{RazelosKrikkis2003, author = {}, title = {The Optimum Design of Radiating and Convective--Radiating Circular Fins}, journal = {Heat Transfer Engineering}, year = {2003}, volume = {24}, number = {3}, pages = {17--41}, } @ARTICLE{ChenDai2002, author = {. and .}, title = {On the efficiency of adaptive finite element methods for elliptic problems with discontinuous coefficients}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {443--462}, } @ARTICLE{Castillo2002, author = {.}, title = {Performance of discontinuous {Galerkin} methods for elliptic PDE`s}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {524--547}, } @ARTICLE{ChangHuang2002, author = {. and Huang}, title = {Efficient algebraic multigrid and their convergence}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {597--618}, } @ARTICLE{MatthiesTobiska2002, author = {. and .}, title = {The Inf--Sup Condition for the Mapped Q\_k--P\_(k--1)\^{} disc Element in Arbitrary Space Dimensions}, journal = {Computing}, year = {2002}, volume = {69}, pages = {119--139}, } @ARTICLE{DoerflerRumpf1998, author = { .}, title = {An adaptive strategy for elliptic problems including a--posteriori controlled boundary approximation}, journal = {Mathematics of Computation}, year = {1998}, volume = {67}, number = {224}, pages = {1361--1382}, } @ARTICLE{SheuTsai2002, author = {}, title = {Flow topology in a steady three--dimensional lid--driven cavity}, journal = {Computers and Fluids}, year = {2002}, volume = {31}, pages = {911--934}, } @ARTICLE{AboubacarMatallahTamaddon-JahromiWebster2002, author = { .}, title = {Numerical prediction of extensional flow in contradiction geometries: hybrid finite volume/element method}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {104}, pages = {125--164}, } @ARTICLE{HillCox2002, author = { .}, title = {On the problem of the determination of force distributions in granular heaps using continuum theory}, journal = {Q. J. Mech. Appl. Math.}, year = {2002}, volume = {55}, number = {4}, pages = {655--668}, } @ARTICLE{TomeMangiavacchiCuminatoCasteloMcKee2002, author = {. and . and . and }, title = {A finite difference technique for simulating unsteady viscoelastic free surface flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {106}, pages = {61--106}, } @ARTICLE{Rasmussen2002, author = {Rasmussen}, title = {Lagrangian Viscoelastic Flow Computations Using a Generalized Molecular Stress Function Model}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {106}, pages = {107--120}, } @ARTICLE{FadiliTardyPearson2002, author = {Fadili and }, title = {A 3D filtration law for power--law fluids in heterogeneous porous media}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {106}, pages = {121--146}, } @ARTICLE{YuPhan-ThienFanTanner2002, author = {Yu and .}, title = {Viscoelastic mobility problem of a system of particles}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {104}, pages = {87--124}, } @ARTICLE{ArnoldBrezziCockburnMarini2002, author = {. and . and . and Marini}, title = {Unified Analysis of Discontinuous {Galerkin} Methods for Elliptic Problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {39}, number = {5}, pages = {1749--1779}, } @ARTICLE{KnapekKoster2002, author = { Koster}, title = {Integral operators on space grids}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {39}, number = {5}, pages = {1794--1809}, } @ARTICLE{DegondLemouPicasso2002, author = {. and . and .}, title = {Viscoelastic fluid model derived from kinetic quations for polymers}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1501--1519}, } @ARTICLE{Chadwick2002, author = {Chadwick}, title = {Axisymmetric identation of a this incompressible elastic layer}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1520--1530}, } @ARTICLE{vanDuijnMikelicPop2002, author = {. and . and .}, title = {Effective equations for two--phase flow with trapping on the micro scale}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1531--1568}, } @ARTICLE{CookNwankoSchleinigerWood2002, author = {Cook and Nwanko and Schleiniger and Wood}, title = {Rapid, unambiguous polymer characterization by flow--referenced capillary viscometry}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1657--1676}, } @ARTICLE{HorganSaccomandiSgura2002, author = { .}, title = {A two--point boundary--value problem for the axial shear of hardening isotropic incompressible nonlinearly elastic materials}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1712--1727}, } @ARTICLE{BochevLiaodelaPena1996, author = {. and . and .}, title = {Analysis and Computation of Adaptive Moving Grids by Deformation}, journal = {Numerical Methods for Partial Differential Equations}, year = {1996}, volume = {12}, pages = {489ff}, } @ARTICLE{Chorin1968, author = {Chorin}, title = {Numerical solution of the {Navier}--{Stokes} equations}, journal = {Math. Comput.}, year = {1968}, volume = {22}, pages = {745--762}, } @ARTICLE{GriebelZumbusch1999, author = {. and .}, title = {Parallel multigrid in an adaptive {PDE} solver based on hashing and space--filling curves}, journal = {Parallel Computing}, year = {1999}, volume = {25}, pages = {827--843}, } @ARTICLE{RannacherBecker1996, author = {. and .}, title = {A Feed--Back Approach to Error Control in Finite Element Methods: Basic Analysis and Examples}, journal = {East-West Journal of Numerical Mathematics}, year = {1996}, volume = {4}, pages = {237--264}, } @ARTICLE{VanKan1986, author = {.}, title = {A second--order accurate pressure--correction scheme for viscous incompressible flow}, journal = {SIAM Journal on Scientific and Statistical Computing}, year = {1986}, volume = {7}, pages = {870--891}, } @ARTICLE{PanGlowinski2002, author = {Pan and .}, title = {Direct Simulation of the Motion of Neutrally Buoyant Circular Cylinders in Plane Poiseuille Flow}, journal = {J. Comput. Phys.}, year = {2002}, volume = {181}, pages = {260--279}, } @ARTICLE{JonesKojouharovLeSmith2002, author = {Jones and Kojouharov and }, title = {Bacterial wall attachment in a flow reactor}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1728--1771}, } @ARTICLE{PreusserRumpf2002, author = {. and .}, title = {A Level Set method for anisotropic geometric diffusion in 3D image processing}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {5}, pages = {1772--1793}, } @ARTICLE{Lesnic2002, author = {Lesnic}, title = {The decomposition method for forward and backward time--dependent problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {27--39}, } @ARTICLE{AzulayGambini2002, author = {}, title = {Which pivot to solve linear systems ?}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {471--484}, } @ARTICLE{OlshanskiiReusken2002, author = {. and .}, title = {{Navier}--{Stokes} equations in rotation form: A robust multigrid solver for the velocity problem}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {23}, number = {5}, pages = {1683--1706}, } @ARTICLE{AchdouJaphetMadayNataf2002, author = { .}, title = {A new cement to glue non--conforming grids with Robin initerface conditions: The finite volume case}, journal = {Numerische Mathematik}, year = {2002}, volume = {92}, pages = {593--620}, note = {DOI 10.1007/s002110100336}, } @ARTICLE{BarrenecheaValentin2002, author = { Valentin}, title = {An unusual stabilized finite element method for as generalized {Stokes} problem}, journal = {Numerische Mathematik}, year = {2002}, volume = {92}, pages = {653--677}, note = {DOI 10.1007/s002110100371}, } @ARTICLE{Veeser2002, author = {Veeser}, title = {Convergent adaptive finite elements for the nonlinear Laplacian}, journal = {Numerische Mathematik}, year = {2002}, volume = {92}, pages = {743--770}, } @ARTICLE{Wang2002, author = {Wang}, title = {On the quadratic finite element approximation to the obstacle problem}, journal = {Numerische Mathematik}, year = {2002}, volume = {92}, pages = {771--778}, } @ARTICLE{XuePhan-ThienTanner2002, author = { . and .}, title = {Upwinding with deferred correction (UPDC): an effective implementation of higher--order convection schemes for implicit finite volume methods}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {108}, pages = {1--24}, } @ARTICLE{ZatloukalVlcekTzoganakisSaha2002, author = {Zatloukal and and Saha}, title = {Improvement in techniques for the determination of extensional rheological data from entrance flows: computational and experimental analysis}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {107}, pages = {13--37}, } @ARTICLE{HyunKimAhnLee2002, author = { Lee}, title = {Large amplitude oscillatory shear as a way to classify the complex fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {107}, pages = {51--65}, } @ARTICLE{Greco2002, author = {.}, title = {Drop deformation for non--Newtonian fluids in slow flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {107}, pages = {111--131}, } @ARTICLE{Codina1999, author = {.}, title = {Numerical Solution of the Incompressible {Navier}--{Stokes} Equations with Coriolis Forced Based on the Discretization of the Total Time Derivative}, journal = {J. Comput. Phys.}, year = {1999}, volume = {148}, pages = {567--496}, } @ARTICLE{HaaseKuhnReitzinger2002, author = { }, title = {Parallel algebraic multigrid methods on distributed memory computers}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {410--427}, } @ARTICLE{MishraTalukdarTrimisDurst2003, author = { .}, title = {Computational efficiency improvements of the radiative transfer problems with or without conduction {}--{} a comparison of the collapsed dimension method and the discrete transfer method}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {46}, pages = {3083--3095}, } @ARTICLE{KaloniLou2002, author = {}, title = {On the stability of thermally driven shear flow of an Oldroyd--B fluid heated from below}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {107}, pages = {97--110}, } @ARTICLE{ReeseSolomativ2002, author = {. and Solomativ}, title = {Mean field heat transfer scaling for non--Newtonian stagnant lid convection}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {107}, pages = {39--49}, } @ARTICLE{OjiroNikiUsui2003, author = { Usui}, title = {A new criterion for the H--matrix property}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {150}, pages = {293--302}, } @ARTICLE{RoosZarin2003, author = {. and Zarin}, title = {The streamline--diffusion method for a convection--diffusion problem with a point source}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {150}, pages = {109--128}, } @ARTICLE{GuvenBayatitoglu2003, author = {Bayatitoglu and Guven}, title = {The radiative transfer solution of a rectangular enclosure using angular domain discrete wavelets}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {46}, pages = {687--694}, } @ARTICLE{BorchersBuschSololichinEigenberger1999, author = { .}, title = {Applicability of the standard k--Epsilon turbulence model to the dynamic simulation of bubble columns. Part II: Comparison of detailed experiments and flow simulations}, journal = {Chemical Engineering Science}, year = {1999}, volume = {54}, pages = {5927--5935}, } @ARTICLE{SokolichinEigenberger1999, author = {. and .}, title = {Applicability of the standard k--Epsilon turbulence model to the dynamic simulation of bubble columns. Part I: Detailed numerical simulations}, journal = {Chemical Engineering Science}, year = {1999}, volume = {54}, pages = {2273--2284}, } @ARTICLE{SokolichinEigenberger1997, author = {. and .}, title = {Dynamic numerical simulation of gas--liquid two--phase flows}, journal = {Chemical Engineering Science}, year = {1997}, volume = {52}, number = {4}, pages = {611--626}, } @ARTICLE{BeckerSokolichinEigenberger1994, author = {. and . and .}, title = {Gas--liquid flow in bubble columns and loop reactors: Part II. Comparison of detailed experiments and flow simulations}, journal = {Chemical Engineering Science}, year = {1994}, volume = {49}, number = {24B}, pages = {5747--5762}, } @ARTICLE{SokolichinEigenberger1994, author = {. and .}, title = {Gas--liquid flow in bubble columns and loop reactors: Part I. Detailed modelling and numerical simulation}, journal = {Chemical Engineering Science}, year = {1994}, volume = {49}, number = {24B}, pages = {5735--5746}, } @ARTICLE{Lui2002, author = {.}, title = {On linear monotone iteration and Schwarz methods for nonlinear elliptic PDEs}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {109--129}, note = {DOI 10.1007/s002110100349}, } @ARTICLE{KaltenbacherSchicho2002, author = {. and .}, title = {A multi--grid method with a priori and a posteriori level choice for the regularization of nonlinear ill--posed problems}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {77--107}, note = {DOI 10.1007/s002110100375}, } @ARTICLE{GaiffeGlowinskiMasson2002, author = {. and . and .}, title = {Domain decomposition and splitting methods for Mortar mixed finite element approximations to parabolic equations}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {53--75}, note = {DOI: 10.1007/s002110100367}, } @ARTICLE{Globisch2003, author = {Globisch}, title = {The hierarchical preconditioning on unstructured three--dimensional grids with locally refined regions}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {150}, pages = {265--282}, } @ARTICLE{ZhouFlamantGauthierLu2002, author = { Flamant and Gauthier and Lu}, title = {{L}agrangian Approach for Simulating the Gas--Particle Flow Structure in a Circulating Fluidized Bed Riser}, journal = {International Journal of Multiphase Flow}, year = {2002}, volume = {28}, pages = {1801--1821}, } @ARTICLE{Dax2003, author = {.}, title = {The Orthogonal Razleigh Quotient Iteration (ORQI) method}, journal = {Linear Algebra and\_its Applications}, year = {2003}, volume = {358}, pages = {23--43}, } @ARTICLE{KnyazevNeymeyr2003, author = {. and Neymeyr}, title = {A geometric theory for preconditioned inverse iteration III: A short and sharp convergence estimate for generalized eigenvalue problems}, journal = {Linear Algebra and\_its Applications}, year = {2003}, volume = {358}, number = {95-114}, } @ARTICLE{Zhuravlev2002, author = {Zhuravlev}, title = {A model of the mechanism of motion of a snake}, journal = {Journal of Applied Mathematics and Mechanics}, year = {2002}, volume = {66}, number = {4}, pages = {523--527}, } @ARTICLE{PhillipsWilliams2002, author = {}, title = {Comparison of creeping and internial flow of an Oldroyd B fluid through planar and axisymmetric contractions}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {108}, pages = {24--47}, } @ARTICLE{vanOsGerritsma2002, author = { and Gerritsma}, title = {A varaible order spectral element scheme applied to the velocity--pressure--total--stress formulation of the upper convected Maxwell model}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {108}, pages = {73--97}, } @ARTICLE{JendrejackdePabloGraham2002, author = { and .}, title = {A method for multiscale simulation of flowing complex fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {123}, pages = {123--142}, } @ARTICLE{LeeShaqfehKhomami2002, author = { .}, title = {A study of viscoelastic free surface flows by the finite element method: Hele--Shaw and slot coating flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {108}, pages = {327--362}, } @ARTICLE{Harlen2002, author = {Harlen}, title = {The negative wake behind a sphere sedimenting through a viscoelastic fluid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {108}, pages = {411--430}, } @ARTICLE{MenaldiSritharan2003, author = {}, title = {Stochastic 2--D {Navier}--{Stokes} Equation}, journal = {Applied Mathematics and Optimization}, year = {2003}, volume = {46}, pages = {31--53}, note = {DOI: 10.1007/s00245-002-0734-6}, } @ARTICLE{BruGinestarMarinVerduMasManteuffel2002, author = { .}, title = {Iterative Schemes for the Neuron Diffusion Equation}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {1307--1323}, } @ARTICLE{He2002, author = {He}, title = {Coupling Boundary Integral and Finite Element Methods for the Oseen Coupled Problem}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {1413--1429}, } @ARTICLE{FontelosFriedmanHu2002, author = {Fontelos and }, title = {Mathematical analysis of a model for the iteration of Angiogenesis}, journal = {SIAM Journal on Mathematical Analysis}, year = {2002}, volume = {33}, number = {6}, pages = {1330--1355}, } @ARTICLE{FiloPluschke2002, author = {}, title = {A free boundary problem in dermal drug delivery}, journal = {SIAM Journal on Mathematical Analysis}, year = {2002}, volume = {33}, number = {6}, pages = {1430--1454}, } @ARTICLE{MouginMagnaudet2002, author = { Magnaudet}, title = {The generalized Kirchhoff equations and their application to the interaction between a rigid body and an arbitrary time--dependent viscous flow}, journal = {International Journal of Multiphase Flow}, year = {2002}, volume = {28}, pages = {1837--1851}, } @ARTICLE{DanilovOmelyanov2002, author = {Danilov and Omelyanov}, title = {Dynamics of the interface between two immiscible liquids with nearly equal densities unter gravity}, journal = {Eur. J. Appl. Math.}, year = {2002}, volume = {13}, pages = {497--516}, } @ARTICLE{GauthierGondretMoidyRabaud2002, author = { Rabaud}, title = {Instabilities in the flow between co--{} and counter--rotating disks}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {473}, pages = {1--12}, note = {DOI: 10.1017/S0022112002002525}, } @ARTICLE{BarkleyGomesHenderson2002, author = { Henderson}, title = {Three--dimensional instability in a flow over a backward--facing step}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {473}, pages = {167--190}, } @ARTICLE{OteroWittenbergWorthingDoering2002, author = { }, title = {Bounds on Rayleigh--Benard convection with an imposed heat flux}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {473}, pages = {191--199}, } @ARTICLE{Saif2002, author = {Saif}, title = {On the capillary interaction between solid plates forming menisci on the surface of a liquid}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {473}, pages = {321--347}, } @ARTICLE{BagchiBalachandar2002, author = {. and .}, title = {Shear versus vortex--induced lift force on a rigid sphere at moderate Re}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {473}, pages = {379--288}, } @ARTICLE{WillisBarenghi2002, author = {Willis and Barenghi}, title = {Hydromagnetic Taylor--Couette flow: wavy modes}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {472}, pages = {399--410}, } @ARTICLE{ParkChoiYooKim2003, author = { .}, title = {Streamline upwind numerical simulation of two--dimensional confined impinging slot jets}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {46}, pages = {251--262}, } @ARTICLE{CarpenBrady2002, author = { .}, title = {Gravitational instability in suspension flow}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {472}, pages = {201--210}, } @ARTICLE{ThaokarKumaran2002, author = { .}, title = {Stability of fluid flow past a membrane}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {472}, pages = {29--50}, } @ARTICLE{IbragimovRjasanow2002, author = {}, title = {Numerical solution of the Boltzmann equation on the uniform grid}, journal = {Computing}, year = {2002}, volume = {69}, pages = {163--186}, } @ARTICLE{DeJongJacobi2003, author = {DeJong and Jacobi}, title = {Localized flow and hear transfer interactions in louvered--fin arrays}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {46}, pages = {443--455}, } @ARTICLE{KhoromskijMelenk2002, author = { .}, title = {An Efficient Direct Solver for the boundary Concentrated {FEM} in 2D}, journal = {Computing}, year = {2002}, volume = {69}, pages = {91--117}, } @ARTICLE{BabuskaChleboun2001, author = {. and .}, title = {Effects of Uncertainties in the Domain on the Solution of {N}eumann Boundary Value Problems in Two Spatial Dimensions}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {240}, pages = {1339--1370}, } @ARTICLE{AliabadiJohnsonAbedi2003, author = {Abedi and Aliabadi and Johnson}, title = {Comparison of finite element and pendulum models for simulating of sloshing}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {535--545}, } @ARTICLE{BernardiHecht2001, author = {}, title = {Error Indicators for the Mortar Finite Element discretization of the Laplace equation}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {240}, pages = {1371--1403}, } @ARTICLE{Stefanelli2002, author = {Stefanelli}, title = {Analysis of a variable time--step discretization of the three--dimensional Fremond Model for shape memory alloys}, journal = {Math. Comput.}, year = {2002}, volume = {71}, number = {240}, pages = {1431--1453}, } @ARTICLE{GaticaHeuer2001a, author = {. and Heuer}, title = {Conjugate Gradient Method for dual--dual mixed formulations}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {240}, pages = {1455--1472}, } @ARTICLE{CaulfieldWoods2002, author = {}, title = {The mixing in a room by a localized finite--mass--flux source of buoyancy}, journal = {Journal of Fluid Mechanics}, year = {2002}, volume = {471}, pages = {33--50}, } @ARTICLE{Cao2001, author = {.}, title = {Convergence of block two--stage iterative methods for symmetric positive definite systems}, journal = {Numerische Mathematik}, year = {2001}, volume = {90}, pages = {47--63}, } @ARTICLE{KimXuZikatanov2003, author = { .}, title = {A multigrid method based on graph matching for convection--diffusion equations}, journal = {Numerical Linear Algebra and Applications}, year = {2003}, volume = {10}, pages = {181--195}, note = {DOI: 10.1002/nla.317}, } @ARTICLE{AmaraDabaghi2001, author = {. and Dabaghi}, title = {An optimal C0 finite element algorithm for the 2D biharmonic problem: theoretical analysis and numerical results}, journal = {Numerische Mathematik}, year = {2001}, volume = {90}, pages = {19--40}, note = {DOI: 10.1007/s002110100284}, } @ARTICLE{CarstensenProhl2001, author = {. and .}, title = {Numerical analysis of relaxed micromagnetics by penalised finite elements}, journal = {Numerische Mathematik}, year = {2001}, volume = {90}, pages = {65--99}, note = {DOI: 10.1007/s002110100268}, } @ARTICLE{Chesters1991, author = {Chesters}, title = {The modelling of coalescence processes in fluid--liquid dispersions: A Review of Current Understanding}, journal = {Transactions IChemE}, year = {1991}, volume = {69}, number = {A}, pages = {259--270}, month = jul, } @ARTICLE{HanHuang2002, author = { Huang}, title = {Exact and Approximating Boundary Conditions for the Parabolic Problems on Unbounded Domains}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {655--666}, } @ARTICLE{BrainmanToledo2002, author = {Brainman and .}, title = {Nested--Dissection ordering for sparse LU with partial pivoting}, journal = {SIAM Journal on Matrix Analysis and Applications}, year = {2002}, volume = {23}, number = {4}, pages = {998--1012}, } @ARTICLE{CampagnaKobbeltSeidel1998, author = { . and .}, title = {Directed Edges {}--{} A Scalable Representation for Triangle Meshes}, journal = {Journal of Graphics Tools}, year = {1998}, volume = {3}, number = {4}, pages = {1--12}, } @ARTICLE{KuzminMoellerTurek2003a, author = {. and . and .}, title = {Multidimensional {FEM}--FCT schemes for arbitrary time--stepping}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {42}, pages = {265--295}, } @ARTICLE{ZietheSpohnTurek2003, author = {. and . and .}, title = {Core formation: A New Modelling Approach}, journal = {ESA}, year = {2003}, volume = {SP-514}, pages = {137--143}, } @ARTICLE{KuzminMoellerTurek2003b, author = {. and .}, title = {High--Resolution {FEM}--FCT Schemes for Multidimensional Conservation Laws {}}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, number = {45-47}, pages = {4915--4946}, month = may, } @ARTICLE{KuzminTurek2003c, author = {. and .}, title = {High--Resolution {FEM}--TVD Schemes Based on a Fully Multidimensional Flux Limiter}, journal = {J. Comput. Phys.}, year = {2004}, volume = {198}, pages = {131--158}, } @ARTICLE{Kuzmin200, author = {.}, title = {A High--Resolution Finite Element Scheme for Convection--Dominated Transport}, journal = {Communications in Numerical Methods in Engineering}, year = {200}, volume = {16}, number = {3}, pages = {215--223}, } @ARTICLE{SmolianskiKuzmin1999, author = {. and .}, title = {Multilayer Taylor--{Galerkin} Schemes for Convection Problems}, journal = {Int. J. Numer. Methods Eng.}, year = {1999}, volume = {46}, number = {5}, pages = {659--670}, } @ARTICLE{Turek2003, author = {.}, title = {Perspektiven f\"{u}r Computational Fluid Dynamics {}--{} Eine kritische Betrachtung aus numerischer Sicht}, journal = {Simulation}, year = {2003}, volume = {2}, pages = {54--58}, note = {ISSN 1610-8949}, } @ARTICLE{KarakashianPascal2003, author = {. and Pascal}, title = {A Posteriori error estimates for a discontinuous {Galerkin} approximation of second--order elliptic problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {6}, pages = {2374--2399}, } @ARTICLE{StynesTobiska2003, author = {. and .}, title = {The SDFEM for a convection--diffusion problem with a boundary layer: optimal error analysis and enhancement of accuracy}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {5}, pages = {1620--1642}, } @ARTICLE{BankXu2003, author = {. and Xu}, title = {Asymptotically exact a posteriori error estimators, part II: general unstructured mesh}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {6}, pages = {2313--2332}, } @ARTICLE{MakridakisNochetto2003, author = { .}, title = {Elliptic reconstruction and a posteriori error estimates for parabolic problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {6}, pages = {1585--1594}, } @ARTICLE{MedjoTebou2004, author = {. and .}, title = {Adjoint--Based Iterative Methods for Robust Control Problems in Fluid Mechanics}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {42}, number = {1}, pages = {302--325}, } @ARTICLE{BankXu2003a, author = {}, title = {Asymptotically exact a posteriori error estimators, part I: grid with superconvergence}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {6}, pages = {2294--2312}, } @ARTICLE{Picasso2003, author = {.}, title = {An anisotropic error indicator based on Zienkiewicz--Zhu error estimator: application to elliptic and parabolic problems}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {24}, number = {4}, pages = {1328--1355}, } @ARTICLE{HagstromGoodrich2002, author = {Hagstrom and Goodrich}, title = {Accurate radiation boundary conditions for the linearized Euler equations in cartesian domains}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {3}, pages = {770--795}, } @ARTICLE{GilGuttermanOnnYavneh2002, author = { .}, title = {Automated transformations for {PDE} systems with application to multigrid solvers}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {3}, pages = {886--904}, } @ARTICLE{LiuSideris2003, author = { Sideris}, title = {Convergence of the Ghost Fluid Method for elliptic equations with interfaces}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {244}, pages = {1731--1746}, month = may, note = {S 0025-5718(03)01525-4}, } @ARTICLE{PebayBaker2003, author = {}, title = {Analysis of triangle quality measures}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {244}, pages = {1817--1839}, note = {S 0025-5718803901485-6}, } @ARTICLE{BelgacemRenard2003, author = {. and Renard}, title = {Hybrid Finite Element methods for the Signorini problem}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {243}, pages = {1117--1145}, note = {S 0025-5718(03)01490-X}, } @ARTICLE{HuangXu2002, author = {}, title = {A conforming Finite Element method for overlapping and nonmatching grids}, journal = {Mathematics of Computation}, year = {2002}, volume = {72}, number = {243}, pages = {1057--1066}, month = nov, } @ARTICLE{Hiptmair2002, author = {.}, title = {Analysis of multilevel methods for Eddy Current problems}, journal = {Mathematics of Computation}, year = {2002}, volume = {72}, number = {243}, pages = {1281--1303}, month = oct, note = {S 0025-5718(02)01468-0}, } @ARTICLE{Dai2003, author = {.}, title = {A Family of Hybrid Conjugate Gradient Methods for Unconstrained Optimization}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {243}, pages = {1317--1328}, month = feb, note = {S 0025-5718803901491-1}, } @ARTICLE{CapizzanoTyrtyshnikov2003, author = {Capizzano and Tyrtyshnikov}, title = {How to prove that a preconditioner cannot be superlinear}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {243}, pages = {1305--1316}, note = {S 0025-5718(03)01506-0}, } @ARTICLE{MorinNochettoSiebert2002, author = {. and Siebert}, title = {Local problems on stars: A posteriori error estimators, convergence and performance}, journal = {Mathematics of Computation}, year = {2002}, volume = {72}, number = {243}, pages = {1067--1097}, month = nov, note = {S 0025-5718(02)01463-1}, } @ARTICLE{Zhang2003, author = {.}, title = {Finite Element Superconvergence on Shishkin Mesh for 2--D Convection--Diffusion Problems}, journal = {Mathematics of Computation}, year = {2003}, volume = {72}, number = {243}, pages = {1147--1177}, month = feb, note = {S 0025-5718(03)01486-8}, } @ARTICLE{GrigoriadisBartzisGoulas2004, author = { Grigoriadis}, title = {Efficient treatment of complex geometries for large eddy simulations of turbulent flows}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {201--222}, } @ARTICLE{LiuWang2004, author = {}, title = {High order finite difference methods for unsteady incompressible flows in multi--connected domains}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {223--255}, } @ARTICLE{TuckerDavidson2004, author = {}, title = {Zonal k--l based large eddy simulations}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {267--287}, } @ARTICLE{LiuBankDupontGarciaSantos2003, author = {. and .}, title = {Symmetric error estimates for moving mesh mixed methods for advection--diffusion equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {40}, number = {6}, pages = {2270--2291}, } @ARTICLE{VeseOsher2002, author = { .}, title = {Numerical methods for p--harmonic flows and applications to image processing}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {40}, number = {6}, pages = {2085--2104}, } @ARTICLE{LaytonSchieweckYotov2003, author = {. and . and .}, title = {Coupling fluid flow with porous media flow}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {40}, number = {6}, pages = {2195--2218}, } @ARTICLE{BorziItoKunisch2002, author = {. and . and .}, title = {Optimal Control Formulation for Determining Optical Flow}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {3}, pages = {818--847}, } @ARTICLE{EwingLiuWang2002, author = { }, title = {A new superconvergence for mixed Finite Element approximations}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {40}, number = {6}, pages = {2133--2150}, } @ARTICLE{Dawson2002, author = {Dawson}, title = {The P\^{} (k+1) {}--{} S\^{} k local discontinuous {Galerkin} method for elliptic equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {40}, number = {6}, pages = {2151--2170}, } @ARTICLE{SchoetzauSchwabToselli2003, author = {. and . and .}, title = {Mixed hp--DGFEM for incompressible flows}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {40}, number = {6}, pages = {2171--2194}, } @ARTICLE{JiangOswald2003, author = { .}, title = {Triangular sqrt(3)--subdivision schemes: the regular case}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {156}, pages = {47--75}, note = {DOI: 10.1016/S0377-0427(02)00904-4}, } @ARTICLE{Minchev2003, author = {Minchev}, title = {Some algorithms for solving special tridiagonal block Toeplitz linear systems}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {156}, pages = {179--200}, note = {DOI: 10.1016/S0377-0427(02)00911-1}, } @ARTICLE{Tidriri2003, author = {.}, title = {Error estimates for the hybrid finite element/finite volume methods for linear hyperbolic and convection--dominated problems}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {156}, pages = {77--92}, note = {DOI: 10.1016/S0377-0427(02)00905-6}, } @ARTICLE{AnsariHegarty2003, author = {Ansari and Hegarty}, title = {Numerical solution of a convection diffusion problem with Robin boundary conditions}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {156}, pages = {221--238}, note = {DOI: 10.1016/S0377-0427(02)00913-5}, } @ARTICLE{BogaerdsHulsonPetersBaaijens2003, author = {. and Hulson and Baaijens}, title = {Time dependent finite element analysis of the linear stability of viscoelastic flows with interfaces}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {116}, pages = {33--54}, note = {DOI: 10.1016/S0377-0257(-3)00099-5}, } @ARTICLE{MikulaSgallari2003, author = {. and Sgallari}, title = {Semi--implicit finite volume scheme for image processing in 3D cylindrical geometry}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {161}, pages = {117--132}, note = {DOI: 10.1016/S0377-0427(03)00549-1}, } @ARTICLE{ShizgalJung2003, author = { .}, title = {Towards the resolution of the Gibbs phenomena}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {161}, pages = {41--65}, note = {DOI: 10.1016/S0377-0427(03)00500-4}, } @ARTICLE{Lamba2003, author = {Lamba}, title = {An adaptive timestepping algorithm for stochastic differential equations}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {161}, pages = {417--430}, note = {DOI: 10.1016/j.cam.2003.05.001}, } @ARTICLE{AxelssonGololobov2003, author = { Gololobov}, title = {A combined method of local Green`s functions and central difference method for singularity pertubed convection--diffusion problems}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {161}, pages = {245--257}, note = {DOI: 10.1016/j.cam.2003.08.005}, } @ARTICLE{SimonciniSzyld2003, author = {Simoncini and Szyld}, title = {Flexible inner--outer Krylow subspace methods}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {40}, number = {6}, pages = {2219--2239}, } @ARTICLE{Herbin2003, author = {.}, title = {A monotonic method for the numerical solution of some free boundary value problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {40}, number = {6}, pages = {2292--2310}, } @ARTICLE{UedaSellierKidaNakanishi2003, author = {Ueda and . and Kida and Nakanishi}, title = {On the low--Reynolds--number flow about two rotating circular cylinders}, journal = {Journal of Fluid Mechanics}, year = {2003}, volume = {495}, pages = {255--281}, note = {DOI: 10.1017/S002211200300627X}, } @ARTICLE{SaurelGavrilyukRenaud2003, author = {Saurel and Gavrilyuk and Renaud}, title = {A multiphase model with internal degrees of freedom: application to shock--bubble interaction}, journal = {Journal of Fluid Mechanics}, year = {2003}, volume = {495}, pages = {283--321}, note = {DOI: 10.1017/S002211200300630X}, } @ARTICLE{BlawydyiewiczCristiniLoewenberg2003, author = {Blawydyiewicz and }, title = {Multiple stationary states for deformable drops in linear {Stokes} flows}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {5}, pages = {L37-L40}, month = apr, note = {DOI: 10.1063/1.1562162}, } @ARTICLE{MinChoiYoo2003, author = { .}, title = {Maximum drag reduction in a turbulent channel flow by polymer additives}, journal = {Journal of Fluid Mechanics}, year = {2003}, volume = {492}, pages = {91--100}, note = {DOI: 10.1017/S0022112003005597}, } @ARTICLE{Loehner2004, author = {.}, title = {Multistage explicit advective prediction for projection--type incompressible flow solvers}, journal = {J. Comput. Phys.}, year = {2004}, volume = {195}, pages = {143--152}, note = {DOI: 10.1016/j.jcp.2003.09.025}, } @ARTICLE{Heinrichs2003, author = {Heinrichs}, title = {Least--squares spectral collocation for discontinuous ans dingular perturbation}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {157}, pages = {329--345}, note = {DOI: 10.1016/S0377-0427(03)00415-1}, } @ARTICLE{Kourta2004, author = {Kourta}, title = {Instability of channel flow with fluid injection and partial vortex shading}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {155--178}, note = {DOI: 10.1016/S0045-7930(03)00042-2}, } @ARTICLE{ZhangStrouboulisBabuska2001, author = {. and . and .}, title = {A Posteriori Estimators for the FEM: Analysis of the Robustness of the Estimators for the {P}oisson Equation}, journal = {Advances in Computational Mathematics}, year = {2001}, volume = {15}, pages = {375--392}, } @ARTICLE{ZhangStrouboulisBabuska2001a, author = {. and . and .}, title = {/nu\%--Superconvergence of Finite Element Solutions and Error Estimators}, journal = {Advances in Computational Mathematics}, year = {2001}, volume = {15}, pages = {293--404}, } @ARTICLE{FernandezHildViano2003, author = { Viano}, title = {Numerical Approximation of the Elastic--Viscoplastic Contact Problem with Non--matching Meshes}, journal = {Numerische Mathematik}, year = {2003}, volume = {94}, pages = {501--522}, note = {DOI: 10.1007/S00211-002-0441-X}, } @ARTICLE{FrigaardNouar2003, author = {. and Nouar}, title = {On three--dimensional linear stability of Poiseuille flow of Bingham fluids}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {10}, pages = {2843--2851}, note = {DOI: 10.1063/1.1602451}, } @ARTICLE{ZavorinLearyElman2003, author = { .}, title = {Complete stagnation of GMRES}, journal = {Linear Algebra and\_its Applications}, year = {2003}, volume = {367}, pages = {165--183}, note = {DOI: 10.1016/S0024-3795(02)00612-2}, } @ARTICLE{PieglTiller2003, author = {. and .}, title = {Approximating Surfaces of Revolution by Nonrational B--splines}, journal = {IEEE Computer Graphics and Applications}, year = {2003}, volume = {May/June}, pages = {46--52}, month = may, } @ARTICLE{HaslingerKozubekKunischPeichl2003, author = {.}, title = {Shape Optimization and Fictitious Domain Approach for Solving Free Boundary Problems of Bernoulli Type}, journal = {Computational Optimization and Applications}, year = {2003}, volume = {26}, pages = {231--251}, } @ARTICLE{Volkwein2003, author = {.}, title = {{Lagrange}--SQP Techniques for the Control Constrained Optimal Boundary Control for the Burgers Equation}, journal = {Computational Optimization and Applications}, year = {2003}, volume = {26}, pages = {253--284}, } @ARTICLE{BradburyEnright2003, author = {. and .}, title = {Fast Contouring of Solutions to Partial Differential Equations}, journal = {ACM Transactions on Mathematical Software}, year = {2003}, volume = {29}, number = {4}, pages = {418--439}, month = dec, } @ARTICLE{SzczepkowskiMalevichMityushev2003, author = {Szczepkowski and Malevich and Mityushev}, title = {Macroscropic properties of similar arrays of cylinders}, journal = {Q. J. Mech. Appl. Math.}, year = {2003}, volume = {56}, number = {4}, pages = {617--628}, } @ARTICLE{BenziBertaccini2003, author = {. and .}, title = {Approximate inverse preconditioning for shifted linear systems}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {231--244}, } @ARTICLE{YangPanGuo2003, author = { .}, title = {Parallel and distributed scientific and engineering computing}, journal = {Parallel Computing}, year = {2003}, volume = {29}, pages = {1505--1508}, note = {DOI: 10.1016/j.parco.2003.08.001}, } @ARTICLE{GuoBeckermann2003, author = {. and .}, title = {Three--dimensional simulation of freckle formation during binary alloy solidification: effect of mesh spacing}, journal = {Numerical Heat Transfer}, year = {2003}, volume = {44}, pages = {559--576}, note = {DOI: 10.1080/10407780390224184}, } @ARTICLE{ManayYezzi2003, author = {}, title = {Anti--Geometric Diffusion for Adaptive Thresholding and Fast Segmentation}, journal = {IEEE Transactions on Image Processing}, year = {2003}, volume = {12}, number = {11}, pages = {1310--1323}, } @ARTICLE{Codina2001, author = {.}, title = {A stabilized finite element method for generalized stationary incompressible flows}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2001}, volume = {190}, pages = {2681--2706}, } @ARTICLE{GilesPierceSueli2004, author = {. and .}, title = {Progress in adjoint error correction for integral functionals}, journal = {Computing and Visualization in Science}, year = {2004}, volume = {6}, pages = {113--121}, note = {DOI: 10.1007/S00791-003-0115-Y}, } @ARTICLE{Scott2003, author = {Scott}, title = {Parallel Frontal Solvers for Large Sparse Linear Systems}, journal = {ACM Transactions on Mathematical Software}, year = {2003}, volume = {29}, number = {4}, pages = {395--417}, month = dec, } @ARTICLE{Zhang2004, author = {.}, title = {Polynomial Preserving Gradient Recovery and A Posteriori Estimate for Bilinear Element on Irregular Quadrilaterals}, journal = {International Journal of Numerical Analysis And Modelling}, year = {2004}, volume = {1}, number = {1}, pages = {1--24}, } @ARTICLE{AcostaDuran2000, author = {. and }, title = {Error Estimates for Q1 isoparametric elements satisfying a weak angle condition}, journal = {SIAM Journal on Numerical Analysis}, year = {2000}, volume = {38}, number = {4}, pages = {1073--1088}, } @ARTICLE{Zhang2004a, author = {.}, title = {Polynomial Preserving Recovery for Anisotropic and Irregular Grids}, journal = {J. Comput. Math.}, year = {2004}, volume = {22}, number = {2}, pages = {331--340}, } @ARTICLE{XuZhang2004, author = {. and .}, title = {Analysis of Recovery Type A Posteriori Error Estimators for Mildly Structured Grids}, journal = {Math. Comput.}, year = {2004}, volume = {73}, pages = {1139--1152}, note = {S0025-5718(03)01600-4}, } @ARTICLE{ElmanErnstOLeary2001, author = {. and }, title = {A multigrid method enhanced by Krylov subspace iteration for discrete Helmholz equations}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1291--1315}, } @ARTICLE{ArioliNoulardRusso2001, author = { .}, title = {Stopping criteria for iterative methods: applications to PDE`s}, journal = {Calcolo}, year = {2001}, volume = {38}, pages = {97--112}, } @ARTICLE{Knupp1995, author = {Knupp}, title = {Mesh generation using vector--fields}, journal = {J. Comput. Phys.}, year = {1995}, volume = {119}, pages = {142--148}, } @ARTICLE{Kim2003, author = {.}, title = {Control of Turbulent Boundary Layers}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {5}, pages = {1093--1105}, month = may, } @ARTICLE{ChenDongarraLuszczekRoche2003, author = {. and . and }, title = {Self--adapting software for numerical linear algebra and LAPACK for clusters}, journal = {Parallel Computing}, year = {2003}, volume = {29}, pages = {1723--1743}, } @ARTICLE{DuanLiang2003, author = {. and .}, title = {Nonconforming elements in least--squares mixed finite element methods}, journal = {Math. Comput.}, year = {2003}, volume = {93}, number = {245}, pages = {1--18}, month = mar, note = {DOI: S0025-5718(03)01520-5}, } @ARTICLE{BacutaBrambleXu2002, author = {. and . and Xu}, title = {Regularity estimates for elliptic boundary value problems in Besov spaces}, journal = {Math. Comput.}, year = {2002}, volume = {72}, number = {244}, pages = {1577--1595}, month = dec, note = {DOI: S0025-5718(02)01502-8}, } @ARTICLE{MihalykoLakatosMatejdeszBlickle2004, author = { Blickle}, title = {Population balance model for particle--to--particle heat transfer in gas--solid systems}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {1325--1334}, note = {DOI: 10.1016/j.iuheatmasstransfer.2003.09.03}, } @ARTICLE{YuLauChan2004, author = { .}, title = {Numerical simulation of gas--particle flow in a single--side backwars--facing step flow}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {163}, pages = {319--331}, note = {DOI: 10.1016/j.cam.2003.08.077}, } @ARTICLE{Zhang2001, author = {.}, title = {Derivative Superconvergent Points in Finite Element Solutions of Harmonic Functions {}--{} a Theoretical Justification}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {240}, pages = {1421--1430}, month = dec, note = {DOS: S0025-5718(01)01398-9}, } @ARTICLE{BeckerBraak2001, author = {. and .}, title = {A finite element pressure gradient stabilization for the {Stokes} equations based on local projections}, journal = {Calcolo}, year = {2001}, volume = {38}, pages = {173--199}, } @ARTICLE{LasserToselli2003, author = {. and .}, title = {An overlapping domain decomposition preconditioner for a class of discontinuous {Galerkin} approximations of advection--diffusion problems}, journal = {Math. Comput.}, year = {2003}, volume = {72}, number = {243}, pages = {1215--1238}, note = {DOI: S0025-5718(03)01484-4}, } @ARTICLE{BennerE.S.Quintana-OrtiG.Quintana-Orti2003, author = {Benner and . and .}, title = {State--space truncation methods for parallel model reduction of large--scale systems}, journal = {Parallel Computing}, year = {2003}, volume = {29}, pages = {1701--1722}, note = {DOI: 10.1016/j.parco.2003.05.013}, } @ARTICLE{DjomehriBiswas2003, author = { .}, title = {Performance enhancement strategies or multi--block overset grid {CFD} applications}, journal = {Parallel Computing}, year = {2003}, volume = {29}, pages = {1791--1810}, month = may, note = {DOI: 10.1016/j.parco.2003.05.019}, } @ARTICLE{ColomerCostaConsulOliva2004, author = {Colomer and Costa and Oliva}, title = {Three--dimensional numerical simulation of convection and radiation in a differentially heated cavity using the discrete ordinates method}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {257--269}, note = {DOI: 10.1016/S0017-9310(03)00387-9}, } @ARTICLE{PudasainiHutter2003, author = {Pudasaini and Hutter}, title = {Rapid shear flows of dry granular masses down curved and twisted channels}, journal = {Journal of Fluid Mechanics}, year = {2003}, volume = {495}, pages = {193--208}, note = {DOI: 10.1017/S0022112003006141}, } @ARTICLE{TornbergEngquist2003, author = { Engquist}, title = {The Segment Projection Method for Interface Tracking}, journal = {Communications on Pure and Applied Mathematics}, year = {2003}, volume = {LVI}, pages = {47--79}, } @ARTICLE{LeeKimChoKim2003, author = { . and Kim}, title = {An algorithm for automatic 2D quadrilateral mesh generation with line constraints}, journal = {Computer Aided Design}, year = {2003}, volume = {35}, pages = {1055--1068}, } @ARTICLE{BinevDahmenDeVore2004, author = { . and .}, title = {Adaptive Finite Element Methods with convergence rater {}}, journal = {Numerische Mathematik}, year = {2004}, volume = {97}, pages = {219--268}, note = {DOI: 10.1007/s00211-003-0492-7}, } @ARTICLE{BenkoMartinVarady2001, author = { }, title = {Algorithms for Reverse Engeneering Boundary Representation Models}, journal = {Computer Aided Design}, year = {2001}, volume = {33}, number = {11}, pages = {839--851}, } @ARTICLE{CaoHuangRussell1999, author = {. and Huang and Russell}, title = {A study of monitor functions for two--dimensional adaptive mesh generation}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {20}, number = {6}, pages = {1978--1994}, } @ARTICLE{LazaarNicaise2002, author = { .}, title = {A nonconforming finite element method with anisotropic mesh grading for the incompressible {Navier}--{Stokes} equations in domaind with edges}, journal = {Calcolo}, year = {2002}, volume = {39}, pages = {123--168}, } @ARTICLE{CaiDouglasYe1999, author = {. and . and Ye}, title = {A stable noconforming quadrilateral finite element method for the stationary {Stokes} and {Navier}--{Stokes} equations}, journal = {Calcolo}, year = {1999}, volume = {36}, pages = {215--232}, } @ARTICLE{LiRenardy2000, author = {.}, title = {Numerical Study of Flows of Two Immiscible Liquids at Low Reynolds Number}, journal = {SIAM Review}, year = {2000}, volume = {42}, number = {3}, pages = {417--439}, month = mar, } @ARTICLE{Sommerfeld2001, author = {.}, title = {Validation of a Stochastic {L}agrangian Modelling Approach for Inter--Particle Collisions in Homogeneous Isotropic Turbulence}, journal = {International Journal of Multiphase Flow}, year = {2001}, volume = {27}, pages = {1829--1858}, } @ARTICLE{HuangSun2003, author = {. and .}, title = {Variational mesh adaption II: error estimates and monitor functions}, journal = {J. Comput. Phys.}, year = {2003}, volume = {184}, pages = {619--648}, } @ARTICLE{Yserentant1992, author = {.}, title = {Old and New Convergence Proofs for Multigrid Methods}, journal = {Acta Numerica}, year = {1992}, pages = {1--44}, } @ARTICLE{Brenner2001, author = {Brenner}, title = {Convergence of the Multigrid {V}--Cycle algorithm for second--order boundary value problems without full elliptic regularity}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {238}, pages = {507--525}, } @ARTICLE{Apel2002, author = {.}, title = {Multigrid Methods for Anisotropic Edge Refinement}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {40}, number = {5}, pages = {1993--2006}, } @ARTICLE{AruliahAscher2002, author = {. and .}, title = {Multigrid preconditioning for Krylov methods for time--harmonic maxwell`s equations in three dimensions}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {702--718}, } @ARTICLE{BabuskaSuri1994, author = {. and .}, title = {The P and H--P Versions of the Finite Element Method, Basic Principles and Properties}, journal = {SIAM Review}, year = {1994}, volume = {36}, number = {4}, pages = {578--632}, } @ARTICLE{ArnoldBoffiFalk2002, author = {. and . and .}, title = {Approximation by Quadrilateral Finite Elements}, journal = {Math. Comput.}, year = {2002}, volume = {71}, number = {239}, pages = {909--922}, } @ARTICLE{BankDupond1981, author = {. and .}, title = {An Optimal Order Process for Solving Finite Element Equations}, journal = {Math. Comput.}, year = {1981}, volume = {36}, number = {153}, pages = {35--51}, } @ARTICLE{BankDupondYserentant1988, author = {. and . and .}, title = {The Hierarchical Basis Multigrid Method}, journal = {Numerische Mathematik}, year = {1988}, volume = {52}, pages = {427--458}, } @ARTICLE{BraessHackbusch1983, author = {. and .}, title = {A new convergence proof for the multigrid method including the {V}--Cycle}, journal = {SIAM Journal on Numerical Analysis}, year = {1983}, volume = {20}, number = {5}, pages = {967--975}, } @ARTICLE{Brenner2003, author = {.}, title = {Convergence of nonconforming {V}--Cycle and {F}--Cycle multigrid algorithms for second order elliptic boundary value problems}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1041--1066}, month = aug, note = {\url{http:/ / www.ams.org/ mcom/ 2004-73-247/ S0025-5718-03-01578-3/ home.html}}, } @ARTICLE{ZienkiewiczZhu1992, author = {. and .}, title = {The Superconvergent Patch Recovery and A Posteriori Error Estimates. Part 1: The Recovery Technique}, journal = {Int. J. Numer. Methods Eng.}, year = {1992}, volume = {33}, pages = {1331--1364}, } @ARTICLE{ZienkiewiczZhu1992a, author = {. and .}, title = {The Superconvergent Patch Recovery and A Posteriori Error Estimates. Part 2: Error Estimators and Adaptivity}, journal = {Int. J. Numer. Methods Eng.}, year = {1992}, volume = {33}, pages = {1365--1382}, } @ARTICLE{BankWeiser1985, author = {. and .}, title = {Some a posteriori error estimators for elliptic partial differential equations}, journal = {Math. Comput.}, year = {1985}, volume = {44}, pages = {283--301}, } @ARTICLE{ChapmanSaadWington2000, author = {. and . and .}, title = {High order ILU preconditioners for {CFD} problems}, journal = {International Journal for Numerical Methods in Fluids}, year = {2000}, volume = {33}, number = {6}, pages = {767--788}, note = {Preprint: \url{http:/ / citeseer.ist.psu.edu}}, } @ARTICLE{HeuvelineRannacher2003, author = {. and .}, title = {Duality--based Adaptivity in the hp--Finite Element Method}, journal = {J. Numer. Math.}, year = {2003}, volume = {2}, pages = {95--113}, } @ARTICLE{DuesterRank2001, author = {. and .}, title = {The p--version of the finite element method compared to an adaptive h--version for the deformation theory of plasticity}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2001}, volume = {190}, pages = {1925--1935}, } @ARTICLE{Kang2003, author = {.}, title = {\$P\_1\$ nonconforming finite element multigrid method for radiation transport}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {2}, pages = {369--384}, } @ARTICLE{OosterleeWashio1998, author = {. and .}, title = {An evaluation of parallel multigrid as a solver and a preconditioner for singularly perturbed problems}, journal = {SIAM Journal on Scientific Computing}, year = {1998}, volume = {19}, pages = {87--110}, } @ARTICLE{PiquetVasseur1998, author = {. and .}, title = {Multigrid preconditioned Krylov subspace methods for three--dimensional numerical solutions of the incompressible {Navier}--{Stokes} equations}, journal = {Numerical Algorithms}, year = {1998}, volume = {17}, pages = {1--32}, } @ARTICLE{Ramage1999, author = {.}, title = {A multigrid preconditioner for stabilized discretizations of advection--diffusion problems}, journal = {J. Comput. Appl. Math.}, year = {1999}, volume = {110}, pages = {187--203}, } @ARTICLE{RoseTarjan1978, author = {. and .}, title = {Algorithmic aspects of vertex elimination on directed graphs}, journal = {SIAM Journal on Applied Mathematics}, year = {1978}, volume = {34}, number = {1}, pages = {176--197}, } @ARTICLE{RoseTarjanLueker1976, author = {. and . and .}, title = {Algorithmic aspects of vertex elimination on graphs}, journal = {SIAM Journal on Computing}, year = {1976}, volume = {5}, } @ARTICLE{Stevenson2003, author = {.}, title = {An analysis of nonconforming multi--grid methods, leading to an improved method for the Morley element}, journal = {Math. Comput.}, year = {2003}, volume = {72}, number = {241}, pages = {55--81}, month = may, note = {S 0025-5718(02)01410-2}, } @ARTICLE{vanderVorst1992, author = {.}, title = {Bi--CGStab: A fast and smoothly converging variant of Bi--CG for the solution of nonsymmetric linear systems}, journal = {SIAM Journal on Scientific and Statistical Computing}, year = {1992}, volume = {13}, number = {2}, pages = {631--644}, } @ARTICLE{WienandsOosterleeC.W.WashioT.2000, author = {. and . and .}, title = {Fourier Analysis of GMRES(m) Preconditioned by Multigrid}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {2}, pages = {582--603}, } @ARTICLE{BabuskaRheinboldt1978, author = {. and .}, title = {A Posteriori Error Estimate for the Finite Element Method}, journal = {Int. J. Numer. Methods Eng.}, year = {1978}, volume = {12}, pages = {1597--1615}, } @ARTICLE{LeshanskyBrady2004, author = {. and .}, title = {Force on a sphere via the generalized reciprocal theorem}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {3}, pages = {843--844}, } @ARTICLE{ToelkeKrafczykSchulzRank1998, author = { .}, title = {Implicit discretization and non--uniform mesh refinement approaches for FD discretizations of LBGK Models}, journal = {Int. J. Mod. Phys.}, year = {1998}, volume = {9}, number = {8}, pages = {1143--1157}, } @ARTICLE{LeeLin2001, author = { .}, title = {A Characteristic {Galerkin} Method for Discrete Boltzmann Equation}, journal = {J. Comput. Phys.}, year = {2001}, volume = {171}, pages = {336--356}, note = {doi: 10.1006/jcph.2001.6791, \url{http:/ / www.idealibrary.com}}, } @ARTICLE{ShiLinYu2003, author = {. and .}, title = {Discontinuous {Galerkin} spectral element lattice Boltzmann method on triangular element}, journal = {International Journal for Numerical Methods in Fluids}, year = {2003}, volume = {42}, pages = {1249--1261}, } @ARTICLE{BernsdorfDurstSchaefer1999, author = {. and . and Sch\"{a}.}, title = {Comparison of cellular automata and finite volume techniques for simulation of incompressible flows in complex geometries}, journal = {International Journal for Numerical Methods in Fluids}, year = {1999}, volume = {29}, pages = {251--264}, } @ARTICLE{XiPengChou1999, author = {. and . and .}, title = {Finite--volume lattice Boltzmann method}, journal = {Physical Review}, year = {1999}, volume = {59}, number = {5}, pages = {6202--6205}, month = may, } @ARTICLE{PengXiDuncanChou1999, author = {. and . and . and .}, title = {Finite volume scheme for lattice Boltzmann method on unstructured meshes}, journal = {Physical Review}, year = {1999}, volume = {59}, number = {4}, pages = {4675--4682}, month = apr, } @ARTICLE{XiPengChou1999a, author = {. and . and .}, title = {Finite--colume lattice Boltzmann schemes in two and three dimensions}, journal = {Physical Review}, year = {1999}, volume = {60}, number = {3}, pages = {3380--3388}, month = sep, } @ARTICLE{NoutgalievDinhTheofanousJoseph2003, author = {. and . and .}, title = {The lattice Boltzmann equation method: theoretical interpretation, numerics and implementations}, journal = {International Journal of Multiphase Flow}, year = {2003}, volume = {29}, pages = {117--169}, } @ARTICLE{ChenChenMatthaeus1992, author = {. and . and .}, title = {Recovery of the {Navier}--{Stokes} equation using a lattice--gas Boltzmann method}, journal = {Physical Review}, year = {1992}, volume = {45}, number = {8}, pages = {5339--5342}, } @ARTICLE{MasselotChopard1998, author = {. and .}, title = {A lattice Boltzmann model for particle transport and deposition}, journal = {Europhys. Lett.}, year = {1998}, volume = {100}, pages = {1--6}, } @ARTICLE{StickmanCooperLiPerea-Reeves2000, author = { .}, title = {Practical Application of Lattice--Gas and Lattice Boltzmann Methods to Dispersion Problems}, journal = {Geosciences}, year = {2000}, note = {\url{http:/ / www.sandia.gov/ eesector/ gs/ gc/ hws/ saltfing.htm}}, } @ARTICLE{MotomanceaMotomanceaNegrea2002, author = {. and .}, title = {A two dimensional model of fluid structure interactions in artificial heart--valves}, journal = {Seria Mecanica}, year = {2002}, volume = {47}, number = {61}, pages = {69--76}, note = {Buletinul Stiintific al Universitatii `Politehnica` din Timisoara}, } @ARTICLE{LeeBarlovicSchreckenbergKim2004, author = {. and . and . and .}, title = {Mechanical Restriction versus Human Overreaction Triggering Congested Traffic States}, journal = {Physical Review Letters}, year = {2004}, volume = {92}, number = {23}, pages = {238702-1 - 238702-4}, } @ARTICLE{BicoQuere2000, author = {. and .}, title = {Liquid trains in a tube}, journal = {Europhys. Lett.}, year = {2000}, volume = {51}, number = {5}, pages = {546--550}, month = sep, } @ARTICLE{GuessousSouhar2004, author = {. and .}, title = {Multilevel block ILU preconditioner for sparse nonsymmetric M--matrices}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {162}, pages = {231--246}, } @ARTICLE{Yserentant1986, author = {.}, title = {The Convergence of Multi--Level Methods for Solving Finite--Element Equations in the Presence of Singularities}, journal = {Math. Comput.}, year = {1986}, volume = {47}, number = {176}, pages = {399--409}, } @ARTICLE{Brenner1989a, author = {.}, title = {An optimal--order nonconforming multigrid method for the biharmonic equation}, journal = {SIAM Journal on Numerical Analysis}, year = {1989}, volume = {26}, number = {5}, pages = {1124--1138}, } @ARTICLE{Brenner1989b, author = {.}, title = {An Optimal--Order Multigrid Method for \$P\_1\$ Nonconforming finite Elements}, journal = {Math. Comput.}, year = {1989}, volume = {52}, number = {185}, pages = {1--15}, } @ARTICLE{Rui2002, author = {.}, title = {Symmetric modified finite volume element method for self--adjoint elliptic and parabolic problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {146}, pages = {373--386}, } @ARTICLE{SirotkinTarvainen2002, author = {. and .}, title = {Parallel Schwarz methods for convection--dominated semilinear diffusion problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {189--211}, } @ARTICLE{MatthewsORiordanShishkin2002, author = { .}, title = {A numerical method for a system of singular pertubed reaction--diffusion equations}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {151--166}, } @ARTICLE{SchulzSteinebach2002, author = { .}, title = {Two--dimensional modelling of the river Rhine}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {11--20}, } @ARTICLE{HoelligReifWipper2002, author = { .}, title = {Multigrid methods with web--splines}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {237--256}, note = {DOI: 10.1007/s002110100324}, } @ARTICLE{HoerschlerMeinkeSchroeder2003, author = {H\"{. and Schr\"{o}.}, title = {Numerical simulation of the flow in a model of the nasal cavity}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {39--45}, } @ARTICLE{CohenKaberPostel2003, author = {. and . and .}, title = {Adaptive multiresolution for finite volume solutions of gas dynamics}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {31--38}, } @ARTICLE{CastilloCockburnSchoetzauSchwab2001, author = {. and \"{ .}, title = {Optimal A Priori error estimates for the hp--version of the local discontinuous {Galerkin} method for convection--duffusion problems}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {238}, pages = {455--478}, note = {S 0025-5718(01)01317-5}, } @ARTICLE{CampbellHymanShashkov2002, author = {. and . and .}, title = {Mimetic Finite Difference Operators for Second--Order Tensor on Unstructured Grids}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {157--173}, } @ARTICLE{HeinrichPietsch2002, author = {. and .}, title = {Nitsche Type Mortaring for some Elliptic Problem with Corner Singularities}, journal = {Computing}, year = {2002}, volume = {68}, pages = {217--238}, } @ARTICLE{KimChoBae2002, author = { .}, title = {Perturbed integral equation method on determining unknown geometry in fluid flow}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {99--120}, } @ARTICLE{LiangMaThou2002, author = { .}, title = {A symmetric finite volume scheme for selfadjoint elliptic problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {121--136}, } @ARTICLE{MazziaPutti2002, author = {. and .}, title = {Mixed--finite element and finite volume discretization for heavy brine simulations in groundwater}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {191--213}, } @ARTICLE{BitterlichKnabner2002, author = {. and .}, title = {An efficient method for solving an inverse problem for the Richards equation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {147}, pages = {153--173}, } @ARTICLE{Turkyilmazoglu2003, author = {.}, title = {Instability of the flow in a vicinity of trailing edge of a class of thin aerofoils}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {353--371}, note = {incomplete copy}, } @ARTICLE{BurgNewmanIII2003, author = {. and .}, title = {Computationally efficient, numerically exact design space derivatives via the Taylor`s series expansion method}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {373--383}, } @ARTICLE{DuMuWu2001, author = {. and . and .}, title = {Efficient parallel algorithms for parabolic problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {5}, pages = {1469--1487}, } @ARTICLE{Medjo2002, author = {.}, title = {Iterative methods for a class of robust control problems in a fluid mechanics}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {39}, number = {5}, pages = {1625--1647}, } @ARTICLE{BarrettBlowey2001a, author = {. and .}, title = {Finite element approximation of a degenerate Allen--Cahn/Cahn--Hilliard system}, journal = {SIAM Journal on Numerical Analysis}, year = {2001}, volume = {39}, number = {5}, pages = {1598--1624}, } @ARTICLE{XuLiChen2002, author = {. and . and .}, title = {A multigrid method for the Mortar--Type Morley element approximation of a plate bending problem}, journal = {SIAM Journal on Numerical Analysis}, year = {2002}, volume = {39}, number = {5}, pages = {1712--1731}, } @ARTICLE{JannelliFazioAmbrosi2003, author = {. and . and .}, title = {A 3D mathematical model for the prediction of mucilage dynamics}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {47--57}, } @ARTICLE{MossiSagaut2003, author = {. and .}, title = {Numerical investigation of fully developed channel flow using shock--capturing schemes}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {249--274}, } @ARTICLE{BermudezRodriguez2001, author = {. and .}, title = {Analysis of a Finite Element method for pressure/potential formulation od elastoacoustic spectral problems}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {238}, pages = {537--552}, note = {S 0025-5718(01)01335-7}, } @ARTICLE{Richter2001, author = {.}, title = {On the stability of a family of finite element methods for hyperbolic problems}, journal = {Math. Comput.}, year = {2001}, volume = {71}, number = {238}, pages = {527--535}, note = {S 0025-5718(01)01334-5}, } @ARTICLE{NgRostWen2002, author = {. and . and .}, title = {On inversion of Toeplitz matrices}, journal = {Linear Algebra and\_its Applications}, year = {2002}, volume = {348}, pages = {145--151}, } @ARTICLE{Geskus2002, author = {.}, title = {A new proof of a theorem on M--matrices}, journal = {Linear Algebra and\_its Applications}, year = {2002}, volume = {348}, pages = {139--144}, } @ARTICLE{Matejas2002, author = {.}, title = {Convergence of scaled iterates by the {Jacobi} method}, journal = {Linear Algebra and\_its Applications}, year = {2002}, volume = {349}, pages = {17--53}, } @ARTICLE{GoetzHuth.S.Buggisch2002, author = {. and . and .}, title = {Flow behaviour of pastes with a deformation history dependent yield stress}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {103}, pages = {149--166}, } @ARTICLE{Renardy2002, author = {.}, title = {Similarity solutions for jet breakup for various models of viscoelastic fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2002}, volume = {104}, pages = {65--74}, } @ARTICLE{Vanselow2002, author = {.}, title = {Error Estimates of a {FEM} with Lumping for Parabolic PDEs}, journal = {Computing}, year = {2002}, volume = {68}, pages = {131--141}, note = {DOI 10.1007/s00607-001-1439-4}, } @ARTICLE{HernandezRubio2002, author = {. and .}, title = {Semilocal Convergence of the Secant Method under Mild Convergence Conditions of Differentiability}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {277--285}, } @ARTICLE{KaraaZhang2002, author = {. and .}, title = {Convergence and Performance of Iterative Methods for Solving Variable Coefficient Convection--Diffusion Equation with a Fourth--Order Compact Difference Scheme}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {457--479}, } @ARTICLE{MorettiRojas-MedarRojas-Meda2002, author = {. and . and .}, title = {The Equations of a Viscous Incompressible Chemically Active Fluid: Existence and Uniqueness of Strong Solutions in an Unbounded Domain}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {287--299}, } @ARTICLE{LuoLin2002, author = { .}, title = {High Accuracy Analysis of the Adini`s Nonconforming Element}, journal = {Computing}, year = {2002}, volume = {68}, pages = {65--79}, } @ARTICLE{Neuss2002, author = {.}, title = {A New Sparse--Matrix Storage Method for Adaptively Solving Large Systems of Reaction--Diffusion--Transport Equations}, journal = {Computing}, year = {2002}, volume = {68}, pages = {19--36}, } @ARTICLE{Croisille2002, author = {.}, title = {Keller`s Box--Scheme for the One--Dimensional Stationary Convection--Diffusion Equation}, journal = {Computing}, year = {2002}, volume = {68}, pages = {37--63}, } @ARTICLE{LaiSchumaker2002, author = {. and .}, title = {Quadrilateral macroelements}, journal = {SIAM Journal on Mathematical Analysis}, year = {2002}, volume = {33}, number = {5}, pages = {1107--1116}, } @ARTICLE{Benney2002, author = {.}, title = {Nonlinear instability theories in hydrodynamics}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {144}, pages = {1--5}, } @ARTICLE{ZhangKouatchouOthman2002, author = {. and .}, title = {On Cyclic Reduction and Finite Difference Schemes}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {213--222}, } @ARTICLE{TanahashiMiyauchiIkeda1997, author = {. and . and .}, title = {Scaling law of coherent fine scale structure in homogenuous isotropic turbulence}, journal = {TSF}, year = {1997}, volume = {11}, pages = {4-17 - 4-22}, } @ARTICLE{CordobaFefferman2001, author = {. and .}, title = {Potato chip singularities of 3D flows}, journal = {SIAM Journal on Mathematical Analysis}, year = {2001}, volume = {33}, number = {4}, pages = {786--789}, } @ARTICLE{AzeradGuillen2001, author = {. and .}, title = {Mathematical justification of the hydrostatic approximation in the primitive equations of geophysical fluid dynamics}, journal = {SIAM Journal on Mathematical Analysis}, year = {2001}, volume = {33}, number = {4}, pages = {847--589}, } @ARTICLE{LeongTanTan2002, author = {. and . and .}, title = {An Iterative Modular Multiplication Algorithm}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {175--180}, } @ARTICLE{KannanChung2002, author = {. and .}, title = {Finite Difference Approximate Solutions for the Two--Dimensional Burgers` System}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {193--200}, } @ARTICLE{Li2002, author = {.}, title = {Uniform Convergence of Discontinuous Finite Element Methods for Singularity Perturbed Reaction--Diffusion Problems}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {44}, pages = {231--240}, } @ARTICLE{Batcho2001, author = {.}, title = {The energy operator and new scaling relations for the incompressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {62}, number = {1}, pages = {70--93}, } @ARTICLE{HariharanYoung2001, author = {. and .}, title = {Comparison of asymptotic solutions of a phase--field model to a sharp--interface model}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {1}, pages = {244--263}, } @ARTICLE{Knessl2001, author = {.}, title = {On two--dimensional convection--diffusion past a circle}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {61}, number = {1}, pages = {310--335}, } @ARTICLE{LoncaricRyabenkiiTsynkov2001, author = {. and .}, title = {Active shilding and control of noise}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {62}, number = {2}, pages = {563--596}, } @ARTICLE{BenammouOmrani2002, author = {. and .}, title = {A finite element method for the Sivashinsky equation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {142}, pages = {419--431}, } @ARTICLE{LiYan2002, author = {. and .}, title = {New error estimates of bi--cubic Hermite finite element methods for biharmonic equations}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {142}, pages = {251--285}, } @ARTICLE{WangLi2002, author = {. and .}, title = {An analysis of a conforming exponentially fittet finite element method for a convection--diffusion problem}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {143}, pages = {291--310}, } @ARTICLE{Lenferink2002, author = {.}, title = {A second order scheme for a time--dependent, singularly perturbed convection--diffusion equation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {143}, pages = {49--68}, } @ARTICLE{Bedivan2002, author = {.}, title = {Error estimates for Least Squares Finite Element Methods}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {1003--1020}, } @ARTICLE{LittleLiChoiSaad2002, author = {. and . and . and .}, title = {Particle Partitioning Strategies for the Parallel Computation of Solid--Liquid Flows}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {1591--1616}, } @ARTICLE{BosseurFloriOrenga2002, author = {. and . and .}, title = {Identification of Boundary Conditions in a Nonlinear Shallow Water Flow}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {1559--1573}, } @ARTICLE{HanLo2002, author = {. and .}, title = {Two New Self--Adaptive Projection Methods for Variational Inequality Problems}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {1529--1537}, } @ARTICLE{Schweizer2001, author = {.}, title = {Bifurcation analysis for surface waves generated by wind}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {62}, number = {2}, pages = {407--423}, } @ARTICLE{BourgeatGipoulouxMarusic-Paloka2001, author = {. and . and .}, title = {Mathematical modelling and numerical simulation of a Non--Newtonian viscous flow through a thin filter}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {62}, number = {2}, pages = {597--626}, } @ARTICLE{OthmerHillen2002, author = {. and .}, title = {The diffusion limit of transport equations II: Chemotaxis equations}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {4}, pages = {1222--1250}, note = {CFD3 CFD4 CFD9}, } @ARTICLE{GervasioLionsQuarteroni2001, author = {. and . and .}, title = {Heterogeneous coupling by virtual control methods}, journal = {Numerische Mathematik}, year = {2001}, volume = {90}, pages = {241--264}, note = {DOI 10.1007/s002110100303}, } @ARTICLE{ChenEwing2001, author = {. and .}, title = {Degenerate two--phase incompressible flow}, journal = {Numerische Mathematik}, year = {2001}, volume = {90}, pages = {215--240}, note = {DOI 10.1007/s002110100291}, } @ARTICLE{WangLiu2002, author = {. and .}, title = {Analysis of finite difference schemes for unsteady {Navier}--{Stokes} equations in vorticity formulation}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {543--576}, note = {DOI 10.1007/s002110100311}, } @ARTICLE{DeckelnickDziuk2002, author = {. and .}, title = {A fully discrete numerical scheme for weighted mean curvature flow}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {423--452}, note = {DOI 10.1007/s002110100322}, } @ARTICLE{Kornhuber2002, author = {.}, title = {On constrained Newton linearization and multigrid for variational inequalities}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {699--721}, note = {10.1007/s002110100341}, } @ARTICLE{GuillaumeMasmoudioZeglaoui2002, author = {. and .}, title = {From compressible to incompressible materials via an asymptotic expansion}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {649--673}, note = {DOI 10.1007/s002110100347}, } @ARTICLE{CalvettiLewisReichel2002, author = {. and . and .}, title = {On the regularizing properties of the GMRES method}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {605--625}, note = {DOI 10.1007/s002110100339}, } @ARTICLE{ApelNicaise2004, author = {. and .}, title = {The inf--sup Condition for Low Order Elements on Anisotropic Meshes}, journal = {Calcolo}, year = {2004}, volume = {41}, pages = {89--113}, note = {DOI 10.1007/s10092-004-0086-5}, } @ARTICLE{Chang2004, author = {.}, title = {A note on the upper bound of the spectral radius for SOR iteration matrix}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {167}, pages = {251--253}, note = {DOI 10.1016/j.cam.2003.03.052}, } @ARTICLE{WangHuang2004, author = {. and .}, title = {Comparison results between {Jacobi} and other iterative methods}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {45--51}, note = {DOI 10.1016/j.cam.2003.10.017}, } @ARTICLE{BellengerCoorevits2004, author = {. and .}, title = {New approach for automation of finite element analysis}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {65--75}, note = {DOI 10.1016/j.cam.2003.05.011}, } @ARTICLE{Homeier2004, author = {Homeier, .}, title = {A modified Newton method with cubic convergence: the multivariate case}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {161--169}, note = {DOI 10.1016/j.cam.2003.12.041}, } @ARTICLE{Dehghan2002, author = {.}, title = {A new ADI Technique for Two--Dimensional Parabolic Equation with an Integral Condition}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {1477--1488}, } @ARTICLE{Berrone2002, author = {.}, title = {Robustness in a posteriori error analysis for {FEM} models}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {389--422}, note = {DOI 10.1007/s002110100370}, } @ARTICLE{Gallagher2002, author = {.}, title = {Mathematical analysis of a structure--preserving approximation of the bidimensional vorticity equation}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {223--236}, note = {DOI 10.1007/s002110100293}, } @ARTICLE{BarrientosGaticaStephan2002, author = {. and . and .}, title = {A mixed finite element method for nonlinear elasticity: two--fold saddle point approach and a--posteriori error estimate}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {197--222}, note = {DOS: 10.1007/s002110100337}, } @ARTICLE{EwingIlievLazarov2001, author = {. and . and .}, title = {A modified finite volume approximation of second--order elliptic equations with discontinuous coefficients}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1335--1351}, } @ARTICLE{Ushakova2001, author = {.}, title = {Conditions of nondegeneracy of three--dimensional cells. A formula of a volume of {} cells}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1274--1290}, } @ARTICLE{Cortez2001, author = {.}, title = {The method of regularized stokeslets}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1204--1225}, } @ARTICLE{Stefanica2001, author = {.}, title = {A numerical study of FETI algorithms for mortar finite element methods {}}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1135--1160}, } @ARTICLE{GodinaudLeRouxLeRoux2001, author = {. and . and .}, title = {A {Lagrange} scheme for a mathematical model of powder compression}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1094--1112}, } @ARTICLE{Liu2002, author = {.}, title = {Numerical solution of forward and backward problem for 2--D hear conduction equation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {459--482}, } @ARTICLE{SunLeeCendes2001, author = {. and .}, title = {Construction of nearly orthogonal Nedelec bases for rapid convergence with multilevel preconditioned solvers}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1053--1076}, } @ARTICLE{HennWitsch2001, author = {. and .}, title = {Iterative multigrid regularization techniques for image matching}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {23}, number = {4}, pages = {1077--1093}, } @ARTICLE{CaoGuo2002, author = {. and .}, title = {A preconditioner with inexact element face solver for three dimensional p--Version finite element methods}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {144}, pages = {131--144}, } @ARTICLE{Dubois2002, author = {.}, title = {Vorticity--velocity--pressure formulation for the {Stokes} problem}, journal = {Mathematical Methods in the applied Sciences}, year = {2002}, volume = {25}, pages = {1091--1119}, note = {DOI 10.1002/mma.328}, } @ARTICLE{RotsteinBrandonNovick-CohenNepomnyashchy2001, author = {. and .}, title = {Phase field equations with memory: The hyperbolic case}, journal = {SIAM Journal on Applied Mathematics}, year = {2001}, volume = {62}, number = {1}, pages = {264--282}, } @ARTICLE{BelhoucineDevilleElazehariBensalah2004, author = {. and . and .}, title = {Explicit algebraic Reynolds stress model of incompressible turbulent flow in rotating square duct}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {179--199}, } @ARTICLE{Winslow1967, author = {.}, title = {Numerical solution of the quasi--linear Poisson--equation in a nonuniform triangle mesh}, journal = {J. Comput. Phys.}, year = {1967}, volume = {1}, pages = {149--172}, } @ARTICLE{Silver2004, author = {.}, title = {Analysis of three--dimensional grids: the eight--point cube}, journal = {Applied Mathematics and Computation}, year = {2004}, volume = {153}, pages = {467--473}, } @ARTICLE{NoubaryNoubary2004, author = {. and .}, title = {On survival of sports records}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {227--234}, } @ARTICLE{Saito2004, author = {.}, title = {A holomorphic semigroup approach to the lumped mass finite element method}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {71--85}, } @ARTICLE{Randrianarivony2004, author = {.}, title = {Anisotropic finite elements for the {Stokes} problem: a posteriori error estimatoe and adaptive mesh}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {255--275}, } @ARTICLE{BrackbillSaltzman1982, author = {. and .}, title = {Adaptive zoning for singular problems in two dimensions}, journal = {J. Comput. Phys.}, year = {1982}, volume = {46}, pages = {342--368}, } @ARTICLE{John2002a, author = {.}, title = {Higher order finite element methods and multigrid solvers in a benchmark problem for the 3D {Navier}--{Stokes} equations}, journal = {International Journal for Numerical Methods in Fluids}, year = {2002}, volume = {40}, pages = {775--798}, note = {DOI 10.1002/fld.377}, } @ARTICLE{MahapatraMahapatra2004, author = {. and .}, title = {Numerical modelling of combined conductive and radiative heat transfer within square enclosure using discrete ordinate method}, journal = {Int. J. Heat Mass Transfer}, year = {2004}, volume = {40}, pages = {533--538}, note = {DOI 10.1007/s00231-003-0500-6}, } @ARTICLE{KumarSinghKumarSinghSingh2004, author = {. and . and .}, title = {Generalised Cousette flow of two immiscible viscous fluids with heat transfer using Brinkham model}, journal = {Int. J. Heat Mass Transfer}, year = {2004}, volume = {40}, pages = {517--523}, note = {DOI 10.1007/s00231-003-0421-4}, } @ARTICLE{NgyenGovindjeeKleinGao2004, author = {. and . and . and .}, title = {A rate--dependent cohesive continuum model for the study of crack dynamics}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {3239--3265}, note = {DOI 10.1016/j.cma.2003.09.024}, } @ARTICLE{StroevenAskesSluys2004, author = {. and . and .}, title = {Numerical determination of representative volumes for granular materials}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {3221--3238}, note = {DOI 10.1016/j.cma.2003.09.023}, } @ARTICLE{KohnoTanahashi2004, author = {. and .}, title = {Numerical analysis of moving interfaces using a level set method coupled with adaptive mesh refinement}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {921--944}, note = {DOI 10.1002/fld.715}, } @ARTICLE{ChandioSujathaWebster2004, author = {. and . and .}, title = {Consistent hybrid finite volume/element formulations: model and complex viscoelastic flows}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {945--971}, note = {DOI 10.1002/fld.723}, } @ARTICLE{Allen2004, author = {.}, title = {An unsteady multiblock multigrid scheme for lifting forward flight rotor simulation}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {973--974}, note = {DOI 10.1002/fld.711}, } @ARTICLE{XieKoshizukaYoshiaki2004, author = {. and . and .}, title = {Modelling of a single drop impact onto liquid film using particle method}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1009--1023}, note = {DOI 10.1002/fld.736}, } @ARTICLE{HanLuBao1994, author = { .}, title = {A Discrete Artificial Boundary Condition for Steady Incompressible Viscous Flow in a No--Slip Channel Using a Fast Iterative Method}, journal = {J. Comput. Phys.}, year = {1994}, volume = {114}, pages = {201--208}, } @ARTICLE{NiazmandRenksizbulut2003, author = {. and .}, title = {Surface effects on transient three--dimensional flows around rotating spheres at moderate Reynolds numbers}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {1405--1433}, } @ARTICLE{GlothHaenelTranVilsmeier2003, author = {. and .}, title = {A front tracking method on unstructured grids}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {547--570}, } @ARTICLE{ZhaoMerrimanOsherWang1998, author = {. and . and . and .}, title = {Capturing the Behavior of Bubbles and Drops Using the Variational Level Set Approach}, journal = {J. Comput. Phys.}, year = {1998}, volume = {143}, pages = {495--518}, } @ARTICLE{ThompsonThamesWayneMastin1974, author = {. and . and .}, title = {Automatic Numerical Generation of Body--Fitted Curvilinear Coordinate System for Field Containing Any Number of Arbitrary Two--Dimensional Bodies}, journal = {J. Comput. Phys.}, year = {1974}, volume = {15}, pages = {299--319}, } @ARTICLE{SussmanSmerekaOsher1994, author = {. and . and .}, title = {A Level Set Approach for Computing Solutions to Incompressible Two--Phase Flow}, journal = {J. Comput. Phys.}, year = {1994}, volume = {114}, pages = {146--159}, } @ARTICLE{BrackbillKotheZemach1992, author = {. and . and .}, title = {A Continuum Method for Modeling Surface Tension}, journal = {J. Comput. Phys.}, year = {1992}, volume = {100}, pages = {335--354}, } @ARTICLE{Lafaurie1994, author = {.}, title = {Modelling Merging and Fragmentation in Multiphase Flows with SURFER}, journal = {J. Comput. Phys.}, year = {1994}, volume = {113}, pages = {143--147}, } @ARTICLE{DuchanoyJongen2003, author = { .}, title = {Efficient simulation of liquid--solid flows with high fraction in complex geometries}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {1453--1471}, } @ARTICLE{Nigam2003, author = {.}, title = {Numerical simulation of buoyant mixture flows}, journal = {International Journal of Multiphase Flow}, year = {2003}, volume = {29}, pages = {983--1015}, } @ARTICLE{Nachbin2003, author = {.}, title = {A terrain--following Boussinesq system}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {63}, number = {3}, pages = {905--922}, } @ARTICLE{Giannakopoulus1988, author = {.}, title = {Directional Control in Grid Generation}, journal = {J. Comput. Phys.}, year = {1988}, volume = {74}, pages = {422--439}, } @ARTICLE{Nataf1989, author = {.}, title = {An Open Boundary Condition for the Computation of the Steady Incompressible {Navier}--{Stokes} Equations}, journal = {J. Comput. Phys.}, year = {1989}, volume = {85}, pages = {104--129}, } @ARTICLE{MaasRindtvanSteenhoven2003, author = {. and . and .}, title = {The influence of hear on the 3D--transition {} of the Karaman vortex street}, journal = {International Journal of Heat and Mass Transfer}, year = {2003}, volume = {46}, pages = {3069--3081}, } @ARTICLE{Renardy2003, author = {.}, title = {Stress integration for the constitutive law of the upper convected Maxwell fluid near the corners in a driven cavity}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {112}, pages = {77--84}, } @ARTICLE{Schmidt1996, author = {.}, title = {Computation of Three Dimensional Dendrites with Finite Elements}, journal = {J. Comput. Phys.}, year = {1996}, volume = {125}, pages = {293--312}, } @ARTICLE{DeckelnickDziuk1995, author = {. and .}, title = {Convergence of a finite element method for non--parametric mean curvature flow}, journal = {Numerische Mathematik}, year = {1995}, volume = {72}, pages = {197--222}, } @ARTICLE{Dziuk1991, author = {.}, title = {An algorithm for evolutionary surfaces}, journal = {Numerische Mathematik}, year = {1991}, volume = {58}, pages = {603--611}, } @ARTICLE{Dai2001, author = {.}, title = {New properties of a nonlinear conjugate gradient method}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, number = {1}, pages = {83--98}, note = {DOS 10.1007/s002110000253}, } @ARTICLE{WangKhooYeo2003, author = { .}, title = {Elastic mesh technique for 3D BIM simulation with an application to underwater explosion bubble dynamics}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {1195--1212}, note = {DOI 10.1016/S0045-7930(02)00105-6}, } @ARTICLE{PengMerrimanOsherZhaoKang1999, author = {. and .}, title = {A {PDE}--Based Fast Local Level Set Method}, journal = {J. Comput. Phys.}, year = {1999}, volume = {155}, pages = {410--438}, } @ARTICLE{GrahamSpenceVainikko2003, author = {. and . and .}, title = {Parallel iterative methods for {Navier}--{Stokes} equations and application to eigenvalue computation}, journal = {Concurrency and Computation: Practice and Experience}, year = {2003}, pages = {1--17}, } @ARTICLE{Dagan2003, author = {.}, title = {Numerical consistency and spurious boundry layer in the projection method}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {1213--1232}, } @ARTICLE{DeckelnickDziuk2000, author = {. and .}, title = {Error esimates for a semi--implicit discrete finite element scheme for the mean curvature flow of graphs}, journal = {Interfaces and Free Boundaries}, year = {2000}, volume = {2}, pages = {341--359}, } @ARTICLE{vanderVorst2002, author = {, .}, title = {Efficient and reliable iterative methods for linear systems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {251--265}, } @ARTICLE{MohammadiPironneau2002, author = {. and .}, title = {Applied optimal shape design}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {193--205}, } @ARTICLE{Zhang2002, author = {.}, title = {A Class of Product--type Krylow--Subspace Methods for Solving Nonsymmetric Linear Systems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {149}, pages = {297--305}, } @ARTICLE{Gerbeau2000, author = {.}, title = {A stabilized finite element method for the incompressible magnetohydrodynamic equations}, journal = {Numerische Mathematik}, year = {2000}, volume = {87}, pages = {83--111}, note = {DOI 10.1007/s002110000193}, } @ARTICLE{BeckermannDiepersSteinbachKarmaTong1999, author = { .}, title = {Modelingh Melt Convection in a Phase--Field Simulations of Solidification}, journal = {J. Comput. Phys.}, year = {1999}, volume = {154}, pages = {468--496}, } @ARTICLE{DiepersBeckermannSteinbach1999, author = { . and .}, title = {Simulation of convection and ripening in a binary alloy mush using a phase--field method}, journal = {Acta Materialia}, year = {1999}, volume = {47}, number = {13}, pages = {3663--3678}, } @ARTICLE{ReddyBeckermann1997, author = {. and .}, title = {Modeling of Macrosegregation Due to Thermosolutal Convection and Contraction--Driven Flow in Direct Chill Continuous Casting of an Al--Cu Round Ingot}, journal = {Metallurgical and Materials Transactions}, year = {1997}, volume = {28B}, pages = {479--488}, } @ARTICLE{LeBorne2000, author = {.}, title = {Ordering Techniques for Two--{} and Three--Dimensional Convection--Dominated Elliptic Boundary Value Problems}, journal = {Computing}, year = {2000}, volume = {64}, pages = {123--155}, } @ARTICLE{Matusu-Necasova1998, author = {.}, title = {Bipolar Isothermal Non--Newtonian Compressible Flow}, journal = {Journal of Mathematical Analysis and Applications}, year = {1998}, volume = {225}, pages = {168--192}, } @ARTICLE{BeckermannWang1995, author = {. and .}, title = {Multiphase/--scale modeling of alloy solidification}, journal = {Annual Review of Heat Transfer}, year = {1995}, volume = {6}, pages = {115--198}, } @ARTICLE{FeistauerFelcmanLukacova-Medvidova1997, author = {. and .}, title = {On the Convergence of a Combined Finite Volume--Finite Element Method for Nonlinear Convection--Diffusion Problems}, journal = {Numerical Methods for Partial Differential Equations}, year = {1997}, volume = {13}, pages = {163--190}, } @ARTICLE{FeistauerFelcmanLukacova-MedvidovaWarnecke1999, author = {. and . and . and .}, title = {Error estimates for a combined Finite Volume--Finite Element method for nonlinear convection--diffusion problems}, journal = {SIAM Journal on Numerical Analysis}, year = {1999}, volume = {36}, number = {5}, pages = {1528--1548}, } @ARTICLE{Fliege2001, author = {.}, title = {OLAF {}--{} A General Modeling System to Evaluate and Optimize the Location on an Air Polluting Facility}, journal = {OR Spektrum}, year = {2001}, volume = {23}, pages = {117--136}, } @ARTICLE{TardosMcNamaraTalu2002, author = {. and . and .}, title = {Slow and intermediate flow of a frictional bulk powder in the Couette geometry}, journal = {Powder Technology}, year = {2002}, volume = {4686}, pages = {1--17}, } @ARTICLE{JongenChouikhi2002, author = {. and .}, title = {Simulation of Free--Surface Flows Induced by Partially Immersed Moving Body: Validation}, journal = {AIChE Journal}, year = {2002}, volume = {48}, number = {3}, pages = {452--462}, } @ARTICLE{KressNilsson2003, author = {. and .}, title = {Boundary conditions and estimates for the linearized {Navier}--{Stokes} equations on staggered grids}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {1093--1112}, } @ARTICLE{MarsdenShkoller2003, author = {. and .}, title = {The Anisotropic {L}agrangian Averaged {E}uler and {N}avier--{S}tokes Equations}, journal = {Archive for Rational Mechanics and Analysis}, year = {2003}, volume = {166}, pages = {27--46}, note = {DOI 10.1007/s00205-002-0207-8}, } @ARTICLE{GanChangFengHu2003, author = {. and . and . and .}, title = {Direct numerical simulation of the sedimentation of solid particles with thermal convection}, journal = {Journal of Fluid Mechanics}, year = {2003}, volume = {481}, pages = {385--411}, } @ARTICLE{Metcalf2002, author = {.}, title = {History of Fortran}, journal = {ACM SIGPLAN Fortran Forum}, year = {2002}, volume = {21}, number = {3}, pages = {19--20}, note = {\url{http:/ / portal.acm.org/ citation.cfm?id=602376.602379}}, } @ARTICLE{Markus2002, author = {.}, title = {Combining Fortran and scripting languages}, journal = {ACM SIGPLAN Fortran Forum}, year = {2002}, volume = {21}, number = {3}, pages = {10--18}, note = {\url{http:/ / portal.acm.org/ citation.cfm?id=602376.602378}}, } @ARTICLE{Walter2003, author = {.}, title = {Germany`s Negative Vote (on F2000)}, journal = {ACM Fortran Forum}, year = {2003}, volume = {22}, number = {1}, pages = {11--21}, month = apr, } @ARTICLE{XuZikatanov2003, author = {. and .}, title = {Some Observations on Babuska and Brezzi Theories}, journal = {Numerische Mathematik}, year = {2003}, volume = {94}, pages = {195--202}, note = {DOI 10.1007s002110100308}, } @ARTICLE{Weickert1999, author = {.}, title = {Coherence--enhancing diffusion of colour images}, journal = {Image and Vision Computing}, year = {1999}, volume = {17}, pages = {201--212}, } @ARTICLE{LiouShihChauChen2002, author = {. and . and . and .}, title = {Three--dimensional simulations of the droplet formation during the inkjet printing process}, journal = {International Communications in Heat and Mass Transfer}, year = {2002}, volume = {29}, number = {8}, pages = {1109--1118}, } @ARTICLE{Ying2002, author = {.}, title = {Convergence study of the Chorin--Marsden formula}, journal = {Math. Comput.}, year = {2002}, volume = {72}, number = {241}, pages = {307--333}, note = {DOI S 0025-5718(02)01423-0}, } @ARTICLE{Marcinkowski2002, author = {.}, title = {A mortar element method for some discretizations of a plate problem}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {361--386}, note = {DOI 10.1007/s002110100389}, } @ARTICLE{SmithJooArmstrongBrown2003, author = {. and . and . and .}, title = {Linear stability analysis of flow of an Oldroyd--B fluid through a linear array of cylinders}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {109}, pages = {13--50}, } @ARTICLE{BertolaCafaro2003, author = {. and .}, title = {Analogy between pipe flow of non--Newtonian fluids and 2--D compressible flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {109}, pages = {1--12}, } @ARTICLE{HuZou2002, author = {. and .}, title = {Two new variants of nonlinear inexact Uzawa algorithms for saddle--point problems}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {333--359}, note = {DOI 10.1007/s002110100386}, } @ARTICLE{GiladiKeller2002, author = {. and .}, title = {Space--time domain decomposition for parabolic problems}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {279--313}, note = {DOI 10.1007/s002110100345}, } @ARTICLE{ClemonsHariharanYoung2002, author = {. and . and .}, title = {Asymptotic solutions of a phase--field model for alloy solidification}, journal = {SIAM Journal on Applied Mathematics}, year = {2002}, volume = {62}, number = {6}, pages = {1952--1979}, } @ARTICLE{Stevenson2002a, author = {.}, title = {A stable, direct solver for the gradient equation}, journal = {Math. Comput.}, year = {2002}, volume = {72}, number = {241}, pages = {41--53}, month = jun, note = {S 0025-5718(02)01436-9}, } @ARTICLE{KnyazevWidlund2001, author = {. and .}, title = {Lavrentiev Regularization + Ritz approximation = Uniform finite element error estimates for differential equations with rough coefficients}, journal = {Math. Comput.}, year = {2001}, volume = {72}, number = {241}, pages = {17--40}, month = jun, note = {S 0025-5718(01)01378-3}, } @ARTICLE{LiuYan2003, author = {. and .}, title = {A posteriori error estimates for optimal control problems governed by parabolic equations}, journal = {Numerische Mathematik}, year = {2003}, volume = {93}, pages = {497--521}, note = {DOI 10.1007/s002110100380}, } @ARTICLE{GradinaruHiptmair2003, author = {. and .}, title = {Mixed finite elements on soarse grids}, journal = {Numerische Mathematik}, year = {2003}, volume = {93}, pages = {471--495}, note = {DOI 10.1007/s002110100382}, } @ARTICLE{BrennerHe2003, author = {. and .}, title = {Lower bounds for three--dimensional nonoverlapping domain decomposition algorithms}, journal = {Numerische Mathematik}, year = {2003}, volume = {93}, pages = {445--470}, note = {DOI 10.1007/s002110100376}, } @ARTICLE{BaoHan2003, author = {. and .}, title = {Error bounds for the finite element approximation of an incompressible material in an unbounded domain}, journal = {Numerische Mathematik}, year = {2003}, volume = {93}, pages = {415--444}, note = {DOI 10.1007/s002110100387}, } @ARTICLE{SteinTezduyarBenney2003, author = {. and . and .}, title = {Computational Methods for Modeling Parachute Systems}, journal = {Computational Modeling}, year = {2003}, pages = {39--45}, } @ARTICLE{ErnstRotherWauerBalzer2003, author = {. and . and . and .}, title = {DLR`s VirtualLab: Scientific Software Just a Mouse Click Away}, journal = {Computing in Science and Engineering}, year = {2003}, volume = {3}, pages = {70--79}, } @ARTICLE{ForbesFoster2003, author = {. and .}, title = {The End of Moore`s Law?}, journal = {Computing in Science and Engineering}, year = {2003}, } @ARTICLE{AlbarrealCalzadaCruzFernandez-CaraGaloMarin2002, author = {. and . and .}, title = {Convergence analysis and error estimates for a parallel algorithm for solving the {Navier}--{Stokes} equations}, journal = {Numerische Mathematik}, year = {2002}, volume = {93}, pages = {201--221}, note = {DOI 10.1007/s00211010383}, } @ARTICLE{Tuma2002, author = {.}, title = {A note on the \$LDL\^{} T\$ decomposition of matrices from saddle--point problems}, journal = {SIAM Journal on Matrix Analysis and Applications}, year = {2002}, volume = {23}, number = {4}, pages = {903--915}, } @ARTICLE{AmaraCapatina-PagaghiucTrujillo2000, author = {. and .}, title = {Numerical approximation of {Navier}--{Stokes} equations with non standard boundary conditions}, journal = {International Journal for Numerical Methods in Fluids}, year = {2000}, pages = {1--6}, } @ARTICLE{CohenCohen1997, author = {. and .}, title = {Finite element methods for active contour models and ballons for 2--d and 3--d images}, journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, year = {1997}, pages = {1131--1147}, } @ARTICLE{KassWitkinTerzopoulus1988, author = {. and . and .}, title = {Snakes: Active contour models}, journal = {International Journal of Computer Vision}, year = {1988}, pages = {321--331}, } @ARTICLE{Terzopoulus1986, author = {.}, title = {Regularization of inverse visual problems involving discontinuities}, journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, year = {1986}, } @ARTICLE{CohenCohen1993, author = {. and .}, title = {Finite element methods for active contour models and ballons for 2--d and 3--d images}, journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, year = {1993}, pages = {1131--1147}, } @ARTICLE{TerzopoulusWitkinKass1987, author = {. and . and .}, title = {Symmetrie--seeking models and 3d object construction}, journal = {International Journal of Computer Vision}, year = {1987}, pages = {211--221}, } @ARTICLE{ZhangLin2003, author = {. and .}, title = {Ultraconvergence of ZZ Patch Recovery at Mesh Symmetry Points}, journal = {Numerische Mathematik}, year = {2003}, volume = {95}, pages = {781--801}, note = {DOI 10.1007/s00211-003-0457-x}, } @ARTICLE{WangLi2003, author = {. and .}, title = {A nonconforming combination of the Finite Element and Finite Volume methods with an anisotropic mesh refinement for a singularly perturbed convection--diffusion equation}, journal = {Math. Comput.}, year = {2003}, volume = {72}, number = {244}, pages = {1689--1709}, note = {S 0025-5718(03)01516-3}, } @ARTICLE{Kaltenbacher2003, author = {.}, title = {{V}--Cycle convergence of some multigrid methods for ill--posed problems}, journal = {Math. Comput.}, year = {2003}, volume = {72}, number = {244}, pages = {1711--1730}, note = {S 0025-5718(03)01533-3}, } @ARTICLE{YamamotoIshiyamaNakajimaNakamuraMori2003, author = {.}, title = {Three--dimensional viscoelastic flows through a rectangular channel with a cavity}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {114}, pages = {13--31}, note = {DOI 10.1016/s0377-0257(03)00115-0}, } @ARTICLE{DeanGlowinski2003, author = {. and .}, title = {Numerical Solution of the Two--Dimensional Elliptic {M}onge--{A}mpere Equation with {D}irichlet Boundary Conditions: An Augmented {L}agrangian Approach}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2003}, volume = {I}, number = {336}, pages = {779--784}, note = {DOI 10.1016/s1631-073X(30)00149-3}, } @ARTICLE{LikhtmanGraham2003, author = {. and .}, title = {Simple constitutive equation for linear poplymer melts derived from molecular theory: Rolie--Poly equation}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {114}, pages = {1--12}, note = {DOI 10.1016/S0377-0257(03)00114-9}, } @ARTICLE{Ragazzo2003, author = {.}, title = {On the motion of solids through an ideal liquid: approximated equations for many body systems}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {63}, number = {6}, pages = {1972--1997}, } @ARTICLE{GuidoSimeoneGreco2003, author = {. and . and .}, title = {Deformation of a Newtonian drop in a viscoelastic matrix under steady shear flow {}--{} Experimental validation of slow tflow theory}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {114}, pages = {65--82}, note = {DOI 10.1016/s0377-0257(03)00118-6}, } @ARTICLE{Oliveira2003, author = {.}, title = {Asymmetric flows of viscoelastic fluids in symmetric planar geometries}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2003}, volume = {114}, pages = {33--63}, note = {DOI 10.1016/s0377-0257(03)00117-4}, } @ARTICLE{Kirchner2002, author = {.}, title = {Thermodynamically consistent modelling of abrasive granular materials. Non--equilibrium theory}, journal = {Proc. R. Soc. Lond.}, year = {2002}, volume = {458}, pages = {2153--2176}, note = {DOI 10.1098/rspa.2002.0963}, } @ARTICLE{Cao2002, author = {.}, title = {On the convergence of iterative methods for solving singular linear systems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {145}, pages = {1--9}, } @ARTICLE{Reusken2002, author = {.}, title = {Convergence analysis of a multigrid method for convection--diffusion equations}, journal = {Numerische Mathematik}, year = {2002}, volume = {91}, pages = {323--349}, note = {DOI 10.1007/s002110100312}, } @ARTICLE{StockieMackenzieRussell2001, author = {. and . and .}, title = {A moving mesh method for one--dimensional hyperbolic conservation laws}, journal = {SIAM Journal on Scientific Computing}, year = {2001}, volume = {22}, number = {5}, pages = {1791--1813}, } @ARTICLE{PanRamiWang2002, author = {. and . and .}, title = {Structured matrices and Newton`s iteration: unified approach}, journal = {Linear Algebra and\_its Applications}, year = {2002}, volume = {343}, number = {344}, pages = {233--265}, } @ARTICLE{ChenEberlyKaltofenSaundersTurnerVillard2002, author = { .}, title = {Efficient matric preconditioners for black box linear algebra}, journal = {Linear Algebra and\_its Applications}, year = {2002}, volume = {343}, number = {344}, pages = {119--146}, } @ARTICLE{TocinoArdanuy2002, author = {. and .}, title = {Runge.Kutta methods for numerical solution of stochastic differential equations}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {219--241}, } @ARTICLE{YoungTsaiEldhoCheng2002, author = {. and . and . and .}, title = {Solution of {Stokes} Flow Using an Iterative DRBEM Based on Compactly--Supported, Positive--Definite Radial Basis Function}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {607--619}, } @ARTICLE{Melenk2002, author = {.}, title = {On condition numbers in \$hp\$--{FEM} with {Gauss}--Lobatto--based shape functions}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {139}, pages = {21--48}, } @ARTICLE{LehnRoesslerSchein2002, author = {. and R\"{o}". and .}, title = {Adaptive schemes for the numerical solution of SDEs {}--{}--{} a comparison}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {297--308}, } @ARTICLE{FangTsuchiyaYamamoto2002, author = {. and . and .}, title = {Finite difference, finite element and finite volume methods applied to two--point boundary problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {139}, pages = {9--19}, } @ARTICLE{Fasshauer2002, author = {.}, title = {Newton Iteration with Multiquadrics for the Solution of Nonlinear PDEs}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {423--438}, } @ARTICLE{DingguoWeiwen2002, author = {. and .}, title = {Globally convergent inexact generalized Newton\"{A}s method for nonsmooth equations}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {37--49}, } @ARTICLE{JungNicaiseTabka2002, author = {. and . and .}, title = {Some multilevel methods on graded meshes}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {151--171}, } @ARTICLE{Cui2002, author = {.}, title = {A suffivient condition for the convergence of the inexact Uzawa algorithm for saddle point problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {139}, pages = {189--196}, } @ARTICLE{Tidriri2002, author = {.}, title = {Analysis of the hybrid finite element/finite volume methods for linear hyperbolic and convection--dominated convection--diffusion problems}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {139}, pages = {323--350}, } @ARTICLE{GasparClaveroLisbona2002, author = {. and .}, title = {Some numerical experiments with multigrid methods on Shishkin meshes}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {21--35}, } @ARTICLE{BhuruthJainGopaul.A.2002, author = {. and . and .}, title = {Preconditioned iterative methods for the nine--point approximation to the convection--diffusion equation}, journal = {J. Comput. Appl. Math.}, year = {2002}, volume = {138}, pages = {73--92}, } @ARTICLE{Acton1998, author = {.}, title = {Multigrid Anisotropic Diffusion}, journal = {IEEE Transactions on Image Processing}, year = {1998}, volume = {7}, number = {3}, pages = {280--291}, } @ARTICLE{SchoetzauSchwab2001, author = {Sch\"{. and .}, title = {\$hp\$--discontinuous {Galerkin} time--stepping for parabolic problems}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2001}, volume = {I}, number = {333}, pages = {1121--1126}, } @ARTICLE{GuermondShen2001, author = {. and .}, title = {Quelques resultats nouveaux sur les methodes de projection}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2001}, volume = {I}, number = {333}, pages = {1111--1116}, } @ARTICLE{George1973, author = {.}, title = {Nested dissection of a regular Finite Element mesh}, journal = {SIAM Journal on Numerical Analysis}, year = {1973}, volume = {10}, number = {2}, pages = {345--363}, } @ARTICLE{MittalAl-Kurdi2002, author = {. and .}, title = {LU--Decomposition and Numerical Structure for Solving Large Sparse Nonsymmetric Linear systems}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {131--155}, } @ARTICLE{HavuPitkaeranta2001, author = {. and .}, title = {An analysis of finite element locking in a parameter dependent model problem}, journal = {Numerische Mathematik}, year = {2001}, volume = {89}, pages = {691--714}, note = {DOI 10.1007/s002110100277}, } @ARTICLE{Dekker2001, author = {.}, title = {Partitioned--GMRES in domain decomposition with approximate subdomain solution}, journal = {BIT Numerical Mathematics}, year = {2001}, volume = {41}, number = {5}, pages = {924--935}, } @ARTICLE{LangVerwer2001, author = {. and .}, title = {ROS3P {}--{}--{} an accurate third--order Rosenbrock solver designed for parabolic problems}, journal = {BIT Numerical Mathematics}, year = {2001}, volume = {41}, number = {4}, pages = {731--738}, } @ARTICLE{SakuraiAokiLeeKato2002, author = {. and . and . and .}, title = {Poisson Equation Solver with Fourth--Order Accuracy by using Interpolated Differential Operator Scheme}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {621--630}, } @ARTICLE{HanHuang2002a, author = {. and .}, title = {A Class of Artificial Boundary Conditions for Heat Equation in Unbounded Domains}, journal = {Computers and Mathematics with Applications}, year = {2002}, volume = {43}, pages = {889--900}, } @ARTICLE{LiRenardy2000a, author = {. and .}, title = {Shear--induced rupturing of a viscous drop in a Bingham liquid}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2000}, volume = {95}, pages = {235--251}, } @ARTICLE{Pukhnachev1998, author = {.}, title = {The quasistationary approximation in tge problem of the motion of an isolated volume of a viscous incompressible capillary liquid}, journal = {Journal of Applied Mechanics}, year = {1998}, volume = {62}, number = {6}, pages = {927--937}, } @ARTICLE{Zegeling2004, author = {.}, title = {Tensor--poroduct adaptive grids based on coordinate transformations}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {166}, pages = {343--360}, } @ARTICLE{Gazzola1997, author = {.}, title = {A note on the evolution {Navier}--{Stokes} equations with a pressure--dependent viscosity}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Physik}, year = {1997}, volume = {48}, pages = {760--773}, } @ARTICLE{ConwayGlasser2004, author = {. and .}, title = {Density waves and coherent structures in granular Couette flows}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {3}, pages = {509--529}, month = mar, } @ARTICLE{Tardos1997, author = {.}, title = {A fluid mechanistic approach to slow, frictional flow of powders}, journal = {Powder Technology}, year = {1997}, volume = {92}, pages = {61--74}, } @ARTICLE{Rombach2002, author = {.}, title = {Festlegung der zul\"{a}ssigen Spannstahlspannungen}, journal = {0}, year = {2002}, volume = {19}, pages = {145--155}, } @ARTICLE{Beuchler2003, author = {.}, title = {AMLI preconditioner for the \$p\$--version of the {FEM}}, journal = {Numerical Linear Algebra and Applications}, year = {2003}, volume = {10}, pages = {721--732}, note = {DOI 10.1002/nla.329}, } @ARTICLE{Kunert2002, author = {.}, title = {Toward Anisotropic Mesh Construction and Error Estimation in the Finite Element Method}, journal = {Numerical Methods for Partial Differential Equations}, year = {2002}, volume = {18}, number = {6}, pages = {625--648}, } @ARTICLE{Zhao2004, author = {.}, title = {Convergence of {V}--cycle and {F}--cycle multigrid methods for the biharminic problem using the Morley element}, journal = {Electronic Transactions on Numerical Analysis }, year = {2004}, volume = {17}, pages = {112--132}, } @ARTICLE{Gautschi2002, author = {.}, title = {The interplay between classical analysis and (numerical) linear algebra {}--{}--{} a tribute to }, journal = {Electronic Transactions on Numerical Analysis }, year = {2002}, volume = {13}, pages = {119--147}, } @ARTICLE{JohnMaubachTobiska1997, author = {. and . and .}, title = {Nonconforming streamline--diffusion--finite--element--method for convection--diffusion problems}, journal = {Numerische Mathematik}, year = {1997}, volume = {78}, pages = {165--188}, } @ARTICLE{CarstensenBartelsJansche2002, author = {. and . and .}, title = {A posteriori error estimates for nonconforming finite element methods}, journal = {Numerische Mathematik}, year = {2002}, volume = {92}, pages = {233--256}, note = {DOI 10.1007/s002110100378}, } @ARTICLE{Verfuerth1998, author = {.}, title = {A posteriori error estimates for nonlinear problems. \$L\^{} r(0,T;L\^{} p(/Omega))\$--error estimates for finite element discretisations of parabolic equations}, journal = {Math. Comput.}, year = {1998}, volume = {67}, number = {224}, pages = {1335--1360}, } @ARTICLE{FormaggiaPerotto2003, author = {. and .}, title = {Anisotropic Error Estimates for Elliptic Problems}, journal = {Numerische Mathematik}, year = {2003}, volume = {94}, pages = {67--92}, note = {DOI 10.1007/s00211-002-0415-z}, } @ARTICLE{ArmentanoDuran2004, author = {. and .}, title = {Asymptotic lower bounds for eigenvalues by nonconforming Finite Element methods}, journal = {Electronic Transactions on Numerical Analysis }, year = {2004}, volume = {17}, pages = {91--101}, } @ARTICLE{Silva2003, author = {.}, title = {Cache aware data laying for the {Gauss}--Seidel smoother}, journal = {Electronic Transactions on Numerical Analysis }, year = {2003}, volume = {15}, pages = {66--77}, } @ARTICLE{Brenner2004, author = {.}, title = {Discrete Sobolev and Poincare inequalities for piecewise polynomial functions}, journal = {Electronic Transactions on Numerical Analysis }, year = {2004}, volume = {18}, pages = {42--48}, } @ARTICLE{BelgacemBrenner2001, author = {. and .}, title = {Some nonstandard Finite Element estimates with applications to 3D poisson and signorini problems}, journal = {Electronic Transactions on Numerical Analysis }, year = {2001}, volume = {12}, pages = {134--148}, } @ARTICLE{DallOsso2004, author = {.}, title = {An iterative back substitution algorithm for the solution of tridiagonal matrix systems with fringes}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {169}, pages = {87--99}, note = {DOI: 10.1016/j.cam.2003.11.004}, } @ARTICLE{LangCaoHuangRussel2003, author = {. and .}, title = {A two--dimensional moving finite element method with local refinement based on a posteriori error estimates}, journal = {Applied Numerical Mathematics}, year = {2003}, volume = {46}, pages = {75--94}, note = {DOI 10.1016/S0168-9274(03)00013-8}, } @ARTICLE{AsakuraAsariNakajima1997, author = {. and . and .}, title = {Simulation of solid--liquid flows in a vertical pipe by a collision model}, journal = {Powder Technology}, year = {1997}, volume = {94}, pages = {201--206}, } @ARTICLE{Shamolin2004, author = {.}, title = {Geometrical representation of the motion of a body through a medium}, journal = {Int. J. Appl. Math. Mech.}, year = {2004}, volume = {40}, number = {4}, pages = {480--486}, } @ARTICLE{AmaraBarucqDuloue2004, author = {.}, title = {A mixed convergent formulation for the three--dimensional {Stokes} equations}, journal = {Calcolo}, year = {2004}, volume = {41}, pages = {37--64}, note = {DOI 10.1007/s10092-004-0083-8}, } @ARTICLE{Rezakhanlou2004, author = {.}, title = {Boltzmann--Grad Limits for Stochastic Hard Sphere Models}, journal = {Communications in Mathematical Physics}, year = {2004}, volume = {248}, pages = {553--637}, note = {DOI 10.1007/s00220-004-1101-z}, } @ARTICLE{ParkSheen2003, author = {. and .}, title = {\$P\_1\$--Nonconforming Quadrilateral Finite Element Methods for Second--Order Elliptic Problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {624--640}, } @ARTICLE{KanschatSuttmeier1999, author = {. and .}, title = {A Posteriori Error Estimates for Nonconforming Finite Element Schemes}, journal = {Calcolo}, year = {1999}, volume = {36}, pages = {129--141}, } @ARTICLE{HouzeauxCodina2004, author = { .}, title = {A finite element model for the simulation of lost foam casting}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {203--226}, note = {DOI: 10.1002/fld.757}, } @ARTICLE{LatyshevYushkanov2004, author = {. and .}, title = {A Method for Solving Boundary Value Problems for Kinetic Equations}, journal = {Computational Mathematics and Mathematical Physics}, year = {2004}, volume = {44}, number = {6}, pages = {1051--1061}, } @ARTICLE{Al-ZoubiBrenner2004, author = {. and .}, title = {Comparative Study of Thermal Flows with Different Finite Volume and Lattice Boltzmann Schemes}, journal = {Int. J. Mod. Phys.}, year = {2004}, volume = {15}, number = {2}, pages = {307--319}, } @ARTICLE{Waltz2004, author = {.}, title = {Parallel adaptive refinement for unsteady flow calculations on 3D unstructured grids}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {37--57}, note = {DOI: 10.1002/fld.674}, } @ARTICLE{DohrmannBochev2004, author = {. and .}, title = {A stabilized finite element method for the {Stokes} problem based on polynomial pressure projections}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {183--201}, note = {DOI: 10.1002/fld.752}, } @ARTICLE{TangTaoHe2004, author = {. and . and .}, title = {Lattice Boltzmann method for simulating gas flow in microchannels}, journal = {Int. J. Mod. Phys.}, year = {2004}, volume = {15}, number = {2}, pages = {335--347}, } @ARTICLE{KaraaZhang2004, author = {. and .}, title = {High Order ADI Method for Solving Unsteady Convection--Diffusion Problems}, journal = {J. Comput. Phys.}, year = {2004}, volume = {198}, pages = {1--9}, } @ARTICLE{TaiZhao2004, author = {. and .}, title = {A finite volume unstructured multigrid method for efficient computation of unsteady incompressible viscous flows}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {59--84}, note = {DOI: 10.1002/fld.742}, } @ARTICLE{PretoriusLehner2004, author = { .}, title = {Adaptive mesh refinement for characteristic codes}, journal = {J. Comput. Phys.}, year = {2004}, volume = {198}, pages = {10--34}, } @ARTICLE{VidovicSegalWesseling2004, author = {. and . and .}, title = {A superlinearly convergent finite volume method for the incompressible {Navier}--{Stokes} equations on staggered unstructured grids}, journal = {J. Comput. Phys.}, year = {2004}, volume = {198}, pages = {159--177}, } @ARTICLE{PasquettiRapetti2004, author = {. and .}, title = {Spectral element methods on triangles and quadrilaterals: comparisons and applications}, journal = {J. Comput. Phys.}, year = {2004}, volume = {198}, pages = {349--362}, } @ARTICLE{SteijlHoeijmakers2004, author = {Hoeijmakers, . and .}, title = {Fourth--order accurate compact--difference discretization method for Helmhotlz and incompressible {Navier}--{Stokes} equations}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {227--244}, note = {DOI: 10.1002/fld.750}, } @ARTICLE{ZhangKwonYoun2004, author = {. and . and .}, title = {Least--squares Meshfree Method for Incompressible {Navier}--{Stokes} Problems}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {263--288}, note = {DOI: 10.1002/fld758}, } @ARTICLE{Aristov2004, author = {.}, title = {Solution to the Boltzmann Equation at Small Knudsen Numbers}, journal = {Computational Mathematics and Mathematical Physics}, year = {2004}, volume = {44}, number = {6}, pages = {1069--1081}, } @ARTICLE{PrivalovPrivalovaSamsonov2003, author = {. and . and .}, title = {On the dynamics of a boomerang}, journal = {Mechanics of Solids}, year = {2003}, volume = {38}, number = {4}, pages = {28--38}, } @ARTICLE{KumarKumarDas2004, author = {. and . and .}, title = {A New Class of Stabilized Mesh--free Finite Elements for the Approximation of the {Stokes} Problem}, journal = {Numerical Methods for Partial Differential Equations}, year = {2004}, volume = {20}, number = {5}, pages = {703--722}, note = {DOI 10.1002/num.20007}, } @ARTICLE{HiranoYoshidaOzoeOkamoto2004, author = {. and . and . and .}, title = {A Study on Preconditioning Iterative Methods an Relativity of Pressure in the Numerical Calculation of Fluid Flow}, journal = {Numerical Methods for Partial Differential Equations}, year = {2004}, volume = {20}, number = {5}, pages = {789--802}, note = {DOI: 10.1002/num20015}, } @ARTICLE{KwakLee2004, author = {. and .}, title = {Multigrid Algorithm for the Cell--Centered Finite Difference Method II: Discontinuous Coefficient Case}, journal = {Numerical Methods for Partial Differential Equations}, year = {2004}, volume = {20}, number = {5}, pages = {742--764}, note = {DOI: 10.1002/num.20001}, } @ARTICLE{Douglas1996, author = {.}, title = {Caching in with multigrid algorithms: problems in two dimensions}, journal = {Parallel Algorithms and Applications}, year = {1996}, volume = {9}, pages = {195--204}, } @ARTICLE{DouglasHuKarlKowarschikRuedeWeiss2000a, author = {. and . and . and . and . and .}, title = {Cache optimization for structured and unstructured grid multigrid}, journal = {Electronic Transactions on Numerical Analysis }, year = {2000}, volume = {10}, pages = {25--40}, } @ARTICLE{DouglasHuRayThorneTuminaro2004, author = {. and .}, title = {Cache aware multigrid for variable coefficient elliptic problems on adaptive mesh refinement hierarchies}, journal = {Numerical Linear Algebra and Applications}, year = {2004}, volume = {11}, pages = {173--187}, } @ARTICLE{DouglasThorne2002, author = {. and .}, title = {A note on cache memory methods for multigrid in three dimensions}, journal = {Contemporary Mathematics}, year = {2002}, volume = {306}, pages = {167--177}, } @ARTICLE{DouglasHaaseHuKarlKowarschikRuedeWeiss2000, author = { .}, title = {Portable memory hierarchy techniques for {PDE} solvers: Part I}, journal = {SIAM News}, year = {2000}, volume = {33}, number = {5}, pages = {1, 8-9}, } @ARTICLE{Keyes2002, author = {.}, title = {Terascale Implicit Methods for Partial Differential Equations}, journal = {Contemporary Mathematics}, year = {2002}, volume = {306}, pages = {29--84}, } @ARTICLE{GroppKaushikKeyesSmith2001, author = {. and . and . and .}, title = {High Performance Parallel Implicit {CFD}}, journal = {Parallel Computing}, year = {2001}, volume = {27}, pages = {337--362}, } @ARTICLE{HayderKeyesMehrotra1998, author = {. and .}, title = {A Comparison of PETSc Library and HPF Implementations of an Archetypal {PDE} Computation}, journal = {Advances in Engineering Software}, year = {1998}, volume = {29}, pages = {415--424}, } @ARTICLE{GriebelZumbusch1999a, author = {. and .}, title = {Parallel multigrid in an adaptive {PDE} solver based on hashing and space--filling curves}, journal = {Parallel Computing}, year = {1999}, volume = {25}, number = {7}, pages = {827--843}, } @ARTICLE{GriebelZumbusch1999b, author = {. and .}, title = {Parallel adaptive subspace correction schemes with applications to elasticity}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1999}, volume = {184}, pages = {303--332}, } @ARTICLE{KnoppLubeRapin2002, author = {. and . and .}, title = {Stabilized finite element methods with shock capturing for advection--diffusion problems}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2002}, volume = {191}, pages = {2997--3013}, } @ARTICLE{NakaoHishidaTanakaShiina2004, author = {. and .}, title = {Solidification characteristics of rising immiscible oil droplets in coolant}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {5339--5349}, } @ARTICLE{AlipchenkovNigmatulinSolovievStonikZaichikZeigarnik2004, author = { .}, title = {A three--fluid model of two--phase dispersed--annular flow}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {5323--5338}, } @ARTICLE{RobinsonJudd2004, author = {. and .}, title = {The dynamics of spherical bubble growth}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {5101--5113}, } @ARTICLE{NakamuraIgarashi2004, author = {. and .}, title = {Variation of Nusselt number with flow regimes behind a circular cylinder for Reynolds numbers from 70 to 30 000}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {5169--5173}, } @ARTICLE{BuickPankajOoiChavez-SagarnagaPearceHoughton2004, author = { . and .}, title = {Motion of granular particles on the wall of a model silo and the associated wall vibrations}, journal = {Journal of Physics D: Applied Physics}, year = {2004}, volume = {37}, pages = {2751--2760}, } @ARTICLE{ArbenzHochstenbach2004, author = { .}, title = {A {Jacobi}--Davidson method for solving complex symmetric eigenvalue problems}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {5}, pages = {1655--1673}, } @ARTICLE{AdamsChartier2004, author = {. and .}, title = {New geometric immersed interface multigrid solvers}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {5}, pages = {1516--1533}, } @ARTICLE{Hoffman2004a, author = {.}, title = {On duality--based a posteriori error estimation in various norms and linear functionals for large eddy simulation}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, number = {1}, pages = {178--195}, } @ARTICLE{BochevP.Gunzburger2004, author = {. and .}, title = {An absolutely stable pressure--poisson stabilized finite element method for the {Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {47}, number = {3}, pages = {1189--1207}, } @ARTICLE{LiuMaTangYan2004, author = {. and . and . and .}, title = {A posteriori error estimates for discontinuous {Galerkin} time--stepping method for optimal control problems governed by parabolic equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {42}, number = {3}, pages = {1032--1061}, } @ARTICLE{HuangRussel1999, author = {. and .}, title = {Moving mesh strategy based on a gradient flow equation for two--dimensional problems}, journal = {SIAM Journal on Scientific Computing}, year = {1999}, volume = {20}, number = {3}, pages = {998--1015}, } @ARTICLE{WangZhang2003, author = {. and .}, title = {MSP: A Class of Parallel Multistep Successive Sparse Approximate Inverse Preconditioning Strategies}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {24}, number = {4}, pages = {1141--1156}, } @ARTICLE{KimmelYavneh2003, author = {. and .}, title = {An algebraic multigrid approach for image analysis}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {24}, number = {4}, pages = {1218--1231}, } @ARTICLE{SalisburyContiBarbagli2004, author = {. and .}, title = {Haptic Rendering: Introductory Concepts}, journal = {IEEE Computer Graphics and Applications}, year = {2004}, volume = {24}, number = {2}, pages = {24--32}, } @ARTICLE{SobelForsbergLaidlawZeleznikKeefePivkinKarniadakisRichardsonSwartz2004, author = { .}, title = {Particle Flurries; Synoptic 3D Pulsatile Flow Visualization}, journal = {IEEE Computer Graphics and Applications}, year = {2004}, volume = {24}, number = {2}, pages = {76--84}, } @ARTICLE{LinSalisbury2004, author = {. and .}, title = {Haptic Rendering {}--{} Beyond Visual Computing}, journal = {IEEE Computer Graphics and Applications}, year = {2004}, volume = {24}, number = {2}, pages = {22--23}, } @ARTICLE{GerstnerGriebel2003, author = {. and .}, title = {Dimension--adaptive tensor--product quadrature}, journal = {Computing}, year = {2003}, volume = {71}, number = {1}, pages = {65--87}, note = {DOI: 10.1007/s00607-003-0015-5}, } @ARTICLE{SchwabTodor2003, author = { .}, title = {Sparse finite elements for stochastic elliptic problems {}--{} higher order moments}, journal = {Computing}, year = {2003}, volume = {71}, number = {1}, pages = {43--63}, note = {DOI: 10.1007/s00607-003-0024-4}, } @ARTICLE{Achatz2003, author = {.}, title = {Higher order sparse grid methods for elliptic partial differential equations with variable coefficients}, journal = {Computing}, year = {2003}, volume = {71}, number = {1}, pages = {1--15}, note = {DOI: 10.1007/s00607-003-0012-8}, } @ARTICLE{GradinaruHiptmair2003a, author = {. and .}, title = {Multigrid for discrete differential forms on sparse grids}, journal = {Computing}, year = {2003}, volume = {71}, number = {1}, pages = {17--42}, note = {DOI: 10.1007/s00607-003-0008-4}, } @ARTICLE{DouglasHaaseHuKarlKowarschikRuedeWeiss2000a, author = { \"{.}, title = {Portable memory hierarchy techniques for {PDE} solvers: Part II}, journal = {SIAM News}, year = {2000}, volume = {33}, number = {6}, } @ARTICLE{PooleLiuMandel2003, author = {. and . and .}, title = {Advancing Analysis Capabilities in Ansys through Solver Technology}, journal = {Electronic Transactions on Numerical Analysis }, year = {2003}, volume = {15}, pages = {106--121}, } @ARTICLE{BastianBirkenJohannsenLangNeussRentz-ReichertWieners1997, author = { .}, title = {UG {}--{} a flexible software toolbox for solving partial differential equations}, journal = {Computing and Visualization in Science}, year = {1997}, volume = {1}, pages = {27--40}, } @ARTICLE{LawsonHansonKincaidKrogh1979, author = {. and . and . and .}, title = {Basic Linear Algebra Subprograms for Fortran Usage}, journal = {ACM Transactions on Mathematical Software}, year = {1979}, volume = {5}, number = {3}, pages = {308--323}, } @ARTICLE{DongarraDuCrozHammarlingDuff1990, author = {. and .}, title = {A Set of Level 3 Basic Linear Algebra Subprograms}, journal = {ACM Transactions on Mathematical Software}, year = {1990}, volume = {16}, number = {1}, pages = {1--17}, } @ARTICLE{WhaleyPetitetDongarra2001, author = {. and . and .}, title = {Automated Empirical Optimization of Software and the ATLAS Project}, journal = {Parallel Computing}, year = {2001}, volume = {27}, number = {1-2}, pages = {3--35}, } @ARTICLE{ActisSzabo2003, author = {. and .}, title = {Analysis of Bounded and Fastened Repairs by the p--Version of the Finite--Element Method}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {1--14}, } @ARTICLE{CalvoDuranyVazquez2003, author = {. and . and .}, title = {Finite elements numerical solution of a coupled profile--velocity--temperature shallow ice sheet approximation model}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {158}, pages = {31--41}, } @ARTICLE{SchoenauerAdolph2003, author = {. and .}, title = {FDEM: how we make the FDM more flexible than the {FEM}}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {158}, pages = {157--167}, } @ARTICLE{AinsworthCoggins2002, author = {. and .}, title = {A Uniformly Stable Family of Mixed hp--Finite Elements with Continuous Pressures for Incompressible Flow}, journal = {IMA Journal of Numerical Analysis}, year = {2002}, volume = {22}, number = {2}, pages = {307--327}, month = apr, } @ARTICLE{MatsuuraOharuTakataTamura2003, author = {. and . and .}, title = {Mathematical approaches to bone reformation phenomena an numerical simulations}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {158}, pages = {107--119}, } @ARTICLE{Diaz-GoanoMinevNandakumar2003, author = {. and . and .}, title = {A fictitious domain/finite element method for particulate flows}, journal = {J. Comput. Phys.}, year = {2003}, volume = {192}, pages = {105--123}, } @ARTICLE{BasuJorgeBadriLin2003, author = {. and . and . and .}, title = {Higher--Order Modeling of Continua by Finite--Element, Boundary--Element, Meshles, and Wavelet Methods}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {15--33}, } @ARTICLE{CareyAndersonCarnesKirk2004, author = {. and . and . and .}, title = {Some aspects of adaptive grid technology related to boundary and interior layers}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {166}, pages = {55--86}, } @ARTICLE{GlowinskiHeRappazWagner2003, author = {. and . and . and .}, title = {Approximation of multi--scale elliptic problems using patches of finite elements}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2003}, volume = {337}, pages = {679--684}, } @ARTICLE{AmatBusquierCandela2003, author = {. and . and .}, title = {Third--order iterative methods without using any Frechet derivative}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {158}, pages = {11--18}, } @ARTICLE{LiuChellappaRosenfeld2003, author = {. and . and .}, title = {Accurate Dense Optical Flow Estimation Using Adaptive Structure Tensors and a Parametric Model}, journal = {IEEE Transactions on Image Processing}, year = {2003}, volume = {12}, number = {10}, pages = {1170--1180}, month = oct, } @ARTICLE{LiLideGroh2003, author = { .}, title = {Effect of magnetic field on g--jitter induced convection and solute striation during solidification in space}, journal = {International Journal of Heat and Mass Transfer}, year = {2003}, volume = {46}, pages = {4799--4811}, } @ARTICLE{GodaHibikiKimIshiiUhle2003, author = {. and . and . and . and .}, title = {Drift--lux model for downward two--phase flow}, journal = {International Journal of Heat and Mass Transfer}, year = {2003}, volume = {46}, pages = {4835--4844}, } @ARTICLE{Pozrikidis2004, author = {.}, title = {Boundary conditions for shear flow past a permeable interface modeled as an array of cylinders}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {1--17}, } @ARTICLE{KimYangSenda2004, author = {. and . and .}, title = {Large eddy simulation of turbulent flow past a square cylinder confined in a channel}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {81--69}, } @ARTICLE{Zhao2004a, author = {.}, title = {Computation of complex turbulent flow using matrix--free implicit dual time--stepping scheme and LRN turbulence model on unstructured grids}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {119--136}, } @ARTICLE{BerthoGiorgiutti-DauphineHulin2003, author = { .}, title = {Intermittend dry granular flow in a vertical pipe}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {11}, pages = {3358--3369}, month = nov, } @ARTICLE{BalmforthCrasterToniolo2003, author = {. and . and .}, title = {Interfacial instability in non--Newtonian fluid layers}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {11}, pages = {3370--3384}, } @ARTICLE{BagchiBalachandar2003, author = {. and .}, title = {Effect of turbulence on the drag and lift of a particle}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {11}, pages = {3496--3513}, } @ARTICLE{Ali2003, author = {.}, title = {Low Mach Number Flows in Time--Dependent Domains}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {63}, number = {6}, pages = {2020--2041}, } @ARTICLE{BuchmannPetersen2003, author = {. and .}, title = {Solving Dirichlet Problems Numerically Using the Feynman--Kac Representation}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {519--540}, } @ARTICLE{ClimentPerea2003, author = {. and .}, title = {Iterative methods for least--square problems based on proper splittings}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {158}, pages = {43--48}, } @ARTICLE{Huhtanen2003, author = {.}, title = {Two splittings of a square matrix}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {563--570}, } @ARTICLE{ZhuYinLuk2004, author = {. and . and .}, title = {Numerical characteristics of a simple finite element formulation for consolidation analysis}, journal = {Communications in Numerical Methods in Engineering}, year = {2004}, volume = {20}, pages = {767--775}, note = {DOI: 10.1002/cnm.662}, } @ARTICLE{He2004, author = {.}, title = {A modified Newton--Raphson method}, journal = {Communications in Numerical Methods in Engineering}, year = {2004}, volume = {20}, pages = {801--805}, note = {DOI: 10.1002/cnm.664}, } @ARTICLE{ToselliSchwab2003, author = {. and .}, title = {Mixed HP--finite element approximations on geometric edge and boundary layer meshes in three dimensions}, journal = {Numerische Mathematik}, year = {2003}, volume = {94}, pages = {771--801}, note = {DOI: 10.1007/s00211-002-0393-1}, } @ARTICLE{AmaraBaranger2003, author = {. and .}, title = {An extra stress {}--{} vorticity formulation of {Stokes} problem for the Oldroyd viscoelastic model}, journal = {Numerische Mathematik}, year = {2003}, volume = {94}, pages = {603--622}, note = {DOI: 10.1007/s00211-002-0440-y}, } @ARTICLE{NakaiWuSogabeArimitsu2004, author = {. and . and .}, title = {A Study of Thickness Optimization of Golf Club Heads to Maximize Release Velocity of Balls}, journal = {Communications in Numerical Methods in Engineering}, year = {2004}, volume = {20}, pages = {747--755}, note = {DOI: 10.1002/cnm.698}, } @ARTICLE{PatankarHuangKoJoseph2001, author = {. and . and . and .}, title = {Lift--off of a single particle in Newtonian and viscoelastic fluids by direct numerical simulation}, journal = {Journal of Fluid Mechanics}, year = {2001}, volume = {438}, pages = {67--100}, } @ARTICLE{KevlahanVasilyev2001, author = {. and .}, title = {An adaptive wavelet method for fluid--structure interaction}, journal = {Direct Large-Eddy Simul. Workshop}, year = {2001}, volume = {4}, pages = {253--260}, } @ARTICLE{FengHuangJoseph1994, author = {. and . and .}, title = {The Turning Couples on an Elliptic Particle Settling in a Vertical Channel}, journal = {Journal of Fluid Mechanics}, year = {1994}, volume = {271}, pages = {1--16}, } @ARTICLE{AmodioPaprzycki1997, author = { Paprzycki}, title = {A cyclic reduction approach to the numerical solution of boundary value ODEs}, journal = {SIAM Journal on Scientific Computing}, year = {1997}, volume = {18}, number = {1}, pages = {56--68}, } @ARTICLE{CarstensenJochimsen2003, author = {. and .}, title = {Adaptive Finite Element Methods for Microstructures? Numerical Experiments for a 2--Well Benchmark}, journal = {Computing}, year = {2003}, volume = {71}, number = {2}, pages = {175}, note = {DOI: 10.1007/s00607-003-0027-1}, } @ARTICLE{AhnKangKwon2003, author = { .}, title = {A Flexible Inverse Laplace Transform Algorithm and its Application}, journal = {Computing}, year = {2003}, volume = {71}, number = {2}, pages = {115--132}, note = {DOI: 10.1007/s00607-003-0023-5}, } @ARTICLE{Novati2003, author = {.}, title = {An Explicit One--Step Method for Stiff Problems}, journal = {Computing}, year = {2003}, volume = {71}, number = {2}, pages = {133--152}, note = {DOI: 10.1007/s00607-003-0021-7}, } @ARTICLE{KadoshCohen-OrYagel2003, author = {. and . and .}, title = {Tricubic Interpolation of Discrete Surfaces for Binary Volumes}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2003}, volume = {9}, number = {4}, pages = {580--586}, month = oct, } @ARTICLE{ChengDey2003, author = {. and .}, title = {Quality Meshing with weighted delaunay refinement}, journal = {SIAM Journal on Computing}, year = {2003}, volume = {33}, number = {1}, pages = {69--93}, } @ARTICLE{SchmidlMilenkovic2004, author = {. and .}, title = {A Fast Impulsive Contact Suite for Rigid Body Simulation}, journal = {IEEE Transactions on Visualization and Computer Graphics}, year = {2004}, volume = {10}, number = {2}, pages = {189--197}, month = mar, } @ARTICLE{CarstensenKlose2003, author = {. and .}, title = {A posteriori finite element error control for the p--Laplace problem}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {792--814}, } @ARTICLE{LimChangJoergensen2004, author = {. and . and .}, title = {A novel partial differential algebraic equation (PDAE) solver: iterative space--time conservation element/solution element (CE/SE) method}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {1309--1324}, } @ARTICLE{Upreti20004, author = {.}, title = {A new robust technique for optimal control of chemical engineering processes}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {1325--1336}, } @ARTICLE{Fabbri2004, author = {.}, title = {Effect of viscous dissipation on the optimization of the heat transfer in internally finned tubes}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {3003--3015}, } @ARTICLE{SlavtchevMendes2004, author = {. and .}, title = {Marangoni instability in binary liquid--liquid systems}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {27}, pages = {3269--3278}, } @ARTICLE{FujimotoTakuda2004, author = {. and .}, title = {Entrapment of air at 45 degrees oblique collision of a water drop with a smooth solid surface at room temperature}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {3301--3305}, } @ARTICLE{SarrisLekakisVlachos2004, author = {. and . and .}, title = {Natural convection in rectangular tanks heated locally from below}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {27}, pages = {3549--3563}, } @ARTICLE{NajibSandriZine2004, author = {. and . and .}, title = {On a posteriori estimates for a lineasized Oldroyd`s problem}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {167}, pages = {345--361}, } @ARTICLE{FultonFokasXenophontos2004, author = {. and . and .}, title = {An analytical method for linear elliptic PDEs and its numerical implementation}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {267}, pages = {465--483}, } @ARTICLE{GalianoJuengelVelasco2003, author = {. and . and .}, title = {A parabolic cross--diffusion szstem for granular materials}, journal = {SIAM Journal on Mathematical Analysis}, year = {2003}, volume = {35}, number = {3}, pages = {561--578}, } @ARTICLE{GlowinskiHeRapparzWagner2004, author = { .}, title = {A multi--domain method for solving numerically multi--scale elliptic problems}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {741--746}, } @ARTICLE{Fernandez-CaraGuerreroImanuvilovPuel2004, author = { .}, title = {Remarks on Exact Controllability for {Stokes} and {Navier}--Stokers Systems}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {375--380}, } @ARTICLE{SaasFailleNatafWillien2004, author = { .}, title = {Domain Decomposition for a Finite Volume Method on Non--matching Grids}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {407--412}, } @ARTICLE{Nicaise2004, author = {.}, title = {A posteriori residual error estimation of a cell--centered finite volume method}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {419--424}, } @ARTICLE{Osher1984, author = {.}, title = {Riemann solvers, the entropy condition, and difference approximations}, journal = {SIAM Journal on Numerical Analysis}, year = {1984}, volume = {21}, number = {2}, pages = {217--235}, } @ARTICLE{ArioliManzini2003, author = {. and .}, title = {Null space algorithm and spanning trees in solving Darcy`s equation}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {839--848}, } @ARTICLE{Axelsson2003, author = {.}, title = {A survey of algebraic multilevel iteration (AMLI) methods}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {863--879}, } @ARTICLE{BertacciniNg2003, author = { .}, title = {Band--Toeplitz preconditioned GMRES iterations for time--dependent PDES}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {901--914}, } @ARTICLE{Spiteri2003, author = {.}, title = {A new characterization of m--matrices and h--matrices}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {1019--1032}, } @ARTICLE{HemkerHoffmannRaalte2003, author = {. and . and .}, title = {Two--level fourier analysis of a multigrid approach for discontinuous {Galerkin} discretization}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {1018--1041}, } @ARTICLE{HertyKlar2003, author = {. and .}, title = {Modeling, simulation, and optimization of traffic flow networks}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {1066--10087}, } @ARTICLE{Blank2003, author = {.}, title = {Preconditioning via a Schur complement method: an application in state estimation}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {942--960}, } @ARTICLE{LacroixVassilevskiWheelerWheeler2003, author = {. and .}, title = {Iterative solution methods for modeling multiphase flow in porous media fully implicitly}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {905--926}, } @ARTICLE{WanChan2003, author = {. and .}, title = {A phase error analysis of multigrid methods for hyperbolic equations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {857--880}, } @ARTICLE{LeeLeveque2003, author = {. and .}, title = {An immersed interface method for incompressible {Navier}--{Stokes} equations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {832--856}, } @ARTICLE{Kanschat2003, author = {.}, title = {Preconditioning Methods for Local Discontinuous {Galerkin} Discretizations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {815--831}, } @ARTICLE{BrysonLevy2003, author = {. and .}, title = {Central schemes for multidimensional Hamilton--{Jacobi} equations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, number = {3}, pages = {767--791}, } @ARTICLE{EsmaeeliTryggvason2004, author = {. and .}, title = {A front tracking method for computations of boiling in complex geometries}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {1037--1050}, } @ARTICLE{CottetMaitre2004, author = {. and .}, title = {A level--set formulation of immersed boundary methods for fluid--structure interaction problems}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {581--586}, } @ARTICLE{CoimbraSerenoRodrigues2004, author = {. and . and .}, title = {Moving finite element method: applications to science and engineering problems}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {597--603}, } @ARTICLE{GerberKermani2004, author = {. and .}, title = {A pressure based Eulerian--Eulerian multi--phase model for non--equilibrium condensation on transonic steam flow}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {2217--2231}, } @ARTICLE{Singh2004, author = {.}, title = {A numerical solution of composite heat transfer problems using meshless method}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {2123--2138}, } @ARTICLE{SalahAskriSlimiNasrallah2004, author = {. and . and .}, title = {Numerical resolution of the radiative transfer equation in a cylindrical enclosure with the finite--volume method}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {2501--2509}, } @ARTICLE{HsiehLien2004, author = {. and .}, title = {Numerical Modeling of buoyancy--driven turbulent flows in enclosures}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {25}, pages = {659--670}, } @ARTICLE{LysakerLundervoldTai2003, author = {. and . and .}, title = {Noise Removal Using Fourth--Order Partial Differential Equation With Applications to Medical Magnetic Resonance Images in Space and Time}, journal = {IEEE Transactions on Image Processing}, year = {2003}, volume = {12}, number = {12}, pages = {1579--1590}, } @ARTICLE{SunZhang2004, author = {. and .}, title = {A Parallel Two--Level Hybrid Method for Tridiagonal Systems and Its Application to Fast Poisson Solvers}, journal = {IEEE Transactions on Parallel and Distributed Systems}, year = {2004}, volume = {15}, number = {2}, pages = {97--106}, } @ARTICLE{CiarletMardare2004, author = {. and .}, title = {On the recovery of a manifold with boundary in \$R\^{} n\$}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {333--340}, } @ARTICLE{MolkiMahendraVengala2004, author = {. and . and .}, title = {Visualization and Modeling of Flow Boiling of R--134A in Minichannels with Transverse Ribs}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {3}, pages = {94--103}, } @ARTICLE{LiLiShaoNieEvans2004, author = {. and . and .}, title = {Optimum parameter for the Sor--like method for augmented systems}, journal = {J. Comput. Math.}, year = {2004}, volume = {81}, number = {6}, pages = {749--763}, } @ARTICLE{Nishio2004, author = {.}, title = {Single--Phase Laminar--Flow Heat Transfer and Two--Phase Oscillating--Flow Heat Transport in Microchannels}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {3}, pages = {31--43}, } @ARTICLE{ColinLalondeCaen2004, author = {. and .}, title = {Validation of a Second--Order Slip Flow Model in Rectangular Microchannels}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {3}, pages = {23--30}, } @ARTICLE{Celata2004, author = {.}, title = {Single--Phase Heat Transfer and Fluid Flow in Micropipes}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {3}, pages = {13--22}, } @ARTICLE{Kandiklar2004, author = {.}, title = {Selected Papers from the First International Conference on Microchannels and Minichannels}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {3}, pages = {1--2}, } @ARTICLE{GunjalRanadeChaudhari2005, author = {. and . and .}, title = {Dynamics of Drop Impact on Solid Surface: Experiments and VOF Simulations}, journal = {AIChE Journal}, year = {2005}, volume = {51}, number = {1}, pages = {59--78}, } @ARTICLE{EhrhardMeisel2002, author = {. and .}, title = {Str\"{o}mungs--{} und Transportprozesse in Mikrokan\"{a}len}, journal = {Nachrichten - Forschungszentrum Karlsruhe}, year = {2002}, volume = {34}, number = {2-3}, pages = {137--142}, } @ARTICLE{DingShuYeoXu2004, author = { .}, title = {Development of least--square--based two--dimensional finite--difference schemes and their application to simulate natural convection in a cavity}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {137--154}, } @ARTICLE{ChenToledo2003, author = {. and .}, title = {Vaidya`s preconditioners: implementation and experimental study}, journal = {Electronic Transactions on Numerical Analysis }, year = {2003}, volume = {16}, pages = {30--49}, } @ARTICLE{GuennouniJbilouSadok2003, author = {. and .}, title = {A block version of BICGSTAB for linear systems with multiple right--hand sides}, journal = {Electronic Transactions on Numerical Analysis }, year = {2003}, volume = {16}, pages = {129--142}, } @ARTICLE{FengMichaelides2005, author = {. and .}, title = {Proteus: a direct forcing method in the simulations of particulate flows}, journal = {J. Comput. Phys.}, year = {2005}, volume = {202}, pages = {20--51}, } @ARTICLE{Carstensen2003, author = {.}, title = {All first--order averaging techniques for a posteriori finite element error control on unstructured grids {} are efficient and reliable}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1153--1165}, } @ARTICLE{Demlow2003, author = {.}, title = {Piecewise linear finite element methods are not localized}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1195--1201}, } @ARTICLE{DahmenFaermannGrahamHackbuschSauter2003, author = {. and . and . and . and .}, title = {Inverse inequalities on non--quasi--uniform meshes and application to the mortar element method}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1107--1138}, } @ARTICLE{Brenner2003a, author = {.}, title = {Korn`s inequalities for piecewise \$H\^{} 1\$ vector fields}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1067--1087}, } @ARTICLE{MargolinTitiWynne2003, author = {. and . and .}, title = {The postprocessing {Galerkin} and nonlinear {Galerkin} methods {}--{}--{} A truncation analysis point of view}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {695--714}, } @ARTICLE{HeMiaoMattheijChen2004, author = { .}, title = {Numerical analysis of a modified finite element nonlinear {Galerkin} method}, journal = {Numerische Mathematik}, year = {2004}, volume = {97}, pages = {725--756}, } @ARTICLE{GiraultNochettoScott2004, author = {. and . and .}, title = {Stability of the finite element {Stokes} projection in \$W\^{} {1,/infty}\$}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {957--962}, } @ARTICLE{MiklosovicMurrayHowleFish2004, author = {. and .}, title = {Leading--edge tubercles delay stall on humpback whale (Megaptera novaeangliae) flippers}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {5}, pages = {L39-L42}, } @ARTICLE{Pena2004, author = {.}, title = {A stable test to check if a matrix is a nonsingular M--Matrix}, journal = {Math. Comput.}, year = {2004}, volume = {73}, number = {247}, pages = {1385--1392}, } @ARTICLE{BuzdinWittum2004, author = {. and .}, title = {Two--frequency decomposition}, journal = {Numerische Mathematik}, year = {2004}, volume = {97}, pages = {269--295}, } @ARTICLE{ClarenzDiewaldRumpf2004, author = {. and . and .}, title = {Processing Textured Surfaces via Anisotropic Geometric Diffusion}, journal = {IEEE Transactions on Image Processing}, year = {2004}, volume = {13}, number = {2}, pages = {248--261}, } @ARTICLE{KimParter2003, author = {. and .}, title = {Semicirculant preconditioning of elliptic operators}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {767--795}, } @ARTICLE{TsaiChengOsherZhao2003, author = {. and . and .}, title = {Fast sweeping algorithms for a class of Hamilton--{Jacobi} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {673--694}, } @ARTICLE{TangTang2003, author = {. and .}, title = {Adaptive mesh methods for one--{} and two--dimensional hyperbolic conservation laws}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {487--515}, } @ARTICLE{ErvinMiles2003, author = {. and .}, title = {Approximation of time--dependent viscoelastic fluid flow: Supg approximation}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {2}, pages = {457--486}, } @ARTICLE{SchaererKaszkurewiczMangiavacchi2004, author = {. and . and .}, title = {A multilevel Schwarz shooting method for the solution of the Poisson equation in two dimensional incompressible flow simulations}, journal = {Applied Mathematics and Computation}, year = {2004}, volume = {153}, pages = {803--831}, } @ARTICLE{BulutErguetAsilBokor2004, author = {. and . and . and .}, title = {Numerical solution of a viscous incompressible flow problem through an orifice by Adomian decompression method}, journal = {Applied Mathematics and Computation}, year = {2004}, volume = {153}, pages = {733--741}, } @ARTICLE{ElbashbeshyBazid2004, author = {. and .}, title = {The effect of temperature--dependent viscosity on heat transfer over continuous moving surface with variable internal heat generation}, journal = {Applied Mathematics and Computation}, year = {2004}, volume = {153}, pages = {721--731}, } @ARTICLE{Wu2004, author = {.}, title = {An iterative block Arnoldi algorithm with modified approximate eigenvectors for large unsymmetric eigenvalue problems}, journal = {Applied Mathematics and Computation}, year = {2004}, volume = {153}, pages = {611--643}, } @ARTICLE{LeeHaBalachandarLee2004, author = {. and . and . and .}, title = {Numerical simulations of flow and heat transfer past a circular cylinder with a periodic array of fins}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {5}, pages = {1273--1286}, } @ARTICLE{SahinOwens2004, author = {. and .}, title = {A numerical investigation of wall effects up to high blockage ratios on two--dimensional flow past a confined circular cylinder}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {5}, pages = {1305--1320}, } @ARTICLE{GramlichMazouchiHomsy2004, author = {. and . and .}, title = {Time--dependent free surface {Stokes} flow with a moving contact line. II. Flow over wedges and trenches}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {5}, pages = {1660--1667}, } @ARTICLE{MazouchiGramlichHomsy2004, author = {. and . and .}, title = {Time--dependent free surface {Stokes} flow with a moving contact line. I. Flow over plane surfaces}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {5}, pages = {1647--1659}, } @ARTICLE{DziubinskiFidosSosno2004, author = {. and . and .}, title = {The flow pattern map of a two--phase non--Newtonian liquid--gas flow in the vertical pipe}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {551--563}, } @ARTICLE{KangAhnOhKim2004, author = {. and . and . and .}, title = {Advanced electromagnetic flowmetry for slug flow: numerical signal prediction and calibration}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {585--614}, } @ARTICLE{ZhangXu2004, author = {. and .}, title = {Finite Element Analysis of Pulsed Laser Bending: The Effect of Melting and Solidification}, journal = {Journal of Applied Mechanics}, year = {2004}, volume = {71}, pages = {321--326}, } @ARTICLE{SuikerFleck2004, author = {. and .}, title = {Frictional Collapse of Granular Assemblies}, journal = {Journal of Applied Mechanics}, year = {2004}, volume = {71}, pages = {350--358}, } @ARTICLE{BarthBochevGunzburgerShadid2004, author = {. and . and .}, title = {A Taxonomy of consistently stabilized finite element methods for the stokes problem}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {5}, pages = {1585--1607}, } @ARTICLE{SchmittAgarPlatteBuijssenPawlowskiDuisberg2005, author = { .}, title = {Ceramic Plate Heat Exchanger for Heterogeneous Gas Phase Reactions}, journal = {Chemical Engineering and Technology}, year = {2005}, volume = {28}, number = {3}, pages = {337--343}, } @ARTICLE{Toh2003, author = {.}, title = {Solving large scale semidefinite programs via an iterative solver on the augmented systems}, journal = {SIAM Journal on Optimization}, year = {2003}, volume = {14}, number = {3}, pages = {670--698}, } @ARTICLE{Meister2003, author = {.}, title = {Viscous fluid flow at all speeds: analysis and numerical simulation}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Physik}, year = {2003}, volume = {54}, pages = {1010--1049}, } @ARTICLE{Radaev2003, author = {.}, title = {The theory of 3D equations of mathematical plasticity}, journal = {Mechanics of Solids}, year = {2003}, volume = {5}, pages = {102--120}, } @ARTICLE{AntoniouPronko2004, author = {. and .}, title = {The hamiltinian formalism in fluid dynamics}, journal = {Theoretical and Mathematical Physics}, year = {2004}, volume = {141}, number = {3}, pages = {1670--1685}, } @ARTICLE{KumarQuateroniFormaggiaLamponi2003, author = {. and . and . and .}, title = {On Parallel Computation of Blood Flow in Human Arterial Network Based on 1--D Modelling}, journal = {Computing}, year = {2003}, volume = {71}, pages = {321--351}, } @ARTICLE{Chen2003, author = {.}, title = {Derivative Estimation with Finite Differences}, journal = {Simulation}, year = {2003}, volume = {79}, number = {10}, pages = {598--609}, } @ARTICLE{ShenZhaoTanT.2004, author = { .}, title = {Decay of vortex velocity and diffusion of temperature in a generalized second grade fluid}, journal = {Journal of Applied Mathematics and Mechanics}, year = {2004}, volume = {25}, number = {10}, pages = {1151--1159}, } @ARTICLE{WooGaudiotWendelborn2004, author = {. and . and .}, title = {Alias Analysis in Java with Reference--Set Representation for Hogh--Performance Computing}, journal = {International Journal of Parallel Programming}, year = {2004}, volume = {32}, number = {1}, pages = {39--76}, } @ARTICLE{Al-SaifZhu2004, author = {Al-Saif, . and .}, title = {Upwind local differential quadrature method for solving coupled viscous flow and heat transfer equations}, journal = {Journal of Applied Mathematics and Mechanics}, year = {2004}, volume = {25}, number = {10}, pages = {1130--1138}, } @ARTICLE{EamesHunt2004, author = {. and .}, title = {Forces on bodies moving unsteadily in rapidly compressed flows}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {505}, pages = {349--364}, } @ARTICLE{RighettiRomano2004, author = {. and .}, title = {Particle--fluid interactions in a plane near--wall turbulent flow}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {505}, pages = {93--121}, } @ARTICLE{Graham2004, author = {.}, title = {The Influence of Small Particles on the Structure of a Turbulent Shear Flow}, journal = {Journal of Fluids Engineering}, year = {2004}, volume = {126}, pages = {613--619}, } @ARTICLE{GaoMashayek2004, author = {. and .}, title = {A Stochastic Model for Gravity Effects in Particle--Laden Turbulent Flows}, journal = {Journal of Fluids Engineering}, year = {2004}, volume = {126}, pages = {620--625}, } @ARTICLE{RoosLinss2001a, author = {. and .}, title = {Gradient Recovery for Singularity Perturbed Boundary Value Problems: One--Dimensional Convection--Diffusion}, journal = {Computing}, year = {2001}, volume = {66}, pages = {163--178}, } @ARTICLE{Guskov2004, author = {.}, title = {An anisotropic mesh parameterization scheme}, journal = {Engineering with Computers}, year = {2004}, volume = {20}, pages = {129--135}, } @ARTICLE{BasaraAlajbegovicBeader2004, author = {. and . and .}, title = {Simulation of single--{} and two--phase flows on sliding unstructured meshes using finite volume method}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1137--1159}, } @ARTICLE{TrindadePereira2004, author = {. and .}, title = {Parallel--in--time simulation of the unsteady {Navier}--{Stokes} equations for incompressible flow}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1123--1136}, } @ARTICLE{HayatKhanSiddiquiAsghar2004, author = {. and . and . and .}, title = {Transient flows of a second grade fluid}, journal = {International Journal of Non-Linear Mechanics}, year = {2004}, volume = {39}, pages = {1621--1633}, } @ARTICLE{KepnerAhalt2004, author = {. and .}, title = {Short communication MatlabMPI}, journal = {Journal of Parallel and Distributed Computing}, year = {2004}, volume = {64}, pages = {997--1005}, } @ARTICLE{AkyildizBelloutVajravelu2004, author = {. and . and .}, title = {Exact solutions of nonlinear differential equations arising in third grade fluid flows}, journal = {International Journal of Non-Linear Mechanics}, year = {2004}, volume = {39}, pages = {1571--1578}, } @ARTICLE{AmarAndreucciBisegnaGianni2004, author = { .}, title = {An elliptic equation with history}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {595--598}, } @ARTICLE{BurmanErn2004, author = {. and .}, title = {Discrete maximum principle for {Galerkin} approximations of the Laplace operator on arbitrary meshes}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {338}, pages = {641--646}, } @ARTICLE{WenYehWangLin2004, author = {. and . and . and .}, title = {On the drag of two--dimensional flow about a circular cylinder}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {10}, pages = {3828--3831}, } @ARTICLE{Wang2004, author = {.}, title = {A mixed finite volume element method based on rectangular mesh for biharmonic equations}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {172}, pages = {117--130}, } @ARTICLE{GaloAlbarrealCalzadaCruzFernandez-CaraMarin2004, author = {. and . and . and . and . and .}, title = {A simultaneous directions parallel algorithm for the {Navier}--{Stokes} equations}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {235--240}, } @ARTICLE{LiuWalkington2001, author = {. and .}, title = {An Eulerian Description of Fluids Containing Visco--Elastic Particles}, journal = {Archive for Rational Mechanics and Analysis}, year = {2001}, volume = {159}, pages = {229--252}, } @ARTICLE{AbboudSalaunSalmon2002, author = {. and . and .}, title = {Integral method for the {Stokes} problem}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {71--76}, } @ARTICLE{Iftimie2002, author = {.}, title = {Remarques sur la limite a-->0 pour les fluides de grade 2}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {83--86}, } @ARTICLE{Ayachour2003, author = {Ayachour}, title = {A fast implementation for GMRES method}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {269--283}, } @ARTICLE{BAI2003, author = {.}, title = {On the convergence of parallel nonstationary multisplitting iteration methods}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {1--11}, } @ARTICLE{HanYin2003, author = {. and .}, title = {A non--overlap domain decomposition method for the forward--backward heat equation}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {35--44}, } @ARTICLE{Hataue2003, author = {.}, title = {On analogy and dissimilarity of dependence of stability on several parameters in flow simulations}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {45--53}, } @ARTICLE{ItohNamekawa2003, author = {. and .}, title = {An improvement in DS--BiCGstab(l) and its application for linear systems in lattice QCD}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {65--75}, } @ARTICLE{Watanabe2003, author = {.}, title = {On incompressible viscous fluid flows with slip boundary conditions}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {161--172}, } @ARTICLE{Riley2004, author = {.}, title = {On the control of laminar flow separation}, journal = {Q. J. Mech. Appl. Math.}, year = {2004}, volume = {57}, number = {2}, pages = {237--244}, } @ARTICLE{CiarletGirault2002, author = {. and .}, title = {Condition inf--sup pour l`element fini de Taylor--Hood P2--iso--P1, 3--D; application aux equations de Maxwell}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {335}, pages = {827--832}, } @ARTICLE{Cottet2002, author = {.}, title = {A particle model for fluid--structure interaction}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {335}, pages = {833--838}, } @ARTICLE{BardosPironneau2002, author = {. and .}, title = {A formalism for the differentiation of conservation laws}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {335}, pages = {839--845}, } @ARTICLE{Dubois2002a, author = {.}, title = {What is a solution to the {Navier}--{Stokes} equations?}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {335}, pages = {27--32}, } @ARTICLE{CoquelGallouetHerardSeguin2002, author = {. and . and . and .}, title = {Closure laws for a two--fluid two--pressure model}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {927--932}, } @ARTICLE{MadayRapettiWohlmuth2002, author = {. and . and .}, title = {Coupling between scalar and vector potentials by the mortar element method}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {933--938}, } @ARTICLE{AcerbiMingione2002, author = {. and .}, title = {Regularity results for electrorheological fluids: the stationary case}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {817--822}, } @ARTICLE{Cheskidov2002, author = {.}, title = {Turbulent boundary layer equations}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {334}, pages = {423--427}, } @ARTICLE{PantaleCaperaaRakotomalala2004, author = {. and . and .}, title = {Development of an object--oriented finite element program: application to metal--forming and impact simulations}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {341--351}, } @ARTICLE{RuasCarneirodeAraujoRamos2004, author = {. and . and .}, title = {Multi--field finite element methods with discontinuous pressures for axisymmetric incompressible flow}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {393--402}, } @ARTICLE{Brenner2003b, author = {.}, title = {Korn`s inequalities for piecewise \$H\^{} 1\$ vector fields}, journal = {Math. Comput.}, year = {2003}, volume = {73}, number = {247}, pages = {1067--1087}, } @ARTICLE{YabeTakano2004, author = {. and .}, title = {Global Convergence Properties of Nonlinear Conjugate Gradient Methods with Modified Secant Condition}, journal = {Computational Optimization and Applications}, year = {2004}, volume = {28}, pages = {203--225}, } @ARTICLE{YabeTakano2004a, author = {. and .}, title = {Global Convergence Properties of Nonlinear Conjugate Gradient Methods with Modified Secant Condition}, journal = {Computational Optimization and Applications}, year = {2004}, volume = {28}, pages = {203--225}, } @ARTICLE{KuuselaLahtinenAla-Nissila2004, author = {. and . and .}, title = {Sedimentation dynamics of spherical particles in confined geometries}, journal = {Physical Review}, year = {2004}, volume = {69}, number = {066310}, pages = {1--9}, } @ARTICLE{BraunshteinKalmanUllmann2004, author = { .}, title = {Performance and Optimization of Composed Fin Arrays}, journal = {Heat Transfer Engineering}, year = {2004}, volume = {25}, number = {4}, pages = {4--12}, } @ARTICLE{Roisman2004, author = {.}, title = {Dynamics of inertia dominated binary drop collisions}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {9}, pages = {3438--3449}, } @ARTICLE{DastgeerZank2004, author = {. and .}, title = {Nonlinear flows in nearly incompressible hydrodynamic fluids}, journal = {Physical Review}, year = {2004}, volume = {69}, number = {066309}, pages = {1--5}, } @ARTICLE{BoseKumaran2004, author = {. and .}, title = {Velocity distribution for a two--dimensional sheared granular flow}, journal = {Physical Review}, year = {2004}, volume = {69}, number = {061301}, pages = {1--16}, } @ARTICLE{SantosGarzoDufty2004, author = {. and .}, title = {Inherent rheology of a granular fluid in uniform shear flow}, journal = {Physical Review}, year = {2004}, volume = {69}, number = {061303}, pages = {1--10}, } @ARTICLE{ElootBisschopVerdonck2004, author = {. and . and .}, title = {Experimental evaluation of the migration of spherical particles in three--dimensional Poisseuille flow}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {7}, pages = {2282--2293}, } @ARTICLE{HwangYang2004, author = {. and .}, title = {Numerical study of vortical structures around a wall--mounted cubic obstacle in channel flow}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {7}, pages = {2382--2394}, } @ARTICLE{KimKangChoi2004, author = {. and . and Choi}, title = {Analysis of convective instability and heat transfer characteristics of nanofluids}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {7}, pages = {2395--2401}, } @ARTICLE{Riley2004a, author = {.}, title = {On the control of laminar flow separation}, journal = {Q. J. Mech. Appl. Math.}, year = {2004}, volume = {57}, number = {2}, pages = {237--244}, } @ARTICLE{DamClerc2004, author = {. and .}, title = {Experimental study of the impact of an ink--jet printed droplet on a solid substrate}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {9}, pages = {3403--3414}, } @ARTICLE{GarckeNestlerStinner2004, author = {. and . and .}, title = {A diffuse interface model for alloys with multiple components and phases}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {64}, number = {3}, pages = {775--799}, } @ARTICLE{KeenanShorter2004, author = {. and .}, title = {Simulation by diffusion on a manifold with boundary: applications to ultrasonic prenatal medical imaging}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {64}, number = {3}, pages = {932--960}, } @ARTICLE{BerlyandKhruslov2004, author = {. and .}, title = {Homogenized non--newtonian viscoelastic rheology of a suspension of interacting particles in a viscous newtonian fluid}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {64}, number = {3}, pages = {1002--1034}, } @ARTICLE{Guermond2004, author = {.}, title = {A finite elemtn technique for solving first--order PDEs in \$L\^{} P\$}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {42}, number = {2}, pages = {714--737}, } @ARTICLE{Paschedag2003, author = {.}, title = {Kopplung zwischen Populationsbilanzen und {CFD} am Beispiel von F\"{a}llungsprozessen}, journal = {Chemie Ingenieur Technik}, year = {2003}, volume = {75}, number = {12}, pages = {1835--1839}, } @ARTICLE{Paschedag2004, author = {.}, title = {Modeling of Mixing and Precipitation Using {CFD} and Population Balances}, journal = {Chemical Engineering and Technology}, year = {2004}, volume = {27}, number = {3}, pages = {232--236}, } @ARTICLE{MouhotPareschi2004, author = {. and .}, title = {Fast methods for the Boltzmann collision integral}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {71--76}, } @ARTICLE{Boor2003, author = {.}, title = {A Leibniz formula for multivariate divided differences}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {856--868}, } @ARTICLE{BergerHelzelLeveque2003, author = {. and . and .}, title = {h--Box methods for the approximation of hyperbolic conservation laws on irregular grids}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {893--918}, } @ARTICLE{Deng2003, author = {.}, title = {An optimal parallel nonoverlapping domain decomposition iterative procedure}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {964--982}, } @ARTICLE{Stevenson2003a, author = {.}, title = {Adaptive solution of operator equations using wavelet frames}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {1074--1100}, } @ARTICLE{SamelsonTemamWangWang2003, author = {. and . and . and .}, title = {Surface pressure poisson equation formulation of the primitive equations: numerical schemes}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {1163--1194}, } @ARTICLE{AringazinMazhitov2004, author = {. and .}, title = {Stochastic Models of {L}agrangian Acceleration of Fluid Particle in Developed Turbulence}, journal = {Int. J. Mod. Phys.}, year = {2004}, volume = {18}, number = {23 \& 24}, pages = {3095--3168}, } @ARTICLE{LiuAshworthEmerson2004, author = {. and .}, title = {Pressure correction methods based on Krylov subspace conception for solving incompressible {Navier}--{Stokes} problems}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1249--1268}, } @ARTICLE{Ekaterinatis2004, author = {.}, title = {High--order accurate numerical solutions of incompressible flows with the artificial compressibility method}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1187--1207}, } @ARTICLE{GaskellJimackSellierThompson2004, author = {. and . and .}, title = {Efficient and accurate time adaptive multigrid simulations of droplet spreading}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {45}, pages = {1161--1186}, } @ARTICLE{ArgentiniBakkerLowe2004, author = {. and . and .}, title = {Efficiently using memory in lattice Boltzmann simulations}, journal = {Future Generation Computer Systems}, year = {2004}, volume = {20}, pages = {973--980}, } @ARTICLE{Sman2004, author = {.}, title = {Diffusion on unstructured triangular grids using Lattice Boltzmann}, journal = {Future Generation Computer Systems}, year = {2004}, volume = {20}, pages = {965--971}, } @ARTICLE{ChopardHoekstra2004, author = {. and .}, title = {Computational Science of lattice Boltzmann modelling}, journal = {Future Generation Computer Systems}, year = {2004}, volume = {20}, pages = {907--908}, } @ARTICLE{PopRaduKnabner2004, author = {. and . and .}, title = {Mixed finite elements for the Richard`s equation: linearization procedure}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {356--373}, } @ARTICLE{Anonymous2004, author = {Anonymous}, title = {Index of the Journal of Computational and Applied Mathematics Vol. 168, Nos. 1--2, 2004}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, number = {1-2}, pages = {VII-X}, } @ARTICLE{Laschet2004, author = {.}, title = {Homogenization of the fluid flow and heat transfer in transpiration cooled multi--layer plates}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {277--288}, } @ARTICLE{Hoppe2004, author = {.}, title = {Adaptive multigrid and domain decomposition methods in the computation of electromagnetic fields}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {245--254}, } @ARTICLE{FortinCoorevits2004, author = {. and .}, title = {Selecting contact particles in dynamics granular mechanics systems}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {207--213}, } @ARTICLE{KashidGerlachFranzkeAckerPlatteAgarTurek2005, author = {. and .}, title = {Internal Circulation within the Liquid Slugs of a Liquid--Liquid Slug--Flow Capillary Microreactor}, journal = {Indian Journal of Engineering and Material Sciences}, year = {2005}, volume = {44}, number = {14}, pages = {5003--5010}, note = {Submitted for publication in the special issue: In honor of Milorad P. Dudukovic}, } @ARTICLE{GunjalKashidRanadeChaudhari2005, author = {. and . and . and .}, title = {Hydrodynamics of Trickle--Bed Reactors: Experiments and {CFD} Modeling}, journal = {Indian Journal of Engineering and Material Sciences}, year = {2005}, note = {Submitted for publication in the special issue: In honor of Milorad P. Dudukovic, to be published August 3, 2005}, } @ARTICLE{HronMalekNecasRajagopal2003, author = {. and .}, title = {Numerical simulations and global existence of solutions of two--dimensional flows of fluids with pressure--and shear--dependent viscosities}, journal = {Mathematics and Computers in Simulations}, year = {2003}, number = {61}, pages = {297--315}, } @ARTICLE{MalekNecasRajagopal2002, author = {. and . and .}, title = {Global Existence of Solutions for Flows of Fluids with Pressure and Shear Dependent Viscosities}, journal = {Applied Mathematics Letters}, year = {2002}, volume = {15}, pages = {961--967}, } @ARTICLE{DelgadoKaperskiLabrosse2004, author = {. and . and .}, title = {Spectrally accurate numerical incompressible flows with deformed boundaries: a finite volume preconditioning method}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {145--154}, } @ARTICLE{Banas2004, author = {.}, title = {Solution of convection--diffusion equation by the method of characteristics}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {31--39}, } @ARTICLE{VierendeelsMerciDick2004, author = {. and . and .}, title = {A multigrid method for natural convective heat transfer with large temperature differences}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {509--517}, } @ARTICLE{SteinebachRademacherRentropSchulz2004, author = {. and .}, title = {Mechanisms of coupling in river flow simulation systems}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {459--470}, } @ARTICLE{SeaidKlarDubroca2004, author = { .}, title = {Flux limiters in the coupling of radiation and hydrodynamic models}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {425--435}, } @ARTICLE{ZegelingKok2004, author = {. and .}, title = {Adaptive moving mesh computations for reaction--diffusion systems}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {519--528}, } @ARTICLE{MichelettiPerottoPicasso2003, author = {. and .}, title = {Stabilized finite elements on anisotropic meshes: a priori error estimates for the advection--diffusion and the {Stokes} problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {3}, pages = {1131--1162}, } @ARTICLE{EdussuriyaWilliamsBailey2004, author = { .}, title = {A cell--centred finite element method for modelling viscoelastic flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {117}, pages = {47--61}, } @ARTICLE{YeShyyTaiChung2004, author = {. and . and . and .}, title = {Assessment of sharp--{} and continuous--interface methods for drop in static equilibrium}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {917--926}, } @ARTICLE{YeShyyTaiChung2004a, author = {. and .}, title = {Assessment of sharp--{} and continuous--interface methods for drop in static equilibrium}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {917--926}, } @ARTICLE{GuentherKlarMaterneWegener2003, author = {. and . and .}, title = {Multivalued fundamental diagrams and stop and go waves for continuum traffic flow equations}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {64}, number = {2}, pages = {468--483}, } @ARTICLE{StijnenHartBovendeerdVosse2004, author = { .}, title = {Evaluation of a fictitious domain method for predicting dynamic response of mechanical heart valves}, journal = {Journal of Fluids and Structures}, year = {2004}, volume = {19}, pages = {835--850}, } @ARTICLE{CaseChakaFriendFruripGolabJohnsonMooreMountainOlsonSchillerStorer2004, author = { .}, title = {The first industrial fluid properties simulation challenge}, journal = {Fluid Phase Equilibria}, year = {2004}, volume = {217}, pages = {1--10}, } @ARTICLE{FriendFruripMageeOlson2004, author = {. and . and . and .}, title = {Establishing benchmarks for the first industrial fluids simulation challenge}, journal = {Fluid Phase Equilibria}, year = {2004}, volume = {217}, pages = {11--15}, } @ARTICLE{ZhangWang2004, author = {. and .}, title = {A block LU--SGS Implicit Dual Time--stepping Algorithm for Hybrid Dynamic Meshes}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {891--916}, } @ARTICLE{LarsenThoemmesKlar2003, author = {. and .}, title = {New frequency--averaged approximations to the equations of radiative heat transfer}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {64}, number = {2}, pages = {565--582}, } @ARTICLE{GremaudMatthewsSchaefferD.G.2003, author = {. and . and .}, title = {Secondary circulation in granular flow through nonaxisymmetric hoppers}, journal = {SIAM Journal on Applied Mathematics}, year = {2003}, volume = {64}, number = {2}, pages = {583--600}, } @ARTICLE{SilvesterMihaljovic2004, author = {. and .}, title = {A black--box multigrid preconditioner for the biharmonic equation}, journal = {BIT Numerical Mathematics}, year = {2004}, volume = {44}, pages = {151--163}, } @ARTICLE{LiZhang2004, author = {. and .}, title = {Gas Kinetic Algorithm Using Boltzmann Model Equation}, journal = {Computers and Fluids}, year = {2004}, volume = {33}, pages = {967--991}, } @ARTICLE{Hedou-Rouillier2004, author = {.}, title = {A finite difference method to solve the forward problem in electroencephalography (EEG)}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {167}, pages = {35--58}, } @ARTICLE{LiChan2004, author = {. and .}, title = {High--order schemes for Hamilton--{Jacobi} equations on triangular meshes}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {167}, pages = {227--241}, } @ARTICLE{KutluayEsenDag2004, author = { .}, title = {Numerical solutions of the Burgers` equation by the keast--squares quadratic B--spline finite element method}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {267}, pages = {21--33}, } @ARTICLE{Chang2004a, author = {.}, title = {Letter to the Editor: A not on the upper bound of the spectral radius for SOR iteration matrix}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {167}, pages = {251--253}, } @ARTICLE{WonKim2004, author = {. and .}, title = {Alignment and aggregation of sperical particles in viscoelastic fluid under shear flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {117}, pages = {141--146}, } @ARTICLE{ClemeurRutgersDebbaut.B.2004, author = { .}, title = {Numerical simulation of abrupt contraction flows using the Double Convected Pom--Pom model}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {117}, pages = {193--209}, } @ARTICLE{Khomami2004, author = {.}, title = {Review of Computational Rheology by and }, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {117}, pages = {71}, } @ARTICLE{HainkeFriedrichMueller2004, author = {. and . and .}, title = {A matrix based correction scheme of the liquid fraction during columnar solidification}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {2883--2887}, } @ARTICLE{DednerRohde2004, author = {. and .}, title = {Numerical approximation of entropy solutions for hyperbolic integro--differential equations}, journal = {Numerische Mathematik}, year = {2004}, volume = {97}, pages = {441--471}, } @ARTICLE{WangLiuJohnston2004, author = { .}, title = {Analysis of a fourth order finite difference method for the incompressible Boussinesq equations}, journal = {Numerische Mathematik}, year = {2004}, volume = {97}, pages = {555--594}, } @ARTICLE{MoellerKuzminTurek2005, author = {. and . and .}, title = {Implicit finite element discretizations based on the flux--corrected transport algorithm}, journal = {International Journal for Numerical Methods in Fluids}, year = {2005}, volume = {Volume 47, Number 10 - 11}, pages = {1197--1203}, month = apr, } @ARTICLE{BurmanErn2002, author = {. and .}, title = {Nonlinear diffusion and discrete maximum principle for stabilized galerkin approximations of the convection--{}--diffusion--reaction equation}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2002}, volume = {191}, number = {35}, pages = {3833--3855}, } @ARTICLE{SemperLiao1995, author = {. and .}, title = {A Moving Grid Finite--Element Method Using Grid Deformation}, journal = {Numerical Methods for Partial Differential Equations}, year = {1995}, volume = {11}, pages = {603--615}, } @ARTICLE{MaischakStephan2004, author = {. and .}, title = {A Least Squares Coupling Method with Finite Elements and Boundary Elements for Transmission Problems}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {995--1016}, } @ARTICLE{FixRoop2004, author = {. and .}, title = {Least Squares Finite--Element Solution of a Fractional Order Two--Point Boundary Value Problem}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1017--1034}, } @ARTICLE{BochevGunzburger2004, author = {. and .}, title = {Least--Squares Finite--Element Methods for Optimization and Control Problems for the {Stokes}--Equations}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1035--1059}, } @ARTICLE{Chen2004, author = {.}, title = {Weighted Least--Squares Approximations to Nonlinear Hyperbolic Equations}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1059--1076}, } @ARTICLE{CaiFleitasJiangLiao2004, author = {. and .}, title = {Adaptive Grid Generation Based on Least--Squares Finite--Element Method}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1077--1086}, } @ARTICLE{Chen2004a, author = {.}, title = {Characteristic--Nonconforming Finite--Element Methods for Advection--Dominated Diffusion Problems}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1087--1100}, } @ARTICLE{Peterson2004, author = {.}, title = {Using the Cayley Transform and Fiite Elements to Solve Linear Time--Dependent Problems}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1101--1110}, } @ARTICLE{Nicolaides2004, author = {.}, title = {A Method for Complex Geometries in Finite--Difference Solutions of Maxwell`s Equations}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1111--1120}, } @ARTICLE{GrinfeldStrang2004, author = {. and .}, title = {The Laplacian Eigenvalues of a Polygon}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1121--1134}, } @ARTICLE{JohnLaytonSahin2004, author = {. and . and .}, title = {Derivation and Analysis of Near Wall Models for Channel and Recirculating Flows}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1135--1152}, } @ARTICLE{Liakos2004, author = {.}, title = {Discretization of the {Navier}--{Stokes} Equations with Slip Boundary Condition II}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1153--1166}, } @ARTICLE{vanJoolenNetaGivoli2004, author = {. and .}, title = {A Stratified Dispersive Wave Model with High--Order Nonreflecting Boundary Conditions}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1167--1180}, } @ARTICLE{CharinaMairSchmidt2004, author = {. and .}, title = {Mixed Velocity, Stress, Current, and Potential Boundary Conditions for Stationary MHD Flow}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1181--1190}, } @ARTICLE{WatersFixCox2004, author = {. and . and .}, title = {The Method of Glowinski and Pironneau for the Unsteady {Stokes} Problem}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1191--1212}, } @ARTICLE{Lee2004, author = {.}, title = {Analysis of a Defect Correction Metzhod for Viscoelastic Fluid Flow}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1213--1230}, } @ARTICLE{TurnerJr.2004, author = {., .}, title = {Modelling of a Crystal Growth Process}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {48}, number = {7-8}, pages = {1231--1243}, } @ARTICLE{LiaoAnderson1992, author = {. and .}, title = {A New Approach to Grid Generation}, journal = {Applicable Analysis}, year = {1992}, volume = {44}, pages = {285--298}, } @ARTICLE{CaoHuangRussell2003, author = {. and . and .}, title = {Approaches for generating moving adaptive meshes: location versus velocity}, journal = {Applied Numerical Mathematics}, year = {2003}, volume = {47}, pages = {121--138}, note = {DOI: 10.1016/S0168-9276(03)00061-8}, } @ARTICLE{Berggren2004, author = {.}, title = {Approximations of very weak solutions to boundary--value problems}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {42}, number = {2}, pages = {860--877}, } @ARTICLE{DekelLeviatan2004, author = {. and .}, title = {The Bramble--Hilbert lemma for convex domains}, journal = {SIAM Journal on Mathematical Analysis}, year = {2004}, volume = {35}, number = {5}, pages = {1203--1212}, } @ARTICLE{DraganescuDupntScott2004, author = {. and . and .}, title = {Failure of the discrete maximum principle for an elliptic finite element problem}, journal = {Math. Comput.}, year = {2004}, volume = {74}, number = {249}, pages = {1--23}, } @ARTICLE{WeiZhuHe2004, author = {. and . and .}, title = {Large eddy simulation and spectrum analysis of flow around two square cylinders arranged side by side}, journal = {Journal of Applied Mathematics and Mechanics}, year = {2004}, volume = {25}, number = {8}, pages = {903--910}, } @ARTICLE{Meng2004, author = {.}, title = {Analytic solutions of incompressible laminar flow in the inlet region of a pipe}, journal = {Journal of Applied Mathematics and Mechanics}, year = {2004}, volume = {25}, number = {8}, pages = {894--897}, } @ARTICLE{RosaMorales2004, author = {. and .}, title = {Experimental and Numerical Development of a Two--Phase Venturi Flow Meter}, journal = {Journal of Fluids Engineering}, year = {2004}, volume = {126}, pages = {457--467}, } @ARTICLE{BiswasBreuerDurst2004, author = {. and . and .}, title = {Backward--Facing Step Flows for Various Expansion Ratios at Low and Moderate Reynolds Numbers}, journal = {Transactions ASME}, year = {2004}, volume = {126}, pages = {362--373}, } @ARTICLE{MargolinShashkov2004, author = {. and .}, title = {Remapping, recovery and repair on a staggered grid}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4139--4155}, } @ARTICLE{FarhatGeuzaine2004, author = {. and .}, title = {Design and analysis of robust ALE time--integrators for the solution of unsteady flow problems on moving grids}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4073--4095}, } @ARTICLE{FormaggiaNobile2004, author = {. and .}, title = {Stability analysis of second--order time accurate schemes for ALE--{FEM}}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4097--4116}, } @ARTICLE{FouresteyPiperno2004, author = { .}, title = {A second--order time--accurate ALE {Lagrange}--{Galerkin} method applied to wind engineering and control of bridge profiles}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4117--4137}, } @ARTICLE{BelletFachinotti2004, author = {. and .}, title = {ALE method for solidification modelling}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4355--4381}, } @ARTICLE{BagabirDrikakis2004, author = {. and .}, title = {Numerical experiments using high--resolution schemes for unsteady, inviscid, compressible flow}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4675--4705}, } @ARTICLE{LagreeBergerDevergeVilainHirschberg2004, author = {. and .}, title = {Characterization of the pressure drop in a 2D symmetrical pipe: Some asymptotical, numerical, and experimental comparisons}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Mechanik}, year = {2004}, volume = {85}, number = {2}, pages = {141--146}, } @ARTICLE{KhooCaiXiongLimWahlqvistFok2004, author = { .}, title = {Computation tools for the combat of cardiovascular heart disease}, journal = {International Journal of Computer Applications in Technology}, year = {2004}, volume = {21}, number = {1-2}, pages = {16--22}, } @ARTICLE{ReesMastersMalanLewis2004, author = { .}, title = {An edge--based finite volume scheme for saturated--unsaturated groundwater flow}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4741--4759}, } @ARTICLE{ChenWang2003, author = {. and .}, title = {An interior estimate of superconvergence for finite element solutions for second--order elliptic problems on quasi--uniform meshes by local projections}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1318--1338}, } @ARTICLE{KasagiMatsunaga1995, author = {. and .}, title = {Three--dimensional particle--tracking velocimetry measurement of turbulence statistics and energy budget in a backward--facing step flow}, journal = {Int. J. Numer. Methods Heat Fluid Flow}, year = {1995}, volume = {16}, pages = {477--485}, } @ARTICLE{KweonKellogg2004, author = {. and .}, title = {Regularity of solutions to the {Navier}--{Stokes} system for compressible flows on a polygon}, journal = {SIAM Journal on Mathematical Analysis}, year = {2004}, volume = {35}, number = {6}, pages = {1451--1485}, } @ARTICLE{Vohralik2004, author = {.}, title = {Equivalence between mixed finite element and multi--point finite element methods}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {525--528}, } @ARTICLE{WihlerFrauenfelderSchwab2003, author = {. and .}, title = {Exponential Convergence of the hp--DGFEM for Diffusion Problems}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {183--205}, } @ARTICLE{Seshaiyer2003, author = {.}, title = {Stability and Convergence of Nonconforming hp Finite--Element Methods}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {165--182}, } @ARTICLE{RiviereWheeler2003, author = {. and .}, title = {A Posteriori Error Estimates for a Discontinuous {Galerkin} Method Applied to Elliptic Problems.}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {141--164}, note = {Log Number: R74}, } @ARTICLE{CaoKelly2003, author = {. and .}, title = {Pointwise and Local Error Estimates for the Quantities of Interest in Two--Dimensional Elasticity}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {69--79}, } @ARTICLE{NopensBeheydtVanrolleghem2005, author = {. and .}, title = {Comparison and pitfalls of different discretised solution methods for population balance models: a simulation study}, journal = {Computers and Chemical Engineering}, year = {2005}, volume = {29}, pages = {367--377}, } @ARTICLE{Reusken2003, author = {.}, title = {Convergence analysis of the {Gauss}--Seidel preconditioner for discretized one dimensional Euler equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1388--1405}, } @ARTICLE{Meister2003a, author = {.}, title = {Viscous flow at all speeds: analysis and numerical simulation}, journal = {Zeitschrift f\"{u}r Angewandte Mathematik und Physik}, year = {2003}, volume = {54}, pages = {1010--1049}, } @ARTICLE{MuellerShores2004, author = {. and .}, title = {A New Sinc--{Galerkin} Method for Convection--Diffusion Equations with Mixed Boundary Conditions}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {47}, pages = {803--822}, } @ARTICLE{BoglaevDuoba2004, author = {. and .}, title = {Iterative Domain Decomposition Algorithms for a Convection--Diffusion Problem}, journal = {Computers and Mathematics with Applications}, year = {2004}, volume = {47}, pages = {501--518}, } @ARTICLE{WangTang2004, author = {. and .}, title = {Non--self--overlapping structured grid generation on an n--sided surface}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {961--982}, } @ARTICLE{LinChenOng2004, author = {. and . and .}, title = {Finite element methods based on a new formulation for the non--stationary incompressible {Navier}--{Stokes} equations}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {1169--1180}, } @ARTICLE{BriesenMarquardt2004, author = {. and .}, title = {Adaptive multigrid solution strategy for the dynamic simulation of petroleum mixture processes}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {29}, pages = {139--148}, } @ARTICLE{BelgacemChiltonSeshaiyer2003, author = {. and . and .}, title = {The hp--Mortar Finite--Element Method for the Mixed Elasticity and {Stokes} Problems}, journal = {Computers and Mathematics with Applications}, year = {2003}, volume = {46}, pages = {35--55}, } @ARTICLE{TerracolSagaut2003, author = {. and .}, title = {A mulilevel--based dynamic approach for subgrid--scale modeling in large--eddy simulation}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {12}, pages = {3671--3682}, } @ARTICLE{Fairag2003, author = {.}, title = {Numerical computations of viscous, incompressible flow problems using a two--level Finite--Element method}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {24}, number = {6}, pages = {1919--1929}, } @ARTICLE{WienersWohlmuth2003, author = {. and .}, title = {Duality estimates and multigrid analysis for saddle point problems arising from mortar discretizations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {24}, number = {6}, pages = {2163--2184}, } @ARTICLE{PitmanNichitaPatraBauerSheridanBursik2003, author = { . and .}, title = {Computing granular avalanches and landslides}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {12}, pages = {3638--3646}, } @ARTICLE{AndreevThessHaberstroh2003, author = {. and . and .}, title = {Visualization of magnetoconvection}, journal = {Physics of Fluids}, year = {2003}, volume = {15}, number = {12}, pages = {3886--3889}, } @ARTICLE{GriebelSchweitzer2002, author = {. and .}, title = {A particle--partition of unity method--part III: A multilevel solver}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {2}, pages = {377--409}, } @ARTICLE{FarhloulNicaisePaquet2004, author = {. and .}, title = {A posteriori error estimation for the dual mixed finite element method of the {Stokes} problem}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {513--518}, } @ARTICLE{WendlerSchintke2005, author = {. and .}, title = {Executing and observing {CFD} applications on the Grid}, journal = {Future Generation Computer Systems}, year = {2005}, volume = {21}, pages = {11--18}, } @ARTICLE{YangHayesJenkinsCant2005, author = {. and . and . and .}, title = {The Cambridge {CFD} grid for large--scale distributed {CFD} applications}, journal = {Future Generation Computer Systems}, year = {2005}, volume = {21}, pages = {45--51}, } @ARTICLE{HurlbertWitteBestKurwitz2004, author = { .}, title = {Scaling two--phase flows to Mars and Moon gravity conditions}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {351--368}, } @ARTICLE{KwakKirisKim2005, author = {. and . and .}, title = {Computational challenges of viscous incompressible flows}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {283--299}, } @ARTICLE{He2003, author = {.}, title = {Two--level method based on finite element and Crank--Nicolson extrapolation for the time--dependent {Navier}--{Stokes} equations}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1263--1285}, } @ARTICLE{DemkowiczmL.Babuska2003, author = { .}, title = {p Interpolation Error Estimates for Edge Finite Elements of Variable Order in Two Dimensions}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1195--1208}, } @ARTICLE{CareyBarthWoodsKirkAndersonChowBangerth2, author = {. and . and . and . and . and .}, title = {Modelling Error and Constitutive Relations in Simulation of Flow and Transport}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {1211--1236}, } @ARTICLE{MarmottantVillermaux2004, author = {. and .}, title = {On spray information}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {498}, pages = {73--111}, } @ARTICLE{BisiSpigaToscani2004, author = {. and . and .}, title = {Grad`s equations and hydrodynamics for weakly inelastic granular flows}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4235--4274}, } @ARTICLE{Wang2004a, author = {.}, title = {Interaction of two circular cylinders in inviscid fluid}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {44412--4425}, } @ARTICLE{SilvaPereira2004, author = {. and .}, title = {The effect of subgrid--scale models on the vortices computed from large--eddy simulations}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4506--4534}, } @ARTICLE{ArnarsonJenkins2004, author = {. \"{O}. and .}, title = {Binary mixtures {} of inelastic spheres: Simplified constitutive constitutive theory}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4543--4550}, } @ARTICLE{CubaudHo2004, author = {. and .}, title = {Transport of bubbles in square microchannels}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4575--4585}, } @ARTICLE{SalmanSoteriou2004, author = {. and .}, title = {Langrangian simulation of evaporation droplet sprays}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4601--4622}, } @ARTICLE{GroszmanRogers2004, author = {. and .}, title = {Turbulent scales of dilute particle--laden flows in microgravity}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4671--4684}, } @ARTICLE{KeatingPiomelliBalarasKaltenbach2004, author = {. and . and .}, title = {A priori and a posteriori test of inflow conditions for large--eddy simulations}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {12}, pages = {4696--4712}, } @ARTICLE{GalettiBruneauZanettiIollo2004, author = {. and . and . and .}, title = {Low--order modelling of laminar flow regimes past a confined square cylinder}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {503}, pages = {161--170}, } @ARTICLE{Hindmarsh2004, author = {.}, title = {Thermoviscous stability of ice--sheet flows}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {502}, pages = {17--40}, } @ARTICLE{Colombini2004, author = {.}, title = {Revisiting the linear theory of sand dune formation}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {502}, pages = {1--16}, } @ARTICLE{Sun2004, author = {.}, title = {Flat--plate aerodynamic at very low Reynolds number}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {502}, pages = {199--206}, } @ARTICLE{LopezMarquezHirsaMiraghaie2004, author = {. and . and . and .}, title = {Symmetry breaking in free--surface cylinder flows}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {502}, pages = {99--126}, } @ARTICLE{Catlin2004, author = {.}, title = {The force--moments on a deforming body introduced impulsively into an inviscid incompressible fluid with uniform vorticity}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {455--480}, } @ARTICLE{DymentBoudlal2004, author = {. and .}, title = {A theoretical model for gas--liquid slug flow in down inclined ducts}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {521--550}, } @ARTICLE{Voss2004, author = {.}, title = {An Arnoldi method for nonlinear Eigenvalue problems}, journal = {BIT Numerical Mathematics}, year = {2004}, volume = {44}, pages = {387--401}, } @ARTICLE{Sangallo2003, author = {.}, title = {Quasi optimality of the SUPG method for the one--dimensional advection--diffusion problem}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1528--1542}, } @ARTICLE{Michel2003, author = {.}, title = {A finite volume scheme for two--phase immiscible flow in porous media}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, number = {4}, pages = {1301--1317}, } @ARTICLE{SetaKono2004, author = {. and .}, title = {Thermal Lattice Boltzmann Method for Liquid--Gas Two--Phase Flows in Two Dimensions}, journal = {JSME International Journal}, year = {2004}, volume = {47}, number = {3}, pages = {572--583}, } @ARTICLE{Fetecau2002, author = {.}, title = {The Rayleigh--{Stokes} problem for an edge in an Oldroyd--B fluid}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2002}, volume = {335}, pages = {979--984}, } @ARTICLE{Coutier-DelgoshaFortes-PatellaReboudHakimiHirsch2005, author = {. and .}, title = {Stability of preconditioned {Navier}--{Stokes} equations associated with a cavitation model}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {319--349}, } @ARTICLE{GiraultRiviereWheeler2004, author = {.}, title = {A discontinuous {Galerkin} method with nonoverlapping domain decomposition for the {Stokes} and {Navier}--{Stokes} problems}, journal = {Math. Comput.}, year = {2004}, volume = {74}, number = {249}, pages = {53--84}, } @ARTICLE{ChiesaMathiesenMelheimHalvorsen2005, author = { .}, title = {Numerical Simulation of Particulate Flow by the {E}ulerian--{L}agrangian and the {E}ulerian--{E}ulerian Approach with Application to a Fluidized Bed}, journal = {Computers and Chemical Engineering}, year = {2005}, volume = {29}, pages = {291--304}, } @ARTICLE{FerryBalachandar2005, author = {. and .}, title = {Equilibrium Eulerian approach for predicting the thermal field of a dispersion of small particles}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {681--689}, } @ARTICLE{DarbandiMokarizadeh2004, author = {. and .}, title = {A modified pressure--based algorithm to solve flow fields with shock and expansion waves}, journal = {Numerical Heat Transfer}, year = {2004}, volume = {46}, number = {B}, pages = {497--504}, } @ARTICLE{MishraTalukdarTrimisDurst2004, author = {. and . and .}, title = {Effect of angular quadrature schemes on the computational efficiency of the discrete transfer method for solving radiative transport problems with participating medium}, journal = {Numerical Heat Transfer}, year = {2004}, volume = {46}, number = {B}, pages = {463--478}, } @ARTICLE{CuiLi2004, author = {. and .}, title = {A discontinuous finite--element formulation for multidimensional radiative transfer in absorbing, emitting, and scattering media}, journal = {Numerical Heat Transfer}, year = {2004}, volume = {46}, number = {B}, pages = {399--428}, } @ARTICLE{LiuZhangWangLiu2004, author = {. and . and .}, title = {Coupling of {Navier}--{Stokes} Equations with Protein Molecular Dynamics and its Application to Hemodynamics}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {1237--1252}, } @ARTICLE{BehrAroraNoseMotomura2004, author = {.}, title = {Performance analysis of ventricular assist devices using finite element flow simulation}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {1201--1210}, } @ARTICLE{MatsudaShaoYoshinoHoshiharaTanaka2004, author = {. and . and . and .}, title = {Fluid flow analysis by finite element method using arbitrarilu deformed elements (Proposal of the GSMR--method)}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {1181--1199}, } @ARTICLE{CaoPetzold2004, author = {. and .}, title = {A posteriori error estimation and global error control for ordinary differential equations by the adjoint method}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, number = {2}, pages = {359--374}, } @ARTICLE{FengProhl2004, author = {. and .}, title = {Error analysis of a mixed finite element method for the Cahn--Hilliard equation}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {47--84}, } @ARTICLE{LarsonNiklasson2004, author = {. and .}, title = {Analysis of a familiy of discontinuous {Galerkin} methods for elliptic problems: the one dimensional case}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {113--130}, } @ARTICLE{ConnellyKokini2004, author = {. and .}, title = {The effect of shear thinning and differential viscoelasticity on mixing in a model 2D mixer as determined using {FEM} with particle tracking}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {123}, pages = {1--17}, } @ARTICLE{XueTannerPhan-Thien2004, author = {. and . and .}, title = {Numerical modelling of transient viscoelastic flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {123}, pages = {33--58}, } @ARTICLE{FrigaardRyan20004, author = {. and .}, title = {Flow of a visco--plastic fluid in a channel of slowly varying width}, journal = {J. Non-Newtonian Fluid Mech.}, year = {20004}, volume = {123}, pages = {67--83}, } @ARTICLE{Johnson2005, author = {.}, title = {A B--spline collocation method for solving the incompressible {Navier}--{Stokes} equations using an ad hoc method: the Boundary Residual method}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {121--149}, } @ARTICLE{ElmanLoghinWathen2003a, author = {. and . and .}, title = {Preconditioning Techniques for Newton`s Method for the Incompressible {Navier}--{}--{Stokes} Equations}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {961--974}, } @ARTICLE{BenziGanderGolub2003, author = {. and . and .}, title = {Optimization of the Hermitian and Skew--Hermitian splitting iteration for saddle--point problems}, journal = {BIT Numerical Mathematics}, year = {2003}, volume = {43}, pages = {881--900}, } @ARTICLE{DacorognaMoser1990, author = {. and .}, title = {On a partial differential equation involving Jacobian determinant}, journal = {Annales de le Institut Henri Poincare}, year = {1990}, volume = {7}, pages = {1--26}, } @ARTICLE{ChenFeng2004, author = {. and .}, title = {An adaptive finite element algorithm with reliable and efficient error control for linear parabolic problems}, journal = {Math. Comput.}, year = {2004}, volume = {73}, number = {247}, pages = {1167--1193}, } @ARTICLE{ReisenbergerMeileBrennMueller2004, author = {. and M\"{u}.}, title = {Aerodynamic behaviour of prismatic bodies with sharp and rounded edges}, journal = {Experiments in Fluids}, year = {2004}, volume = {37}, pages = {547--558}, } @ARTICLE{GouldScott2004, author = {. and .}, title = {A Numerical Evaluation of HSL Packages for the Direct Solution of Large Sparse, Symmetric Linear Systems of Equations}, journal = {ACM Transactions on Mathematical Software}, year = {2004}, volume = {30}, number = {3}, pages = {300--325}, } @ARTICLE{DavisGilbertLarimoreNg2004, author = { . and .}, title = {A Column Approximate Minimum Degree Ordering Algorithm}, journal = {ACM Transactions on Mathematical Software}, year = {2004}, volume = {30}, number = {3}, pages = {353--376}, } @ARTICLE{BrezinaFalgoutMaclachlanManteuffelMccormickRuge2004, author = { .}, title = {Adaptive smoothed aggregation(\$/alpha\$SA)}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {6}, pages = {1896--1920}, } @ARTICLE{Xu2004, author = {.}, title = {Discontinuous {Galerkin} BGK method for viscous flow equations: one--dimensional systems}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {6}, pages = {1941--1963}, } @ARTICLE{AnBai2005, author = {. and .}, title = {Directional secant method for nonlinear equations}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {175}, pages = {291--304}, } @ARTICLE{LimPeskin2004, author = {. and .}, title = {Simulations of the whirling instability by the immersed boundary method}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {6}, pages = {2066--2083}, } @ARTICLE{GoezLainSommerfeld2004, author = {G\"{. and . and .}, title = {Study of the Numerical Instabilities in {L}agrangian Tracking of Bubbles and Particles in Two--Phase Flow}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {2727--2733}, } @ARTICLE{Goetz2005, author = {G\"{o}.}, title = {Simulating particles in {Stokes} flow}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {175}, pages = {415--427}, } @ARTICLE{Oberman2004, author = {.}, title = {A convergent monotone difference scheme for motion of level sets by mean curvature}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {365--379}, } @ARTICLE{LasoRamirezPicasso2004, author = {. and . and .}, title = {Implicit micro--macro methods}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {215--226}, } @ARTICLE{HousiadasBeris2004, author = {. and .}, title = {An efficient fully implicit spectral scheme for DNS of turbulend viscoelastic channel flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {243--262}, } @ARTICLE{BarraultMadayNguyenPatera2004, author = {. and . and . and .}, title = {An `Empirical Interpolation` Method: Application to Efficient Reduced--basis Discretization of Partial Differential Equations}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {667--672}, } @ARTICLE{Toselli2004, author = {.}, title = {Domain decomposition method of dial--primal FETI type for edge element approximations in three dimensions}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {673--678}, } @ARTICLE{PlatteKuzminFredebeulTurek2005, author = {. and .}, title = {Novel Simulation Approaches for Cyclic Steady--state Fixed--bed Processes Exhibiting Sharp Fronts and Shocks}, journal = {International Series of Numerical Mathematics}, year = {2005}, volume = {151}, pages = {207--223}, note = {Trends and Applications in Constructive Approximation (Eds.) , \& , Birkh\"{a}user Verlag Basel/Switzerland}, } @ARTICLE{MikuleviciusRozovskii2004, author = {. and .}, title = {Stochastic {N}avier--{}--{S}tokes Equations for Turbulent Flows}, journal = {SIAM Journal on Mathematical Analysis}, year = {2004}, volume = {35}, number = {5}, pages = {1250--1310}, } @ARTICLE{AykanatPinarCtalyurek2004, author = {. and . and Ctaly\"{u}rek, \"{U}.V.}, title = {Permuting Sparse Rectangular Matrices into Block--Diagonal Form}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {25}, number = {6}, pages = {1860--1879}, } @ARTICLE{Demlow2004, author = {.}, title = {Localized Pointwise Error Estimates for Mixed Finite Element Methods}, journal = {Math. Comput.}, year = {2004}, volume = {73}, number = {248}, pages = {1623--1653}, note = {DOI: S 0025-5718(04)01650-3}, } @ARTICLE{FortinBertrandFortinChamberlandBoulanger-NadeauMalikiNajeh2004, author = {. and . and . and . and . and .}, title = {An Adaptive Remeshing Strategy for Shear--Thinning Fluid Flow Simluations}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {2363--2375}, note = {DOI: 10.1016/j.compchemeng.2004.06.007}, } @ARTICLE{MarraMolaQuartapelleRiviello2004, author = {. and . and . and .}, title = {Calculation of Impulsively Started Incompressible Viscous Flows}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {877--902}, note = {DOI: 10.1002/fld.743}, } @ARTICLE{Navarro-MartinezTutty2005, author = {. and .}, title = {Numerical Solution of G\"{o}rtler Vortices in Hypersonic Compression Ramps}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {225--247}, note = {DOI: 10.1016/j.compfluid.2004.05.002}, } @ARTICLE{DuarteGormazNatesan2004, author = { .}, title = {Arbitrary {L}agrangian--{E}ulerian Method for {N}avier--{S}tokes Equations with Moving Boundaries}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {4819--4836}, note = {DOI: 10.1016/j.cma.2004.05.003}, } @ARTICLE{TanThompsonHourigan2005, author = {. and . and .}, title = {Evaluating Fluid Forces on Bluff Bodies Using Partial Velocity Data.}, journal = {Journal of Fluids and Structures}, year = {2005}, volume = {20}, pages = {5--24}, note = {DOI: 10.1016/j.fluidstructs.2004.08.002}, } @ARTICLE{NamkoongChoiYoo2005, author = {. and . and .}, title = {Computation of Dynamic Fluid--Structure Interaction in Two--Dimensional Laminar Flows Using Combined formulation}, journal = {Journal of Fluids and Structures}, year = {2005}, volume = {20}, pages = {51--69}, note = {DOI: 10.1016/j.fluidstructs.2004.06.008}, } @ARTICLE{Homeier2005, author = {.}, title = {On Newton--Type Methods with Cubic Convergence}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {176}, pages = {425--432}, note = {DOI: 10.1016/j.cam.2004.07.027}, } @ARTICLE{Baranyi2005, author = {.}, title = {Lift and Drag Evaluation in Translating and Rotating Non--Inertial Systems}, journal = {Journal of Fluids and Structures}, year = {2005}, volume = {20}, pages = {25--34}, note = {DOI: 10.1016/j.fluidstructs.2004.10.009}, } @ARTICLE{BorneCookeville2003, author = {. and Cookeville}, title = {H--Matrices for Convection--Diffusion Problems with Constant Convection}, journal = {Computing}, year = {2003}, volume = {70}, pages = {261--274}, note = {DOI: 10.1007/s00607-003-1474-4}, } @ARTICLE{OsVanPhillips2004, author = {, . and .}, title = {The Prediction of Complex Flows of {} Polymer Melts Using Spectral Elements {}}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {287--301}, note = {DOI: 10.1016/j.jnnfm.2003.12.009}, } @ARTICLE{SimonsenStern2005, author = {. and .}, title = {Flow Pattern Around an Appended Tanker Hull Form in Simple Maneuvering Conditions {}}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {169--198}, note = {DOI: 10.1016/j.compfluid.2004.05.001}, } @ARTICLE{GelhardLubeOlshanskiiStarcke2005, author = {. and . and . and .}, title = {Stabilized Finite Element Schemes with LBB--Stable Elements for Incompressible Flows {}}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {177}, pages = {243--267}, note = {DOI: 10.1016/j.cam.2004.09.017}, } @ARTICLE{Olshanskii1999, author = {.}, title = {An Iterative Solver for the Oseen Problem and Numerical solution of Incompressible {Navier}--{Stokes} Equations}, journal = {Numerical Linear Algebra and Applications}, year = {1999}, volume = {6}, pages = {353--378}, } @ARTICLE{Ehrhard2001, author = {.}, title = {Laminar Mixed Conevction in Two--Dimensional Far Wakes Above Heated/Cooled Bodies: Model and Experiments}, journal = {Journal of Fluid Mechanics}, year = {2001}, volume = {439}, pages = {165--198}, } @ARTICLE{OlshanskiiReusken2004, author = {. and .}, title = {Convergence Analysis of a Multigrid Method for a convection--Dominated Model Problem}, journal = {SIAM Journal on Numerical Analysis}, year = {2004}, volume = {42}, number = {3}, pages = {1261--1291}, note = {DOI:10.1137/S0036142902418679}, } @ARTICLE{Ehrhard1994, author = {.}, title = {The Spreading of Hanging drops}, journal = {Journal of Colloid and Interface Science}, year = {1994}, volume = {168}, pages = {242--246}, } @ARTICLE{MiladinovaLebontoshev2004, author = {. and . and .}, title = {Thin--Film Flow of a Power--Law Liquid Falling Down an Inclined Plate}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {69--78}, note = {DOI: 10.1016/j.jnnfm.2004.01.021}, } @ARTICLE{Suri2005, author = {.}, title = {Stable hp Mixed Finite Elements Based on the Hellinger--Reissner Principle}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {174}, pages = {213--225}, note = {DOI: 10.1016/j.cam.2004.04.008}, } @ARTICLE{NoackMezicTadmorBanaszuk2004, author = { .}, title = {Optimal Mixing in Rercirculation Zones}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, number = {4}, pages = {867--888}, note = {DOI: 10.1063/1.1645276}, } @ARTICLE{AmodioSgura2005, author = {. and .}, title = {Higher--Order Finite Difference Schemes for the Solution of Second--Order BVPs}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {176}, pages = {59--76}, note = {DOI:10.1016/j.cam.2004.07.008}, } @ARTICLE{Zube2004, author = {.}, title = {Number Systems, \$alpha\$--Splines and Refinement}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {172}, pages = {207--231}, note = {DOI: 10.1016/j.cam.2004.02.004}, } @ARTICLE{TomeGrossiCasteloCuminatoMangiavacchiFerreiraDeSoussaMcKee2004, author = { .}, title = {A Numerical Method for Solving Three--Dimensional Generalized Newtonian Free surface Flows}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {123}, pages = {85--103}, note = {DOI: 10.1016/j.jnnfm.2004.06.012}, } @ARTICLE{BramblePasiack2003, author = {. and .}, title = {A New Approximation Technique for Div--Curl Systems}, journal = {Mathematics of Computation}, year = {2003}, volume = {73}, number = {248}, pages = {1739--1762}, } @ARTICLE{IkedaDurbin2004, author = {. and .}, title = {Mesh Stretch Effects on Convection in Flow Simulations}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {110--125}, note = {DOI: 10.1016/j.jcp.2004.02.006}, } @ARTICLE{LiLi2004, author = { .}, title = {A Novel Approach of Divergence--Free Reconstruction for Adaptive Mesh Refinement}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {1--15}, note = {DOI: 10.1016/j.jcp.2004.01.027}, } @ARTICLE{AlonsoRussoPadraRodriguez2004, author = { .}, title = {Accurate Pressure Post--Process of a {} Finite Element Method for Elastoacoustics}, journal = {Numerische Mathematik}, year = {2004}, volume = {98}, pages = {389--425}, note = {DOI: 10.1007/s00211-004-0523-z}, } @ARTICLE{DoduRabut2004, author = {. and .}, title = {Irrotational or Divergence--Free Interpolation}, journal = {Numerische Mathematik}, year = {2004}, volume = {98}, pages = {477--498}, note = {DOI:10.1007/s00211-004-0541-x}, } @ARTICLE{TanZhangHuangTang2004, author = {. and . and .}, title = {Moving Mesh methods with Locally {} Varying Time Steps.}, journal = {J. Comput. Phys.}, year = {2004}, volume = {200}, pages = {347--367}, note = {DOI: 10.1016/j.jcp.2004.04.007}, } @ARTICLE{OliveiraSorensen2005, author = {. and .}, title = {A New Class of Preconditioners for Large--Scale Linear Systems from Interior Point Methods for Linear Programming}, journal = {Linear Algebra and\_its Applications}, year = {2005}, volume = {394}, pages = {1--24}, note = {DOI: 10.1016/j.Iaa.2004.08.019}, } @ARTICLE{BartelsCarstensenDolzmann2004, author = {. and . and .}, title = {Inhomogeneous Dirichlet Conditions in a Priori and a Posteriori Finite Element {} Error {} Analysis}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {1--24}, note = {DOI: 10.1007/s00211-004-0548-3}, } @ARTICLE{BartelsCarstensen2004, author = {. and .}, title = {Averaging Techniques Yield Reliable {} a Posteriori Finite Element Error {} Control {} for Obstacle Problems.}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {225--249}, note = {DOI: 10.1007/s00211-004-0553-6}, } @ARTICLE{StrakosTichy2002, author = {. and .}, title = {On Error Estimation in the Conjugate Gradient Method and Why It Works in Finite Precision Computations. {}}, journal = {Electronic Transactions on Numerical Analysis }, year = {2002}, volume = {13}, pages = {56--80}, } @ARTICLE{ChoJun2004, author = {. and .}, title = {\$gamma\$--Adaptive Mesh Generation for Shell Finite Element Analysis}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {291--316}, note = {DOI: 10.1016/j.jep.2004.02.007}, } @ARTICLE{Yun2005, author = {.}, title = {A Note on SOR (\$omega\$) Splitting of an M--Matrix}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {176}, pages = {461--462}, note = {DOI: 10.1016/j.cam.2004.07.029}, } @ARTICLE{XiePasquali2004, author = {. and .}, title = {A New , Convenient Way of Imposing Open--Flow Boundary Conditions in Two--and Three--Dimensional {} Viscoelastic Flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {159--176}, note = {DOI: 10.1016/j.jnnfm.2004.02.011}, } @ARTICLE{FoteinopoulouMavrantzasTsamopoulos2004, author = { . and .}, title = {Numerical Simulation of Bubble Growth in Newtonian and Viscoelastic filaments Undergoing Stretching}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, number = {f7200}, pages = {177--200}, note = {DOI: 10.1016/j.jnnfm.2004.02.012}, } @ARTICLE{FangLozinskiOwens2004, author = { .}, title = {Towards More Realistic Kinetic Models for Concentrated Solutions and Melts. {}}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {79--90}, note = {DOI: 10.1016/j.jnnfm.2003.11.011}, } @ARTICLE{ArreguiCendanVazquez2005, author = {. and . and .}, title = {A Duality for the Compressible Reynold`s Equation.Application to Simulation of Read/Write Processes in Magnetic Storage Devices.}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {175}, pages = {31--40}, note = {DOI: 10.1016/j.cam.2004.06.009}, } @ARTICLE{YuanWu2004, author = {. and .}, title = {An Implicit Three--Dimensional Fully Non--Hydrostatic Model for Free--Surface Flows.}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {709--733}, note = {DOI: 10.1002/fld.778}, } @ARTICLE{ColonnaRebay2004, author = {. and .}, title = {Numerical Simulation of Dense Gas Flows on Unstructured Grids with an Implicit High Resolution Upwind euler Solver.}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {735--765}, note = {DOI: 10.1002/fld.762}, } @ARTICLE{ZhangJiaWang2004, author = {. and . and .}, title = {2D Nearly Orthogonal Mesh Generation}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {685--707}, note = {DOI: 10.1002/fld.770}, } @ARTICLE{DeanGlowinski2004, author = {. and .}, title = {Numerical Solution of the Two--Dimensional Elliptic Monge--Ampere Equation with Dirichlet Boundary Conditions: A Least Squares Approach}, journal = {Mathematique}, year = {2004}, volume = {1339}, pages = {887--892}, note = {DOI: 10.1016/j.crma.2004.09.018}, } @ARTICLE{SahinOwens2004a, author = {. and .}, title = {On the Effects of Viscoelasticity on Two--Dimensional Vortex Dynamics in the Cylinder Wake {}}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {123}, pages = {121--139}, note = {DOI: 10.1016/j.jnnfm.2004.08.002}, } @ARTICLE{RiveraHenicheAscanioTanguy2004, author = {. and . and . and .}, title = {A Virtual Element Model for Centered and Eccentric Mixer Configurations}, journal = {Computers and Chemical Engineering}, year = {2004}, volume = {28}, pages = {2459--2468}, note = {DOI: 10.1016/j.compchemeng.2004.06.012}, } @ARTICLE{HorsleyHorsleyWilsonJones2004, author = {. and . and . and .}, title = {Non--Newtonian Effects on Fall Velocities of Vertically Aligned spheres}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {124}, pages = {147--152}, note = {DOI: 10.1016/j.jnnfm.2004.09.002}, } @ARTICLE{KimKimAhnLee2004, author = {. and . and .}, title = {An Efficient Iterative Solver and High--Resolution Computations of the Oldroyd--B Fluid Past a Confined Cylinder}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {123}, pages = {161--173}, note = {DOI: 10.1016/j.jnnfm.2004.08.003}, } @ARTICLE{NieArmaly2004, author = {. and .}, title = {Reverse Flow in Three--Dimensional Backward--Facing Step Flow {}}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {4713--4720}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.05.027}, } @ARTICLE{TsaiChengOsherBurchardSapiro2004, author = {. and . and . and . and .}, title = {Visibility and its Dynamics in a {PDE} Based Implicit framework}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {260--290}, note = {DOI: 10.1016/j.jcp.2004.02.015}, } @ARTICLE{LuoBewley2004, author = {. and .}, title = {On the Contravariant Form of the {Navier}--{Stokes} Equations in Time--Dependent Curvilinear Coordinate Systems}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {355--375}, note = {DOI: 10.1016/j.jcp.2004.02.012}, } @ARTICLE{Nakajimaokuda2004, author = {. and .}, title = {Parallel Iterative Solvers with Selective Blocking Preconditioning for Simulations of Fault--Zone Contact {}}, journal = {Numerical Linear Algebra and Applications}, year = {2004}, volume = {11}, pages = {831--852}, note = {DOI: 10.1002/nla.349}, } @ARTICLE{WarsaBenziWareingMorel2004, author = {. and . and . and .}, title = {Preconditioning a Mixed Discontinuous Finite Element Method for Radiation Diffusion {}}, journal = {Numerical Linear Algebra and Applications}, year = {2004}, volume = {11}, pages = {795--811}, note = {DOI: 10.1002/nla.347}, } @ARTICLE{ChungPakChang2004, author = {. and . and .}, title = {A Numerical Study of Two--Phase Flow Using a Two--Dimensional Two--Fluid Model}, journal = {Numerical Heat Transfer}, year = {2004}, volume = {45}, pages = {1049--1066}, note = {DOI: 10.1080/10407780490453891}, } @ARTICLE{RanjanIrudayarajRedyyMujundar2004, author = {.}, title = {Finite--Element Simulation and Validation of Stepwise Drying of Bananas}, journal = {Numerical Heat Transfer}, year = {2004}, volume = {45}, pages = {997--1012}, note = {DOI: 10.1080/10407780490453963}, } @ARTICLE{AboubacarPhillipsTamaddon-JahromiSnigerevWebster2004, author = { .}, title = {High--Order Finite Volume Methods for Viscoelastic Flow Problems}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {16--40}, note = {DOI: 10.1016./J.JCP.2004.01.033}, } @ARTICLE{JohnstonLiu2004, author = {. and .}, title = {Accurate, Stable and Efficient {Navier}--{Stokes} Solvers Based on Explicit Treatment of the Pressure Term}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {221--259}, note = {DOI: 10.1016/j.jcp.2004.02.009}, } @ARTICLE{MarquezMeddahiSelgas2004, author = { .}, title = {A New BFM\_FEM Coupling Strategy {} for Two--Dimensional Fluid--Solid Interaction Problems}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {205--220}, note = {DOI: 10.1016/j.jcp.2004.02.005}, } @ARTICLE{RylanderJin2004, author = {. and .}, title = {Perfectly Matched Layer for the Time Domain Finite Element Method {}}, journal = {J. Comput. Phys.}, year = {2004}, volume = {200}, pages = {238--250}, note = {DOI: 10.1016/j.jcp.2004.03.016}, } @ARTICLE{BurkardtGunzburgerPeterson2002, author = {. and . and .}, title = {Insensitive {} Functionals, Inconsistent Gradients, Spurious Minima , and Regularized Functionals in Flow Optimization Problems}, journal = {International Journal of Computational Fluid Dynamics}, year = {2002}, volume = {16}, pages = {171--185}, note = {DOI: 10.1080/10618560290034663}, } @ARTICLE{JohnKaya2004, author = {. and .}, title = {A Finite Variational Multiscale Method for the {Navier}--{Stokes} Equations}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, pages = {000--000}, note = {DOI: 10.1137/030601533}, } @ARTICLE{John2005, author = {.}, title = {An Assessment of Two Models for the Subgrid Scale Tensor in The Rational LES Model}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {173}, pages = {57--80}, note = {DOI: 10.1016/j.cam.2004.02.022}, } @ARTICLE{DuncaJohnLayton2004, author = {. and . and .}, title = {The Communication Error of the Space Averaged {Navier}--{Stokes} Equations on a Bounded Domain}, journal = {Advances in Mathematical Fluid mechanics}, year = {2004}, pages = {53--78}, } @ARTICLE{DomenichiniBaccani2004, author = {. and .}, title = {A formulation of {Navier}--{Stokes} Problem in Cylindrical Coordinates applied to the 3D Entry Jet in a Duct}, journal = {J. Comput. Phys.}, year = {2004}, volume = {200}, pages = {177--191}, note = {DOI:10.1016/j.jcp.2004.04.002}, } @ARTICLE{MuellerOhlaZauterEbner2004, author = {. and .}, title = {Effect of Elements on Porosity in Spray--Formed Copper--Alloy Billets}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {78--86}, note = {DOI:10.1016/j.mse.2004.02.060}, } @ARTICLE{MesquitaBarbosa2004, author = {. and .}, title = {Spray Forming High Speed Steel--Properties and Processing}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {87--95}, note = {DOI: 10.1016/j.mser.2004.02.035}, } @ARTICLE{MarkusCuiFrotsching2004, author = {. and .}, title = {Analysis of deposit Growth in Spray Forming with Multiple Atomizers}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {166--174}, } @ARTICLE{HattelPrydsPedersen2004, author = {. and . and .}, title = {An Integrated Numerical Model for the Prediction of {G}aussian and Billet Shapes}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {184--189}, } @ARTICLE{WangYuNg2004, author = {. and . and .}, title = {Particle Velocities, Sizes and Flux Distribution in Plasma Spray with Two Powder Injection Ports {}}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {122--136}, } @ARTICLE{SrivastavaMandalOjha2004, author = {. and .}, title = {Evolution of Microstructure in Spray Formed AI--18\%Si Alloy}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {14--20}, note = {DOI: 10.1016/j.msea.2004.02.031}, } @ARTICLE{YangHannula2004, author = { .}, title = {Soundness of Spray Formed Disc Shape Tools of Hot--Work Steels {}}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {39--44}, } @ARTICLE{ZhangLinHillertSellebyShiYanZhangLuoLiWang2004, author = {. and . and . and . and . and . and . and . and . and .}, title = {Microstructure and Mechanical Properties of Spray Formed Ultrahigh--Carbon Steels}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {45--49}, } @ARTICLE{HughesFranca1987, author = {. and .}, title = {A New Finite Element Formulation for Computational Fluid Dynamics: {VII.} {T}he {S}tokes--Problem with Various Well--Posed Boundary Conditions: Symmetric Formulations that Converge for All Velocity/Pressure Spaces}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1987}, volume = {65}, pages = {85--96}, } @ARTICLE{ZhangLin2005, author = {. and .}, title = {Locating Superconvergent Points of Finite Element Methods in 3D}, journal = {International Journal of Numerical Analysis And Modelling}, year = {2005}, volume = {2}, number = {1}, pages = {19--30}, } @ARTICLE{Date2005, author = {.}, title = {Solution of Transport Equations on Unstructured Meshes with Cell--Centered Colocated Variables . Part.1: Discretization}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {1117--1127}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.09.036}, } @ARTICLE{PimpalnerkarKulkarniDate2005, author = {. and . and .}, title = {Solution of Transport Equations on Unstructured Meshes with Cell--Centered Colocated Variables . Part.2: Applications}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {1128--1136}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.09.037}, } @ARTICLE{WapperomRenardy2005, author = {. and .}, title = {Numerical Prediction of the Boundary Layers in the Flow Around a Cylinder Using a Fixed Velocity Field {}}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {125}, pages = {35--48}, note = {DOI: 101016/j.jnnfm.2004.09.005}, } @ARTICLE{Eid2005, author = {.}, title = {Higher Order Isoparametric Finite Element Solution of {Stokes} Flow}, journal = {Applied Mathematics and Computation}, year = {2005}, volume = {162}, pages = {1083--1101}, note = {DOI: 10.1016/j.amc.2004.01.014}, } @ARTICLE{RenoufDuboisAlart2004, author = {. and . and .}, title = {A Parallel Version of the Non Smooth Contact Dynamics Algorithm Applied to the Simulation of Granular Media}, journal = {J. Comput. Appl. Math.}, year = {2004}, volume = {168}, pages = {375--382}, note = {DOI: 10.1016/j.cam.2003.05.019}, } @ARTICLE{MayorDAnnaBarratLoreto2005, author = {. and .}, title = {Observing Brownian Motion and Measuring Temperatures in Vibration--Fluidized Granular Matter}, journal = {New Journal of Physics}, year = {2005}, volume = {7}, pages = {1--16}, note = {DOI: 10.1088/1367-2630/7/1/028}, } @ARTICLE{PullinWang2004, author = {. and .}, title = {Unsteady Forces on an Accelerating Plate and Application to Hovering Insect Flight}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {509}, pages = {1--21}, note = {DOI: 10.1017/S0022112004008821}, } @ARTICLE{Waltz2004a, author = {.}, title = {Microfluidics Simulation Using Adaptive Unstructured Grids}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {46}, pages = {939--960}, note = {DOI: 10.1002/fld.753}, } @ARTICLE{BazFrommerSpiteri2005, author = {. and . and .}, title = {Asynchronous Iterations with Flexible Communication: Contracting Operators.}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {176}, pages = {91--103}, note = {DOI: 10.1016/j.cam.2004.07.009}, } @ARTICLE{PerazzoGratton2004, author = {. and .}, title = {{Navier}--{Stokes} Solutions for Parallel Flow in Rivulets on an Inclined Plane}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {507}, pages = {367--379}, note = {DOI: 10.1017/S0022112004008791}, } @ARTICLE{Li2005, author = {.}, title = {Comparison Results for Solving Preconditioned Linear Systems}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {176}, pages = {319--329}, note = {DOI: 10.1016/j.cam.2004.07.022}, } @ARTICLE{BushHasha2004, author = {. and .}, title = {On the Collision of Laminar Jets:Fluid Chains and Fishbones}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {511}, pages = {285--310}, note = {DOI: 10.1017/S002211200400967X}, } @ARTICLE{Chen2005, author = {.}, title = {Orthogonality Correction Technique in Superconvergence Analysis}, journal = {International Journal of Numerical Analysis And Modelling}, year = {2005}, volume = {2}, pages = {31--42}, } @ARTICLE{Allali2005, author = {.}, title = {A Priori and a Posteriori Error Estimates for Boussinesq Equations}, journal = {International Journal of Numerical Analysis And Modelling}, year = {2005}, volume = {2}, pages = {179--196}, } @ARTICLE{Meyer2003, author = {.}, title = {Stable Evaluation of Jacobian Matrices on Highly Refined Finite Element Meshes}, journal = {Computing}, year = {2003}, volume = {70}, pages = {359--373}, note = {DOI: 10.1007/s00607-003-0010-x}, } @ARTICLE{HaasdonkOhlbergerRumpfSchmidtSiebert2003, author = {. and .}, title = {Multiresolution Visualization of Higher Order Adaptive Finite Element Simulations}, journal = {Computing}, year = {2003}, volume = {70}, pages = {181--204}, note = {DOI: 10.1007/s00607-003-1476-2}, } @ARTICLE{Agouzal2005, author = {.}, title = {A Posteriori Error Estimation For Finite Element Discretizations of Quasi--Newtonian {Stokes} Flows}, journal = {International Journal of Numerical Analysis And Modelling}, year = {2005}, volume = {2}, pages = {221--239}, } @ARTICLE{HughesFrancaBalestra1986, author = {. and . and .}, title = {A New Finite Element Formulation for Computational Fluid Dynamics: {V.} {C}ircumventing the {B}abuska--{B}rezzi Condition. {A} Stable {P}etrov--{G}alerkin Formulation of the {S}tokes Problem Accomodating Equal--Order Interpolations}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {1986}, volume = {59}, number = {1}, pages = {85--99}, } @ARTICLE{PhelanMaloneWinter1989, author = {. and . and .}, title = {A Purely Hyperbolic Model For Unsteady Viscoelastic Flow}, journal = {J. Non-Newtonian Fluid Mech.}, year = {1989}, volume = {32}, pages = {197--224}, } @ARTICLE{Ehrhard1993, author = {.}, title = {Experiments on Isothermal and Non--Isothermal Spreading}, journal = {Journal of Fluid Mechanics}, year = {1993}, volume = {257}, pages = {463--483}, } @ARTICLE{GuWu2005, author = {. and .}, title = {Alternating--Direction Iterative Technique for Mixed Finite Element Methods for {Stokes} Equations}, journal = {Applied Mathematics and Computation}, year = {2005}, volume = {162}, pages = {1035--1047}, note = {DOI: 10.1016/j.amc.2004.01.021}, } @ARTICLE{YangHeTangRao2005, author = { .}, title = {Inexact Block Newton Methods for Solving Nonlinear Equations}, journal = {Applied Mathematics and Computation}, year = {2005}, volume = {162}, pages = {1207--1218}, note = {DOI: 10.1016/j.amc.2004.03.002}, } @ARTICLE{YamamotoHayakawa2003, author = {. and .}, title = {Error Estimation With Guaranteed Accuracy of Finite Element Method in Nonconvex Polygonal Domains}, journal = {J. Comput. Appl. Math.}, year = {2003}, volume = {159}, pages = {173--183}, note = {DOI: 10.1016/S0377-0427(03)00569-7}, } @ARTICLE{CarstensenFunken2000a, author = {. and .}, title = {Fully Reliable Localized Error Control in the {FEM}}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {21}, pages = {1465--1484}, } @ARTICLE{ChenWang2003a, author = {. and .}, title = {An Interior Estimate of Superconvergence for Finite Element Solutions for Second--Order Elliptic Problems on Quasi--Uniform Meshes by Local Projections}, journal = {SIAM Journal on Numerical Analysis}, year = {2003}, volume = {41}, pages = {1318--1338}, note = {DOI: 10.1137/S0036142902410039}, } @ARTICLE{NewberyRaymentGrant2004, author = {. and . and .}, title = {A Particle Image Velocimetry Investigation of In--Flight and Deposition Behaviour of Steel Droplets During Electric Arc Sprayforming}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {137--145}, } @ARTICLE{ZimmermannLandes2004, author = {. and .}, title = {A Particle Image Shape Imaging (PSI) Investigation of Particles in a Plasma Jet}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {153--157}, } @ARTICLE{HoileRaymentGrantRoche2004, author = { . and . and .}, title = {Oxide Formation in the Sprayform Tool Process}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {50--57}, } @ARTICLE{HoileRaymentGrantRoche2004a, author = {. and . and . and .}, title = {Microstructure Development in Alloy Splats During Rapid Solidification}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {175--183}, } @ARTICLE{NannelliSucci1992, author = {. and .}, title = {The Lattice Boltzmann Equation on Irregular Lattices}, journal = {Journal of Statistical Physics}, year = {1992}, volume = {68}, pages = {401--407}, } @ARTICLE{HofDoorneWesterweelNieuwstadtFaisstEckhardtWedinKerswellWaleffe2004, author = {.}, title = {Experimental Observation of Nonlinear Traveling Waves in Turbulent Pipe Flow}, journal = {Science Magazine}, year = {2004}, volume = {305}, pages = {1594--1597}, } @ARTICLE{BurmanErn1997, author = { .}, title = {Stabilized {Galerkin} Approximation of Convection--Diffusion--Reaction Equations: Discrete Maximum Principle And Convergence}, journal = {Mathematics of Computation}, year = {1997}, pages = {000--000}, } @ARTICLE{KeshtibanBelblidiaWebster2005, author = {. and .}, title = {Computation of Incompressible and Weekly--Compressible viscoelastic Liquids Flow: Finite Element/Volume Schemes}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, pages = {1--21}, } @ARTICLE{ZhouAdvaniWetzel2004, author = { .}, title = {Slow Drag in Granular Materials Under High Pressure}, journal = {Physical Review}, year = {2004}, volume = {69}, note = {DOI: 10.1103/PhysRevE.69.061306}, } @ARTICLE{LiaoPanSu1994, author = {. and . and .}, title = {Numerical Grid Generator Based on Moser`s Deformation Method}, journal = {Numerical Methods for Partial Differential Equations}, year = {1994}, volume = {10}, pages = {21--31}, } @ARTICLE{BreschDesjardinsLin2003, author = {. and . and .}, title = {On Some Compressible Fluid Models: Korteweg, Lubrication, and Shallow Water Systems}, journal = {Communications in Partial Differential Equations}, year = {2003}, volume = {28}, pages = {843--868}, note = {DOI: 10.1081/PDE-120020499}, } @ARTICLE{Mazzone2003, author = {.}, title = {A Single Phase Variational Problem Involving the Area of Level Surfaces}, journal = {Communications in Partial Differential Equations}, year = {2003}, volume = {28}, pages = {991--1004}, note = {DOI: 10.1081/PDE-120021183}, } @ARTICLE{ChoeKim2003, author = {. and .}, title = {Strong Solutions of the {Navier}--{Stokes} Equations for Nonhomogeneous Incompressible Fluids}, journal = {Communications in Partial Differential Equations}, year = {2003}, volume = {28}, pages = {1183--1201}, note = {DOI: 10.1081/PDE-120021191}, } @ARTICLE{AxelssonKaratson2004, author = {. and .}, title = {Superlinearly Convergent CG Methods via Equivalent Preconditioning for Nonsymmetric Elliptic Operators}, journal = {Numerische Mathematik}, year = {2004}, volume = {99}, pages = {197--223}, note = {DOI: 10.1007/s00211-004-0557-2}, } @ARTICLE{FritzHueeberWohlmuth2004, author = { .}, title = {A Comparision of Mortar and Nitsche Techniques for Linear Elasticity}, journal = {Calcolo}, year = {2004}, volume = {41}, pages = {115--137}, note = {DOI: 10.1007/s10092-004-0087-4}, } @ARTICLE{JuengelWang2003, author = { .}, title = {Convergence of Nonlinear Schr\"{o}dinger--Poisson Systems to the Compressible Euler Equations}, journal = {Communications in Partial Differential Equations}, year = {2003}, volume = {28}, pages = {1005--1022}, note = {DOI: 10.1081/PDE-120021184}, } @ARTICLE{DutrifoyHmidi2004, author = {. and .}, title = {The Incompressible Limit of Solutions of the Two--Dimensional Compressible Euler System with Degenerating Initial Data}, journal = {Communications on Pure and Applied Mathematics}, year = {2004}, volume = {LVII}, pages = {1159--1177}, } @ARTICLE{VazquezZuazua2003, author = {. and .}, title = {Large Time Behavior for a Simplified 1D Model of Fluid--Solid Interaction}, journal = {Communications in Partial Differential Equations}, year = {2003}, volume = {28}, pages = {1705--1738}, note = {DOI: 10.1081/PDE-120024530}, } @ARTICLE{Kotalik2005, author = {.}, title = {Modelling of an Argon Plasma Flow}, journal = {Czechoslovak Journal of Physics}, year = {2005}, volume = {55}, pages = {1--16}, } @ARTICLE{BachMoehwaldRothardtPrehmEnglHartzDroesler2004, author = {. and . and . and . and . and . and .}, title = {Particle image Velocimetry in Thermal Spraying}, journal = {Material Science and Engineering}, year = {2004}, volume = {383}, pages = {146--152}, note = {DOI: 10.1016/j.msea.2004.02.044}, } @ARTICLE{GlowinskiGuidoboniPan2005, author = {. and .}, title = {Wall--Driven Incompressible Viscous Flow in a Two--Dimensional Semi--Circular Cavity}, journal = {J. Comput. Phys.}, year = {2005}, pages = {1--16}, } @ARTICLE{BenziTuma1999, author = {. and .}, title = {A Comparitive Study of Sparse Approximative Inverse Preconditioners}, journal = {Applied Numerical Mathematics}, year = {1999}, volume = {30}, pages = {305--340}, } @ARTICLE{StrzodkaDroskeRumpf2004, author = { .}, title = {Image Registration by a Regularized Gradient Flow {}--A Streaming Implementation in DX9 Graphics Hardware}, journal = {Computing}, year = {2004}, note = {To appear}, } @ARTICLE{RenardyRenardy2004, author = {. and .}, title = {Similarity Solutions for Breakup of Jets of Power Law Fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2004}, volume = {122}, pages = {303--312}, note = {DOI: 10.1016/j.jnnfm.2004.01.026}, } @ARTICLE{Juncu2005, author = {.}, title = {Conjugate Heat/Mass Transfer From a Circular Cylinder with an Internal Heat/Mass Source in Laminar Crossflow at Low Reynolds Numbers}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {419--424}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.09.020}, } @ARTICLE{PengShiauHwang2003, author = {. and . and .}, title = {Transition in a 2--D lid--driven cavity flow}, journal = {Computers and Fluids}, year = {2003}, volume = {32}, pages = {337--352}, } @ARTICLE{HirabayashiOhtaRuefenachtChopard2004, author = { .}, title = {A lattice Boltzmann study of blood flow in stented aneurism}, journal = {Future Generation Computer Systems}, year = {2004}, volume = {20}, pages = {925--934}, } @ARTICLE{KunzBogerStinebringChyczewskiGibelingVen, author = {. and .}, title = {A preconditioned {Navier}--{Stokes} method for two--phase flows with application to cavitation prediction}, journal = {Computers and Fluids}, year = {2000}, volume = {29}, number = {8}, pages = {849--875}, note = {\url{http:/ / www.personal.psu.edu/ faculty/ r/ f/ rfk102/ PROF/ AIAACFD99paper/ AIAAcfd99.pdf}}, } @ARTICLE{BrezinaClearyFalgoutHensonJonesManteuffelMcCormickRuge2000, author = { .}, title = {Algebraic multigrid based on element interpolation (AMGe)}, journal = {SIAM Journal on Scientific Computing}, year = {2000}, volume = {22}, number = {5}, pages = {1570--1592}, } @ARTICLE{TanahashiOotsuFukushimaMiyauchi2002, author = {. and . and . and .}, title = {Measurement of coherent fine--scale eddies in turbulent mixing layer by DPIV}, journal = {Engineering Turbulence Modelling and Experiments}, year = {2002}, volume = {5}, pages = {525--534}, } @ARTICLE{TardosKhanSchaeffer1998, author = {. and . and .}, title = {Forces of a slowly rotating, rough cylinder in a Couette device containing a dry, frictional powder}, journal = {Physics of Fluids}, year = {1998}, volume = {10}, number = {2}, pages = {335--341}, } @ARTICLE{CorliFan2004, author = {. and .}, title = {The Riemann Problem for Reversible Reactive Flows With Metastability}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {65}, pages = {426--457}, note = {DOI: 10.1137/S0036139903429671}, } @ARTICLE{LinPayne2004, author = {. and .}, title = {Spatial Decay Bounds in the Time Dependent Pipe Flow of an Incompressible Viscous Fluid}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {65}, pages = {458--474}, note = {DOI: 10.1137/040606326}, } @ARTICLE{HigdonKennedyCavendishCafeoRyne2004, author = {. and . and . and . and .}, title = {Combining Field Data and Computer Simulations for Calibration and Prediction}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, pages = {448--466}, note = {DOI: 10.1137/S1064827503426693}, } @ARTICLE{Borisyuk2003, author = {.}, title = {Experimental Study of Wall Pressure Fluctuations in a Pipe Behind a Stenosis}, journal = {International Journal of Fluid Mechanics Research}, year = {2003}, volume = {30}, pages = {264--278}, } @ARTICLE{Volynsky2003, author = {.}, title = {An Approximated Mathematical Simulation of a Flow Through Jointed Porous Formations}, journal = {International Journal of Fluid Mechanics Research}, year = {2003}, volume = {30}, pages = {337--346}, } @ARTICLE{Silva2005, author = {.}, title = {Resolvent Estimates for Plane Couette Flow}, journal = {SIAM Journal on Applied Mathematics}, year = {2005}, volume = {65}, pages = {667--683}, note = {DOI: 10.1137/S0036139903426940}, } @ARTICLE{Bila2004, author = {.}, title = {Application of Symmetry Analysis to a {PDE} Arising in the Car Windshield Design}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {65}, pages = {113--130}, note = {DOI: 10.1137/S0036139903434031}, } @ARTICLE{LeMaitreReaganDebusschereNajmGhanemKnio2004, author = { .}, title = {Natural Convection in a Closed Cavity Under Stochastic Non--Boussinesq Conditions}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, pages = {375--394}, note = {DOI: 10.1137/S1064827503422853}, } @ARTICLE{BirginCastilloMartinez2005, author = {. and . and .}, title = {Numerical Comparision of Augmented {L}agrangian Algorithms for Nonconvex Problems}, journal = {Computational Optimization and Applications}, year = {2005}, volume = {31}, pages = {31--55}, } @ARTICLE{MunnierZuazua2005, author = {. and .}, title = {Large Time Behaviour for a Simplified N--Dimensional Model of Fluid--Solid Interaction}, journal = {Communications in Partial Differential Equations}, year = {2005}, volume = {30}, pages = {377--417}, note = {DOI: 10.1081/PDE-200050080}, } @ARTICLE{Silvestre2005, author = {.}, title = {The Two Membranes Problem}, journal = {Communications in Partial Differential Equations}, year = {2005}, volume = {30}, pages = {245--257}, note = {DOI: 10.1081/PDE-200044490}, } @ARTICLE{Kim2005, author = {.}, title = {A Free Boundary Problem With Curvature}, journal = {Communications in Partial Differential Equations}, year = {2005}, volume = {30}, pages = {121--138}, note = {DOI: 10.1081/PDE-200044474}, } @ARTICLE{DengHouYu2005, author = {.}, title = {Geometric Properties and Nonblowup of 3D Incompressible Euler Flow}, journal = {Communications in Partial Differential Equations}, year = {2005}, volume = {30}, pages = {225--243}, note = {DOI: 10.1081/PDE-200044488}, } @ARTICLE{Renardy2005, author = {.}, title = {Post--Breakup Asymptotics for a Giesekus Jet}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {126}, pages = {1--5}, note = {DOI: 10.1016/j.jnnfm.2004.12.004}, } @ARTICLE{BischofBueckerRasch2004, author = {. and . and .}, title = {Sensitivity Analysis of Turbulence Models Using Automatic Differentiation}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, pages = {510--522}, note = {DOI: 10.1137/S1064827503426723}, } @ARTICLE{Zhou2005, author = {.}, title = {Asymptotic Stability for the 3D {N}avier--{}--{S}tokes Equations}, journal = {Communications in Partial Differential Equations}, year = {2005}, volume = {30}, pages = {323--333}, note = {DOI: 10.1081/PDE-200037770}, } @ARTICLE{FarnellDavidBarton2004, author = {. and . and .}, title = {Numerical Simulations of a Filament in a Flowing Soap Film}, journal = {International Journal for Numerical Methods in Fluids}, year = {2004}, volume = {44}, pages = {313--330}, note = {DOI: 10.1002/fld.640}, } @ARTICLE{DoiMakino2005, author = {. and .}, title = {Sedimentation of Particles of General Shape}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1884625}, } @ARTICLE{HwangAndersonHulsen2005, author = {. and . and .}, title = {Chaotic Advection in a Cavity Flow With Rigid Particles}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1884465}, } @ARTICLE{MeyersGeurtsBaelmans2005, author = {. and . and .}, title = {Optimality of the Dynamic Procedure for Large--Eddy Simulations}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1879054}, } @ARTICLE{Shen2005, author = {.}, title = {Use of Degenerated Reynolds Equation in Solving the Microchannel Flow Problem}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1867474}, } @ARTICLE{Sauer-BudgePeraire2004, author = {. and .}, title = {Computing Bounds for Linear Functionals of Exact Weak Solutions to the Advection--Diffusion--Reaction Equation}, journal = {SIAM Journal on Scientific Computing}, year = {2004}, volume = {26}, pages = {636--652}, note = {DOI: 10.1137/S1064827503427121}, } @ARTICLE{BarberonHelluy2005, author = {. and .}, title = {Finite Volume Simulation of Cavitating Flows}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {832--858}, note = {DOI: 10.1016/j.compfluid.2004.06.004}, } @ARTICLE{BoffiGastaldi2002, author = {. and .}, title = {On the Quadrilateral {} {Q}2 {}--{}--{} {P} Element for the {Stokes} Problem}, journal = {International Journal for Numerical Methods in Fluids}, year = {2002}, volume = {39}, pages = {1001--1011}, note = {DOI:10.1002/fld.358}, } @ARTICLE{WojikMarr2005, author = {. and .}, title = {Mikroverfahrenstechnische Prinzipien in der Fl\"{u}ssig/Fl\"{u}ssig {}--{}--{} Extraktion}, journal = {Chemie Ingenieur Technik}, year = {2005}, volume = {77}, number = {6}, pages = {653--668}, note = {DOI:10.1002/cite.200407102}, } @ARTICLE{RumseyMelissaRiversMorrison2005, author = {. and . and .}, title = {Study of {CFD} Variation on Transport Configurations from the Second Drag--prediction Workshop}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {785--816}, note = {DOI: 10.1016/j.compfluid.2004.07.003}, } @ARTICLE{ReynaudDebrayJean-PierreMaitre2005, author = {. and . and .}, title = {Hydrodynamics and Heat Transfer in Two--dimensional Minichannels}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {3197--3211}, note = {DOI: 10.1016/j.ijheatmasstransfer.2005.02.020}, } @ARTICLE{SenguptaRussellMinkowyczMashayek2005, author = { .}, title = {Numerical Simulation Data for Assessment of Particle--laden Turbulent Flow Models}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {3035--3046}, note = {DOI: 10.1016/j.ijheatmasstransfer.2005.03.003}, } @ARTICLE{LecoqAnthoreCichockiSzymczakFeuillebois2004, author = { . and .}, title = {Drag Force on a Sphere Moving Towards a Corrugated Wall}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {513}, pages = {247--264}, note = {DOI: 10.1017/S0022112004009942}, } @ARTICLE{KnaepenKassinosCarati2004, author = {. and .}, title = {Magnetohydrodynamic Turbulence at Moderate Magnetic Reynolds Number}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {513}, pages = {199--220}, note = {DOI: 10.1017/S0022112004000023}, } @ARTICLE{Girimaji2004, author = {Girimaji, .}, title = {A New Perspective on Realizability of Turbulence Models}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {512}, pages = {191--210}, note = {DOI: 10.1017/S0022112004009656}, } @ARTICLE{HodgesJensenRallison2004, author = {. and . and .}, title = {Sliding, Alipping and Rolling:The Sedimentation of a Viscous Drop Down a Gently Inclined Plane}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {512}, pages = {95--131}, note = {DOI: 10.1017/S0022112004009814}, } @ARTICLE{Zeyao2005, author = {.}, title = {Concatenation Algorithms for Parallel Numerical Simulation of Radiation Hydrodynamics Coupled with Neutron Transport}, journal = {International Journal of Parallel Programming}, year = {2005}, volume = {33}, pages = {57--71}, note = {DOI: 10.1007/s10766-004-1461-1}, } @ARTICLE{Carstensen2005, author = {.}, title = {A Unifying Theory of a Posteriori Finite Element Error Control}, journal = {Numerische Mathematik}, year = {2005}, note = {DOI: 10.1007/s00211-004-0577-y}, } @ARTICLE{CarstensenCausinSacco2005, author = { .}, title = {A Posteriori Dual--mixed Adaptive Finite Element Error Control for Lame and {Stokes} Equations}, journal = {Numerische Mathematik}, year = {2005}, note = {DOI: 10.1007/s00211-005-0616-3}, } @ARTICLE{SenguptaGanerwalDipankar2004, author = {. and .}, title = {High Accuracy Compact Schemes and Gibbs Phenomenon}, journal = {Journal of Scientific Computing}, year = {2004}, volume = {21}, pages = {253--268}, } @ARTICLE{SenguptaGuntakaDey2004, author = {. and .}, title = {{N}avier--{}--{S}tokes Solution by New Compact Scheme for Incompressible Flows}, journal = {Journal of Scientific Computing}, year = {2004}, volume = {21}, pages = {269--282}, } @ARTICLE{TruscottTurner2005, author = {. and .}, title = {A Heterogeneous Three--dimensional Computational Model for Wood Drying}, journal = {Applied Mathematical Modelling}, year = {2005}, volume = {29}, pages = {381--410}, note = {DOI: 10.1016/j.apm.2004.09.008}, } @ARTICLE{Nicolas2005, author = {.}, title = {Effects of Static Contact Angles on Inviscid Gravity--Capillary Waves}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1829111}, } @ARTICLE{MonnierWitomski2004, author = {. and .}, title = {Analysis of a Local Hydrodynamic Model with Marangoni Effect}, journal = {Journal of Scientific Computing}, year = {2004}, volume = {21}, pages = {369--403}, } @ARTICLE{AboubacarAguayoPhillipsPhillipsTamaddon-JahromiSnigerevWebster2005, author = { .}, title = {Modelling Pom--Pom Type Models with High--order Finite Volume Schemes}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {126}, pages = {207--220}, note = {DOI: 10.1016/j.jnnfm.2004.09.012}, } @ARTICLE{SirakovAinserHaoucheGuillet2005, author = { .}, title = {Three--dimensional Numerical Simulation of Viscoelastic Contraction Flows Using the Pom--Pom Differential Constitutive Model}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {126}, pages = {163--173}, note = {DOI: 10.1016/j.jnnfm.2004.08.013}, } @ARTICLE{CaraeniFuchs2005, author = {. and .}, title = {Compact Third--order Multidimensional Upwind Discretization for Steady and Unsteady Flow Simulations}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {419--441}, note = {DOI: 10.1016/j.compfluid.2004.03.002}, } @ARTICLE{FangGilbertXu2005, author = {. and . and .}, title = {Viscosity Dominated Flows with Temperature--dependent Thermal Conductivity}, journal = {Mathematical Methods in the applied Sciences}, year = {2005}, volume = {28}, pages = {1201--1217}, note = {DOI: 10.1002/mma616}, } @ARTICLE{CodinaSoto1999, author = {. and .}, title = {Finite Element Implementation of Two--Equation and Algebraic Stress Turbulence Models for Steady Incompressible Flows}, journal = {International Journal for Numerical Methods in Fluids}, year = {1999}, volume = {30}, pages = {309--333}, } @ARTICLE{NormanMiksis2005, author = {. and .}, title = {Dynamics of a Gas Bubble Rising in an Inclined Channel at Finite Reynolds Number}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, note = {DOI: 10.1063/1.1842220}, } @ARTICLE{TitarevToro2005, author = {. and .}, title = {WENO Schemes Based on Upwind and Centred TVD Fluxes}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {705--720}, note = {DOI: 10.1016/j.compfluid.2004.05.009}, } @ARTICLE{FedorchenkoAn-Bang2004, author = {. and .}, title = {The Formation and Dynamics of a Blob on Free and Wall Sheets Induced by a Drop Impact on Surfaces}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, note = {DOI: 10.1063/1.1793071}, } @ARTICLE{GuerraSuSilvaFreire2005, author = {. and . and .}, title = {The Near Wall Behaviour of an Impinging Jet}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {2829--2840}, note = {DOI: 10.1016/j.ijheatmasstransfer.2005.01.027}, } @ARTICLE{ByunLeeWookBaek2004, author = {. and .}, title = {Radiative Heat Transfer in Discretely Heated Irregular Geometry with an Absorbing, Emitting, and Anisotropically Scattering Medium Using Combined Monte--Carlo and Finite Volume Method}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {4195--4203}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.05.008}, } @ARTICLE{LovienoPassoniTordella2004, author = { .}, title = {A Large--Eddy Simulation Near--Wall Treatment}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, note = {DOI: 10.1063/1.1783371}, } @ARTICLE{KroegerNoelle2005, author = {. and .}, title = {Numerical Comparison of the Method of Transport to a Standard Scheme}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {541--560}, note = {DOI: 10.1016/j.compfluid.2003.12.002}, } @ARTICLE{YouZhaoCaiQiXu2004, author = { .}, title = {Experimental Investigation of Interparticle Collision Rate in Particulate Flow}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {1121--1138}, note = {DOI: 10.1016/j.ijmultiphaseflow.2004.05.009}, } @ARTICLE{Megias-AlguacilFeiglDresslerFischerWindhab2005, author = {.}, title = {Droplet Deformation Under Simple Shear Investigated by Experiment, Numerical simulation and Modeling}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {126}, pages = {153--161}, note = {DOI: 10.1016/j.jnnfm.2004.07.019}, } @ARTICLE{LucaMaffettoneGrecoSimeoneGuido2005, author = {. and .}, title = {Analysis of Start--up Dynamics of a Single Drop Through an Ellipsoidal Drop Model for Non--Newtonian Fluids}, journal = {J. Non-Newtonian Fluid Mech.}, year = {2005}, volume = {126}, pages = {145--151}, note = {DOI: 10.1016/j.jnnfm.2004.11.004}, } @ARTICLE{OezsoyRambaudStitouRiethmuller2005, author = { .}, title = {Vortex Characteristics in Laminar cavity at Very Low Mach Number}, journal = {Experiments in Fluids}, year = {2005}, volume = {38}, pages = {133--145}, note = {DOI: 10.1007/s00348-004-0845-8}, } @ARTICLE{CiarletCiarlet2004, author = {. and .}, title = {Another Approach to Linearized Elasticity and Korn`s Inequality}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {307--312}, note = {DOI: 10.1016/j.crma.2004.06.021}, } @ARTICLE{Laforest2004, author = {.}, title = {A Posteriori Error Estimate for Front--Tracking:Systems of Conservation Laws}, journal = {SIAM Journal on Mathematical Analysis}, year = {2004}, volume = {35}, pages = {1347--1370}, note = {DOI: 10.1137/S0036141002416870}, } @ARTICLE{KitawagaSugiyamaMurai2004, author = {. and . and .}, title = {Experimental Detection of Bubble--Bubble Interactions in a Wall--sliding Bubble Swarm}, journal = {International Journal of Multiphase Flow}, year = {2004}, volume = {30}, pages = {1213--1234}, note = {DOI: 10.1016/j.ijmultiphaseflow.2004.07.002}, } @ARTICLE{BorrellYoonGaryLeal2004, author = {. and . and .}, title = {Experimental Analysis of the Coalescence Process via Head--on Collisions in a Time--dependent Flow}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, note = {DOI: 10.1063/1.1795291}, } @ARTICLE{DelaleTunc2004, author = {. and .}, title = {A Bubble Fission Model for Collapsing Cavitation Bubbles}, journal = {Physics of Fluids}, year = {2004}, volume = {16}, note = {DOI: 10.1063/1.1808112}, } @ARTICLE{SleijpenWubs2003, author = {. and .}, title = {Exploiting Multilevel Preconditioning Techniques In Eigenvalue Computations}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, pages = {1249--1272}, note = {DOI: 10.1137/S1064827599361059}, } @ARTICLE{DobesRicchiutoDeconinck2005, author = {. and . and .}, title = {Implicit Space--time Residual Distribution Method for Unsteady Laminar Viscous Flow}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {593--615}, note = {DOI: 10.1016/j.compfluid.2003.09.007}, } @ARTICLE{ErikssonJohnsonLogg2003, author = {. and . and .}, title = {Explicit Time--stepping for Stiff ODES}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, pages = {1142--1157}, note = {DOI: 10.1137/S1064827502409626}, } @ARTICLE{ZhangShu2005, author = {. and .}, title = {An Analysis of and a Comparison Between the Discontinuous {G}alerkin and the Spectral Finite Volume Methods}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {581--592}, note = {DOI: 10.1016/j.compfluid.2003.05.006}, } @ARTICLE{CroisilleGreff2005, author = {. and .}, title = {An Efficient Box--scheme for Convection--Diffusion Equations with Sharp Contrast in the Diffusion Coefficients}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {461--489}, note = {DOI: 10.1016/j.compfluid.2003.12.003}, } @ARTICLE{ChoJunImKim2004, author = {. and .}, title = {An Improved Interface Element with Variable Nodes for Non--matching Finite Element Meshes}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {194}, pages = {3022--3046}, note = {DOI: 10.1016/j.cma.2004.08.002}, } @ARTICLE{EymardGalloueetHerbin2004, author = {. and . and .}, title = {A Finite Volume Scheme for Anisotropic Diffusion Problems}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2004}, volume = {339}, pages = {299--302}, note = {DOI: 10.1016/j.crma.2004.05.023}, } @ARTICLE{DoleanNatafRapin2005, author = {. and . and .}, title = {New Constructions of Domain Decomposition Methods for Systems of PDEs}, journal = {C. R., Math., Acad. Sci. Paris}, year = {2005}, volume = {340}, pages = {693--696}, note = {DOI: 10.1016/j.crma.2005.03.026}, } @ARTICLE{GleesonRocheWestGelb2004, author = {. and . and .}, title = {Modelling Annular Micromixers}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {64}, pages = {1294--1310}, note = {DOI: 10.1137/S0036139902420407}, } @ARTICLE{CrowdyMarshall2004, author = {. and .}, title = {Constructing Multiply Connected Quadrature Domains}, journal = {SIAM Journal on Applied Mathematics}, year = {2004}, volume = {64}, pages = {1334--1359}, note = {DOI: 10.1137/S0036139903438545}, } @ARTICLE{Dax2003a, author = {.}, title = {A Modified Iterative Refinement Scheme}, journal = {SIAM Journal on Scientific Computing}, year = {2003}, volume = {25}, pages = {1199--1213}, note = {DOI: 10.1137/S1064827502409055}, } @ARTICLE{XueSunMa2004, author = {. and . and .}, title = {On the Reconstruction of Three--dimensional Complex Geological Objects Using Delaunay Triangulation}, journal = {Future Generation Computer Systems}, year = {2004}, volume = {20}, pages = {1227--1234}, note = {DOI: 10.1016/j.future.2003.11.012}, } @ARTICLE{AndersonElliottPember2004, author = {. and . and .}, title = {An Arbitrary {L}agrangian--{E}ulerian Method with Adaptive Mesh Refinement for the Solution of the Euler Equations}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {598--617}, note = {DOI: 10.1016/j.jcp.2004.02.021}, } @ARTICLE{QiuShu2005, author = {. and .}, title = {Hermite WENO Schemes and their Applications as Limiters for {Runge}--{}--{Kutta} Discontinuous {Galerkin} Method II: Two Dimensional Case}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {642--663}, note = {DOI: 10.1016/j.compfluid.2004.05.005}, } @ARTICLE{CrouseillesDegondLemou2004, author = {. and . and .}, title = {A Hybrid Kinetic/Fluid Model for Solving the Gas Dynamics Boltzmann--BGK Equation}, journal = {J. Comput. Phys.}, year = {2004}, volume = {199}, pages = {776--808}, note = {DOI: 10.1016/j.jcp.2004.03.007}, } @ARTICLE{KoldaLewisTorczon2003, author = {. and . and .}, title = {Optimization by Direct Search: New Perspectives on Some Classical and Modern Methods}, journal = {SIAM Review}, year = {2003}, volume = {45}, number = {3}, pages = {385--482}, } @ARTICLE{CockburnKanschatSchoetzau2005, author = {. and . and .}, title = {The Local Discontinuous {Galerkin} Method for Linearized Incompressible Fluid Flow: A Review}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {491--506}, note = {DOI: 10.1016/j.compfluid.2003.08.005}, } @ARTICLE{XuLI2004, author = {. and .}, title = {Microchannel Flow in the Slip Regime: Gas--Kinetic BGK--Burnett Solutions}, journal = {Journal of Fluid Mechanics}, year = {2004}, volume = {513}, pages = {87--110}, note = {DOI: 10.1017/S0022112004009826}, } @ARTICLE{Utyuzhnikov2005, author = {.}, title = {Some New Approaches to Building and Implementation of Wall--functions for Modeling of Near--wall Turbulent Flows}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {771--784}, note = {DOI: 10.1016/j.compfluid.2004.07.004}, } @ARTICLE{Linss2004, author = {.}, title = {Error Expansion for a First--order Upwind Difference Scheme Applied to a Model Convection--Diffusion Problem}, journal = {IMA Journal of Numerical Analysis}, year = {2004}, volume = {24}, pages = {239--253}, } @ARTICLE{Chen2004b, author = {.}, title = {{N}avier--{}--{S}tokes Simulation of the Flow Field in Laterally Driven Microcomb Structures}, journal = {International Journal of Heat and Mass Transfer}, year = {2004}, volume = {47}, pages = {4205--4213}, note = {DOI: 10.1016/j.ijheatmasstransfer.2004.05.007}, } @ARTICLE{Badea2004, author = {.}, title = {On the {S}chwarz--{N}eumann Method with an Arbitrary Number of Domains}, journal = {IMA Journal of Numerical Analysis}, year = {2004}, volume = {24}, pages = {215--238}, } @ARTICLE{CorreHanssLerat2005, author = {. and . and .}, title = {A Rsidual--based Compact Scheme for the Unsteady Compressible {N}avier--{}--{S}tokes Equations}, journal = {Computers and Fluids}, year = {2005}, volume = {34}, pages = {561--580}, note = {DOI: 10.1016/j.compfluid.2003.08.003}, } @ARTICLE{ThangamSpeziale1992, author = {. and .}, title = {Turbulent Flow Past a Backward--Facing Step: A Critical Evaluation of Two--Equation Models}, journal = {AIAA Journal}, year = {1992}, volume = {30}, } @ARTICLE{Torczon1997, author = {.}, title = {On the Convergence of Pattern Search Algorithms}, journal = {SIAM Journal on Optimization}, year = {1997}, volume = {7}, number = {1}, pages = {1--25}, } @ARTICLE{Slawig2003, author = {.}, title = {A Formula for the Derivative with Respect to Domain Variations in {N}avier--{S}tokes Flow Based on an Embedding Domain Method}, journal = {SIAM Journal on Control and Optimization}, year = {2003}, volume = {42}, number = {2}, pages = {495--512}, } @ARTICLE{Gunzburger2000, author = {.}, title = {Adjoint Equation--Based Methods for Control Problems in Incompressible, Viscous Flows}, journal = {Flow, Turbulence and Combustion}, year = {2000}, volume = {65}, pages = {249--272}, } @ARTICLE{BurmanHansbo2004, author = {. and .}, title = {Edge stabilization for {G}akerkin approximations of convection--diffusion--reaction problems}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2004}, volume = {193}, pages = {1437--1453}, } @ARTICLE{vandenEshofSleijpenvanGijzen2005, author = {. and . and .}, title = {Relaxation Strategies for Nested {K}rylov Methods}, journal = {J. Comput. Appl. Math.}, year = {2005}, volume = {177}, pages = {347--365}, } @ARTICLE{CaoHuangRussell2002, author = {. and . and .}, title = {A Moving Mesh Method Based on the Geometric Conservation Law}, journal = {SIAM Journal on Scientific Computing}, year = {2002}, volume = {24}, number = {1}, pages = {118--142}, } @ARTICLE{GrajewskiHronTurek2005, author = {. and . and .}, title = {Dual weighted a posteriori error estimation for a new nonconforming linear finite element on quadrilaterals}, journal = {Applied Numerical Mathematics}, year = {2005}, volume = {54}, number = {3-4}, pages = {504--518}, month = aug, note = {Ausgabe dieser Zeitschrift im Lehrstuhlhandapparat vorhanden}, } @ARTICLE{TurekBertranOlivier2001, author = {. and . and .}, title = {Analysis of unsteady airfoils at low speeds}, journal = {Notes on Numerical Fluid Mechanics}, year = {2001}, } @ARTICLE{Madlik2004, author = {.}, title = {Numerical Simulations of the Three--dimensional Large Deformations in Solids and Fluids}, journal = {Theoretical and Computational Fluid Dynamics}, year = {2004}, month = aug, } @ARTICLE{KooKleinstreuer2005, author = {. and .}, title = {Laminar Nanofluid Flow in Micro {}--{}--{} Sinks}, journal = {Int. J. Heat Mass Transfer}, year = {2005}, volume = {48}, pages = {2652--2661}, note = {DOI:10.1016/j.ijheatmasstransfer.2005.01.029}, } @ARTICLE{LeBris2005, author = {, Claude.}, title = {Computational Chemistry from the Perspective of Numerical Analysis}, journal = {Cambridge University Press}, year = {2005}, pages = {363--444}, note = {DOI:10.1017/S096249290400025X}, } @ARTICLE{Stynes2005, author = {.}, title = {Steady--state Convection--diffusion Problems}, journal = {Acta Numerica}, year = {2005}, pages = {445--508}, note = {DOI:10.1017/S0962492904000261}, } @ARTICLE{RayMacDonald2005, author = {, .}, title = {Experimental Study and {} Finite Element Analysis of Simple {} X--{} and T--{} Branch Tube Hydroforming Processes}, journal = {ijmecsci}, year = {2005}, volume = {47}, pages = {1498--1518}, note = {DOI: 10.1016/j.ijmecsci.2005.06.007}, } @ARTICLE{LaiLiu2005, author = {. and .}, title = {Fast Direct Solver for the Biharmonic Equation on a Disk and its Application to Incompressible Flows}, journal = {J.amc}, year = {2005}, volume = {164}, pages = {679--695}, note = {DOI:10.1016/j.amc.2004.004.064}, } @ARTICLE{SumanKumar2005, author = {. and .}, title = {An Analytical Model for Fluid Flow and Heat Transfer in a Micro--{}--Heat Pipe of Polygonal Shape}, journal = {International Journal of Heat and Mass Transfer}, year = {2005}, volume = {48}, pages = {4498--4509}, note = {DOI:10.1016/j.ijheatmasstransfer.2005.05.001}, } @ARTICLE{AllenNaitoh2005, author = {. and .}, title = {Experimental Study of the Production of Vortex Rings Using a Variable Diameter Orifice}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, pages = {061701}, note = {DOI:10.1063/1.1921949}, } @ARTICLE{StoneStone2005, author = {. and .}, title = {Imaging and Quantifying Mixing in a Model Droplet Micromixer}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, pages = {063103}, note = {DOI:10.1063/1.1929547}, } @ARTICLE{LyubimovStraubeLyubimova2005, author = {Lyubimov, . and Straube, . and Lyubimova, .}, title = {Capture of Particles of Dust by Convective Flow}, journal = {Physics of Fluids}, year = {2005}, volume = {17}, pages = {063302}, note = {DOI:10.1063/1.1926187}, } @ARTICLE{KilicHallerNeishtadt2005, author = {. and .}, title = {Unsteady Fluid Flow Separation by the Method of Averaging}, journal = {Physics of Fluids}, year = {2005}, pages = {067104}, note = {DOI:10.1063/1.1924698}, } @ARTICLE{SolbergPapadopoulos2005, author = {Solberg, . and .}, title = {An Analysis of Dual Formulations for the Finite Element Solution of Two--body Contact Problems}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2005}, volume = {194}, pages = {2734--2780}, note = {DOI:10.1016/j.cma.2004.06.045}, } @ARTICLE{FangChengHuangLiuHanZengYuan2005, author = {. and . and . and Liu, Zhong--Xin. and Han, Bao--Ping. and Zeng, Rong--Hui. and Yuan, Wei--Kang.}, title = {Prediction of Flooding Velocity in a Trickle Bed}, journal = {International Journal of Multiphase Flow}, year = {2005}, volume = {31}, pages = {666--673}, note = {DOI:10.1016/j.ijmultiphaseflow.2004.12.002}, } @ARTICLE{Dolejsi1998, author = {.}, title = {Anisotropic Mesh Adaptation for Finite Volume and Finite Element Methods on Triangular Meshes}, journal = {Computing and Visualization in Science}, year = {1998}, pages = {165--178}, note = {\url{http:/ / www.karlin.mff.cuni.cz/ ~dolejsi/ articles/ ama.ps.gz}}, } @ARTICLE{FrancaMadureiraValentin2005, author = {. and Madureira, . and .}, title = {Towards Multiscale Functions: Enriching Finite Element Spaces with Local but Not Bubble--like Functions {}}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2005}, volume = {194}, pages = {3006--3021}, } @ARTICLE{Ravindran2005, author = {.}, title = {Real--time Computational Algorithm for Optimal control of an MHD Flow System}, journal = {SIAM J.Sci.Comput}, year = {2005}, volume = {26}, number = {4}, pages = {1369--1388}, note = {DOI:10.1137/S1064827502400534}, } @ARTICLE{Yserentant1983, author = {.}, title = {On the Convergence of Multi--level Mehtods for Strongly Nonuniform Families of Grids and Any Number of Smoothing Steps Per Level}, journal = {Computing}, year = {1983}, volume = {30}, } @ARTICLE{ToelkeKrafczykRank2002a, author = {. and Krafczyk, Manfred. and .}, title = {A Multigrid--solver for the Discrete {B}oltzmann--equation.}, journal = {Journal of Statistical Physics}, year = {2002}, volume = {107}, pages = {573--591}, } @ARTICLE{BankDoughlas1985, author = {. and Doughlas, .}, title = {Sharp Estimates for Multigrid Rates of Convergence with General Smoothing and Acceleration}, journal = {SIAM Journal on Numerical Analysis}, year = {1985}, volume = {22}, pages = {617--633}, } @ARTICLE{MazzaferroFerroGoldschmit2005, author = {Mazzaferro, Ferro, . and Goldschmit, .}, title = {An Algorithm for Rotating Axisymmetric {} Flows: Model, Validation {} and Industrial {} Applications}, journal = {International Journal for Numerical Methods in Fluids}, year = {2005}, volume = {48}, pages = {1101--1121}, note = {DOI:10.1002/fld.962}, } @ARTICLE{Cao005, author = {.}, title = {On the Error of {} Linear Interpolation and the Orientation, Aspect Ratio,}, journal = {SIAM Journal on Numerical Analysis}, year = {005}, volume = {43}, number = {1}, pages = {19--40}, note = {DOI:10.1137/s0036142903433492}, } @ARTICLE{Melenk2005, author = {.}, title = {hp--Interpolation of Nonsmooth Functions and an Application to hp--A posteriori Error Estimation}, journal = {SIAM Journal on Numerical Analysis}, year = {2005}, volume = {43}, number = {1}, pages = {127--155}, note = {10.1137/ S0036142903432930}, } @ARTICLE{Ainsworth2005, author = {.}, title = {robust Posteriori Error Estimation for Nonconforming Finite Element Approximation}, journal = {SIAM Journal on Numerical Analysis}, year = {2005}, volume = {42}, number = {6}, pages = {2320--2341}, note = {DOI:1.1137/S0036142903425112}, } @ARTICLE{Huang2005, author = {.}, title = {Measuring Mesh Qualities and Application to Variational Mesh Adaptation}, journal = {SIAM Journal on Scientific Computing}, year = {2005}, volume = {26}, number = {5}, pages = {1643--1666}, note = {DOI:10.1137/S1064827503429405}, } @ARTICLE{AuteriParoliniQuartapelle2002, author = {. and . and .}, title = {Numerical Investigation on the Stability of the Singular Driven Cavity Flow}, journal = {J. Comput. Phys.}, year = {2002}, volume = {183}, pages = {1--25}, note = {10.1006/jcph.2002.7145}, } @ARTICLE{ZhangChildressLibchaberShelley2000, author = { , .}, title = {Flexible Filaments in a Flowing Soap Film as a Model for One--dimensional Flags in a Two--dimensional Wind}, journal = {Nature}, year = {2000}, volume = {408}, pages = {835--839}, month = dec, } @ARTICLE{LeoperRapetti2005, author = {. and .}, title = {Numerical Solution of the Monge--Ampere Equation by a {N}ewton`s Algorithm}, journal = {J.crcma}, year = {2005}, volume = {1340}, pages = {319--324}, note = {DOI:10.1016/j.crma.2004.12.018}, } @ARTICLE{DobrzynskiFreyMohammadiPironneau2005, author = {. and . and .}, title = {Fast and accurate simulations of Air--cooled Structures}, journal = {Computer Methods in Applied Mechanics and Engineering}, year = {2005}, } @ARTICLE{MuchaRautmann2005, author = {. and .}, title = {Convergence of Rothe`s Scheme for the {N}avier--{S}tokes {} Equations with Slip Conditions in {2D} Domains}, journal = {ZAMM}, year = {2005}, pages = {1--10}, } @ARTICLE{DeganiSchaefer2005, author = {. and .}, title = {Gaussian Quadrature without Orthogonal Polynomials}, journal = {arXiv}, year = {2005}, note = {\url{http:/ / arxiv.org/ pdf/ math.NA/ 0506199}}, } @ARTICLE{HronOuazziTurek2005, author = {. and . and .}, title = {Finite element methods for the simulation of incompressible powder flow}, journal = {Communications in Numerical Methods in Engineering}, year = {2005}, number = {21}, pages = {581--596}, } @ARTICLE{TurekBeckerKilian2006, author = {. and . and .}, title = {Hardware--oriented numerics and concepts for {PDE} software}, journal = {Future Generation Computer Systems}, year = {2006}, volume = {22}, number = {1-2}, pages = {217--238}, } @ARTICLE{ChiouWangHutter2005, author = {. and .}, title = {Influence of Obstacles on Rapid Granular Flows}, journal = {Act Mech}, year = {2005}, volume = {175--122}, pages = {105--}, note = {DOI:10.1007/s00707-004-0208-9}, } @ARTICLE{StellaGiangiPagliaCasataSimoneGaudenzi2005, author = {. and . and . and . and . and .}, title = {A Numerical Simulation of {} Fluid {}--structure Interaction in Internal Flows}, journal = {Numerical Heat Transfer}, year = {2005}, volume = {47}, pages = {403--418}, note = {DOI:10.1080/10407790590919180}, } @ARTICLE{ArtoliHoekstraSloot2006, author = {. and . and .}, title = {Optimizing Lattice Boltzmann Simulations for Unsteady Fluids}, journal = {Computers and Fluids}, year = {2006}, volume = {35}, pages = {227--240}, note = {DOI:10.1016/j.compfluid.2004.12.002}, } @ARTICLE{MeiYuShyy2002, author = {. and . and .}, title = {Force Evaluation in the Lattice Boltzmann Method Involving Curved Geometry {}}, journal = {Physical Review}, year = {2002}, volume = {65}, pages = {041203(1)--041203(14)}, note = {DOI:10.1103/PhysRevE.65.041203}, } @ARTICLE{DengXueZhangZhong2005, author = {. and . and . and .}, title = {An Inexact Newton Method Derived from Efficiency Analysis}, journal = {Jglobopt}, year = {2005}, volume = {31}, pages = {287--315}, } @ARTICLE{BlanesCasas2005, author = {. and .}, title = {On the Necessity of Negative Coefficients for Operator Splitting Schemes of Order Higher than Two}, journal = {Applied Numerical Mathematics}, year = {2005}, volume = {54}, pages = {23--37}, } @ARTICLE{LiaoQiTam2005, author = {. and . and .}, title = {A {} {G}radient--based Continuous Method for Large--scale {} Optimization Problems}, journal = {Acta Materialia}, year = {2005}, volume = {31}, pages = {271--286}, } @ARTICLE{KumarVarshneySharma2005, author = {. and . and .}, title = {Computational Technique for Flow in Blood Vessels {} with Porous Effects}, journal = {Appl.Math.Mech.}, year = {2005}, volume = {26}, number = {1}, pages = {63--72}, } @ARTICLE{KumarVarshneySharma2005a, author = { . and .}, title = {Performance Modeling and Analysis of Blood Flow in Elastic Arteries {}}, journal = {Appl.Math.Mech.}, year = {2005}, volume = {26}, number = {3}, pages = {345--354}, month = mar, } @ARTICLE{LingKansa2004, author = {. and .}, title = {Preconditioning for Radial Basis Functions {} with Domain Decomposition Methods}, journal = {j.mcm.}, year = {2004}, volume = {40}, pages = {1413--1427}, } @ARTICLE{Vil`danov2004, author = {.}, title = {Integrable {} Boundary Value Problem for the Boussinesq Equation {}}, journal = {Theoretical and Mathematical Physics}, year = {2004}, volume = {141}, number = {2}, pages = {1494--1508}, } @ARTICLE{SpeetjensClercx2005, author = {. and .}, title = {A Spectral Solver for the {N}avier--{}--{S}tokes Equations in the Velocity {}--{}--Vorticity {} Formulation {}}, journal = {International Journal of Computational Fluid Dynamics}, year = {2005}, volume = {19}, number = {3}, pages = {191--209}, } @ARTICLE{TezukaMatsumotoSuzukiMatsubara2005, author = {. and . and . and .}, title = {A Platform for Parallel {CFD} {FEM} Computations}, journal = {International Journal of Computational Fluid Dynamics}, year = {2005}, volume = {19}, number = {3}, pages = {235--242}, } @ARTICLE{BrandesSchwambornGerndtJeitnerKerekuSchulzBrunstNagelNeumannMueller--PfefferkornTrenklerKarlTaoHoppe2005, author = {. and M\"{u}ller--. and . and . and . and .}, title = {Monitoring Cache Behaviour on Parallel SMP Architectures and Related Programmming Tools {} {}}, journal = {Future Generation Computer Systems}, year = {2005}, volume = {21}, pages = {1298--1311}, } @ARTICLE{GieringKaminskiSlawig2005, author = {.}, title = {Generating efficient Derivative Code with TAF Adjoint and Tangent Linear Euler Flow Around an Airfoil}, journal = {Future Generation Computer Systems}, year = {2005}, volume = {21}, pages = {1345--1355}, } @BOOK{GrossmannRoos1994, author = { .}, title = {Numerik partieller Differentialgleichungen}, publisher = {Oxford, Clarendon Press}, year = {1994}, note = {2.Auflage }, } @BOOK{Schwarz1997, author = {.}, title = {Numerische Mathematik}, publisher = {Oxford, Clarendon Press}, year = {1997}, } @BOOK{NozickaGuddatHollatz1972, author = {. and . and .}, title = {Theorie der linearen Optimierung}, publisher = {Akademie Verlag, Berlin}, year = {1972}, } @BOOK{Nitsche1968, author = {.}, title = {Praktische Mathematik}, publisher = {Bibliograph. Inst., Mannheim}, year = {1968}, } @BOOK{ForsytheWasow1967, author = {. and .}, title = {Finite--difference methods for partial differential equations. 6. pr.}, publisher = {John Wiley \& Sons Ltd.}, year = {1967}, } @BOOK{Wasow1965, author = {.}, title = {Asymptotic expansions for ordinary differential equations}, publisher = {John Wiley \& Sons Ltd.}, year = {1965}, } @BOOK{Vogel1970, author = {.}, title = {Lineares Optimieren, 2. Auflage}, publisher = {Geest und Portig, Leipzig}, year = {1970}, } @BOOK{Schwarz1984, author = {.}, title = {Methode der finiten Elemente,2. Auflage}, publisher = {Oxford, Clarendon Press}, year = {1984}, } @BOOK{WernerSchaback1972, author = {. and .}, title = {Praktische Mathematik, Bd. 2}, publisher = {Springer, Berlin}, year = {1972}, } @BOOK{Werner1970, author = {.}, title = {Praktische Mathematik,Bd. 1}, publisher = {Springer, Berlin}, year = {1970}, } @BOOK{MangoldtKnopp196869, author = {. and .}, title = {Einf\"{u}hrung in die h\"{o}here Mathematik,Band 1, 13. Auflage}, publisher = {Hirzel, Stuttgart}, year = {1968/69}, } @BOOK{MangoldtKnopp196869a, author = {. and .}, title = {Einf\"{u}hrung in die h\"{o}here Mathematik,Band 2, 13. Auflage}, publisher = {Hirzel, Stuttgart}, year = {1968/69}, } @BOOK{MangoldtKnopp1972, author = {. and .}, title = {Einf\"{u}hrung in die h\"{o}here Mathematik,Band 3, 13. Auflage}, publisher = {Hirzel, Stuttgart}, year = {1972}, } @BOOK{MangoldtKnopp1972a, author = { .}, title = {Einf\"{u}hrung in die h\"{o}here Mathematik,Band 4, 1. Auflage}, publisher = {Hirzel, Stuttgart}, year = {1972}, } @BOOK{Michlin1969, author = {.}, title = {Numerische Realisierung von Variationsmethoden}, publisher = {Akademie Verlag, Berlin}, year = {1969}, } @BOOK{Minorsky1962, author = {.}, title = {Nonlinear oscillations}, publisher = {N.J.: van Nostrand,, Princeton}, year = {1962}, } @BOOK{KuenziMuellerNievergelt1968, author = {Kuenzi and M\"{u}ller and Nievergelt}, title = {Einf\"{u}hrungskurs in die dynamische Programmierung}, publisher = {Springer, Berlin}, year = {1968}, } @BOOK{Hartree1964, author = {.}, title = {Numerical analysis,2. ed.}, publisher = {Oxford, Clarendon Press}, year = {1964}, } @BOOK{Apostol1957, author = {.}, title = {Mathematical analysis}, publisher = {Addison-Wesley}, year = {1957}, } @BOOK{AnsorgeHass1970, author = { .}, title = {Konvergenz von Differenzenverfahren f\"{u}r lineare und nichtlineare Anfangswertaufgaben}, publisher = {Springer, Berlin}, year = {1970}, } @BOOK{KuenziSoeiTienTan1966, author = {Kuenzi and }, title = {Lineare Optimierung gro"ser Systeme}, publisher = {Springer, Berlin}, year = {1966}, } @BOOK{Schwartz1969, author = {.}, title = {Nonlinear functional analysis}, publisher = {Gordon und Breach, New York}, year = {1969}, } @BOOK{Ostrowski196062, author = {.}, title = {Vorlesungen \"{u}ber Differential--{} und Integralrechnung,Band 1}, publisher = {Birkh\"{a}user}, year = {1960-62}, } @BOOK{Ostrowski196062a, author = {.}, title = {Vorlesungen \"{u}ber Differential--{} und Integralrechnung,Band 2}, publisher = {Birkh\"{a}user}, year = {1960-62}, } @BOOK{Ostrowski196062c, author = {.}, title = {Vorlesungen \"{u}ber Differential--{} und Integralrechnung,Band 3}, publisher = {Birkh\"{a}user}, year = {1960-62}, } @BOOK{Rudin1953, author = {.}, title = {Principles of mathematical analysis}, publisher = {McGraw Hill}, year = {1953}, } @BOOK{Richtmeyer1967, author = {.}, title = {Difference methods for initial--value problems,2. ed.}, publisher = {Interscience Publ., New York}, year = {1967}, } @BOOK{OrtegaRheinboldt1970, author = {. and .}, title = {Iterative solution of nonlinear equations in several variables}, publisher = {Academic Press, New York}, year = {1970}, } @BOOK{Reichardt1968, author = {.}, title = {Vorlesungen \"{u}ber Vektor--{} und Tensorrechnung,2. Auflage}, publisher = {Dt. Verlag der Wissenschaften, Berlin}, year = {1968}, } @BOOK{LjusternikSobolev1968, author = { Sobolev}, title = {Elemente der Funktionalanalysis,4. Auflage}, publisher = {Akademie Verlag, Berlin}, year = {1968}, } @BOOK{KuenziTzschachZehnder1967, author = {. and Zehnder}, title = {Numerische Methoden der mathematischen Optimierung mit Algol--{} und Fortran--Programmen}, publisher = {Oxford, Clarendon Press}, year = {1967}, } @BOOK{Landau1930, author = {.}, title = {Grundlagen der Analysis}, publisher = {Akademie-Verlag-Gesellschaft, Leipzig}, year = {1930}, } @BOOK{Haemmerlin1970, author = {.}, title = {Numerische Mathematik,Band 1}, publisher = {Bibliograph. Inst., Mannheim}, year = {1970}, } @BOOK{Guber1967, author = {.}, title = {Lineare Algebra und analytische Geometrie, Band 1}, publisher = {Merkel, Erlangen}, year = {1967}, } @BOOK{Guber1968, author = {.}, title = {Lineare Algebra und analytische Geometrie, Band 2}, publisher = {Merkel, Erlangen}, year = {1968}, } @BOOK{Graeub1967, author = {.}, title = {Linear algebra, 3. ed.}, publisher = {Springer, Berlin}, year = {1967}, } @BOOK{GrauertLiebFischer1967, author = { .}, title = {Differential--{} und Integralrechnung, Band 1}, publisher = {Springer, Berlin}, year = {1967}, } @BOOK{GrauertLiebFischer1967a, author = { Lieb and .}, title = {Differential--{} und Integralrechnung, Band 2}, publisher = {Springer, Berlin}, year = {1967}, } @BOOK{GrauertLiebFischer1967b, author = { Lieb and .}, title = {Differential--{} und Integralrechnung, Band 3}, publisher = {Springer, Berlin}, year = {1967}, } @BOOK{Courant1963, author = {.}, title = {Vorlesungen \"{u}ber Differential--{} und Integralrechnung, Band 1}, publisher = {Springer, Berlin}, year = {1963}, } @BOOK{Courant1969, author = {.}, title = {Vorlesungen \"{u}ber Differential--{} und Integralrechnung, Band 2}, publisher = {Springer, Berlin}, year = {1969}, } @BOOK{Kaufmann1967, author = {.}, title = {Graphs, dynamic programming and finite games}, publisher = {Academic Press, New York}, year = {1967}, } @BOOK{Joergens1970, author = {J\"{o}.}, title = {Lineare Integraloperatoren}, publisher = {Oxford, Clarendon Press}, year = {1970}, } @BOOK{Hildebrand1956, author = {.}, title = {Introduction to numerical analysis}, publisher = {McGraw Hill}, year = {1956}, } @BOOK{Heuser1975, author = {.}, title = {Funktionalanalysis}, publisher = {Oxford, Clarendon Press}, year = {1975}, } @BOOK{Henrici1962, author = {.}, title = {Discrete variable methods in ordinary differential equations}, publisher = {John Wiley \& Sons Ltd.}, year = {1962}, } @BOOK{Sander1973, author = {.}, title = {Dualit\"{a}t bei Optimierungsaufgaben}, publisher = {Oldenbourg, M\"{u}nchen}, year = {1973}, } @BOOK{Rudin1974, author = {.}, title = {Real and complex analysis, 2. ed.}, publisher = {McGraw Hill}, year = {1974}, } @BOOK{Koethe1966, author = {.}, title = {Topologische lineare R\"{a}ume, Band 1, 2. Auflage}, publisher = {Springer, Berlin}, year = {1966}, } @BOOK{Knoedel1969, author = {Kn\"{o}.}, title = {Graphentheoretische Methoden und ihre Anwendungen}, publisher = {Springer, Berlin}, year = {1969}, } @BOOK{Keller1968, author = {.}, title = {Numerical methods for two--point boundary value problems}, publisher = {Waltham, Mass.: Blaisdell}, year = {1968}, } @BOOK{KuenziOettli1969, author = {K\"{. and .}, title = {Nichtlineare Optimierung}, publisher = {Springer, Berlin}, year = {1969}, } @BOOK{Boehmer1976, author = {B\"{o}.}, title = {Spline functions}, publisher = {Springer, Berlin}, year = {1976}, } @BOOK{BellmannCooke1963, author = {. and .}, title = {Differential--difference equations}, publisher = {Academic Press, New York}, year = {1963}, } @BOOK{Bauer1966, author = {.}, title = {Differential--{} und Integralrechnung, Band 1}, publisher = {Merkel, Erlangen}, year = {1966}, } @BOOK{Bauer1966a, author = {.}, title = {Differential--{} und Integralrechnung, Band 2}, publisher = {Merkel, Erlangen}, year = {1966}, } @BOOK{TrefethenBau1997, author = {. and .}, title = {Numerical Linear Algebra}, publisher = {SIAM}, year = {1997}, note = {ISBN 0-89871-361-7}, } @BOOK{Demmel1997, author = {.}, title = {Applied Numerical Linear Algebra}, publisher = {SIAM}, year = {1997}, note = {ISBN 0-89871-389-7}, } @BOOK{Schoenauer1998, author = {Sch\"{.}, title = {Scientific Supercomputing,Architecture and Use of Shared on Distributed Memory Parallel Computers}, publisher = {Self-Edition by Wille Sch\"{o}nauer}, year = {1998}, note = {ISBN 3-00-005484-7}, } @BOOK{OxfordDuden1998, author = {}, title = {Oxford Duden}, publisher = {Oxford University Press}, year = {1998}, } @BOOK{Golub1996, author = {.}, title = {Scientific Computing,Eine Einf\"{u}hrung in das wissenschaftliche Rechnen und die Parallele Numerik}, publisher = {, Stuttgart, Leipzig}, year = {1996}, note = {ISBN 3-519-02969-3}, } @BOOK{Joseph1990, author = {.}, title = {Fluid Dynamics of Viscoelastic Liquids}, publisher = {Springer, Berlin}, year = {1990}, } @BOOK{VDIGesellschaftEnergietechnik1996, author = {VDI-Gesellschaft Energietechnik}, title = {Ventilatoren im industriellen Einsatz}, publisher = {VDI-Verlag, D\"{u}sseldorf}, year = {1996}, note = {ISBN 3-18-091249-9}, } @BOOK{DeBoorHoelligRiemenschneider1993, author = {DeBoor and . and Riemenschneider}, title = {Box splines}, publisher = {Springer, Berlin}, year = {1993}, } @BOOK{DeVoreLorentz1993, author = {. and .}, title = {Constructive approximation}, publisher = {Springer, Berlin}, year = {1993}, } @BOOK{Hackbusch1996, author = {.}, title = {Theorie und Numerik elliptischer Differentialgleichungen, 2. Auflage}, publisher = {Oxford, Clarendon Press}, year = {1996}, } @BOOK{Schumaker1993, author = {.}, title = {Spline functions}, publisher = {Krieger, Malabar, Florida}, year = {1993}, } @BOOK{Skrypnik1986, author = {.}, title = {Nonlinear elliptic boundary value problems}, publisher = {B. }, year = {1986}, } @BOOK{StuartHumphries1996, author = {. and .}, title = {Dynamical systems and numerical analysis}, publisher = {Cambridge University Press}, year = {1996}, } @BOOK{Braun1993, author = {Braun}, title = {Differential Equations and Their Applications}, publisher = {nn}, year = {1993}, note = {ISBN 3-540-97894-1}, } @BOOK{CuckerShub1997, author = {Cucker and Shub}, title = {Foundations of Computational Mathematics}, publisher = {nn}, year = {1997}, note = {ISBN 3-540-61647-0}, } @BOOK{EgorovKomechShublin1999, author = {. and Shublin}, title = {Elements of the Modern Theory of Partial Differential Equations}, publisher = {nn}, year = {1999}, note = {ISBN 3-540-65377-5}, } @BOOK{KloedenPlaten1999, author = {Kloeden and Platen}, title = {Numerical Solution of Stochastic Differential Equations}, publisher = {nn}, year = {1999}, note = {ISBN 3-540-54062-8}, } @BOOK{Murray1993, author = {.}, title = {Mathematical Biology}, publisher = {nn}, year = {1993}, note = {ISBN 3-540-57204-X}, } @BOOK{RoosStynesTobiska1996, author = {. and . and .}, title = {Numerical Methods for Singularly Pertubed Differential Equations}, publisher = {nn}, year = {1996}, note = {ISBN 3-540-60718-8}, } @BOOK{Acheson1998, author = {.}, title = {Elementary fluid dynamics}, publisher = {Oxford, Clarendon Press}, year = {1998}, note = {ISBN 0-19-859679-0}, } @BOOK{Chemin1998, author = {.}, title = {Perfect incompressible fluids}, publisher = {Oxford, Clarendon Press}, year = {1998}, note = {ISBN 0-19-850397-0}, } @BOOK{Schwab1998, author = {.}, title = {p--{} and hp--finite element methods}, publisher = {Oxford, Clarendon Press}, year = {1998}, note = {ISBN 0-19-850390-3}, } @BOOK{QuarteroniValli1999, author = {. and .}, title = {Domain decomposition methods for partial differential equations}, publisher = {Oxford, Clarendon Press}, year = {1999}, note = {ISBN 0-19-850178-1}, } @BOOK{Quartapelle1993, author = {.}, title = {Numerical Solution of the Incompressible {N}avier--{}--{S}tokes Equations}, publisher = {Birkh\"{a}user}, year = {1993}, note = {ISBN 3-7643-2935-1}, } @BOOK{QuarteroniValli1997, author = {. and .}, title = {Numerical Approximation of Partial Differential Equations}, publisher = {Springer, Berlin}, year = {1997}, note = {ISBN 3-540-57111-6}, } @BOOK{KarniadakisSherwin1999, author = {. and .}, title = {Spectral hp Element Methods for {CFD}}, publisher = {Oxford University Press}, year = {1999}, note = {ISBN 0-19-510226-6}, } @BOOK{Maschke2000, author = {Maschke}, title = {Hallo iMac}, publisher = {nn}, year = {2000}, note = {ISBN 3-908489-35-0}, } @BOOK{Schuler2000, author = {Schuler}, title = {Photoshop 5.5}, publisher = {nn}, year = {2000}, note = {ISBN 3-908490-13-8}, } @BOOK{Carey1997, author = {.}, title = {Computational Grids: Generation, Adaptation, and Solution Strategies}, publisher = {Taylor und Francis}, year = {1997}, } @BOOK{BeitzGrote1997, author = {. and .}, title = {Dubbel {}--{} Taschenbuch f\"{u}r den Maschinenbau,19. Auflage}, publisher = {Springer, Berlin}, year = {1997}, note = {ISBN 3-540-62467-8}, } @BOOK{Finlayson1992, author = {.}, title = {Numerical Methods for Problems with Moving Fronts}, publisher = {Ravenna Park Publishing, Inc., Seattle, Washington USA}, year = {1992}, note = {ISBN 0-9631765-0-1}, } @BOOK{QuarteroniSaccoSaleri2000, author = {. and . and .}, title = {Numerical Mathematics, Texts in Applied Mathematics 37}, publisher = {Springer, Berlin}, year = {2000}, note = {ISBN 0-387-98959-5}, } @BOOK{Renardy2000, author = {.}, title = {Mathematical Analysis of Viscoelastic Flows CBMS--NSF Regional Conference Series in Applied Mathematics}, publisher = {SIAM}, year = {2000}, note = {ISBN 0-89871-457-5}, } @BOOK{ChenEwingShi2000, author = {. and . and .}, title = {Numerical Treatment of Multiphase Flows in Porous Media/Proceedings, Beijing, China 1999}, publisher = {Springer, Berlin}, year = {2000}, note = {ISBN 3-540-67566-3}, } @BOOK{Wesseling2001, author = {.}, title = {Principles of computational fluid dynamics}, publisher = {Springer, Berlin}, year = {2001}, note = {ISBN 3-540-67853-0}, } @BOOK{Liseikin1999, author = {.}, title = {Grid Generation Methods}, publisher = {Springer, Berlin}, year = {1999}, note = {ISBN 3-540-65686-3}, } @BOOK{Lang2000, author = {.}, title = {Adaptive Multilevel Solution of Nonlinear Parabolic {PDE} Systems}, publisher = {Springer, Berlin}, year = {2000}, note = {ISBN 3-540-67900-6}, } @BOOK{Sethian1999, author = {.}, title = {Level Set Methods and Fast Marching Methods}, publisher = {Cambridge University Press}, year = {1999}, note = {ISBN 0-521-64557-3}, } @BOOK{DoneaRoigHuerta1998, author = {. and . and .}, title = {High--Order Accurate Time--Stepping Schemes for Convection--Diffusion Problems}, publisher = {International Center for Numerical Methods in Engineering, Barcelona}, year = {1998}, } @BOOK{NikolaidisPitas2001, author = {. and .}, title = {3--D Image Processing Algorithms}, publisher = {John Wiley \& Sons Ltd.}, year = {2001}, note = {ISBN 0-471-37736-8}, } @BOOK{Pitas2000, author = {.}, title = {Digital Image Processing Algorithms and Applications}, publisher = {John Wiley \& Sons Ltd.}, year = {2000}, note = {ISBN 0-471-37739-2}, } @BOOK{ReecePreston2000, author = {. and .}, title = {Finite Element Methods in Electrical Power Engineering}, publisher = {Oxford University Press}, year = {2000}, note = {ISBN 0198565046}, } @BOOK{MohammadiPironneau2001, author = {. and .}, title = {Applied Shape Optimization for Fluids}, publisher = {Oxford University Press}, year = {2001}, note = {ISBN 0198507437}, } @BOOK{Larson1999, author = {.}, title = {Structure and rheology of complex fluids}, publisher = {Oxford University Press}, year = {1999}, note = {ISBN 0-19-512197-X}, } @BOOK{Hirsch2000, author = {.}, title = {Numerical computation of internal and external flows, Band 1}, publisher = {John Wiley \& Sons Ltd.}, year = {2000}, note = {ISBN 0-471-92385-0}, } @BOOK{Hirsch2000a, author = {.}, title = {Numerical computation of internal and external flows, Band 2}, publisher = {John Wiley \& Sons Ltd.}, year = {2000}, note = {ISBN 0-471-92385-0}, } @BOOK{Born2000, author = {.}, title = {Windows 2000 Professional}, publisher = {Microsoft Press}, year = {2000}, note = {ISBN 3-86063-134-9}, } @BOOK{Duden2001, author = {Duden}, title = {Stilw\"{o}rterbuch}, publisher = {Dudenverlag}, year = {2001}, note = {ISBN 3-411-04028-9}, } @BOOK{Duden2001a, author = {Duden}, title = {Deutsches Universalw\"{o}rterbuch}, publisher = {Dudenverlag}, year = {2001}, note = {ISBN 3-411-05504-9}, } @BOOK{BuenningKrauseLarisch2001, author = {. and . and .}, title = {Windows 2000 im Netzwerkeinsatz}, publisher = {Hanser, M\"{u}nchen}, year = {2001}, note = {ISBN 3-446-21498-4}, } @BOOK{Handschuch1999, author = {.}, title = {Solaris 7 {}--{} Systemadministration,2., vollst. \"{u}berarb. Auflage}, publisher = {Springer, Berlin}, year = {1999}, note = {ISBN 3-540-63421-5}, } @BOOK{Lienemann2001, author = {.}, title = {TCP/IP--Praxis}, publisher = {Heise, Hannover}, year = {2001}, note = {ISBN 3-88229-184-2}, } @BOOK{Siemers1999, author = {.}, title = {Prozessorbau}, publisher = {Hanser, M\"{u}nchen}, year = {1999}, note = {ISBN 3-446-19330-8}, } @BOOK{Loehner2001, author = {.}, title = {Applied {CFD} Techniques {}--{} An Introduction based on Finite Element Methods}, publisher = {John Wiley \& Sons Ltd.}, year = {2001}, note = {ISBN 0471-498432}, } @BOOK{Buyya1999, author = {.}, title = {High Performance Cluster Computing:Programming and Applications Volume 2}, publisher = {Prentice Hall}, year = {1999}, note = {ISBN 0-13-013785-5}, } @BOOK{Buyya1999a, author = {.}, title = {High Performance Cluster Computing:Architectures and Systems Volume 1}, publisher = {Prentice Hall}, year = {1999}, note = {ISBN 0-13-013784-7}, } @BOOK{Fleischer2001, author = {.}, title = {Detaillierte Modellierung von Gas--Fl\"{u}ssigkeits--Reaktoren/Fortschritt--Berichte VDI, Reihe 3, Nr. 691}, publisher = {VDI-Verlag, D\"{u}sseldorf}, year = {2001}, note = {ISBN 3-18-369103-5}, } @BOOK{Sonar2001, author = {.}, title = {Angewandte Mathematik, Modellbildung und Informatik}, publisher = {Vieweg}, year = {2001}, note = {ISBN 3-528-03179-4}, } @BOOK{Herrmann2000, author = {.}, title = {Rechnerarchitektur}, publisher = {Vieweg}, year = {2000}, note = {ISBN 3-528-15598-1}, } @BOOK{Heun2000, author = {.}, title = {Grundlegende Algorithmen}, publisher = {Vieweg}, year = {2000}, note = {ISBN 3-528-03140-9}, } @BOOK{Herzberger1998, author = {.}, title = {\"{U}bungsbuch zur Numerischen Mathematik}, publisher = {Vieweg}, year = {1998}, note = {ISBN 3-528-06948-1}, } @BOOK{Davis2001, author = {.}, title = {Theory of Solidification}, publisher = {Cambridge University Press}, year = {2001}, note = {ISBN 0521650801}, } @BOOK{KacurMikulaSevcovic2001, author = {. and . and .}, title = {ACTA MATHEMATICA UNIVERSITATIS COMENIANAE, Volume LXX, Number 1}, publisher = {Comenius University Press, Slovakia}, year = {2001}, note = {ISBN 0862-9544 ISSN}, } @BOOK{Schroeder1990, author = {.}, title = {Mathematik im Reich der T\"{o}ne}, publisher = {Oxford, Clarendon Press}, year = {1990}, note = {ISBN 3-322-00476-7}, } @BOOK{GoossensRahtz2000, author = {. and .}, title = {Mit LATEX ins Web}, publisher = {Addison-Wesley}, year = {2000}, note = {ISBN 3-8273-1629-4}, } @BOOK{Toro1999, author = {.}, title = {Riemann Solvers and Numerical Methods for Fluid Dynamics. A practical introduction}, publisher = {Springer, Berlin}, year = {1999}, note = {ISBN 3-540-65966-8}, } @BOOK{PONS2001, author = {PONS}, title = {PONS Gro"sw\"{o}rterbuch f\"{u}r Experten und Universit\"{a}t Hauptband}, publisher = {Klettverlag}, year = {2001}, note = {ISBN 3-12-517159-8}, } @BOOK{PONS2001a, author = {PONS}, title = {PONS Gro"sw\"{o}rterbuch f\"{u}r Experten und Universit\"{a}t Beilage/Beibuch}, publisher = {Klettverlag}, year = {2001}, note = {ISBN 3-12-517158-X}, } @BOOK{Laney1998, author = {.}, title = {Computational Gasdynamics}, publisher = {Cambridge University Press}, year = {1998}, note = {ISBN 0-521-62558-0}, } @BOOK{NattererWuebbeling2001, author = {. and .}, title = {Mathematical Methods in Image Reconstruction}, publisher = {SIAM}, year = {2001}, note = {ISBN 0-89871-472-9}, } @BOOK{GodlewskiRaviart1996, author = {. and .}, title = {Numerical Approximation of Hyperbolic Systems of Conservation Laws}, publisher = {Springer, Berlin}, year = {1996}, note = {ISBN 0-387-94529-6}, } @BOOK{PieglTiller1997, author = {. and .}, title = {The NURBS Book}, publisher = {Springer, Berlin}, year = {1997}, note = {ISBN 3-540-61545-8}, } @BOOK{Seydel2002, author = {.}, title = {Tools for Computational Finance}, publisher = {Springer, Berlin}, year = {2002}, note = {ISBN 3-540-43609}, } @BOOK{NorburyRoulstone2002, author = {. and Roulstone, I.}, title = {Large--Scale Atmosphere--Ocean Dynamics I,Analytical methods and numerical models}, publisher = {Cambridge University Press}, year = {2002}, note = {ISBN 052180681 X}, } @BOOK{NorburyRoulstone2002a, author = {. and .}, title = {Large--Scale Atmosphere--Ocean Dynamics II,Geometric methods and models}, publisher = {Cambridge University Press}, year = {2002}, note = {ISBN 0521807573}, } @BOOK{PeriauxJolyPironneauOnate2001, author = { . and . and .}, title = {Innovative Tools for Scientific Computation in Aeronautical Engineering}, publisher = {CIMNE, Barcelona}, year = {2001}, note = {ISBN 84-89925-78-X}, } @BOOK{Brockhaus2002, author = {Brockhaus}, title = {Der Brockhaus {}--{} Naturwissenschaft und Technik 3 Bde und CD}, publisher = {Brockhaus/Spektrum}, year = {2002}, note = {ISBN 3-7653-1065-4}, } @BOOK{Capasso2002, author = {.}, title = {Mathematical Modelling for Polymer Processing}, publisher = {Springer, Berlin}, year = {2002}, } @BOOK{GoodmaoORourke1997, author = {Goodmao and ORourke}, title = {Handbook of Discrete and Computational Geometry}, publisher = {nn}, year = {1997}, note = {ISBN 0-8493-8524-5}, } @BOOK{Boussinesq1903, author = {.}, title = {Theorie Analytique de la chaleur}, publisher = {Gauthier-Villars}, year = {1903}, note = {Paris }, } @BOOK{GiraultRaviart1986, author = {. and .}, title = {Finite Element Methods for {N}avier--{}--{S}tokes Equations}, publisher = {Springer, Berlin}, year = {1986}, note = {Berlin-Heidelberg }, } @BOOK{Hackbusch1985, author = {.}, title = {Multi--Grid Methods and Applications}, publisher = {Springer, Berlin}, year = {1985}, note = {ISBN 3-540-12761-5}, } @BOOK{BungartzGriebelZengerJuni2002, author = {. and . and .}, editor = {M\"{o}. and . and .}, title = {Einf\"{u}hrung in die Computergraphik}, publisher = {Vieweg}, year = {Juni 2002}, volume = {2. \"{u}berarbeitete und erweiterte Auflage}, edition = {Mathematische Grundlagen der Informatik}, note = {Lehrstuhlhandapparat, ISBN 3-528-16769-6}, } @BOOK{MeisterStruckmeier2002, author = {. and .}, title = {Hyperbolic Partial Differential Equations}, publisher = {Vieweg}, year = {2002}, note = {Lehrstuhlhandapparat, ISBN 3-528-03188-3}, } @BOOK{NielsonHagenMueller1993, author = {. and . and M\"{u}.}, title = {Focus on Scientific Visualization}, publisher = {Springer, Berlin}, year = {1993}, } @BOOK{HagenMuellerNielson1997, author = {. and M\"{u}. and .}, title = {Scientific Visualization: Overviews,Methodologies, Techniques}, publisher = {IEEE Computer Society Press}, year = {1997}, } @BOOK{Balzert2000, author = {.}, title = {Lehrbuch der Software--Technik}, publisher = {Spektrum akad. Verlag}, year = {2000}, } @BOOK{BeckerRannacher1995, author = {. and .}, title = {Finite Element Solution of the Incompressible {N}avier--{}--{S}tokes Equations on Anisotropically Refined Meshes}, publisher = {Vieweg}, year = {1995}, series = {Notes Numer. Fluid Mech. 49, 52-62}, } @BOOK{Turek1999h, author = {.}, title = {Efficient Solvers for Incompressible Flow Problems: An Algorithmic and Computational Approach}, publisher = {Springer, Berlin}, year = {1999}, } @BOOK{Lendecke2003, author = {.}, title = {Samba f\"{u}r Unix/Linux--Administratoren}, publisher = {dpunkt.verlag}, year = {2003}, volume = {1. Auflage}, note = {ISBN 3.-89864-193-7}, } @BOOK{BallHunt2000a, author = {. and .}, editor = {. and .}, title = {ICIAM 99}, publisher = {Oxford University Press}, year = {2000}, volume = {Proceedings of the Fourth International Congress on Industrial and Applied Mathematics}, month = jul, note = {Edingburgh 5 - 9 July 1999, ISBN 0198505140}, } @BOOK{DoneaHuerta2003, author = {. and .}, title = {Finite Element Methods for Flow Problems}, publisher = {John Wiley \& Sons Ltd.}, year = {2003}, month = may, note = {ISBN 0-471-49666-9}, } @BOOK{Fluent2002, author = {Fluent}, title = {Fluent 6.1}, publisher = {Fluent Inc.}, year = {2002}, note = {\url{http:/ / www.fluent.com} }, } @BOOK{Comsol2003, author = {Comsol}, title = {FEMLAB 2.3}, publisher = {Comsol Group}, year = {2003}, note = {\url{http:/ / www.femlab.com} }, } @BOOK{Walter1978, author = {.}, title = {Differentialgeometrie}, publisher = {Wissenschaftsverlag Bibliographisches Institut}, year = {1978}, note = {ISBN 3-411-01543-8}, } @BOOK{KluenterLaser2003, author = {Kl\"{. and .}, title = {LDAP verstehen, Open LDAP einsetzen}, publisher = {dpunkt.verlag}, year = {2003}, volume = {1. Auflage}, address = {Ringstr. 19, 69115 Heidelberg}, note = {ISBN 3-89864-217-8}, } @BOOK{Brodlieetal.1992, author = {Brodlie}, title = {Scientific Visualization}, publisher = {Springer, Berlin}, year = {1992}, } @BOOK{AbramowskiMueller1992, author = {. and .}, title = {Geometrisches Modellieren}, publisher = {BI-Wissenschaftsverlag}, year = {1992}, } @BOOK{DeuflhardBornemann1994, author = {. and Bornemann}, title = {Numerische Mathematik II {}--{} Integration gew\"{o}hnlicher Differentialgleichungen}, publisher = {de Gruyter}, year = {1994}, } @BOOK{SilberschatzGagneGalvin2002, author = {. and . and .}, title = {Operating Systems Concepts (6th Edition)}, publisher = {John Wiley \& Sons Ltd.}, year = {2002}, } @BOOK{Comer1999, author = {.}, title = {Computer Networks and Internets}, publisher = {Prentice Hall}, year = {1999}, } @BOOK{Tanenbaum2002, author = {.}, title = {Computer Networks (4th Edition)}, publisher = {Prentice Hall}, year = {2002}, } @BOOK{HairerNorsettWanner1987, author = {Hairer and Norsett and Wanner}, title = {Solving Ordinary Differential Equations I}, publisher = {Springer, Berlin}, year = {1987}, volume = {I - Nonstiff Problems}, } @BOOK{HairerNorsettWanner1992, author = { Wanner}, title = {Solving Ordinary Differential Equations II}, publisher = {Springer, Berlin}, year = {1992}, volume = {II - Stiff Problems}, } @BOOK{FoleyvanDamFeinerHughes1996, author = {. and Hughes, .}, title = {Computer Graphics--{} Principles and Practice (2nd Edition)}, publisher = {Addison-Wesley}, year = {1996}, } @BOOK{GonzalezWoods1992, author = {. and .}, title = {Digital Image Processing}, publisher = {Addison-Wesley}, year = {1992}, } @BOOK{GonzalesWoods1992, author = {. and .}, title = {Digital Image Processing}, publisher = {Addison-Wesley}, year = {1992}, } @BOOK{HundsdorferVerwer2003, author = {. and .}, editor = {. and . and . and . and }, title = {Numerical Solution of Time--Dependent Advection--Diffusion--Reaction Equations}, publisher = {Springer, Berlin}, year = {2003}, note = {ISBN 3-540-03440-4}, } @BOOK{Kilian2002, author = {.}, title = {ScaRC als verallgemeinerter Mehrgitter--{} und Gebietszerlegungsansatz f\"{u}r parallele Rechnerplattformen}, publisher = {Logos Verlag, Berlin}, year = {2002}, note = {ISBN 3-8325-0092-8}, } @BOOK{John2003, author = {.}, title = {Large Eddy Simulation of Turbulent Incompressible Flows}, publisher = {Springer, Berlin}, year = {2003}, note = {ISBN 3-540-40643-3}, } @BOOK{KrauseJaegerResch2003, author = {. and J\"{a}. and .}, editor = {. and J\"{a}. and .}, title = {High Performance Computing in Science and Engineering ` 03}, publisher = {Springer, Berlin}, year = {2003}, note = {Buch war ein Geschenk, ISBN 3-540-40850-9}, } @BOOK{Oksendal2003, author = {.}, title = {Stochastic Differential Equations}, publisher = {Springer, Berlin}, year = {2003}, volume = {Sixth Edition}, note = {ISBN 3-540-04758-1}, } @BOOK{Wesseling1992, author = {.}, title = {An Introduction to Multigrid Methods}, publisher = {John Wiley \& Sons Ltd.}, year = {1992}, note = {\url{http:/ / www.mgnet.org/ mgnet-books-wesseling.html} }, } @BOOK{Kelly1998, author = {.}, title = {Element Shape Testing}, publisher = {AnSyS Inc.}, year = {1998}, volume = {9 th edition}, note = {chapter 13 in Ansys Theory Reference; www.ansys.com }, } @BOOK{Kuzmin2001, author = {.}, editor = {.}, title = {Positive Finite Element Schemes Based on the Flux--Corrected Transport Procedure}, publisher = {Elsevier}, year = {2001}, volume = {Computational Fluid and Solid Mechanics}, note = {887-888 }, } @BOOK{SamimyBreuerLealSteen2003, author = {. and . and . and .}, title = {A Gallery of Fluid Motion}, publisher = {Cambridge University Press}, year = {2003}, note = {ISBN 052153500X}, } @BOOK{Anderson1995, author = {.}, title = {Computational Fluid Dynamics: The Basics with Applications}, publisher = {McGraw Hill}, year = {1995}, edition = {6}, note = {ISBN 0070016852}, } @BOOK{Hackbusch2002, author = {.}, title = {Iterative L\"{o}sung gro"ser schwachbesetzer Gleichungssysteme}, publisher = {Oxford, Clarendon Press}, year = {2002}, month = mar, note = {ISBN 351912372X}, } @BOOK{Braess1997, author = {.}, title = {Finite Elemente}, publisher = {Springer, Berlin}, year = {1997}, edition = {2}, note = {ISBN 3540619054}, } @BOOK{BabuskaStrouboulis2001, author = {. and .}, title = {The Finite Element method and its reliability}, publisher = {Oxford University Press}, year = {2001}, } @BOOK{Ciarlet1978, author = {.}, editor = {. and . and .}, title = {The finite element method for elliptic problems}, publisher = {North-Holland Publishing Company, Amsterdam, New-York, Oxford}, year = {1978}, series = {Studies in mathematics and its applications, Vol. 4}, note = {ISBN 0444850287}, } @BOOK{DavisRabinowitz1975, author = {. and Rabinowitz}, title = {Methods of numerical integration}, publisher = {Academic Press, New York}, year = {1975}, edition = {2}, note = {ISBN 0122063503}, } @BOOK{Meister1999, author = {.}, title = {Numerik linearer Gleichungssysteme}, publisher = {Vieweg}, year = {1999}, note = {ISBN 3528031352}, } @BOOK{AinsworthOden2000, author = {. and .}, title = {A posteriori error estimation in finite element analysis}, publisher = {John Wiley \& Sons Ltd.}, year = {2000}, } @BOOK{BangerthRannacher2003, author = {. and .}, title = {Adaptive Finite Element Methods for Differential Equations}, publisher = {Birkh\"{a}user}, year = {2003}, series = {Lectures in Mathematics}, note = {ISBN 37764370092}, } @BOOK{KuzminLoehnerTurek2004, author = {. and . and .}, title = {Flux--Corrected Transport}, publisher = {Springer, Berlin}, year = {2005}, series = {Scientific Computation}, note = {Subtitle: Principles, Algorithms and Applications, ISBN 3-540-23730-5}, } @BOOK{Braess2003, author = {.}, title = {Finite Elemente}, publisher = {Springer, Berlin}, year = {2003}, edition = {3rd}, note = {3., korrigierte und erg\"{a}nzte Auflage, ISBN 3-540-00122-0}, } @BOOK{Bathe2001, author = {.}, title = {Finite--Elemente--Methoden}, publisher = {Springer, Berlin}, year = {2001}, edition = {2}, note = {ISBN 3540668063 }, } @BOOK{Fernando2004, author = {.}, editor = {.}, title = {{GPU} Gems {}}, publisher = {Addison-Wesley}, year = {2004}, volume = {Programming Techniques, Tips, and Tricks for Real-Time Graphics}, month = nov, note = {ISBN 0-321-22832}, } @BOOK{SmithBjoerstadGropp1996, author = {. and . and .}, title = {Domain Decomposition}, publisher = {Cambridge University Press}, year = {1996}, } @BOOK{GoedeckerHoisie2001, author = {. and .}, title = {Performance Optimization of Numerically Intensive Codes}, publisher = {SIAM}, year = {2001}, } @BOOK{Ammeraal1998, author = {.}, title = {Computer Graphics for Java Programmers}, publisher = {John Wiley \& Sons Ltd.}, year = {1998}, } @BOOK{ErikssonEstepJohnson2005, author = {. and . and .}, title = {Angewandte Mathematik: Body and Soul [Band 2]}, publisher = {Springer, Berlin}, year = {2005}, note = {Exemplar von Herrn Turek }, } @BOOK{BollhoeferMehrmann2004, author = {. and .}, title = {Numerische Mathematik}, publisher = {Vieweg}, year = {2004}, note = {ISBN 3-528-03220-0}, } @BOOK{SchlagerThibud2005, author = {. and .}, title = {Wissenschaftliche mit LATEX arbeiten}, publisher = {Pearson Studium}, year = {2005}, note = {ISBN 3-8273-7078-7}, } @BOOK{MittelbachGoossens2005, author = {. and .}, title = {The Latex Companion {}}, publisher = {Addison-Wesley}, year = {2005}, edition = {Second Edition}, note = {ISBN 0201362996}, } @BOOK{Gunzburger2002, author = {.}, title = {Perspectives in Flow Control and Optimization}, publisher = {SIAM}, year = {2002}, month = dec, note = {ISBN 089871527X}, } @BOOK{Troeltzsch2005, author = {Tr\"{.}, title = {Optimalsteuerung bei partiellen Differentialgleichungen}, publisher = {Vieweg}, year = {2005}, month = jul, note = {ISBN 3528032243}, } @BOOK{GilgSinger2005, author = {. and .}, title = {Repetitorium der Algebra}, publisher = {Shaker, Postfach 101818, 52018 Aachen}, year = {2005}, note = {ISBN 3-8322-4280-5}, } @BOOK{Braess2001a, author = {.}, title = {Finite Elements}, publisher = {Cambridge University Press}, year = {2001}, edition = {2nd}, note = {ISBN 0-521-01195-7}, } @BOOK{Swan2005, author = {.}, title = {Practical English Usage}, publisher = {Oxford University Press}, year = {2005}, note = {Buch wurde aus LS-Mitteln bezahlt }, } @INPROCEEDINGS{FrolkovicGeiser2000, author = {}, title = {Numerical simulations of radionuclides transport in double porosity media with sorption}, year = {2000}, pages = {28--36}, } @INPROCEEDINGS{Hinze2000, author = {.}, title = {The SQP--method for tracking type control of the instationary {Navier}--{Stokes} equations}, year = {2000}, } @INPROCEEDINGS{Hron2000, author = {.}, title = {Mathematical description of soft tissue mechanics}, year = {2000}, pages = {37--46}, } @INPROCEEDINGS{TornbergEngquist2000, author = {}, title = {The segment projection method applied to multiphase flow simulations}, year = {2000}, pages = {011--019}, } @INPROCEEDINGS{Bern2001, author = {.}, title = {A Computational Geometry Perspective on Adaptive Mesh Generation}, booktitle = {Error Estimation and Solution Adaptive Discretization in {CFD}}, publisher = {NATO/NASA/VKI Lecture Series}, year = {2001}, pages = {1--38}, series = {NATO/NASA/VKI Lecture Series}, note = {NASA Ames Research Center, September 10 -14, 2001}, } @INPROCEEDINGS{GilesPierce2001, author = {. and .}, title = {Adjoint Error Correction for Integral Outputs}, booktitle = {Error Estimation and Solution Adaptive Discretization in {CFD}}, year = {2001}, pages = {39--86}, series = {NATO/NASA/VKI Lecture Series}, note = {NASA Ames Research Center, September 10 -14, 2001}, } @INPROCEEDINGS{PateraPeraire2001, author = {. and .}, title = {Finite Elements and Reduced Basis Methods}, booktitle = {Error Estimation and Solution Adaptive Discretization in {CFD}}, year = {2001}, pages = {145--220}, note = {NASA Ames Research Center, September 10 - 14, 2001}, } @INPROCEEDINGS{Prudhomme2001, author = {.}, title = {Global Error Estimates by Residual Methods}, booktitle = {Error Estimation and Solution Adaptive Discretization in {CFD}}, year = {2001}, pages = {221--284}, series = {NATO/NASA/VKI Lecture Series}, note = {NASA Ames Research Center, September 10-14, 2001}, } @INPROCEEDINGS{SuliHouston2001, author = {. and .}, title = {Adaptive Finite Element Approximation of Hyperbolic Problems}, booktitle = {Error Estimation and Solution Adaptive Discretization in {CFD}}, year = {2001}, pages = {285--346}, series = {NATO/NASA/VKI Lecture Series}, note = {NASA Ames Research Center, September 10 - 14, 2001}, } @INPROCEEDINGS{Busse1989, author = {.}, editor = {.}, title = {Fundamentals in Thermal Convection}, booktitle = {Mantle Convection: Plate Tectonics and Global Dynamics}, year = {1989}, pages = {23--95}, series = {Gordon and Breach Science Publishers}, } @INPROCEEDINGS{CuthillMcKee1969, author = {. and .}, title = {Reducing the bandwidth of sparse symmetric matrices}, booktitle = {ACM/CSC--ER Proceedings of the 1969 24th national conference}, publisher = {ACM Press}, year = {1969}, pages = {157--172}, note = {ACM Nat. Conf., \url{http:/ / portal.acm.org/ citation.cfm?id=805928}}, } @INPROCEEDINGS{HughesBrooks1979, author = {. and .}, editor = {.}, title = {A multidimensional upwind scheme with no crosswind diffusion}, booktitle = {Finite element methods for convection dominated flows}, year = {1979}, note = {New York: ASME 1979}, } @INPROCEEDINGS{Johnson1989, author = {.}, editor = {. and .}, title = {The streamline diffusion finite element method for compressible and incompressible fluid flow}, booktitle = {Numerical Analysis 1989, Finite Element Method in Fluids VII}, publisher = {Longman Sci. Tech.}, year = {1989}, pages = {155--181}, series = {Pitman Res. Notes Math. Ser., 228}, note = {Huntsville}, } @INPROCEEDINGS{Tobiska1989, author = {.}, editor = {. and .}, title = {Full and weighted upwind finite element methods}, booktitle = {Splines in Numerical Analysis}, year = {1989}, note = {Internationales Seminar ISAM 1989 in Weissig}, } @INPROCEEDINGS{VariousXX, author = {Various}, title = {Proc. IEEE Conferences on Visualization (Visualization `XX)}, booktitle = {IEEE Visualization}, year = {XX}, note = {Eine Reihe von Tagungsb\"{a}nden dieser j\"{a}hrlichen Konferenz befindet sich in der Bereichsbibliothek Informatik}, } @INPROCEEDINGS{AcademyofScienceoftheCzechRepublic2002, author = {Academy of Science of the Czech Republic}, editor = {. and . and . and }, title = {Applications of Mathematics}, booktitle = {Lecture Notes of the Seventh International School on `Mathematical Theory in Fluid Mechanics`}, year = {2002}, volume = {Volume 47}, number = {Number 6}, pages = {461--543}, } @INPROCEEDINGS{TurekSchaeferRannacher1998, author = {. and Sch\"{a}. and .}, editor = {. and . and . and . and .}, title = {Evaluation of a {CFD} Benchmark for Laminar Flows}, booktitle = {ENUMATH 97: Proceedings of the 2nd European Conference on Numerical Mathematics and Advanced Applications}, publisher = {World Science Publishing}, year = {1998}, pages = {549--563}, } @INPROCEEDINGS{TurekBeckerOswald1998a, author = {. and . and .}, title = {Parallel Multilevel Algorithms for Solving the Incompressible {Navier}--{}--{Stokes} Equations}, booktitle = {Proceedings of the High Performance Workshop at the HLRZ Stuttgart}, year = {1998}, } @INPROCEEDINGS{TurekBeckerRannacher2001, author = {. and . and .}, editor = {. et al.}, title = {An Efficient {Navier}--{Stokes} Solver for Automotive Aerodynamics}, booktitle = {to appear in Mathematik Schl\"{u}sseltechnologie f\"{u}r die Zukunft}, publisher = {nn}, year = {2001}, } @INPROCEEDINGS{TurekOuazziHron2002, author = {. and .}, editor = {B\"{a}nsch, E.}, title = {A Computational Comparison of two {FEM} Solvers For Nonlinear Incompressible Flow}, booktitle = {Challenges in Scientific Computing CISC 2002}, publisher = {Springer, Berlin}, year = {2002}, pages = {87--109}, series = {LNCSE}, address = {Berlin}, note = {ISBN 3-540-40887-8}, } @INPROCEEDINGS{TurekDecheng2002, author = {. and .}, editor = {B\"{a}nsch, E.}, title = {The Fictitious Boundary Method for the Implicit Treatment of Dirichlet Boundary Conditions with Applications to Incompressible Flow Simulation}, booktitle = {Challenges in Scientific Computing CISC 2002}, publisher = {Springer, Berlin}, year = {2002}, pages = {37--68}, series = {LNCSE}, address = {Berlin}, } @INPROCEEDINGS{TurekKilian1998, author = {. and .}, editor = {. and . and . and . and .}, title = {An Example for Parallel ScaRC and Its Application to the Incompressible {Navier}--{Stokes} Equations}, booktitle = {ENUMATH 97: Proceedings of the 2nd European Conference on Numerical Mathematics and Advanced Applications}, publisher = {World Science Publishing}, year = {1998}, pages = {389--396}, } @INPROCEEDINGS{TurekBeckerKilian1998, author = {. and . and .}, editor = {. and . and .}, title = {Some concepts of the software package FEAST}, booktitle = {VECPAR`98 {}--{} Third International Conference for Vector and Parallel Processing}, publisher = {Springer, Berlin}, year = {1998}, volume = {1573}, series = {Lecture Notes in Computer Science}, } @INPROCEEDINGS{TurekBeckerKilian1999, author = {. and . and .}, editor = { .}, title = {Consequences of modern hardware design for numerical simulations and their realization in FEAST}, booktitle = {Proceedings Euro--Par`99 Parallel Processing}, publisher = {Springer, Berlin}, year = {1999}, note = {Toulouse, France, August/September 1999}, } @INPROCEEDINGS{TurekBeckerBuijssenKilian2002, author = {. and . and . and .}, editor = {. and .}, title = {High Performance {FEM} Simulation via FEAST and Application to Parallel {CFD} via FeatFlow}, booktitle = {NIC Symposium 2001}, year = {2002}, volume = {9}, pages = {493--502}, series = {NIC-Serie}, note = {5.-6. Dezember 2001, Forschungszentrum J\"{u}lich}, } @INPROCEEDINGS{TurekBuijssen2002, author = {. and .}, editor = {. and .}, title = {Sources of Parallel Inefficiency for Incompressible {CFD} simulation}, booktitle = {Proceedings 8th International Euro--Par Conference}, publisher = {Springer, Berlin}, year = {2002}, pages = {701--704}, series = {LNCS}, note = {Paderborn, Germany, August 27-30}, } @INPROCEEDINGS{Turek1992, author = {.}, title = {NSDIV2D {}--{}--{} flow prediction tools via discrete divergence--free finite element models}, booktitle = {DGLR--{}--Bericht 92--{}--07}, year = {1992}, note = {8. DGLR-Fach-Symposium Str\"{o}mungen mit Abl\"{o}sung, K\"{o}ln-Porz}, } @INPROCEEDINGS{TurekSchreiber1993b, author = {. and .}, title = {An Efficient Finite Element Solver for the Nonstationary Incompressible {Navier}--{Stokes} Equations in Two and Three Dimensions}, booktitle = {Proc. Workshop `Numerical Methods for the {Navier}--{Stokes} Equations`}, publisher = {Vieweg}, year = {1993}, volume = {47}, pages = {25--28}, series = {Notes on Numerical Fluid Mechanics}, } @INPROCEEDINGS{Turek1994h, author = {.}, editor = {.}, title = {Tools for predicting incompressible flows via nonconforming finite element models}, booktitle = {Computational Methods in Water Resources X}, publisher = {Kluwer}, year = {1994}, pages = {1255--1262}, } @INPROCEEDINGS{TurekMuellerProhlRannacher1994, author = {. and . and . and .}, editor = {. and .}, title = {Implicit Time--Discretization of the Nonstationary Incompressible {Navier}--{Stokes} Equations}, booktitle = {Proc. 10th GAMM--Seminar}, publisher = {Vieweg}, year = {1994}, } @INPROCEEDINGS{Turek1996f, author = {.}, title = {Recent Benchmark Computations of Laminar Flow Around a Cylinder}, booktitle = {Proc. 3rd World Conference in Applied Computational Fluid Mechanics}, year = {1996}, note = {Freiburg}, } @INPROCEEDINGS{TurekOswald1996, author = {. and .}, title = {A Parallel Multigrid Algorithm for Solving the Incompressible {N}avier--{}--{S}tokes Equations with Nonconforming Finite Elements in Three Dimensions}, booktitle = {Proc. Parallel CFD`96, Capri/Italy}, year = {1996}, } @INPROCEEDINGS{TurekRannacherSchreiber1996a, author = {. and . and .}, editor = {. and J\"{. and .}, title = {{Numerische Modellierung von Gasbrennern: Berechnung schwachkompressibler Gasstr\"{o}mungen}}, booktitle = {{Mathematik - Schl\"{u}sseltechnologie f\"{u}r die Zukunft}}, publisher = {Springer, Berlin}, year = {1996}, } @INPROCEEDINGS{TurekSchaefer1996, author = {. and Sch\"{a}.}, editor = {.}, title = {Benchmark computations of laminar flow around cylinder}, booktitle = {Flow Simulation with High--Performance Computers II}, publisher = {Vieweg}, year = {1996}, volume = {52}, pages = {547--566}, series = {Notes on Numerical Fluid Mechanics}, note = {co. , , }, } @INPROCEEDINGS{Turek1992c, author = {.}, title = {On Ordering Strategies in a Multigrid Algorithm}, booktitle = {Proc. 8th GAMM--{}--Seminar}, publisher = {Vieweg}, year = {1992}, volume = {41}, series = {Notes on Numerical Fluid Mechanics}, } @INPROCEEDINGS{AltieriBeckerTurek1998, author = {. and . and .}, editor = {. and . and .}, title = {On the realistic performance of Linear Algebra components in iterative solvers}, booktitle = {High Performance Scientific and Engineering Computing: {} Proceedings of the International FORTWIHR Conference on HPSEC, Munich, March 16--18, 1998}, publisher = {Springer, Berlin}, year = {1999}, volume = {8}, pages = {3--12}, series = {Lecture Notes in Computational Science and Engineering}, note = {ISBN 3-540-65730-4}, } @INPROCEEDINGS{Turek1994b, author = {.}, editor = {. and .}, title = {Multigrid techniques for simple discretely divergence--free finite element spaces}, booktitle = {Multigrid Methods IV}, year = {1994}, pages = {321--332}, series = {Multigrid Methods IV}, note = {\url{http:/ / www.featflow.de/ ture/ paper/ mg_emg93.ps.gz}}, } @INPROCEEDINGS{TurekBeckerKilian2003b, author = {. and . and .}, title = {Hardware--oriented Numerics and concepts for {PDE} software}, booktitle = {FUTURE 1095}, publisher = {Elsevier}, year = {2003}, pages = {1--23}, note = {International Conference on Computational Science ICCS2002, Amsterdam}, } @INPROCEEDINGS{JablonowskiBrunetBlissHaber1993, author = {. and . and . and .}, title = {VASE: the Visualization and Application Steering Environment}, booktitle = {SuperComputing 1993}, year = {1993}, pages = {560--569}, } @INPROCEEDINGS{BernEppstein1995, author = { .}, title = {Mesh Generation and Optimal Triangulation}, year = {1995}, note = {\url{http:/ / www.ics.uci.edu/ ~eppstein/ pubs/ BerEpp-CEG-95.pdf}}, } @INPROCEEDINGS{DarmofalHaimes1995, author = {}, title = {An Analysis of 3--D Particle Path Integration Algorithms}, booktitle = {AIAA Paper {}}, year = {1995}, pages = {95--1713}, note = {htttp://raphael.mit.edu/pv3/pv3.html}, } @INPROCEEDINGS{DouglasHuKarlKowarschikRuedeWeiss2000, author = { . and . and }, editor = {.}, title = {Fixed and adaptive cache aware algorithms for multigrid methods}, booktitle = {Multigrid methods VI. Proceedings of the 6th European multigrid conference, Gent, Belgium, September 27--30, 1999}, publisher = {Springer, Berlin}, year = {2000}, volume = {14}, pages = {87--93}, series = {Lect. Notes Comput. Sci. Eng.}, } @INPROCEEDINGS{Rannacher1998, author = {.}, editor = {. and .}, title = {Error Control in Finite Element Computations}, booktitle = {Proc. Summer School Error Control and Adaptivity in Scientific Computing}, publisher = {Kluwer}, year = {1998}, pages = {247--278}, } @INPROCEEDINGS{Ruede1999, author = {.}, editor = {. and . and .}, title = {Technological trends and their impact on the future of supercomputers}, booktitle = {High Performance Scientific and Engineering Computing}, publisher = {Springer, Berlin}, year = {1999}, volume = {8}, series = {LNCSE}, } @INPROCEEDINGS{KwokChen2000, author = {Kwok and .}, title = {A simple and effective mesh quality metric for hexahedral and wedge elements}, booktitle = {Proceedings of the 9th International Meshing Roundtable}, year = {2000}, volume = {9}, address = {\url{http:/ / imr.sandia.gov}}, } @INPROCEEDINGS{Knupp2001a, author = {Knupp}, title = {Hexahedral Mesh Untangling \& Algebraic Mesh Quality Metrics}, booktitle = {Proceedings of the 9th International Meshing Roundtable}, publisher = {Springer, Berlin}, year = {2001}, volume = {17}, number = {3}, pages = {261--268}, series = {Engineering with Computers}, address = {\url{http:/ / imr.sandia.gov}}, month = oct, } @INPROCEEDINGS{Durbeck1999, author = {Durbeck}, title = {Evaporation: A technique for visualizing mesh quality}, booktitle = {Proceedings of the 8th International Meshing Roundtable}, year = {1999}, volume = {8}, address = {\url{http:/ / imr.sandia.gov}}, } @INPROCEEDINGS{Knupp1999, author = {Knupp}, title = {Martix norms \& condition numbers}, booktitle = {Proceedings of the 8th International Meshing Roundtable}, year = {1999}, address = {\url{http:/ / imr.sandia.gov}}, } @INPROCEEDINGS{OwenStatenCanannSaigal1998, author = { }, title = {Advancing Front Quadrilateral Meshing Using Triangle Transformations}, booktitle = {Proceedings of the 7th International Meshing Roundtable}, year = {1998}, volume = {7}, address = {\url{http:/ / ims.sandia.gov}}, } @INPROCEEDINGS{Pellegrini1990, author = {Pellegrini}, title = {Stabbing and ray shooting in 3 dimensional space}, booktitle = {Proceedings of the 6th annual symposium on computational geometry}, publisher = {ACM Press}, year = {1990}, volume = {6}, pages = {177--186}, month = may, note = {ISBN 0-89791-362-0}, } @INPROCEEDINGS{GuskovVidimceSweldensSchroeder2000, author = { \"{o}der}, title = {Normal Meshes}, booktitle = {Proceedings of the 27th annual conference on computer graphics and interactive techniques}, publisher = {ACM Press}, year = {2000}, volume = {27}, pages = {95--102}, note = {ISBN 1-58113-208-5}, } @INPROCEEDINGS{BarequetKumar1997, author = {}, editor = {. and .}, title = {Repairing CAD Models}, booktitle = {IEEE Visualization `97}, publisher = {IEEE Computer Society Press}, year = {1997}, pages = {362--370}, address = {\url{http:/ / cs.jhu.edu/ ~barquet}}, } @INPROCEEDINGS{KuzminTurek2004d, author = {. and .}, editor = {. and . and Neittaanm\"{a}.}, title = {Finite Element Discretization and Iterative Solution Techniques for Multiphase Flows in Gas--Liquid Reactors}, booktitle = {Conjugate Gradient Algorithms and Finite Element Methods}, publisher = {Springer, Berlin}, year = {2004}, pages = {297--324}, note = {Jyv\"{a}skyl\"{a}, Finland, June 11-12,2002; (ISBN-Nr. 3-540-21319-8)}, } @INPROCEEDINGS{KuzminTurek2003b, author = {. and .}, editor = {.}, title = {Explicit and Implicit High--Resolution Finite Element Schemes Based on the Flux--Corrected--Transport Algorithm}, booktitle = {Proceedings of the 4th European Conference on Numerical Mathematics and Advanced Applications}, publisher = {Springer, Berlin}, year = {2003}, pages = {133--143}, } @INPROCEEDINGS{KuzminMoellerTurek2004, author = {. and M\"{o} .}, editor = {. and . and Neittaanm\"{a}.}, title = {Implicit Flux--Corrected Transport Algorithm for Finite Element Simulaiton of the Compressible Euler Equations}, booktitle = {Conjugate Gradient Algorithms and Finite Element Methods}, publisher = {Springer, Berlin}, year = {2004}, pages = {325--354}, note = {Jyv\"{a}skyl\"{a}, Finland, June 11-12, 2002 (ISBN-Nr. 3-540-21319-8)}, } @INPROCEEDINGS{ZietheSpohnTurek2002a, author = {. and . and .}, title = {Core Formation: A New Modelling Approach}, booktitle = {ESLAB 36, Earth--Like Planets and Moons}, year = {2002}, note = {ESTEC Noordwijk (Poster)}, } @INPROCEEDINGS{Kuzmin2000e, author = {.}, title = {Numerical Simulation of Mass Transfer and Chemical Reactions in Gas--Liquid Flows}, booktitle = {Proceedings of the 3rd European Conference on Numerical Mathematics and Advanced Applications}, year = {2000}, volume = {World Scientific}, pages = {237--244}, } @INPROCEEDINGS{Kuzmin2000f, author = {.}, editor = {Neittaanm\"{. and .}, title = {The Free--{Lagrange} {FEM} for Problems in Time--Dependent Domains}, booktitle = {Proceedings of Analysis and Appproximation of Boundary Value Problems}, year = {2000}, pages = {115--125}, series = {Reports of the Department of Mathematical Information Technology A 2}, address = {University of Jyv\"{a}skyl\"{a}}, } @INPROCEEDINGS{KuzminTurek2000, author = {. and .}, title = {Efficient Numerical Techniques for Flow Simulation in Bubble Column Reactor}, booktitle = {Preprints of the 5th German--Japanese Symposium on Bubble Columns}, year = {2000}, pages = {99--104}, series = {VDI/GVC}, } @INPROCEEDINGS{KuzminTurek2004a, author = {. and .}, title = {Numerical Simulation of Turbulent Bubbly Flows}, booktitle = {submitted to 3rd International Symposium on Two--Phase Flow Modelling and Experimentation}, year = {2004}, note = {Pisa, 22. - 24. September 2004}, } @INPROCEEDINGS{RevellesUrenaLastra2000, author = { }, title = {An efficient parametric algorithm for Octree Traversal}, booktitle = {The 8th International Conference in Central Europe on Computer Graphics, Visualization and Interactive Digital Media 2000}, year = {2000}, pages = {212--219}, address = {\url{http:/ / wscg.zcu.cz/ wscg2000/ wscg2000.htm}}, month = feb, note = {Turek-ID: GRID1, SOFT1, SOFT6}, } @INPROCEEDINGS{Mukherjee2002, author = {Mukherjee}, title = {A hybrid, variational 3D smoother for orphaned shell meshes}, booktitle = {Proceedings of the 11th International Roundtable, Sandia National Laboratories}, year = {2002}, pages = {379--390}, address = {\url{http:/ / imr.sandia.gov}}, } @INPROCEEDINGS{Blacker2000, author = {Blacker}, title = {Meeting the Challenge for Automated Conformal Hexahedral Meshing}, booktitle = {Proceedings of the 11th International Roundtable, Sandia National Laboratories}, year = {2000}, pages = {11--19}, address = {\url{http:/ / img.sandia.gov}}, } @INPROCEEDINGS{ZhuBlackerSmith2002, author = {}, title = {Background overlay grid size funcions}, booktitle = {Proceedings of the 11th International Meshing Roundtable, Ithaca, New York, USA}, year = {2002}, pages = {65--74}, address = {\url{http:/ / imr.sandia.gov}}, } @INPROCEEDINGS{ChanvanderHorst1997, author = {. and .}, editor = {. and .}, title = {Approximate and incomplete factorizations}, booktitle = {Parallel Numerical Algorithms; ICASE/LaRC Interdisciplinary Series in Science and Engineering}, publisher = {Kluwer}, year = {1997}, pages = {167--202}, note = {Preprint: \url{http:/ / citeseer.ist.psu.edu}}, } @INPROCEEDINGS{VuikvanKanWesseling2000, author = {. and . and .}, title = {A black box multigrid preconditioner for second order elliptic partial differential equations}, booktitle = {European Congress on Computational Methods in Applied Sciences and Engineering, ECCOMAS 2000, Barcelona}, year = {2000}, month = sep, } @INPROCEEDINGS{HoffmanJohnson2001, author = {. and .}, title = {Adaptive Finite Element Methods for Incompressible Fluid Flow}, booktitle = {Error Estimation and Solutioh Adaptive Discretization in {CFD}}, year = {2001}, pages = {87--144}, series = {NATO/NASA/VKI Lecture Series}, note = {NASA Ames Research Center, September 2001}, } @INPROCEEDINGS{Buijssen2004a, author = {.}, editor = {. and . and . and .}, title = {parpp3d++ {}--{} a Parallel {HPC} Code for the Incompressible Nonstationary {Navier}--{}--{Stokes} Equations}, booktitle = {High Performance Computing in Science and Engineering}, publisher = {Springer, Berlin}, year = {2004}, pages = {169--178}, series = {Transaction of the Second Joint HLRB and KONWIHR Result and Reviewing Workshop}, month = mar, note = {ISBN 3-540-44326-6}, } @INPROCEEDINGS{KeiterRombach2002, author = {Keiter, . and .}, title = {Accurate handling of pressure peaks in FE--simulations of granular media}, booktitle = {15th ASCE Engineering Mechanics Conference, June 2--5, 2002}, year = {2002}, address = {Columbia University, New York, NY}, } @INPROCEEDINGS{Lukacova-MedvidovaGrigerekNecasova2000, author = {. and . and .}, title = {Numerical solution of bipolar barotropic non--Newtonian fluids}, booktitle = {Proceedings of the Num. methods in contin. mech. 2000}, year = {2000}, } @INPROCEEDINGS{SeppaenenVauhonenKaipioSomersalo2002, author = {.}, title = {Inteference of velocity field based on tomographic measurements in process industry}, booktitle = {4th International Conference on Inverse Problems in Engineering, Rio de Janeiro, Brazil}, year = {2002}, } @INPROCEEDINGS{EgorovMakarovRudinskySmirnovZhmakin1998, author = { .}, title = {Modelling analysis of oxygen transport during Czochralski growth of silicon crystals}, booktitle = {Mat. Res. Soc. Symp.}, year = {1998}, volume = {490}, pages = {181--186}, } @INPROCEEDINGS{Griebel2002, author = {.}, editor = {. and .}, title = {Computer Science and Numerical Mathematics}, booktitle = {NIC Symposium}, year = {2002}, volume = {9}, pages = {477--478}, address = {John von Neumann Institute for Computing, J\"{u}lich; http.//www.fz-juelich.de/nic-series/volume9}, note = {ISBN 3-00-009055-X}, } @INPROCEEDINGS{MatthiesTobiska2000, author = {. and .}, title = {Simulation of free surfaces of magnetic liquids}, booktitle = {Proceedings of ALGORITHMY 2000, Conference on Scientific Computing}, year = {2000}, pages = {1--10}, } @INPROCEEDINGS{Pukhnachov1994a, author = {.}, editor = {. and . and .}, title = {Nonlocal symmetries in nonlinear heat equations}, booktitle = {Energy Methods in Continuum Mechanics}, year = {1994}, volume = {Proceedings of the Workshop on Energy Methods for Free Boundary Problems in Continuum Mechanics, Oviedo, Spain, March 21-23}, month = mar, } @INPROCEEDINGS{DoghriNikolovPierardZealouk2000, author = {. and . and . and .}, title = {A micro--mechanically based constitutive model for small deformation of semi--crystalline polymers}, booktitle = {5th Belgian National Congress on Theoretical and Applied Mech., Louvain--la--Neuve, May 2000}, year = {2000}, month = may, } @INPROCEEDINGS{DoghriNikolovPierardZealouk2000a, author = {. and . and . and .}, title = {Multi--scale modeling and simulation of the small deformation of semu--crystalline polymers}, booktitle = {Int. Conf. on Comput. Engin. Science, ICES`2K, Los Angeles, Aug. 2000}, year = {2000}, month = aug, } @INPROCEEDINGS{Pukhnachov1996a, author = {.}, title = {Thermocapillary motion in an emulsion}, booktitle = {Third Microgragity Fluid Physics Conference}, year = {1996}, volume = {NASA Conference Publication}, number = {3338}, month = jun, } @INPROCEEDINGS{Braess2001, author = {.}, title = {Enhanced Assumed Strain Elements and Nearly Incompressible Material}, booktitle = {European Conference on Computational Mechanics, ECCM--2001, Cracow, Poland, June 26--29}, year = {2001}, } @INPROCEEDINGS{KobbeltVorsatzLabsikSeidel1999, author = {. and .}, editor = {. and .}, title = {A shrink Wrapping Approach to Remeshing Polygonal Surfaces}, booktitle = {EUROGRAPHICS `99}, year = {1999}, volume = {18}, number = {3}, } @INPROCEEDINGS{TakanashiMurakiDoiKaufman1998, author = {. and .}, title = {3D Active Net for Volume Extraction}, booktitle = {SPIE Electronic Imaging `98}, year = {1998}, number = {3298}, pages = {184--192}, series = {Proceedings SPIE}, } @INPROCEEDINGS{LuerigKobbeltErtl1998, author = { .}, title = {Deformable surfaces for feature based indirect finite volume rendering}, booktitle = {Computer Graphics International}, year = {1998}, pages = {752--760}, series = {IEEE Proceedings}, } @INPROCEEDINGS{BleiHoSommerfeld2002, author = {. and . and .}, title = {A stochastic droplet collision model with consideration of impact efficiency}, booktitle = {ILASS--Europe 2002, Zaragoza, 9.--11. September 2002}, year = {2002}, } @INPROCEEDINGS{RombachSpecker2003, author = {. and .}, editor = {. and .}, title = {Segmentbr\"{u}cken}, booktitle = {Betonkarlender}, year = {2003}, volume = {II}, number = {1}, pages = {177--212}, } @INPROCEEDINGS{RombachAbendeh2004, author = {. and .}, title = {Temperature Induced Deformations in March--Cast Segments}, booktitle = {IABSE Symposium `Metropolitan Habitats and Infrastructures`, Shanghai, 22--24.09.2004}, year = {2004}, } @INPROCEEDINGS{RombachNeumann2004, author = {. and .}, title = {3--D Finite Element method modelling of granular flow in solos}, booktitle = {EM2004: 17th ASCE Engineering Mechanics Conference, June 13--16 2004, University of Delaware, Newark, DE}, year = {2004}, } @INPROCEEDINGS{WanTurekRivkind2003a, author = {. and . and .}, editor = {. and . and . and .}, title = {An Efficient Multigrid {FEM} Solution Technique for Incompressible Flow with Moving Rigid Bodies}, booktitle = {Numerical Mathematics and Advanced Applications}, publisher = {Springer, Berlin}, year = {2003}, pages = {844--853}, note = {Enumath 2003 Prague; ISBN-Nr. 3-540-21460-7}, } @INPROCEEDINGS{OuazziTurek2003f, author = {. and .}, editor = {. and .}, title = {Numerical Methods and Simulation Techniques for Flow with Shear and Pressure Dependent Viscosity}, booktitle = {Numerical Mathematics and Advanced Applications}, publisher = {Springer, Berlin}, year = {2003}, pages = {668--676}, month = aug, note = {Enumath 2003 Prague; ISBN-Nr. 3-540-21460-7}, } @INPROCEEDINGS{MoellerKuzminTurek2003, author = { .}, editor = {. and .}, title = {Implicit {FEM}--FCT algorithm for compressible flows}, booktitle = {Numerical Mathematics and Advanced Applications}, year = {2003}, pages = {641--650}, month = sep, note = {Enumath 2003; ISBN-Nr. 3-540-21460-7}, } @INPROCEEDINGS{DouglasHuIskandarani2000, author = {. and .}, editor = {Neittaanm\"{. and . and .}, title = {Preprocessing costs of cache based multigrid}, booktitle = {Proceedings of the Third European Conference on Numerical Mathematics and Advanced Applications}, publisher = {World Scientific}, year = {2000}, pages = {362--370}, } @INPROCEEDINGS{Keyes2000, author = {.}, title = {Four Horizons for Enhancing the Performance of Parallel Simulations based on Partial Differential Equations}, booktitle = {Europar 2000 Parallel Processing}, year = {2000}, volume = {1900}, pages = {1--17}, series = {Lecture Notes in Computer Science}, } @INPROCEEDINGS{Keyes2000a, author = {.}, editor = {. and .}, title = {Trends in Algorithms for Nonuniform Applications on Hierarchical Distributed Architectures}, booktitle = {Proceedings of the Workshop on Computational Aerosciences for the 21st Century}, publisher = {Kluwer}, year = {2000}, pages = {103--137}, address = {Dordrecht}, } @INPROCEEDINGS{GroppKaushikKeyesSmith1999, author = { .}, title = {Towards Realistic Performance Bounds for Implicit {CFD} Codes}, booktitle = {Proceedings of Parallel CFD`99}, year = {1999}, pages = {241--248}, } @INPROCEEDINGS{MarkGlanvilleAkeleyKilgard2003, author = { .}, title = {Cg: A System for Programming Graphics Hardware in a C--like Language}, booktitle = {Proceedings of ACM SIGGRAPH 2003}, publisher = {ACM Press}, year = {2003}, pages = {896--907}, } @INPROCEEDINGS{KruegerWestermann2003, author = {. and .}, title = {Linear algebra operators for {GPU} implementation of numerical algorithms}, booktitle = {ACM Transactions on Graphics (TOG)}, year = {2003}, volume = {22}, number = {3}, pages = {908--916}, } @INPROCEEDINGS{BolzFarmerGrinspunSchroeder2003, author = {. and Schr\"{.}, title = {Sparse Matrix Solvers on the GPU: Conjugate Gradients and Multigrid}, booktitle = {Proceedings of ACM SIGGRAPH 2003 {} {}}, year = {2003}, volume = {22}, number = {3}, pages = {917--924}, } @INPROCEEDINGS{Zumbusch1999, author = {.}, title = {Dynamic loadbalancing in a lightweight adaptive parallel multigrid {PDE} solver}, booktitle = {Proceedings of 9th SIAM Conference on Parallel Processing for Scientific Computing, San Antonio, Texas}, year = {1999}, pages = {10}, } @INPROCEEDINGS{WhaleyDongarra1998, author = {. and .}, title = {Automatically Tuned Linear Algebra Software}, booktitle = {SuperComputing 1998: High Performance Networking and Computing}, year = {1998}, } @INPROCEEDINGS{HysingTurek2005, author = {. and .}, title = {The Eikonal equation: numerical efficiency vs. algorithmic complexity on quadrilateral grids}, booktitle = {Accepted to be published in the conference Proceedings of Algoritmy 2005, Conference on Scientific Computing.}, year = {2005}, } @INPROCEEDINGS{WathenLoghinKayElmanSilvester2003, author = {. and . and . and .}, editor = {. and . and . and .}, title = {A New Preconditioner for the Oseen Equations}, booktitle = {Proceedings of ENUMATH 2001, the 4th European Conference on Numerical Mathematics and Advanced Applications, Ischia, July 2001}, publisher = {Springer, Berlin}, year = {2003}, volume = {XXIV}, pages = {979--988}, month = jun, note = {ISBN: 88-470-0180-3 }, } @INPROCEEDINGS{GanesanTobiska2005, author = {.}, title = {Finite Element Simulation of a Droplet Impinging a Hrizon}, booktitle = {Proceedings of Algoritmy 2005}, year = {2005}, pages = {1--11}, address = {Institut for Analysis and Numerik, Otto-Von-Guericke-Universit\"{a}t Magdeburg, D-39106 Magdeburg.}, } @INPROCEEDINGS{HronTurek2005, author = {. and .}, editor = {. and .}, title = {A Monolithic {FEM} Solver for ALE Formulation of Fluid Structure Interaction with Configurations for Numerical Benchmarking}, booktitle = {Computational Methods for Coupled Problems in Science and Engineering}, year = {2005}, volume = {First Edition}, pages = {148}, note = {Konferenzband `First International Conference on Computational Methods for Coupled Problems in Science and Engineering` (Santorini, May 25th - 27th)}, } @INPROCEEDINGS{SivakumarNishiyama.H.2004, author = {. and .}, title = {Numerical Analysis on the Impact Behavior of Molten Metal Droplets Using a Modified Splat--Quench Solidification Model}, booktitle = {Transactions of the ASME}, year = {2004}, volume = {126}, pages = {1014--1022}, address = {Electromagnetic Intelligent Fluids Laboratory, Institute of Fluid Science , Tohoku University, 2-1-1 Katahira, Aoba-Ku, Sendai 980-8577, Japan}, } @INPROCEEDINGS{AuberyCharunyakornPutivisutisak2005, author = { .}, title = {A New Approach for the Simulation of Multi--Louver Fin Surfaces}, booktitle = {The 8th Asian Symposium on Visualization {}}, year = {2005}, pages = {(36-1)-(36-6)}, address = {Department of Mechanical Engineering , Faculty of Engineering, Chulalongkorn University, Bangkok, Thailand}, } @INPROCEEDINGS{TurekAltieriBeckerKilianOswaldWallis1998, author = { .}, title = {Some Basic Concepts of FEAST}, booktitle = {Proc. 14th GAMM Seminar `Concepts of Numerical Software`}, publisher = {Vieweg}, year = {1998}, series = {to appear in: Notes on Numerical Fluid Mechanics}, } @INPROCEEDINGS{BeckerRannacher2001, author = {. and .}, title = {An optimal control approach to a posteriori error estimates in finite element methods}, booktitle = {Acta Numerica}, publisher = {Cambridge University Press}, year = {2001}, pages = {1--102}, } @INPROCEEDINGS{DouglasDupond1976, author = {. and .}, title = {Interiour penalty procedures for elliptic and parabolic {G}alerkin methods}, booktitle = {Computing methods in applied sciences (2nd Int. Sympo., Versailles, 1975)}, publisher = {Springer, Berlin}, year = {1976}, volume = {Lecture Note in Physics}, number = {58}, note = {Lecture Note in Physics}, } @INPROCEEDINGS{BieglerGhattasHeinkenschlossvanBloemenWaanders2003, author = {. and .}, title = {Large--Scale {PDE}--Constrained Optimization}, booktitle = {Lecture Notes in Computational Science and Engineering}, publisher = {Springer, Berlin}, year = {2003}, } @INPROCEEDINGS{GalopppoGovindarajuHensonManocha2005, author = { .}, title = {LU--GPU: Efficient Algorithm for Solving Dense Linear Systems on Graphics Hardware}, booktitle = {ACM/IEEE SC/05}, publisher = {ACM Press}, year = {2005}, note = {\url{http:/ / gamma.cs.unc.edu/ LU-GPU/ lugpu05.pdf}}, } @INPROCEEDINGS{GoeddekeStrzodkaTurek2005b, author = {.}, editor = {H\"{u}. and . and R\"{u}.}, title = {Accelerating Double Precision {FEM} Simulations with {GPUs}}, booktitle = {Simulationstechnique 18th Symposium in Erlangen, September 2005}, publisher = {SCS Publisching House e.V.}, year = {2005}, volume = {Frontiers in Simulation}, pages = {139--144}, note = {ASIM 2005}, } @INPROCEEDINGS{BurnsDaoudVaigl1994, author = { .}, editor = {.}, title = {{LAM: An Open Cluster Environment for MPI}}, booktitle = {{Proceedings of Supercomputing Symposium}}, publisher = {nn}, year = {1994}, pages = {379--386}, note = {\url{http:/ / www.lam-mpi.org/ download/ files/ lam-papers.tar.gz}}, } @INPROCEEDINGS{SquyresLumsdaine2003, author = {.}, editor = {. and .}, title = {{A Component Architecture for LAM/MPI}}, booktitle = {{Recent Advances in Parallel Virtual Machine and Message Passing Interface, 10th European PVM/MPI Users` Group Meeting, Venice, Italy, September 29 - October 2, 2003, Proceedings}}, publisher = {Springer, Berlin}, year = {2003}, number = {2840}, pages = {379--387}, series = {Lecture Notes in Computer Science}, month = sep, } @UNPUBLISHED{Pinnau2001, author = {.}, title = {H\"{o}here Numerik II {}--{} Theorie und Numerik elliptischer Differentialgleichungen}, year = {2001}, note = {Vorlesungsskript, TU Darmstadt}, } @UNPUBLISHED{Foerster1994, author = {.}, title = {Analysis und Numerik}, year = {1994}, note = {Vorlesungsskript Universit\"{a}t Hildesheim}, } @UNPUBLISHED{Buijssen2002a, author = {.}, title = {3D Grid Generation}, year = {2002}, month = mar, note = {Handout FeatFlow Springschool 2002}, } @UNPUBLISHED{Turek2001d, author = {.}, title = {Numerische Methoden f\"{u}r Partielle Differentialgleichungen}, year = {2001}, note = {Vorlesungsskript}, } @UNPUBLISHED{InternationalCenterforNumericalMethodsinEngineering2003, author = {International Center for Numerical Methods in Engineering}, title = {{GiD} {}--{} The personal pre and post processor}, year = {2003}, note = {\url{http:/ / gid.cimne.upc.es}}, } @UNPUBLISHED{AdvancedVisualSystems2003, author = {Advanced Visual Systems}, title = {{AVS/Express}}, year = {2003}, note = {\url{http:/ / www.avs.com}}, } @UNPUBLISHED{Ortega2003, author = {.}, title = {General Mesh Viewer}, year = {2003}, note = {\url{http:/ / www-xdiv.lanl.gov/ XCM/ gmv/ }}, } @UNPUBLISHED{LuberHastreiter1999, author = {}, title = {Script of Lecture Scientific Visualization}, year = {1999}, note = {Universit\"{a}t Erlagen}, } @UNPUBLISHED{Kruska2000, author = {Kruska}, title = {{Wissenschaftliche Visualisierung, Skript zur Vorlesung}}, year = {2000}, note = {\url{http:/ / phong.informatik.uni-leipzig.de/ kruska/ visuallecture.html}}, } @UNPUBLISHED{Rannacher2001a, author = {.}, title = {{Numerik gew\"{o}hnlicher Differentialgleichungen}}, year = {2001}, note = {\url{http:/ / gaia.iwr.uni-heidelberg.de/ httpdoc/ Lectures/ Notes/ notes.html}}, } @UNPUBLISHED{Ansys2004, author = {ANSYS}, title = {CFX--5: Features}, year = {2004}, note = {\url{http:/ / www.ansys.com/ products/ cfx-features.asp}}, } @UNPUBLISHED{Cilea2004, author = {Cilea}, title = {Fluent Suite}, year = {2004}, note = {\url{http:/ / www.cilea.it/ SERVIZI/ B/ CFD/ FluentSuite/ fluent_main_eng.htm}}, } @UNPUBLISHED{PETSc2004, author = {PETSc}, title = {PETSc: Features}, year = {2004}, note = {\url{http:/ / www-unix.mcs.anl.gov/ petsc/ petsc-2/ features/ index.html}}, } @UNPUBLISHED{DouglasRuedeHuBittemcourt1998, author = {. and R\"{u}. and . and .}, title = {A guide to designing cache aware multigrid algorithms}, year = {1998}, note = {Submitted}, } @UNPUBLISHED{DouglasRuedeHuBittencourt1998, author = {. and .}, title = {A guide to designing cache aware multigrid algorithms}, year = {1998}, note = {Report}, } @MISC{Becker1998, author = {.}, title = {FEAST {}--{} Finite Element \& Solution Tools}, note = {Vortrag, Chemnitzer FEM-Symposion, Chemnitz, Germany}, year = {1998}, month = sep, } @MISC{Turek2001a, author = {.}, title = {On the Next Generation of {CFD} Tools}, note = {Vortrag, Fluid Conference, Vancouver}, year = {2001}, month = aug, } @MISC{Turek2001b, author = {.}, title = {Hardware--oriented Numerics for PDEs {}--{} Consequences for Incompressible Flow Solvers}, note = {Vortrag, MIT-Conference, Boston}, year = {2001}, month = jun, } @MISC{Turek2001c, author = {.}, title = {Computational Prediction of Natural Convection Flows in Enclosures with FeatFlow}, note = {Vortrag, MIT-Conference, Boston}, year = {2001}, month = jun, } @MISC{Turek2001e, author = {.}, title = {Introduction to Hardware--Oriented Numerics (for PDEs)}, note = {Vortrag, GAMM2001, Zuerich}, year = {2001}, month = feb, } @MISC{Turek2000b, author = {.}, title = {Efficient Simulation of Nonsteady Incompressible Flow}, note = {Vortrag, ALGORITMY2000, Slovakia}, year = {2000}, month = sep, } @MISC{Turek2000c, author = {.}, title = {Concepts of High Performance {FEM} Simulation and Realization in the FEAST Software}, note = {Vortrag, IMACS2000, Lausanne}, year = {2000}, month = aug, } @MISC{Turek2000e, author = {.}, title = {On the construction of Discrete Projection Methods}, note = {Vortrag, FluidMechanics Conference due to J. Heywood///`s birthday, Napoli}, year = {2000}, month = may, } @MISC{Turek2000f, author = {.}, title = {Efficient numerical solution of incompressible flow problems {}--{} Towards adaptive error control}, note = {Vortrag, BAMC, Manchester}, year = {2000}, month = apr, } @MISC{Turek2000g, author = {.}, title = {On adaptive time stepping for incompressible flow}, note = {Vortrag, GAMM Workshop, Kiel}, year = {2000}, } @MISC{Turek1999, author = {.}, title = {Realistic Performance Evaluation of Numerical Simulations}, note = {Vortrag, SPEC Workshop, Wuppertal}, year = {1999}, month = sep, } @MISC{Turek1999a, author = {.}, title = {Multilevel Pressure Schur Complement methods for solving the incompressible {Navier}--{Stokes} equations: The case of GLOBAL MPSC}, note = {Vortrag, ICIAM 99, Edinburgh}, year = {1999}, month = jul, } @MISC{Turek1999b, author = {.}, title = {Trends in processor technology and their impact on Numerics for PDE`s}, note = {Vortrag, FVCA II, Duisburg}, year = {1999}, month = jul, } @MISC{Turek1999e, author = {.}, title = {On implementation techniques for boundary conditions: Applications to {CFD}}, note = {Vortrag, ENUMATH99, Yvaeskylae}, year = {1999}, month = jul, } @MISC{Turek1999f, author = {.}, title = {Realistic Performance Evaluation of Numerical Simulations}, note = {Vortrag, Mannheimer Supercomputer-Tage 1999, Mannheim}, year = {1999}, month = jun, } @MISC{Turek1999g, author = {.}, title = {Simulationswerkzeuge fuer Stroemungsvorgaenge im Automobilbau}, note = {Vortrag, BMBF Statusseminar 1999, Erlangen}, year = {1999}, month = feb, } @MISC{Kuzmin2000, author = {.}, title = {Efficient numerical techniques for flow simulation in Bubble Column reactors}, note = {Vortrag, 5th German-Japanese Symposium on Bubble Columns , Hotel Mercure, Dresden, Germany}, year = {2000}, month = may, } @MISC{Kuzmin2000a, author = {.}, title = {Numerical Tools for Simulation of Gas--Liquid Reactors {}}, note = {Vortrag, Computational Methods for Multidimensional Reactive Flows, University of Heidelberg, Germany}, year = {2000}, month = sep, } @MISC{Kuzmin2000b, author = {.}, title = {{FEM}--Simulationen von Blasenstr\"{o}ungen in Gas--Fl\"{u}ssigkeits--Reaktoren}, note = {Vortrag, DFG-Kolloquium, SPP Analyse, Modellbildung und Berechnung mehrphasiger Str\"{o}mungen}, year = {2000}, month = oct, } @MISC{Kuzmin2000c, author = {.}, title = {Finite Element Simulation of Gas--Liquid Reactors}, note = {Vortrag, 5th French-Russian-Finnish Workshop on Experimentation, Mathematical Modeling and Computation in Engineering SciencesUniversity of Jyvaskyla, University of Jyvaskyla, Finland}, year = {2000}, month = jun, } @MISC{BuijssenTurek2002, author = {. and .}, title = {Sources of Parallel Inefficiency for Incompressible {CFD} Simulations}, note = {Vortrag, EuroPar 2002}, year = {2002}, month = aug, } @MISC{BeckerKilianTurek1998a, author = {. and . and .}, title = {Some concepts of the software package FEAST}, note = {Vortrag, Vector and Parallel Processing - VECPAR98, Porto, Portugal}, year = {1998}, month = jun, } @MISC{BeckerKilianTurek1999a, author = {. and . and .}, title = {Consequences of modern hardware design for numerical simulation and their realization in FEAST}, note = {Vortrag, EuroPar99 Parallel Processing, Toulouse, France}, year = {1999}, month = aug, } @MISC{BeckerKilianTurek2001, author = {. and . and .}, title = {Experiences with the High Performance {FEM} package FEAST}, note = {Vortrag, International FORTWIHR conference, Erlangen, Germany}, year = {2001}, month = mar, } @MISC{ZietheSpohnTurek2002, author = {. and . and .}, title = {Core Formation: A New Modelling Approach}, note = {Vortrag, 36th ESLAB Symposion Earth-like Planets and Moons, Nordwijk}, year = {2002}, } @MISC{Turek1994i, author = {.}, title = {Tools For Predicting Incompressible Flows Via Nonconforming Finite Element Models}, note = {Vortrag, X International Conference on Computational Methods in Water Resources, SFB 359}, year = {1994}, month = jul, } @MISC{SchreiberTurek1993a, author = {. and .}, title = {An efficient finite element solver for the nonstationary incompressible {Navier}--{Stokes} equations in two and three dimensions}, note = {Vortrag, International Workshop on Numerical Methods for the Navier-Stokes Equations}, year = {1993}, } @MISC{Buijssen2002b, author = {Buijssen, .}, title = {3D Grid Generation}, note = {Vortrag, FeatFlow SpringSchool 2002}, year = {2002}, month = mar, } @MISC{TurekWanRivkind2002, author = {. and . and .}, title = {The Fictitious Boundary Method for the implicit treatment of Dirichlet boundary conditions with applications to incompressible flow simulation}, note = {Vortrag, CICS Berlin}, year = {2002}, month = oct, } @MISC{SchmachtelTurek2001, author = {. and .}, title = {Numerical Analysis for the Linearized Incompressible {Navier}--{Stokes} Equations and the Newton Method}, note = {Vortrag, AMFLOW, Heidelberg}, year = {2001}, } @MISC{Turek2004, author = {.}, title = {{FEM} Techniques for in Flow Simulation}, note = {Vortrag, ASIM}, year = {2004}, month = mar, } @MISC{Turek2004a, author = {.}, title = {{FEM} Techniques for Multiphase Flow Simulation}, note = {Vortrag, International Workshop: Transport in Fluid Multiphase Systems From Experimental Data to Machnistics Models, SFB 540 Aachen}, year = {2004}, month = mar, } @MISC{Buijssen2004, author = {Buijssen, .}, title = {{FEM} Techniques and High Performance Computing Approaches for Incompressible Flow Simulation}, note = {Vortrag, Vortrag im Rahmen der Graduate School}, year = {2004}, month = apr, } @MISC{BuijssenTurek2003a, author = {. and .}, title = {parpp3d++ {}--{} a Parallel {HPC} Research Code for {CFD}}, note = {Vortrag, 6th HLRS Workshop, Stuttgart}, year = {2003}, month = oct, } @MISC{KoesterTurek2004, author = {. and .}, title = {An Optimal--Order--Multigrid Method for Quadratic Conforming Finite Element {}}, note = {Vortrag, Vortrag Ehrenfriedersdorf, 17th Chemnitz FEM Symposium 2004 vom 20. - 22. September 2004 }, year = {2004}, month = sep, } @MISC{OuazziTurek2001, author = {. and .}, title = {Multigrid Methods for Stabilized Nonconforming Finite Element for Incompressible Flow}, note = {Vortrag, AMFLOW 2001, Workshop on Adaptive Methods for Flow Computation}, year = {2001}, month = oct, } @MISC{OuazziTurek2003a, author = {. and .}, title = {Efficient Numerical Methods and Simulation Techniques for Granular Flow}, note = {Vortrag, DFG Assistant Workshop, at Technical University of Braunschweig }, year = {2003}, month = feb, } @MISC{OuazziTurek2003b, author = {. and .}, title = {Numerical Methods and Simulation Techniques for flow with pressure dependent viscosity {}}, note = {Vortrag, Workshop Uni Freiburg, }, year = {2003}, month = apr, } @MISC{OuazziTurek2004a, author = {. and .}, title = {Efficient Numerical Methods and Simulation Techniques for Granular flow}, note = {Vortrag, DFG Assistant Workshop, at Karlsruhe University }, year = {2004}, month = sep, } @MISC{GrajewskiTurek2004, author = {. and .}, title = {Concepts of patchwise mesh refinement {} in the context of DWR {}}, note = {Vortrag, 17th Chemnitz FEM Symposium}, year = {2004}, month = sep, } @MISC{OuazziTurek2003c, author = {. and .}, title = {Numerical Methods and Simulation Techniques for Flow with Shear and Pressure dependent Viscosity {}}, note = {Vortrag, Enumath, at Charles University Prague Institute of Chemical Technology, PRAGUE, CZECH REPUBLIC }, year = {2003}, month = aug, } @MISC{OuazziTurek2003e, author = {. and .}, title = {Efficient Numerical Methods and Simulation Techniques for Granular Flow}, note = {Vortrag, DFG Assistent Workshop , TU M\"{u}nchen}, year = {2003}, month = sep, } @MISC{KashidPlatteAgarTurek2004, author = {. and . and . and .}, title = {Computational modelling of slug flow in a capillary millireactor}, note = {Vortrag, First Indo-German conference on `PDE scientific computing and optimization in Applications`, 08. - 10. September 2004 in Trier}, year = {2004}, month = sep, } @MISC{OuazziTurek2004b, author = {. and .}, title = {Edge--oriented stabilisation for nonconforming finite element methods {}}, note = {Vortrag, Miniworkshop G\"{o}ttingen}, year = {2004}, month = dec, } @MISC{Hysing2005, author = {.}, title = {The Eikonal Equations}, note = {Vortrag, Algoritmy}, year = {2005}, month = mar, } @MISC{Turek2005, author = {.}, title = {{FEM} Techniques for Incompressible Flow Problems with Time Dependent Interfaces}, note = {Vortrag, Oberwolfach}, year = {2005}, month = feb, } @MISC{MoellerKuzminTurek2004a, author = {. and . and .}, title = {An Iterative {FEM}--FCT Algorithm for the Compressible Euler Equations}, note = {Vortrag, European Congress on Computational Methods in Applied Sciences and Engineering, 24.-28.-7.2004, University of Jyv\"{a}skyl\"{a}, Finnland.}, year = {2004}, month = jul, } @MISC{MoellerKuzminTurek2004b, author = {. and . and .}, title = {Implicit Finite Element Discretizations Based on the Flux--Corrected Transport Algorithm}, note = {Vortrag, ICFD Conference on Numerical Methods for Fluid Dynamics , 29.3.-1.4. 2004, University of Oxford, England }, year = {2004}, month = mar, } @MISC{MoellerKuzmin2003, author = {. and .}, title = {An implicit {FEM}--FCT algorithm for scalar transport problems and the compressible Euler equations}, note = {Vortrag, Workshop `High-resolution schemes for convection-dominated flows: 30 years of FCT` , September 29-30, 2003, University of Dortmund}, year = {2003}, month = sep, } @MISC{MoellerKuzmin2005, author = {. and .}, title = {Adaptive Grid Refinement for High--Resolution Finite Element Schemes based on Algebraic Flux Correction}, note = {Vortrag, FEF05 - Thirteenth Conference on Finite Elements for Flow Problems, 4. - 6. April 2005}, year = {2005}, month = apr, } @MISC{Becker2005, author = {.}, title = {Hardware--oriented numerics and the FEAST framework}, note = {Vortrag, Workshop: Supercomputers on a chip GPUs as Mathematical Coprocessors in Finite-Element-Simulations}, year = {2005}, month = apr, } @MISC{Goeddeke2005, author = {G\"{.}, title = {The {GPU} as a {FEM}--coprocessor: Algorithmic design goals}, note = {Vortrag, Workshop: Supercomputers on a chip GPUs as Mathematical Coprocessors in Finite-Element-Simulations}, year = {2005}, month = apr, } @MISC{Turek2005a, author = {.}, title = {Effiziente Diskretisierungs--{} und L\"{o}sungsmethoden f\"{u}r Lattice--Boltzmann Gleichungen}, note = {Vortrag, LB-Arbeitsgruppe Workshop}, year = {2005}, month = apr, } @MISC{Turek2005b, author = {.}, title = {Hardware--Oriented Numerics {}--{} High Performance {FEM} Simulation of PDEs}, note = {Vortrag, Kolloquiumsvortrag D\"{u}sseldorf}, year = {2005}, month = apr, } @MISC{Koester2005, author = {K\"{.}, title = {Optimization Techniques for Incompressible Flow Problems {}}, note = {Vortrag, Graduate School}, year = {2005}, month = apr, } @MISC{Turek2005c, author = {.}, title = {{FEM} Techniques for Particulate Flow}, note = {Vortrag, International Workshop on Current Topics in Mathematical Fluid Mechanics MFM 2005, 17. - 18. Juni 2005, IST-Lisbon, Portugal. In honor of Professor }, year = {2005}, month = jun, } @MISC{TurekHronKoesterOuazzi2005, author = {. and . and K\"{o}. and .}, title = {Edge--Oriented {FEM} Stabilization Techniques for Incompressible Flow}, note = {Vortrag, Enumath}, year = {2005}, month = jul, } @MISC{TurekWan2005, author = {. and .}, title = {{FEM} Techniques for Particulate Flow}, note = {Vortrag, Enumath}, year = {2005}, month = jul, } @MISC{KuzminMoeller2005, author = {. and M\"{o}.}, title = {On the use of Limiting Techniques for Adaptive Mesh Refinement}, note = {Vortrag, Enumath 2005}, year = {2005}, month = jul, } @MISC{HronTurek2005a, author = {. and .}, title = {A monolithic multigrid {FEM} solver for FSI}, note = {Vortrag, FSW Workshop Erlangen}, year = {2005}, month = jun, } @MISC{HronTurek2005b, author = {. and .}, title = {Proposal for the numerical FSI benchmarks}, note = {Vortrag, FSW Workshop Erlangen}, year = {2005}, month = jun, } @MISC{OuazziTurek2004c, author = {. and .}, title = {Efficient Numerical Methods and Simulation Techniques for Granular flow}, note = {Vortrag, DFG Assistant Workshop, at Duisburg-Essen University}, year = {2004}, month = feb, } @MISC{HronTurek2005c, author = {. and .}, title = {A monolith {FEM} solver for fluid structure interaction}, note = {Vortrag, Ecomas Thematic Conference: Coupled Problems 2005 Computational Methods for Coupled Problems in Science and Engineering, 25-27 May 2005, Santorini Island, Greece}, year = {2005}, month = may, } @MISC{Hysing2005c, author = {.}, title = {A new implicit surface tension implementation for interfacial flows}, note = {Vortrag, Level Set Workshop on Direct and Inverse Problems}, year = {2005}, month = sep, } @MISC{Goeddeke2005a, author = {.}, title = {Introduction to data--stream based computations on graphics hardware}, note = {Vortrag, ASIM 2005 - 18th Symposium on Simulation Technique Workshop Parallel Computing and Graphics Processors, Erlangen, Germany, September 12th-15th 2005}, year = {2005}, month = sep, } @MISC{GoeddekeStrzodkaTurek2005a, author = {. and .}, title = {Accelerating Double Precision {FEM} Simulations with {GPUs}}, note = {Vortrag, ASIM 2005 - 18th Symposium on Simulation Technique Workshop Parallel Computing and Graphics Processors, Erlangen, Germany, September 12th-15th 2005}, year = {2005}, month = sep, } @MISC{TurekHronKoesterOuazzi2005a, author = {. and . and K\"{o} .}, title = {Edge--oriented {FEM} Stabilization Accuracy, Robustness and Efficient Solvers}, note = {Vortrag, 8th European Multigrid Conference Eccomas Thematic Conference on Multigrid, Multilevel and Multiscale Methods, Scheveningen, The Hague vom 27. - 30. September 2005}, year = {2005}, month = sep, } @MISC{Buijssen2003b, author = {Buijssen, .}, title = {parpp3d++ {}--{} a Parallel {HPC} Research Code for CFD: Block Smoothers in Parallel Multigrid Methods, Hitachi SR8000 vs. Linuxcluster}, note = {Vortrag, 6th HLRS Workshop, Stuttgart}, year = {2003}, month = oct, } @MISC{Becker2002, author = {.}, title = {FEAST: The Future}, note = {Vortrag, Featflow Springschool}, year = {2002}, month = mar, } @TECHREPORT{BlumHarigMueller1990, author = {. and . and M\"{u}.}, title = {Finite element analysis tools}, type = {{Preprints SFB 123}, {Nr.} 554}, year = {1990}, note = {Release 1.1 User Manual}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Zhou1995, author = {.}, title = {How accurate is the streamline--diffusion method?}, type = {{Preprints SFB 359}, {Nr.} 95-22}, year = {1995}, note = {Math. Comp. 66, 31-44}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Mueller1975, author = {.}, title = {Approximation unbeschr\"{a}nkter Funktionen bez\"{u}glich einer Korovkin--Metrik}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 001}, year = {1975}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1976, author = {.}, title = {Auswertungsalgorithmen bester numerischer Stabilit\"{a}t f\"{u}r Poynome}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 002}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1976, author = {.}, title = {On the degree of Lp--approximation by integral Schoenberg splines}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 003}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1976a, author = {.}, title = {On the multivariate polynomials of least deciation from zero on the unit ball}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 004}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1976a, author = {M\"{u}ller, .}, title = {Die G\"{u}ter der Lp--Approximation durch Kantorovic--Polynome}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 006}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fromm1976, author = {.}, title = {L1--approximation to zero}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 007}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1976b, author = {.}, title = {Lp--approximation by the method of integral Meyer--K\"{o}nig and Zeller operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 008}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1976c, author = {M\"{u}ller, .}, title = {Degree fo approximation by integral Schoenberg splines}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 009}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Ulrich1976, author = {.}, title = {Ein Reduktionsverfahren der konvexen Optimierung}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 010}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1976b, author = {.}, title = {On mulitvariate polynomials of least deviation from zero on the unit cube}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 012}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{VeselicWeidmann1977, author = {. and .}, title = {Potential scattering in a homogeneous electrostatic field}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 013}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ReimerPittnauer1977, author = {. and .}, title = {Intervall--Funktionale von {Gauss}--Legandre--Typ}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 014}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Maier1977, author = {.}, title = {Lp--approximation by Kontorovic operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 015}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KustererReimer1977, author = {. and .}, title = {Stable evaluation of polynomials in time log n.}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 016}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Grundmann1977, author = {.}, title = {Saturation theorems for Bernstein polynomials in Banachsubspaces fo C (I,IR)}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 017}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1977, author = {.}, title = {On a class of {Jacobi}--like procedures for diagonalizing arbitrary real matrices}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 018}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Grundmann1977a, author = {.}, title = {Approximation of the identity in linear normed spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 019}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1977a, author = {.}, title = {A note on a new class of elementary matrices}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 020}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{AqanovicVeselic1977, author = {. and .}, title = {On a singular contact problem for two--dimensional Laplace equation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 021}, year = {1977}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1978, author = {.}, title = {On real eigenvalues of real tridiagonal matrices}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 023}, year = {1978}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1978a, author = {.}, title = {On optimal linearisation of a quadratic eigenvalue problem}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 024}, year = {1978}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1978, author = {.}, title = {Numerische Behandlung von Verzweigungsproblemen bei gew\"{o}hnlichen Differentialgleichungen}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 025}, year = {1978}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1978b, author = {.}, title = {A global {Jacobi} method for a symmetric indefinite Problem Sx= lamda Tx}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 026}, year = {1978}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1979, author = {.}, title = {Numerische Behandlung von Verzweigungsproblemen bie gew\"{o}hnlichen Randwertaufgaben}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 027}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1979, author = {.}, title = {On the efficient solution of nonlinear finite element equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 029}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1979a, author = {.}, title = {{Zur numerischen Behandlung gest\"{o}rter Verzweigungsprobleme}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 030}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1979b, author = {.}, title = {{Eine Verallgemeinerung eines Satzes von Crandall und Rabinowitz aus der Verzweigungstheorie}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 031}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{WeberWerner1979, author = {. and .}, title = {On the accurate determination on nonisolated solution of nonlinear equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 032}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1979c, author = {.}, title = {Numerical solution of Hopf bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 033}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MeyerVeselic1979, author = {. and .}, title = {On some new inclusion theorems for eigenvalues of partitioned matrices}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 034}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MaierSwetitsMueller1979, author = {. and . and M\"{.}, title = {The local L1 saturation class of integrated Meyer--K\"{o}nig and Zeller operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 035}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Veselic1979a, author = {.}, title = {On the chaqracterisation of the bound and the scattering states for time--dependent Hamiltonians}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 036}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1979e, author = {.}, title = {Shooting methods for bifurcations problemns in ordinary differential equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 037}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1979a, author = {.}, title = {An efficient algortihm for large sparse nonlinear programming problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 038}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Grundmann1979, author = {.}, title = {{Bemerkungen zur Whittaker-Konstanten}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 039}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1979, author = {.}, title = {Best approximations to polynomials in the mean and norms of coefficient--functionals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 041}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1979a, author = {.}, title = {Extremal bases for normed vector spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 042}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1980, author = {.}, title = {An extension of the Freud--Popov lemma}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 043}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suendermann1980, author = {.}, title = {Lebesgue constants in Lagrangrian interpolation at the Fekete points}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 044}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MittelmannWeber1980, author = {. and .}, title = {Numerical methods for bifurcation problems {}--{} a survey and classification}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 045}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1980, author = {.}, title = {Numerical computation of the Fourier transform using Laguerre functions and the Fast Fourier Transform}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 046}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1980, author = {.}, title = {On the efficient solution of nonlinear finite element problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 047}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1980a, author = {.}, title = {On the nurmerical approximation of secondary bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 048}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1980a, author = {.}, title = {On the numerical solution of contact problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 049}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Grundmann1980, author = {.}, title = {Langsame Approximation der Identit\"{a}t}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 050}, year = {1980}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1981, author = {M\"{u}ller, .}, title = {{\"{U}ber die Methode der integrierten Schoenberg-Splines}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 051}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{RackReimer1981, author = {. and .}, title = {The numerical stability of Lagrangian--form--evaluation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 052}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{WeberWerner1981, author = {. and .}, title = {On the numerical solution of some finite--dimensional bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 053}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1981, author = {.}, title = {Extremal spline bases}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 054}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Weber1981, author = {.}, title = {Numerical solution of a class of nonlinear bundary value problems for analytic functions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 055}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Hackbusch1981, author = {.}, title = {On Multi--grid methods for variational inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 057}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1982, author = {.}, title = {Multi--grid methods for simple bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 058}, year = {1982}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1982a, author = {.}, title = {A fast solver for nonlinear eigenvalue problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 059}, year = {1982}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1982, author = {.}, title = {On the uniform norm of an aribtrary coefficient functional over finite dimensional subspace of C(D)}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 061}, year = {1982}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1983, author = {.}, title = {The radius of convergence of a cardinal Lagrangian--Spline series of odd degree}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 062}, year = {1983}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1983a, author = {.}, title = {Best constants occuring with the modulus of continuity in the error estimate for spline interpolants of odd degree on equidistant grids}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 063}, year = {1983}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mittelmann1983, author = {.}, title = {Multigrid solution of bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 065}, year = {1983}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuepperKuszta1983, author = {. and .}, title = {Feedback stimulated bifurcation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} }, year = {1983}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1984, author = {.}, title = {{Optimale Knoten f\"{u}r K-fache Lagrangesche Numerische Differentation}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 067}, year = {1984}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suendermann1984, author = {.}, title = {On Projection Constants of Polynomial Spaces on the unit ball in several Variables}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 068}, year = {1984}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Grundmann1984, author = {.}, title = {{Die Whittakerkonstante}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 069}, year = {1984}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1984, author = {.}, title = {The main roots of the Euler--Frobenius polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 070}, year = {1984}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MuellerHaeusler1984, author = {M\"{. and H\"{.}, title = {{Der Remez-Algorithmus f\"{u}r die rationale Tschebyscheff-Approximation}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 071}, year = {1984}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1985, author = {.}, title = {Interpolation on the sphere and bounds for the lagrangian square sums}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 072}, year = {1985}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1985a, author = {.}, title = {An element algebraic representation of polynomial spline interpolants for equidistant lattics and its condition}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 073}, year = {1985}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Siepmann1985, author = {.}, title = {Cardinal interpolation by polynomial Splines: Interpolation of Data with exponential Growth}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 074}, year = {1985}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ReimerSuendermann1985, author = {. and S\"{u}.}, title = {A Remez--type algorithm for the calculation of extremal fundamental systems for polynomial spaces on the sphere}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 075}, year = {1985}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1986, author = {.}, title = {Approximation by Cheney--Sharma--Kantorovic polynomials in the Lp--metric}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 077}, year = {1986}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Quak1986, author = {.}, title = {Lp--error estimates for positive linear operators using the second--order Tau--modulus}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 078}, year = {1986}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Quak1986a, author = {.}, title = {Mulitvariate Lp--error estimates for positive linear operators vis the First--order tau--modulus}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 079}, year = {1986}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ReimerSuendermann1986, author = {. and S\"{u}.}, title = {G\"{u}nstige Knoten f\"{u}r die Interpolation mit homogenen harmonischen Polynomen}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 080}, year = {1986}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1987, author = {.}, title = {Saturation of a new method of Kantorovic type Meyer--K\"{o}nig and Zeller operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 081}, year = {1987}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Quak1987, author = {.}, title = {For the numerical solution of O.D. Es by a--methods using the Tau--modulus}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 083}, year = {1987}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1988, author = {.}, title = {Optimal refinement near singularities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 087}, year = {1988}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1988a, author = {.}, title = {Die Wirkung der Radon--Transformation auf Polynomr\"{a}ume}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 088}, year = {1988}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LindeReimerSuendermann1988, author = {. and . and S\"{.}, title = {{Numerische Berechnung extremaler Fundamentalsysteme f\"{u}r Polynomr\"{a}ume \"{u}ber der Vollkugel}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 090}, year = {1988}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1988b, author = {.}, title = {Cardinal Hermite--Spline--Interpolation on the Equidistant Lattice}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 091}, year = {1988}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1988c, author = {.}, title = {{Zur reellen Darstellung periodischer Hermite-Spline-Interpolierender bei \"{a}quidistantem Gitter mit Knotenshift}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 092}, year = {1988}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1989, author = {.}, title = {Fundamentalsysteme mit lokal erf\"{u}llten Extremalbedingungen 1. und 2. Ordnung f\"{u}r R\"{a}ume sh\"{a}rischer harmonischer Polynome}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 093}, year = {1989}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{RosierSuendermann1989, author = {. and S\"{.}, title = {Numerical Inversion of the Radon Transform on Poly--nomial Spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 094}, year = {1989}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1989a, author = {.}, title = {The Construction of Periodic Birkhoff--Spline--Interpolants on the Equidistant Lattice}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 095}, year = {1989}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1990, author = {.}, title = {Representation of Spline Interpolants on the Finite Equidistant Grid}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 096}, year = {1990}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheMueller1990, author = {. and M\"{.}, title = {New Local Error Estimates for Meyer--K\"{o}nig and Zeller Kantorovic--Operators using Maximal Functions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} }, year = {1990}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mache1990, author = {.}, title = {Weighted Simultaneous Lp--Approximation by the Method of Kantorovic--Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 098}, year = {1990}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LupasMache1990, author = {. and .}, title = {Approximation by Vn--Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 099}, year = {1990}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HermannKaiser1990, author = {. and .}, title = {Shooting Methods for Two--Point BVPs with Partially Separated Endconditions. Part I.}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 100}, year = {1990}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Kubach1991, author = {.}, title = {Remark on Best Harmonic Approximation on Ellipsoids}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 101}, year = {1991}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mache1991, author = {.}, title = {Equivalence Theorem on Weighted Simultaneous Lp--Approximation by the Method of Kantorovic Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 103}, year = {1991}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Frank1992, author = {.}, title = {Fuzzy--Mengen, Fuzzy--Logik und ihre Anwendungen}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 104}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumLiskyRannacher1992a, author = {. and . and .}, title = {A Domain Splitting Algorithm for Parabolic Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} }, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HermannKaiser1992, author = {. and .}, title = {Shooting Methods for Two--Point BVPs with Partially Separated Endconditions. Part II}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 106}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1992, author = {.}, title = {On the Existence of {Gauss}--like Node--Distributions on High--Dimensional Spheres}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 107}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LupasMache1992, author = {. and .}, title = {The Degree of Approximation by a Class of Linear Positive Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 108}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Heilmann1992, author = {.}, title = {Rate of Approximation of Weighted Derivatives by Linear Combinations of SMD Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 108a}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Heilmann1992a, author = {.}, title = {Saturation of Linear Combinations of Positive Linear Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 110}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mache1992, author = {.}, title = {A Method for Summability of {Lagrange} Interpolation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 111}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Tan1992, author = {.}, title = {Interpolationg Multivariate Rational Splines of Special Forms in R.}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 112}, year = {1992}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1993, author = {.}, title = {Equally Weighted Quadrature Rules on the Sphere. Survey and Results}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 113}, year = {1993}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheZhou1993, author = {. and .}, title = {Best Direct and Converse Results by {Lagrange}--Type Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 114}, year = {1993}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheZhou1993a, author = {. and .}, title = {Characterization Theorems for the Approximation by a Family Operators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 115}, year = {1993}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Albrecht1993, author = {.}, title = {The {Runge}--{Kutta} Theory in a Nutshell}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 116}, year = {1993}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1994, author = {.}, title = {Uniform Inequalities for Gegenbauer Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 117}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1994a, author = {.}, title = {A Short Proof of a Result of Kogbetliantz on the Positivity of Certain Ces\`{a}ro Means}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 118}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1994b, author = {.}, title = {Zonal Spherical Polynomials with Minimal L1--Norm}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 119}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Rosier1994, author = {.}, title = {Biorthogonal series expansions of x--ray and k--plane transform}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 120}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LupasMacheMueller1994, author = {. and . and M\"{u}.}, title = {Weighted Lp--Approximation of Derivatives by the Mehtod of Gammaoperators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 121}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LupasMache1994, author = {. and .}, title = {On Numerical Differentiation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 122}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Felten1994, author = {.}, title = {A modulus of smoothness based on an algebraic addition}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 123}, year = {1994}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Felten1995, author = {.}, title = {Characterization of best algebraic approximation by an algebraic modulus of smoothness}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 124}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1995, author = {.}, title = {Some New Coincide Conditions in Minisum Multifacility Location Problems with Mixed Gauges}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 125}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1995a, author = {.}, title = {How to Detect Nondifferentiabilities and to Reduce Dimension in Minisum Mulitfacility Location Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 126}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mache1995, author = {.}, title = {A Link between Bernstein Polynomials and Durrmeyer Polynomials with {Jacobi} Weight}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 128}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FeltenMache1995, author = {. and .}, title = {Error Estimates of an Algebraic Convolution Operator by an Algebraic Modulus of Smoothness}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 129}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1995, author = {.}, title = {Leading Coefficients and Extreme Points of Homogeneous Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 130}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fredebeul1995, author = {.}, title = {A ( )--stable Formulae of Adams--Type}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 131}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Felten1995a, author = {.}, title = {A Smoothness Concept for Weighted Lp Spaces on [--1,1]}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 132}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FeltenKilgore1996, author = {. and .}, title = {Some Inequalities for Derivatives of Trigonometric and Algebraic Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 136}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1996, author = {.}, title = {Radon--Transform, Laplace--Series and Matrix--Transforms}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 137}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1996a, author = {.}, title = {The Average Size of Certain Gram--Determinants and Interpolation on Non--Compact Sets}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 138}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeMaier1996, author = {. and .}, title = {A Two--Stage Approach for Computing Cubature Formulae for the Sphere}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 139T}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1996b, author = {.}, title = {Node Distributions on the Unit Sphere with Fundamental Matrices Diagonal--Dominant in the Four Norm}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 140}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulWeber1996, author = {. and .}, title = {{\"{U}ber dne Einsatz von Quasi-Newton-Verfahren in DAGL-L\"{o}sern}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 141T}, year = {1996}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeNickel1997, author = {. and .}, title = {An Interior Point Method for Multifacility Location Problems with Forbidden Regions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 142}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BuhmannPinkus1997, author = {. and .}, title = {Identifiying Linear Combinations of Ridge Functions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 143}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1997, author = {.}, title = {A note on On Pareto optima, the Fermat--Weber problem and polyhedral gauges}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 144}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheLupasMaierMueller1997, author = {. and . and . and M\"{u}.}, title = {Linear Combinations of Gammaoperators in Lp--Spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 145}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Felten1997, author = {.}, title = {Direct and Inverse Estimates for Bernstein Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 147}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1997a, author = {.}, title = {Coincidence Conditions in Mulitfacility Location Problems with Positive and Negative Weights}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 148}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fischer1997, author = {.}, title = {Merit Functions and Stability for Complementarity Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 149}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fredebeul1997, author = {.}, title = {Insight into Linear Multistep Formulae from the Infinite Point of View}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 150}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulGreve1997, author = {. and .}, title = {A treatment of Rosenbrock methods within Albrecht`s linear {Runge}--{Kutta} theory}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 151}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeMichelotPlastria1998, author = {. and }, title = {A Polynomial Time Coincidence Detector for Multifacility Location Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 152}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1998, author = {.}, title = {Multithreaded Dimension Reduction in Location Theory}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 153}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1998a, author = {.}, title = {Special Cases and Efficient Algorithms for Multithreaded Dimension Reduction in Location Theory}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 154}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumLangerSchroeder1998, author = {. and . and Schr\"{o}.}, title = {LiMA {}--{} A Generic Class Library for Mesh Representation in 2D/3D}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 155T}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Lupas1998, author = {.}, title = {Polynomials which minimize certain functionals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 156}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1998b, author = {.}, title = {Solving Convex Location Problems with Gauges in Polynomial Time}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 158}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FischerFacchinei1998, author = {. and .}, title = {On the Identification of Zero Variables in an Interior--Point Framework}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 159}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FacchineiFischerKanzow1998, author = {. and . and .}, title = {On the Identification of Zero Variables in an Interior--Point Framework: Complete Numerical Results}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 160T}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege1998c, author = {.}, title = {Skript zur Vorlesung Innere--Punkt--Methoden}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 161T}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LupasMacheMaierMueller1998, author = {. and . and . and M\"{u}.}, title = {Certain Results Involving Gammaoperators}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 162}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Buhmann1998, author = {.}, title = {A New Class of Radial Basis Functions with Compact Support}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 164}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{DavydovNuernberger1998, author = {. and N\"{u}.}, title = {Interpolation By C1Splines of Degree q>4 on Triangulations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 165}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{DavydovSommer1998, author = {. and .}, title = {Interpolation by Weak Chebyshev Spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 166}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{DavydovNuernbergerZeilfelder1998, author = {. and N\"{ .}, title = {Interpolation by Splines on Triangulations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 167}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Moeller1999, author = {M\"{.}, title = {Exact computation of the generalized inverse and the least--squares solution}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 168}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Buhmann1999, author = {.}, title = {Approximation and interpolation with radial functions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 169}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerSauer1999, author = {M\"{ .}, title = {H--bases for polynomial interpolation and system solving}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 170}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FischerJiang1999, author = {. and .}, title = {Merit Functions {}--{} A Survey}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 172}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumSuttmeier1999, author = {. and .}, title = {An Adaptive Finite Element Discretisation for a Simplified Signorini Problem}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 173}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulKornmaier1999, author = {. and .}, title = {Ordnungsdynamische Dense Output {Runge}--{Kutta}--Fehlberg Verfahren I: Techniken der Konstruktion}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 174}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulKornmaier1999a, author = {. and .}, title = {{Ordnungsdynamische Dense Output Runge-Kutta-Fehlberg Verfahren II: Strategien der Anwendung}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 175}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumSuttmeier1999a, author = {. and .}, title = {Weighted Error Estimates for Finite Element Solutions of Variational Inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 179}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Maier1999, author = {.}, title = {Newton and {Lagrange} Representation of Kergin Interpolants}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 181}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mueller1999, author = {M\"{.}, title = {The central approximation theorems for the method of the left Gamma quasi--interpolants in Lp--spaces}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 182}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FischerJeyakumarLuc2000, author = {. and . and .}, title = {Solution Point Characterization and Convergence Analysis of a Descent Algorithm for Nonsmooth Continuous Complementarity Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 183}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheMache2000, author = {. and .}, title = {Approximation by Bernstein Quasi--Interpolants}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 184}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerSauer2000, author = {M\"{ .}, title = {H--Bases I: The Foundation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 186}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerSauer2000a, author = { .}, title = {H--Bases II: Applications to Numerical Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 187}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MacheMueller2000, author = {. and M\"{u}ller}, title = {The Method of Left Baskakov Quasi--Interpolants}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 188}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1999, author = {.}, title = {Discretized Newman--Shapiro Operators and Jackson`s Inequality on the Sphere}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 178}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fredebeul2000, author = {.}, title = {A Shooting Method for Symbolic Computation of Splines}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 189}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{PlatteFredebeul2000, author = {. and .}, title = {Dynamische und station\"{a}re Simulation des Reaktionsverhaltens eines Str\"{o}mungsumkehrreaktors}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 190T}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulPlatte2000, author = {. and .}, title = {Zur Anwendung direkter L\"{o}ser bei der direkten Berechnung periodisch station\"{a}rer Zust\"{a}nde eines Str\"{o}mungsumkehrreaktors}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 197T}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FredebeulKornmaierMueller2001, author = {. and . and M\"{u}ller}, title = {Multiple Order Double Output {Runge}--{Kutta}--Fehlberg Formulae: Strategies for Efficient Application}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 198}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Mache2001, author = {.}, title = {Generalized Methods using Integral Transforms and Concolution Structures with Orthogonal Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 199}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeFischer2001, author = {. and .}, title = {Script zur Vorlesung `Diskrete Optimierung`}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 200T}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumSuttmeier2001, author = {. and .}, title = {Adaptive Finite Elements for contact problems: Case studies in high--speed machining}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 201}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fischer2001, author = {.}, title = {Local Behavior of an Iterative Framework for generalized Equations with Nonisolated Solutions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 203}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fischer2001a, author = {.}, title = {Skript zur Vorlesung `Kontinuierliche Optimierung`}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 204T}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{RivkindSolonnikov2001, author = {. and .}, title = {Jeffrey--Hamel asyptotics for steady state {Navier}--{Stokes} flow in domains with sector like outlets to infinity}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 205}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{RivkindSolonnikov2001a, author = {. and .}, title = {On nonsymmetric two--dimensional viscous flow through an aperture}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 206}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeHeseler2002, author = {. and .}, title = {Constructing Approximations to the Efficient Set of Couvex Quadratic Multiobjective Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 211}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerSauer2002, author = { .}, title = {Multivariate refinable functions of high approximation order via quotient ideals of Laurent polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 212}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ChuiHeStoecklerSun2002, author = {. and . and St\"{o}ckler, J. and .}, title = {Compactly Supported Tight Affine Frames with Integer Dilations and Maximum Vanishing Moments}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 213}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ChuiHeStoeckler2002, author = {. and . and St\"{o}.}, title = {Tight Frames with Maximum Vanishing Moments and Minimum Support}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 214}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{JetterStoeckler2002, author = {. and St\"{o}.}, title = {An Identitiy for Mulitvariate Bernstein Polynomials}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 223}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ChuiStoeckler2002, author = {. and St\"{o}.}, title = {Recent Development of Spline Wavelet Frames with Compact Support}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 224}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LasserMacheObermaier2002, author = {. and . and .}, title = {Approximation Methods by using Orthogonal Polynomial Expansions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 185}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{deBruinMache2002, author = {. and .}, title = {(0,1) P\`{a}l--type Interpolation: A Genearal Method for Regularity}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 207}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{deBruinMache2002a, author = {, . and .}, title = {(0,2) P\`{a}l--type Interpolation: A General Method for Regularity}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 208}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Acker1998, author = {.}, title = {{Arbeiten mit GMV unter FEATFLOW}}, type = {{Preprints SFB 359}, {Nr.} 98-50}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{WehrseBaschekShaviv1994, author = {. and . and .}, title = {Vertical Structure and Spectra of Accretion Disks}, type = {{Preprints SFB 359}, {Nr.} 94-20}, year = {1994}, month = mar, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Bader1983, author = {.}, title = {Numerische Behandlung von Randwertproblemen f\"{u}r Funktionaldifferentialgleichungen}, type = {{Preprints SFB 123}, {Nr.} 227}, year = {1983}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Wittum1988, author = {.}, title = {Distributive Iterationen f\"{u}r indefinite Systeme}, type = {{Preprints SFB 123}, {Nr.} 454}, year = {1988}, month = apr, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Wittum1988a, author = {.}, title = {On the Convergence of Multi--Grid Methods with Transforming Smoothers. Theory with Applicaitons to the {Navier}--{Stokes} Equations}, type = {{Preprints SFB 123}, {Nr.} 468}, year = {1988}, month = jun, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1988, author = {.}, title = {Numerical Analysis of Nonstationary Fluid Flow (A survey)}, type = {{Preprints SFB 123}, {Nr.} 492}, year = {1988}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Bader1988, author = {.}, title = {Solution of Boundary Value Problems by Collocation Methods}, type = {{Preprints SFB 123}, {Nr.} 494}, year = {1988}, month = dec, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BlumRannacher1988, author = {. and .}, title = {On Mixed Finite Element Methods in Plate Bending Analysis {}--{} I. The First Herrmann Scheme}, type = {{Preprints SFB 123}, {Nr.} 498}, year = {1988}, month = dec, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BlumHarigMueller1989, author = {. and . and M\"{u}.}, title = {Extrapolation Techniques for Finite Element Approximations of {Stokes} and {Navier}--{Stokes} Flow over a Backward Facing Step}, type = {{Preprints SFB 123}, {Nr.} 499}, year = {1989}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1989, author = {.}, title = {On the Numerical Analysis of the Nonstationary {Navier}--{Stokes} Equations (The Role of Regularity and Stability Theory)}, type = {{Preprints SFB 123}, {Nr.} 542}, year = {1989}, month = oct, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Walter1989, author = {.}, title = {{Implementierung eines Finite-Elemente-Verfahrens f\"{u}r die Euler-Gleichung der Gasdynamic}}, type = {{Preprints SFB 123}, {Nr.} 500}, year = {1989}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{WittumLiebau1989, author = {. and .}, title = {On Truncated Incomplete Decompositions}, type = {{Preprints SFB 123}, {Nr.} 509}, year = {1989}, month = mar, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BlumHarigMueller1990a, author = {. and . and M\"{u}.}, title = {FEAT {}--{} Finite Element Analysis Tools {}--{} Release 1.1 User Manual}, type = {{Preprints SFB 123}, {Nr.} 554}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Dreyer1990, author = {.}, title = {Nonlinear ILU--Decomposition as Smoother in Multigrid Methods}, type = {{Preprints SFB 123}, {Nr.} 572}, year = {1990}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1990, author = {.}, title = {Defect Correction Techniques in the Finite Element Method}, type = {{Preprints SFB 123}, {Nr.} 586}, year = {1990}, month = aug, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1990a, author = {.}, title = {L Infiniti--Stability Estimates and Asymptotic Error Expansion for Parabolic Finite Element Equations}, type = {{Preprints SFB 123}, {Nr.} 589}, year = {1990}, month = aug, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1991, author = {.}, title = {On the Convergence of the Newton--Raphson Method for Strongly Nonlinear Finite Element Equations}, type = {{Preprints SFB 123}, {Nr.} 627}, year = {1991}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{DieciBader1992, author = {. and .}, title = {On Approximation Invariant Tori Block Inerations and Multigrid Methods for the Associated Systems}, type = {{Preprints SFB 123}, {Nr.} 658}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{HeywoodRannacher1992, author = {. and .}, title = {On the Question of Turbulence Modeling by Approximate Interial Manifolds and the Nonlinear {Galerkin} Method}, type = {{Preprints SFB 123}, {Nr.} 678}, year = {1992}, month = jun, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Schieweck1992, author = {.}, title = {A Parallel Multigrid Algorithm for Solving the {Navier}--{Stokes} Equations on a Transputer System}, type = {{Preprints SFB 123}, {Nr.} 699}, year = {1992}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Kanschat1996, author = {.}, title = {Parallel and Adaptive {Galerkin} Methods for Radiative Transfer Problems}, type = {{Preprints SFB 359}, {Nr.} 96-29}, year = {1996}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{NabhRannacher1996, author = {. and .}, title = {A Comparative Study of Nonlinear {G}alerkin Finite Element Methods for Dissipative Evolution Problems}, type = {{Preprints SFB 359}, {Nr.} 96-18}, year = {1996}, month = apr, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Becker1995a, author = {.}, title = {An Adaptive Finite Element Method for the Incompressible {Navier}--{Stokes} Equations on Time--dependent Domains}, type = {{Preprints SFB 359}, {Nr.} 95-44}, year = {1995}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Kroemker1995, author = {.}, title = {The Graphics Program cnom 2.0 User Manual}, type = {{Preprints SFB 359}, {Nr.} }, year = {1995}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{FuehrerKanschat1995, author = {. and .}, title = {A Posteriori Error Control in Radiative Transfer}, type = {{Preprints SFB 359}, {Nr.} 95-31}, year = {1995}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Rannacher1993, author = {.}, title = {Domain Decomposition in the Nonstationary Streamline Diffusion Finite Element Method}, type = {{Preprints SFB 359}, {Nr.} 93-56}, year = {1993}, month = oct, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{ChenRannacher1993, author = {. and .}, title = {Superconvergence Properties of Finite Element Schemes for the {Navier}--{Stokes} Problem}, type = {{Preprints SFB 359}, {Nr.} 93-37}, year = {1993}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{ChenRannacher1993a, author = {. and .}, title = {Local Error Expansions and Richardson Extrapolation for the Streamline Diffusion Finite Element Method}, type = {{Preprints SFB 359}, {Nr.} 93-22}, year = {1993}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{RannacherZhou1993, author = {. and .}, title = {Mesh Adaptation via a Predictor Corrector Strategy in the Streamline Diffusion Method for Nonstationary Hyperbolic Systems}, type = {{Preprints SFB 359}, {Nr.} 93-17}, year = {1993}, month = apr, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{FuehrerRannacher1994, author = {. and .}, title = {Error Analysis for the Finite Element Approximation of a Radiative Transfer Model}, type = {{Preprints SFB 359}, {Nr.} 94-46}, year = {1994}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BeckerRannacher1994, author = {. and .}, title = {Finite Element Solution of the Incompressible {Navier}--{}--{Stokes} Equations on Anisotropically Refined Meshes}, type = {{Preprints SFB 359}, {Nr.} 94-xx}, year = {1994}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Duschl1991, author = {. J.}, title = {Modelling viscosity in accretion disks}, type = {{Preprints SFB 359}, {Nr.} 91-08}, year = {1991}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{ZhouRannacher1994, author = {. and .}, title = {Pointwise Superconvergence of the Streamline Diffusion Finite Element Method}, type = {{Preprints SFB 359}, {Nr.} 94-72}, year = {1994}, month = dec, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Zhou1995a, author = {.}, title = {How accurate is the Streamline Diffusion Finite Element Method}, type = {{Preprints SFB 359}, {Nr.} 95-22}, year = {1995}, month = apr, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{RannacherZhou1994, author = {. and .}, title = {Analysis of a Domain--Splitting Method for Nonstationary Convection--Diffusion Problems}, type = {{Preprints SFB 359}, {Nr.} 94-16}, year = {1994}, month = mar, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Fuehrer1993, author = {.}, title = {A comparative study of finite element solvers for hyperbolic problems with applications to radiative transfer}, type = {{Preprints SFB 359}, {Nr.} 93-65}, year = {1993}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Bockhorn1990, author = {.}, title = {Simulation chemischer Reaktionen in turbulenten Str\"{o}mungen}, type = {{Preprints SFB 359}, {Nr.} 90-12}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{HauschildtWehrse1990, author = {. and .}, title = {Solution of the special relativistic equation of radiative transfer in rapidly expanding spherical shells}, type = {{Preprints SFB 359}, {Nr.} 90-24}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{JuengerReinelt1992, author = {. and .}, title = {{Schnittebenenverfahren in der Kombinatorischen Optimierung}}, type = {{Preprints SFB 359}, {Nr.} 92-30}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{JuengerReineltThienel1992, author = {. and . and .}, title = {Provably Good Solutions for the Traveling Salesman Problem}, type = {{Preprints SFB 359}, {Nr.} 92-31}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Harig1992, author = {.}, title = {A 3D Finite Element Upwind Approximation of the Stationary {Navier}--{Stokes} Equations}, type = {{Preprints SFB 359}, {Nr.} 92-09}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{MaasPope1992, author = {. and .}, title = {Implementation of Simplified Chemical Kinetics Based on Intrinsic Low--Dimensional Manifolds}, type = {{Preprints SFB 359}, {Nr.} 92-06}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Harig1990, author = {.}, title = {Visualization of 3--D Finite Element Solution of the {Navier}--{Stokes} Equations}, type = {{Preprints SFB 359}, {Nr.} 90-02}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{QuienMueller1990, author = {. and M\"{.}, title = {Ray Tracing on Transputers and Late Gothic Vaults}, type = {{Preprints SFB 359}, {Nr.} 90-04}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BaderGehrke1990, author = {. and .}, title = {On the Performance of Transputer Networks for solving linear Systems of Equations}, type = {{Preprints SFB 359}, {Nr.} 90-07}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{SauterWittum1990, author = {. and .}, title = {On the Computation of the Eigenmodes of Lake Constance by means of a Multi--Grid Method}, type = {{Preprints SFB 359}, {Nr.} 90-10}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{KanschatSuttmeier1998, author = {. and .}, title = {A Posteriori Error Estimates for Nonconforming Finite Element Schemes}, type = {{Preprints SFB 359}, {Nr.} 98-56}, year = {1998}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Braack1998, author = {.}, title = {An Adaptive Finite Element Method for Reactive--Flow Problems}, type = {{Preprints SFB 359}, {Nr.} 98-25}, year = {1998}, month = apr, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Hebeker1997, author = {.}, title = {An a Posteriori Error Estimator for Elliptic Boundary and Interface Problems in Layered Domains}, type = {{Preprints SFB 359}, {Nr.} 97-46}, year = {1997}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Suttmeier1997, author = {.}, title = {Adaptive Finite Element Approximation of Problems in Elasto--Plasticity Theory}, type = {{Preprints SFB 359}, {Nr.} 97-11}, year = {1997}, month = mar, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Fuehrer1997, author = {.}, title = {A Posteriori Error Control for the Numerical Approximation of Hyperbolic Problems}, type = {{Preprints SFB 359}, {Nr.} 97-05}, year = {1997}, month = feb, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Grimmer1997a, author = {.}, title = {A Comparative Study of Transport--Oriented Discretizations of Convection--dominated Problems with Application to the Incompressible {Navier}--{Stokes} Equations}, type = {{Preprints SFB 359}, {Nr.} 97-35}, year = {1997}, month = oct, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Fredebeul1995a, author = {.}, title = {A--BDF: A Generalization of the Backward--Difference--Formulae}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 127}, year = {1995}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{SiepmannSuendermann1982, author = {. and .}, title = {On a minimal property of cubic periodic Lagrangian Splines}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 060}, year = {1982}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Koester2002, author = {.}, title = {FEATFLOW in a Windows environment}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 217}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FliegeVicente2003, author = { .}, title = {A Multicriteria Approach to Bilevel Optimization}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 227}, year = {2003}, month = feb, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GoedertSuttmeier2002, author = {. and .}, title = {On the numerical simulation on induced anisotropy in polar ice}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 220}, year = {2002}, month = jul, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LauraBeutel2003, author = {.}, title = {On a lower bound of the second moment of the quadratic Schoenberg operator}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 226}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Turek1998c, author = {.}, title = {Konsequenzen eines numerischen `Elch Tests` f\"{u}r Computersimulationen}, type = {{Preprints SFB 359}, {Nr.} 98-46}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1999i, author = {.}, title = {Trends in processor technology and their impact on Numerics for PDE`s}, type = {{Preprints SFB 359}, {Nr.} 99-31}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1992e, author = {.}, title = {Visualization tools for the nonstationary {Navier}--{}--{Stokes} equations}, type = {{Preprints SFB 123}, {Nr.} 680}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{TurekBuijssen1998, author = {. and .}, title = {OMEGA2D {}--{} A Tutorial for a Preprocessing Tool in FEAT2D and FEATFLOW}, type = {{Preprints SFB 359}, {Nr.} 98-37}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{KuzminMoellerTurek2002a, author = {. and M\"{. and .}, title = {Multidimensional {FEM}--FCT schemes for arbitrary time--stepping}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 215}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2000a, author = {. and .}, title = {Efficient numerical techniques for flow simulation in bubble column reactors}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 196}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2001, author = {. and .}, title = {Flux correction tools for finite elements}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 209}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerKuzminTurek2002, author = {M\"{o}. and . and .}, title = {Implicit flux--corrected transport algorithm for finite element simulation of the compressible Euler equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 221}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2002a, author = {. and .}, title = {Finite element discretization and iterative solution techniques for multiphase flows in gas--liquid reactors}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 222}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2003a, author = {. and .}, title = {High--resolution {FEM}--TVD schemes based on a fully multidimensional flux limiter}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 229}, year = {2003}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminMoellerTurek2003, author = {. and M\"{o}. and .}, title = {High resolution {FEM}--FCT schemes for multidimenstional conservation laws}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 231}, year = {2003}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ChuiWenjieStoeckler2003, author = {. and \"{o}.}, title = {Nonstationary Tight Wavelet Frames on Bounded Intervals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 230}, year = {2003}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{RannacherTurek1990, author = {. and .}, title = {A Simple Nonconforming Quadrilateral {S}tokes Element}, type = {{Preprints SFB 123}, {Nr.} 567}, year = {1990}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1989, author = {.}, title = {A multigrid {Stokes} solver using divergence free finite elements}, type = {{Preprints SFB 123}, {Nr.} 527}, year = {1989}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{TurekOuazziSchmachtel2002a, author = {. and . and .}, title = {Multigrid Methods for Stabilized Nonconforming Finite Elements for Incompressible Flow involving the Deformation Tensor Formulation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 216}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekBeckerKilian2003a, author = {. and . and .}, title = {Hardware--oriented Numerics and concepts for {PDE} software}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 235}, year = {2003}, note = {to be published in a special issue on PDE Software in the Elsevier journal Future Generation Computer Systems}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BeckerRannacherTurek2000, author = {. and . and .}, title = {An Efficient {Navier}--{}--{Stokes} Solver for Automotive Aerodynamics}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 191}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HronMalekTurek1992, author = {. and M\`{. and .}, title = {A numerical investigation of flows of shear--thinning fluids with applications to blood rheology}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 192}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{AckerTurek2000, author = {. and .}, title = {Postprocessing of FEATFLOW Data with the Particle Tracing Tool GMVPT (Version 1.2)}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 193}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Turek2000h, author = {.}, title = {{CFD} for Incompressible Flow: Numerical Efficiency vs. Gigaflops}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 194}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Turek2000i, author = {.}, title = {Harte Nuss: Numerische Simulation von Str\"{o}mungen}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 195}, year = {2000}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BertranOlivierTurek2001, author = {. and . and .}, title = {Analysis of Unsteady Airfoils at low Speeds}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 202}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{PlatteTurekAgarFredebeul2001, author = {. and . and . and .}, title = {Cyclic Steady State Behaviour of a Reverse Flow Reactor in Analysis of High Temperature Reactions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 210}, year = {2001}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekSchmachtel2002a, author = {. and .}, title = {Fully coupled and operator--splitting approaches for natural convection flows in enclosures}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 218}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BuijssenTurek2002a, author = {. and .}, title = {Sources of Parallel Inefficiency for Incompressible {CFD} Simulations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 219}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ZietheSpohnTurek2002b, author = {. and . and .}, title = {Core Formation: A New Modelling Approach}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 225}, year = {2002}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HronOuazziTurek2003, author = {. and .}, title = {A Computational Comparison of two {FEM} Solvers For Nonlinear Incompressible Flow}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 228}, year = {2003}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Buijssen2003a, author = {Buijssen, .}, title = {User Guide parpp3d++; A Parallel C++ Version of the pp3d Module in Featflow}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 234}, year = {2003}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekWanRivkind2003, author = {. and . and .}, title = {The Fictitious Boundary Method for the implicit treatment of Dirichlet boundary conditions with applications to incompressible flow simulations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 236}, year = {2003}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BuijssenTurek2003, author = {Buijssen, . and .}, title = {parpp3d++ {}--{} a Parallel {HPC} Code for the Incompressible Nonstationary {Navier}--{}--{Stokes} Equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 237}, year = {2003}, month = jul, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GrajewskiHronTurek2003, author = {. and . and .}, title = {Numerical Analysis and a--posteriori Error Control for a New Nonconformining Linear Finite Element on Quadrilaterals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 238}, year = {2003}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{PGDeViSoR2003, author = {PG DeViSoR}, title = {{Endbericht der Projektgruppe DeViSoR}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 240T}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{OuazziTurek2003, author = {. and .}, title = {Numerical Methods and Simulation Techniques for Flow with Shear and Pressure dependent Viscosity}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 241}, year = {2003}, month = oct, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer2003a, author = {.}, title = {On the Implementation of a Generalized Hyperinterpolation Process}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 243}, year = {2003}, month = nov, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{WanTurekRivkind2003, author = {. and . and .}, title = {An efficient multigrid {FEM} solution technique for incompressible flow with moving rigrid bodies}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 245}, year = {2003}, month = nov, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GrajewskiHronTurek2004, author = {. and . and .}, title = {Dual Weighted A Posteriori Error Estimation for a New Nonconforming Linear Finite Element on Quadrilaterals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 247}, year = {2004}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GrundmannMoeller1976, author = {. and M\"{o}.}, title = {Invariant integration formulas for the n--simplex by combinatorial methods}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 005}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer1997, author = {.}, title = {Nonnegative Polynomials with a Constraint which Minimize Certain Integrals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 146}, year = {1997}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumLangerSchroeder1998a, author = {. and . and Schr\"{o}.}, title = {LiMA {}--{} A Generic Class Library for Mesh Representation in 2 D/ 3 D}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 157}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumBraessSuttmeier2003, author = {. and . and .}, title = {A Cascadic Multigrid Algorithm for Variational Inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 232}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BlumSchroederSuttmeier2003, author = {. and Schr\"{o}. and .}, title = {A Posteriori Error bounds for finite element schemes for a model friction problem}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 233}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Moeller2003, author = {.}, title = {Extension of positive functionals and cubature formulae}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 239}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{JetterStoeckler2003, author = {. and .}, title = {New Polynomial Preserving Operators on Simplices}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 242}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege2003, author = {.}, title = {Zur Entwicklung eines guten Programmstils}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 244T}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Achtziger2003, author = {.}, title = {On a Definition of Singular Optima and corresponding optimality conditions}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 246}, year = {2003}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminMoeller2004, author = {. and M\"{o}.}, title = {Algebraic Flux Correction I. Scalar Conservation Laws}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 249}, year = {2004}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminMoeller2004a, author = {. and M\"{o}.}, title = {Algebraic Flux Correction II. Compressible Euler Equations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 250}, year = {2004}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Moeller2004, author = {M\"{.}, title = {Counting zeros of polynomials by their B\`{e}zier ordinates}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 251}, year = {2004}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege2004, author = {.}, title = {An Efficient Interior--Point Method for Convex Multicriteria Optimization Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 248}, year = {2004}, month = feb, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2004b, author = {. and .}, title = {Multidimensional {FEM}--TVD Paradigm for Convection--Dominated Flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 253}, year = {2004}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurek2004c, author = {. and .}, title = {Numerical Simulation of Turbulent Bubbly Flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 254}, year = {2004}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KashidRivkindTurek2004, author = {. and . and .}, title = {Numerical Laboratory for master`s course in ROBOTICS AND AUTOMATION}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 255T}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HeermannWeyersFliege2004, author = {. and . and .}, title = {A New Adaptive Algorithm for Convex Quadratic Multicriteria Optimization}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 252}, year = {2004}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004, author = {.}, title = {On a direct approach to adaptive FE--discretisations for elliptic variational inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 256}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004a, author = {.}, title = {Goal--oriented error--estimation and postprocessing for FE--discretisations of variational inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 257}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004b, author = {.}, title = {Reliable, goal--oriented postprocessing for FE--discretisations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 258}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004c, author = {.}, title = {Finite Element analysis of the torsion problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 259}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004e, author = {.}, title = {On concepts of {PDE}--Software: The cellwise oriented approach in DEAL}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 260T}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004f, author = {.}, title = {Consistent error estimation of FE--approximations of variational inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 261}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KashidAckerPlatteAgarTurek2004, author = {. and . and . and . and .}, title = {Internal circulations inside the liquid slugs of liquid--liquid slug flow capillary microreactor}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 262}, year = {2004}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004g, author = {.}, title = {Adaptive Computational Methods for Variational Inequalities}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 263}, year = {2004}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{AchtzigerPoppZimmermann2004, author = {.}, title = {Optimization and Parallelization of Loop Nests via Linear Vector--Valued Schedules}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 264}, year = {2004}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004i, author = {.}, title = {Optimised mesh design for FE--{Galerkin} schemes using global norms}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 267}, year = {2004}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Suttmeier2004j, author = {.}, title = {On Plasticity with Hardening: An Adaptive Finite Element Discretisation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 265T}, year = {2004}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Fliege2004a, author = {.}, title = {An Efficient Interior--Point Method for Convex Multicriteria Optimization Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 248}, year = {2004}, month = feb, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LeibigDekorskyFliege2004, author = {. and . and .}, title = {Accelerated Power Control for CDMA Systems with Beamforming or Multiuser Detection}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 266}, year = {2004}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{ChuiHeStoeckler2004, author = {. and St\"{o}.}, title = {Nonstationary Tight Wavelet Frames, II: Unbounded Intervals}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 268}, year = {2004}, month = jul, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HronOuazziTurek2004, author = { .}, title = {Finite Element Methods for the Simulation of Incompressible Powder Flow}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 269}, year = {2004}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekKuzmin2004, author = {. and .}, title = {Algebraic Flux Correction III. Incompressible Flow Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 270}, year = {2004}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekBuijssenGrajewskiWobker2004, author = {. and .}, title = {High Performance {FEM} Simulation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 271}, year = {2004}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{PlatteKuzminFredebeulTurek2004, author = {. and .}, title = {Novel simulation approaches for cyclic--steady--state fixed--bed processes exhibiting sharp fronts and shocks}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 272}, year = {2004}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{NalpantidisPlatteAgarTurek2004, author = {.}, title = {Influence of axially structured catalyst in reverse flow reactor on the hybrid N2O decomposition}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 273}, year = {2004}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GellerKrafczykToelkeTurekHron2004, author = { .}, title = {Benchmark computations based on Lattice--Boltzmann, Finite Element and Finite Volume Methods for laminar Flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 274}, year = {2004}, month = oct, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerKuzminTurek2004, author = {. and .}, title = {Implicit {FEM}--FCT algorithm for compressible flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 276}, year = {2004}, month = nov, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{WanTurek2004, author = {. and .}, title = {Direct Numerical Simulation of Particulate Flow via Multigrid {FEM} Techniques and the Fictitious Boundary Method}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 275}, year = {2004}, month = nov, note = {Submitted to IGNMF 2005}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{WanTurek2004a, author = {. and .}, title = {An Efficient Multigrid--{FEM} Method for the Simulation of Liquid--Solid Two Phase Flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 277}, year = {2004}, month = dec, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KashidPlatteAgarTurek2004a, author = {. and . and . and .}, title = {Computational modelling of slug flow in a capillary microreactor}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 279}, year = {2004}, month = dec, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{HysingTurek2004, author = {. and .}, title = {The Eikonal Equation: Numerical Efficiency vs. Algorithmic Complexity on Quadrilateral Grids}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 280}, year = {2004}, month = dec, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Selimovic2005, author = {.}, title = {Foot points and total curvature}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 281}, year = {2005}, month = feb, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{TurekRivkindHronGlowinski2005a, author = {. and . and . and .}, title = {Numerical analysis of a new time--stepping theta--scheme for incompressible flow simulations}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 282}, year = {2005}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Kuzmin2005, author = {.}, title = {On the design of flux limiters for finite element discretizations with a consistent mass matrix}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 283}, year = {2005}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{OuazziTurek2005, author = {. and .}, title = {Unified edge--oriented stabilization of nonconforming finite element methods for incompressible flow problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 284}, year = {2005}, month = mar, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerKuzminTurek2005a, author = {. and .}, title = {Implicit finite element discretizations based on the flux corrected transport algorithm}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 285}, year = {2005}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Hron2005, author = {.}, title = {Fluid--structure interaction with applications in biomechanics}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 286}, year = {2005}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FengKoesterZhang2005, author = { K\"{o} .}, title = {Cylinder Flow Benchmark with Commercial Software Package {}--{} A Comparative Study}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 287}, year = {2005}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KashidAgar2005, author = {. and .}, title = {Hydrodynamics of liquid--liquid slug flow capillary microreactor}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 288}, year = {2005}, month = apr, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Reimer2005, author = {.}, title = {Asymptotics in Generalized Hyperinterpolation}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 289}, year = {2005}, month = may, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MittelmannWeber1981, author = {. and .}, title = {A Bibliography on nunmerical Mehtods for bifurcation problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 056}, year = {1981}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MaierMueller1976, author = {. and M\"{u}.}, title = {Die lokale Lp--Saturationsklasse des Verfahrens der integralen Meyer--K\"{o}nig und Zeller Operatoren}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 011}, year = {1976}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{LubinskyMache1998, author = {. and .}, title = {(C,I) Means of Orthonormal Expansions for Exponential Weights}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 163}, year = {1998}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{VeselicWenzel1979, author = {. and .}, title = {A quadratically convergent {Jacobi}--like method for real matrices with complex eigenvalue}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 028}, year = {1979}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{FischerMunsonFacchineiFerrisKanzow1999, author = {. and . and . and . and .}, title = {The Semismooth Algorithm for Large Scale Complementarity Problems}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 180}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{DavydovNuernbergerZeilfelder1999, author = {. and N\"{u} .}, title = {Bivariate Spline--Interpolation with optimal approximation order}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 171}, year = {1999}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KashidGerlachGoetzFranzkeAckerPlatteAgarTurek2004, author = {. and . and . and . and . and .}, title = {Internal circulation within the liquid slugs of liquid--liquid slug flow capillary microreactor}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 278}, year = {2004}, month = dec, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Turek1995a, author = {.}, title = {A comparative study of some time--stepping techniques for the incompressible {Navier}--{Stokes} equations: From fully implicit nonlinear schemes to semi--implicit projection methods}, type = {{Preprints SFB 359}, {Nr.} 95-10}, year = {1995}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{KilianTurek1998, author = {. and .}, title = {An example for parallel ScaRC and its application to the incompressible {Navier}--{Stokes} equations}, type = {{Preprints SFB 359}, {Nr.} 98-06}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{RannacherSchaeferTurek1998, author = {. and Sch\"{. and .}, title = {Evaluation of a {CFD} Benchmark for Laminar Flows}, type = {{Preprints SFB 359}, {Nr.} 98-23}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{MuellerProhlRannacherTurek1994, author = {. and . and .}, title = {Implicit time--discretization of the nonstationary incompressible {Navier}--{Stokes} equations}, type = {{Preprints SFB 359}, {Nr.} 94-34}, year = {1994}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1994f, author = {.}, title = {Multigrid techniques for a divergence--free finite element discretization}, type = {{Preprints SFB 359}, {Nr.} 94-21}, year = {1994}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1993a, author = {.}, title = {An efficient solution techniques for the radiative transfer equation}, type = {{Preprints SFB 359}, {Nr.} 93-08}, year = {1993}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{SchreiberTurek1993, author = {. and .}, title = {Multigrid results for the nonconforming Morley element}, type = {{Preprints SFB 359}, {Nr.} 93-67}, year = {1993}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{AckerTurek1999, author = {. and .}, title = {Postprocessing of 2D FEATFLOW Data with the Particle Tracing Tool GMVPT}, type = {{Preprints SFB 359}, {Nr.} 99-28}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{AltieriBeckerTurek1999, author = {. and . and .}, title = {Proposal for SPARSE BANDED BLAS Techniques}, type = {{Preprints SFB 359}, {Nr.} 99-11}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BeckerRungeTurek1999, author = {. and . and .}, title = {Performance Rating via the FEAST INDICES}, type = {{Preprints SFB 359}, {Nr.} 99-09}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BeckerTurekWaguetFEASTGroup1998, author = {. and . and . and FEAST Group}, title = {Application of DEVISORGRID 1.0 in the FEATFLOW software}, type = {{Preprints SFB 359}, {Nr.} 98-53}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{AltieriBeckerKilianOswaldTurekWallis1998, author = {. and . and . and . and . and .}, title = {Some Basic Concepts of FEAST}, type = {{Preprints SFB 359}, {Nr.} 98-28}, year = {1998}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1991a, author = {.}, title = {Ein robustes und effizientes Mehrgitterverfahren zur L\"{o}sung der instation\"{a}ren, inkompressiblen 2--D {Navier}--{Stokes}--Gleichungen mit diskret divergenzfreien finiten Elementen}, type = {{Preprints SFB 123}, {Nr.} 642}, year = {1991}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1991b, author = {.}, title = {On Ordering Strategies in a Multigrid Algorithm}, type = {{Preprints SFB 123}, {Nr.} 653}, year = {1991}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BeckerRungeTurek1999b, author = { .}, title = {The FEAST INDICES {}--{} Realistic evaluation of modern software components and processor technologies}, type = {{Preprints SFB 359}, {Nr.} 99-17}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{MalekTurek1992, author = {. and .}, title = {Divergence--Free Finite Elements}, type = {{Preprints SFB 123}, {Nr.} 687}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{HarigSchreiberTurek1994, author = {. and . and .}, title = {FEAT3D {}--{} Finite Element Analysis Tools in 3 Dimensions User Manual. Release 1.2}, type = {{Preprints SFB 359}, {Nr.} 94-19}, year = {1994}, month = mar, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1999j, author = {.}, title = {Trends in processor technology and their impact on Numerics for PDE`s}, type = {{Preprints SFB 359}, {Nr.} 99-31}, year = {1999}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BuijssenTurek1999, author = {. and .}, title = {Batch--Oriented MPEG Generation with {GMV} in Background Mode}, type = {{Preprints SFB 359}, {Nr.} 99-29}, year = {1999}, month = jun, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{SchaeferTurek1996, author = {Sch\"{. and .}, title = {Benchmark Computations of Laminar Flow Around a Cylinder}, type = {{Preprints SFB 359}, {Nr.} 96-03}, year = {1996}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{TurekWehrse1994a, author = {. and .}, title = {Spectral Appearance of Dust Enshrouded Stars: A Finite Element Approach to 2D Radiative Transfer}, type = {{Preprints SFB 359}, {Nr.} 94-43}, year = {1994}, month = jul, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1994j, author = {.}, title = {A Generalized Mean Intensity Approach for the Numerical Solution of the Radiative Transfer Equation}, type = {{Preprints SFB 359}, {Nr.} 94-04}, year = {1994}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{AltieriBeckerTurek1998a, author = {. and . and .}, title = {On the Realistic Performance of Certain Linear Algebra Components in Iterative Solvers}, type = {{Preprints SFB 359}, {Nr.} 98-31}, year = {1998}, month = may, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{HeywoodRannacherTurek1992, author = {. and . and .}, title = {Artificial Boundaries and Flux and Pressure Conditions for the Incompressible {N}avier--{}--{S}tokes Equations}, type = {{Preprints SFB 123}, {Nr.} 681}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1994d, author = {.}, title = {On Discrete Projection Methods for the incompressible {Navier}--{Stokes} Equations: An Algorithmical Approach}, type = {{Preprints SFB 359}, {Nr.} 94-70}, year = {1994}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Turek1992d, author = {.}, title = {Tools for Simulationg Nonstationary Incompressible Flow via Discretely Divergence--Free Finite Element Models}, type = {{Preprints SFB 123}, {Nr.} 679}, year = {1992}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BraackRannacherSchmich2004, author = {. and . and .}, title = {A Computational Study of a Splitting Scheme for Weakly Compressible Stationary Viscous Flow}, type = {{Preprints SFB 359}, {Nr.} 04-18}, year = {2004}, note = {\url{http:/ / www.iwr.uni-heidelberg.de/ sfb359/ PP/ Preprint2004-18.pdf}}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{GaldiHeuveline2004, author = {. and .}, title = {Lift and Sedimentation of Particles in the Flow of a Viscoelastic Liquid in a Channel}, type = {{Preprints SFB 359}, {Nr.} 04-36}, year = {2004}, note = {\url{http:/ / www.iwr.uni-heidelberg.de/ sfb359/ PP/ Preprint2004-36.pdf}}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Becker2000, author = {.}, title = {An optimal--control approach to a--posteriori error estimation for Finite Element discretizations of the {Navier}--{Stokes} equations}, type = {{Preprints SFB 359}, {Nr.} 00-34}, year = {2000}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{Becker2000a, author = {.}, title = {Mesh adaption for stationary flow control}, type = {{Preprints SFB 359}, {Nr.} 00-35}, year = {2000}, month = nov, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{BollhoeferMehrmann1999a, author = {Bollh\"{. and .}, title = {A new approach to algebraic multileven methods based on sparse approximate inverses}, type = {{Preprints SFB 393}, {Nr.} 99-22}, year = {1999}, month = aug, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{KunertMghazliNicaise2003, author = {. and . and .}, title = {A posteriori error estimation for a finite volume discretization on anisotropic meshes}, type = {{Preprints SFB 393}, {Nr.} 03-16}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{CreuseKunertNicaise2003, author = {. and . and .}, title = {A posteriori error estimation for the {Stokes} problem: Anisotropic and isotropic discretizations}, type = {{Preprints SFB 393}, {Nr.} 03-01}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{Beuchler2003a, author = {.}, title = {Optimal preconditioners for the \$p\$--Version of the fem}, type = {{Preprints SFB 393}, {Nr.} 03-03}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{ApelNicaise2003, author = {. and .}, title = {The inf--sup condition for the Bernardi--Fortin--Raugel element on anisotropic meshes}, type = {{Preprints SFB 393}, {Nr.} 03-15}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{Beuchler2003b, author = {.}, title = {A Dirichlet--Dirichlet DD--pre--conditioner for \$p\$--{FEM}}, type = {{Preprints SFB 393}, {Nr.} 03-12}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{ApelPester2003, author = {. and .}, title = {Clement--type interpolation on spherical domains {}--{}--{} interpolation error estimates and application to a posteriori error estimation}, type = {{Preprints SFB 393}, {Nr.} 03-13}, year = {2003}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{Kunert2000a, author = {.}, title = {Anisotropic mesh construction and error estimation in the finite element method}, type = {{Preprints SFB 393}, {Nr.} 00-01}, year = {2000}, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{Grosman2002, author = {.}, title = {Robust local problem error estimation for a singularly perturbed reaction--diffusion problem on anisotropic finite element meshes}, type = {{Preprints SFB 393}, {Nr.} 02-07}, year = {2002}, month = may, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{Grosman2004, author = {.}, title = {The robustness of the hierarchical a posteriori error estimator for reaction--diffusion equation on anisotropic meshes}, type = {{Preprints SFB 393}, {Nr.} 04-02}, year = {2004}, month = mar, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{MeyerUnger2004, author = {. and .}, title = {Projection methods for contact problems in elasticity}, type = {{Preprints SFB 393}, {Nr.} 0404}, year = {2004}, month = apr, institution = {Fakult\"{a}t f\"{u}r Mathematik, Techn. Universit\"{a}t Chemnitz}, } @TECHREPORT{AckerTurek1999a, author = {. and .}, title = {3D Presentation of FEATFLOW data with {GMV}}, type = {{Preprints SFB 359}, {Nr.} 99-19}, year = {1999}, institution = {Universit\"{a}t Heidelberg}, } @TECHREPORT{GoeddekeStrzodkaTurek2005, author = {. and . and .}, title = {Accelerating Double Precision {FEM} Simulations with {GPUs}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 292}, year = {2005}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KoesterPlatteTurek2005, author = {. and . and .}, title = {Speed--up of {FEM} simulation for granular flow via optimised Numerical linear Algebra software}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 290}, year = {2005}, month = jun, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{BuijssenWobkerTurek2005a, author = {. and . and .}, title = {High Performance {FEM} Simulation in {CFD} and {CSM}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 293}, year = {2005}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Hysing2005a, author = {.}, title = {A new implicit surface tension implementation for interfacial flows}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 295}, year = {2005}, month = aug, note = {Submitted to Int. J. Numer. Meth. Fluids}, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{GrajewskiKoesterKilianTurek2005, author = {. and .}, title = {Numerical Analysis and Practical Aspects of a Robust and Efficient Grid Deformaton Method in the Finite Element Context}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 294}, year = {2005}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KoesterTurek2005, author = {. and .}, title = {A note on optimal multigrid convergence for higher--order {FEM}}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 296}, year = {2005}, month = aug, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{MoellerKuzmin2005a, author = {. and .}, title = {Adaptive Mesh Refinement for High--Resolution Finite Element Schemes}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 297}, year = {2005}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{KuzminTurekHaario2005, author = {. and . and .}, title = {Finite element simulation of turbulent bubbly flows in gas--liquid reactors}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 298}, year = {2005}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Moritzen2005, author = {.}, title = {Modifications on the {Jacobi}--Davidson Method for Usage in Topology Optimization of Structures}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 291T}, year = {2005}, month = sep, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Kuzmin2005b, author = {.}, title = {On the design of general--purpose flux limiters for implicit {FEM} with a consistent mass matrix}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 299}, year = {2005}, month = oct, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } @TECHREPORT{Goeddeke2005b, author = {.}, title = {GPGPU--Basic Math Tutorial}, type = {{Ergebnisberichte des Instituts f\"{u}r Angewandte Mathematik}, {Nr.} 300}, year = {2005}, month = nov, institution = {FB Mathematik, Universit\"{a}t Dortmund}, } %% Extended UTXO Specification \newcommand\version{-1} \title{The Extended UTXO Ledger Model} \pagestyle{plain} \date{11th March 2020} \author{} \documentclass[a4paper]{article} % correct bad hyphenation here \hyphenation{} \usepackage{natbib} \usepackage{url} % *** MATHS PACKAGES *** % \usepackage[cmex10]{amsmath} \usepackage{amssymb} \usepackage{stmaryrd} \usepackage{amsthm} % *** ALIGNMENT PACKAGES *** % \usepackage{array} \usepackage{float} %% Try to improve placement of figures. Doesn't work well with subcaption package. \usepackage{subcaption} \usepackage{caption} \usepackage{subfiles} \usepackage{geometry} \usepackage{listings} \usepackage[dvipsnames]{xcolor} \usepackage{verbatim} \usepackage{alltt} %\usepackage{todonotes} \usepackage[disable]{todonotes} % These have to go at the end of the packages. \usepackage[colorlinks=true,linkcolor=MidnightBlue,citecolor=ForestGreen,urlcolor=Plum]{hyperref} \usepackage[capitalise, noabbrev]{cleveref} % Mild hack to get cleveref to refer to things as "Rules": we alias the % enumerate counters to "rule". This would be annoying if we ever wanted to % refer to items or lists as something other than "rule", but we don't. \crefname{rule}{rule}{rules} \crefname{rule}{Rule}{Rules} \crefalias{enumi}{rule} \crefalias{enumii}{rule} % Stuff for splitting figures over page breaks %\DeclareCaptionLabelFormat{continued}{#1~#2 (Continued)} %\captionsetup[ContinuedFloat]{labelformat=continued} % *** MACROS *** % A command for making notes with bold titles and an independent % numbering system. This is used for a list of notes in the appendix. \newcounter{note} \newcommand{\note}[1]{ \bigskip \refstepcounter{note} \noindent\textbf{Note \thenote. #1} } \newcommand{\todochak}[1]{\todo[inline,color=purple!40,author=chak]{#1}} \newcommand{\todompj}[1]{\todo[inline,color=yellow!40,author=Michael]{#1}} \newcommand{\todokwxm}[1]{\todo[inline,color=blue!20,author=kwxm]{#1}} \newcommand{\todojm}[1]{\todo[inline,color=purple!40,author=Jann]{#1}} \newcommand{\red}[1]{\textcolor{red}{#1}} \newcommand{\redfootnote}[1]{\red{\footnote{\red{#1}}}} \newcommand{\blue}[1]{\textcolor{blue}{#1}} \newcommand{\bluefootnote}[1]{\blue{\footnote{\blue{#1}}}} %% A version of ^{\prime} for use in text mode \makeatletter \DeclareTextCommand{\textprime}{\encodingdefault}{% \mbox{$\m@th'\kern-\scriptspace$}% } \makeatother \renewcommand{\i}{\textit} % Just to speed up typing: replace these in the final version \renewcommand{\t}{\texttt} % Just to speed up typing: replace these in the final version \newcommand{\s}{\textsf} % Just to speed up typing: replace these in the final version \newcommand{\msf}[1]{\ensuremath{\mathsf{#1}}} \newcommand{\mi}[1]{\ensuremath{\mathit{#1}}} %% A figure with rules above and below. \newcommand\rfskip{7pt} \newenvironment{ruledfigure}[1]{\begin{figure}[#1]\hrule\vspace{\rfskip}}{\vspace{\rfskip}\hrule\end{figure}} %% Various text macros \newcommand{\true}{\textsf{true}} \newcommand{\false}{\textsf{false}} \newcommand{\hash}[1]{\ensuremath{#1^{\#}}} \newcommand{\List}[1]{\ensuremath{\s{List}[#1]}} \newcommand{\Set}[1]{\ensuremath{\s{Set}[#1]}} \newcommand{\FinSet}[1]{\ensuremath{\s{FinSet}[#1]}} \newcommand{\Interval}[1]{\ensuremath{\s{Interval}[#1]}} \newcommand{\FinSup}[2]{\ensuremath{\s{FinSup}[#1,#2]}} \newcommand{\support}{\msf{support}} \newcommand{\script}{\ensuremath{\s{Script}}} \newcommand{\scriptAddr}{\msf{scriptAddr}} \newcommand{\ctx}{\ensuremath{\s{Context}}} \newcommand{\toData}{\ensuremath{\s{toData}}} \newcommand{\mkContext}{\ensuremath{\s{mkContext}}} % Macros for eutxo things. \newcommand{\TxId}{\ensuremath{\s{TxId}}} \newcommand{\txId}{\msf{txId}} \newcommand{\txrefid}{\mi{id}} \newcommand{\Address}{\ensuremath{\s{Address}}} \newcommand{\DataHash}{\ensuremath{\s{DataHash}}} \newcommand{\hashData}{\msf{dataHash}} \newcommand{\idx}{\mi{index}} \newcommand{\inputs}{\mi{inputs}} \newcommand{\outputs}{\mi{outputs}} \newcommand{\forge}{\mi{forge}} \newcommand{\fee}{\mi{fee}} \newcommand{\addr}{\mi{addr}} \newcommand{\val}{\mi{value}} %% \value is already defined \newcommand{\validator}{\mi{validator}} \newcommand{\redeemer}{\mi{redeemer}} \newcommand{\datum}{\mi{datum}} \newcommand{\datumHash}{\mi{datumHash}} \newcommand{\datumWits}{\mi{datumWitnesses}} \newcommand{\Data}{\ensuremath{\s{Data}}} \newcommand{\outputref}{\mi{outputRef}} \newcommand{\txin}{\mi{in}} \newcommand{\id}{\mi{id}} \newcommand{\lookupTx}{\msf{lookupTx}} \newcommand{\getSpent}{\msf{getSpentOutput}} \newcommand{\slotnum}{\ensuremath{\s{SlotNumber}}} \newcommand{\spent}{\msf{spentOutputs}} \newcommand{\unspent}{\msf{unspentOutputs}} \newcommand{\txunspent}{\msf{unspentTxOutputs}} \newcommand{\eutxotx}{\msf{Tx}} \newcommand{\qty}{\ensuremath{\s{Quantity}}} \newcommand{\token}{\ensuremath{\s{Token}}} \newcommand{\currency}{\ensuremath{\s{CurrencyId}}} \newcommand{\nativeCur}{\ensuremath{\mathrm{nativeC}}} \newcommand{\nativeTok}{\ensuremath{\mathrm{nativeT}}} \newcommand{\qtymap}{\ensuremath{\s{Quantities}}} \newcommand\B{\ensuremath{\mathbb{B}}} \newcommand\N{\ensuremath{\mathbb{N}}} \newcommand\Z{\ensuremath{\mathbb{Z}}} \renewcommand\H{\ensuremath{\mathbb{H}}} %% \H is usually the Hungarian double acute accent \newcommand{\emptyBs}{\ensuremath{\emptyset}} \newcommand{\emptymap}{\ensuremath{\{\}}} %% ------------- Start of document ------------- %% \begin{document} \maketitle \section{Introduction: The Extended UTXO Model} \label{sec:intro} The Cardano blockchain~\citep{Cardano, Cardano-ledger-spec} uses a variant of the \textit{Unspent Transaction Output} (UTXO) model used by Bitcoin. Transactions consume \textit{unspent outputs} (UTXOs) from previous transactions and produce new outputs which can be used as inputs to later transactions. Unspent outputs are the liquid funds on the blockchain. Users do not have individual accounts, but rather have a software \textit{wallet} on a smartphone or PC which manages UTXOs on the blockchain and can initiate transactions involving UTXOs owned by the user. Every core node on the blockchain maintains a record of all of the currently unspent outputs, the \textit{UTXO set}; when outputs are spent, they are removed from the UTXO set. This document contains a description of some extensions of the UTXO model: the main aim of these extensions is to facilitate the implementation of \textit{smart contracts}, programs which perform automated and irrevocable transfer of funds on the blockchain, subject to certain conditions being met. A smart contract may involve multiple transactions, and our aim is to define a transaction model which enables the implementation of highly expressive contracts. An important feature of our UTXO models is \textit{scripts}, programs which run on the blockchain to check the validity of transactions. In Cardano, scripts will be programs in the Plutus Core language~\citep{Plutus-Core-spec}. The Extended UTXO models are largely agnostic as to the scripting language. \subsection{Structure of the document} \label{sec:doc-structure} The papers~\citep{Zahnentferner18-Chimeric} and \citep{Zahnentferner18-UTxO} give a formal specification of a basic UTXO model. See \cref{note:basic-utxo} for some background on this model. \medskip \noindent This document proposes two extensions of the basic UTXO model (EUTXO stands for \textit{Extended UTXO}): \begin{itemize} \item \textbf{EUTXO-1} (\cref{sec:eutxo-1}): this extends the basic UTXO model with enhanced scripting features, allowing the implementation of complex smart contracts. \item \textbf{EUTXO-2} (\cref{sec:eutxo-2}): this adds multicurrency features to EUTXO-1, allowing users to define \textit{custom currencies} and \textit{non-fungible tokens}. \end{itemize} \medskip The rationale for providing two separate extensions is that (1) introducing the extensions separately clarifies the structure of the models and makes it easier to explain the relevant design decisions, and (2) it is possible that a particular blockchain might not need the full power of EUTXO-2 and so could use the simpler EUTXO-1 model, perhaps with less computational overhead. \medskip For ease of reference we have kept exposition to a minimum in the main text. Some aspects of the models are explained in more detail in \cref{appendix:comments}, with cross-references in the main text. Further explanation and many examples are contained in the book~\citep{Plutus-book}. \section{Notation} This section defines some basic notation. We generally follow the notation established by \citep{Zahnentferner18-UTxO}, except that we make use of finitely-supported functions in most places that \citep{Zahnentferner18-UTxO} use maps. \subsection{Basic types and operations} \label{sec:basic-notation} This section describes some types, notation, and conventions used in the remainder of the document. \begin{itemize} \item Types are typeset in $\mathsf{sans~serif}$. \item \B{} denotes the type of booleans, $\{\false, \true\}$. \item \N{} denotes the type of natural numbers, $\{0, 1, 2, \ldots\}$. \item \Z{} denotes the type of integers, $\{\ldots, -2, -1, 0, 1, 2, \ldots\}$. \item We regard $\N$ as a subtype of $\Z$ and convert freely between natural numbers and non-negative integers. \item \H{} denotes the type of bytestrings, $\bigcup_{n=0}^{\infty}\{0,1\}^{8n}$. \emptyBs{} denotes the empty bytestring. A bytestring is a sequence of 8-bit bytes: the symbol $\H$ is used because bytestrings are often presented as sequences of hexadecimal digits. \item If a type $M$ is a monoid, we use $+$ for the monoidal operation and $0$ for the unit of the monoid. If $M$ is a commutative monoid, we use $\sum$ for the extension of $+$ to a finite set of elements of type $M$. If $M$ is a group, we use $-$ for the group inverse operation. This should never be ambiguous. \item A record type with fields $\phi_1, \ldots, \phi_n$ of types $T_1, \ldots, T_n$ is denoted by $(\phi_1 : T_1, \ldots, \phi_n : T_n)$. If $t$ is a value of a record type $T$ and $\phi$ is the name of a field of $T$ then $t.\phi$ denotes the value of $\phi$ for $t$. \item If $T$ is a type then $\FinSet{T}$ is the type of finite sets with elements of type $T$. \item A list $l$ of type $\List{T}$ is either the empty list $[]$ or a list $e :: l$ with $head$ $e$ of type $T$ and $tail$ $l$ of type $\List{T}$. A list has only a finite number of elements. We denote the $i$th element of a list $l$ by $l[i]$ and the length of $l$ by $\left|l\right|$. \item $x \mapsto f(x)$ denotes an anonymous function. \item A cryptographic collision-resistant hash of a value $c$ is denoted $\hash{c}$. \item For a type $A$ which forms a total order, $\Interval{A}$ is the type of intervals over that type. Intervals may be bounded or unbounded, and open or closed at either end. The type $\Interval{A}$ forms a lattice under inclusion. \end{itemize} \subsection {Finitely-supported functions} \label{sec:fsfs} Finitely-supported functions are a generalisation of maps to monoidal values. They always return an answer (which will in all but finitely many cases be zero), and can be queried for the set of non-zero points in their domain. For two types $K$ and $V$ where $V$ is a monoid, $\FinSup{K}{V}$ denotes the type of \textit{finitely-supported functions} from $K$ to $V$. That is, there is a function $\support : \FinSup{K}{V} \rightarrow \FinSet{K}$ such that $k \in \support(f) \Leftrightarrow f(k) \neq 0$. Equality on finitely-supported functions is defined as pointwise equality. Similarly, if $V$ has a partial order, then a partial order on finitely-supported functions is also defined pointwise. If the type $M$ is a monoid then we define the sum of two finitely-supported functions $f, g \in \FinSup{K}{M}$ to be the function $f+g \in \FinSup{K}{M}$ given by \[(f+g)(k) = f(k) + g(k) \] Note that the type $\FinSup{K}{M}$ is a monoid with this operation, and the empty function as identity element. If the type $M$ is a group, then we can similarly define the inverse of a finitely-supported function $f$ as the function $(-f)$ with the same support, given by \[ (-f)(k) = -f(k) \] Again, $\FinSup{K}{M}$ is a group with this operation. See \cref{note:finitely-supported-functions} for discussion of using finitely-supported functions computationally. \subsection{The \Data{} type} We also define a type \Data{} which can be used to pass information into scripts in a type-safe manner: see \cref{fig:data-defn}. The definition is given here in EBNF form, but can easily be translated to a Haskell type, for instance. \begin{ruledfigure}{H} \begin{alltt} \Data = "I" \(\Z\) | "B" \(\H\) | "Constr" \(\N (\List{\Data})\) | "List" \(\List{\Data}\) | "Map" \(\List{\Data\times\Data}\) \end{alltt} \caption{The \Data{} type} \label{fig:data-defn} \end{ruledfigure} \noindent Thus values of type \Data{} are nested sums and products built up recursively from the base types of integers and bytestrings. This allows one to encode a large variety of first-order data structures: for example, we could encode values of Haskell's \verb|Maybe Integer| type using \verb|Constr 0 []| to represent \verb|Nothing| and \verb|Constr 1 [I 41]| to encode \verb|Just 41|. The \texttt{List} and \texttt{Map} constructors are strictly redundant, but are included for convenience to allow straightforward encoding of lists and records. We assume that the scripting language has the ability to parse values of type \Data{}, converting them into a suitable internal representation. \section{EUTXO-1: Enhanced scripting} \label{sec:eutxo-1} The EUTXO-1 model adds the following new features to the model proposed in~\citep{Zahnentferner18-UTxO}: \begin{itemize} \item Every transaction has a \textit{validity interval}, of type $\Interval{\slotnum}$. A core node will only process the transaction if the current slot number lies within the transaction's validity interval. \item The redeemer script of~\citet{Zahnentferner18-UTxO} has been replaced with a \textit{redeemer object} (\textit{redeemer} for short) of type \Data{}. \item Each unspent output now has an object of type \Data{} associated with it: we call this the output's \textit{datum} (or occasionally \emph{datum object}) (see \cref{note:datum}). Only the hash $\datumHash$ of the datum is stored in the output: the full value must be provided when the output is spent, much like the validator. \item Validator scripts make use of information about the pending transaction (ie, the transaction which is just about to take place, assuming that validation succeeds). This information is contained in a structure which we call \ctx{} (see \cref{sec:context} for its definition). We may refer to this information as the \textit{validation context} in cases where ambiguity may arise. \item Validation of an output is performed by running the validator with three inputs: \begin{enumerate} \item the datum, \item the redeemer, \item the \ctx{} information, encoded as \Data{}. \end{enumerate} \end{itemize} \todokwxm{There's the issue about whether validators return \texttt{true/false} or \texttt{()/Error}. Remember to fix this when things have settled down.} \subsection{A Formal Description of the EUTXO-1 Model} \label{section:eutxo-spec} In this section we give a formal description of the EUTXO-1 model. The description is given in a straightforward set-theoretic form, which (a) admits an almost direct translation into Haskell, and (b) should easily be amenable to mechanical formalisation. This will potentially allow us to argue formally about smart contracts and to develop tools for automatic contract analysis. The definitions in this section are essentially the definitions of UTXO-based cryptocurrencies with scripts from \citep{Zahnentferner18-UTxO}, except that we have added the new features mentioned above (the validity interval, the datum and the \ctx{} structure), changed the type of the redeemer from \script{} to \Data{}, and used finitely-supported functions in place of maps. \Cref{fig:eutxo-1-types} lists the types and operations used in the the basic EUTXO model. Some of these are defined, the others must be provided by the ledger. %% \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \multicolumn{3}{l}{\textsc{Ledger primitives}}\\ \qty{} && \mbox{an amount of currency}\\ \slotnum && \mbox{a slot number}\\ \Address && \mbox{the ``address'' of a script in the blockchain}\\ \DataHash && \mbox{the hash of an object of type \Data{}}\\ \TxId && \mbox{the identifier of a transaction}\\ \txId : \eutxotx \rightarrow \TxId && \mbox{a function computing the identifier of a transaction}\\ \script && \mbox{the (opaque) type of scripts}\\ \scriptAddr : \script \rightarrow \Address && \mbox{the address of a script}\\ \hashData : \Data \rightarrow \DataHash && \mbox{the hash of a data object}\\ \llbracket \cdots \rrbracket : \script \rightarrow \Data \times \Data \times \Data \rightarrow \B && \mbox{application of a script to its arguments}\\ \\ \multicolumn{3}{l}{\textsc{Defined types}}\\ \s{Output } &=&(\addr: \Address,\\ & &\ \val: \qty,\\ & &\ \datumHash: \DataHash)\\ \\ \s{OutputRef } &=&(\txrefid: \TxId, \idx: \s{Int})\\ \\ \s{Input } &=&(\outputref: \s{OutputRef},\\ & &\ \validator: \script,\\ & &\ \datum: \Data,\\ & &\ \redeemer: \Data)\\ \\ \eutxotx\s{ } &=&(\inputs: \FinSet{\s{Input}},\\ & &\ \outputs: \List{\s{Output}},\\ & &\ \i{validityInterval}: \Interval{\slotnum},\\ & &\ \datumWits: \FinSup{\DataHash}{\Data},\\ & &\ \fee: \qty,\\ & &\ \forge: \qty) \\ \\ \s{Ledger } &=&\!\List{\eutxotx}\\ \end{array} \end{displaymath} \caption{Primitives and basic types for the EUTXO-1 model} \label{fig:eutxo-1-types} \end{ruledfigure} \subsubsection{Remarks} \paragraph{ETUXO-1 on Cardano.} The Cardano implementation of EUTXO-1 uses the primitives given in \cref{fig:eutxo-1-types-cardano}. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \qty{} &=& \Z\\ \slotnum &=& \N\\ \Address &=& \H\\ \DataHash &=& \H\\ \TxId &=& \H\\ \txId : \eutxotx \rightarrow \TxId &=& t \mapsto \hash{t}\\ \script &=& \mbox{a Plutus Core program}\\ \scriptAddr : \script \rightarrow \Address &=& s \mapsto \hash{s}\\ \llbracket \cdots \rrbracket : \script \rightarrow \Data \times \Data \times \Data \rightarrow \B &=& \mbox{typechecking the program and running}\\ &&\mbox{the Plutus Core interpreter}\\ \end{array} \end{displaymath} \caption{Cardano primitives for the EUTXO-1 model} \label{fig:eutxo-1-types-cardano} \end{ruledfigure} \paragraph{Inputs and outputs.} Note that a transaction has a \textsf{Set} of inputs but a \textsf{List} of outputs. See \cref{note:inputs-and-outputs} for a discussion of why. \paragraph{Validator addresses in outputs.} The \textit{addr} field of an output should contain the address of the validator script for that output: this requirement is enforced in \cref{rule:validator-scripts-hash} of \cref{fig:eutxo-1-validity} below. \paragraph{Scripts and hashes.} Note that datum objects and validators are provided as parts of transaction inputs, even though they are conceptually part of the output being spent. The reasons for this are explained in \cref{note:scripts}. \paragraph{Datum witnesses.} The transaction may include the full value of the datum for each output that it creates. See \cref{note:datum-witnesses} for more discussion. \paragraph{Fees.} Users are charged a fee for the on-chain storage and execution costs of a transaction, and this is included in the EUTXO models. The details are not important for the purposes of the models, but see \cref{note:fees} for some more discussion. \paragraph{Special types of transaction.} In a practical implementation it might be useful to include special cases for common transaction types such as pay-to-pukbey transactions in order to increase efficiency and decrease storage requirements (and hence reduce fees). These have been omitted from this model because it subsumes all of the other transaction types we're likely to encounter, and also because it's difficult to give a definitive list of such special cases. \paragraph{Ledger structure.} We model a ledger as a simple list of transactions: a real blockchain ledger will be more complex than this, but the only property that we really require is that transactions in the ledger have some kind of address which allows them to be uniquely identified and retrieved. \subsubsection{The \ctx{} type} \label{sec:context} Recall from the introduction to \cref{sec:eutxo-1} that when a transaction input is being validated, the validator is supplied with an object of type \ctx{} which contains information about the pending transaction. The \ctx{} type for the current version of EUTXO-1 is defined in \cref{fig:ptx-1-types}, along with some related types. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \s{OutputInfo } &=&(\val: \qty,\\ & &\ \i{validatorHash}: \Address,\\ & &\ \datumHash: \DataHash)\\ \\ \s{InputInfo } &=&(\outputref: \s{OutputRef},\\ & &\ \i{validatorHash}: \Address,\\ & &\ \i{datumHash}: \DataHash,\\ & &\ \i{redeemerHash}: \DataHash,\\ & &\ \val: \qty)\\ \\ \ctx\s{ } &=&(\i{inputInfo}: \List{\s{InputInfo}},\\ & &\ \i{thisInput}: \N,\\ & &\ \i{outputInfo}: \List{\s{OutputInfo}},\\ & &\ \i{validityInterval}: \Interval{\slotnum},\\ & &\ \datumWits: \FinSup{\DataHash}{\Data},\\ & &\ \fee: \qty,\\ & &\ \forge: \qty)\\ \\ \mkContext: \eutxotx \times \s{Input} \times \s{Ledger} \rightarrow\\ \ctx &=& \mbox{\parbox[t]{6cm}{summarises a transaction in the context of an input and a ledger state}}\\ \\ %% Without the break after the right arrow, the = sign is well %% over halfway across the pages, which is horrible. This makes %% it a bit better, but not much. \toData: \ctx \rightarrow \Data &=& \mbox{encodes a \ctx{} as \Data} \end{array} \end{displaymath} \caption{The \ctx{} type for the EUTXO-1 model} \label{fig:ptx-1-types} \end{ruledfigure} \subsection{Remarks} \paragraph{The contents of \ctx{}.} The \ctx{} type is essentially a summary of the information contained in the $\eutxotx$ type in \cref{fig:eutxo-1-types}. The \fee{}, \forge{}, and \i{validityInterval} fields are copied directly from the pending transaction. The \i{outputInfo} field contains information about the outputs which will be produced if the pending transaction validates successfully: it contains only the address of the relevant validator, and the hash of the datum.\footnote{See \cref{note:datum-objects-in-ptx} for further explanation.} The \i{inputInfo} field contains information about the inputs to the pending transaction, but provides only the hashes of the validators and redeemers for the inputs. The \i{thisInput} field is an index pointing to the element of \i{inputInfo} relating to the input currently undergoing validation. % Note: in the code at the moment, the hashes of the validator % and redeemer scripts in inputInfo are allowed to be absent % when we have pubkey inputs. We're ignoring that special case here. \paragraph{Defining \mkContext{} and \toData.} Assuming we have an appropriate hashing function, it is straightforward to define \mkContext. For the implementation of \toData, note that the $inputs$ field is a \FinSet{} in \eutxotx{}, but a \List{} in \ctx{}. Therefore \toData{} has to introduce an ordering of the transaction inputs. Contract authors cannot make any assumptions about this ordering and therefore should ensure that their scripts pass or fail regardless of what particular permutation of transaction inputs they are presented with. Apart from that, the function \toData{} is implementation-dependent and we will not discuss it further. \paragraph{Determinism.} The information provided in the \ctx{} structure is sufficiently limited that the validation process becomes \textit{deterministic}, which has important implications for fee calculations. See \cref{note:validation-determinism} for further discussion. \subsection{Validity of EUTXO-1 transactions} \label{sec:eutxo-1-validity} A number of conditions must be satisfied in order for a transaction $t$ to be considered valid with respect to a ledger $l$. \Cref{fig:validation-functions-1} defines some auxiliary functions used in validation. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{lll} \multicolumn{3}{l}{\lookupTx : \s{Ledger} \times \TxId \rightarrow \eutxotx{}}\\ \lookupTx(l,id) &=& \textsf{find}(l, tx \mapsto \txId(tx) = id)\\ \\ \multicolumn{3}{l}{\txunspent : \eutxotx \rightarrow \FinSet{\s{OutputRef}}}\\ \txunspent(t) &=& \{(\txId(t),1), \ldots, (\txId(id),\left|t.outputs\right|)\}\\ \\ \multicolumn{3}{l}{\unspent : \s{Ledger} \rightarrow \FinSet{\s{OutputRef}}}\\ \unspent([]) &=& \emptymap \\ \unspent(t::l) &=& (\unspent(l) \setminus t.\inputs) \cup \txunspent(t)\\ \\ \multicolumn{3}{l}{\getSpent : \s{Input} \times \s{Ledger} \rightarrow \s{Output}}\\ \getSpent(i,l) &=& l\langle i.\outputref.\id \rangle.\outputs[i.\outputref.\idx] \end{array} \end{displaymath} \caption{Auxiliary functions for EUTXO-1 validation} \label{fig:validation-functions-1} \end{ruledfigure} \noindent The function $\lookupTx(l,id)$ looks up the unique transaction $T$ whose $\TxId$ is $id$, and can of course fail. However, during validation Rule~\ref{rule:all-inputs-refer-to-unspent-outputs} of Figure~\ref{fig:eutxo-1-validity} ensures that all of the transaction inputs refer to existing unspent outputs, and in these circumstances $\lookupTx$ will always succeed for the transactions of interest. We can now define what it means for a transaction $t$ of type $\eutxotx$ to be valid for a ledger $l$ during the slot $\msf{currentSlot}$: see \cref{fig:eutxo-1-validity}. Our definition combines Definitions 6 and 14 from \citep{Zahnentferner18-UTxO}, differing from the latter in \cref{rule:all-inputs-validate}. \todokwxm{Check this.} \begin{ruledfigure}{H} \begin{enumerate} \item \label{rule:slot-in-range} \textbf{The current slot is within the validity interval} \begin{displaymath} \msf{currentSlot} \in t.\i{validityInterval} \end{displaymath} \item \label{rule:all-outputs-are-non-negative} \textbf{All outputs have non-negative values} \begin{displaymath} \textrm{For all } o \in t.\outputs,\ o.\val \geq 0 \end{displaymath} \item \label{rule:all-inputs-refer-to-unspent-outputs} \textbf{All inputs refer to unspent outputs} \begin{displaymath} \{i.\outputref: i \in t.\inputs \} \subseteq \unspent(l). \end{displaymath} \item \label{rule:forging} \textbf{Forging} \\ A transaction with a non-zero \forge{} field is only valid if the ledger $l$ is empty (that is, if it is the initial transaction). \item \label{rule:value-is-preserved} \textbf{Value is preserved} \begin{displaymath} t.\forge + \sum_{i \in t.\inputs} \getSpent(i, l).\val = t.\fee + \sum_{o \in t.\outputs} o.\val \end{displaymath} \item \label{rule:no-double-spending} \textbf{No output is double spent} \begin{displaymath} \textrm{If } i_1, i_2 \in t.\inputs \textrm{ and } i_1.\outputref = i_2.\outputref \textrm{ then } i_1 = i_2. \end{displaymath} \item \label{rule:all-inputs-validate} \textbf{All inputs validate} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \llbracket i.\validator\rrbracket (i.\datum,\, i.\redeemer,\, \toData(\mkContext(t,i,l))) = \true. \end{displaymath} \item \label{rule:validator-scripts-hash} \textbf{Validator scripts match output addresses} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \scriptAddr(i.\validator) = \getSpent(i, l).\addr \end{displaymath} \item \label{rule:datum-objects-hash} \textbf{Datum objects match output hashes} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \hashData(i.\datum) = \getSpent(i, l).\datumHash \end{displaymath} \end{enumerate} \caption{Validity of a transaction $t$ in the EUTXO-1 model} \label{fig:eutxo-1-validity} \end{ruledfigure} \todokwxm{Do we really needs the $\llbracket\cdots\rrbracket$ business?} \noindent We say that a ledger $l$ is \textit{valid} if either $l$ is empty or $l$ is of the form $t::l^{\prime}$ with $l^{\prime}$ valid and $t$ valid for $l^{\prime}$. In practice, validity imposes a limit on the sizes of the $\validator$ \script{}, the $\redeemer$ and $\datum$ \Data{} fields, and the result of \toData. The validation of a single transaction must take place within one slot, so the evaluation of $\llbracket ~ \rrbracket$ cannot take longer than one slot. \todokwxm{Do we need this $\uparrow$?} %%\newpage \section{EUTXO-2: multicurrency support and non-fungible tokens} \label{sec:eutxo-2} We now extend the EUTXO-1 model further by introducing features which allow, among other things, the implementation of new currencies and \textit{non-fungible tokens} (NFTs). \paragraph{Multiple currencies.} The EUTXO-2 model allows an unlimited number of \textit{currencies}. Each custom currency has a unique identifier. \todokwxm{We may wish to implement a DEX to enable exchange of custom currencies. This is a little problematic in Ethereum because the Ether currency itself does not conform to ERC-20 and so has to be wrapped in another currency before it can participate in DEX trades. Will the fact that Ada is treated differently from custom currencies in EUTXO cause us difficulties here?} \paragraph{NFTs.} A non-fungible token (NFT) is a unique object which can be transferred to another user, but not duplicated. NFTs have proven useful in a number of blockchain applications (see~\citep{ERC-721} for example); for example, they can represent ownership of some object in a game. We can implement NFTs as custom currencies whose supply is limited to a single coin. \subsection{The definition of EUTXO-2} In order to support these extensions, we introduce several new types. Custom currencies are represented by unique \textit{currency identifiers} and each currency has a number of \textit{tokens} which partition each custom currency into a number of sub-currencies. The basic idea is that ordinary currencies have a single token whose sub-currency has an unlimited supply and NFTs have a number of tokens with the sub-currency for each token limited to a supply of one. The changes to the basic EUTXO-1 types are now quite simple: see \cref{fig:eutxo-2-types}. We change the type of the $\val$ field in the \s{Output} type to be \qtymap{}, representing values of all currencies; we also change the type of the \forge{} field on transactions to \qtymap{}, to allow the creation and destruction of funds in all currencies; the supply of a currency can be reduced by forging a negative amount of that currency, as in EUTXO-1. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \multicolumn{3}{l}{\textsc{Ledger primitives}}\\ \currency && \mbox{an identifier for a custom currency}\\ \token && \mbox{a type consisting of identifiers for individual tokens}\\ \\ \multicolumn{3}{l}{\textsc{Defined types}}\\ \qtymap &=& \FinSup{\currency}{\FinSup{\token}{\qty}}\\ \\ \s{Output}_2 &=&(\addr: \Address,\\ & &\ \val: \qtymap\\ & &\ \datumHash: \DataHash)\\ \\ \s{OutputRef}_2 &= &(\txrefid: \TxId, \idx: \s{Int})\\ \\ \s{Input}_2 &=&( \outputref: \sf{OutputRef}_2,\\ & &\ \validator: \script,\\ & &\ \datum: \Data,\\ & &\ \redeemer: \Data)\\ \\ \eutxotx_2 &=&(\inputs: \FinSet{\s{Input}_2},\\ & &\ \outputs: \List{\s{Output}_2},\\ & &\ \i{validityInterval}: \Interval{\slotnum},\\ & &\ \datumWits: \FinSup{\DataHash}{\Data},\\ & &\ \fee: \qtymap,\\ & &\ \forge: \qtymap)\\ \\ \s{Ledger}_2 &=&\!\List{\eutxotx_2}\\ \end{array} \end{displaymath} \caption{Extra primitives and basic types for the EUTXO-2 model} \label{fig:eutxo-2-types} \end{ruledfigure} \subsubsection{Remarks} \paragraph{ETUXO-2 on Cardano.} The Cardano implementation of EUTXO-2 uses the primitives given in \cref{fig:eutxo-2-types-cardano}. Cardano also defines an \emph{native currency} and \emph{native currency token}. This allows defining a native currency that behaves as a simple \qty{}. This is used in Fig~\ref{fig:cardano-fee-validity}. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \currency &=& \H\\ \token &=& \H\\ \nativeCur &=& \emptyBs\\ \nativeTok &=& \emptyBs\\ \end{array} \end{displaymath} \caption{Cardano primitives for the EUTXO-2 model} \label{fig:eutxo-2-types-cardano} \end{ruledfigure} \paragraph{\qtymap{}. } The \qtymap{} type represents a collection of funds from a number of currencies and their subcurrencies. \qtymap{} is a finitely-supported function \emph{to} another finitely-supported function. This is well-defined, since finitely-supported functions form a monoid. \subsection{The \ctx{} type for EUTXO-2} \label{sec:pendingtx-2} The \ctx{} type must be also be updated for the EUTXO-2 model. All that is required is to replace \qty{} by \qtymap{} everywhere in \cref{fig:ptx-1-types} except for the \fee{} field: for reference the details are given in \cref{fig:ptx-2-types}. \begin{ruledfigure}{H} \begin{displaymath} \begin{array}{rll} \s{OutputInfo}_2\s{ } &=&(\val: \qtymap,\\ & &\ \i{validatorHash}: \Address,\\ & &\ \datumHash: \DataHash)\\ \\ \s{InputInfo}_2\s{ } &=& (\outputref: \s{OutputRef},\\ & &\ \i{validatorHash}: \Address,\\ & &\ \i{datumHash}: \DataHash,\\ & &\ \i{redeemerHash}: \DataHash),\\ & &\ \val: \qtymap)\\ \\ \ctx_2\s{ } &=&(\i{inputInfo}: \List{\s{InputInfo}_2},\\ & &\ \i{thisInput}: \N,\\ & &\ \i{outputInfo}: \List{\s{OutputInfo$_2$}},\\ & &\ \i{validityInterval}: \Interval{\slotnum},\\ & &\ \datumWits: \FinSup{\DataHash}{\Data},\\ & &\ \fee: \qtymap,\\ & &\ \forge: \qtymap)\\ \\ \mkContext_2: \eutxotx_2 \times \s{Input} \times \s{Ledger} \rightarrow&\\ \ctx_2 &=& \mbox{\parbox[t]{6cm}{summarises a transaction in the context of an input and a ledger state}}\\ \\ \toData_2: \ctx_2 \rightarrow \Data &=& \mbox{encodes a $\ctx_2$} \end{array} \end{displaymath} \caption{The \ctx{} type for the EUTXO-2 model} \label{fig:ptx-2-types} \end{ruledfigure} \subsection{Validity of EUTXO-2 transactions} \label{sec:eutxo-2-validity} \bigskip \noindent The validity conditions from \cref{fig:eutxo-1-validity} must also be updated to take account of multiple currencies. We can now adapt the definition of validity for EUTXO-1 (\cref{fig:eutxo-1-validity}) to obtain a definition of validity for EUTXO-2: see \cref{fig:eutxo-2-validity}. \begin{ruledfigure}{H} \begin{enumerate} \item \label{rule:slot-in-range-2} \textbf{The current slot is within the validity interval} \begin{displaymath} \msf{currentSlot} \in t.\i{validityInterval} \end{displaymath} \item \label{rule:all-outputs-are-non-negative-2} \textbf{All outputs have non-negative values} \begin{displaymath} \textrm{For all } o \in t.\outputs,\ o.\val \geq 0 \end{displaymath} \item \label{rule:all-inputs-refer-to-unspent-outputs-2} \textbf{All inputs refer to unspent outputs} \begin{displaymath} \{i.\outputref: i \in t.\inputs \} \subseteq \unspent(l). \end{displaymath} \item \label{rule:forging-2} \textbf{Forging}\\ A transaction with a non-zero \forge{} field is only valid if either: \begin{enumerate} \item the ledger $l$ is empty (that is, if it is the initial transaction). \item \label{rule:custom-forge} for every key $h \in \support(t.\forge)$, there exists $i \in t.\inputs$ with $$ h = \getSpent(i,l).\addr; $$ in other words, some input must spend an output whose address is $h$. \end{enumerate} \item \label{rule:value-is-preserved-2} \textbf{Values are preserved} \begin{displaymath} t.\forge + \sum_{i \in t.\inputs} \getSpent(i, l) = t.\fee + \sum_{o \in t.\outputs} o.\val \end{displaymath} \item \label{rule:no-double-spending-2} \textbf{No output is double spent} \begin{displaymath} \textrm{If } i_1, i_2 \in t.\inputs \textrm{ and } i_1.\outputref = i_2.\outputref \textrm{ then } i_1 = i_2. \end{displaymath} \item \label{rule:all-inputs-validate-2} \textbf{All inputs validate} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \llbracket i.\validator\rrbracket(i.\datum,\, i.\redeemer,\, \toData_2(\mkContext_2(t, i, l))) = \true \end{displaymath} \item \label{rule:validator-scripts-hash-2} \textbf{Validator scripts match output addresses} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \scriptAddr(i.\validator) = \getSpent(i, l).\addr \end{displaymath} \item \label{rule:datum-objects-hash-2} \textbf{Datum objects match output hashes} \begin{displaymath} \textrm{For all } i \in t.\inputs,\ \hashData(i.\datum) = \getSpent(i, l).\datumHash \end{displaymath} \end{enumerate} \caption{Validity of a transaction $t$ in the EUTXO-2 model} \label{fig:eutxo-2-validity} \end{ruledfigure} \subsection{Remarks} \paragraph{Monetary policies.} Rule~\ref{rule:custom-forge} can be used to enforce monetary policies for custom currencies: see Note~\ref{note:monetary-policies} for a detailed explanation. \paragraph{Preservation of value over \qtymap{}.} In \cref{rule:value-is-preserved-2}, $+$ and $\sum$ operate over \qtymap{}, which is a finitely-supported function (which, with their operations, are defined in \cref{sec:fsfs}). Preservation of value in this model essentially requires that the quantities of each of the individual currencies involved in the transaction are preserved. \paragraph{Preservation of value and forging.} Recall that values in $\forge$ can be negative whereas values in outputs must be non-negative. This allows currency to be destroyed as well as created. The \cref{rule:value-is-preserved-2} implies that a transaction is invalid if it attempts to destroy more of a currency than is actually available in its inputs. \paragraph{Validation on Cardano.} Cardano adds an additional rule in Fig~\ref{fig:cardano-fee-validity}, which asserts that fees are paid exclusively in the native currency. \begin{ruledfigure}{H} \textbf{Fees are paid in the native currency} \begin{displaymath} \support(t.\fee) = \{ \nativeCur \} \textrm{ and } \support(t.\fee(\nativeCur)) = \{ \nativeTok \} \end{displaymath} \caption{Validity of a transaction $t$ in the EUTXO-2 model} \label{fig:cardano-fee-validity} \end{ruledfigure} \subsection{The EUTXO-2 model in practice.} See~\cite{Plutus-book} for examples of contracts which make use of the features of the EUTXO-2 model. See also \cref{note:monetary-policies,note:eutxo-2-implications,note:eutxo-2-performance} for comments on some technical aspects of the model. \appendix \section{Comments} \label{appendix:comments} \note{Computing with finitely-supported functions.} \label{note:finitely-supported-functions} We intend that finitely-supported functions are implemented as finite maps, with a failed map lookup corresponding to returning 0. However, there are two apparent difficulties: \begin{enumerate} \item The domain of a map does not correspond to the support of the function: values may be mapped to zero, thus appearing in the domain but not the support. \item Pointwise equality is hard to compute. \end{enumerate} However, both of these are easily ameliorated. We say that a set $w$ is a \textit{weak support} of a finitely-supported function $f$ if $\support(f) \subseteq w$. That is, a weak support contains all the points that are non-zero, but possibly also some points that are zero. It is easy to see that the domain of a map is a weak support for the finitely-supported function it represents. We can compute the support from the weak support by simply checking the function at each value and removing those that are zero. This is potentially expensive, but we only need to do it when we need the support, which we only do during the computation of \cref{rule:value-is-preserved-2}. Pointwise equality between two finitely-supported functions $f$ and $g$ is equivalent to checking pointwise equality only over the union of $\support(f)$ and $\support(g)$; or similarly over the union of a weak support of $f$ and of $g$. In particular, for finitely-supported functions represented as maps, we can check pointwise equality over the union of their domains. The same applies to checking partial ordering pointwise, which can similarly be done over the union of the weak support. \note{The Basic UTXO model: Outputs and scripts.} \label{note:basic-utxo} There is no well-defined notion of ownership for UTXOs. In many transactions an output will accrue to a single user who is then entitled to spend it at a later date. However, in general the notion of ownership is more complex: an output of a transaction might require the cooperation of several users before it could be spent, or it might not be spendable until some other condition has been met (for example a certain period of time may have to pass). At the extremes, an output could be spendable by anyone, or by no-one. In order to deal with this complexity, an output can be locked by a \textit{script}% \footnote{In the Cardano setting, scripts are Plutus Core programs~\citep{Plutus-Core-spec}.} which must be supplied with suitable evidence to unlock the output. In the basic model, each input to a transaction comes with a \i{validator} script which checks that the transaction is allowed to spend the output. In order to spend an output, the transaction supplies an object of type $\Data$, called the \i{redeemer}, which provides evidence that the transaction has the authority to do so;\footnote{The validator plays a role similar to that of BitCoin's \texttt{scriptPubKey} and the redeemer to \texttt{scriptSig}. } a process called \i{validation} is then performed which checks that the redeemer satisfies the conditions required by the validator. Before a transaction can proceed, all inputs must be successfully validated: if one or more inputs fails to validate then the transaction is rejected. A simple example of this is a \i{pay-to-pubkey} script, where the redeemer consists of a signature for the current transaction produced using a private key belonging to the owner of the output. The validator script (provided by the owner of the output) would check the signature using a known public key: if the public key corresponds to the private key then validation succeeds, otherwise it fails. Thus the output can only be spent by the owner of the relevant private key See \cref{note:scripts} for more information about validators in the EUTXO setting. \note{Inputs and outputs.} \label{note:inputs-and-outputs} A transaction has a \textsf{Set} of inputs but a \textsf{List} of outputs. This is for two reasons: \begin{itemize} \item We need a way to uniquely identify a transaction output, so that it can be referred to by a transaction input that spends it. The pair of a transaction id and an output index is sufficient for this, but other schemes are conceivable. \item Equality of transaction outputs is defined structurally. But that means that if we had two outputs paying $X$ to address $A$, then they would be equal and therefore if we kept them in a \s{Set} one would be lost. \end{itemize} \noindent An alternative design would be to include a unique nonce in transaction outputs (effectively: their index in the list), and then we could use this to identify them (and distinguish them from each other), and so we could keep them in a \s{Set} instead. \note{The datum.} \label{note:datum} The introduction of the datum increases the expressivity of the model considerably. For example, one can use a datum to propagate state between transactions, and this can be used to give a contract the structure of a finite state machine; the fact that the datum is part of the output and not the transaction means that the state can change without the transaction changing, which makes it easier to have an ``identity'' for an ongoing contract. \note{Fees and Costs.} \label{note:fees} Users may have to pay a fee in order to have a transaction executed. In a public blockchain an important reason for this is to deter hostile agents from carrying out denial of service attacks by submitting transactions which take a long time or use excessive amounts of memory. The precise details of fees in Cardano are outwith the scope of this document, and indeed have not been fully decided at the time of writing. However, we expect that the fee will include a component based on the size of the transaction (including its associated scripts), and also a so-called \textit{gas} charge to cover execution costs. We will have a model specifying the costs of individual operations during script execution; costs will be monitored dynamically during execution, and if the gas consumed ever exceeds the amount covered by the fee then the transaction will fail. \note{Scripts and Hashes.} \label{note:scripts} The spendability of an output is determined by its validator, and thus the validator for an output must be known at the time when the output is created (a completely new validator may be created, or an existing validator may be re-used). Conceptually the validator is part of the output, so it may be rather unexpected that \cref{fig:eutxo-1-types} defines the validator to be part of an \textit{input}, with the output only containing the address of the validator. The rationale for this is that a validator $V$ for an output $O$ is not required until $O$ is actually spent, which may be some time after $O$ was created. Recall from \cref{note:fees} that the cost of executing a transaction depends on the size of transaction, including the associated scripts. Thus the transaction that produces the validator only pays for the size of a hash (32 bytes) and the transaction that runs it pays for the full size of the script. This strategy also helps to reduce on-chain storage requirements, since validators can be stored off-chain until needed (and the presence of the hash in the output can be used to check that the correct validator is in fact being used when validation occurs), but unspent outputs persist on-chain in the UTXO set until they are eventually spent. The same strategy applies to datum objects. \note{Datum witnesses.} \label{note:datum-witnesses} Although a datum is only recorded as a hash in a transaction output, it is useful to be able to record the full value of the datum on the transaction that \emph{creates} an output: this allows observers to determine the full datum without it having to be kept in the UTXO set. This mechanism is \emph{optional}, since it incurs an increase in transaction size (and hence cost), and some clients may want to transmit the information off-chain instead to minimize these costs. Hence there is a $\datumWits$ field on transactions, which \emph{may} contain mappings from the $\DataHash$es used in the transaction to their \Data{} values. This information is also present in \ctx{}. \note{Datum objects in \ctx{}.} \label{note:datum-objects-in-ptx} In \cref{fig:ptx-1-types,fig:ptx-2-types} the \textsf{OutputInfo} does not include the datum attached to the output. These may be found in $\datumWits$. Having access to the value of the datum allows a validator to inspect an outgoing datum, for example to confirm that its contents are correct in some sense. This can be useful when a datum is used to propagate information about the state of a contract to later transactions. See~\cite{Plutus-book} for examples of this. \todokwxm{More specific reference?} \note{Determinism of the validation process.} \label{note:validation-determinism} The \ctx{} type is the only information about the ``outside world'' available to a validator at the time of validation. Allowing the validator access to this information gives the EUTXO models a considerable amount of power, as can be seen from the example contracts in~\cite{Plutus-book}. However, it is important not to make too much information available to the validator. The choice of the \ctx{} type above means that the information available to the validator is essentially independent of the state of the blockchain, and in particular, it is independent of time (note that the check that the current slot number is within a transaction's validity range takes place \textit{before} validation is initiated, and the slot number is not passed to the validator (although the validity range is)). This implies that validation is \textit{determinisitic} and validators can be run off-chain in order to determine their execution cost before on-chain validation actually occurs. This helps users to calculate transaction fees in advance and avoid the possibility of their transactions failing due to an insufficient fee having been paid (and also avoids overpayment due to overestimating the fees). \todokwxm{This may be a little optimistic. For example, in the crowdfunding contract we don't know in advance how many contributions will be made, or whether a campaign will succeed or fail. Thus we won't know how much the final transaction will cost until just before it happens.} \note{Monetary policies for custom currencies.} \label{note:monetary-policies} The new \textbf{Forging} rule in \cref{fig:eutxo-2-validity} enables custom currencies to implement their own monetary policies: for example, one might wish to place some limit on the amount of a currency that can be forged, or restrict the creation of the currency to owners of particular public keys. The idea is that a custom currency has a monetary policy which is defined by some script $H$, and the address $h = \scriptAddr(H)$ is used as the identifier of the currency. Whenever a new quantity of the currency is forged, \cref{rule:custom-forge,rule:all-inputs-validate-2,rule:validator-scripts-hash-2} imply that $H$ must be executed; $H$ is provided with the \forge{} field of the transaction via the \ctx{} object, and so it knows how much of the currency is to be forged and can respond appropriately. The advantage of this scheme is that custom currencies can be handled entirely within the smart contract system, without the need to introduce any extra blockchain infrastructure such as a central registry of custom currencies. In practice some refinement of this scheme will be required in order to (a) allow re-use of a monetary policy for different currencies, and (b) prevent unauthorised forging of a currency. To deal with (a) we can make the validator script unique by including a nonce. This still doesn't prevent unauthorised people from using the script $H$ to produce currency, but this can be prevented by, for instance, embedding a reference to an unspent output in the script and requiring that the currency can only be forged if the referenced output is spent at the same time, so it can only be forged once. \smallskip \todokwxm{Is this exactly what we want? I think that gives you a single opportunity to forge the currency, and then you can't make any more. I suppose that you could restrict the ability to forge to a particular individual by requiring them to provide something signed by their private key when they want to create new money. Also, is this a case where an output of value zero would be useful?} \smallskip \todokwxm{The use of the term ``forging'' is a bit confusing here, since it also means ``counterfeiting''.} % \todojm{I think there are two different concerns: % (1) Using the same monetary policy multiple times; % (2) Preventing unauthorised forging of a currency. % The "referencing an unspent output" trick accomplishes % both but is mainly aimed at (2). You could also achieve (1) by % embedding any kind of random data in the validator script, like a % nonce. % } \note{Implications of the EUTXO-2 model.} \label{note:eutxo-2-implications} The EUTXO-2 model and the techniques described in \cref{note:monetary-policies} allow us to implement fungible (normal) and non-fungible token currencies, as well as ``mixed states'': \begin{itemize} \item Standard (fungible) currencies are implemented by issuing currencies with a single \token{}. \item Non-fungible token currencies are implemented by only ever issuing single quantities of many unique \token{}s. \item Note that there is nothing in this model which enforces uniqueness: having multiples of a single \token{} merely means that those can be used fungibly. If a currency wants to make sure it only issues unique tokens it must track this itself. These ``mixed'' token currencies can have many \token{}s, but these can have more than unit quantities in circulation. These can be useful to model distinct categories of thing where there are fungible quantities within those, for example share classes. \end{itemize} \note{Performance issues for EUTXO-2.} \label{note:eutxo-2-performance} The EUTXO-2 model will lose some efficiency in comparison to the EUTXO-1 model, simply because the data structures are more complicated. This would even apply to transactions which only involve the native currency (if there is one), since it would be necessary to check whether the \qtymap{} contains anything that needs to be processed. If this is a concern then one could implement a model with two types of transaction, essentially just the disjoint union of the EUTXO-1 and EUTXO-2 transaction types. A simple case distinction at the start of a transaction could then select either a fast native-currency-only computation or a slower multicurrency computation. This would be harder to maintain though. \smallskip Another optimisation would be possible if one wished to implement custom currencies but not NFTs: since in this case every currency would only have a single token, the tokens could be omitted and the \qtymap{} replaced with a map from currency ids to quantities. \smallskip A more significant cost may be that we can no longer use \verb|{-# UNPACK #-}| when our \qty{} type stops being a simple combination of wrappers and products around primitives, but this is again an issue with any multi-currency proposal. \bibliographystyle{plainnat} %% ... or whatever \bibliography{extended-utxo-specification} \end{document} \input{../default/latex/header.tex} \addbibresource{latex/lit.bib} \subject{V000} % Versuchsnummer \title{Vorlage} % Versuchstitel \date{% Durchführung: DATUM \hspace{3em} Abgabe: DATUM } \begin{document} \maketitle \thispagestyle{empty} \tableofcontents \newpage \input{latex/zielsetzung.tex} \input{latex/theorie.tex} \input{latex/durchfuehrung.tex} \input{latex/auswertung.tex} \input{latex/diskussion.tex} \printbibliography{} \end{document} \documentclass[12pt]{article} \usepackage[authoryear]{natbib} \setcitestyle{notesep={: }} \begin{document} This thesis bases on the empirical work of Frumkes \citep{brain}. The next theorem shows how to earn more money if you sell bulk trash. This work is founded by Dennis and Matthias \citep{money}. \begin{thebibliography}{} \bibitem[Frumkes(2001)]{brain} . (2001). ``\textit{How to Raise Your I.Q. by Eating Gifted Children}.'' iUniverse. \bibitem[Dennis and Matthias(2017)]{money} ., . (2017). ``\textit{How we earn more money}.'' Fachschaft VWL. \end{thebibliography} \end{document}\hypertarget{_issue1335_test_8php}{}\doxysection{vendor/phpunit/phpunit/tests/end-\/to-\/end/regression/\+Git\+Hub/1335/\+Issue1335\+Test.php File Reference} \label{_issue1335_test_8php}\index{vendor/phpunit/phpunit/tests/end-\/to-\/end/regression/GitHub/1335/Issue1335Test.php@{vendor/phpunit/phpunit/tests/end-\/to-\/end/regression/GitHub/1335/Issue1335Test.php}} \doxysubsection*{Data Structures} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{class_issue1335_test}{Issue1335\+Test}} \end{DoxyCompactItemize} CarreGameEngine/Documentation/Doxygen/latex/structbt_soft_body_1_1e_p_solver.tex1-10 \hypertarget{structbt_soft_body_1_1e_p_solver}{ \section{btSoftBody::ePSolver Struct Reference} \label{structbt_soft_body_1_1e_p_solver}\index{btSoftBody::ePSolver@{btSoftBody::ePSolver}} } \hyperlink{structbt_soft_body_1_1e_p_solver}{ePSolver} : positions solvers {\tt \#include $<$btSoftBody.h$>$} \subsection*{Public Types} \begin{CompactItemize} \item enum \hyperlink{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86}{\_\-} \{ , \hyperlink{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86d1de203597e3f891a95c3aaf4f805e84}{Anchors}, \hyperlink{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf8618826149c80118f73e906b4966c69891}{RContacts}, \hyperlink{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf861ff50fff0983fbc2e4ae19997d7bfb4e}{SContacts}, \hyperlink{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86f5b5c91f8cb6d3d29ee7bb11e1ce5610}{END} \} \end{CompactItemize} \subsection{Detailed Description} \hyperlink{structbt_soft_body_1_1e_p_solver}{ePSolver} : positions solvers Definition at line 102 of file btSoftBody.h. \subsection{Member Enumeration Documentation} \hypertarget{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86}{ \index{btSoftBody::ePSolver@{btSoftBody::ePSolver}!\_\-@{\_\-}} \index{\_\-@{\_\-}!btSoftBody::ePSolver@{btSoftBody::ePSolver}} \subsubsection[\_\-]{\setlength{\rightskip}{0pt plus 5cm}enum {\bf btSoftBody::ePSolver::\_\-}}} \label{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86} \begin{Desc} \item[Enumerator: ]\par \begin{description} \index{Anchors@{Anchors}!btSoftBody::ePSolver@{btSoftBody::ePSolver}}\index{btSoftBody::ePSolver@{btSoftBody::ePSolver}!Anchors@{Anchors}}\item[{\em \hypertarget{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86d1de203597e3f891a95c3aaf4f805e84}{ Anchors} \label{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86d1de203597e3f891a95c3aaf4f805e84} }]Linear solver. \index{RContacts@{RContacts}!btSoftBody::ePSolver@{btSoftBody::ePSolver}}\index{btSoftBody::ePSolver@{btSoftBody::ePSolver}!RContacts@{RContacts}}\item[{\em \hypertarget{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf8618826149c80118f73e906b4966c69891}{ RContacts} \label{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf8618826149c80118f73e906b4966c69891} }]Anchor solver. \index{SContacts@{SContacts}!btSoftBody::ePSolver@{btSoftBody::ePSolver}}\index{btSoftBody::ePSolver@{btSoftBody::ePSolver}!SContacts@{SContacts}}\item[{\em \hypertarget{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf861ff50fff0983fbc2e4ae19997d7bfb4e}{ SContacts} \label{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf861ff50fff0983fbc2e4ae19997d7bfb4e} }]Rigid contacts solver. \index{END@{END}!btSoftBody::ePSolver@{btSoftBody::ePSolver}}\index{btSoftBody::ePSolver@{btSoftBody::ePSolver}!END@{END}}\item[{\em \hypertarget{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86f5b5c91f8cb6d3d29ee7bb11e1ce5610}{ END} \label{structbt_soft_body_1_1e_p_solver_5d6ab41a09da7333bc2047b4ea14bf86f5b5c91f8cb6d3d29ee7bb11e1ce5610} }]Soft contacts solver. \end{description} \end{Desc} Definition at line 102 of file btSoftBody.h. \begin{Code}\begin{verbatim}102 { enum _ { 103 Linear, 104 Anchors, 105 RContacts, 106 SContacts, 107 END 108 };}; \end{verbatim} \end{Code} The documentation for this struct was generated from the following file:\begin{CompactItemize} \item C:/Users/New/Documents/Games\_\-Technology/Year4\_\-Semester1/ICT397/$\sim$My Work/Assignment2/ICT397Carre/CarreGameEngine/Dependencies/BulletPhysicsEngine/include/BulletSoftBody/btSoftBody.h\end{CompactItemize} \hypertarget{classPwdService}{}\section{Pwd\+Service Class Reference} \label{classPwdService}\index{Pwd\+Service@{Pwd\+Service}} Service for handling P\+WD ftp command. {\ttfamily \#include $<$pwdservice.\+h$>$} Inheritance diagram for Pwd\+Service\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=184pt]{de/dce/classPwdService__inherit__graph} \end{center} \end{figure} Collaboration diagram for Pwd\+Service\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=184pt]{d4/de5/classPwdService__coll__graph} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hyperlink{classPwdService_a5008a00dba5c8867e11196c0c70e6e79}{Pwd\+Service} (int command\+Channel\+Socket) \end{DoxyCompactItemize} \subsection*{Additional Inherited Members} \subsection{Detailed Description} Service for handling P\+WD ftp command. \subsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{classPwdService_a5008a00dba5c8867e11196c0c70e6e79}\label{classPwdService_a5008a00dba5c8867e11196c0c70e6e79}} \index{Pwd\+Service@{Pwd\+Service}!Pwd\+Service@{Pwd\+Service}} \index{Pwd\+Service@{Pwd\+Service}!Pwd\+Service@{Pwd\+Service}} \subsubsection{\texorpdfstring{Pwd\+Service()}{PwdService()}} {\footnotesize\ttfamily Pwd\+Service\+::\+Pwd\+Service (\begin{DoxyParamCaption}\item[{int}]{command\+Channel\+Socket }\end{DoxyParamCaption})} \begin{DoxyParams}{Parameters} {\em command\+Channel\+Socket} & Socket to communication with client on command channel \\ \hline \end{DoxyParams} The documentation for this class was generated from the following files\+:\begin{DoxyCompactItemize} \item src/controller/services/srv/pwdservice.\+h\item src/controller/services/srv/pwdservice.\+cpp\end{DoxyCompactItemize} bernhard-schaaf-christmann/scaloppinageheim/dia-puzzle.tex \documentclass[12pt]{article} \usepackage{graphicx} \usepackage{amsmath} \newcommand{\pic}[1]{\mathord{\includegraphics[height=1.6ex]{#1}}} \begin{document} \thispagestyle{empty} \begin{equation*} \begin{split} &\pic{"ad"} + \pic{"serg"} + \pic{"bem"} + \pic{"min"} + \pic{"iss"}\\ +& %\pic{"stem"} ( \pic{"E24"} \setminus \pic{"rsm"} ) + ( \{\pic{"cws"}\} \oplus \{\pic{"as"}\} \oplus \{\pic{"caw"}\} )\\ %+ ( \pic{"cws"} \triangle \pic{"as"} \triangle \pic{"caw"} )\\ +& %\pic{"pa"} + \textrm{Tel}(131) +( \pic{"thol"} - \pic{"olid"} ) + \pic{"eliz"} = ?? \end{split} \end{equation*} \end{document} \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command}{\section{strange.\-unittests.\-Test\-Sequence\-Command Class Reference} \label{classstrange_1_1unittests_1_1_test_sequence_command}\index{strange.\-unittests.\-Test\-Sequence\-Command@{strange.\-unittests.\-Test\-Sequence\-Command}} } \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command_a826599b97594702465066fb05e01fcd1}{void {\bfseries Test\-Missing\-Execute} ()}\label{classstrange_1_1unittests_1_1_test_sequence_command_a826599b97594702465066fb05e01fcd1} \item \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command_a55d8ded5eb253080ba581beb4b03b449}{void {\bfseries Test\-Successful\-Execute} ()}\label{classstrange_1_1unittests_1_1_test_sequence_command_a55d8ded5eb253080ba581beb4b03b449} \item \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command_a00296442b91a0586755900740a58a92a}{void {\bfseries Test\-Retain\-Release} ()}\label{classstrange_1_1unittests_1_1_test_sequence_command_a00296442b91a0586755900740a58a92a} \item \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command_aaa3fe25f76b836afaa6d1426e6f25ee0}{void {\bfseries Test\-Cancel} ()}\label{classstrange_1_1unittests_1_1_test_sequence_command_aaa3fe25f76b836afaa6d1426e6f25ee0} \item \hypertarget{classstrange_1_1unittests_1_1_test_sequence_command_a90f39312bbaf40b85fcdf0e1bad69275}{void {\bfseries Test\-Sequence\-Id} ()}\label{classstrange_1_1unittests_1_1_test_sequence_command_a90f39312bbaf40b85fcdf0e1bad69275} \end{DoxyCompactItemize} The documentation for this class was generated from the following file\-:\begin{DoxyCompactItemize} \item Assets/scripts/strange/tests/extensions/sequencer/Test\-Sequence\-Command.\-cs\end{DoxyCompactItemize} \documentclass[11pt]{article} \usepackage{times} \usepackage{epsf} \usepackage{epsfig} \usepackage{amsmath, alltt, amssymb, xspace} \usepackage{wrapfig} \usepackage{fancyhdr} \usepackage{url} \usepackage{verbatim} \usepackage{fancyvrb} \usepackage{float} \usepackage{subfigure} \usepackage{cite} \usepackage{hyperref} \hypersetup{% pdfborder = {0 0 0} } \topmargin -0.50in % distance to headers \oddsidemargin 0.0in \evensidemargin 0.0in \textwidth 6.5in \textheight 8.9in %\centerfigcaptionstrue %\def\baselinestretch{0.95} \newcommand\discuss[1]{\{\textbf{Discuss:} \textit{#1}\}} %\newcommand\todo[1]{\vspace{0.1in}\{\textbf{Todo:} \textit{#1}\}\vspace{0.1in}} \newtheorem{problem}{Problem}[section] %\newtheorem{theorem}{Theorem} %\newtheorem{fact}{Fact} \newtheorem{define}{Definition}[section] %\newtheorem{analysis}{Analysis} \newcommand\vspacenoindent{\vspace{0.1in} \noindent} %\newenvironment{proof}{\noindent {\bf Proof}.}{\hspace*{\fill}~\mbox{\rule[0pt]{1.3ex}{1.3ex}}} %\newcommand\todo[1]{\vspace{0.1in}\{\textbf{Todo:} \textit{#1}\}\vspace{0.1in}} %\newcommand\reducespace{\vspace{-0.1in}} % reduce the space between lines %\def\baselinestretch{0.95} \newcommand{\fixmefn}[1]{ \footnote{\sf\ \ \fbox{FIXME} #1} } \newcommand{\todo}[1]{ \vspace{0.1in} \fbox{\parbox{6in}{TODO: #1}} \vspace{0.1in} } \newcommand{\mybox}[1]{ \vspace{0.2in} \noindent \fbox{\parbox{6.5in}{#1}} \vspace{0.1in} } \newcounter{question} \setcounter{question}{1} \newcommand{\myquestion} {{\vspace{0.1in} \noindent \bf Question \arabic{question}:} \addtocounter{question}{1} \,} \newcommand{\myproblem} {{\noindent \bf Problem \arabic{question}:} \addtocounter{question}{1} \,} \newcommand{\copyrightnotice}[1]{ \vspace{0.1in} \fbox{\parbox{6in}{ This lab was developed for the Labtainer framework by the Naval Postgraduate School, Center for Cybersecurity and Cyber Operations under sponsorship from the DoD CySP program. This work is in the public domain, and cannot be copyrighted.}} \vspace{0.1in} } \newcommand{\idea}[1]{ \vspace{0.1in} {\sf IDEA:\ \ \fbox{\parbox{5in}{#1}}} \vspace{0.1in} } \newcommand{\questionblock}[1]{ \vspace{0.1in} \fbox{\parbox{6in}{#1}} \vspace{0.1in} } \newcommand{\argmax}[1]{ \begin{minipage}[t]{1.25cm}\parskip-1ex\begin{center} argmax #1 \end{center}\end{minipage} \; } \newcommand{\bm}{\boldmath} \newcommand {\bx} {\mbox{\boldmath $x$}} \newcommand {\by} {\mbox{\boldmath $y$}} \newcommand {\br} {\mbox{\boldmath $r$}} \newcommand{\tstamp}{\today} %\rfoot[\fancyplain{\tstamp} {\tstamp}] {\fancyplain{}{}} \pagestyle{fancy} \lhead{\bfseries Labtainers} \chead{} \rhead{\small \thepage} \lfoot{} \cfoot{} \rfoot{} 10-100 % % Copyright (C) 1998-2021 http://www.texnia.com % % This file may be distributed and/or modified under the conditions of % the MIT License. A version can be found at the end of this file. % % Repository: https://github.com/jbezos/titlesec % % Notes % ~~~~~ % % The following tags are used: % ttl@ : the generic tag used through the style % ttlh@ : a shape definition % ttlf@ : a macro containing the title format % ttls@ : id. the title space % ttlp@ : page key related macros % ttll@ : level number % % The ttlf@ and ttls@ contains data in the form {..}{..}. % Perhaps in future releases they should be converted % to a prop-like list, similar to that proposed by the % latex team. % % Admittedly, the current implementation seems too % complicated, but that's necessary in order to provide % certain compatibility with the sections as defined by the % used class. Other packages opt for providing the sections % as defined by standard classes ignoring the class; for % instance sectsty which does a simple task in a simple and % nice way. However, that was not my goal. % % Release % ~~~~~~~ \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{titlesec}[2021/07/05 v2.14 Sectioning titles] % Initialization % ~~~~~~~~~~~~~~ \newif\ifttl@ps \ttl@psfalse % The \ttl@label switch is used when printing the label in titles. % A numberless variant makes it to true. % There is a \ttl@toclabel as well, which is true iff the % title is numbered; used in toc entries (except default part % and chapter) and marks (only in titlesec pagestyles). \newif\ifttl@label \newif\ifttl@toclabel \newbox\ttl@box % A provision for the report style: \@ifundefined{if@mainmatter} {\let\if@mainmatter\iftrue}{} \@ifundefined{if@openright} {\let\if@openright\iftrue}{} % and the ams styles as well \@ifundefined{@chapapp} {\let\@chapapp\chaptername}{} \def\ttl@trylist{\ttl@try{}} \def\ttl@getkeys#1#2{% \if\expandafter @\@gobble#1@\@empty \edef\ttl@b{\expandafter\@gobble\string#1}% \let\ttl@a\ttl@b \else \ttl@keys \ttl@getkeys{#1}{#2}% \fi} % A more meaningful error for \@notdefinable \expandafter\AtEndOfPackage\expandafter{\expandafter \gdef\expandafter\@notdefinable\expandafter{\@notdefinable}} \def\@notdefinable{% \PackageError{titlesec}% {Incompatible package}% {Titlesec cannot continue defining its own macros because\MessageBreak \@backslashchar\reserved@a\space is already used by other package, the class\MessageBreak or the document.}} % +-----------------+ % | C L A S S E S | % +-----------------+ \def\ttl@useclass#1#2{% \@ifstar {\ttl@labelfalse#1{#2}[]}% {\ttl@labeltrue\@dblarg{#1{#2}}}} \def\ttl@straightclass{\ttl@useclass\ttl@straight@i} \def\ttl@partclass{\ttl@useclass\ttl@part@i} \def\ttl@topclass{\ttl@useclass\ttl@top@i} \def\ttl@pageclass{\ttl@useclass\ttl@page@i} % Here \scantokens is used to make sure the unescaped name % has `letters' and no `others'. Mainly for hyperref, so there % should be no problems. \newcommand\titleclass[1]{% \edef\ttl@a{\expandafter\@gobble\string#1}% \ifx\scantokens\@undefined\else \scantokens\expandafter{\expandafter \def\expandafter\ttl@a\expandafter{\ttl@a}}% \fi \@ifnextchar[{\@tempswatrue\ttl@class@i{#1}}% {\@tempswafalse\ttl@class@ii{#1}}} \def\ttl@class@i#1[#2]{% \@namedef{ttll@\ttl@a}{#2}% \expandafter\providecommand\csname\ttl@a title\endcsname{}%%%% \@ifundefined{ttl@toplevel}{}% {\expandafter\let\csname ttlss@\ttl@a\expandafter\endcsname \csname ttlss@\ttl@toplevel\endcsname}% \edef\ttl@toplevel{\ttl@a}% \ttl@class@ii{#1}} \def\ttl@class@ii#1#2{% \@ifundefined{ttl@#2class}% {\PackageError{titlesec}{Unknown sectioning class}% {Valid names are top, page and straight}}% {\expandafter\let\csname ttl@compat\ttl@a\endcsname\relax \@ifundefined{\ttl@a mark}% {\@namedef{\ttl@a mark}{\@gobble}}% {}% \edef#1{% \expandafter\noexpand\csname ttl@#2class\endcsname{\ttl@a}}}% \if@tempswa \expandafter\@gobble \else \expandafter\@firstofone \fi {\@ifnextchar[% {\ttl@class@iii}% {\@ifundefined{ttll@\ttl@a}% {\PackageError{titlesec}{Unknown sectioning level}% {\string\titleclass\space with no optional arguments\MessageBreak only changes the class of an *existing* level}}}}} \def\ttl@class@iii[#1]{% \edef\ttl@b{\expandafter\@gobble\string#1}% \expandafter\let\csname ttlss@\ttl@a\expandafter\endcsname \csname ttlss@\ttl@b\endcsname \expandafter\edef\csname ttlss@\ttl@b\endcsname{\ttl@a}% \let\ttl@a\ttl@toplevel \count@\csname ttll@\ttl@toplevel\endcsname \ttl@class@iv} \def\ttl@class@iv{% \@ifundefined{ttlss@\ttl@a}{}% {\advance\count@\@ne \edef\ttl@a{\csname ttlss@\ttl@a\endcsname}% \expandafter\edef\csname ttll@\ttl@a\endcsname{\the\count@}% \ttl@class@iv}} % Typesetting Classes: General tools % ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ % The following command handles the *n spacing % Some tricks are necessary to multiply a % skip by a non integer number \newskip\beforetitleunit \beforetitleunit=1ex\@plus.3ex\@minus.06ex \newskip\aftertitleunit \aftertitleunit=1ex\@plus.1ex \newdimen\ttl@plus \newdimen\ttl@minus \def\ttl@assign#1{% \@ifstar {\ttl@assign@i{#1}}% {\ttl@assign@d{#1}}} \def\ttl@assign@i#1#2\relax#3{% \ttl@plus\z@ \ttl@minus\z@ \afterassignment\ttl@assign@ii \dimen@\the#3, % <- space #1 = #2\dimen@ plus #2\ttl@plus minus #2\ttl@minus} \def\ttl@assign@ii#1 {% <- space \if#1,\else\afterassignment\ttl@assign@ii\fi \csname ttl@\string#1\endcsname} \def\ttl@assign@d#1#2\relax#3{\setlength#1{#2}} % To be used with \v/vspace to make them calc-savvy \def\ttl@calc#1#2{% {\setlength\@tempskipa{#2}% #1\@tempskipa}} \def\ttl@calcneg#1#2{% {\setlength\@tempskipa{#2}% #1{-\@tempskipa}}} % Gets from ttls@ and passes the spacing parameters: \def\ttl@startargs#1#2{% Get the first arguments, with the spacing \@ifundefined{ttlp@#2}% {\let\ttl@key@page\@empty}% {\ttlp@fetch{#2}}% \begingroup \def\ttl@b{ttls@#2}% \edef\ttl@key@numberless{\ifttl@label//\else/*\fi}% \def\ttl@a##1{\csname ttl@key@##1\endcsname}% Used as elt in try \ttl@trylist \xdef\ttl@b{\ttl@c}% \endgroup \ifx\ttl@b\@empty \PackageError{titlesec}{Format/spacing not found}% {I was unable to find the format corresponding to #2.\MessageBreak Maybe you haven't set it with \string\titleformat\space and \string\titlespacing} \fi \expandafter#1\ttl@b{#2}} % Used in ttl@select \def\ttl@savefn#1[#2]#3{% \ifcase#1% \footnotemark[#2]% \gdef\ttl@fn{\footnotetext[#2]{#3}}% \else \footnotemark \gdef\ttl@fn{\footnotetext{#3}}% \fi} \def\ttl@nest@error{% \PackageError{titlesec}{Nested titles}{Titles must not be nested}} \def\ttl@hmode@error{% \PackageError{titlesec}{Entered in horizontal mode} {The argument cannot contain horizontal material\MessageBreak such as text, \string\noindent, \string\makebox, etc.}} % \ttl@select not only selects the right version to be % used. It also take steps to ensure that a mark % is not lost inside a box by saving it into \ttl@mk, % which in turn is used by the sect and chap commands. % As of 2019 and due the LaTex % kernel modifies \markboth, we consider two possibilities (2.13). \newif\ifttl@explicit \def\ttl@gmk#1{\gdef\ttl@mk{#1}} \def\ttl@select#1#2#3#4{% \ttl@Hy@saveanchor \global\let\ttl@mk\@empty % global because of rigidchapters \global\let\ttl@fn\@empty \begingroup \if@inlabel\else % Keep item's \everypar \everypar{\setbox\z@\lastbox\ttl@strut}% \fi \let\ttl@straight@i\ttl@nest@error \let\ttl@top@i \ttl@nest@error \let\ttl@part@i \ttl@nest@error \let\ttl@page@i \ttl@nest@error \let\ttl@newpage\newpage \def\newpage{\ttl@savewrite\ttl@newpage}% \expandafter\ifx\csname markboth \endcsname\relax \def\markboth##1##2{\protect\ttl@gmk{\protect\markboth{##1}{##2}}}% \def\markright##1{\protect\ttl@gmk{\protect\markright{##1}}}% \else \@namedef{markboth }##1##2{\protect\ttl@gmk{\markboth{##1}{##2}}}% \@namedef{markright }##1{\protect\ttl@gmk{\markright{##1}}}% \fi \def\@mkboth##1##2{\protect\ttl@gmk{\protect\@mkboth{##1}{##2}}}% \def\footnote{\@ifnextchar[% {\ttl@savefn\z@}{\ttl@savefn\@ne[]}}% \edef\ttl@key@numberless{\ifttl@label//\else/*\fi}% \def\ttl@b{ttlf@#1}% \def\ttl@a##1{\csname ttl@key@##1\endcsname}% Used as elt in try \ttl@trylist \ifx\ttl@c\@empty \PackageError{titlesec}{No format for this command}% {If you (re)set the class of a sectioning command, you may\MessageBreak you may need to (re)define its format with \string\titleformat}% \fi \ifttl@explicit \def\ttl@passexplicit{\ttl@case{#4}}% \ttl@c{#4}{#2}{#3}{}% ttl@c is returned by ttl@try with ttlf@... \else \let\ttl@passexplicit\ttl@case \ttl@c{#2}{#3}{#4}% ttl@c is returned by ttl@try with ttlf@... \fi \endgroup} \let\ttl@savewrite\@empty \def\ttl@finmarks{% \ttl@savewrite \ttl@mk % Contains a possible mark, returned by \ttl@select \ttl@fn} % And a footnote \def\ttl@try#1{% \edef\ttl@c{#1}% #1 is a list in the form \ttl@a{key}\ttl@a{key} \@ifundefined{\ttl@b\ttl@c}{}{% \edef\ttl@c{\expandafter\noexpand\csname\ttl@b\ttl@c\endcsname}% \def\ttl@a##1{\csname ttl@extra@##1\endcsname}% #1% \let\ttl@try\@gobble}} % locally modified to `break' testings % \ttl@write writes marks and toc. tocdepth is taken care of when % the toc is typesetted and not here. Used always through % ttl@savewrite, which is reset to \@empty to avoid duplicated % calls. \def\ttl@write#1#2{% \ttl@blinemarks \csname#1mark\endcsname{#2}% \def\ttl@a{\protect\numberline{\@nameuse{the#1}}}% \@nameuse{ttl@toc#1}% eg, \ttl@tocpart modifies \ttl@a \ttl@addcontentsline{#1}{#2}% Depends on toctitles, uses \ttl@a \ttl@elinemarks \global\ttl@toclabelfalse \global\let\ttl@savewrite\@empty} \newif\ifttl@premark % to be used in ttlps.def \ttl@premarkfalse % 2019-06-20. Added the \lastskip stuff, because a mark 'forgets' the % last skip. \def\ttl@premark#1#2{% \let\ttl@lastskip\relax \ifvmode \ifdim\lastskip=\z@\else \edef\ttl@lastskip{\the\lastskip}% \vskip-\ttl@lastskip\relax \fi \fi \protected@xdef\ttl@prevmarks{\ttl@marks}% \ttl@blinemarks \csname#1mark\endcsname{#2}% \ttl@elinemarks \ifx\ttl@lastskip\relax\else \vskip\ttl@lastskip\relax \fi \gdef\ttl@prevmarks{\ttl@marks}} % Must be preceded by a default \ttl@savewrite, which is used % in starred variants--\@empty in top and straight classes. % In straight class, it is preceded by the setting of % prev marks to provide a "fixed" top mark. Otherwise, % the default prev mark (= curr mark) is used (restored % after ttl@labelling in straight). This is the command % to be hacked if you want to change the behaviour of % starred variants. \def\ttl@labelling#1#2{% \let\ttl@Hy@saveanchor\@empty \ifttl@label % 1st - if star \def\ttl@savewrite{\ttl@write{#1}{#2}}% \@nameuse{ttl@#1label}% eg, sets if mainmatter in chapter. \ifttl@label % 2nd - eg, if not main matter \ifnum\@nameuse{ttll@#1}>\c@secnumdepth\relax \ttl@labelfalse % 3rd - if too deep \else \ttl@Hy@refstepcounter{#1}% \@nameuse{ttl@#1out}% \fi \fi \fi \let\ifttl@toclabel\ifttl@label \ifx\ttl@savewrite\@empty\else % If marks \ifttl@ps \ifttl@premark \global\ttl@premarkfalse \else % if no \pretitlemark \ttl@premark{#1}{#2}% \fi \fi \ifttl@label\else\ttl@Hy@steplink{#1}\fi \fi} % Executed by ttl@labelling if the name of section is chapter: \def\ttl@chapterlabel{\if@mainmatter\else\ttl@labelfalse\fi} % Executed by ttl@labelling if chapter has a number. Note % you can define messages for other sectioning levels (eg, % \ttl@sectionout). \def\ttl@chapterout{\typeout{\chaptertitlename\space\thechapter.}} % Straight class % ~~~~~~~~~~~~~ % Default for nobottomtitles. Changed by nobottomtitles* \def\ttl@addstretch{\advance\@tempskipa-\pagestretch} % 1:name 2:level 3:indent 4:before 5:after 6:afind [7]:cap 8:title % The second argument of ttl@sect is the level, which % is empty if the star version is used. In this case % neither the toc nor the marks are written. \def\ttl@straight@i#1[#2]#3{% \def\@currentlabelname{#2}% for nameref \gdef\ttl@savemark{\csname#1mark\endcsname{#3}}% \let\ttl@savewrite\@empty \def\ttl@savetitle{#3}% \gdef\thetitle{\csname the#1\endcsname}% \if@noskipsec \leavevmode \fi \par \ttl@labelling{#1}{#2}% \ttl@startargs\ttl@straight@ii{#1}{#3}} % 1:left 2:right 3:before 4:after 5:afterindent 6:name 7:title \def\ttl@straight@ii#1#2#3#4#5#6#7{% \ttl@assign\@tempskipa#3\relax\beforetitleunit \@ifundefined{ttl@ps@#6}{}% {\PackageWarning{titlesec}{Page style in straight class ignored}}% \if@nobreak \ttl@titlespace{\@tempskipa}% \else \@ifundefined{#6break}% {\addpenalty{\@secpenalty}}% {\csname#6break\endcsname}% \addvspace{\@tempskipa}% \ifdim\bottomtitlespace<\z@ \else \begingroup \@tempskipb\pagegoal \@tempskipa\pagegoal \ttl@addstretch % \relax if nobottomtitle* \advance\@tempskipa-\bottomtitlespace\relax % not a register \pagegoal\@tempskipa \def\@textbottom{\vskip\z@\@plus.0001fil}% \penalty9999 \pagegoal\@tempskipb \endgroup \fi \fi \@afterindenttrue \ifcase#5 \@afterindentfalse\fi \ttl@assign\@tempskipb#4\relax\aftertitleunit \ttl@select{#6}{#1}{#2}{#7}% \ttl@finmarks \@ifundefined{ttlp@#6}{}{\ttlp@write{#6}}% \if@noskipsec \global\@nobreakfalse \everypar{% \if@noskipsec \global\@noskipsecfalse \clubpenalty\@M \hskip-\parindent \begingroup \@svsechd\unskip{\hspace{\@tempskipb}}% \endgroup \else \clubpenalty\@clubpenalty\everypar{}% \fi}% \else \par\nobreak \vspace{\@tempskipb}% \@afterheading \fi \ignorespaces} % Part class % ~~~~~~~~~~ \providecommand\partmark[1]{\markboth{}{}} \def\ttl@part@i#1[#2]#3{% \gdef\ttl@savemark{\csname#1mark\endcsname{#3}}% \ifx\ttl@notocparts\@undefined \def\ttl@savewrite{\ttl@write{#1}{#3}}% Not #2! \else \let\ttl@savewrite\@empty \fi \def\ttl@savetitle{#3}% \ttl@labelling{#1}{#2}% \ttl@startargs\ttl@part@ii{#1}{#3}} \def\ttl@part@ii#1#2#3#4#5#6#7{% \ttl@assign\@tempskipa#3\relax\beforetitleunit \vspace*{\@tempskipa}% \@ifundefined{ttl@ps@#6}{}% {\PackageWarning{titlesec}{Page style in part class ignored}}% \global\@afterindenttrue \ifcase#5 \global\@afterindentfalse \fi \ttl@assign\@tempskipb#4\relax\aftertitleunit \ttl@select{#6}{#1}{#2}{#7}% \ttl@finmarks \@ifundefined{ttlp@#6}{}{\ttlp@write{#6}}% \par\nobreak \vspace{\@tempskipb}% \@afterheading} % Page class % ~~~~~~~~~~ \def\ttl@page@i#1[#2]#3{% \gdef\ttl@savemark{\csname#1mark\endcsname{#3}}% \ifx\ttl@notocparts\@undefined \def\ttl@savewrite{\ttl@write{#1}{#3}}% Not #2! \else \let\ttl@savewrite\@empty \fi \def\ttl@savetitle{#3}% \ttl@labelling{#1}{#2}% \ttl@startargs\ttl@page@ii{#1}{#3}} \def\ttl@page@ii#1#2#3#4#5#6#7{% \ttl@assign\@tempskipa#3\relax\beforetitleunit \if@openright \cleardoublepage \else \clearpage \fi \@ifundefined{ttl@ps@#6}% {\thispagestyle{plain}}% {\thispagestyle{\@nameuse{ttl@ps@#6}}}% \if@twocolumn \onecolumn \@tempswatrue \else \@tempswafalse \fi \vspace*{\@tempskipa}% \@afterindenttrue \ifcase#5 \@afterindentfalse\fi \ttl@assign\@tempskipb#4\relax\aftertitleunit \ttl@select{#6}{#1}{#2}{#7}% \ttl@finmarks \@ifundefined{ttlp@#6}{}{\ttlp@write{#6}}% \vspace{\@tempskipb}% \newpage \if@twoside \if@openright \null \@ifundefined{ttl@ps@#6}% {\thispagestyle{empty}}% {\thispagestyle{\@nameuse{ttl@ps@#6}}}% \newpage \fi \fi \if@tempswa \twocolumn \fi \ignorespaces} % Top class and some makechapterhead stuff % ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ % % \ttl@mkchap is the new make(s)chapterhead. \def\ttl@mkchap#1#2#3#4#5#6#7{% \gdef\ttl@savemark{\csname#6mark\endcsname{#7}}% \let\ttl@savewrite\@empty \let\ttl@Hy@saveanchor\@empty \@ifundefined{ttl@ps@#6}{}% {\thispagestyle{\@nameuse{ttl@ps@#6}}}% \let\ifttl@toclabel\ifttl@label \ttl@mkchap@i{#1}{#2}{#3}{#4}{#5}{#6}{#7}} % But \ttl@mkchap@i is used by both makechapterhead and % the top class. \def\ttl@mkchap@i#1#2#3#4#5#6#7{% \ttl@assign\@tempskipa#3\relax\beforetitleunit \vspace*{\@tempskipa}% \global\@afterindenttrue \ifcase#5 \global\@afterindentfalse\fi \ttl@assign\@tempskipb#4\relax\aftertitleunit \ttl@topmode{\@tempskipb}{% \ttl@select{#6}{#1}{#2}{#7}}% \ttl@finmarks % Outside the box! \@ifundefined{ttlp@#6}{}{\ttlp@write{#6}}} \def\ttl@top@i#1[#2]#3{% \gdef\ttl@savemark{\csname#1mark\endcsname{#3}}% \let\ttl@savewrite\@empty \def\ttl@savetitle{#3}% \ttl@labelling{#1}{#2}% \ttl@startargs\ttl@top@ii{#1}{#3}} \def\ttl@top@ii#1#2#3#4#5#6#7{% \@ifundefined{#6break}% {\if@openright \cleardoublepage \else \clearpage \fi}% {\csname#6break\endcsname}% \@ifundefined{ttl@ps@#6}% {\thispagestyle{plain}}% {\thispagestyle{\@nameuse{ttl@ps@#6}}}% \global\@topnum\z@ \@ifundefined{#6tolists}% {\addtocontents{lof}{\protect\ttl@tocsep}% \addtocontents{lot}{\protect\ttl@tocsep}} {\@nameuse{#6tolists}}% \if@twocolumn \@topnewpage[\ttl@mkchap@i{#1}{#2}{#3}{#4}{#5}{#6}{#7}]% \else \ttl@mkchap@i{#1}{#2}{#3}{#4}{#5}{#6}{#7}% \@afterheading \fi \ignorespaces} % \def\ttl@noskipsectrue{% % \if@noskipsec % \PackageError{titlesec}{Invalid shape for top class}% % {The selected shape only makes sense when merged into\MessageBreak % a paragraph. That is impossible in the top class}% % \else \newcommand\chaptertitlename{\@chapapp} \def\ttl@tocsep{\addvspace{10\p@}} % +-----------------+ % | S H A P E S | % +-----------------+ % % Reformatting Titles: Interface % ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ % The surrounding space is stored in a macro % named \ttls@
whose content is % {left}{right}{before}{after}{afterindent}. % But if there is the page key, the name is % \ttls@
/ \newcommand\titlespacing{% \@ifstar{\ttl@spacing@i{\z@}}{\ttl@spacing@i{\@ne}}} \def\ttl@spacing@i#1#2#3#4#5{% \ttl@getkeys{#2}{titlesec}% \@ifnextchar[{% \ttl@spacing@ii{#1}{#3}{#4}{#5}% }{% \ttl@spacing@ii{#1}{#3}{#4}{#5}[\z@]}} \def\ttl@spacing@ii#1#2#3#4[#5]{% \expandafter\def\csname ttls@\ttl@a\endcsname {{#2}{#5}{#3}{#4}{#1}}} % The section name is built in \ttl@a. % The format is stored in a macro named \ttlf@
, % or \ttlf@
/ if there is the page spec, % or \ttlf@.../* if numberless is true % whose content is % \ttl@{format}{label}{sep}{before}{after} \newtoks\ttl@toksa \newcommand\titleformat{% \@ifstar{\ttl@format@s}% {\ttl@format@i}} \def\ttl@format@s#1#2{% \edef\ttl@a{\expandafter\@gobble\string#1}% \@ifundefined{ttlf@\ttl@a}% {\PackageError{titlesec}{Not allowed in `easy' settings} {The sectiong command you are trying to redefine\MessageBreak is not handled by the starred variant (eg, \string\part)}}{} \expandafter\expandafter\expandafter \ttl@format@si\csname ttlf@\ttl@a \endcsname {#2}} \def\ttl@format@si#1#2#3#4#5#6#7{% \@namedef{ttlf@\ttl@a}{#1{#7}{#3}{#4}{#5}{#6}}} \def\ttl@format@i#1{% \@ifnextchar[{\ttl@format@ii{#1}}{\ttl@format@ii{#1}[hang]}} \def\ttl@format@ii#1[#2]#3#4#5#6{% \ttl@getkeys{#1}{titlesec}% \ttl@toksa{{#3}{#4}{#5}{#6}}% Save arguments \@ifnextchar[{% \ttl@format@iii{#2}% }{% \ttl@format@iii{#2}[]}} % First, we get the shape -- if not defined it loads % the corresponding file. \def\ttl@format@iii#1[#2]{% \@ifundefined{ttlh@#1}{% \begingroup \makeatletter \InputIfFileExists{#1.tss}{}{% \@ifundefined{ttlhx@#1}% {\PackageError{titlesec}{Unknown shape}% {Shapes are defined in files with extension tss\MessageBreak Either you have misspelled the shape\MessageBreak or there is no a #1.tss file}}% {\global\expandafter \let\csname ttlh@#1\expandafter\endcsname \csname ttlhx@#1\endcsname}}% \endgroup}{}% \@temptokena{#2}% \ifttl@explicit \edef\ttl@b{% \def\expandafter\noexpand\csname ttlf@\ttl@a\endcsname####1% {\expandafter\noexpand\csname ttlh@#1\endcsname \the\ttl@toksa{\the\@temptokena}}}% \else \edef\ttl@b{% \def\expandafter\noexpand\csname ttlf@\ttl@a\endcsname {\expandafter\noexpand\csname ttlh@#1\endcsname \the\ttl@toksa{\the\@temptokena}}}% \fi \ttl@b \csname ttl@compat\ttl@a\endcsname} % Styles % ~~~~~~ % 1:global 2:label 3:sep 4:style 5:after 6:left 7:right 8:title % \ttl@ and \ttlh@ take the following eight % arguments: % {format}{label}{sep}{before}{after}{left}{right}{title} % where before and after refer to the format. % With the option explicit, #4 contains the title and #8 is % empty. \def\ttl@strut{\strut} \def\ttlh@display#1#2#3#4#5#6#7#8{% \gdef\ttl@makeline##1{\ttl@calc\hspace{#6}##1\ttl@calc\hspace{#7}}% \setlength\leftskip{#6}% \setlength\rightskip{#7}% \interlinepenalty\@M \ttl@changecentercr \ttl@defnostruts \ttl@beginlongest #1\ifhmode\ttl@hmode@error\fi \ttl@glcmds \parindent\z@ \ifttl@label {#2\ttl@strut\@@par}\nobreak\ttl@calc\vspace{#3}% \fi #4{#8}% \kern\z@\ttl@strut\@@par \nobreak\ttl@midlongest#5\@@par \ttl@endlongest} \def\ttlh@hang#1#2#3#4#5#6#7#8{% \gdef\ttl@makeline##1{\ttl@calc\hspace{#6}##1\ttl@calc\hspace{#7}}% \setlength\leftskip{#6}% \setlength\rightskip{#7}% \interlinepenalty\@M \ttl@changecentercr \ttl@defnostruts \ttl@beginlongest #1{\ifhmode\ttl@hmode@error\fi \ttl@glcmds \parindent\z@ \begingroup \ifttl@label \noindent \sbox\z@{#2\ttl@strut\ttl@calc\hspace{#3}}% \hangindent\wd\z@ \box\z@ \fi #4{#8}% \kern\z@\ttl@strut\@@par \endgroup \nobreak\ttl@midlongest#5\@@par}% \ttl@endlongest} \def\ttlh@runin#1#2#3#4#5#6#7#8{% \global\@noskipsectrue \gdef\ttl@makeline##1{##1}% \ttl@changecentercr \ttl@defnostruts #1{\ifhmode\ttl@hmode@error\fi \global\sbox\ttl@box{% \ttl@calc\hspace{#6}% \ifttl@label{\ttl@strut#2}\ttl@calc\hspace{#3}\fi #4{#8}#5\unskip}}% \gdef\@svsechd{\unhbox\ttl@box}} % ---------- \gdef\ttlhx@block#1#2#3#4#5#6#7#8{% \gdef\ttl@makeline##1{\ttl@calc\hspace{#6}##1\ttl@calc\hspace{#7}}% \setlength\leftskip{#6}% \setlength\rightskip{#7}% \interlinepenalty\@M \ttl@changecentercr \ttl@defnostruts \ttl@beginlongest #1% \ifhmode\ttl@hmode@error\fi \ttl@glcmds \parindent\z@ \leavevmode \ifttl@label {#2}% \setlength\@tempskipa{#3}% \ifdim\@tempskipa=\z@\else\ttl@calc\hspace{#3}\fi \fi #4{#8}% \kern\z@\ttl@strut\@@par \nobreak\ttl@midlongest#5\@@par \ttl@endlongest} \gdef\ttlhx@frame#1#2#3#4#5#6#7#8{% \def\ttl@filleft##1{\hfill}% \def\ttl@filright##1{\hfill}% \gdef\ttl@makeline##1{% \ttl@calc\hspace{#6}##1\ttl@calc\hspace{#7}}% \interlinepenalty\@M \ttl@changecentercr \ttl@defnostruts #1\ifhmode\ttl@hmode@error\fi \parindent\z@ \leavevmode \@tempdima\fboxrule \addtolength\@tempdima{#3}% \setlength\leftskip{#6}% \setlength\rightskip{#7}% \lower\@tempdima\hbox{% \everypar{}% \setbox\z@\hbox{#2}% \addtolength\hsize{-#6}% \addtolength\hsize{-#7}% \@tempdima\dp\z@ % 2002/3/22 \advance\@tempdima.5\ht\z@ \vbox{% \hbox to \hsize{% \leaders\hrule\@height\fboxrule\ttl@filleft{#3}% \ifttl@label\lower.5\ht\z@\box\z@\fi \leaders\hrule\@height\fboxrule\ttl@filright{#3}}% \vskip-\lineskip \ifttl@label\vskip-\@tempdima\fi \hbox{% \vrule\@width\fboxrule \kern-\fboxrule \vbox{% \ttl@calc\vspace{#3}% \leavevmode \addtolength\leftskip {#3}\addtolength\leftskip{-#6}% \addtolength\rightskip{#3}\addtolength\rightskip{-#7}% \ttl@strut#4{#8}\kern\z@\ttl@strut\@@par \ttl@calc\vspace{#3}}% \kern-\fboxrule \vrule\@width\fboxrule}% \hrule\@height\fboxrule}}% \@@par\nobreak#5\@@par} \gdef\ttlhx@leftmargin#1#2#3#4#5#6#7#8{% \global\@noskipsectrue \addtolength\@tempskipb{#6}% \xdef\ttl@makeline##1{\hskip-\the\@tempskipb\relax##1}% \leftskip\z@skip \rightskip\z@skip \ttl@changecentercr \ttl@defnostruts #1\ifhmode\ttl@hmode@error\fi \parindent\z@ \global\setbox\ttl@box\vtop{% \setlength\hsize{#6}% \linewidth\hsize \everypar{}% \color@begingroup \ifttl@label{\ttl@strut#2\ttl@strut}\ttl@calc\hspace{#3}\fi \ttl@strut#4{#8}\kern\z@\ttl@strut\@@par \color@endgroup}% \advance\@tempskipa\ht\ttl@box \advance\@tempskipa\dp\ttl@box \advance\@tempskipa-\pagegoal \advance\@tempskipa\pagestretch \@tempskipb\pagegoal \pagegoal-\@tempskipa \ifdim\bottomtitlespace<\z@\else \def\@textbottom{\vskip\z@\@plus.0001fil}% \fi \penalty9999 \pagegoal\@tempskipb \dp\ttl@box=\z@ \gdef\@svsechd##1##2{% \llap{\box\ttl@box##2}% \if@afterindent\hskip\parindent\fi #5}} \let\ttlhx@margin\ttlhx@leftmargin \gdef\ttlhx@rightmargin#1#2#3#4#5#6#7#8{% \global\@noskipsectrue \addtolength\@tempskipb{#6}% \xdef\ttl@makeline##1{##1\hskip-\the\@tempskipb}% \leftskip\z@skip \rightskip\z@skip \ttl@changecentercr \ttl@defnostruts #1\ifhmode\ttl@hmode@error\fi \parindent\z@ \global\setbox\ttl@box\vtop{% \setlength\hsize{#6}% \linewidth\hsize \everypar{}% \color@begingroup \ifttl@label{\ttl@strut#2\ttl@strut}\ttl@calc\hspace{#3}\fi \ttl@strut#4{#8}\kern\z@\ttl@strut\@@par \color@endgroup}% \advance\@tempskipa\ht\ttl@box \advance\@tempskipa\dp\ttl@box \advance\@tempskipa-\pagegoal \advance\@tempskipa\pagestretch \@tempskipb\pagegoal \pagegoal-\@tempskipa \ifdim\bottomtitlespace<\z@\else \def\@textbottom{\vskip\z@\@plus.0001fil}% \fi \penalty9999 \pagegoal\@tempskipb \dp\ttl@box=\z@ \gdef\@svsechd##1##2{% \rlap{\hskip\textwidth##2\box\ttl@box}% \if@afterindent\hskip\parindent\fi}} \gdef\ttlhx@wrap#1#2#3#4#5#6#7#8{% \global\@noskipsectrue \gdef\ttl@makeline##1{##1}% \ttl@changecentercr \ttl@defnostruts \begingroup #1\ifhmode\ttl@hmode@error\fi \titlewidth\z@ \def\\{\@ifstar{\@ifnextchar[{\ttl@bs}{\newline}}% {\@ifnextchar[{\ttl@bs}{\newline}}}% \def\ttl@bs[##1]{\newline}% \let\@centercr\\% \advance\rightskip 1\leftskip plus 1fil \leftskip=\z@ \parindent\z@ \let\iftitlemeasuring\@firstoftwo \global\setbox\ttl@box\vtop{\setlength\hsize{#6}% \color@begingroup \ifttl@label{#2}\ttl@calc\hspace{#3}\fi #4{#8}\kern\z@\ttl@strut \@@par \color@endgroup}% \let\iftitlemeasuring\@secondoftwo \ttl@boxprocess \global\titlewidth\titlewidth \global\titlewidthfirst\titlewidthfirst \global\titlewidthlast\titlewidthlast \endgroup \edef\ttl@maxdimen{\the\titlewidth}% #1\ifhmode\ttl@hmode@error\fi \global\setbox\ttl@box\vtop{\setlength\hsize{\ttl@maxdimen}% \color@begingroup \ifttl@label{#2}\ttl@calc\hspace{#3}\fi#4{#8}\kern\z@\ttl@strut \@@par \color@endgroup}% \advance\@tempskipa1.5\baselineskip \advance\@tempskipa\ht\ttl@box \advance\@tempskipa\dp\ttl@box \advance\@tempskipa-\pagegoal \advance\@tempskipa\pagestretch \@tempskipb\pagegoal \pagegoal-\@tempskipa \ifdim\bottomtitlespace<\z@\else \def\@textbottom{\vskip\z@\@plus.0001fil}% \fi \penalty9999 \pagegoal\@tempskipb \@tempdima\ht\ttl@box \advance\@tempdima\dp\ttl@box \@tempdimb\@tempdima \divide\@tempdima\baselineskip \count@\@tempdima \advance\count@ \ifdim\@tempdimb<\the\count@.5\baselineskip\@ne\else\tw@\fi \dp\ttl@box=\z@ \xdef\@svsechd##1##2{% \noexpand\llap{\box\ttl@box##2}% \setbox\z@\hbox{\hskip\ttl@maxdimen\relax##2}% \global\hangindent\wd\z@ \global\hangafter-\the\count@\relax}} \gdef\ttlhx@drop#1#2#3#4#5#6#7#8{% \global\@noskipsectrue \gdef\ttl@makeline##1{##1}% \ttl@changecentercr \ttl@defnostruts #1\ifhmode\ttl@hmode@error\fi \parindent\z@ \global\setbox\ttl@box\vtop{\setlength\hsize{#6}% \color@begingroup \ifttl@label{#2}\ttl@calc\hspace{#3}\fi #4{#8}\kern\z@\ttl@strut \@@par \color@endgroup}% \advance\@tempskipa1.5\baselineskip \advance\@tempskipa\ht\ttl@box \advance\@tempskipa\dp\ttl@box \advance\@tempskipa-\pagegoal \advance\@tempskipa\pagestretch \@tempskipb\pagegoal \pagegoal-\@tempskipa \ifdim\bottomtitlespace<\z@\else \def\@textbottom{\vskip\z@\@plus.0001fil}% \fi \penalty9999 \pagegoal\@tempskipb \@tempdima\ht\ttl@box \advance\@tempdima\dp\ttl@box \@tempdimb\@tempdima \divide\@tempdima\baselineskip \count@\@tempdima \advance\count@ \ifdim\@tempdimb<\the\count@.5\baselineskip\@ne\else\tw@\fi \dp\ttl@box=\z@ \xdef\@svsechd##1##2{% \noexpand\llap{\box\ttl@box##2}% \setbox\z@\hbox{\noexpand\ttl@calc\noexpand\hspace{#6}\relax##2}% \global\hangindent\wd\z@ \global\hangafter-\the\count@\relax}} % +-----------------+ % | T O O L S | % +-----------------+ % % calcwidth % ~~~~~~~~~ % Implemented after code from soul (but much modified...) \newdimen\titlewidth \newdimen\titlewidthlast \newdimen\titlewidthfirst \let\ttl@glcmds\relax \let\ttl@beginlongest\@empty \let\ttl@midlongest\@empty \let\ttl@endlongest\@empty \let\iftitlemeasuring\@secondoftwo \def\ttl@xbeginlongest#1\ttl@endlongest{% \titlewidth\z@ \titlewidthlast\z@ \let\iftitlemeasuring\@firstoftwo \setbox\ttl@box\vbox{% \def\ttl@glcmds{% \def\\{\@ifstar{\@ifnextchar[{\ttl@bs}{\newline}}% {\@ifnextchar[{\ttl@bs}{\newline}}}% \def\ttl@bs[####1]{\newline}% \let\@centercr\\% \def\ttl@midlongest####1\@@par{}% Very dirty... \advance\rightskip 1\leftskip plus 1fil \leftskip=\z@}% #1}% \let\iftitlemeasuring\@secondoftwo \ttl@boxprocess #1} \def\ttl@boxprocess{% \setbox\ttl@box=\vbox{% \unvcopy\ttl@box \unskip\unpenalty \global\setbox\@ne=\lastbox}% \ifvoid\@ne \else \setbox\tw@=\hbox{\hskip-\leftskip\unhbox\@ne\hskip-\rightskip}% \titlewidthfirst\wd\tw@ \ifdim\titlewidth<\titlewidthfirst \titlewidth\titlewidthfirst \fi \ifdim\titlewidthlast=\z@ \titlewidthlast\titlewidthfirst \fi \expandafter\ttl@boxprocess \fi} % Rules % ~~~~~ \providecommand\titleline{% \@ifstar{\ttl@line@i{\hb@xt@\titlewidth}}% {\ttl@line@i{}}} \def\ttl@line@i#1{% \@ifnextchar[{\ttl@line{#1}}{\ttl@line{#1}[s]}} \def\ttl@line#1[#2]#3{% \vskip\topskip \hrule \@height \z@ \nobreak \vskip-\topskip \begingroup \parindent\z@ \everypar{}% \leftskip\z@ \rightskip\z@ % #1 is either \hb@xt@\titlewidth or empty: \@makebox[\hsize][#2]{\ttl@makeline{#1{#3}}}% \par \endgroup \hrule height \z@ \nobreak} \providecommand\titlerule{\@ifstar{\ttl@row}{\ttl@rule}} \let\ttl@leaders\xleaders % For titletoc compatibility \def\ttl@row{\@ifnextchar[{\ttl@row@i}{\ttl@row@i[\wd\z@]}} \def\ttl@row@i[#1]#2{% \ifvmode\expandafter\titleline\fi {\sbox\z@{#2}% \ttl@calcneg\hspace{#1}% \hskip\wd\z@ \ttl@leaders\hb@xt@#1{\hss\box\z@}% \hfill\kern\z@}} \def\ttl@rule{\@ifnextchar[{\ttl@rule@i}{\ttl@rule@i[.4\p@]}} \def\ttl@rule@i[#1]{% \ifvmode\expandafter\titleline\fi {\leaders\hrule height #1\hfill\kern\z@}} % Par shapes and space % ~~~~~~~~~~~~~~~~~~~~ \providecommand\filright{% \gdef\ttl@filleft##1{\hskip##1}% \gdef\ttl@filright##1{\hfill}% \let\\\@centercr \advance\rightskip\z@ \@plus 1fil\relax} \providecommand\filleft{% \gdef\ttl@filleft##1{\hfill}% \gdef\ttl@filright##1{\hskip##1}% \let\\\@centercr \advance\leftskip\z@ \@plus 1fil \parfillskip\z@} \providecommand\filcenter{\filleft\filright \gdef\ttl@filleft##1{\hfill}} \providecommand\fillast{% \gdef\ttl@filleft##1{\hfill}% \gdef\ttl@filright##1{\hfill}% \let\\\@centercr \filleft\advance\rightskip\z@ \@plus -1fil \parfillskip\z@ \@plus 2fil\relax} \newcommand\filinner{% \if@twoside \ifodd\count\z@\filleft\else\filright\fi \else \filleft \fi} \newcommand\filouter{% \if@twoside \ifodd\count\z@\filright\else\filleft\fi \else \filright \fi} \newcommand\wordsep{\fontdimen\tw@\font \@plus \fontdimen\thr@@\font \@minus \fontdimen4\font} % Struts. % ~~~~~~ % A way to remove the struts added by styles. May be redefined below % with option nostruts. \def\ttl@defnostruts{\def\nostruts{\let\ttl@strut\@empty}} % +-----------------+ % | O P T I O N S | % +-----------------+ \DeclareOption{pagestyles}{\let\sectiontitle\@empty} \DeclareOption{extramarks}{\let\ttl@fetchmark\@empty} \DeclareOption{floatps}{% \ifx\sectiontitle\@empty \let\ttl@replace\space \else \PackageWarning{titlesec}{Ignoring `floatps' without `pagestyles'. This option is now deprecated.}% \fi} \DeclareOption{psfloats}{% \ifx\sectiontitle\@empty \let\ttl@replace\@empty \else \PackageWarning{titlesec}{Ignoring `psfloats' without `pagestyles'}% \fi} \DeclareOption{loadonly}{\let\ttl@extract\@empty} \DeclareOption{outermarks}{% \def\ttl@titlemarks{\outertitlemarks}} \DeclareOption{topmarks}{ \def\ttl@titlemarks{\toptitlemarks}} \DeclareOption{botmarks}{% \def\ttl@titlemarks{\bottitlemarks}} \DeclareOption{innermarks}{% \def\ttl@titlemarks{\innertitlemarks}} \DeclareOption{footmarks}{} % Backward compat \DeclareOption{explicit}{\ttl@explicittrue} \DeclareOption{clearempty}{% \def\cleardoublepage{% \clearpage{\ps@empty\if@twoside\ifodd\c@page\else \hbox{}\newpage\if@twocolumn\hbox{}\newpage\fi\fi\fi}}} \DeclareOption{rigidchapters}{% \def\ttl@topmode#1#2{\vbox to #1{#2\vfil}}% \def\ttl@chapafter{.26\textheight}} \DeclareOption{rubberchapters}{% \def\ttl@topmode#1#2{{#2}\ttl@calc\vspace{#1}}% \def\ttl@chapafter{40\p@}} \DeclareOption{bottomtitles}{% \def\bottomtitlespace{-1\p@}} \DeclareOption{nobottomtitles}{% \def\bottomtitlespace{.2\textheight}} \DeclareOption{nobottomtitles*}{% \let\ttl@addstretch\relax \def\bottomtitlespace{.2\textheight}} \DeclareOption{calcwidth}{% \let\ttl@beginlongest\ttl@xbeginlongest} \DeclareOption{aftersep}{% \let\ttl@titlespace\@gobble} \DeclareOption{largestsep}{% \let\ttl@titlespace\addvspace} \DeclareOption{oldparttoc}{% \def\ttl@tocpart{\def\ttl@a{\thepart\hspace{1em}}}} \DeclareOption{newparttoc}{% \let\ttl@tocpart\relax} \DeclareOption{notocpart*}{% \let\ttl@notocparts\@empty} \DeclareOption{rm}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\rmfamily}} \DeclareOption{sf}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\sffamily}} \DeclareOption{tt}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\ttfamily}} \DeclareOption{md}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\mdseries}} \DeclareOption{bf}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\bfseries}} \DeclareOption{up}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\upshape}} \DeclareOption{it}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\itshape}} \DeclareOption{sl}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\slshape}} \DeclareOption{sc}{% \protected@xdef\ttl@fonts{\ttl@fonts\protect\scshape}} \DeclareOption{big}{% \gdef\ttl@sizes#1{\ifcase#1\relax\Huge\or\Large\or\large \or\normalsize\or\or\or\huge\fi}} \DeclareOption{medium}{% \gdef\ttl@sizes#1{\ifcase#1\relax\huge\or\Large\or\large \or\normalsize\or\or\or\LARGE\fi}} \DeclareOption{small}{% \gdef\ttl@sizes#1{\ifcase#1\relax\LARGE\or\large \or\normalsize\or\normalsize\or\or\or\Large\fi}} \DeclareOption{tiny}{% \gdef\ttl@sizes#1{\ifcase#1\relax\large\or\normalsize\or \normalsize\or\normalsize\or\or\or\normalsize\fi}} \DeclareOption{raggedleft}{% \gdef\ttl@fil{\filleft}} \DeclareOption{center}{% \gdef\ttl@fil{\filcenter}} \DeclareOption{raggedright}{% \gdef\ttl@fil{\filright}} \DeclareOption{uppercase}{% \gdef\ttl@case{\MakeUppercase}} \DeclareOption{compact}{% \gdef\ttl@space{1}% \gdef\ttl@chapafter{30\p@}} % Deprecated. To be remmoved in a major upgrade (3.0) \DeclareOption{indentfirst}{% \gdef\@afterindentfalse{\let\if@afterindent\iftrue}% \@afterindenttrue \def\titlespacing{% \@ifstar{\ttl@spacing@i{\@ne}}{\ttl@spacing@i{\@ne}}}} \DeclareOption{nonindentfirst}{% \def\titlespacing{% \@ifstar{\ttl@spacing@i{\z@}}{\ttl@spacing@i{\z@}}}} % New names \DeclareOption{indentafter}{% \gdef\@afterindentfalse{\let\if@afterindent\iftrue}% \@afterindenttrue \def\titlespacing{% \@ifstar{\ttl@spacing@i{\@ne}}{\ttl@spacing@i{\@ne}}}} \DeclareOption{noindentafter}{% \def\titlespacing{% \@ifstar{\ttl@spacing@i{\z@}}{\ttl@spacing@i{\z@}}}} % newlinetospace \let\ttl@blinemarks\relax \let\ttl@elinemarks\relax \DeclareRobustCommand\ttl@linetosp{% \@ifstar{\ttl@linetosp@i}{\ttl@linetosp@i}}% \def\ttl@linetosp@i{% \ifdim\lastskip>\z@\else\space\fi \ignorespaces} \DeclareOption{newlinetospace}{% \def\ttl@blinemarks{% \let\ttl@e\\% \def\\{\ttl@linetosp}}% \def\ttl@elinemarks{\let\\\ttl@e}}% % toctitles \def\ttl@addcontentsline#1#2{% \addcontentsline{toc}{#1}{\ifttl@toclabel\ttl@a\fi#2}% \nobreak} \DeclareOption{toctitles}{% \def\ttl@addcontentsline#1#2{% \addcontentsline{toc}{#1}{\ifttl@toclabel\ttl@a\fi\ttl@savetitle}% \nobreak}} % pageatnewline \def\ttl@changecentercr{% \let\ttl@centercr\@centercr \def\@centercr{\@ifstar{\ttl@centercr*}{\ttl@centercr*}}} \DeclareOption{pageatnewline}{\let\ttl@changecentercr\relax} \def\ttl@fonts{} % nostruts \DeclareOption{nostruts}{% \let\ttl@strut\@empty \def\ttl@defnostruts{% \let\ttl@strut\@empty \def\nostruts{\let\ttl@strut\@empty}}} \ExecuteOptions{rubberchapters,bottomtitles,aftersep,oldparttoc,% innermarks} \ProcessOptions % +-----------------+ % | H Y P E R R E F | % +-----------------+ % % These two commands are provided by hyperref. But if they % are not defined at \begin{document} hyperref has not been % loaded or it is an old version. \AtBeginDocument{% \ifx\ttl@Hy@steplink\@undefined \let\ttl@Hy@steplink\@gobble \let\ttl@Hy@refstepcounter\refstepcounter \fi} % +-----------------+ % | PAGE STYLES | % +-----------------+ % % This is generic: \newcommand\assignpagestyle[2]{% \@namedef{ttl@ps@\expandafter\@gobble\string#1}{#2}} % Old pagestyles % ~~~~~~~~~~~~~~ \providecommand\newpagestyle{% \let\ttl@compatps\@empty % marks the ``old interface'' \let\ttl@coreps\@empty \makeatletter \edef\ttl@d{% \noexpand\input{titleps.sty}% \catcode`\noexpand\@=\the\catcode`\@}% \ttl@d \newpagestyle} \providecommand\renewpagestyle{% \let\ttl@compatps\@empty % marks the ``old interface'' \let\ttl@coreps\@empty \makeatletter \edef\ttl@d{% \noexpand\input{titleps.sty}% \catcode`\noexpand\@=\the\catcode`\@}% \ttl@d \renewpagestyle} \providecommand\widenhead{% \let\ttl@compatps\@empty % marks the ``old interface'' \let\ttl@coreps\@empty \makeatletter \edef\ttl@d{% \noexpand\input{titleps.sty}% \catcode`\noexpand\@=\the\catcode`\@}% \ttl@d \widenhead} % New pagestyles % ~~~~~~~~~~~~~~ \@ifundefined{sectiontitle}{}{% \let\ttl@coreps\@empty \input{titleps.sty}} % +-----------------+ % | K E Y S | % +-----------------+ \def\ttl@keys{% \let\ttl@keys\relax \@ifundefined{define@key}{\RequirePackage{keyval}}{}% \def\ttl@getkeys##1##2{% \let\ttl@a\@empty \if\expandafter @\@gobble##1@\@empty % if there is a single token \edef\ttl@b{\expandafter\@gobble\string##1}% \let\ttl@a\ttl@b \else \ttl@labelfalse % A temporary flag: true if there is page key \setkeys{##2}{##1}% \ifttl@label \@ifundefined{ttlp@\ttl@b}{% \expandafter\let\csname ttlp@\ttl@b\endcsname\@empty}{}% \fi \edef\ttl@a{\ttl@b\ttl@a}% \fi}% % \define@key{titlesec}{name}{% \edef\ttl@b{\expandafter\@gobble\string##1}}% \define@key{titlesec}{numberless}[true]{% \csname @tempswa##1\endcsname \if@tempswa \edef\ttl@a{\ttl@a/*}% \fi}% \define@key{titlesec}{page}{% \ttl@labeltrue % Used as flag \edef\ttl@a{/##1\ttl@a}}% % \def\ttl@extra@numberless{\ttl@labeltrue}% The actual meaning \let\ttl@key@numberless\@empty %\let\ttl@key@matter\@empty \let\ttlp@append\@gobbletwo \def\ttl@setkeys##1{% \def\ttl@trylist{\ttl@try{}}% \@for\ttl@b:=##1\do{% \begingroup \let\ttl@a\relax \def\ttl@try####1{% \noexpand\ttl@try{####1\ttl@a{\ttl@b}}% \noexpand\ttl@try{####1}}% \xdef\ttl@trylist{\ttl@trylist}% \endgroup}}% \ttl@setkeys{page,numberless}% matter % \if@twoside \newcounter{ttlp@side}% \newcount\ttlp@side \def\ttlp@theside{\ifodd\c@page o\else e\fi}% \def\ttlp@append##1##2{% {\let\@elt\relax \expandafter\xdef\csname ttlp@##1\endcsname{% \csname ttlp@##1\endcsname\@elt ##2}}}% \def\ttlp@write##1{% {\let\ttlp@theside\relax \protected@write\@auxout{}% {\string\ttlp@append{##1}{\ttlp@theside}}}}% \def\ttlp@fetch##1{% \stepcounter{ttlp@side}% \global\advance\ttlp@side\@ne \@whilenum\ttlp@side<\c@ttlp@side\do{% \expandafter\@next\expandafter\@tempa\csname ttlp@##1\endcsname{}{}% \global\advance\ttlp@side\@ne}% \expandafter\@next\expandafter\ttl@b\csname ttlp@##1\endcsname{% \xdef\ttl@key@page{/\if\ttl@b oodd\else even\fi}% }{% \xdef\ttl@key@page{/\ifodd\c@page odd\else even\fi}% \@@warning{Unsynchronized `##1' title on page \thepage}}}% \else \let\ttlp@write\@gobble \def\ttlp@fetch##1{\gdef\ttl@key@page{/odd}}% \fi} % +-----------------+ % | C O M P A T | % +-----------------+ % Easy setup, i.e., that of package options, is % taken care of, if necessary. \renewcommand\secdef[2]{% \@ifstar {\ttl@labelfalse #2} {\ttl@labeltrue \ifx#1\@chapter \if@mainmatter\else\ttl@labelfalse\fi \ifnum\ttll@chapter>\c@secnumdepth\ttl@labelfalse\fi \else\ifx#1\@part \ifnum\ttll@part>\c@secnumdepth\ttl@labelfalse\fi \fi\fi \let\ifttl@toclabel\ifttl@label \@dblarg{#1}}} \@ifundefined{ttl@extract}{}{\endinput} \newcommand\titlelabel[1]{% \def\@seccntformat##1{#1}} \expandafter\ifx\csname chapter\endcsname\relax \def\ttl@compatpart{\titleclass{\part}{part}\relax} \else \def\ttl@compatchapter{% \def\@makechapterhead{% \ttl@labeltrue \if@mainmatter\else\ttl@labelfalse\fi \ifnum\ttll@chapter>\c@secnumdepth\ttl@labelfalse\fi \ttl@startargs\ttl@mkchap{chapter}}% \def\@makeschapterhead{% \ttl@labelfalse \if@mainmatter\else\ttl@labelfalse\fi \ifnum\ttll@chapter>\c@secnumdepth\ttl@labelfalse\fi \ttl@startargs\ttl@mkchap{chapter}}} \def\ttl@compatpart{\titleclass{\part}{page}\relax} \fi \def\ttl@@extract#1\@startsection#2#3#4#5#6#7#8{% \@tempskipa=#5 \@tempskipb=#6 \ifdim\@tempskipa<\z@ \toks@{\titlespacing*#8{#4}}% \@tempskipa-\@tempskipa \else \toks@{\titlespacing#8{#4}}% \fi \@ifundefined{ttl@space}{}{% \ttl@assign\@tempskipa*\ttl@space\relax\beforetitleunit}% \ifdim\@tempskipb<\z@ \if@tempswa \titleformat#8[runin]% {\ttl@fonts\ttl@sizes{#3}}{\@seccntformat{#2}}% {\z@}\ttl@passexplicit \else \titleformat#8[runin]% {#7}{\@seccntformat{#2}}% {\z@}\ttl@passexplicit \fi \@tempskipb-\@tempskipb \else \if@tempswa \titleformat#8% {\ttl@fil\ttl@fonts\ttl@sizes{#3}}{\@seccntformat{#2}}% {\z@}\ttl@passexplicit \else \titleformat#8% {#7}{\@seccntformat{#2}}% {\z@}\ttl@passexplicit \fi \@ifundefined{ttl@space}{}{% \ttl@assign\@tempskipb*\ttl@space\relax\aftertitleunit}% \fi \edef\ttl@a{\the\toks@{\the\@tempskipa}{\the\@tempskipb}} \ttl@a} \def\ttl@extract#1{% \ifx#1\@undefined \let#1\relax % Avoid error if undefined \fi \expandafter\in@\expandafter\@startsection\expandafter{#1}% \ifin@ \expandafter\ttl@@extract#1#1% \else \PackageWarningNoLine{titlesec}% {Non standard sectioning command \string#1\MessageBreak detected. Using default spacing and no format}% \titlespacing*#1{\z@}{*3}{*2}% \fi} \@tempswafalse \ifx\ttl@fonts\@empty \def\ttl@fonts{\bfseries} \else \@tempswatrue \fi \expandafter\ifx\csname ttl@sizes\endcsname\relax \gdef\ttl@sizes#1{\ifcase#1\relax\Huge\or\Large\or\large \or\normalsize\or\or\or\huge\fi} \else \@tempswatrue \fi \expandafter\ifx\csname ttl@fil\endcsname\relax \let\ttl@fil\@empty \else \@tempswatrue \fi \expandafter\ifx\csname ttl@case\endcsname\relax \let\ttl@case\@firstofone \else \@tempswatrue \fi \if@tempswa \expandafter\ifx\csname chapter\endcsname\relax\else \titleformat\chapter[display]% {\@ifundefined{ttl@fil}{\raggedright}{\ttl@fil}\ttl@fonts\ttl@sizes6} {\@chapapp\space\thechapter}{.8\baselineskip}{\ttl@sizes\z@\ttl@passexplicit} \fi \fi \ttl@extract\section \ttl@extract\subsection \ttl@extract\subsubsection \ttl@extract\paragraph \ttl@extract\subparagraph \let\ttl@extract\@undefined \let\ttl@@extract\@undefined \def\ttl@toplevel{part} \expandafter\ifx\csname chapter\endcsname\relax \@namedef{ttll@part}{0} \titleclass{\section}{straight}[\part] \titlespacing*{\part} {\z@} {4ex} {3ex} \else \let\ttl@save@mkchap\@makechapterhead \let\ttl@save@mkschap\@makeschapterhead \def\@makechapterhead#1{% \gdef\ttl@savemark{\chaptermark{#1}}% \ttl@save@mkchap{#1}% \@ifundefined{ttl@ps@chapter}{}% {\thispagestyle{\@nameuse{ttl@ps@chapter}}}} \def\@makeschapterhead#1{% \gdef\ttl@savemark{\chaptermark{#1}}% \ttl@save@mkschap{#1}% \@ifundefined{ttl@ps@chapter}{}% {\thispagestyle{\@nameuse{ttl@ps@chapter}}}} \@namedef{ttll@part}{-1} \@namedef{ttlss@part}{chapter} \@namedef{ttll@chapter}{0} \titleclass{\section}{straight}[\chapter] % The following is unoperant, unless when \chapter / \part % format is redefined \titlespacing*{\part} {\z@} {\z@\@plus1fil} {\z@\@plus1fil} \titlespacing*\chapter {\z@}% {50\p@}% {\ttl@chapafter}% \fi \titleclass{\subsection} {straight}[\section] \titleclass{\subsubsection}{straight}[\subsection] \titleclass{\paragraph} {straight}[\subsubsection] \titleclass{\subparagraph} {straight}[\paragraph] \endinput MIT License ----------- Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. liu2231665/TAAN-MTL @article{Yang2016DeepMR, title={Deep Multi-task Representation Learning: A Tensor Factorisation Approach}, author={ and }, journal={ArXiv}, year={2016}, volume={abs/1605.06391} } \section{PyLith Application (\protect\object{PyLithApp})} The top-level object is the PyLith application with three facilities: \begin{inventory} \facilityitem{mesher}{Importer for the finite-element mesh;} \facilityitem{problem}{Problem to run, such as the materials, boundary conditions, etc.; and} \facilityitem{petsc}{PETSc settings} \end{inventory} \subsection{Mesh Information (\facility{mesher})} Geometrical and topological information for the finite element mesh may be provided by exporting an Exodus II format file from CUBIT/Trelis, by exporting a GMV file and an accompanying Pset file from LaGriT, or by specifying the information in PyLith mesh ASCII format. See Chapter \vref{cha:examples} for examples. PyLith supports linear cells in 2D (Figure \vref{fig:2D:cells}), and 3D (Figure \vref{fig:3D:cells}). The vertex ordering must follow the convention shown in Figures \vref{fig:2D:cells}-\vref{fig:3D:cells}. PyLith no longer supports use of quadratic cells using the PyLith ASCII mesh format. In the next release, we plan to support higher order discretizations via PETSc finite-element features from meshes with linear cells as input. The mesh information defines the vertex coordinates and specifies the vertices composing each cell in the mesh. The mesh information must also define at least one set of vertices for which displacement (Dirichlet) boundary conditions will be provided. In most realistic problems, there will be several vertex groups, each with a unique identifying label. For example, one group might define a surface of the mesh where displacement (Dirichlet) boundary conditions will be applied, another might define a surface where traction (Neumann) boundary conditions will be applied, while a third might specify a surface that defines a fault. Similarly, the mesh information contains cell labels that define the material type for each cell in the mesh. For a mesh with a single material type, there will only be a single label for every cell in the mesh. See Chapters \vref{cha:material:models} and \vref{cha:boundary:interface:conditions} for more detailed discussions of setting the materials and boundary conditions. \begin{figure}[htbp] \includegraphics[scale=0.6]{runpylith/figs/tri3}\hspace*{0.5in}% \includegraphics[scale=0.6]{runpylith/figs/quad4} \caption{Linear cells available for 2D problems are the triangle (left) and the quadrilateral (right).} \label{fig:2D:cells} \end{figure} \begin{figure}[htbp] \includegraphics[scale=0.6]{runpylith/figs/tet4}\hspace*{0.5in}% \includegraphics[scale=0.6]{runpylith/figs/hex8} \caption{Linear cells available for 3D problems are the tetrahedron (left) and the hexahedron (right).} \label{fig:3D:cells} \end{figure} \subsubsection{\object{Mesh Importer}} The default mesher component is \object{MeshImporter}, which provides the capabilities of reading the mesh from files. The \object{MeshImporter} has several properties and facilities: \begin{inventory} \propertyitem{reorder\_mesh}{Reorder the vertices and cells using the reverse Cuthill-McKee algorithm (default is False)} \facilityitem{reader}{Reader for a given type of mesh (default is \object{MeshIOAscii}).} \facilityitem{distributor}{Handles distribution of the mesh among processors.} \facilityitem{refiner}{Perform global uniform mesh refinement after distribution among processors (default is no refinement).} \end{inventory} Reordering the mesh so that vertices and cells connected topologically also reside close together in memory improves overall performance and can improve solver performance as well. \userwarning{The coordinate system associated with the mesh must be a Cartesian coordinate system, such as a generic Cartesian coordinate system or a geographic projection.} \subsubsection{\object{MeshIOAscii}} The \object{MeshIOAscii} object is intended for reading small, simple ASCII files containing a mesh constructed by hand. We use this file format extensively in the examples. Appendix \vref{sec:format:MeshIOAscii} describes the format of the files. The properties and facilities of the \object{MeshIOAscii} object include: \begin{inventory} \propertyitem{filename}{Name of the mesh file.} \facilityitem{coordsys}{Coordinate system associated with the mesh.} \end{inventory} \subsubsection{\object{MeshIOCubit}} \label{sec:MeshIOCubit} The \object{MeshIOCubit} object reads the NetCDF Exodus II files output from CUBIT/Trelis. Beginning with CUBIT 11.0, the names of the nodesets are included in the Exodus II files and PyLith can use these nodeset names or revert to using the nodeset ids. The properties and facilities associated with the \object{MeshIOCubit} object are: \begin{inventory} \propertyitem{filename}{Name of the Exodus II file.} \propertyitem{use\_nodeset\_names}{Identify nodesets by name rather than id (default is True).} \facilityitem{coordsys}{Coordinate system associated with the mesh.} \end{inventory} \subsubsection{\object{MeshIOLagrit}} \label{sec:MeshIOLagrit} The \object{MeshIOLagrit} object is used to read ASCII and binary GMV and PSET files output from LaGriT. PyLith will automatically detect whether the files are ASCII or binary. We attempt to provide support for experimental 64-bit versions of LaGriT via flags indicating whether the FORTRAN code is using 32-bit or 64-bit integers. The \object{MeshIOLagrit} properties and facilities are: \begin{inventory} \propertyitem{filename\_gmv}{Name of GMV file.} \propertyitem{filename\_pset}{Name of the PSET file.} \propertyitem{flip\_endian}{Flip the endian of values when reading binary files (default is False).} \propertyitem{io\_int32}{Flag indicating that PSET files use 32-bit integers (default is True).} \propertyitem{record\_header\_32bt}{Flag indicating FORTRAN record header is 32-bit (default is True).} \facilityitem{coordsys}{Coordinate system associated with mesh.} \end{inventory} \userwarning{The PyLith developers have not used LaGriT since around 2008 and the most recent release appears to have been in 2010.} \subsubsection{\object{Distributor}} The distributor uses a partitioner to compute which cells should be placed on each processor, computes the overlap among the processors, and then distributes the mesh among the processors. The type of partitioner is set via PETSc settings. The properties and facilities of the \object{Distributor} include: \begin{inventory} \propertyitem{partitioner}{Name of mesh partitioner ['chaco','parmetis'].} \propertyitem{write\_partition}{Flag indicating that the partition information should be written to a file (default is False).} \facilityitem{data\_writer}{Writer for partition information (default is \object{DataWriterVTK} for VTK output).} \end{inventory} \begin{cfg}[\object{Distributor} parameters in a \filename{cfg} file] [pylithapp.mesh_generator.distributor]

partitioner

= chaco ; Options are 'chaco' (default) and 'parmetis'. \end{cfg} METIS/ParMETIS are not included in the PyLith binaries due to licensing issues. \subsubsection{\object{Refiner}} The refiner is used to decrease node spacing by a power of two by recursively subdividing each cell by a factor of two. In a 2D triangular mesh a node is inserted at the midpoint of each edge, splitting each cell into four cells (see Figure \vref{fig:uniform:refinement:2x}). In a 2D quadrilateral mesh a node is inserted at the midpoint of each edge and at the centroid of the cell, splitting each cell into four cells. In a 3D tetrahedral mesh a node is inserted at the midpoint of each edge, splitting each cell into eight cells. In a 3D hexahedral mesh a node is inserted at the midpoint of each edge, the centroid of each face, and at the centroid of the cell, splitting each cell into eight cells. \begin{figure}[htbp] \includegraphics[scale=0.6]{runpylith/figs/refinement2x} \caption{Global uniform mesh refinement of 2D and 3D linear cells. The blue lines and orange circles identify the edges and vertices in the original cells. The purple lines and green circles identify the new edges and vertices added to the original cells to refine the mesh by a factor of two.} \label{fig:uniform:refinement:2x} \end{figure} Refinement occurs after distribution of the mesh among processors. This allows one to run much larger simulations by (1) permitting the mesh generator to construct a mesh with a node spacing larger than that needed in the simulation and (2) operations performed in serial during the simulation setup phase, such as, adjusting the topology to insert cohesive cells and distribution of the mesh among processors uses this much smaller coarse mesh. For 2D problems the global mesh refinement increases the maximum problem size by a factor of $4^{n}$, and for 3D problems it increases the maximum problem size by a factor of $8^{n}$, where $n$ is the number of recursive refinement levels. For a tetrahedral mesh, the element quality decreases with refinement so $n$ should be limited to 1-2. \subsection{Problem Specification (\protect\facility{problem})} The problem component specifies the basic parameters of the simulation, including the physical properties, the boundary conditions, and interface conditions (faults). The current release of PyLith contains two types of problems, \object{TimeDependent} for use in static, quasistatic, and dynamic simulations and \object{GreensFns} for computing static Green's functions. The general properties facilities include: \begin{inventory} \propertyitem{solver}{Type of solver to use ({\tt linear} or {\tt nonlinear});} \facilityitem{solution}{Solution field;} \facilityitem{normalizer}{Scales used to nondimensionalize the problem (default is \object{NondimElasticQuasistatic});} \facilityitem{materials}{Array of materials comprising the domain (default is [material]);} \facilityitem{bc}{Array of boundary conditions (default is none);} \facilityitem{interfaces}{Array of interface conditions, i.e., faults (default is none); and} \facilityitem{gravity\_field}{Gravity field used to construct body forces (default=\object{NullComponent});} \end{inventory} \begin{cfg}[Problem parameters in a \filename{cfg} file] [pylithapp.timedependent]

solver

= linear solution = pylith.problems.SolnDisp normalizer = spatialdata.units.NondimElasticQuasistatic materials = [elastic, viscoelastic] bc = [boundary_east, boundary_bottom, boundary_west] interfaces = [SanAndreas, SanJacinto] gravity_field = spatialdata.spatialdb.GravityField \end{cfg} The following sections discuss the \facility{solution} and \facility{normalizer}. Materials, boundary conditions, and interface conditions are discussed in Chapter \vref{cha:physics} and the gravity field spatial database is discussed in Section \vref{sec:gravity:field}. \subsubsection{Solution Field (\facility{solution})} The \facility{solution\_field} specifies the subfields of the solution field along with their discretization. Table \vref{tab:solution:containers} shows predefined containers for common subfield collections. Users can create their own containers if they add different material formulations. \important{The order of the subfields within the solution field must be consistent between the \object{Solution} field object and the point-wise functions. The predefined containers are setup to help ensure that this is true.} \important{When a Lagrange multiplier for fault interfaces is included, it should always be the last solution subfield. It is special case, because it is discretization only on the cohesive cells and not over the entire domain.} \begin{table}[htbp] \caption{Predefined containers for solution subfields.} \label{tab:solution:containers} \begin{tabular}{lll} \toprule \thead{Object} & \thead{Subfields} & \thead{Use Cases} \\ \midrule \object{SolnDisp} & displacement & Elasticity w/o inertia and faults \\ \object{SolnDispVel} & displacement, velocity & Elasticity w/inertia and w/o faults \\ \object{SolnDispLagrange} & displacement, lagrange\_fault & Elasticity w/faults \\ \object{SolnDispVelLagrange} & displacement, lagrange\_fault & Elasticity w/faults \\ \object{SolnDispPres} & displacement, pressure & Incompressible elasticity w/o faults \\ \object{SolnDispPresLagrange} & displacement, pressure, lagrange\_fault & Incompressible elasticity w/o faults \\ \bottomrule \end{tabular} \end{table} Each subfield is a \object{SolutionSubfield} object with the following properties: \begin{inventory} \propertyitem{alias}{User-specified name for subfield to use in output (default is the PyLith-specified name);} \propertyitem{basis\_order}{Order for basis functions (default=1);} \propertyitem{quadrature\_order}{Order of quadrature to use in integration (default=1);} \propertyitem{dimension}{Topological dimension associated with subfield (default=-1 and should not be changed); and} \propertyitem{finite\_element\_space}{Finite-element space ({\tt polynomial} or {\tt point; defaults=polynomial}). Point space corresponds to delta functions at the quadrature points;} \end{inventory} \begin{cfg}[Setting discretization information for a \object{SolnDispLagrange} component in a \filename{cfg} file] [pylithapp.problem] solution = pylith.problems.SolnDispLagrange [pylithapp.problem.solution.subfields]

displacement.basis_order

= 1

displacement.quadrature_order

= 1

lagrange_fault.basis_order

= 1

lagrange_fault.quadrature_order

= 1 \end{cfg} \subsubsection{Nondimensionalization (\facility{normalizer})} PyLith rescales all parameters provided by the user so that the simulation solves the equations using nondimensional quantities. This permits application of PyLith to problems across a vast range of spatial and temporal scales. The scales used to nondimensionalize the problem are length, pressure, density, and time. PyLith provides two normalizer objects to make it easy to provide reasonable scales for the nondimensionalization. The \object{NondimElasticQuasistatic} normalizer (which is the default) has the following properties: \begin{inventory} \propertyitem{length\_scale}{Distance to nondimensionalize length (default is 1.0 km).} \propertyitem{shear\_modulus}{Shear modulus to nondimensionalize pressure (default is 3.0e+10 Pa).} \propertyitem{relaxation\_time}{Relaxation time to nondimensionalize time (default is 1.0 year).} \end{inventory} \begin{cfg}[\object{NondimElasticQuasistatic} parameters in a \filename{cfg} file] [pylithapp.timedependent.normalizer]

length_scale

= 1.0*km

shear_modulus

= 3.0e+10*Pa

relaxation_time

= 1.0*yr \end{cfg} The \object{NondimElasticDynamic} normalizer has the following properties: \begin{inventory} \propertyitem{shear\_wave\_speed}{Shear wave speed used to nondimensionalize length and pressure (default is 3.0 km/s).} \propertyitem{mass\_density}{Mass density to nondimensionalize density and pressure (default is 3.0e+3 kg/m$^{3}$).} \propertyitem{wave\_period}{Period of seismic waves used to nondimensionalize time (default is 1.0 s).} \end{inventory} \begin{cfg}[\object{NondimElasticDynamic} parameters in a \filename{cfg} file] [pylithapp.timedependent.normalizer]

shear_wave_speed

= 3.0*km/s

mass_density

= 3.0e+3*kg/m**3

wave_period

= 1.0*s \end{cfg} \important{The default nondimensionalization is reasonable for many problems; however, it may be necessary to change the default values in some cases. When doing this, keep in mind that the nondimensionalization generally applies to the minimum values encountered for a problem. For example, in a quasistatic problem, the \property{length\_scale} should be on the order of the minimum cell size. Similarly, the \property{relaxation\_time} should be on the order of the minimum relaxation time or time scale associated with time-dependent boundary and interface conditions.} \subsubsection{Solution Observers (\facility{solution\_observers})} \label{sec:solution:observers} The solution observers get notified of updates to the solution. Table \vref{solution:observers} lists the current implementations of solution observers, which are used for output. \begin{table}[htbp] \caption{Solution observers.} \label{tab:solution:observers} \begin{tabular}{lll} \toprule \thead{Object} & \thead{Use Cases} \\ \midrule \object{OutputSoln} & Output of the solution over the domain; \\ \object{OutputSolnBoundary} & Output of the solution over an external boundary \\ \object{OutputSolnPoints} & Output of the solution at discrete points \\ \bottomrule \end{tabular} \end{table} All of the solution observers have the following properties: \begin{inventory} \propertyitem{data\_fields}{List of solution subfields to observer/output (default=all which will output all of the subfields);} \facilityitem{writer}{Writer for data (default=\object{DataWriterHDF5});} \facilityitem{trigger}{Trigger defining how often output is written (default=\object{OutputTriggerStep}); and} \facilityitem{field\_filter}{Filter for output fields (default=\object{FieldFilterNone}).} \end{inventory} \object{OutputSolnBoundary} adds a property: \begin{inventory} \propertyitem{label}{Label (name of nodeset/pset) identifier of boundary (required);} \end{inventory} See Section \vref{sec:output} for detailed information about the available components for the \facility{writer}, \facility{trigger}, and \facility{field\_filter} facilities. \begin{cfg}[Setting \object{OutputSolnBoundary} parameters in a \filename{cfg} file] [pylithapp.problem.solution_observers.boundary]

label

= boundary_groundsurf

writer.filename

= output/step01-grounssurf.h5 \end{cfg} \paragraph{Output at Arbitrary Points (\protect\object{OutputSolnPoints})} \label{sec:output:points} In many situations with recorded observations, one would like to extract the solution at the same locations as the recorded observation. Rather than forcing the finite-element discretization to be consistent with the observation points, PyLith includes a specialized solution observer, \object{OutputSolnPoints}, to interpolate the solution to arbitrary points. The locations are specified in a text file. The \object{OutputSolnPoints} includes: \begin{inventory} \propertyitem{data\_fields}{List of solution subfields to observer/output (default=all which will output all of the subfields);} \facilityitem{reader}{Reader for points list (default is \object{PointsList}); and} \end{inventory} \subsubsection{\object{PointsList} Reader} This object corresponds to a simple text file containing a list of points (one per line) where output is desired. See \vref{sec:format:PointsList} for file format specifications. The points are specified in the coordinate system specified by \object{OutputSolnPoints}. The coordinates will be transformed into the coordinate system of the mesh prior to interpolation. The properties available to customize the behavior of \object{PointsList} are: \begin{inventory} \propertyitem{filename}{Names of file containing list of points.} \propertyitem{comment\_delimiter}{Delimiter at beginning of line to identify comments (default is \#).} \propertyitem{value\_delimiter}{Delimiter used to separate values (default is whitespace).} \end{inventory} \subsection{PETSc Settings (\protect\facility{petsc})} \label{sec:petsc:options} PyLith relies on PETSc for the finite-element data structures, linear and nonlinear solvers, and time-stepping algorithms. PETSc has its own object-oriented interface for specifying runtime options. Instead of trying to maintain a Pyre interface to all of the PETSc options, we use a single \facility{petsc} facility to collect all of the PETSc options and pass them to PETSc. PETSc time-stepping options are discussed in Section \vref{sec:problems:timedependent}. \subsubsection{Monitor/Logging Settings} Table \vref{tab:pesc:options:monitor} shows the main monitoring options offered by PETSc. Our recommended settings for all simulations include: \begin{cfg}[Recommended PETSc monitoring settings as set in a \filename{cfg} file.] [pylithapp.petsc] # Trigger errors if linear or nonlinear solver fails to converge.

ksp_error_if_not_converged

= true

snes_error_if_not_converged

= true # Monitor converged reasons

ksp_converged_reason

= true

snes_converged_reason

= true # Monitor time-stepping and nonlinear solver

ts_monitor

= true

snes_monitor

= true

snes_linesearch_monitor

= true \end{cfg} When optimizing and troubleshooting solver settings, we usually turn on all the monitoring. \begin{table}[htbp] \caption{Description of PETSc monitoring settings.} \label{tab:petsc:options:monitor} \begin{tabular}{lp{4.0in}} \toprule \thead{Option} & \thead{Description} \\ \midrule % log \property{log\_view} & Show logging objects and events. \\ % TS \property{ts\_monitor} & Show time-stepping progress. \\ % KSP \property{ksp\_monitor} & Show preconditioned residual norm. \\ \property{ksp\_view} & Show linear solver parameters. \\ \property{ksp\_error\_if\_not\_converged} & Generate an error if linear solver does not converge. \\ \property{ksp\_converged\_reason} & Indicate why iterating stopped in linear solve. \\ % SNES \property{snes\_monitor} & Show residual norm for each nonlinear solve iteration. \\ \property{snes\_view} & Show nonlinear solver parameters. \\ \property{snes\_error\_if\_not\_converged} & Generate an error if nonlinear solver does not converge. \\ \property{snes\_converged\_reason} & Indicate why iterating stopped in nonlinear solve. \\ \property{snes\_linesearch\_monitor} & Show line search information in nonlinear solve. \\ \bottomrule \end{tabular} \end{table} \subsubsection{Solver Settings} For most problems we use the GMRES method from Saad and Schultz for the linear solver with solver tolerances around 1.0e-10. When running large problems, we often raise the solver tolerances by one or two orders of magnitude to reduce runtime while still achieving suitable accuracy. See \href{http://www.mcs.anl.gov/petsc/petsc-as/documentation/linearsolvertable.html}{PETSc linear solver table} for a list of PETSc options for linear solvers and preconditioners. \usertip{It is important to keep in mind the resolution of the model and observations when setting solver tolerances. For example, matching observations with an accuracy of 1.0\si{\milli\meter} does not require solving the equations to an accuracy of 0.0001\si{\milli\meter}.} \begin{table}[htbp] \caption{Recommended starting point for PETSc solver tolerances.} \label{tab:petsc:options:solver} \begin{tabular}{lcp{4.5in}} \toprule \thead{Property} & \thead{Value} & \thead{Description} \\ \midrule % KSP \property{ksp\_rtol} & 1.0e-10 & Stop iterating when the preconditioned KSP residual norm has this amount relative to its starting value.\\ \property{ksp\_atol} & 1.0e-12 & Stop iterating when the preconditioned KSP residual normal is smaller than this value.\\ % SNES \property{snes\_rtol} & 1.0e-10 & Stop iterating when the SNES residual norm has this amount relative to its starting value.\\ \property{snes\_atol} & 1.0e-10 & Stop iterating when the SNES residual normal is smaller than this value.\\ \bottomrule \end{tabular} \end{table} \paragraph{Settings for small problems} When running small test problems (about 1k or less unknowns) it is very handy to use a robust preconditioner so that issues related to the boundary conditions, material parameters, etc. are more obvious. We recommend using Incomplete (ILU) factorization. \begin{cfg}[Recommended PETSc solver settings for small problems] [pylithapp.petsc]

pc_type

= ilu

ksp_type

= gmres \end{cfg} \paragraph{Settings for medium problems} When running slightly larger problems (about 10k or less unknowns), the Additive Schwarz Method (ASM) using Incomplete LU (ILU) factorization preconditioner is usually more efficient. \begin{cfg}[Recommended PETSc solver settings for medium problems] [pylithapp.petsc]

pc_type

= asm

ksp_type

= gmres \end{cfg} \paragraph{Efficient settings for elasticity without a fault} Algebraic multigrid preconditioner usually works very well on elasticity problems. \begin{cfg}[Recommended PETSc solver settings for solving elasticity problems without a fault] [pylithapp.petsc]

pc_type

= ml

ksp_type

= gmres \end{cfg} \important{The ML algebraic multigrid preconditioner is only available if you build PETSc with the ML package. These features are included in the PyLith binary packages.} \paragraph{Efficient settings for elasticity with a fault} The Lagrange multiplier solution subfield introduces a saddle point in the system of equations, so we use a Schur complement approach. These settings are available in \filename{\$PYLITH\_DIR/share/settings/solver\_fault\_exact.cfg}. \begin{cfg}[Recommended PETSc solver settings for solving elasticity problems with a fault] [pylithapp.petsc]

pc_type

= fieldsplit

pc_use_amat

= true

pc_fieldsplit_type

= schur

pc_fieldsplit_schur_factorization_type

= full

pc_fieldsplit_dm_splits

= true

fieldsplit_displacement_ksp_type

= preonly

fieldsplit_displacement_pc_type

= lu

fieldsplit_lagrange_multiplier_fault_pc_type

= jacobi

fieldsplit_lagrange_multiplier_fault_ksp_type

= gmres

fieldsplit_lagrange_multiplier_fault_ksp_rtol

= 1.0e-11

fieldsplit_lagrange_multiplier_fault_ksp_converged_reason

= true \end{cfg} \userwarning{The split fields and algebraic multigrid preconditioning currently fails in problems with a nonzero null space. This most often occurs when a problem contains multiple faults that extend through the entire domain and create subdomains without any Dirichlet boundary conditions. The current workaround is to use the Additive Schwarz preconditioner without split fields. See Section \vref{sec:Troubleshooting} for the error message encountered in this situation.} \paragraph{Efficient settings for incompressible elasticity} The pressure solution subfield introduces a saddle point in the system of equations, so we again use a Schur complement approach. This time we can use algebraic multigrid preconditioning on each block. These settings are available in \filename{\$PYLITH\_DIR/share/settings/solver\_incompressible\_elasticity.cfg}. \begin{cfg}[Recommended PETSc solver settings for solving incompressible elasticity problems without a fault] [pylithapp.petsc]

pc_type

= fieldsplit

pc_fieldsplit_type

= schur

pc_fieldsplit_schur_fact_type

= full

pc_fieldsplit_schur_precondition

= full

fieldsplit_displacement_pc_type

= lu

fieldsplit_pressure_pc_type

= lu \end{cfg} % End of file wenshen/bitcon-ansi-c0 \hypertarget{struct_m_i_byte_array}{ \section{ByteArray Struct Reference} \label{struct_m_i_byte_array}\index{ByteArray@{ByteArray}} } Structure for \hyperlink{struct_m_i_byte_array}{ByteArray} objects. {\ttfamily \#include $<$ByteArray.h$>$} Collaboration diagram for ByteArray: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=282pt]{struct_m_i_byte_array__coll__graph} \end{center} \end{figure} \subsection*{Data Fields} \begin{DoxyCompactItemize} \item \hyperlink{struct_m_i_object}{Object} \hyperlink{struct_m_i_byte_array_ad0814be49cef65d1662b278c4c591509}{base} \item \hyperlink{struct_m_i_shared_data}{SharedData} $\ast$ \hyperlink{struct_m_i_byte_array_a2092be3761112a91b7da551b8f834b02}{sharedData} \item uint32\_\-t \hyperlink{struct_m_i_byte_array_a894bdfa2d603d8343f8ef01dda6fcd23}{offset} \item uint32\_\-t \hyperlink{struct_m_i_byte_array_aebb70c2aab3407a9f05334c47131a43b}{length} \item void($\ast$ \hyperlink{struct_m_i_byte_array_aeb0776cff292839518da5f5a48884fd9}{onErrorReceived} )(\hyperlink{_m_i_constants_8h_a78789cd8e7333545dd73768531261968}{Error} error, char $\ast$,...) \end{DoxyCompactItemize} \subsection{Detailed Description} Structure for \hyperlink{struct_m_i_byte_array}{ByteArray} objects. \begin{DoxySeeAlso}{See also} \hyperlink{_m_i_byte_array_8h}{ByteArray.h} \end{DoxySeeAlso} Definition at line 29 of file ByteArray.h. \subsection{Field Documentation} \hypertarget{struct_m_i_byte_array_ad0814be49cef65d1662b278c4c591509}{ \index{ByteArray@{ByteArray}!base@{base}} \index{base@{base}!ByteArray@{ByteArray}} \subsubsection[{base}]{\setlength{\rightskip}{0pt plus 5cm}{\bf Object} {\bf base}}} \label{struct_m_i_byte_array_ad0814be49cef65d1662b278c4c591509} \hyperlink{struct_m_i_object}{Object} base structure Definition at line 30 of file ByteArray.h. \hypertarget{struct_m_i_byte_array_aebb70c2aab3407a9f05334c47131a43b}{ \index{ByteArray@{ByteArray}!length@{length}} \index{length@{length}!ByteArray@{ByteArray}} \subsubsection[{length}]{\setlength{\rightskip}{0pt plus 5cm}uint32\_\-t {\bf length}}} \label{struct_m_i_byte_array_aebb70c2aab3407a9f05334c47131a43b} Length of byte array. Set to \_\-BYTE\_\-ARRAY\_\-UNKNOWN\_\-LENGTH if unknown. Definition at line 33 of file ByteArray.h. \hypertarget{struct_m_i_byte_array_a894bdfa2d603d8343f8ef01dda6fcd23}{ \index{ByteArray@{ByteArray}!offset@{offset}} \index{offset@{offset}!ByteArray@{ByteArray}} \subsubsection[{offset}]{\setlength{\rightskip}{0pt plus 5cm}uint32\_\-t {\bf offset}}} \label{struct_m_i_byte_array_a894bdfa2d603d8343f8ef01dda6fcd23} Offset from the beginning of the byte data to the beginning of this array Definition at line 32 of file ByteArray.h. \hypertarget{struct_m_i_byte_array_aeb0776cff292839518da5f5a48884fd9}{ \index{ByteArray@{ByteArray}!onErrorReceived@{onErrorReceived}} \index{onErrorReceived@{onErrorReceived}!ByteArray@{ByteArray}} \subsubsection[{onErrorReceived}]{\setlength{\rightskip}{0pt plus 5cm}void($\ast$ {\bf onErrorReceived})({\bf Error} error, char $\ast$,...)}} \label{struct_m_i_byte_array_aeb0776cff292839518da5f5a48884fd9} Definition at line 34 of file ByteArray.h. \hypertarget{struct_m_i_byte_array_a2092be3761112a91b7da551b8f834b02}{ \index{ByteArray@{ByteArray}!sharedData@{sharedData}} \index{sharedData@{sharedData}!ByteArray@{ByteArray}} \subsubsection[{sharedData}]{\setlength{\rightskip}{0pt plus 5cm}{\bf SharedData}$\ast$ {\bf sharedData}}} \label{struct_m_i_byte_array_a2092be3761112a91b7da551b8f834b02} Underlying byte data Definition at line 31 of file ByteArray.h. The documentation for this struct was generated from the following file:\begin{DoxyCompactItemize} \item Object/\hyperlink{_m_i_byte_array_8h}{ByteArray.h}\end{DoxyCompactItemize} Shawn was a Nursing Assistant and within 2 days she had Convulsions and Central Nervous system involvement. Her symptoms still continue. Shawn in her own words: 4th June 2021: As most of you know I took the Vaccine on 4th January 2021 and since my life has forever changed! I have lost many friends, most of my co-workers have abandoned me. My job terminated me! Everything I loved to do I can no longer enjoy, my family is suffering! I miss my job, my patients, that was my calling! It has all been taken from me and American Senior Communities, Good Samaritan Home and Rehabilitative Center dropped me like I was nothing! Workmans Comp denied me even though my DON told me to go to the ER. It’s on my chart – I’m allergic to the Moderna Vaccine yet can’t get a real diagnosis or any medical help! 26th June 2021: Please keep me in your prayers!! I have taken a few steps back again this week. I lost use of my legs on Wednesday, I was able to get a walker on Thursday to help me around but it’s definitely taking a physical and mental toll! I’m trying to keep in mind the good days I have too but these bad ones are really bad! 28 August 2021: Shawn still continues to suffer 8 months out from the vaccine, “I was sitting on my bed talking with my husband and best friend, normal conversation and within seconds things went south! They thought I was dead! It took me so long coming out of this I thought I was going to die! Whatever happened left me paralyzed from my waist down until this morning I was able to use my legs again! \begin{pf} Предположим противное: пусть $\zeta\left(1+it_0\right)=0$. Тогда при $\sigma \to 1+:$\\ $\displaystyle \zeta^3(\sigma)\zeta^4\left(\sigma+it_0\right)\zeta\left(\sigma+2it_0\right) = O\left( \frac{1}{(\sigma-1)^3}(\sigma-1)^4\cdot1 \right) = O_{\sigma \to 1}(\sigma -1)$. (Т.к. $\zeta(\sigma) \to +\infty$ при $\sigma \to 1+$, точнее, $\displaystyle \zeta(\sigma)=O\left( \frac{1}{\sigma-1} \right)$ ибо полюс порядка $1$; $\zeta\left( 1+it_0 \right)=0 \xRightarrow{\text{из мультипл.}} \zeta\left( \sigma+it_0 \right)=O(\sigma-1)$; $\zeta\left( 1+2it_0 \right)$ — какая-то константа, полюса там нет из аналитичности функции в $\Re(s)>0$ везде, кроме $1$). Итак, получили $\zeta^3(\sigma)\zeta^4\left(\sigma+it_0\right)\zeta\left(\sigma+2it_0\right) = O_{\sigma \to 1}(\sigma -1)$, но по Лемме \ref{l3_lm9} её модуль $\geq 1$ при любом $\sigma >1$. Противоречие.\\ (Из Леммы \ref{l3_lm9} также можно ещё одним способом получить, что в полуплоскости $\Re(s)>1$ у $\zeta$-функции нет корней: если бы существовал корень $s=\sigma+it$, то $\lvert \zeta^3(\sigma)\zeta^4(s)\zeta(\sigma+2it) \rvert \geq 1$, противоречие). \end{pf} \begin{lemma} \label{l4_lm10} $\displaystyle \frac{\zeta'(s)}{\zeta(s)} + \frac{1}{s-1}$ аналитична при $\Re(s) \geq 1$. \end{lemma} \begin{pf} Знаем, что при $\Re(s) > 1$ оба слагаемых — аналитические функции. Мы также доказали, что $\displaystyle \zeta(s)=\frac{f(s)}{s-1}$, где $f(s)$ точно аналитична при $\Re(s) > 0$ и $f(s) \ne 0$ при $\Re(s) \geq 1.$ Отсюда следует, что $\displaystyle \frac{\zeta'(s)}{\zeta(s)} = \frac{f'(s)}{f(s)} - \frac{1}{s-1}$, где $f$ аналитична при $\Re(s) > 0$, а значит, что $f'$ тоже. В $\Re(s)\geq 1$ у знаменателя нет нулей. \end{pf} Положим $\displaystyle F(s) := -\frac{1}{s}\frac{\zeta'(s)}{\zeta(s)} - \frac{1}{s-1}$.\\ \begin{lemma} \label{l2_lm11} Справедливы следующие утверждения \begin{enumerate}[nolistsep] \item[1)] $F(s)$ аналитична в $\Re(s) \geq 1$. \item[2)] $\displaystyle F(s) = \int_1^{+\infty} \frac{\psi(x)-x}{x^{1+s}}dx$ при $\Re(s) > 1$. \end{enumerate} \end{lemma} \begin{pf} По порядку. \begin{enumerate}[nolistsep] \item[1)] $\displaystyle F(s) = -\frac{1}{s}\left( \frac{\zeta'(s)}{\zeta(s)} + \frac{s}{s-1} \right) = -\frac{1}{s}\left( \frac{\zeta'(s)}{\zeta(s)} + \frac{1}{s-1} +1 \right)$ — $\displaystyle -\frac{1}{s}$ аналитична, а второй множитель аналитичен по Лемме \ref{l4_lm10}. \item[2)] При $\displaystyle \Re(s)>1 \ \frac{\zeta'(s)}{\zeta(s)} = -s\int_1^{+\infty} \frac{\psi(x)dx}{x^{1+s}}, \, \frac{1}{s-1}=\int_1^{+\infty}\frac{dx}{x^s}$. Оба интеграла сходятся абсолютно, поэтому можно их складывать: $$F(s) = \int_1^{+\infty}\frac{\psi(x)dx}{x^{1+s}} - \int_1^{+\infty} \frac{dx}{x^s} = \int_1^{+\infty} \frac{\psi(x)-x}{x^{1+s}}dx.$$ \end{enumerate} \end{pf} \begin{theorem} \label{l4_thm7} В интегральном представлении $F(s)$ можно перейти к пределу в $\Re(s)>1$, т.е. $$F(1) = \int_1^{+\infty} \frac{\psi(x)-x}{x^2}dx.$$ \end{theorem} \begin{lemma} \label{l4_lm12} Если интеграл $\displaystyle \int_1^{+\infty} \frac{\psi(x)-x}{x^2}dx$ сходится (это будет следовать из Теоремы \ref{l4_thm7}), то $\psi(x) \sim x$. \end{lemma} \begin{pf} Возьмём $\varepsilon >0:$ $$\int_x^{(1+\varepsilon)x}\frac{\psi(u)-u}{u^2}du \geq \varepsilon x \frac{\psi(x)-(1+\varepsilon)x}{(1+\varepsilon)^2x^2} = \frac{\varepsilon}{\left(1+\varepsilon^2\right)x^2}\left(\frac{\psi(x)}{x}-(1+\varepsilon)\right).$$ Из сходимости интеграла слева при фиксированном $\varepsilon$ получаем $\displaystyle \int_x^{(1+\varepsilon)x}\frac{\psi(u)-u}{u^2}du \xrightarrow{x \to \infty} 0$. Следовательно, $\displaystyle \varlimsup\limits_{x\to\infty} \frac{\varepsilon}{(1+\varepsilon)^2}\left( \frac{\psi(x)}{x}-(1+\varepsilon) \right) \leq 0$ при фиксированном $\varepsilon$. Отсюда $\displaystyle \varlimsup\limits_{x\to\infty} \frac{\psi(x)}{x} \leq 1+\varepsilon$, а т.к. это верно для любого $x$, то $\displaystyle \varlimsup\limits_{x\to\infty} \frac{\psi(x)}{x} \leq 1$. И наоборот, меняя знак неравенства, получаем $\displaystyle \varlimsup\limits_{x\to\infty} \frac{\psi(x)}{x} \geq 1-\varepsilon$, а т.к. это верно для любого $\varepsilon$, то $\displaystyle \varlimsup\limits_{x\to\infty} \frac{\psi(x)}{x} \geq 1$. \end{pf} \begin{pf} (Теоремы \ref{l4_thm7}).\\ Положим $\displaystyle F_T(s) = \int_1^T \frac{\psi(x)-x}{x^{1+s}}dx, \, T>1$. Поскольку $\displaystyle \int_n^{n+1}\frac{dx}{x^s}$ — целая функция, т.к. $\psi(x)$ на отрезке $[n, n+1]$ постоянна, то $\displaystyle \int_1^T \frac{\psi(x)-x}{x^{1+s}}dx$ является суммой целых функций вида $\displaystyle \int_n^{n+1}\frac{dx}{x^s} \ \Rightarrow F_T(s)$ — целая. Нужно показать, что $F_T(1) \to F(1)$ при $T \to \infty$. По определению предела, возьмём $\varepsilon > 0$ и рассмотрим следующий интеграл $$I(T) = \frac{1}{2\pi i}\int_\Gamma \left( F(1+s)-F_T(1+s) \right)T^s\left(\frac{s}{R^2}+\frac{1}{s}\right)ds, \ R = \frac{1}{\varepsilon}.$$ $F(s)$ аналитична в $\Re(s) \geq 1 \ \Rightarrow \ F(1+s)$ аналитична в $\Re(s) \geq 0$. То есть, она аналитична на отрезке $[-iR, iR]$ \textit{(см. Рис. 1)}. Если $F(s)$ аналитична в точке, то она аналитична в некоторой окрестности этой точки. Применяя это к каждой точке нашего отрезка, получаем его покрытие открытыми кругами и выделяем конечное подпокрытие по компактности $[-iR, iR]$. Теперь выбираем $h$ так, чтобы прямоугольник был внутри объединения кругов, т.е. чтобы $F(1+s)$ была аналитична на нарисованном контуре $(h=h(\varepsilon))$. \begin{center} \includegraphics[scale=0.7]{Lecture04_Gamma}~\\ (Рис. 1) \end{center} Значит, в $I(T): \ F(1+s)$ — аналитична в области (по построению), $F_T(1+s)$ — везде целая, $T^s$ — целая (экспонента), $\displaystyle \frac{s}{R^2}$ — целая, $\displaystyle \frac{1}{s}$ — полюс порядка $1$ в нуле. Следовательно, по теореме Коши о вычетах $$I(T) = \left(F(1)-F_T(1)\right)T^0 = F(1)-F_T(1).$$ \end{pf} \begin{lemma} \label{l4_lm13}~\\ При $\displaystyle \sigma = \Re(s)>0: \quad \lvert F(1+s)-F_T(1+s) \rvert \leq A\frac{T^{-\sigma}}{\sigma}$;\\ при $\displaystyle \sigma = \Re(s)<0: \quad \lvert F_T(1+s) \rvert \leq A\frac{T^{-\sigma}}{-\sigma}$,\\ где $A$ такое, что $\displaystyle \left| \frac{\psi(x)}{x}-1 \right| \leq A$ при $x \geq 1$. \end{lemma} \begin{pf}~\\ $\displaystyle \sigma>0: \quad \left| F(1+s)-F_T(1+s) \right| = \left| \int_T^{+\infty}\frac{\psi(x)-x}{x^{2+s}}dx \right| \leq A\int_T^{+\infty}\frac{dx}{x^{1+\sigma}} = A\frac{T^{-\sigma}}{\sigma}$;\\ $\displaystyle \sigma<0: \quad \left| F_T(1+s) \right| = \left| \int_1^T\frac{\psi(x)-x}{x^{2+s}}dx \right| \leq A\int_1^T\frac{dx}{x^{1+\sigma}} = A\frac{T^{-\sigma}}{-\sigma}$. \end{pf}0 \chapter{Security} \label{ch_security} The \textit{libsafecrypto} library must adhere to the Secure Coding Guideliness described in \cite{safecrypto_secure_coding_guidelines}. \textbf{\textit{TODO - Summarise the Secure Coding Guidelines here.}} % Color schemes as obtained from http://colorbrewer2.org/ % % Apache-Style Software License for ColorBrewer software and ColorBrewer Color % Schemes % % Copyright (c) 2002 , , and The Pennsylvania State % University. % % Licensed under the Apache License, Version 2.0 (the "License"); you may not % use this file except in compliance with the License. You may obtain a copy of % the License at % % http://www.apache.org/licenses/LICENSE-2.0 % % Unless required by applicable law or agreed to in writing, software % distributed under the License is distributed on an "AS IS" BASIS, WITHOUT % WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the % License for the specific language governing permissions and limitations under % the License. % % Redistribution and use in source and binary forms, with or without % modification, are permitted provided that the following conditions are met: % 1. Redistributions as source code must retain the above copyright notice, this % list of conditions and the following disclaimer. % 2. The end-user documentation included with the redistribution, if any, must % include the following acknowledgment: This product includes color % specifications and designs developed by Cynthia Brewer % (http://colorbrewer.org/). Alternately, this acknowledgment may appear in the % software itself, if and wherever such third-party acknowledgments normally % appear. % 3. The name "ColorBrewer" must not be used to endorse or promote products % derived from this software without prior written permission. For written % permission, please contact at . % 4. Products derived from this software may not be called "ColorBrewer", nor % may "ColorBrewer" appear in their name, without prior written permission of % Cynthia Brewer. % ============================================================================= % sequential color schemes % ============================================================================= % Multi hue \pgfutil@definecolor{BuGn-A}{RGB}{247,252,253} \pgfutil@definecolor{BuGn-B}{RGB}{237,248,251} \pgfutil@definecolor{BuGn-C}{RGB}{229,245,249} \pgfutil@definecolor{BuGn-D}{RGB}{204,236,230} \pgfutil@definecolor{BuGn-E}{RGB}{178,226,226} \pgfutil@definecolor{BuGn-F}{RGB}{153,216,201} \pgfutil@definecolor{BuGn-G}{RGB}{102,194,164} \pgfutil@definecolor{BuGn-H}{RGB}{65,174,118} \pgfutil@definecolor{BuGn-I}{RGB}{44,162,95} \pgfutil@definecolor{BuGn-J}{RGB}{35,139,69} \pgfutil@definecolor{BuGn-K}{RGB}{0,109,44} \pgfutil@definecolor{BuGn-L}{RGB}{0,88,36} \pgfutil@definecolor{BuGn-M}{RGB}{0,68,27} \pgfutil@definecolor{BuPu-A}{RGB}{247,252,253} \pgfutil@definecolor{BuPu-B}{RGB}{237,248,251} \pgfutil@definecolor{BuPu-C}{RGB}{224,236,244} \pgfutil@definecolor{BuPu-D}{RGB}{191,211,230} \pgfutil@definecolor{BuPu-E}{RGB}{179,205,227} \pgfutil@definecolor{BuPu-F}{RGB}{158,188,218} \pgfutil@definecolor{BuPu-G}{RGB}{140,150,198} \pgfutil@definecolor{BuPu-H}{RGB}{140,107,177} \pgfutil@definecolor{BuPu-I}{RGB}{136,86,167} \pgfutil@definecolor{BuPu-J}{RGB}{136,65,157} \pgfutil@definecolor{BuPu-K}{RGB}{129,15,124} \pgfutil@definecolor{BuPu-L}{RGB}{110,1,107} \pgfutil@definecolor{BuPu-M}{RGB}{77,0,75} \pgfutil@definecolor{GnBu-A}{RGB}{247,252,240} \pgfutil@definecolor{GnBu-B}{RGB}{240,249,232} \pgfutil@definecolor{GnBu-C}{RGB}{224,243,219} \pgfutil@definecolor{GnBu-D}{RGB}{204,235,197} \pgfutil@definecolor{GnBu-E}{RGB}{186,228,188} \pgfutil@definecolor{GnBu-F}{RGB}{168,221,181} \pgfutil@definecolor{GnBu-G}{RGB}{123,204,196} \pgfutil@definecolor{GnBu-H}{RGB}{78,179,211} \pgfutil@definecolor{GnBu-I}{RGB}{67,162,202} \pgfutil@definecolor{GnBu-J}{RGB}{43,140,190} \pgfutil@definecolor{GnBu-K}{RGB}{8,104,172} \pgfutil@definecolor{GnBu-L}{RGB}{8,88,158} \pgfutil@definecolor{GnBu-M}{RGB}{8,64,129} \pgfutil@definecolor{OrRd-A}{RGB}{255,247,236} \pgfutil@definecolor{OrRd-B}{RGB}{254,240,217} \pgfutil@definecolor{OrRd-C}{RGB}{254,232,200} \pgfutil@definecolor{OrRd-D}{RGB}{253,212,158} \pgfutil@definecolor{OrRd-E}{RGB}{253,204,138} \pgfutil@definecolor{OrRd-F}{RGB}{253,187,132} \pgfutil@definecolor{OrRd-G}{RGB}{252,141,89} \pgfutil@definecolor{OrRd-H}{RGB}{239,101,72} \pgfutil@definecolor{OrRd-I}{RGB}{227,74,51} \pgfutil@definecolor{OrRd-J}{RGB}{215,48,31} \pgfutil@definecolor{OrRd-K}{RGB}{179,0,0} \pgfutil@definecolor{OrRd-L}{RGB}{153,0,0} \pgfutil@definecolor{OrRd-M}{RGB}{127,0,0} \pgfutil@definecolor{PuBu-A}{RGB}{255,247,251} \pgfutil@definecolor{PuBu-B}{RGB}{241,238,246} \pgfutil@definecolor{PuBu-C}{RGB}{236,231,242} \pgfutil@definecolor{PuBu-D}{RGB}{208,209,230} \pgfutil@definecolor{PuBu-E}{RGB}{189,201,225} \pgfutil@definecolor{PuBu-F}{RGB}{166,189,219} \pgfutil@definecolor{PuBu-G}{RGB}{116,169,207} \pgfutil@definecolor{PuBu-H}{RGB}{54,144,192} \pgfutil@definecolor{PuBu-I}{RGB}{43,140,190} \pgfutil@definecolor{PuBu-J}{RGB}{5,112,176} \pgfutil@definecolor{PuBu-K}{RGB}{4,90,141} \pgfutil@definecolor{PuBu-L}{RGB}{3,78,123} \pgfutil@definecolor{PuBu-M}{RGB}{2,56,88} \pgfutil@definecolor{PuBuGn-A}{RGB}{255,247,251} \pgfutil@definecolor{PuBuGn-B}{RGB}{246,239,247} \pgfutil@definecolor{PuBuGn-C}{RGB}{236,226,240} \pgfutil@definecolor{PuBuGn-D}{RGB}{208,209,230} \pgfutil@definecolor{PuBuGn-E}{RGB}{189,201,225} \pgfutil@definecolor{PuBuGn-F}{RGB}{166,189,219} \pgfutil@definecolor{PuBuGn-G}{RGB}{103,169,207} \pgfutil@definecolor{PuBuGn-H}{RGB}{54,144,192} \pgfutil@definecolor{PuBuGn-I}{RGB}{28,144,153} \pgfutil@definecolor{PuBuGn-J}{RGB}{2,129,138} \pgfutil@definecolor{PuBuGn-K}{RGB}{1,108,89} \pgfutil@definecolor{PuBuGn-L}{RGB}{1,100,80} \pgfutil@definecolor{PuBuGn-M}{RGB}{1,70,54} \pgfutil@definecolor{PuRd-A}{RGB}{247,244,249} \pgfutil@definecolor{PuRd-B}{RGB}{241,238,246} \pgfutil@definecolor{PuRd-C}{RGB}{231,225,239} \pgfutil@definecolor{PuRd-D}{RGB}{212,185,218} \pgfutil@definecolor{PuRd-E}{RGB}{215,181,216} \pgfutil@definecolor{PuRd-F}{RGB}{201,148,199} \pgfutil@definecolor{PuRd-G}{RGB}{223,101,176} \pgfutil@definecolor{PuRd-H}{RGB}{231,41,138} \pgfutil@definecolor{PuRd-I}{RGB}{221,28,119} \pgfutil@definecolor{PuRd-J}{RGB}{206,18,86} \pgfutil@definecolor{PuRd-K}{RGB}{152,0,67} \pgfutil@definecolor{PuRd-L}{RGB}{145,0,63} \pgfutil@definecolor{PuRd-M}{RGB}{103,0,31} \pgfutil@definecolor{RdPu-A}{RGB}{255,247,243} \pgfutil@definecolor{RdPu-B}{RGB}{254,235,226} \pgfutil@definecolor{RdPu-C}{RGB}{253,224,221} \pgfutil@definecolor{RdPu-D}{RGB}{252,197,192} \pgfutil@definecolor{RdPu-E}{RGB}{251,180,185} \pgfutil@definecolor{RdPu-F}{RGB}{250,159,181} \pgfutil@definecolor{RdPu-G}{RGB}{247,104,161} \pgfutil@definecolor{RdPu-H}{RGB}{221,52,151} \pgfutil@definecolor{RdPu-I}{RGB}{197,27,138} \pgfutil@definecolor{RdPu-J}{RGB}{174,1,126} \pgfutil@definecolor{RdPu-K}{RGB}{122,1,119} \pgfutil@definecolor{RdPu-L}{RGB}{122,1,119} \pgfutil@definecolor{RdPu-M}{RGB}{73,0,106} \pgfutil@definecolor{YlGn-A}{RGB}{255,255,229} \pgfutil@definecolor{YlGn-B}{RGB}{255,255,204} \pgfutil@definecolor{YlGn-C}{RGB}{247,252,185} \pgfutil@definecolor{YlGn-D}{RGB}{217,240,163} \pgfutil@definecolor{YlGn-E}{RGB}{194,230,153} \pgfutil@definecolor{YlGn-F}{RGB}{173,221,142} \pgfutil@definecolor{YlGn-G}{RGB}{120,198,121} \pgfutil@definecolor{YlGn-H}{RGB}{65,171,93} \pgfutil@definecolor{YlGn-I}{RGB}{49,163,84} \pgfutil@definecolor{YlGn-J}{RGB}{35,132,67} \pgfutil@definecolor{YlGn-K}{RGB}{0,104,55} \pgfutil@definecolor{YlGn-L}{RGB}{0,90,50} \pgfutil@definecolor{YlGn-M}{RGB}{0,69,41} \pgfutil@definecolor{YlGnBu-A}{RGB}{255,255,217} \pgfutil@definecolor{YlGnBu-B}{RGB}{255,255,204} \pgfutil@definecolor{YlGnBu-C}{RGB}{237,248,177} \pgfutil@definecolor{YlGnBu-D}{RGB}{199,233,180} \pgfutil@definecolor{YlGnBu-E}{RGB}{161,218,180} \pgfutil@definecolor{YlGnBu-F}{RGB}{127,205,187} \pgfutil@definecolor{YlGnBu-G}{RGB}{65,182,196} \pgfutil@definecolor{YlGnBu-H}{RGB}{29,145,192} \pgfutil@definecolor{YlGnBu-I}{RGB}{44,127,184} \pgfutil@definecolor{YlGnBu-J}{RGB}{34,94,168} \pgfutil@definecolor{YlGnBu-K}{RGB}{37,52,148} \pgfutil@definecolor{YlGnBu-L}{RGB}{12,44,132} \pgfutil@definecolor{YlGnBu-M}{RGB}{8,29,88} \pgfutil@definecolor{YlOrBr-A}{RGB}{255,255,229} \pgfutil@definecolor{YlOrBr-B}{RGB}{255,255,212} \pgfutil@definecolor{YlOrBr-C}{RGB}{255,247,188} \pgfutil@definecolor{YlOrBr-D}{RGB}{254,227,145} \pgfutil@definecolor{YlOrBr-E}{RGB}{254,217,142} \pgfutil@definecolor{YlOrBr-F}{RGB}{254,196,79} \pgfutil@definecolor{YlOrBr-G}{RGB}{254,153,41} \pgfutil@definecolor{YlOrBr-H}{RGB}{236,112,20} \pgfutil@definecolor{YlOrBr-I}{RGB}{217,95,14} \pgfutil@definecolor{YlOrBr-J}{RGB}{204,76,2} \pgfutil@definecolor{YlOrBr-K}{RGB}{153,52,4} \pgfutil@definecolor{YlOrBr-L}{RGB}{140,45,4} \pgfutil@definecolor{YlOrBr-M}{RGB}{102,37,6} \pgfutil@definecolor{YlOrRd-A}{RGB}{255,255,204} \pgfutil@definecolor{YlOrRd-B}{RGB}{255,255,178} \pgfutil@definecolor{YlOrRd-C}{RGB}{255,237,160} \pgfutil@definecolor{YlOrRd-D}{RGB}{254,217,118} \pgfutil@definecolor{YlOrRd-E}{RGB}{254,204,92} \pgfutil@definecolor{YlOrRd-F}{RGB}{254,178,76} \pgfutil@definecolor{YlOrRd-G}{RGB}{253,141,60} \pgfutil@definecolor{YlOrRd-H}{RGB}{252,78,42} \pgfutil@definecolor{YlOrRd-I}{RGB}{240,59,32} \pgfutil@definecolor{YlOrRd-J}{RGB}{227,26,28} \pgfutil@definecolor{YlOrRd-K}{RGB}{189,0,38} \pgfutil@definecolor{YlOrRd-L}{RGB}{177,0,38} \pgfutil@definecolor{YlOrRd-M}{RGB}{128,0,38} % ----------------------------------------------------------------------------- % Single hue \pgfutil@definecolor{Blues-A}{RGB}{247,251,255} \pgfutil@definecolor{Blues-B}{RGB}{239,243,255} \pgfutil@definecolor{Blues-C}{RGB}{222,235,247} \pgfutil@definecolor{Blues-D}{RGB}{198,219,239} \pgfutil@definecolor{Blues-E}{RGB}{189,215,231} \pgfutil@definecolor{Blues-F}{RGB}{158,202,225} \pgfutil@definecolor{Blues-G}{RGB}{107,174,214} \pgfutil@definecolor{Blues-H}{RGB}{66,146,198} \pgfutil@definecolor{Blues-I}{RGB}{49,130,189} \pgfutil@definecolor{Blues-J}{RGB}{33,113,181} \pgfutil@definecolor{Blues-K}{RGB}{8,81,156} \pgfutil@definecolor{Blues-L}{RGB}{8,69,148} \pgfutil@definecolor{Blues-M}{RGB}{8,48,107} \pgfutil@definecolor{Greens-A}{RGB}{247,252,245} \pgfutil@definecolor{Greens-B}{RGB}{237,248,233} \pgfutil@definecolor{Greens-C}{RGB}{229,245,224} \pgfutil@definecolor{Greens-D}{RGB}{199,233,192} \pgfutil@definecolor{Greens-E}{RGB}{186,228,179} \pgfutil@definecolor{Greens-F}{RGB}{161,217,155} \pgfutil@definecolor{Greens-G}{RGB}{116,196,118} \pgfutil@definecolor{Greens-H}{RGB}{65,171,93} \pgfutil@definecolor{Greens-I}{RGB}{49,163,84} \pgfutil@definecolor{Greens-J}{RGB}{35,139,69} \pgfutil@definecolor{Greens-K}{RGB}{0,109,44} \pgfutil@definecolor{Greens-L}{RGB}{0,90,50} \pgfutil@definecolor{Greens-M}{RGB}{0,68,27} \pgfutil@definecolor{Greys-A}{RGB}{255,255,255} \pgfutil@definecolor{Greys-B}{RGB}{247,247,247} \pgfutil@definecolor{Greys-C}{RGB}{240,240,240} \pgfutil@definecolor{Greys-D}{RGB}{217,217,217} \pgfutil@definecolor{Greys-E}{RGB}{204,204,204} \pgfutil@definecolor{Greys-F}{RGB}{189,189,189} \pgfutil@definecolor{Greys-G}{RGB}{150,150,150} \pgfutil@definecolor{Greys-H}{RGB}{115,115,115} \pgfutil@definecolor{Greys-I}{RGB}{99,99,99} \pgfutil@definecolor{Greys-J}{RGB}{82,82,82} \pgfutil@definecolor{Greys-K}{RGB}{37,37,37} \pgfutil@definecolor{Greys-L}{RGB}{37,37,37} \pgfutil@definecolor{Greys-M}{RGB}{0,0,0} \pgfutil@definecolor{Oranges-A}{RGB}{255,245,235} \pgfutil@definecolor{Oranges-B}{RGB}{254,237,222} \pgfutil@definecolor{Oranges-C}{RGB}{254,230,206} \pgfutil@definecolor{Oranges-D}{RGB}{253,208,162} \pgfutil@definecolor{Oranges-E}{RGB}{253,190,133} \pgfutil@definecolor{Oranges-F}{RGB}{253,174,107} \pgfutil@definecolor{Oranges-G}{RGB}{253,141,60} \pgfutil@definecolor{Oranges-H}{RGB}{241,105,19} \pgfutil@definecolor{Oranges-I}{RGB}{230,85,13} \pgfutil@definecolor{Oranges-J}{RGB}{217,71,1} \pgfutil@definecolor{Oranges-K}{RGB}{166,54,3} \pgfutil@definecolor{Oranges-L}{RGB}{140,45,4} \pgfutil@definecolor{Oranges-M}{RGB}{127,39,4} \pgfutil@definecolor{Purples-A}{RGB}{252,251,253} \pgfutil@definecolor{Purples-B}{RGB}{242,240,247} \pgfutil@definecolor{Purples-C}{RGB}{239,237,245} \pgfutil@definecolor{Purples-D}{RGB}{218,218,235} \pgfutil@definecolor{Purples-E}{RGB}{203,201,226} \pgfutil@definecolor{Purples-F}{RGB}{188,189,220} \pgfutil@definecolor{Purples-G}{RGB}{158,154,200} \pgfutil@definecolor{Purples-H}{RGB}{128,125,186} \pgfutil@definecolor{Purples-I}{RGB}{117,107,177} \pgfutil@definecolor{Purples-J}{RGB}{106,81,163} \pgfutil@definecolor{Purples-K}{RGB}{84,39,143} \pgfutil@definecolor{Purples-L}{RGB}{74,20,134} \pgfutil@definecolor{Purples-M}{RGB}{63,0,125} \pgfutil@definecolor{Reds-A}{RGB}{255,245,240} \pgfutil@definecolor{Reds-B}{RGB}{254,229,217} \pgfutil@definecolor{Reds-C}{RGB}{254,224,210} \pgfutil@definecolor{Reds-D}{RGB}{252,187,161} \pgfutil@definecolor{Reds-E}{RGB}{252,174,145} \pgfutil@definecolor{Reds-F}{RGB}{252,146,114} \pgfutil@definecolor{Reds-G}{RGB}{251,106,74} \pgfutil@definecolor{Reds-H}{RGB}{239,59,44} \pgfutil@definecolor{Reds-I}{RGB}{222,45,38} \pgfutil@definecolor{Reds-J}{RGB}{203,24,29} \pgfutil@definecolor{Reds-K}{RGB}{165,15,21} \pgfutil@definecolor{Reds-L}{RGB}{153,0,13} \pgfutil@definecolor{Reds-M}{RGB}{103,0,13} % ============================================================================= % diverging color schemes % ============================================================================= \pgfutil@definecolor{BrBG-A}{RGB}{84,48,5} \pgfutil@definecolor{BrBG-B}{RGB}{140,81,10} \pgfutil@definecolor{BrBG-C}{RGB}{166,97,26} \pgfutil@definecolor{BrBG-D}{RGB}{191,129,45} \pgfutil@definecolor{BrBG-E}{RGB}{216,179,101} \pgfutil@definecolor{BrBG-F}{RGB}{223,194,125} \pgfutil@definecolor{BrBG-G}{RGB}{246,232,195} \pgfutil@definecolor{BrBG-H}{RGB}{245,245,245} \pgfutil@definecolor{BrBG-I}{RGB}{199,234,229} \pgfutil@definecolor{BrBG-J}{RGB}{128,205,193} \pgfutil@definecolor{BrBG-K}{RGB}{90,180,172} \pgfutil@definecolor{BrBG-L}{RGB}{53,151,143} \pgfutil@definecolor{BrBG-M}{RGB}{1,133,113} \pgfutil@definecolor{BrBG-N}{RGB}{1,102,94} \pgfutil@definecolor{BrBG-O}{RGB}{0,60,48} \pgfutil@definecolor{PiYG-A}{RGB}{142,1,82} \pgfutil@definecolor{PiYG-B}{RGB}{197,27,125} \pgfutil@definecolor{PiYG-C}{RGB}{208,28,139} \pgfutil@definecolor{PiYG-D}{RGB}{222,119,174} \pgfutil@definecolor{PiYG-E}{RGB}{233,163,201} \pgfutil@definecolor{PiYG-F}{RGB}{241,182,218} \pgfutil@definecolor{PiYG-G}{RGB}{253,224,239} \pgfutil@definecolor{PiYG-H}{RGB}{247,247,247} \pgfutil@definecolor{PiYG-I}{RGB}{230,245,208} \pgfutil@definecolor{PiYG-J}{RGB}{184,225,134} \pgfutil@definecolor{PiYG-K}{RGB}{161,215,106} \pgfutil@definecolor{PiYG-L}{RGB}{127,188,65} \pgfutil@definecolor{PiYG-M}{RGB}{77,172,38} \pgfutil@definecolor{PiYG-N}{RGB}{77,146,33} \pgfutil@definecolor{PiYG-O}{RGB}{39,100,25} \pgfutil@definecolor{PRGn-A}{RGB}{64,0,75} \pgfutil@definecolor{PRGn-B}{RGB}{118,42,131} \pgfutil@definecolor{PRGn-C}{RGB}{123,50,148} \pgfutil@definecolor{PRGn-D}{RGB}{153,112,171} \pgfutil@definecolor{PRGn-E}{RGB}{175,141,195} \pgfutil@definecolor{PRGn-F}{RGB}{194,165,207} \pgfutil@definecolor{PRGn-G}{RGB}{231,212,232} \pgfutil@definecolor{PRGn-H}{RGB}{247,247,247} \pgfutil@definecolor{PRGn-I}{RGB}{217,240,211} \pgfutil@definecolor{PRGn-J}{RGB}{166,219,160} \pgfutil@definecolor{PRGn-K}{RGB}{127,191,123} \pgfutil@definecolor{PRGn-L}{RGB}{90,174,97} \pgfutil@definecolor{PRGn-M}{RGB}{0,136,55} \pgfutil@definecolor{PRGn-N}{RGB}{27,120,55} \pgfutil@definecolor{PRGn-O}{RGB}{0,68,27} \pgfutil@definecolor{PuOr-A}{RGB}{127,59,8} \pgfutil@definecolor{PuOr-B}{RGB}{179,88,6} \pgfutil@definecolor{PuOr-C}{RGB}{230,97,1} \pgfutil@definecolor{PuOr-D}{RGB}{224,130,20} \pgfutil@definecolor{PuOr-E}{RGB}{241,163,64} \pgfutil@definecolor{PuOr-F}{RGB}{253,184,99} \pgfutil@definecolor{PuOr-G}{RGB}{254,224,182} \pgfutil@definecolor{PuOr-H}{RGB}{247,247,247} \pgfutil@definecolor{PuOr-I}{RGB}{216,218,235} \pgfutil@definecolor{PuOr-J}{RGB}{178,171,210} \pgfutil@definecolor{PuOr-K}{RGB}{153,142,195} \pgfutil@definecolor{PuOr-L}{RGB}{128,115,172} \pgfutil@definecolor{PuOr-M}{RGB}{94,60,153} \pgfutil@definecolor{PuOr-N}{RGB}{84,39,136} \pgfutil@definecolor{PuOr-O}{RGB}{45,0,75} \pgfutil@definecolor{RdBu-A}{RGB}{103,0,31} \pgfutil@definecolor{RdBu-B}{RGB}{178,24,43} \pgfutil@definecolor{RdBu-C}{RGB}{202,0,32} \pgfutil@definecolor{RdBu-D}{RGB}{214,96,77} \pgfutil@definecolor{RdBu-E}{RGB}{239,138,98} \pgfutil@definecolor{RdBu-F}{RGB}{244,165,130} \pgfutil@definecolor{RdBu-G}{RGB}{253,219,199} \pgfutil@definecolor{RdBu-H}{RGB}{247,247,247} \pgfutil@definecolor{RdBu-I}{RGB}{209,229,240} \pgfutil@definecolor{RdBu-J}{RGB}{146,197,222} \pgfutil@definecolor{RdBu-K}{RGB}{103,169,207} \pgfutil@definecolor{RdBu-L}{RGB}{67,147,195} \pgfutil@definecolor{RdBu-M}{RGB}{5,113,176} \pgfutil@definecolor{RdBu-N}{RGB}{33,102,172} \pgfutil@definecolor{RdBu-O}{RGB}{5,48,97} \pgfutil@definecolor{RdGy-A}{RGB}{103,0,31} \pgfutil@definecolor{RdGy-B}{RGB}{178,24,43} \pgfutil@definecolor{RdGy-C}{RGB}{202,0,32} \pgfutil@definecolor{RdGy-D}{RGB}{214,96,77} \pgfutil@definecolor{RdGy-E}{RGB}{239,138,98} \pgfutil@definecolor{RdGy-F}{RGB}{244,165,130} \pgfutil@definecolor{RdGy-G}{RGB}{253,219,199} \pgfutil@definecolor{RdGy-H}{RGB}{255,255,255} \pgfutil@definecolor{RdGy-I}{RGB}{224,224,224} \pgfutil@definecolor{RdGy-J}{RGB}{186,186,186} \pgfutil@definecolor{RdGy-K}{RGB}{153,153,153} \pgfutil@definecolor{RdGy-L}{RGB}{135,135,135} \pgfutil@definecolor{RdGy-M}{RGB}{64,64,64} \pgfutil@definecolor{RdGy-N}{RGB}{77,77,77} \pgfutil@definecolor{RdGy-O}{RGB}{26,26,26} \pgfutil@definecolor{RdYlBu-A}{RGB}{165,0,38} \pgfutil@definecolor{RdYlBu-B}{RGB}{215,48,39} \pgfutil@definecolor{RdYlBu-C}{RGB}{215,25,28} \pgfutil@definecolor{RdYlBu-D}{RGB}{244,109,67} \pgfutil@definecolor{RdYlBu-E}{RGB}{252,141,89} \pgfutil@definecolor{RdYlBu-F}{RGB}{253,174,97} \pgfutil@definecolor{RdYlBu-G}{RGB}{254,224,144} \pgfutil@definecolor{RdYlBu-H}{RGB}{255,255,191} \pgfutil@definecolor{RdYlBu-I}{RGB}{224,243,248} \pgfutil@definecolor{RdYlBu-J}{RGB}{171,217,233} \pgfutil@definecolor{RdYlBu-K}{RGB}{145,191,219} \pgfutil@definecolor{RdYlBu-L}{RGB}{116,173,209} \pgfutil@definecolor{RdYlBu-M}{RGB}{44,123,182} \pgfutil@definecolor{RdYlBu-N}{RGB}{69,117,180} \pgfutil@definecolor{RdYlBu-O}{RGB}{49,54,149} \pgfutil@definecolor{RdYlGn-A}{RGB}{165,0,38} \pgfutil@definecolor{RdYlGn-B}{RGB}{215,48,39} \pgfutil@definecolor{RdYlGn-C}{RGB}{215,25,28} \pgfutil@definecolor{RdYlGn-D}{RGB}{244,109,67} \pgfutil@definecolor{RdYlGn-E}{RGB}{252,141,89} \pgfutil@definecolor{RdYlGn-F}{RGB}{253,174,97} \pgfutil@definecolor{RdYlGn-G}{RGB}{254,224,139} \pgfutil@definecolor{RdYlGn-H}{RGB}{255,255,191} \pgfutil@definecolor{RdYlGn-I}{RGB}{217,239,139} \pgfutil@definecolor{RdYlGn-J}{RGB}{166,217,106} \pgfutil@definecolor{RdYlGn-K}{RGB}{145,207,96} \pgfutil@definecolor{RdYlGn-L}{RGB}{102,189,99} \pgfutil@definecolor{RdYlGn-M}{RGB}{26,150,65} \pgfutil@definecolor{RdYlGn-N}{RGB}{26,152,80} \pgfutil@definecolor{RdYlGn-O}{RGB}{0,104,55} \pgfutil@definecolor{Spectral-A}{RGB}{158,1,66} \pgfutil@definecolor{Spectral-B}{RGB}{213,62,79} \pgfutil@definecolor{Spectral-C}{RGB}{215,25,28} \pgfutil@definecolor{Spectral-D}{RGB}{244,109,67} \pgfutil@definecolor{Spectral-E}{RGB}{252,141,89} \pgfutil@definecolor{Spectral-F}{RGB}{253,174,97} \pgfutil@definecolor{Spectral-G}{RGB}{254,224,139} \pgfutil@definecolor{Spectral-H}{RGB}{255,255,191} \pgfutil@definecolor{Spectral-I}{RGB}{230,245,152} \pgfutil@definecolor{Spectral-J}{RGB}{171,221,164} \pgfutil@definecolor{Spectral-K}{RGB}{153,213,148} \pgfutil@definecolor{Spectral-L}{RGB}{102,194,165} \pgfutil@definecolor{Spectral-M}{RGB}{43,131,186} \pgfutil@definecolor{Spectral-N}{RGB}{50,136,189} \pgfutil@definecolor{Spectral-O}{RGB}{94,79,162} % ============================================================================= % qualitative color schemes % ============================================================================= \pgfutil@definecolor{Accent-A}{RGB}{127,201,127} \pgfutil@definecolor{Accent-B}{RGB}{190,174,212} \pgfutil@definecolor{Accent-C}{RGB}{253,192,134} \pgfutil@definecolor{Accent-D}{RGB}{255,255,153} \pgfutil@definecolor{Accent-E}{RGB}{56,108,176} \pgfutil@definecolor{Accent-F}{RGB}{240,2,127} \pgfutil@definecolor{Accent-G}{RGB}{191,91,23} \pgfutil@definecolor{Accent-H}{RGB}{102,102,102} \pgfutil@definecolor{Dark2-A}{RGB}{27,158,119} \pgfutil@definecolor{Dark2-B}{RGB}{217,95,2} \pgfutil@definecolor{Dark2-C}{RGB}{117,112,179} \pgfutil@definecolor{Dark2-D}{RGB}{231,41,138} \pgfutil@definecolor{Dark2-E}{RGB}{102,166,30} \pgfutil@definecolor{Dark2-F}{RGB}{230,171,2} \pgfutil@definecolor{Dark2-G}{RGB}{166,118,29} \pgfutil@definecolor{Dark2-H}{RGB}{102,102,102} \pgfutil@definecolor{Paired-A}{RGB}{166,206,227} \pgfutil@definecolor{Paired-B}{RGB}{31,120,180} \pgfutil@definecolor{Paired-C}{RGB}{178,223,138} \pgfutil@definecolor{Paired-D}{RGB}{51,160,44} \pgfutil@definecolor{Paired-E}{RGB}{251,154,153} \pgfutil@definecolor{Paired-F}{RGB}{227,26,28} \pgfutil@definecolor{Paired-G}{RGB}{253,191,111} \pgfutil@definecolor{Paired-H}{RGB}{255,127,0} \pgfutil@definecolor{Paired-I}{RGB}{202,178,214} \pgfutil@definecolor{Paired-J}{RGB}{106,61,154} \pgfutil@definecolor{Paired-K}{RGB}{255,255,153} \pgfutil@definecolor{Paired-L}{RGB}{177,89,40} \pgfutil@definecolor{Pastel1-A}{RGB}{251,180,174} \pgfutil@definecolor{Pastel1-B}{RGB}{179,205,227} \pgfutil@definecolor{Pastel1-C}{RGB}{204,235,197} \pgfutil@definecolor{Pastel1-D}{RGB}{222,203,228} \pgfutil@definecolor{Pastel1-E}{RGB}{254,217,166} \pgfutil@definecolor{Pastel1-F}{RGB}{255,255,204} \pgfutil@definecolor{Pastel1-G}{RGB}{229,216,189} \pgfutil@definecolor{Pastel1-H}{RGB}{253,218,236} \pgfutil@definecolor{Pastel1-I}{RGB}{242,242,242} \pgfutil@definecolor{Pastel2-A}{RGB}{179,226,205} \pgfutil@definecolor{Pastel2-B}{RGB}{253,205,172} \pgfutil@definecolor{Pastel2-C}{RGB}{203,213,232} \pgfutil@definecolor{Pastel2-D}{RGB}{244,202,228} \pgfutil@definecolor{Pastel2-E}{RGB}{230,245,201} \pgfutil@definecolor{Pastel2-F}{RGB}{255,242,174} \pgfutil@definecolor{Pastel2-G}{RGB}{241,226,204} \pgfutil@definecolor{Pastel2-H}{RGB}{204,204,204} \pgfutil@definecolor{Set1-A}{RGB}{228,26,28} \pgfutil@definecolor{Set1-B}{RGB}{55,126,184} \pgfutil@definecolor{Set1-C}{RGB}{77,175,74} \pgfutil@definecolor{Set1-D}{RGB}{152,78,163} \pgfutil@definecolor{Set1-E}{RGB}{255,127,0} \pgfutil@definecolor{Set1-F}{RGB}{255,255,51} \pgfutil@definecolor{Set1-G}{RGB}{166,86,40} \pgfutil@definecolor{Set1-H}{RGB}{247,129,191} \pgfutil@definecolor{Set1-I}{RGB}{153,153,153} \pgfutil@definecolor{Set2-A}{RGB}{102,194,165} \pgfutil@definecolor{Set2-B}{RGB}{252,141,98} \pgfutil@definecolor{Set2-C}{RGB}{141,160,203} \pgfutil@definecolor{Set2-D}{RGB}{231,138,195} \pgfutil@definecolor{Set2-E}{RGB}{166,216,84} \pgfutil@definecolor{Set2-F}{RGB}{255,217,47} \pgfutil@definecolor{Set2-G}{RGB}{229,196,148} \pgfutil@definecolor{Set2-H}{RGB}{179,179,179} \pgfutil@definecolor{Set3-A}{RGB}{141,211,199} \pgfutil@definecolor{Set3-B}{RGB}{255,255,179} \pgfutil@definecolor{Set3-C}{RGB}{190,186,218} \pgfutil@definecolor{Set3-D}{RGB}{251,128,114} \pgfutil@definecolor{Set3-E}{RGB}{128,177,211} \pgfutil@definecolor{Set3-F}{RGB}{253,180,98} \pgfutil@definecolor{Set3-G}{RGB}{179,222,105} \pgfutil@definecolor{Set3-H}{RGB}{252,205,229} \pgfutil@definecolor{Set3-I}{RGB}{217,217,217} \pgfutil@definecolor{Set3-J}{RGB}{188,128,189} \pgfutil@definecolor{Set3-K}{RGB}{204,235,197} \pgfutil@definecolor{Set3-L}{RGB}{255,237,111} document/Source/Chapters/chapter5.tex \chapter{Results} In this chapter we examine the rendered output results of our implementation of our BRDF models applied to different input patches such as the Blaze grating and the Elaphe $\ref{fig:elpahespecies}$ and Xenopeltis $\ref{fig:xenospeicies}$ snake skins nanostructures. We are discussing and comparing both, their BRDF maps $\ref{fig:brdfmapexplanation}$ and the corresponding renderings on a snake geometry as shown in section $\ref{sec:snakegeomrenderings}$, for various input parameters. Last we also show a real experimental image showing the effect of diffraction for setting readers perception about expected results. \section{BRDF maps} A BRDF map shows a shader's output for all possible viewing directions for a given fixed, incident light direction. We assume that each viewing direction is expressed in spherical coordinates (See appendix $\ref{sec:sphericalcoordinates}$) $(\theta_v, \phi_v)$ and is represented in the map at a point \begin{align} (x,y) = (sin(\theta_v)cos(\phi_v), sin(\theta_v)sin(\phi_v)) \end{align} with its origin at the center of the map. The light direction for normal incidence $(\theta_i, \phi_i)$ has been fixed to $(0,0)$ for our rendered results. \begin{figure}[H] \centering \subfigure[BRDF map schema]{ \includegraphics[scale=0.3]{resultsnew/brdfmapschema.png} \label{fig:brdfmapschema} } ~ \subfigure[Light reflection geometrical setup]{ \includegraphics[scale=1.05]{results/Lightreflectiongeometry.png} \label{fig:lightreflectiongeometry} } ~ \caption[BRDF Map]{BRDF maps$\footnotemark$ for different patches: $\Theta=(\theta_i,\phi_i)$ is the direction of light propagation} \label{fig:brdfmapexplanation} \end{figure} \footnotetext{image source of figure: \begin{itemize} \item \ref{fig:brdfmapschema}: Taken from D.S.Dhillon et al's Paper $\cite{daljitpaper}$ \item \ref{fig:lightreflectiongeometry}: Taken from \texttt{http://math.nist.gov/\textasciitilde FHunt/appearance/brdf.html} \end{itemize} } \begin{figure}[H] \centering \subfigure[Blaze grating with scale of 2.500 $\mu m$]{ \includegraphics[scale=0.5]{evaluation/blaze_res.png} \label{fig:blazegratingpatch} } ~ \subfigure[Elaphe patch with scale of 3.270 $\mu m$]{ \includegraphics[scale=0.5]{evaluation/elaphe_res.png} \label{fig:elpahegratingpatch} } ~ \subfigure[Xenopeltis patch with scale of 3.210 $\mu m$]{ \includegraphics[scale=0.5]{evaluation/xeno_res.png} \label{fig:xenogratingpatch} } \caption[Our Diffraction Gratings]{Cutouts of our nano-scaled surface gratings used for rendering with actual our shader with a scale indicator (red line) for each patch. Note that for actual rendering, we use larger patches.} \label{fig:gratingpatches} \end{figure} Figure $\ref{fig:brdfmapsdiffpatches}$ shows the BRDF maps of the full lambda space sampling approach (\textbf{FLSS}) as described in section $\ref{sec:fragmentshader}$ applied on different nanoscale surface gratings shown in figure $\ref{fig:gratingpatches}$. In subfigure $\ref{fig:brdfmapBlaze}$ we see the BRDF map for the Blazed grating, showing high relative brightness for its first order diffraction, i.e. for the Blazed gratings most of the diffracted spectral energy lies in its first order. Notice that the surface of blazed grating is forming a step structure for which the angle between the step normal and the grating normal is denoted by $\emph{blaze angle}$. Every blazed grating is manufactured in the Littrow$\footnote{For further information please see \texttt{http://en.wikipedia.org/wiki/Blazed\textunderscore grating}.}$ configuration. This means that the blaze angle is chosen such that for a chosen mode m and a chosen wavelength $\lambda$ with maximum intensity and the incidence angle are identical. Thus a blazed grating has its maximal efficiency for the chosen wavelength of the light used. Higher diffraction modes are still perceivable (second and higher diffraction orders) but with a much lower relative brightness. The asymmetry in the brightness of the pattern is due to the asymmetric geometry of the grating $\ref{fig:blazegratingpatch}$. \\ The finger-like structures contained in the Elaphe surface grating $\ref{fig:elpahegratingpatch}$ are considerably regularly aligned and hence diffraction occurs along the horizontal axis for the BRDF map as shown in figure $\ref{fig:brdfmapElaphe}$. The reason for not seeing any strong diffraction color contribution along other directions in the BRDF map is due to the fact that these ‘nano-fingers’ overlap across layers and thus do not exhibit any well-formed periodicity along finger direction. \\ For the Xenopeltis surface grating shown in figure $\ref{fig:xenogratingpatch}$, we observe diffraction along many different, almost vertical directions in the BRDF map $\ref{fig:brdfmapXeno}$ since the layers of the finger-like structures do not overlap and are shifted significantly along their length but still exhibit some local consistency. A similar argument holds true for diffraction across locally periodic finger patches with slightly different orientations. \begin{figure}[H] \centering \subfigure[Blazed grating]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_gratings/blazed.png} \label{fig:brdfmapBlaze} } ~ \subfigure[Elaphe grating]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_gratings/elaphe.png} \label{fig:brdfmapElaphe} } ~ \subfigure[Xenopeltis grating]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_gratings/xeno.png} \label{fig:brdfmapXeno} } \caption[BRDF Map: FLSS Approach applied on various Gratings]{BRDF maps for different patches} \label{fig:brdfmapsdiffpatches} \end{figure} %%TODO eplain what gound truth is refering to, what it is supposed to denote. Figure $\ref{fig:brdfmapsdiffrenderingapproaches}$ shows BRDF maps of all our BRDF models applied on the Blaze grating. Figure $\ref{fig:brdfmapblazeallLambda}$ shows the FLSS shading approach result for our blazed grating and it is used in order to compare with our other rendering approaches. \\ Figure $\ref{fig:brdfmapblazeonlyreq}$ shows the BRDF map for the NMM approach, introduced in section $\ref{sec:nmmapproach}$, which is close to the FLSS approach as verified in section $\ref{sec:approachesverifications}$ (see figure $\ref{fig:blazneval}$). Nevertheless there is a small, noticeable difference: For the NMM approach we see a white, circular spot around the map center. Nevertheless, apart from this white spot, the NMM approach resembles the FLSS approach. The reason for this differences is due to the fact that the NMM approach treats the center of a BRDF map as a special case, as described in section $\ref{sec:nmmapproach}$. Technically, every location around a small $\epsilon$-circumference from the map center gets white color assigned. \\ Figure $\ref{fig:brdfmapblazepq}$ shows the BRDF map for the PQ approach which relies on sinc-interpolation. The PQ BRDF map and the FLSS results are visual alike. In contrast to the evaluation plots in figure $\ref{fig:evaluationdiffshaderpq}$, the BRDF map for the PQ approach matches well with the one for the FLSS approach. Compared to FLSS, one difference we notice is that the first order of diffraction is a little spread for the PQ approach. \\ Last, let us consider figure $\ref{fig:brdfmapblazegem}$ which shows the BRDF map produced by using Nvidia Gem's implementation $\cite{cpugems}$ of Stam's BRDF model corresponding to periodic like structures with regularly repeated bumps (along y-axis of the BRDF map). This corresponds to a one dimensional diffraction grating, along the x-axis. This model only uses the spacing $d$ of a given grating. It also always produces highly symmetric results and fails to confirm with the Littrow configuration. % brdf maps patches \begin{figure}[H] \centering \subfigure[FLSS]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/flss.png} \label{fig:brdfmapblazeallLambda} } ~ \subfigure[NMMS]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/nmm.png} \label{fig:brdfmapblazeonlyreq} } \subfigure[PQ]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/pqsinc.png} \label{fig:brdfmapblazepq} } ~ \subfigure[Nvidia Gem]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/gem.png} \label{fig:brdfmapblazegem} } \caption[BRDF Map: Our Approaches applied on a Blazed Grating]{BRDF maps for Blazed grating comparing our different rendering approaches} \label{fig:brdfmapsdiffrenderingapproaches} \end{figure} Figure $\ref{fig:brdfmapsdifflambdastepsblaze}$ and figure $\ref{fig:brdfmapsdifflambdastepselaphe65}$ show the BRDF maps for different wavelength step sizes used in the fragment shader for the FLSS approach applied to the blazed grating and the Elaphe snake shed, respectively. Within our fragment shaders the outermost loop iterates over the range $[380nm, 780nm]$ for a given step size $\lambda_{step}$ to integrate over the wavelength spectrum as illustrated in algorithm $\ref{alg:fragmentshaderall}$. Having bigger step sizes implies having fewer $\lambda$-samples which will reduce the overall runtime of a shader but it will also introduce artifacts and therefore lower the overall shading quality. For an Elaphe surface grating, artifacts are perceivable when $\lambda_{step} \geq 10nm$. Results produced by using $5nm$ step sizes do not differ from those produced by using $\lambda_{step}= 1nm$. This allows us to set $\lambda_{step}$ at $5nm$. For a Blazed grating we may chose even bigger step sized without losing any rendering quality(see figure $\ref{fig:brdfmapsdifflambdastepsblaze}$). \begin{figure}[H] \centering \subfigure[$\lambda_{step=1 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/1.png} \label{fig:brdfmapsDiffLambdaStepsL1Blaze} } ~ \subfigure[$\lambda_{step=5 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/5.png} \label{fig:brdfmapsDiffLambdaStepsL5Blaze} } ~ \subfigure[$\lambda_{step=10 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/10.png} \label{fig:brdfmapsDiffLambdaStepsL10Blaze} } \subfigure[$\lambda_{step=25 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/25.png} \label{fig:brdfmapsDiffLambdaStepsL25Blaze} } ~ \subfigure[$\lambda_{step=50 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/50.png} \label{fig:brdfmapsDiffLambdaStepsL50Blaze} } ~ \subfigure[$\lambda_{step=100 nm}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_lambdasteps/100.png} \label{fig:brdfmapsDiffLambdaStepsL100Blaze} } \caption[BRDF Map: Varying step sizes FLSS Blazed Grating]{Blazed grating at $2.5 \mu m$: Different $\lambda$ step sizes} \label{fig:brdfmapsdifflambdastepsblaze} \end{figure} % elpahe steps \begin{figure}[H] \centering \subfigure[$\lambda_{step=1 nm}$]{ \includegraphics[scale=0.43]{nresults/brdfmaps/flss_elaphe_lambdasteps/1.png} \label{fig:brdfmapsDiffLambdaStepsL1Elaphe65} } ~ \subfigure[$\lambda_{step=5 nm}$]{ \includegraphics[scale=0.43]{nresults/brdfmaps/flss_elaphe_lambdasteps/5.png} \label{fig:brdfmapsDiffLambdaStepsL5Elaphe65} } ~ \subfigure[$\lambda_{step=10 nm}$]{ \includegraphics[scale=0.43]{nresults/brdfmaps/flss_elaphe_lambdasteps/10.png} \label{fig:brdfmapsDiffLambdaStepsL10Elaphe65} } \subfigure[$\lambda_{step=25 nm}$]{ \includegraphics[scale=0.43]{nresults/brdfmaps/flss_elaphe_lambdasteps/25.png} \label{fig:brdfmapsDiffLambdaStepsL25Elaphe65} } ~ \subfigure[$\lambda_{step=50 nm}$]{ \includegraphics[scale=0.43]{nresults/brdfmaps/flss_elaphe_lambdasteps/50.png} \label{fig:brdfmapsDiffLambdaStepsL50Elaphe65} } ~ \caption[BRDF Map: Varying step sizes FLSS Elaphe Grating]{Elaphe grating at $65 \mu m$: Different $\lambda$ step sizes} \label{fig:brdfmapsdifflambdastepselaphe65} \end{figure} The figures $\ref{fig:pqblaze}$ and $\ref{fig:pqelaphe}$ show a comparison of the BRDF maps produced by the FLSS approach (on the left) and the PQ shading approach (on the right) applied on all our patches. For Blazed grating, as already mentioned, we notice that both approaches, FLSS and PQ, look similar but have notable differences. In the PQ map, the first order diffraction color contribution is spread. In general, a Blazed Grating is manufactured in a way that a large fraction of the incident light is diffracted preferentially into the first order. Therefore, most of the energy in its BRDF map lies in the first order of diffraction at its blaze angle. This implies that largest portion of the color contribution, visible on the corresponding BRDF map, lies at that angle. In figure $\ref{fig:pqblaze}$, in contrast to the results produced by the FLSS approach, we see color fringes at the first order modes in the BRDF map produced by our PQ approach. This implies that the PQ approach does not produce reliable results which also affirms our evaluation plots shown in figure $\ref{fig:evaluationdiffshaderpq}$. The BRDF resulting BRDF map for the Elaphe and Xenopeltis gratings using the PQ approach are similar to those resulting by the FLSS approach. Nevertheless they also exhibit some artifacts, similar to these discussed for the Blazed grating. \begin{figure}[H] \centering \subfigure[FLSS Approach: Blazed grating]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/flss.png} \label{fig:fullLambdaBlaze} } ~ \subfigure[PQ Sinc Interpolation Approach: Blazed grating]{ \includegraphics[scale=0.25]{nresults/brdfmaps/approaches_blazed/pqsinc.png} \label{fig:pqBlaze} } \caption[BRDF Map: PQ vs FLSS Approach on Blazed Grating]{A comparison between the PQ- and the FLSS approach applied on an Blazed grating.} \label{fig:pqblaze} \end{figure} \begin{figure}[H] \centering \subfigure[FLSS Approach: Elaphe grating]{ \includegraphics[scale=0.25]{nresults/brdfmaps/flss_pq_elaphe/flss.png} \label{fig:fullLambdaElaphe} } ~ \subfigure[PQ Approach: Elaphe grating]{ \includegraphics[scale=0.25]{nresults/brdfmaps/flss_pq_elaphe/pq.png} \label{fig:pqElaphe} } \caption[BRDF Map: PQ vs FLSS Approach on Elaphe Grating]{A comparison between the PQ- and the FLSS approach applied on an Elaphe grating.} \label{fig:pqelaphe} \end{figure} In figure $\ref{fig:pqxeno}$ we are also comparing the PQ approach with the FLSS applied on a Xenopeltis grating as shown in figure $\ref{fig:xenogratingpatch}$. For the subfigures $\ref{fig:pqxeno}$(a), $\ref{fig:pqxeno}$(b), $\ref{fig:pqxeno}$(c) and $\ref{fig:pqxeno}$(d) we were using the full size of the Xenopeltis grating, which is equal to 65$\mu m$ $\times$ 65$\mu m$ as our input patch. This implies that we were using 1 period for the PQ approach for producing the renderings shown in subfigures $\ref{fig:pqXenoti0}$ and $\ref{fig:pqXenoti20}$. On the other hand, figure $\ref{fig:fullLambdaXenoti10}$ shows a PQ rendering using a smaller patch of the Xenopeltis grating as shown in figure $\ref{fig:pqXenoti10}$. The this patch is only 3.25$\mu m$ $\times$ 3.25$\mu m$ which corresponds to 20 periods. \begin{figure}[H] \centering \subfigure[FLSS Approach: Xenopeltis grating $\theta_i = 0 \degree$]{ \includegraphics[scale=0.19]{nresults/brdfmaps/flss_xeno_thetai/0.png} \label{fig:fullLambdaXenoti0} } ~ \subfigure[PQ Approach: Xenopeltis grating $\theta_i = 0 \degree$]{ \includegraphics[scale=0.19]{nresults/brdfmaps/pq_xeno_thetai/0.png} \label{fig:pqXenoti0} } \subfigure[FLSS Approach: Xenopeltis grating $\theta_i = 20 \degree $]{ \includegraphics[scale=0.19]{nresults/brdfmaps/flss_xeno_thetai/20.png} \label{fig:fullLambdaXenoti20} } ~ \subfigure[PQ Approach: Xenopeltis grating $\theta_i = 20 \degree $]{ \includegraphics[scale=0.19]{nresults/brdfmaps/pq_xeno_thetai/20.png} \label{fig:pqXenoti20} } \subfigure[PQ Approach: Applied on the patch in figure $\ref{fig:pqXenoti10}$]{ \includegraphics[scale=0.19]{nresults/brdfmaps/special/xeno_pq_20.png} \label{fig:fullLambdaXenoti10} } ~ \subfigure[A 3.25$\mu m$ patch from our Xenopeltis grating]{ \includegraphics[scale=5.0]{nresults/brdfmaps/special/xeno6x2.png} \label{fig:pqXenoti10} } \caption[BRDF Map: PQ vs FLSS Approach on Xenopeltis Grating]{A comparison between the PQ- and the FLSS approach applied on an Xenopeltis grating.} \label{fig:pqxeno} \end{figure} Figure $\ref{fig:brdfmapsdiffsigmasizeblaze}$ shows BRDF maps for the FLSS approach when applied to the Blazed grating, while varying the value for the spatial variance $\sigma_s$ for the coherence window. This is akin to changing the coherence length for the incident light. We see that the lower the coherence length gets, the fewer of the interacting grating periods are involved. Having fewer periods cause the production of overlapping blurry diffraction bands (i.e. blobs, see e.g. figure $\ref{fig:brdfmapsDiffSigmaStepsL5Blaze}$) for different wavelengths $\lambda$. Thus, when reducing the number of periods this will produces blurry artifacts which ends in having poorly resolved colors. \begin{figure}[H] \centering \subfigure[$\sigma_{s=3.25 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/3_25.png} \label{fig:brdfmapsDiffSigmaStepsL1Blaze} } ~ \subfigure[$\sigma_{s=6.5 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/6_5.png} \label{fig:brdfmapsDiffSigmaStepsL5Blaze} } ~ \subfigure[$\sigma_{s=15 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/15.png} \label{fig:brdfmapsDiffSigmaStepsL10Blaze} } \subfigure[$\sigma_{s=30 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/30.png} \label{fig:brdfmapsDiffSigmaStepsL25Blaze} } ~ \subfigure[$\sigma_{s=45 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/45.png} \label{fig:brdfmapsDiffSigmaStepsL50Blaze} } ~ \subfigure[$\sigma_{s=65 \mu m}$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_blazed_sigmas/65.png} \label{fig:brdfmapsDiffSigmaStepsL100Blaze} } \caption[BRDF Map: Different spatial variance $\sigma_s$ values]{Blazed grating with periodicity of $2.5 \mu m$: Different $\sigma_s$} \label{fig:brdfmapsdiffsigmasizeblaze} \end{figure} Figures $\ref{fig:brdfmapstayloriterationsblaze}$ and $\ref{fig:brdfmapstayloriterationselaphe65}$ show the BRDF maps for the FLSS approach using a different number of terms $N$ used in the Taylor series approximation. For both input patches we clearly, visually observe the convergence of the taylor series for higher values of N. We visually observe convergence of the Taylor series for all our patches for a very large value of $N$$\footnote{Using N equal to 40 lead to visual convergence for all our used gratings.}$. \\ As discussed in section $\ref{sec:taylorapproximation}$ there exists a certain value of $N$ for which our approach converges. For all our shading approaches, applied on our gratings, we visually observed a convergence of their BRDF maps when using $N \geq 39$ DFT terms. Furthermore, for a Blazed grating it suffices to use only $N \geq 7$ - and for an Elaphe grating only $N \geq 9$ DFT terms. Notice, that these numbers of required DFT terms were empirically determined with a trial and error strategy. \\ However, by making use of Taylor error term estimates, as introduced in the appendix section $\ref{chap:taylorseriesapproxappendix}$, we can derive an upper bound for $N$. This computation is dependent on many aspects, such as on the grating spacing, the sampling rate $dH$ and the used wavelength spectrum. Thus, it is usually simpler to empirically determine an actual value for $N$ to be used. \\ In algorithm $\ref{alg:matlabprecomp}$ we compute different powers of DFT terms of our height fields $h$ by evaluating the expression $DFT(h)^n \cdot i^n$, where $i$ denotes the imaginary number $i$ and $n$ is a natural number. Note that when applying the DFT operator on a real or complex number (e.g. such as one particular value of our height field) we get a complex number. Using our Taylor series approximation has basically four possible convergence images, each having its own convergence radius. The reason for this is due to the fact we are multiplying the DFT terms by $i^n$. \\ In order to understand the reason behind this, let us consider the complex unit $i$. Raising $i$ to the power of a natural number $n$ leaves us four possible results as shown in equation $\ref{eq:thefourunits}$: \begin{align} i^n = \left\{ \begin{array}{rl} 1 &\mbox{ if $n \equiv_4 0$} \\ i &\mbox{ if $n \equiv_4 1$} \\ -1 &\mbox{ if $n \equiv_4 2$} \\ -i &\mbox{ otherwise} \end{array} \right. \label{eq:thefourunits} \end{align} Thus multiplying a complex number $c$, such as particular value of a DFT term, by $i$ raised to a certain power, will permute the the value of $c$. Since we compute $\sum_{k=0}^N DFT(h)^k \cdot i^k$ we can split this summation into four partial summations, according to their applied power factor $i$. Each of these summands converges with a different convergence radius. \begin{figure}[H] \centering \subfigure[$N=1$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/0.png} \label{fig:brdfmapsTaylorN0Blaze} } ~ \subfigure[$N=2$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/1.png} \label{fig:brdfmapsTaylorN1Blaze} } ~ \subfigure[$N=4$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/3.png} \label{fig:brdfmapsTaylorN2Blaze} } \subfigure[$N=24$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/23.png} \label{fig:brdfmapsTaylorN3Blaze} } ~ \subfigure[$N=26$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/25.png} \label{fig:brdfmapsTaylorN4Blaze} } ~ \subfigure[$N=40$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_xeno_taylor/39.png} \label{fig:brdfmapsTaylorN5Blaze} } \caption[BRDF Map: Xenopeltis Grating Convergence]{Blazed grating at $65 \mu m$: $N$ Taylor Iterations} \label{fig:brdfmapstayloriterationsblaze} \end{figure} %taylor var elaphe \begin{figure}[H] \centering \subfigure[$N=1$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/0.png} \label{fig:brdfmapsTaylorN0Elaphe65} } ~ \subfigure[$N=2$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/1.png} \label{fig:brdfmapsTaylorN1Elaphe65} } ~ \subfigure[$N=5$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/4.png} \label{fig:brdfmapsTaylorN2Elaphe65} } \subfigure[$N=10$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/9.png} \label{fig:brdfmapsTaylorN3Elaphe65} } ~ \subfigure[$N=11$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/10.png} \label{fig:brdfmapsTaylorN4Elaphe65} } ~ \subfigure[$N=40$]{ \includegraphics[scale=0.16]{nresults/brdfmaps/flss_elaphe_taylor/39.png} \label{fig:brdfmapsTaylorN5Elaphe65} } \caption[BRDF Map: Elaphe Grating Convergence]{Elaphe grating at $65 \mu m$: $N$ Taylor Iterations} \label{fig:brdfmapstayloriterationselaphe65} \end{figure} Figure $\ref{fig:brdfmapsxenodiffthetaiangles}$ shows the BRDF maps of the FLSS approach applied on the Xenopeltis snake shed, using different $\theta_i$ incident angles. When slightly moving the incident angle $\theta_i$, we can observe how the BRDF map changes. For higher values of $\theta_i$ we start seeing additional diffraction color contributions on the right side of the BRDF map. \begin{figure}[H] \centering \subfigure[Xenopeltis grating $\theta_i=0 \degree$]{ \includegraphics[scale=0.22]{nresults/brdfmaps/flss_xeno_thetai/0o.png} \label{fig:brdfmapXenoti0} } ~ \subfigure[Xenopeltis grating $\theta_i=10 \degree$]{ \includegraphics[scale=0.15]{nresults/brdfmaps/flss_xeno_thetai/10.png} \label{fig:brdfmapXenoti10} } ~ \subfigure[Xenopeltis grating $\theta_i=20 \degree$]{ \includegraphics[scale=0.15]{nresults/brdfmaps/flss_xeno_thetai/20.png} \label{fig:brdfmapXenoti20} } \caption[BRDF Map: Varying Viewing Angles]{BRDF maps for Xenopeltis grating: different $\theta_i$ angles} \label{fig:brdfmapsxenodiffthetaiangles} \end{figure} \section{Rendering Surface Geometries} \label{sec:snakegeomrenderings} In this section we are going to present our actual renderings. These renderings are simulating the effects of diffraction produced when a directional light source encounters different nano-scaled surfaces on a given curved snake skin mesh. We will see that diffraction colors change dramatically with changes in the light direction, surface normals and the viewing direction. This is typical for diffraction colors observed in nature. Therefore, the aim of this section is tho provide a visual comparison between diffraction patterns occurring in nature as shown in figure $\ref{fig:snakespecies}$ and the renderings produced by our shader. Last we compare the results produced by our shader against those produced by Stam's approach. For this purpose we compare applied these methods on a synthetic grating (on a CD) and on a natural grating (on a Xenopeltis grating). \\ For rendering we are going to rely on our FLSS approach. Unfortunately, this approach is rather slow and can barely be considered as being interactively performing. The NMM approach on the other hand, has interactive runtime. In general, both approaches are valid according to to their evaluation plots shown in figure $\ref{fig:evaluationdiffshaderalllambda}$ and in figure $\ref{fig:evaluationdiffshadernminmax}$. Therefore, their rendered results may be considered as being accurate. However, the reason for choosing the FLSS approach instead of the NMM approach is that the renderings resulting from the NMM approach look purplish compared to those produced by the FLSS approach. This color-tone shift towards the purple color region for NMM renderings does not correspond to the reality and is a result of its non-uniform wavelength spectrum sampling. Further information about NMM's tone shift issue and how it can be addressed is discussed in the appendix section $\ref{chap:diffflssnmm}$. \\ The Laboratory of Artificial and Natural Evolution in Geneva provided us with a triangular mesh of a snake. This mesh was produced by a 3D scan for an Elaphe snake species and it consists of \emph{11696} vertices and \emph{22950} faces. Note that, for all our renderings, we used this snake mesh. \\ For the snake species under consideration, their macroscopic geometry is highly similar. Only the geometry of their nano-structures varies and is responsible for a snake's iridescence. Thus, we can use the same snake surface model to render diffraction for different species. Table $\ref{tab:hardwarespecifications}$ lists the system specifications of the machine I used in order to produce the rendered images. \begin{table}[H] \centering \begin{tabular}{|r|l|} \hline Processor & Intel i7 CPU 970 @ 3.20 GHz (12 CPUs) \\ Memory & 12288 MB RAM \\ Graphics Card & GeForce GTX 770 \\ Graphics Clock & 1150 MHz \\ Graphics Memory & 4096 MB \\ Graphics Memory Bandwidth & 230.4 $GB/s$ \\ \hline \end{tabular} \caption[Hardware Specifications]{Hardware specifications of the machine used to render snake surface. Statistics are provided using the tool $\emph{NVIDIA Geforce Experience}$.} \label{tab:hardwarespecifications} \end{table} Figure $\ref{fig:renderingdifferentsnankegratings}$ shows renderings produced by the FLSS approach applied on our snake mesh for different, given input patches. Due do the fact that a Blazed grating has its maximum intensity for a certain direction and the geometry of the snake mesh is curved i.e. it is non-flat, we can expect rather less diffraction color contribution as shown in figure $\ref{fig:renderingelaphegrating}$. \\ %%TODO write something about similar configuration for blazed grating In contrast, for both the renderings, we see colorful patterns on the skin of our snake species, Elaphe and Xenopeltis, due to the effect of diffraction. We see much less colorful patterns for Elaphe as shown in figure $\ref{fig:renderingelaphegrating}$ than for Xenopeltis as shown in figure $\ref{fig:renderingxenograting}$. This is consistent with the observations in the real world as shown in figure $\ref{fig:snakespecies}$. As observable in figure $\ref{fig:elpahegratingpatch}$, the substructures (the finger like structures) in the height field of a Elaphe snake skin overlap (i.e are not very regularly aligned) along the y-axis. This is why the Elaphe species is less iridescent than the other specie. The Xenopeltis snake has a brownish body with no pigmentation, which makes the iridescence more spectacular than on Elaphe as seen in figure $\ref{fig:snakespecies}$. \begin{figure}[H] \centering \subfigure[Blazed grating]{ \includegraphics[scale=0.23]{nresults/snakerenderings/flss_gratings/blazed1552.png} \label{fig:renderingblazegrating} } ~ \subfigure[Elaphe grating]{ \includegraphics[scale=0.23]{nresults/snakerenderings/flss_gratings/elaphe.png} \label{fig:renderingelaphegrating} } ~ \subfigure[Xeno grating]{ \includegraphics[scale=0.23]{nresults/snakerenderings/flss_gratings/xeno.png} \label{fig:renderingxenograting} } \caption[Snake Renderings: Our Approaches applied on our Gratings]{Diffraction of different snake skin gratings rendered on a snake geometry} \label{fig:renderingdifferentsnankegratings} \end{figure} Figure $\ref{fig:renderingelaphe65}$ shows a set of subfigures for rendering the effect of diffraction produced by the FLSS approach (used as our reference approach), applied on our snake mesh using the Elaphe nano structure. Figure $\ref{fig:renderingelaphe65dt}$ shows the final diffraction color contribution result with texture-blending. We note that the diffraction color contribution is not significant in this subfigure which resembles quite well to the reality as shown in figure $\ref{fig:elpahespecies}$. In subfigure $\ref{fig:renderingelaphe65ns}$ we see the light cone in order to show the direction of the light source besides the rendered results. Subfigure $\ref{fig:renderingelaphe65ft}$ is a sample visualization of a Fourier transformation for an Elaphe nano-scale surface structure as shown in figure $\ref{fig:renderingelaphe65ns}$. \begin{figure}[H] \centering \subfigure[Diffraction Patten]{ \includegraphics[scale=0.25]{results/snakerenderings/elaphe65/1.png} \label{fig:renderingelaphe65dp} } ~ \subfigure[Diffraction + Texture]{ \includegraphics[scale=0.25]{results/snakerenderings/elaphe65/2.png} \label{fig:renderingelaphe65dt} } \subfigure[Texture with Light Direction]{ \includegraphics[scale=0.2]{nresults/viewingcone.png} \label{fig:renderingelaphe65tl} } ~ \subfigure[Nanostructure]{ \includegraphics[scale=0.11]{results/snakerenderings/elaphe65/4.png} \label{fig:renderingelaphe65ns} } ~ \subfigure[8th DTF Term]{ \includegraphics[scale=0.31]{nresults/elaphe8dft.png} \label{fig:renderingelaphe65ft} } \caption[Snake Renderings: FLSS Approach for Elaphe Grating]{Diffraction for Elaphe snake skin produced by our reference approach.} \label{fig:renderingelaphe65} \end{figure} Figure $\ref{fig:renderingxeno65}$ shows a set of subfigures for the effect of diffraction for the Xenopeltis snake surface. In our Xenopeltis rendering as shown in figure $\ref{fig:renderingXeno65FT}$ when using a grating as shown in figure $\ref{fig:renderingXeno65NS}$ we see a lot of iridescent color contribution. Comparing this to a real image $\ref{fig:renderingXeno65DP}$ we notice much resemblance regarding the reflectance strength and colorful pattern. \begin{figure}[H] \centering \subfigure[Diffraction Patten]{ \includegraphics[scale=0.5]{results/snakerenderings/xeno65/1.png} \label{fig:renderingXeno65DP} } \subfigure[Nanostructure]{ \includegraphics[scale=0.15]{results/snakerenderings/xeno65/4.png} \label{fig:renderingXeno65NS} } ~ \subfigure[3th DTF Term]{ \includegraphics[scale=0.563]{nresults/xeno3dft.png} \label{fig:renderingXeno65FT} } \caption[Snake Renderings: FLSS Approach for Xenopeltis Grating]{Diffraction for Xenopeltis snake skin produced by our reference approach.} \label{fig:renderingxeno65} \end{figure} Figure $\ref{fig:renderingdifferentzoomlevelselaphe}$ shows the diffraction pattern for an Elaphe snake shed at different zoom levels for a fixed incident light and a fixed viewing direction. We changed the zoom-levels by adjusting the field of view angle of our camera. For each image in this figure, the one to its right side is a five times zoomed-in version of the region within its red box. The close up perspectives exhibit complex and colorful diffraction patterns. \begin{figure}[H] \centering \subfigure[$zoom = 1 \times$]{ \includegraphics[scale=0.5]{nresults/snakerenderings/flss_elaphe_zoom/10.png} \label{fig:renderingZoomElaphe05} } ~ \subfigure[$zoom = 3 \times$]{ \includegraphics[scale=0.28]{nresults/snakerenderings/flss_elaphe_zoom/5.png} \label{fig:renderingZoomElaphe02} } ~ \subfigure[$zoom = 6 \times$]{ \includegraphics[scale=0.24]{nresults/snakerenderings/flss_elaphe_zoom/1.png} \label{fig:renderingZoomElaphe01} } \caption[Snake Renderings: FLSS Approach for a varying FoV]{Showing the diffraction pattern on an Elaphe grating for different camera zoom levels by varying the field of view.} \label{fig:renderingdifferentzoomlevelselaphe} \end{figure} Figure $\ref{fig:renderingelaphelightrotations6}$ shows how the diffraction pattern changes when the incident light direction is moved slightly. This figure gives us an impression of what kind of complex, perspective-dependent pattern the diffraction phenomenon produces. The figure column from figure $\ref{fig:renderingElapheRotX2}$ to figure $\ref{fig:renderingElapheRotX6}$ models a scene in which a virtual sun (i.e. a directional light source) moves from south (S) to north (N). For each of this subfigure we can see in its inset (red box) how the diffraction color changes according to how the position of the sun changes. The figure column from figure $\ref{fig:renderingElapheRotY2}$ to figure $\ref{fig:renderingElapheRotY6}$ models a scenario when a sun moves from south-east (SE) to west (W). Similarly, the diffraction patterns changes according to the position of the sun. \begin{figure}[H] \centering \subfigure[$(-3.3130, 0.0, -0.9999)$]{ \includegraphics[scale=0.345]{nresults/snakerenderings/flss_xeno_lightdir/1/1.png} \label{fig:renderingElapheRotX2} } ~ \subfigure[$(0.0995, 0.0993, -0.9900)$]{ \includegraphics[scale=0.245]{nresults/snakerenderings/flss_xeno_lightdir/2/2.png} \label{fig:renderingElapheRotY2} } \subfigure[$(-0.1989, 0.0, -0.9799)$]{ \includegraphics[scale=0.245]{nresults/snakerenderings/flss_xeno_lightdir/1/2.png} \label{fig:renderingElapheRotX4} } ~ \subfigure[$(0.0995, 0.2940, -0.9505)$]{ \includegraphics[scale=0.245]{nresults/snakerenderings/flss_xeno_lightdir/2/3.png} \label{fig:renderingElapheRotY4} } \subfigure[$(-0.3897, 0.0, -0.9208)$]{ \includegraphics[scale=0.245]{nresults/snakerenderings/flss_xeno_lightdir/1/3.png} \label{fig:renderingElapheRotX6} } ~ \subfigure[$(0.0995, 0.4770, -0.8731)$]{ \includegraphics[scale=0.245]{nresults/snakerenderings/flss_xeno_lightdir/2/4.png} \label{fig:renderingElapheRotY6} } \caption[Snake Renderings: FLSS Approach for varying Light Directions]{Showing the change of the diffraction pattern on an Elaphe grating for different light directions. Figure $\ref{fig:renderingElapheRotX2}$ shows the geographic direction in order to describe a local orientation. The first subfigure column the sun moves from south (S) to north (N). In the second column, the sun moves from south-east to west (W).} \label{fig:renderingelaphelightrotations6} \end{figure} Figure $\ref{fig:experimentelaphe65}$ shows a photo of a real experiment which is demonstrating color production due to the effect of diffraction produced when a laser beam hits an Elaphe grating. The exact parameters for this experimental setup are unknown. However, an exemplary laboratory setup is shown in figure $\ref{fig:experimentalsetuppew}$. Nevertheless, this photo of a real experiment gives us an impression of how much our model resembles to the reality. When comparing our BRDF map renderings of the Elaphe grating in figure $\ref{fig:brdfmapElaphe}$ (when using the FLSS approach) with the photo from figure $\ref{fig:experimentelaphe65}$, we notice similarities in the diffraction patterns. \begin{figure}[H] \centering \includegraphics[scale=0.9]{nresults/labsetup.png} \caption[Lab Setup]{Illustration of the laboratory setup for the diffraction experiments} \label{fig:experimentalsetuppew} \end{figure} \begin{figure}[H] \centering \includegraphics[scale=0.2]{results/experiment/elaphe/g2.png} \caption{Diffraction Elaphe: Pattern for a sample setup} \label{fig:experimentelaphe65} \end{figure} In the following we provide a brief comparison between Stam's$\footnote{A reference implementation of Stam's Diffraction Shader\cite{diffstam} is provided by Nvidia's GPU Gems at \texttt{http://http.developer.nvidia.com/GPUGems/gpugems\textunderscore 08.html}}$ and our FLSS method. For this purpose we use two different kinds of gratings, a synthetic, regularly aligned grating and a natural, complex structured grating. These gratings are shown in figure $\ref{fig:stameggratings}$. \begin{figure}[H] \centering \includegraphics[scale=0.5]{background/gratingsstamreworked.png} \caption[Comparing Stam's Approach: Gratings]{Alignment of nano-structures in diffraction gratings. On the left a complex, natural grating of the Elaphe snake species and on the right a synthetic, very regularly aligned grating of a CD.} \label{fig:stameggratings} \end{figure} Figure $\ref{fig:stameggratingsgoodeg}$ shows an example of a case where Stam's approach performs well. Considering the red-line in the figure we notice that the nano-scaled structures of a compact disc are very regularly aligned along the surface. Tracks of a CD are uniformly spaced and bumps along a track are distributed according to a poisson distribution$\footnote{See \texttt{http://en.wikipedia.org/wiki/Poisson\textunderscore distribution}}$. We notice that the diffraction patterns in both approaches look similar. \begin{figure}[H] \centering \includegraphics[scale=0.5]{background/stamgoodeg.png} \caption[Comparing Stam's approach: Good Example]{Comparison of our approach against a reference implementation of Stam's method provided by Nvidia Gem. For synthetic diffraction gratings, which have a very regular structure, Stam's approach is doing well. Both approaches produce similar diffraction patterns.} \label{fig:stameggratingsgoodeg} \end{figure} Finally, tried to reproduce a real Xenopeltis image (as shown in figure $\ref{fig:xenopeltisrealimage}$) for a unknown light and viewing direction. For this purpose I used our FLSS rendering approach and compared its results against Nvidia Gem's implementation. The results are shown in figure $\ref{fig:attemptxeno}$. Even the results of Stam's method look convincing, they have some issues. In some regions in his renderings close to specular regions, there are missing colors. The color distribution in renderings produces by his approach is rather discrete in Stam's. Also notice that Stam's approach always produces a symmetric reflectance map (i.e. BRDF map) exhibiting results in regions where it must not. \begin{figure}[H] \centering \subfigure[Real Xenopeltis Photograph]{ \includegraphics[scale=0.6]{nresults/snakerenderings/gem_flss_xeno/xeno.png} \label{fig:xenopeltisrealimage} } \subfigure[FLSS]{ \includegraphics[scale=0.25]{nresults/snakerenderings/gem_flss_xeno/flss.png} \label{fig:flssattemptxeno} } ~ \subfigure[Stam]{ \includegraphics[scale=0.25]{nresults/snakerenderings/gem_flss_xeno/gem.png} \label{fig:gemattemptxeno} } \caption[Snake Renderings: Stam's vs. FLSS Approach on Xenopeltis]{Comparison of our FLSS approach(see figure $\ref{fig:flssattemptxeno}$) against a reference implementation of Stam's method(see figure $\ref{fig:gemattemptxeno}$) provided by Nvidia Gem. We attempt to reproduce a real Xenopeltis skin coloration(see figure $\ref{fig:xenopeltisrealimage}$). For natural diffraction gratings, which have a rather complex structure, Stam's approach is qualitatively different from the real image.} \label{fig:attemptxeno} \end{figure} In this chapter we have discussed the results produced by our rendering approaches. For the actual snake renderings, we were using the FLSS approach since the NMM approach produces renderings that exhibit an unpleasant purplish color-tone shift. A discussion how to address this issue is discussed in the appendix chapter $\ref{chap:diffflssnmm}$. Relying on the FLSS approach for producing the renderings of the section $\ref{sec:snakegeomrenderings}$ is appropriate since the aim of this section is to produce convincing renderings. However, when speed matters using the NMM approach is the way to go. Even we did not provide any runtime measurement analysis, we usually achieved interactive runtime using the NMM approach. In average, we get \emph{about 5-10 FPS} using the NMM approach relying on a hardware setup as listed in table $\ref{tab:hardwarespecifications}$.MyMiDiII/bmstu-aa \begin{titlepage} \newgeometry{ left=30mm, right=10mm, top=20mm, bottom=17mm } \noindent \begin{minipage}{0.15\textwidth} \includegraphics[width=\linewidth]{../data/bmstu_logo.jpg} \end{minipage} \fontsize{12pt}{12pt}\selectfont \noindent \begin{minipage}{0.85\textwidth}\centering \textbf{Министерство науки и высшего образования Российской Федерации}\\ \textbf{Федеральное государственное бюджетное образовательное учреждение высшего образования}\\ \textbf{«Московский государственный технический университет~\\имени Н.Э.~Баумана}\\ \textbf{(национальный исследовательский университет)»}\\ \textbf{(МГТУ им. Н.Э.~Баумана)} \end{minipage} \begin{spacing}{0.2} ~\\ \end{spacing} \noindent \rule{17cm}{3pt} ~\\ \begin{spacing}{0.1} ~\\ \end{spacing} \noindent ФАКУЛЬТЕТ \uline { ~~~~~~~~~~~~~~~~~~~~~~~~~~~ «Информатика и системы управления» ~~~~~~~~~~~~~~~~~~~~~~~~~~~ } \newline\newline \noindent КАФЕДРА \uline{ ~~~~~~~~~ «Программное обеспечение ЭВМ и информационные технологии» ~~~~~~~~~ } \newline\newline \newline\newline \newline\newline \newline\newline \newline \fontsize{18pt}{18pt}\selectfont \begin{center} \textbf{РУБЕЖНЫЙ КОНТРОЛЬ №1}\\ \textbf{по курсу «Анализ алгоритмов»}\\ ~\\ \fontsize{16pt}{16pt}\selectfont \text{«Моделирование конвейерной обработки»} \end{center} ~\\ \fontsize{14pt}{14pt}\selectfont \noindent\text{Студент} \uline{ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ } \newline\newline \noindent\text{Группа} \uline{ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ИУ7-53Б ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ } \newline\newline \noindent\text{Оценка (баллы)} \uline{ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ } \newline\newline \noindent\text{Преподаватель} \uline{ ~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ } \newline\newline ~\\ ~\\ ~\\ \vspace{17mm} \begin{center} \the\year~г. \end{center} \restoregeometry \end{titlepage} \hypertarget{class_m_v_digit_set}{}\doxysection{M\+V\+Digit\+Set Class Reference} \label{class_m_v_digit_set}\index{MVDigitSet@{MVDigitSet}} {\bfseries{I\+T\+EM}} adds On-\/\+Display {\itshape Per-\/\+Digit} S\+ET feature to \mbox{\hyperlink{class_menu_value}{Menu\+Value}}. {\ttfamily \#include $<$Menu\+Lib.\+h$>$} Inheritance diagram for M\+V\+Digit\+Set\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=5.000000cm]{class_m_v_digit_set} \end{center} \end{figure} \doxysubsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{class_m_v_digit_set_a052cbf35f3f3bbc8f45d54e03c3a3651}\label{class_m_v_digit_set_a052cbf35f3f3bbc8f45d54e03c3a3651}} {\bfseries M\+V\+Digit\+Set} (const char $\ast$\+\_\+\+C\+Name, int \+\_\+\+Range\+Min=0x0000, int \+\_\+\+Range\+Max=0x\+F\+F\+FF, int \+\_\+\+V\+Base=D\+EC) \item \mbox{\Hypertarget{class_m_v_digit_set_a0f317011860feb9b76bd2db24e46f4e7}\label{class_m_v_digit_set_a0f317011860feb9b76bd2db24e46f4e7}} {\bfseries M\+V\+Digit\+Set} (int \+\_\+\+E\+Name\+Addr, byte \+\_\+\+E\+Name\+Size, bool \+\_\+\+Name\+Settable=false, int \+\_\+\+Range\+Min=0x0000, int \+\_\+\+Range\+Max=0x\+F\+F\+FF, int \+\_\+\+V\+Base=D\+EC) \end{DoxyCompactItemize} \doxysubsection*{Protected Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{class_m_v_digit_set_acce2d213bbaa7633205c651f74e3b931}\label{class_m_v_digit_set_acce2d213bbaa7633205c651f74e3b931}} virtual byte \mbox{\hyperlink{class_m_v_digit_set_acce2d213bbaa7633205c651f74e3b931}{Nav\+Value}} (byte \+\_\+\+Key, byte \+\_\+\+Row=0) override \begin{DoxyCompactList}\small\item\em Value Navigate. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection*{Additional Inherited Members} \doxysubsection{Detailed Description} {\bfseries{I\+T\+EM}} adds On-\/\+Display {\itshape Per-\/\+Digit} S\+ET feature to \mbox{\hyperlink{class_menu_value}{Menu\+Value}}. Definition at line 104 of file Menu\+Lib.\+h. The documentation for this class was generated from the following files\+:\begin{DoxyCompactItemize} \item C\+:/\+Users/\+T\+G\+I\+T-\/\+T\+E\+C\+H/\+Documents/\#\+Projects/\+Tode-\/\+R\+C/firmware/\+Tode-\/\+E32-\/20\+C9/\mbox{\hyperlink{_menu_lib_8h}{Menu\+Lib.\+h}}\item C\+:/\+Users/\+T\+G\+I\+T-\/\+T\+E\+C\+H/\+Documents/\#\+Projects/\+Tode-\/\+R\+C/firmware/\+Tode-\/\+E32-\/20\+C9/\mbox{\hyperlink{_menu_lib_8cpp}{Menu\+Lib.\+cpp}}\end{DoxyCompactItemize} 10-100 %\documentclass[wsdraft]{ws-procs11x85} \documentclass{ws-procs11x85} \usepackage{ws-procs-thm} % comment this line when `amsthm / theorem / ntheorem` package is used \usepackage{hyperref} \usepackage{xcolor} \hypersetup{ colorlinks, linkcolor={red!50!black}, citecolor={blue!50!black}, urlcolor={blue!80!black} } \begin{document} \title{TITLE} \author{\textsuperscript{1}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2}, \textsuperscript{2,3} and \textsuperscript{1-3}$^\dag$} \address{ \textsuperscript{1}Department of Bioengineering, Stanford University, Stanford, CA, 94305 \\ \textsuperscript{2}Department of Biomedical Data Science, Stanford University, Stanford, CA, 94305 \\ \textsuperscript{3}Department of Medicine, Stanford University, Stanford, CA, 94305 \\ $^\dag$E-mail: \\ } \begin{abstract} ABSTRACT \end{abstract} \keywords{Pharmacogenomics; Biocuration; Text mining.} % required, do-not-remove \copyrightinfo{\copyright\ 2019 The Authors. Open Access chapter published by World Scientific Publishing Company and distributed under the terms of the Creative Commons Attribution Non-Commercial (CC BY-NC) 4.0 License.} CONTENTS \bibliographystyle{ws-procs11x85} \bibliography{bibliography} \end{document} %%% \renewcommand\bibname{References\\ {\normalfont\it References can be typed in your preferred bibliography style.}}1-10 \documentclass{article} \usepackage{incgraph} \usepackage{amsmath} \usepackage{qtree} \begin{document} \begin{inctext} \Tree [.$=$ [.$M_x$ ] [.$+$ [.$M_x$ ] [.$C_1$ ] ] ] \end{inctext} \end{document} \begin{displaymath} \frac{1}{2}+\frac{1}{10}\cos (2t) + \frac{1}{5}\sin (2t) +\lambda e^{-t} \quad (\lambda \in \R) \end{displaymath} \documentclass{article} \usepackage{scrextend} \usepackage[utf8]{vietnam} \usepackage{titling} \usepackage{setspace} \usepackage[a4paper, total={170mm,257mm},left=25mm,right=25mm, top=20mm]{geometry} \usepackage{unicode-math} \usepackage{amsfonts} \usepackage{amsmath} \DeclareMathOperator*{\argmax}{arg\,max} \usepackage[symbol]{footmisc} \makeatletter \newcommand\footnoteref[1]{\protected@xdef\@thefnmark{\ref{#1}}\@footnotemark} \makeatother \usepackage{footnotehyper} \usepackage{perpage} %the perpage package \MakePerPage{footnote} \usepackage{scrextend} \usepackage{scrextend} \title{\textbf{Phát hiện COVID-19 và viêm phổi \\ bằng ảnh X-quang sử dụng mạng neural tích chập\footnote[1]{Đây là báo cáo đồ án môn học Nhập môn Thị giác máy tính - CS231.L23.KHCL tại trường Đại học Công nghệ Thông tin - ĐHQG HCM}}} \setlength{\parindent}{0pt} \usepackage{indentfirst} \setlength{\parindent}{0pt} \usepackage{ragged2e} \usepackage[english]{babel} \usepackage[square,numbers]{natbib} \usepackage[colorlinks, citecolor = black, urlcolor = gray, bookmarks = false, hypertexnames = true ]{hyperref} \bibliographystyle{abbrvnat} \author{ \textbf{}\\ \small \and \textbf{}\\ \small } \date{Tháng 07/2021} \usepackage{graphicx} \usepackage[ruled, lined, linesnumbered, commentsnumbered, longend]{algorithm2e} \usepackage{xcolor} \usepackage{mathtools} \usepackage[]{algorithm2e} \begin{document} \maketitle \begin{abstract} \justifying \noindent Hiện nay, tình hình COVID-19 của Việt Nam và cả thế giới đang trở nên vô cùng phức tạp. Đó là lý do tại sao chúng ta cần phải có hành động kịp thời để có thể ứng phó với cơn đại dịch này và bước đầu trong công cuộc ứng phó này phải kể đến việc xác định đâu là các nhiễm COVID-19. Vì thế nên trong bài báo cáo này, chúng tôi sẽ giới thiệu một phương pháp xét nghiệm COVID có độ hiệu quả cao trong thời gian ngắn, sau đó sẽ áp dụng phương pháp đó lên nhiều mô hình máy học khác nhau và tiến hành phân tích ưu nhược điểm của các mô hình này để có một cái nhìn tổng quan hơn về việc áp dụng trí tuệ nhân tạo vào xét nghiệm nhanh COVID-19. \end{abstract} % /--------------------------------/ \section{Giới thiệu} \subsection{Về số liệu thống kê} Tình hình dịch COVID-19 hiện nay là một vấn đề cấp bách không chỉ với Việt Nam, mà còn ở toàn thế giới. Cụ thể, tính đến $18h$ ngày $16/6/2021$ theo giờ Việt Nam, toàn thế giới ghi nhận hơn $177.470.620$ ca nhiễm COVID-19, trong đó, $3.839.931$ ca tử vong và $161.919.653$ ca đã bình phục. Ngoài ra, trong vòng $24h$ từ ngày $15/6$ đến $16/6/2021$, cả thế giới ghi nhận hơn $313.790$ ca mắc COVID-19 và $9.749$ ca tử vong \cite{worldometers}.\\ Còn ở Việt Nam, dữ liệu được cập nhật đến $18$h ngày $21/5/2021$, tổng cộng có $4.941$ người nhiễm COVID-19, trong đó có $2.689$ ca khỏi bệnh và $41$ ca tử vong. Và trong vòng ngày $19/6/2021$, có thêm $470$ ca nhiễm mới \cite{ncov}.\\ \subsection{Về phương pháp xét nghiệm} Để đối phó với dịch bệnh viêm đường hô hấp cấp này, các y bác sĩ đã có khá nhiều những phương pháp để có thể chẩn đoán nhanh những ca mắc virus SARS-CoV-2. Tuy nhiên, phương pháp được cho là có hiệu quả và phổ biến hơn cả chính là phương pháp xét nghiệm nhanh COVID bằng phản ứng chuỗi Polymerase. \subsubsection{Xét nghiệm sinh học phản ứng chuỗi Polymerase thời gian thực} \textit{Phản ứng chuỗi Polymerase} (PCR) là một phản ứng nhân bản DNA dựa trên các chu kì nhiệt. Và \textit{xét nghiệm sinh học phân tử chuỗi Polymerase} (RT PCR) là phương pháp xét nghiệm xác định sự hiện diện của virus thông qua phát hiện vật liệu di truyền của virus SARS-CoV-2 \cite{ncov}. Đây là một phương pháp có độ chính xác cao nhưng lại đòi hỏi các hệ thống máy chuyên dụng và cần phải được thực hiện tại các phòng thí nghiệm.\\ Ưu điểm của RT PCR đó chính là nó có độ uy tín và độ chính xác cao. Tuy nhiên nhược điểm lại chính là việc nó thường chỉ được sử dụng cho những ca nghi nhiễm virus SARS-CoV-2, không thể áp dụng rộng rãi với toàn dân vì đây là phương pháp tốn khá nhiều nhân công (thường là những người có trình độ cao về y học) và phải ở trong môi trường đặc thù như phòng thí nghiệm, cho nên phương pháp này có giá thành khá cao: $734.000$ VNĐ cho 1 lần xét nghiệm \textit{(theo thông tư 5834 của Bộ Y tế)}. Phương pháp này cũng không có tính ổn định tuyệt đối vì vẫn có trường hợp 2 - 3 lần đầu là âm tính nhưng lần tiếp theo là dương tính. Ngoài ra, cũng xuất hiện các trường hợp dương tính giả và âm tính giả khi xét nghiệm nhanh Covid bằng RT PCR (tại vì số lượng virus nhân lên chưa đủ lớn và xuất hiện nhiều trong đường hô hấp, hoặc do kĩ thuật lấy mẫu và điều kiện lấy mẫu chưa chuẩn, hoặc thậm chí là do quá trình vận chuyển và bảo quản mẫu xét nghiệm chưa đúng cách). \subsubsection{Chẩn đoán hình ảnh} Ngoài RT PCR, còn có một phương pháp khác giúp bác sĩ có thể xác định nhanh những ca COVID. Đó chính là phương pháp chẩn đoán hình ảnh bằng ảnh X-quang.\\ Phương pháp chẩn đoán hình ảnh này là một phương pháp nhanh hơn rất nhiều so với xét nghiệm nhanh Covid bằng RT PCR. Ngoài ra, giá thành của chẩn đoán hình ảnh bằng ảnh X-quang là thấp, chỉ vào khoảng $100.000$ VNĐ trên 1 người trên 1 lần chụp.\\ Tuy nhiên, nhược điểm của phương pháp này chính là việc các bác sĩ khó có thể chẩn đoán chính xác hoàn toàn các ca nhiễm COVID chỉ sử dụng ảnh X-quang, mà các bác sĩ còn phải cần dùng những đặc điểm dịch tễ và những biểu hiện lâm sàng khác để có thể đưa ra các chẩn đoán phù hợp. Điều này vô hình chung lại gây nên sự thiếu hụt về nguồn nhân lực về y tế vì phương pháp này chỉ có thể sử dụng bởi những y bác sĩ hàng đầu về chẩn đoán hình ảnh. Một nhược điểm khác trong chẩn đoán hình ảnh đó chính là tình trạng nhầm lẫn giữa COVID-19 và bệnh viêm phổi.\\ \subsubsection{Chẩn đoán hình ảnh sử dụng trí tuệ nhân tạo} Để cải thiện những hạn chế mà phương pháp chẩn đoán hình ảnh truyền thống tạo ra, chúng tôi giới thiệu một phương pháp chẩn đoán mới có thể giúp cho các y bác sĩ giảm thiểu được không chỉ về chi phí cần bỏ ra cho những lần xét nghiệm, mà còn giúp giảm thiểu những hạn chế về nhân lực như đã nêu ở phần trên.\\ Vì chẩn đoán hình ảnh bằng ảnh X-quang có một chi phí tương đối thấp, cho nên ta có thể áp dụng việc chẩn đoán này để có thể xét nghiệm Covid diện rộng mà không bị trở ngại về chi phí. Ngoài ra, áp dụng trí tuệ nhân tạo (AI) để chẩn đoán cũng giúp tăng đáng kể tốc độ chẩn đoán COVID khi AI cho ra kết quả ngay lập tức thay vì ta phải chờ 2 3 tiếng như RT PCR. Chúng ta có thể sử dụng việc chẩn đoán hình ảnh diện rộng để bước đầu xác định được những ca dương tính, sau đó chúng ta sẽ tiến hành làm xét nghiệm nhanh Covid bằng RT PCR để có thể kiểm tra lại. \\ Ngoài ra đối với những người đã phơi nhiễm hoặc bị nhiễm COVID-19 thì trong thời gian cách ly và chữa trị, chúng ta có thể áp dụng chẩn đoán hình ảnh đến khi nào người đó có kết quả chẩn đoán là âm tính thì ta có thể kiểm tra lại bằng RT PCR. Điều này giúp chúng ta có thể tiết kiệm ngân sách một cách rõ rệt (cụ thể, giảm chi phí đi gấp 7 lần so với việc sử dụng RT PCR mỗi lần cần kiểm tra). \\ Và vì AI không cần đến nguồn nhân lực dồi dào và đặc thù để có thể sử dụng nên ta có thể tiết kiệm được nguồn nhân lực, nhất là nguồn nhân lực của y bác sĩ. % /--------------------------------/ \section{Cơ sở luận cho chẩn đoán COVID-19 bằng ảnh X-quang} Để có thể áp dụng AI cho việc chẩn đoán COVID-19, cũng như phân biệt giữa COVID-19 và viêm phổi, chúng ta cần phải biết được một số đặc điểm nhất định của cả 2 nền bệnh khi chỉ nhìn vô ảnh chụp X-quang. Như chúng ta có thể thấy ở ảnh dưới đây: \newpage \begin{figure}[h!] \centering \includegraphics[width=\textwidth]{covid+pneumonia+normal.png} \end{figure} \begin{center} Hình 1. Ảnh X-quang cho các trường hợp:\\ Bệnh nhân COVID-19 (A), bệnh nhân viêm phổi (B) và người không bị bệnh về đường hô hấp (C) \end{center} Ảnh X-quang phổi của bệnh nhân viêm phổi và bệnh nhân COVID khác với phổi bình thường đó chính là việc phổi của những người mắc bệnh có xuất hiện các đốm trắng nhiều hoặc ít ở những vị trí khác nhau trong phổi. \\ Những đốm trắng này trong y khoa được gọi là \textit{hình ảnh kính mờ} (ground glass pattern). Hình ảnh kính mờ là tổn thương đông đặc không hoàn toàn, có tỷ trọng cao hơn nhu mô phổi xung quanh vẫn có thể thấy đường bờ các mạch máu hoặc phế quản bên trong tổn thương đó.\\ Một bác sĩ chuyên gia về chẩn đoán hình ảnh có thể nói rằng những hình ảnh kính mờ này chính là nguyên nhân gây nên những đốm trắng trong ảnh. Và các bác sĩ có thể dùng đặc trưng này để phân biệt bệnh nhân COVID hay viêm phổi. Vì thế cho nên có thể biết được rằng khi sử dụng những mạng học sâu, các mô hình máy học sẽ học dựa trên những đặc điểm hình thái này và cho ra kết quả chẩn đoán phù hợp nhất với từng ca bệnh. % /--------------------------------/ \section{Các phương pháp} Trong đồ án này chúng tôi tận dụng sức mạnh của hai mạng neural là VGG19 \cite{vgg19} và ResNet50 \cite{resnet50}. Cả hai kiến trúc mạng này đã chứng minh được hiệu quả của mình qua các ứng dụng trong thực tế. Bên cạnh đó chúng tôi cũng sử dụng COVIDx dataset, một bộ dữ liệu được sử dụng rất nhiều trong các nghiên cứu về COVID-19 hiện nay. \subsection{COVIDx Dataset} COVIDx Dataset \cite{covidxdataset} là một bộ dữ liệu được tổng hợp từ nhiều tập dữ liệu khác nhau, cụ thể: \citet{data1}, \citet{data2}, \citet{data3}, \citet{data4}, và \citet{data5}. Ngoài ra, bộ dữ liệu này cũng cung cấp một công cụ chuyển ảnh y khoa từ định dạng .mri thành định dạng .jpg. Thêm vào đó, tác giả cũng cung cấp một đoạn mã giúp hỗ trợ tiền xử lý, lược bỏ những thành phần không cần thiết cho bộ dữ liệu đã được tổng hợp.\\ Bộ dữ liệu gồm hơn $20.000$ ảnh X-quang phổi từ những bệnh nhân khác nhau, được chia làm 2 tập: tập train và tập test, và được phân thành 3 lớp lần lượt là: viêm phổi (train: 5963, test: 105), COVID-19 (train: 4649, test: 274) và bình thường (train: 8751, test: 100).\\ Mô hình sẽ lấy vào input là một ảnh chụp X-quang phổi và sẽ cho ra ouput là xác suất ảnh chụp X-quang đó rơi vào từng lớp viêm phổi, COVID-19 và bình thường. \subsection{Chi tiết thực hiện} Cả hai mạng học sâu là VGG19 và ResNet50 chúng tôi đề xuất đều đã được pretrained trên ImageNet \cite{imagenet}. Sau đó chúng tôi tiến huấn luyện trên tập dữ liệu COVIDx với thuật toán tối ưu hóa là Adam và chiến lược là learning rate sẽ giảm khi nếu loss của tập validation không cải thiện sau một khoảng thời gian (patience).\\ Đối với mạng VGG19, các siêu tham số chúng tôi sử dụng cho huấn luyện là: learning rate = 5e-4, số lượng epoch = 13, kích thước batch = 32, factor = 0.1, patience = 3 và độ phân giải của ảnh đầu vào = 480x480.\\ Trong khi đó đối với mạng ResNet50, chúng tôi thực hiện huấn luyện 2 lần riêng biệt, cụ thể: \begin{itemize} \item Các siêu tham số cho lần đầu tiên là: learning rate = 5e-4, số lượng epoch = 14, kích thước batch = 32, factor = 0.1, patience = 3, độ phân giải của ảnh đầu vào = 224x224. \item Các siêu tham số cho lần thứ hai là: learning rate = 5e-3, số lượng epoch = 50, kích thước batch = 32, factor = 0.1, patience = 3, độ phân giải của ảnh đầu vào = 224x224. \end{itemize} Ngoài ra, chúng tôi cũng cắt giảm một số hình ảnh trong lớp viêm phổi và bình thường để đảm bảo sự cân bằng về số lượng hình ảnh trong cả 3 lớp của bài toán. Đồ án này chúng tôi đã xây dựng và đánh giá chủ yếu trên tư viện học sâu Keras và framework Tensorflow. % /--------------------------------/ \section{Kết quả thực nghiệm và đánh giá} Để tiến hành kiểm tra chất lượng của các mô hình, chúng tôi đã sử các phương pháp đánh giá như độ chính xác, possitive predictive value (PPV), sensitivity và F1-score. Như trong bảng 1, độ chính xác mô hình VGG-19 của chúng tôi đạt 86\%, kết quả này tốt hơn 3\% so với \cite{covidxdataset}. Tuy nhiên, ResNet50 của chúng tôi thể hiện rất kém, điều này có thể lý giải vì độ phân giải ảnh đầu vào của chúng tôi chỉ là 224x224 (do hạn chế phần cứng), thay vì 480x480. \begin{center} \begin{tabular}{|c|c| c|c| c|c| c|c| c|c|} \hline Kiến trúc & Số lượng tham số (triệu) & Độ chính xác & Độ phân giải ảnh đầu vào\\ \hline VGG19 \cite{covidxdataset} & 20.37 & 83$\%$ & \\ \hline VGG19 & 29.76 trainable + 20.25 non-trainable & 86$\%$ & 480 x 480\\ \hline ResNet50 \cite{covidxdataset} & 24.97 & 90.6$\%$ & \\ \hline ResNet50 (14 epochs) & 25.93 trainable + 23.77 non-trainable & 77$\%$ & 224 x 224\\ \hline ResNet50 (50 epochs) & 25.93 trainable + 23.77 non-trainable & 84$\%$ & 224 x 224 \\ \hline COVID-net \cite{covidxdataset} & 11.75 & 93.3$\%$ & 480 x 480\\ \hline \end{tabular} \end{center} \begin{center} Bảng 1. So sánh độ chính xác và lượng tham số giữa các mô hình \end{center} Tiếp theo, chúng tôi tiến hành so sánh dựa trên thống số sensitivity. Mạng VGG19 của chúng tôi thể hiện vượt trội so với mạng VGG của \cite{covidxdataset} trên độ đo này. Ngoài ra các con số khác trong bảng này trên mạng ResNet50 (epoch) và VGG19 của chúng tôi đều đạt kết quả hết sức khả qua, tất cả đều từ 80$\%$ trở lên. Việc đạt được các kết quả sentivity cao là rất quan trọng bởi điều này nói lên rằng các mô hình bỏ sót rất ít các bệnh nhân trên các lớp. Đây là việc hết sức cần thiết bởi chúng ta không hề muốn một mô hình tuy dự đoán chính xác nhưng lại bỏ sót các trường hợp nhiễm COVID-19 trong bối cảnh dịch bệnh như hiện nay. \begin{center} \begin{tabular}{|c|c| c|c| c|c| c|c| c|c|} \hline Kiến trúc & Không mắc bệnh về đường hô hấp & Viêm phổi & COVID-19\\ \hline VGG19 (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 98\% & 90\% & 58.7\% \\ \hline VGG19 & 96\% & 86\% & 82\%\\ \hline ResNet50 (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 97\% & 92\% & 83\% \\ \hline ResNet50 (14 epochs) & 96\% & 85\% & 67\%\\ \hline ResNet50 (50 epochs) & 86\% & 90\% & 80\%\\ \hline COVID-net (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 95\% & 94\% & 91\%\\ \hline \end{tabular} \end{center} \begin{center} Bảng 2. So sánh các mô hình dựa trên thông số sensitivity \end{center} So sánh trên độ đo PPV cho thấy VGG19 của chúng tôi đạt kết quả rất ấn tượng trên lớp COVID-19 là 99\% và tốt nhất trong số các kết quả trong bảng 3. Tuy nhiên, các kết quả ResNet50 và VGG19 khác của chúng tôi trên các lớp khác cũng chỉ ở mức chấp nhận được, đạt từ 70\% trở lên.\\ \begin{center} \begin{tabular}{|c|c| c|c| c|c| c|c| c|c|} \hline Kiến trúc & Không mắc bệnh về đường hô hấp & Viêm phổi & COVID-19\\ \hline VGG19 (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 83.1\% & 75\% & 98.4\%\\ \hline VGG19 & 70\% & 80\% & 99\%\\ \hline ResNet50 (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 88.2\% & 86.8\% & 98.8\% \\ \hline ResNet50 (14 epochs) & 56\% & 74\% & 97\%\\ \hline ResNet50 (50 epochs) & 73\% & 71\% & 96\%\\ \hline COVID-net (\citeauthor{covidxdataset},\citeyear{covidxdataset}) & 90.5\% & 91.3\% & 98.9\%\\ \hline \end{tabular} \end{center} \begin{center} Bảng 3. So sánh trên thông số PPV giữa các mô hình \end{center} \section{Hướng phát triển} Để tăng tính thuyết phục trong việc ứng dụng vào thực tế, chúng tôi dự định sẽ thực hiện thêm bài toán phát toán phát hiện vật thể. Việc vẽ thêm các bounding box sẽ giúp cho mô hình trở nên đáng tin cậy hơn khi chúng ta chỉ rõ ra được các điểm bất thường trên phổi của bệnh nhân.\\ Bên cạnh đó, hiện nay có nhiều bệnh nhân tuy nhiễm COVID-19 nhưng họ không hề có các triệu chứng điển hình của bệnh này. Việc xếp những người có triệu chứng không điển hình này vào cùng nhóm với những người có triệu chứng điển hình sẽ khiến cho mô hình trở nên kém chính xác và dễ nhầm lẫn với những người có phổi khoẻ mạnh. Do đó, chúng tôi sẽ tách những người nhiễm COVID-19 ra thành 2 lớp riêng biệt là triệu chứng điển và triệu chứng không điển hình.\\ Hơn thế nữa, để tăng độ chính xác chúng tôi cũng sẽ tận dụng các thông tin khác của bệnh nhân như tuổi, giới tính, chúng tộc, cân nặng, các yếu tố dịch tễ... Những thông tin này sẽ giúp ích rất nhiều cho mô hình trong việc tìm ra bệnh nhân nhiễm COVID-19. \newpage \medskip \bibliography{ref} \end{document}\NeedsTeXFormat{LaTeX2e} \ProvidesPackage{inkscape}[02.2021 Little utility for Inkscape exported bundles of PDF and LaTeX outputs.] \RequirePackage{import} \RequirePackage{xifthen} \RequirePackage{pdfpages} \RequirePackage{transparent} \def\inkscaperoot{./} \def\inkscapedefaultwidth{\columnwidth} \newcommand{\inkscapepicture}[2][\inkscapedefaultwidth]{% \def\svgwidth{#1} \import{\inkscaperoot}{#2.pdf_tex} }\hypertarget{struct____attribute____}{}\doxysection{\+\_\+\+\_\+attribute\+\_\+\+\_\+ 구조체 참조} \label{struct____attribute____}\index{\_\_attribute\_\_@{\_\_attribute\_\_}} {\ttfamily \#include $<$ieee80211.\+h$>$} \doxysubsection*{Public 속성} \begin{DoxyCompactItemize} \item unsigned short \mbox{\hyperlink{struct____attribute_____a452081f1fdc6e65cea7b0101e5db940e}{subtype}}\+: 4 \item unsigned short \mbox{\hyperlink{struct____attribute_____a47a641129569257bd0b8582daab1ff15}{type}}\+: 2 \item unsigned short \mbox{\hyperlink{struct____attribute_____acb5957fed2db4d6dcb17dc7b8bdc9b1a}{version}}\+: 2 \item unsigned short \mbox{\hyperlink{struct____attribute_____aff40f775cdf3baff77e051177316fe6a}{order}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a0fa2b8d08904e73fbe74d99afac0da58}{wep}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a16c4ea65c94b5d3d374f670f77f636d9}{more\+\_\+data}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a44e1e2847dd0132c0bdcf0b0be7b9111}{power\+\_\+management}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a47ebcb761d38a10bb6fceb7e4ebee9a8}{retry}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____af6eb8b9bdfb2d2e3a74c67ec8b95d4f2}{more\+\_\+fragments}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a7c4c5409252c8aeaeee60961a917bdfc}{from\+\_\+ds}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____a26effd98feb48d459563423cc4a344aa}{to\+\_\+ds}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a644bb10ea2cb502516333fb68c65fed2}{channel\+\_\+agility}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____ad99a2a7926802cddeb47049e9ca8ee34}{pbcc}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____aab2397bdca03b3419ce53feb84ff881b}{short\+\_\+preamble}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____ae1eef3c285d53319c64eb56f085d203e}{privacy}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____aabdb3c9df8405035cdcb2f785269173f}{cfpoll\+\_\+request}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a3d91298378e16035082408a156ad129b}{cfpoll\+\_\+able}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a5e0ea9092418b2140082eafaac5ff25a}{ibss}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a2056ac5feff4af232e55e37d27a8c30f}{ess}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a8ac5e6430aa82a11eb7695fce039c87d}{rsvd5}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____ace0a3c42d87f4bd1f3f2200e53967154}{rsvd4}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a043f55f8b0079050d93ee85b21972414}{dsss\+\_\+ofdm}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a45c315ee98234249a0b2a19324849562}{rsvd3}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____ab5849ef806e7f791984ae6ec89d6205e}{rsn\+\_\+enabled}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____aa6237fa3ceb6c2f956d18950ff9c32bb}{g\+\_\+short\+\_\+slottime}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a975bfe6d40d9ad2f7352e8eaf96af854}{rsvd2}}\+: 1 \item unsigned char \mbox{\hyperlink{struct____attribute_____a25d59abbc8348ec8c185c52c4bfbe2df}{rsvd1}}\+: 1 \item unsigned short \mbox{\hyperlink{struct____attribute_____aedadd11718438a519e5f8c96f1875484}{seqnum}}\+: 12 \item unsigned short \mbox{\hyperlink{struct____attribute_____aa4486bbe589c34fbf41782e8c038cd24}{flag}}\+: 4 \end{DoxyCompactItemize} \doxysubsection{상세한 설명} ieee80211.\+h 파일의 117 번째 라인에서 정의되었습니다. \doxysubsection{멤버 데이터 문서화} \mbox{\Hypertarget{struct____attribute_____a3d91298378e16035082408a156ad129b}\label{struct____attribute_____a3d91298378e16035082408a156ad129b}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!cfpoll\_able@{cfpoll\_able}} \index{cfpoll\_able@{cfpoll\_able}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{cfpoll\_able}{cfpoll\_able}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::cfpoll\+\_\+able} ieee80211.\+h 파일의 156 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aabdb3c9df8405035cdcb2f785269173f}\label{struct____attribute_____aabdb3c9df8405035cdcb2f785269173f}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!cfpoll\_request@{cfpoll\_request}} \index{cfpoll\_request@{cfpoll\_request}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{cfpoll\_request}{cfpoll\_request}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::cfpoll\+\_\+request} ieee80211.\+h 파일의 155 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a644bb10ea2cb502516333fb68c65fed2}\label{struct____attribute_____a644bb10ea2cb502516333fb68c65fed2}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!channel\_agility@{channel\_agility}} \index{channel\_agility@{channel\_agility}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{channel\_agility}{channel\_agility}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::channel\+\_\+agility} ieee80211.\+h 파일의 151 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a043f55f8b0079050d93ee85b21972414}\label{struct____attribute_____a043f55f8b0079050d93ee85b21972414}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!dsss\_ofdm@{dsss\_ofdm}} \index{dsss\_ofdm@{dsss\_ofdm}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{dsss\_ofdm}{dsss\_ofdm}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::dsss\+\_\+ofdm} ieee80211.\+h 파일의 162 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a2056ac5feff4af232e55e37d27a8c30f}\label{struct____attribute_____a2056ac5feff4af232e55e37d27a8c30f}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!ess@{ess}} \index{ess@{ess}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{ess}{ess}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::ess} ieee80211.\+h 파일의 158 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aa4486bbe589c34fbf41782e8c038cd24}\label{struct____attribute_____aa4486bbe589c34fbf41782e8c038cd24}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!flag@{flag}} \index{flag@{flag}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{flag}{flag}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::flag} ieee80211.\+h 파일의 192 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a7c4c5409252c8aeaeee60961a917bdfc}\label{struct____attribute_____a7c4c5409252c8aeaeee60961a917bdfc}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!from\_ds@{from\_ds}} \index{from\_ds@{from\_ds}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{from\_ds}{from\_ds}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::from\+\_\+ds} ieee80211.\+h 파일의 130 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aa6237fa3ceb6c2f956d18950ff9c32bb}\label{struct____attribute_____aa6237fa3ceb6c2f956d18950ff9c32bb}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!g\_short\_slottime@{g\_short\_slottime}} \index{g\_short\_slottime@{g\_short\_slottime}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{g\_short\_slottime}{g\_short\_slottime}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::g\+\_\+short\+\_\+slottime} ieee80211.\+h 파일의 165 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a5e0ea9092418b2140082eafaac5ff25a}\label{struct____attribute_____a5e0ea9092418b2140082eafaac5ff25a}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!ibss@{ibss}} \index{ibss@{ibss}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{ibss}{ibss}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::ibss} ieee80211.\+h 파일의 157 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a16c4ea65c94b5d3d374f670f77f636d9}\label{struct____attribute_____a16c4ea65c94b5d3d374f670f77f636d9}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!more\_data@{more\_data}} \index{more\_data@{more\_data}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{more\_data}{more\_data}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::more\+\_\+data} ieee80211.\+h 파일의 125 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____af6eb8b9bdfb2d2e3a74c67ec8b95d4f2}\label{struct____attribute_____af6eb8b9bdfb2d2e3a74c67ec8b95d4f2}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!more\_fragments@{more\_fragments}} \index{more\_fragments@{more\_fragments}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{more\_fragments}{more\_fragments}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::more\+\_\+fragments} ieee80211.\+h 파일의 129 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aff40f775cdf3baff77e051177316fe6a}\label{struct____attribute_____aff40f775cdf3baff77e051177316fe6a}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!order@{order}} \index{order@{order}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{order}{order}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::order} ieee80211.\+h 파일의 123 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____ad99a2a7926802cddeb47049e9ca8ee34}\label{struct____attribute_____ad99a2a7926802cddeb47049e9ca8ee34}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!pbcc@{pbcc}} \index{pbcc@{pbcc}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{pbcc}{pbcc}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::pbcc} ieee80211.\+h 파일의 152 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a44e1e2847dd0132c0bdcf0b0be7b9111}\label{struct____attribute_____a44e1e2847dd0132c0bdcf0b0be7b9111}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!power\_management@{power\_management}} \index{power\_management@{power\_management}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{power\_management}{power\_management}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::power\+\_\+management} ieee80211.\+h 파일의 126 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____ae1eef3c285d53319c64eb56f085d203e}\label{struct____attribute_____ae1eef3c285d53319c64eb56f085d203e}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!privacy@{privacy}} \index{privacy@{privacy}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{privacy}{privacy}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::privacy} ieee80211.\+h 파일의 154 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a47ebcb761d38a10bb6fceb7e4ebee9a8}\label{struct____attribute_____a47ebcb761d38a10bb6fceb7e4ebee9a8}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!retry@{retry}} \index{retry@{retry}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{retry}{retry}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::retry} ieee80211.\+h 파일의 128 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____ab5849ef806e7f791984ae6ec89d6205e}\label{struct____attribute_____ab5849ef806e7f791984ae6ec89d6205e}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsn\_enabled@{rsn\_enabled}} \index{rsn\_enabled@{rsn\_enabled}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsn\_enabled}{rsn\_enabled}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsn\+\_\+enabled} ieee80211.\+h 파일의 164 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a25d59abbc8348ec8c185c52c4bfbe2df}\label{struct____attribute_____a25d59abbc8348ec8c185c52c4bfbe2df}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsvd1@{rsvd1}} \index{rsvd1@{rsvd1}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsvd1}{rsvd1}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsvd1} ieee80211.\+h 파일의 167 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a975bfe6d40d9ad2f7352e8eaf96af854}\label{struct____attribute_____a975bfe6d40d9ad2f7352e8eaf96af854}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsvd2@{rsvd2}} \index{rsvd2@{rsvd2}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsvd2}{rsvd2}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsvd2} ieee80211.\+h 파일의 166 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a45c315ee98234249a0b2a19324849562}\label{struct____attribute_____a45c315ee98234249a0b2a19324849562}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsvd3@{rsvd3}} \index{rsvd3@{rsvd3}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsvd3}{rsvd3}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsvd3} ieee80211.\+h 파일의 163 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____ace0a3c42d87f4bd1f3f2200e53967154}\label{struct____attribute_____ace0a3c42d87f4bd1f3f2200e53967154}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsvd4@{rsvd4}} \index{rsvd4@{rsvd4}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsvd4}{rsvd4}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsvd4} ieee80211.\+h 파일의 161 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a8ac5e6430aa82a11eb7695fce039c87d}\label{struct____attribute_____a8ac5e6430aa82a11eb7695fce039c87d}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!rsvd5@{rsvd5}} \index{rsvd5@{rsvd5}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{rsvd5}{rsvd5}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::rsvd5} ieee80211.\+h 파일의 160 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aedadd11718438a519e5f8c96f1875484}\label{struct____attribute_____aedadd11718438a519e5f8c96f1875484}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!seqnum@{seqnum}} \index{seqnum@{seqnum}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{seqnum}{seqnum}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::seqnum} ieee80211.\+h 파일의 191 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____aab2397bdca03b3419ce53feb84ff881b}\label{struct____attribute_____aab2397bdca03b3419ce53feb84ff881b}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!short\_preamble@{short\_preamble}} \index{short\_preamble@{short\_preamble}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{short\_preamble}{short\_preamble}} {\footnotesize\ttfamily unsigned char \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::short\+\_\+preamble} ieee80211.\+h 파일의 153 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a452081f1fdc6e65cea7b0101e5db940e}\label{struct____attribute_____a452081f1fdc6e65cea7b0101e5db940e}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!subtype@{subtype}} \index{subtype@{subtype}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{subtype}{subtype}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::subtype} ieee80211.\+h 파일의 119 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a26effd98feb48d459563423cc4a344aa}\label{struct____attribute_____a26effd98feb48d459563423cc4a344aa}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!to\_ds@{to\_ds}} \index{to\_ds@{to\_ds}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{to\_ds}{to\_ds}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::to\+\_\+ds} ieee80211.\+h 파일의 131 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a47a641129569257bd0b8582daab1ff15}\label{struct____attribute_____a47a641129569257bd0b8582daab1ff15}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!type@{type}} \index{type@{type}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{type}{type}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::type} ieee80211.\+h 파일의 120 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____acb5957fed2db4d6dcb17dc7b8bdc9b1a}\label{struct____attribute_____acb5957fed2db4d6dcb17dc7b8bdc9b1a}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!version@{version}} \index{version@{version}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{version}{version}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::version} ieee80211.\+h 파일의 121 번째 라인에서 정의되었습니다. \mbox{\Hypertarget{struct____attribute_____a0fa2b8d08904e73fbe74d99afac0da58}\label{struct____attribute_____a0fa2b8d08904e73fbe74d99afac0da58}} \index{\_\_attribute\_\_@{\_\_attribute\_\_}!wep@{wep}} \index{wep@{wep}!\_\_attribute\_\_@{\_\_attribute\_\_}} \doxysubsubsection{\texorpdfstring{wep}{wep}} {\footnotesize\ttfamily unsigned short \+\_\+\+\_\+attribute\+\_\+\+\_\+\+::wep} ieee80211.\+h 파일의 124 번째 라인에서 정의되었습니다. 이 구조체에 대한 문서화 페이지는 다음의 파일로부터 생성되었습니다.\+:\begin{DoxyCompactItemize} \item include/\mbox{\hyperlink{ieee80211_8h}{ieee80211.\+h}}\end{DoxyCompactItemize} \begin{figure}[h!t] \centering \includegraphics{./Cpapota_1.pdf} % Cpapota_1.pdf: 0x0 pixel, 0dpi, 0.00x0.00 cm, bb= \caption{Podaire d'un point de la parabole} \label{fig:Cpapota_1} \end{figure} \begin{enumerate} \item \begin{enumerate} \item On peut factoriser $\varphi(t)$ en utilisant des coefficients indéterminés ou une division euclidienne polynomiale. On obtient \begin{displaymath} \varphi(t) = (t+1)(t^3-t^2+3t+1) \end{displaymath} \item Notons $\psi(t) = t^3-t^2+3t+1$. La dérivée $\psi'(t) =3t^2-2t+3$ est une fonction du second degré dont le discriminant est strictement négatif ($=-32$). Elle est à valeurs strictement positives. On en déduit que $\psi$ est strictement croissante. Comme $\psi(-1)=4$ et $\psi(0)=1$, elle s'annule en un unique réel $\alpha\in ]-1,0[$. \end{enumerate} \item L'axe focal de la parabole $\mathcal P$ est l'axe $Ox$ car il contient le sommet et le foyer. Le sommet d'une parabole est à égale distance du foyer et du point d'intersection de l'axe focal avec la directrice (figure \ref{fig:Cpapota_1}). On en déduit que la directrice est la droite d'équation $x=-1$. On en tire l'équation de $\mathcal{P}$ qui est \begin{displaymath} y^2 = 4x \end{displaymath} \item \begin{enumerate} \item Les coordonnées de $M(t)$ sont $(t^2,2t)$. En calculant la vitesse et en utilisant un déterminant, on forme l'équation de la tangente $D_t$. \begin{displaymath} \begin{vmatrix} x-t^2 & 2t \\y-2t & 2 \end{vmatrix} =0 \Leftrightarrow x-ty+t^2 = 0 \end{displaymath} \item Le vecteur $\overrightarrow n$ de coordonnées $(1,-t)$ est normal à $D_t$. Le projeté $N(t)$ de $A$ sur $D_t$ est de la forme $A+\lambda \overrightarrow n$. On calcule $\lambda$ en injectant dans l'équation de $D_t$ l'expression des coordonnées de $N(t)$. On trouve \begin{displaymath} \lambda = -\frac{(1+t)^2}{1+t^2} \end{displaymath} On en déduit \begin{displaymath} \text{coordonnées de }N(t): \left(\frac{-2t}{1+t^2} ,\frac{-2+t+t^3}{1+t^2}\right) \end{displaymath} \end{enumerate} \item \begin{enumerate} \item On définit des fonctions $u$ et $v$ par $u(t)=x(N(t))$ et $v(t)=y(N(t))$. Le calcul des dérivées conduit à : \begin{align*} u'(t)=2\frac{t^2-1}{(1+t^2)^2} & & v'(t)=\frac{t^4+2t^2+4t+1}{(1+t^2)^2}=\frac{(t+1)\psi(t)}{(1+t^2)^2} \end{align*} On en déduit le tableau des variations. \begin{center} \renewcommand{\arraystretch}{1.2} \begin{tabular}{|cccccccccccc|} \hline & $-\infty$ & & $-1$ & & $\alpha$ & & $0$ & & $1$ & & $+\infty$\\ \hline u & $0$ & $\nearrow$ & $1$ & & $\searrow$ & & & & & $\nearrow$ & $0$\\ \hline $v$ & $-\infty$ & $\nearrow$ & $-2$ & $\searrow$ & & & $\nearrow$ & & & & $+\infty$ \\ \hline \end{tabular} \end{center} \item Il existe deux branches infinies pour $t$ en $-\infty$ ou en $+\infty$. Dans chaque cas, la droite d'équation $x=0$ est asymptote.\newline En $-\infty$, la courbe est à droite de l'asymptote. En $+\infty$, la courbe est à gauche de l'asymptote. \item Le point $A\in \mathcal{E}$ car comme il appartient à $\mathcal{P}$, il est son propre projeté orthogonal sur la tangente en $A$. En fait $A=M(-1)=N(-1)$.\newline Il apparait sur le tableau de variations que $A$ est un point stationnaire car $u'(-1)=v'(-1)=0$. On peut factoriser la vitesse: \begin{displaymath} \overrightarrow{N}'(t)=\frac{t+1}{(1+t^2)^2} \left( \underset{=\overrightarrow d(t)}{\underbrace{4(t-1)\overrightarrow i +\psi(t)\overrightarrow j}} \right) \end{displaymath} Le choix de $\overrightarrow d (t)$ comme vecteur directeur de $D_t$ montre que la tangente en $A$ admet $\overrightarrow d(-1)$ comme vecteur directeur. La direction en $A$ est donc $\overrightarrow i + \overrightarrow j$. \end{enumerate} \item \begin{enumerate} \item On se donne trois points arbitraires de coordonnées $(x_1,y_1)$, $(x_2,y_2)$, $(x_3,y_3)$ et on considère le système $\mathcal S$ aux inconnues $a$, $b$, $c$ : \begin{displaymath} \mathcal{S}\hspace{0.5cm} \left\lbrace \begin{aligned} x_1 a + y_1 b +c &=0 \\ x_2 a + y_2 b +c &=0 \\ x_3 a + y_3 b +c &=0 \end{aligned} \right. \end{displaymath} Si les trois points sont alignés, il existe une droite d'équation $ax+by+c=0$ (avec $(a,b)\neq(0,0)$) qui les contient tous les trois. Le système admet donc une solution autre que $(0,0,0)$ et le déterminant doit être nul.\newline Réciproquement, si le déterminant est nul, le système admet une solution $(a,b,c)\neq(0,0,0)$. En fait $(a,b)\neq(0,0)$ car $(a,b)=(0,0)$ entraine $c=0$. On peut donc considérer la droite d'équation $ax+by+c=0$, elle contient les trois points qui sont donc alignés. \item Le calcul se fait chaque fois en utilisant $L_3 \leftarrow L_3 - L_2$ puis $L_2 \leftarrow L_2 - L_1$ (ce qui ne change pas le déterminant). On utilise ensuite la multilinéarite. \begin{multline*} \begin{vmatrix} 1 & t_1 & t_1^2 \\ 1 & t_2 & t_2^2 \\ 1 & t_3 & t_3^2 \end{vmatrix} = \begin{vmatrix} 1 & t_1 & t_1^2 \\ 1 & t_2 & t_2^2 \\ 0 & t_3-t_2 & t_3^2-t_2^2 \end{vmatrix} = \begin{vmatrix} 1 & t_1 & t_1^2 \\ 0 & t_2-t_1 & t_2^2-t_1^2 \\ 0 & t_3-t_2 & t_3^2-t_2^2 \end{vmatrix} \\ = (t_2-t_1)(t_3-t_2) \begin{vmatrix} 1 & t_1 & t_1^2 \\ 0 & 1 & t_2+t_1 \\ 0 & 1 & t_3+t_2 \end{vmatrix} \\ = (t_2-t_1)(t_3-t_2) \begin{vmatrix} 1 & t_1 & t_1^2 \\ 0 & 1 & t_2+t_1 \\ 0 & 0 & t_3+t_1 \end{vmatrix} = (t_2-t_1)(t_3-t_2)(t_3-t_1) \end{multline*} \begin{multline*} \begin{vmatrix} 1 & t_1 & t_1^3 \\ 1 & t_2 & t_2^3 \\ 1 & t_3 & t_3^3 \end{vmatrix} = \begin{vmatrix} 1 & t_1 & t_1^3 \\ 0 & t_2-t_1 & t_2^3-t_1^3 \\ 0 & t_3-t_2 & t_3^3-t_2^3 \end{vmatrix}\\ = (t_2-t_1)(t_3-t_2) \begin{vmatrix} 1 & t_1 & t_1^3 \\ 0 & 1 & t_2^2+t_1t_2+t_1^2 \\ 0 & 1 & t_3^2+t_2t_3+t_2^2 \end{vmatrix} \\ = (t_2-t_1)(t_3-t_2)\left(t_3^2-t_1^2 +t_2(t_3-t_1)\right)\\ = (t_2-t_1)(t_3-t_2)(t_3-t_1)(t_1+t_2+t_3) \end{multline*} \item D'après la question a. on peut caractériser l'alignement par la nullité d'un déterminant que l'on transforme par multilinéarité par rapport aux lignes. Finalement: $N(t_1)$, $N(t_2)$, $N(t_3)$ sont alignés si et seulement si \begin{displaymath} \begin{vmatrix} -2t_1 & -2+t_1+t_1^3 & 1+t_1^2 \\ -2t_2 & -2+t_2+t_2^3 & 1+t_2^2 \\ -2t_3 & -2+t_3+t_3^3 & 1+t_3^2 \\ \end{vmatrix} =0 \end{displaymath} On note respectivement $D$ et $\Delta$ les déterminants calculés en b. On développe ce déterminant par multilinéarité par rapport aux colonnes. On obtient la somme des déterminants \begin{displaymath} D_1= \begin{vmatrix} -2t_1 & -2 & 1 \\ -2t_2 & -2 & 1 \\ -2t_3 & -2 & 1 \\ \end{vmatrix} =0 \end{displaymath} \begin{displaymath} D_2= \begin{vmatrix} -2t_1 & -2 & t_1^2 \\ -2t_2 & -2 & t_2^2 \\ -2t_3 & -2 & t_3^2 \\ \end{vmatrix} =-4 D \end{displaymath} \begin{displaymath} D_3 \begin{vmatrix} -2t_1 & t_1 & 1+t_1^2 \\ -2t_2 & t_2 & 1+t_2^2 \\ -2t_3 & t_3 & 1+t_3^2 \\ \end{vmatrix} =0 \end{displaymath} \begin{displaymath} D_4= \begin{vmatrix} -2t_1 & t_1^3 & 1 \\ -2t_2 & t_2^3 & 1 \\ -2t_3 & t_3^3 & 1 \\ \end{vmatrix} =-2\Delta \end{displaymath} \begin{displaymath} D_5= \begin{vmatrix} -2t_1 & t_1^3 & t_1^2 \\ -2t_2 & t_2^3 & t_2^2 \\ -2t_3 & t_3^3 & t_3^2 \\ \end{vmatrix} =2t_1t_2t_3 D \end{displaymath} En simplifiant par $2D$ qui est non nul car les points sont deux à deux distincts on obtient la relation demandée. \end{enumerate} \item \begin{enumerate} \item Soit $t\in \R\setminus\{-1,+1\}$ fixé. Considérons un $u\neq t$ dans $\R\setminus\{-1,+1\}$. La droite $(N(t),N(u)$ coupe $\mathcal E$ en un point $N(\theta)$ où dépend de $t$ et $u$ selon la formule de la question précédente. Lorsque $u$ tend vers $t$, la direction de la droite $(N(t),N(u))$ tend vers celle de la tangente. La droite $\Delta_t$ recoupe donc $\mathcal{E}$ seulement en $N(\theta)$ avec \begin{displaymath} t^2\theta - 2t -\theta = 2 \Rightarrow \theta = 2\frac{1+t}{t^2-1}=\frac{2}{t-1} \end{displaymath} On peut remarquer qu'en $N(1)$, la tangente est verticale et ne recoupe pas la courbe. \item L'étude des variations des coordonnées montre que la courbe paramétrée $N$ est simple. Un point $N(t)$ est donc son propre tangentiel si et seulement si \begin{displaymath} t=\frac{1}{t-1}\Leftrightarrow t^2-t-1=0\Leftrightarrow t=1\pm\sqrt{5} \end{displaymath} En ces points, la courbe traverse sa tangente. Il s'agit de points d'inflexions. \item Notons $F(t_1,t_2,t_3) = t_1t_2t_3-(t_1+t_2+t_3)-2$ et $\theta_i = \frac{2}{t_i-1}$ pour $i=1,2,3$. Exprimons $F(\theta_1,\theta_2,\theta_3)$: après réduction au même dénominateur, développement et simplification, on obtient \begin{displaymath} F(\theta_1,\theta_2,\theta_3) = -\frac{2F(t_1,t_2,t_3)}{(t_1-1)(t_2-1)(t_3-1)} \end{displaymath} Lorsque $N(t_1)$, $N(t_2)$, $N(t_3)$ sont alignés, \begin{displaymath} F(t_1,t_2,t_3)=0 \Rightarrow F(\theta_1,\theta_2,\theta_3)=0 \end{displaymath} et les trois tangentiels sont alignés (voir fig \ref{fig:Cpapota_2}). \begin{figure}[h!t] \centering \includegraphics{./Cpapota_2.pdf} % Cpapota_2.pdf: 0x0 pixel, 0dpi, 0.00x0.00 cm, bb= \caption{Tangentiels et alignements} \label{fig:Cpapota_2} \end{figure} \end{enumerate} \end{enumerate} % Template for PLoS % Version 1.0 January 2009 % % To compile to pdf, run: % latex plos.template % bibtex plos.template % latex plos.template % latex plos.template % dvipdf plos.template \documentclass[10pt]{article} % amsmath package, useful for mathematical formulas \usepackage{amsmath} % amssymb package, useful for mathematical symbols \usepackage{amssymb} % graphicx package, useful for including eps and pdf graphics % include graphics with the command \includegraphics \usepackage{graphicx} % cite package, to clean up citations in the main text. Do not remove. \usepackage{cite} \usepackage{color} % Use doublespacing - comment out for single spacing \usepackage{setspace} %\doublespacing \onehalfspacing % Text layout \topmargin 0.0cm \oddsidemargin 0.5cm \evensidemargin 0.5cm \textwidth 16cm \textheight 21cm % Bold the 'Figure #' in the caption and separate it with a period % Captions will be left justified \usepackage[labelfont=bf,labelsep=period,justification=raggedright]{caption} % Use the PLoS provided bibtex style \bibliographystyle{plos2009} % Remove brackets from numbering in List of References \makeatletter \renewcommand{\@biblabel}[1]{\quad#1.} \makeatother % Leave date blank \date{} \pagestyle{myheadings} %% ** EDIT HERE ** \usepackage{url} \usepackage{longtable} \usepackage{fixltx2e} \usepackage{lineno} %% ** EDIT HERE ** %% PLEASE INCLUDE ALL MACROS BELOW %DIF 65a65 \newcommand{\R}[1]{\label{#1}\linelabel{#1}} %DIF > %DIF ------- %% END MACROS SECTION %DIF PREAMBLE EXTENSION ADDED BY LATEXDIFF %DIF UNDERLINE PREAMBLE %DIF PREAMBLE \RequirePackage[normalem]{ulem} %DIF PREAMBLE \RequirePackage{color}\definecolor{RED}{rgb}{1,0,0}\definecolor{BLUE}{rgb}{0,0,1} %DIF PREAMBLE \providecommand{\DIFadd}[1]{{\protect\color{blue}\uwave{#1}}} %DIF PREAMBLE \providecommand{\DIFdel}[1]{{\protect\color{red}\sout{#1}}} %DIF PREAMBLE %DIF SAFE PREAMBLE %DIF PREAMBLE \providecommand{\DIFaddbegin}{} %DIF PREAMBLE \providecommand{\DIFaddend}{} %DIF PREAMBLE \providecommand{\DIFdelbegin}{} %DIF PREAMBLE \providecommand{\DIFdelend}{} %DIF PREAMBLE %DIF FLOATSAFE PREAMBLE %DIF PREAMBLE \providecommand{\DIFaddFL}[1]{\DIFadd{#1}} %DIF PREAMBLE \providecommand{\DIFdelFL}[1]{\DIFdel{#1}} %DIF PREAMBLE \providecommand{\DIFaddbeginFL}{} %DIF PREAMBLE \providecommand{\DIFaddendFL}{} %DIF PREAMBLE \providecommand{\DIFdelbeginFL}{} %DIF PREAMBLE \providecommand{\DIFdelendFL}{} %DIF PREAMBLE %DIF END PREAMBLE EXTENSION ADDED BY LATEXDIFF \begin{document} % Title must be 150 characters or less \begin{flushleft} {\Large \textbf{ss3sim: An R package for fisheries stock assessment simulation with Stock Synthesis} } % Insert Author names, affiliations and corresponding author email. \\ $^{1,\ast}$, $^{2}$\DIFaddbegin \DIFadd{, }\DIFaddend $^{3}$, $^{3}$\DIFaddbegin \DIFadd{, }\DIFaddend $^{4}$ \\ \bf{1} Department of Biological Sciences, Simon Fraser University, Burnaby BC, V5A 1S6, Canada \\ \bf{2} Quantitative Ecology and Resource Management, University of Washington, Box 352182, Seattle, WA 98195-2182, USA \\ \bf{3} School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA 98195-5020, USA \\ \bf{4} Center for the Advancement of Population Assessment Methodology (CAPAM), 8901 La Jolla Shores Drive, La Jolla, CA 92037, USA \\ $\ast$ E-mail: \end{flushleft} \linenumbers \DIFdelbegin %DIFDELCMD < \modulolinenumbers[2] %DIFDELCMD < %%% \DIFdelend \DIFaddbegin \modulolinenumbers[1] \DIFaddend \section*{Abstract} Simulation testing is an important approach to evaluating fishery stock assessment methods. In the last decade, the fisheries stock assessment modeling framework Stock Synthesis (SS3) has become \DIFdelbegin \DIFdel{widely-used }\DIFdelend \DIFaddbegin \DIFadd{widely used }\DIFaddend around the world. However, there lacks a generalized and scriptable framework for SS3 simulation testing. Here, we introduce ss3sim, an R package that facilitates \DIFdelbegin \DIFdel{large-scale, rapid, and reproducible }\DIFdelend \DIFaddbegin \R{B3:2}\DIFadd{reproducible, flexible, and rapid }\DIFaddend end-to-end simulation testing with SS3. ss3sim requires an existing SS3 model configuration along with plain-text control files describing alternative population dynamics, fishery properties, sampling scenarios, and assessment approaches. ss3sim then generates an underlying \DIFdelbegin \DIFdel{truth}\DIFdelend \DIFaddbegin \DIFadd{`truth' }\R{B4}\DIFadd{from a specified operating model}\DIFaddend , samples from that truth, modifies and runs an estimation model, and synthesizes the results. The simulations can be run in parallel, \DIFdelbegin \DIFdel{speeding computation}\DIFdelend \DIFaddbegin \R{B16:2}\DIFadd{reducing runtime}\DIFaddend , and the source code is free to be modified under an open-source MIT license. ss3sim is designed to explore structural differences between the underlying truth and assumptions of an estimation model, or between multiple estimation model configurations. For example, ss3sim can be used to answer questions about model misspecification, retrospective patterns, and the relative importance of different types of fisheries data. We demonstrate the software with \DIFdelbegin \DIFdel{a simple }\DIFdelend \DIFaddbegin \DIFadd{an }\DIFaddend example, discuss how ss3sim complements other simulation software, and outline specific research questions that ss3sim could address. \DIFdelbegin %DIFDELCMD < \clearpage %DIFDELCMD < %DIFDELCMD < %%% \DIFdelend \section*{Introduction} Fisheries stock assessment models are \DIFdelbegin \DIFdel{crucial to }\DIFdelend \DIFaddbegin \R{B5}\DIFadd{an invaluable tool for }\DIFaddend providing scientific advice \DIFdelbegin \DIFdel{and to }\DIFdelend \DIFaddbegin \R{B6}\DIFadd{regarding stock status, historical productivity, and changes in stock composition as well as }\DIFaddend evaluating the impact of alternative management actions on fishery resources \cite{gulland1983, hilborn1992}. Although a variety of stock assessment approaches are available, it is often not straightforward to \DIFdelbegin \DIFdel{choose among competing approaches }\DIFdelend \DIFaddbegin \R{B7}\DIFadd{select among competing alternatives }\DIFaddend that may lead to different \DIFdelbegin \DIFdel{modeling outcomes }\DIFdelend \DIFaddbegin \R{B8}\DIFadd{conclusions about stock status }\DIFaddend and associated scientific advice to management. Simulation testing is a critical component to \DIFdelbegin \DIFdel{testing }\DIFdelend \DIFaddbegin \DIFadd{understanding the behavior of }\DIFaddend fishery stock assessment methods, particularly given the potential for model misspecification \DIFdelbegin %DIFDELCMD < \cite{hilborn1987, hilborn1992, rosenberg1994, peterman2004, deroba2013a}%%% \DIFdelend \DIFaddbegin \cite{hilborn1987, hilborn1992, rosenberg1994, peterman2004, deroba2014}\DIFaddend . With simulation testing we can evaluate the precision and bias of alternative assessment \DIFdelbegin \DIFdel{methods }\DIFdelend \DIFaddbegin \DIFadd{approaches }\DIFaddend in a controlled environment where we know the true dynamics of \DIFaddbegin \DIFadd{hypothetical }\DIFaddend fisheries resources under exploitation. Recent simulation studies have been key to improving \DIFdelbegin \DIFdel{strategies }\DIFdelend \DIFaddbegin \R{B9}\DIFadd{structural assumptions }\DIFaddend for dealing with, for example, time-varying natural mortality ($M$) \DIFdelbegin %DIFDELCMD < \cite{lee2011, jiao2012, deroba2013, johnson2013} %%% \DIFdelend \DIFaddbegin \cite{lee2011, jiao2012, deroba2013, johnson2014}\DIFadd{, }\R{B10}\DIFaddend uncertainty in steepness of the stock-recruit relationship \cite{lee2012}, and environmental variability \cite{schirripa2009}, as well as determining \DIFdelbegin \DIFdel{what makes fisheries data informative }%DIFDELCMD < \cite{magnusson2007, wetzel2011a, ono2013, yin2004}%%% \DIFdelend \DIFaddbegin \R{B11}\DIFadd{the utility and influence on assessment outcomes of various fishery-dependent and -independent data sources }\cite{magnusson2007, wetzel2011a, ono2014, yin2004}\DIFaddend . \DIFaddbegin \DIFadd{There }\R{E1}\DIFadd{is a suite of tools available for conducting fishery stock assessments and }\DIFaddend Stock Synthesis (SS3, the third version of the software) is \DIFdelbegin \DIFdel{a }\DIFdelend \DIFaddbegin \DIFadd{one, }\DIFaddend widely-used\DIFdelbegin \DIFdel{fisheries stock assessment }\DIFdelend \DIFaddbegin \DIFadd{, }\DIFaddend modeling framework \cite{methot2013}. SS3 implements statistical age-structured population modeling \DIFdelbegin \DIFdel{, }\DIFdelend using a wide range of \DIFdelbegin \DIFdel{minimally-processed }\DIFdelend \DIFaddbegin \DIFadd{minimally processed }\DIFaddend data \cite{maunder2013, methot2013}. \DIFdelbegin \DIFdel{By using this generalized framework, individuals conducting fisheries stock assessments and peer reviewers can focus on the underlying science, instead of the model code }%DIFDELCMD < \cite{methot2013}%%% \DIFdel{.}\DIFdelend \DIFaddbegin \R{A1:1}\DIFadd{The generalized model structure of SS3 allows flexible scaling to a variety of data and life-history situations, }\R{A1:3}\DIFadd{from data-poor (e.g.~}\cite{wetzel2011a, cope2013}\DIFadd{) to data-rich (e.g.~}\cite{haltuch2013}\DIFadd{). }\DIFaddend Owing in part to these advantages, \DIFaddbegin \R{A3}\DIFaddend SS3 \DIFdelbegin \DIFdel{is one of the world's most commonly-used stock assessment modelling frameworks, particularly in the United States and Australia}\DIFdelend \DIFaddbegin \DIFadd{has been used worldwide to formally assess 61 fishery stocks by 2012: 35 stocks in the US, 10 tuna/billfish stocks in three oceans, four European stocks, and 12 Australian stocks }\cite{methot2013}\DIFadd{. These assessments are conducted by both national agencies (e.g.~NOAA in the USA, CSIRO in Australia) as well as regional fisheries management organizations (e.g.~IATTC, ICCAT, IOTC in the Pacific, Atlantic and Indian oceans respectively). In addition to completed formal stock assessments}\DIFaddend , \DIFdelbegin \DIFdel{where it has been used in 35 and 12 stock assessmentsas of 2012, respectively }\DIFdelend \DIFaddbegin \DIFadd{exploratory SS3 applications for many other stocks are underway }\DIFaddend \cite{methot2013}. \DIFaddbegin \DIFaddend Stock Synthesis is also commonly used as a framework for stock assessment simulation testing \cite{helu2000, yin2004, schirripa2009, lee2011, jiao2012, lee2012, crone2013a, hurtadoferro2013}\DIFdelbegin \DIFdel{. }%DIFDELCMD < %DIFDELCMD < %%% \DIFdel{Although SS3 is increasingly becoming a standard for fisheries stock assessment, }\DIFdelend \DIFaddbegin \DIFadd{, }\R{B12}\DIFadd{but }\DIFaddend there lacks a generalized \DIFdelbegin \DIFdel{framework for simulation }\DIFdelend \DIFaddbegin \DIFadd{structure for simulation for }\DIFaddend testing with SS3. As a result, most stock assessment simulation-testing work \DIFaddbegin \DIFadd{using SS3 }\DIFaddend to date has \DIFdelbegin \DIFdel{used custom frameworks }\DIFdelend \DIFaddbegin \DIFadd{relied on custom frameworks }\cite{helu2000, yin2004, magnusson2007, wetzel2011a, jiao2012, wilberg2006, deroba2013, deroba2014, crone2013a, hurtadoferro2013}\DIFadd{. }\R{A1:2}\DIFadd{Although custom-designed modeling frameworks can be }\DIFaddend tailored to the \DIFdelbegin \DIFdel{particular needs of each study}%DIFDELCMD < \cite{helu2000, yin2004, magnusson2007, wetzel2011a, jiao2012, wilberg2006, deroba2013a, deroba2013, crone2013a, hurtadoferro2013}%%% \DIFdel{. }\DIFdelend \DIFaddbegin \DIFadd{specific needs of a particular stock assessment or simulation study, the use of a generalized framework allows other scientists to validate, reuse, and build upon previous work, thereby improving efficiency and resulting in more reliable outcomes. } \DIFaddend The programming language R \cite{rcoreteam2013} is an ideal language \DIFaddbegin \DIFadd{in which }\DIFaddend to write such a generalized framework \DIFdelbegin \DIFdel{in }\DIFdelend because (1) R has become the standard for statistical computing and visualization and (2) the R package \DIFdelbegin \DIFdel{\texttt{r4ss} }\DIFdelend \DIFaddbegin \DIFadd{r4ss }\DIFaddend \cite{r4ss2013} facilitates reading, processing, and plotting of SS3 model output. \DIFdelbegin %DIFDELCMD < %DIFDELCMD < %%% \DIFdelend Here we introduce ss3sim, an R package that facilitates \DIFdelbegin \DIFdel{large-scale, rapid, and reproducible }\DIFdelend \DIFaddbegin \R{B3:1}\DIFadd{reproducible, flexible, and rapid }\DIFaddend end-to-end simulation testing with the widely-used SS3 framework. We begin by outlining the general structure of ss3sim and describing its functions, and then demonstrate the software with a simple example. We conclude by discussing how ss3sim complements other simulation testing software and by outlining some research questions that our freely accessible and general \DIFdelbegin \DIFdel{SS3 }\DIFdelend simulation-testing framework could address. \section*{The ss3sim framework} \subsection*{Design goals of ss3sim} \DIFaddbegin \R{B3:3}\DIFaddend We designed ss3sim \DIFaddbegin \R{B13}\DIFadd{simulations }\DIFaddend to be reproducible, flexible, and rapid. \emph{Reproducible}: ss3sim simulations are produced using R code, plain-text control files, and SS3 model configurations. ss3sim also allows for random seeds to be set when generating observation and process error. In combination, these features make simulations repeatable across computers and operating systems (Windows, OS X, and Linux). \emph{Flexible}: ss3sim inherits the flexibility of SS3 and can therefore implement many available stock assessment configurations by either modifying existing SS3 model configurations or by modifying \DIFdelbegin \DIFdel{built-in }\DIFdelend generic life-history model configurations \DIFaddbegin \DIFadd{that are built into ss3sim }\DIFaddend (Text S1). Furthermore, ss3sim summarizes the simulation output into plain-text comma-separated-value (\texttt{.csv}) files allowing the output to be processed in R or other statistical software. Finally, the ss3sim \DIFaddbegin \DIFadd{source }\DIFaddend code is written under an open-source MIT license and can be freely modified. \emph{Rapid}: ss3sim relies on SS3, which uses AD Model \DIFdelbegin \DIFdel{Builder as a back-end optimization platform }%DIFDELCMD < \cite{fournier2012}{]} %%% \DIFdelend \DIFaddbegin \R{B14}\DIFadd{Builder }\cite{fournier2012} \DIFaddend --- \DIFdelbegin \DIFdel{the most }\DIFdelend \DIFaddbegin \R{B15}\DIFadd{a }\DIFaddend rapid and robust non-linear optimization software \cite{bolker2013} \DIFaddbegin \DIFadd{--- as a back-end optimization platform}\DIFaddend . ss3sim also facilitates the deployment of simulations across multiple computers or computer cores (i.e.~parallelization), thereby \DIFdelbegin \DIFdel{accelerating computation}\DIFdelend \DIFaddbegin \R{B16}\DIFadd{reducing runtime}\DIFaddend . By using the vetted SS3 framework \DIFdelbegin %DIFDELCMD < \cite{methot2013} %%% \DIFdelend with the tested ss3sim package\DIFdelbegin %DIFDELCMD < \cite{johnson2013, ono2013}%%% \DIFdelend , the time to develop \DIFaddbegin \DIFadd{and run }\DIFaddend a large-scale simulation \DIFaddbegin \DIFadd{study }\DIFaddend can be reduced substantially, \DIFdelbegin \DIFdel{shifting focus to the research questions themselves}\DIFdelend \DIFaddbegin \R{B17}\DIFadd{allowing for more time to refine research questions and interpret results instead of spending it developing and testing custom simulation frameworks}\DIFaddend . \subsection*{The general structure of an ss3sim simulation} ss3sim consists of both low-level functions that modify SS3 configuration files and high-level functions that combine these low-level functions into a complete simulation experiment (Figure 1, Table 1). In this paper we will focus on the structure and use of the high-level function \texttt{run\_ss3sim}; however, the low-level functions can be used on their own as part of a customized simulation \DIFaddbegin \R{B18}\DIFadd{(see Text S1)}\DIFaddend . An ss3sim simulation requires three types of input: (1) a base SS3 model configuration describing the underlying true population dynamics, or operating model (OM); (2) a base SS3 model configuration to assess \DIFdelbegin \DIFdel{that truth based on data generated using }\DIFdelend \DIFaddbegin \DIFadd{the observed data generated by }\DIFaddend the OM, also known as the estimation model or method (EM); and (3) a set of plain-text files (case files) describing alternative model configurations and deviations from these base models (e.g.~different fishing mortality or $M$ trajectories\DIFaddbegin \DIFadd{; }\R{B21:1}\DIFadd{Figure 2}\DIFaddend ). We refer to each unique combination of OM, EM, and case files as a scenario. Scenarios are usually run for multiple iterations \DIFdelbegin \DIFdel{, possibly adding }\DIFdelend \DIFaddbegin \DIFadd{with }\DIFaddend unique process and observation error \DIFdelbegin \DIFdel{to the OM each time}\DIFdelend \DIFaddbegin \DIFadd{in each iteration}\DIFaddend . An ss3sim simulation therefore refers to the combination of all scenarios and iterations. The \texttt{run\_ss3sim} function works by modifying SS3 configuration files as specified in the case-file arguments (\texttt{change} functions), running the OM, sampling from the time-series of true population dynamics to generate \DIFdelbegin \DIFdel{a }\DIFdelend \DIFaddbegin \DIFadd{an observed }\DIFaddend dataset (\texttt{sample} functions), running the EM to get maximum-likelihood estimates \DIFaddbegin \DIFadd{and standard errors }\DIFaddend of parameters and \DIFdelbegin \DIFdel{to derive }\DIFdelend \DIFaddbegin \R{B19}\DIFadd{derived }\DIFaddend quantities, and synthesizing the output for easy data manipulation and visualization (\texttt{get} functions) (Figure 1). \section*{An example simulation with ss3sim} To demonstrate ss3sim, we will work through a simple example in which we examine the effect of (1) high vs.~low precision of a fishery independent index of abundance and (2) \DIFdelbegin \DIFdel{fixing }\DIFdelend \DIFaddbegin \R{B20}\DIFadd{fixing $M$ at an assumed value }\DIFaddend vs.~estimating $M$. All files to run this example are included in the package data, and a more detailed description is available in the accompanying vignette (Text S1). ss3sim requires R version 3.0.0 or greater and SS3 (see Text S1 for more detailed instructions). In R, \DIFdelbegin \DIFdel{the development version of }\DIFdelend ss3sim can be installed \DIFaddbegin \DIFadd{and loaded }\DIFaddend with: \DIFdelbegin %DIFDELCMD < \begin{verbatim}%DIFDELCMD < %DIFDELCMD < install.packages(devtools) %DIFDELCMD < devtools::install_github("ss3sim", username = "seananderson", %DIFDELCMD < dependencies = TRUE) %DIFDELCMD < \end{verbatim} %DIFDELCMD < %%% \DIFdelend \DIFaddbegin \begin{verbatim} install.packages("ss3sim") library("ss3sim") \end{verbatim} \DIFaddend \noindent \DIFaddbegin \DIFadd{Alternatively, the development version of ss3sim can be installed from }\url{https://github.com/ss3sim/ss3sim}\DIFadd{. }\DIFaddend You can read the documentation and \DIFdelbegin \DIFdel{open the }\DIFdelend vignette (Text S1) with: \DIFdelbegin %DIFDELCMD < \begin{verbatim}%DIFDELCMD < %DIFDELCMD < ?ss3sim %DIFDELCMD < help(package = "ss3sim") %DIFDELCMD < vignette("ss3sim-vignette") %DIFDELCMD < \end{verbatim} %DIFDELCMD < %%% \DIFdelend \DIFaddbegin \begin{verbatim} ?ss3sim vignette("ss3sim-vignette") \end{verbatim} \DIFaddend \subsection*{Setting up the SS3 model configurations} ss3sim comes with built-in SS3 model configurations that represent three general life histories: cod-like (slow-growing and long-lived), flatfish-like (fast-growing and long-lived), and sardine-like (fast-growing and short-lived). These model configurations are based on North Sea cod (\emph{Gadus morhua}; , \DIFaddbegin \DIFadd{NMFS, NOAA; }\DIFaddend pers.~comm.), yellowtail flounder (\emph{Limanda ferruginea}; , \DIFdelbegin \DIFdel{pers.~}\DIFdelend \DIFaddbegin \DIFadd{NMFS, NOAA; pers. }\DIFaddend comm.), and Pacific sardine (\emph{Sardinops sagax caeruleus}) \cite{hill2012} (Text S1). We recommend modifying these built-in model configurations to match a desired scenario, although it is possible to \DIFdelbegin \DIFdel{modify }\DIFdelend \DIFaddbegin \DIFadd{create a new ss3sim model by modifying }\DIFaddend an existing SS3 model \DIFdelbegin \DIFdel{configurations to work with ss3sim }\DIFdelend \DIFaddbegin \DIFadd{configuration }\DIFaddend (Text S1). We will base our example around the built-in cod-like model setup. \subsection*{Setting up the case files} The high-level function \texttt{run\_ss3sim} can run all simulation steps based on a specified scenario ID and a set of semicolon-delimited plain-text files that describe alternative cases \DIFdelbegin \DIFdel{(Figure }\DIFdelend \DIFaddbegin \R{B21:2}\DIFadd{(Figures }\DIFaddend 1 \DIFaddbegin \DIFadd{and 2}\DIFaddend ). These \DIFaddbegin \DIFadd{case }\DIFaddend files contain argument values that \DIFdelbegin \DIFdel{will be }\DIFdelend \DIFaddbegin \DIFadd{are }\DIFaddend passed to the low-level ss3sim R functions (e.g.~\texttt{change\_index}, a function that controls how the fishery and survey indices are sampled; Table 1). To use \texttt{run\_ss3sim}\DIFaddbegin \DIFadd{, }\DIFaddend all case files must be named according to the type of case (e.g.~\texttt{E} for estimation or \texttt{F} for fishing mortality), a numeric value representing the case number, and an alphanumeric identifier representing the species or stock (e.g.~\texttt{cod}; Table 1, Text S1). We combine these case IDs with hyphens to create scenario IDs. For example, one of our scenarios will have the scenario ID \texttt{D1-E0-F0-M0-R0-cod}. This scenario ID tells \texttt{run\_ss3sim} to read the case files corresponding to the first data (\texttt{D}) case (i.e.~\texttt{index1-cod.txt}, \texttt{lcomp1-cod.txt}, \texttt{agecomp1-cod.txt\DIFdelbegin %DIFDELCMD < }%%% \DIFdelend )\DIFaddbegin }\DIFaddend , the zero case for estimation (\texttt{E}; i.e.~\texttt{E0-cod.txt\DIFdelbegin %DIFDELCMD < }%%% \DIFdelend )\DIFaddbegin }\DIFaddend , and so on. To investigate the effect of different levels of precision of a \DIFdelbegin \DIFdel{fishery independent }\DIFdelend \DIFaddbegin \DIFadd{fishery-independent }\DIFaddend index of abundance, we will manipulate the argument \texttt{sds\_obs} that gets passed to the function \texttt{change\_index}. In data case \DIFdelbegin \DIFdel{0}\DIFdelend \DIFaddbegin \DIFadd{\texttt{D0}}\DIFaddend , we will specify the standard deviation of the index of abundance at 0.1. and in case \DIFdelbegin \DIFdel{1 }\DIFdelend \DIFaddbegin \DIFadd{\texttt{D1} }\DIFaddend we will increase the standard deviation to 0.4. We can do this by including the line: \texttt{sds\_obs; list(0.1)} in the file \texttt{index0-cod.txt} and the line: \texttt{sds\_obs; list(0.4)} in the file \texttt{index1-cod.txt}. We will also set up a base-case file describing fishing mortality (\texttt{F0-cod.txt}), a file describing a stationary $M$ trajectory (\texttt{M0-cod.txt}), and specify that we do not want to run a retrospective analysis in the file \texttt{R0-cod.txt}. We will set up the file \texttt{E0-cod.txt} to fix \DIFdelbegin \DIFdel{the estimation of }\DIFdelend $M$ at the true value and \DIFaddbegin \DIFadd{not estimate it, and }\DIFaddend case \texttt{E1-cod.txt} to estimate a stationary\DIFaddbegin \DIFadd{, time-invariant }\DIFaddend $M$ (Text S1). All of these text files are available in the package data in the folder \texttt{inst/extdata/eg-cases/}. As an example, here is what the complete \texttt{index0-cod.txt} file looks like: \begin{verbatim} fleets; 2 years; list(seq(1974, 2012, by = 2)) sds_obs; list(0.1) \end{verbatim} \noindent \texttt{fleets}, \texttt{years}, and \texttt{sds\_obs} refer to the arguments in the function \texttt{change\_index} and users can read the help for this function with \texttt{?change\_index} in R. \DIFdelbegin \DIFdel{To start, we will load the }\DIFdelend \DIFaddbegin \subsection*{\DIFadd{Validating the simulation setup}} \R{A4:1}\DIFadd{Before running and interpreting the results of a simulation, it is important to validate the testing framework at several levels. First, it is important to test that the functions that manipulate model configurations (i.e.~the \texttt{change} functions) are set up properly. }\DIFaddend ss3sim \DIFdelbegin \DIFdel{package into an R session and locate three sets of folders within the package data: the folder with the OM, the folder with the EM, and the folder with the plain-text case files: }\DIFdelend \DIFaddbegin \DIFadd{comes with prepackaged models that have been tested extensively with the \texttt{change} functions, as well as documented R functions that include examples and unit tests. We describe strategies for testing the \texttt{change} functions on new SS3 model setups in Text S1. }\DIFaddend \DIFdelbegin %DIFDELCMD < \begin{verbatim}%DIFDELCMD < %DIFDELCMD < library(ss3sim) %DIFDELCMD < d <- system.file("extdata", package = "ss3sim") %DIFDELCMD < om <- paste0(d, "/models/cod-om") %DIFDELCMD < em <- paste0(d, "/models/cod-em") %DIFDELCMD < case_folder <- paste0(d, "/eg-cases") %DIFDELCMD < \end{verbatim} %DIFDELCMD < %DIFDELCMD < %%% \subsection*{\DIFdel{Running the simulations}} %DIFAUXCMD %DIFDELCMD < %DIFDELCMD < %%% \DIFdel{It is important to validate a simulation testing framework with minimal or no }\DIFdelend \DIFaddbegin \DIFadd{Second, the components of the simulation framework must work together as expected (integration tests }\cite{wilson2014}\DIFadd{). One approach to testing for such issues is to run simulation tests with similar OM and EM setups and relatively low }\DIFaddend process and observation error \DIFdelbegin \DIFdel{to ensure unbiased and consistent recovery of parameters under ideal conditions }%DIFDELCMD < \cite{hilborn1992, rykiel1996}%%% \DIFdelend \DIFaddbegin \cite{hilborn1992}\DIFaddend . ss3sim makes this form of \DIFdelbegin \DIFdel{model }\DIFdelend validation simple by allowing users to specify \DIFdelbegin \DIFdel{process error (i.e.~recruitment deviations) and sampling }\DIFdelend \DIFaddbegin \DIFadd{levels of process and observation }\DIFaddend error (Text S1). \DIFdelbegin \DIFdel{Since, }\DIFdelend \DIFaddbegin \R{A4:2}\DIFadd{Assuming that the user specifies sufficient error to avoid numerical instability, this approach can reveal issues that would otherwise be obscured by noise. } \DIFadd{Finally, it is important to validate that the model-fitting algorithms converged to global maxima. ss3sim retains all SS3 model output for future examination, as well as performance diagnostics such as maximum gradient, whether or not the covariance matrix was successfully calculated, run time, and the number of parameters stuck on bounds. These metrics, in combination with visual checks, are useful to determine if the results of a study are robust and meaningful. } \subsection*{\DIFadd{Running the simulations}} \DIFadd{Since }\DIFaddend we have already validated the cod-like model setup (Text S1), we can now run our \DIFaddbegin \DIFadd{example }\DIFaddend simulation scenario. \DIFdelbegin \DIFdel{We }\DIFdelend \DIFaddbegin \DIFadd{To start, we will locate three sets of folders within the package data: the folder with the OM, the folder with the EM, and the folder with the plain-text case files: } \begin{verbatim} d <- system.file("extdata", package = "ss3sim") om <- paste0(d, "/models/cod-om") em <- paste0(d, "/models/cod-em") case_folder <- paste0(d, "/eg-cases") \end{verbatim} \DIFadd{We can then run the simulation with one call to the \texttt{run\_ss3sim} function. We }\DIFaddend will set \texttt{bias\_adjust = TRUE} to enable a procedure that aims to produce mean-unbiased estimates of recruitment and biomass despite log-normal recruitment deviations \cite{methot2011}. We can run 100 iterations of the simulation scenarios with the following code: \begin{verbatim} run_ss3sim(iterations = 1:100, scenarios = c("D0-E0-F0-M0-R0-cod", "D1-E0-F0-M0-R0-cod", "D0-E1-F0-M0-R0-cod", "D1-E1-F0-M0-R0-cod"), case_folder = case_folder, om_model_dir = om, em_model_dir = em, bias_adjust = TRUE) \end{verbatim} \DIFaddbegin \noindent \DIFaddend This produces a folder structure in our working directory containing all of the SS3 output files \DIFaddbegin \R{B21:3}\DIFadd{(Figure 2)}\DIFaddend . We can then collect the output with one function call: \begin{verbatim} get_results_all() \end{verbatim} \noindent This command creates two files in our working directory: \texttt{ss3sim\_scalars.csv} and \texttt{ss3sim\_ts.csv}, which contain scalar output estimates \DIFdelbegin \DIFdel{(e.g.~}\DIFdelend \DIFaddbegin \R{B22}\DIFadd{(model parameters and derived quantities such as }\DIFaddend steepness and maximum sustainable yield) and time-series estimates (e.g.~recruitment and biomass \DIFaddbegin \DIFadd{for }\DIFaddend each year). These estimates come from the report files produced from each run of SS3 \DIFdelbegin \DIFdel{and are read by the \texttt{r4ss} }\DIFdelend \DIFaddbegin \DIFadd{as extracted by the r4ss }\DIFaddend R package. The \texttt{.csv} files contain separate columns for OM and EM values, making it simple to calculate error metrics, such as relative or absolute error. In addition to parameter estimates, the \texttt{.csv} files contain performance metrics, \DIFdelbegin \DIFdel{such as the maximum gradient, whether the covariance matrix was successfully calculated, and the number of parameters stuck on a bound, }\DIFdelend which in combination can be used to gauge model performance and convergence. These results are organized into ``long'' data format, with columns for scenario and iteration, facilitating quick analysis and plotting using common R packages such as \DIFdelbegin \DIFdel{\texttt{ggplot2} }\DIFdelend \DIFaddbegin \DIFadd{\textbf{ggplot2} }\DIFaddend \cite{wickham2009}. For \DIFdelbegin \DIFdel{our }\DIFdelend \DIFaddbegin \DIFadd{the }\DIFaddend example simulation, the relative error in spawning stock biomass over time is, as expected, smaller when the true value of $M$ is specified rather than estimated (Figure \DIFdelbegin \DIFdel{2}\DIFdelend \DIFaddbegin \DIFadd{3}\DIFaddend , top panels E0 vs.~E1). Furthermore, lower precision in the research survey index of abundance results in greater relative error in spawning stock biomass in recent years (Figure \DIFdelbegin \DIFdel{2}\DIFdelend \DIFaddbegin \DIFadd{3}\DIFaddend , top panels D0 vs.~D1), and greater relative error in terminal-year depletion \DIFaddbegin \DIFadd{(the ratio of terminal year spawning biomass to unfished spawning biomass) }\DIFaddend and fishing mortality, but not in spawning stock biomass at maximum sustainable yield, or $M$ (Figure \DIFdelbegin \DIFdel{2}\DIFdelend \DIFaddbegin \DIFadd{3}\DIFaddend , lower panels). \section*{How ss3sim complements other simulation software} The general purpose of ss3sim is to explore model behaviour and performance across combinations of EM configurations and alternative dynamics of fisheries resources under exploitation specified by the OM. In particular, ss3sim provides a suite of functions for dynamically creating structural differences in both OMs and EMs. This expedites testing the properties of alternative stock assessment model configurations, whether the differences are between OMs and EMs \DIFdelbegin %DIFDELCMD < \cite{johnson2013}%%% \DIFdelend \DIFaddbegin \cite{johnson2014}\DIFaddend , or between multiple versions of EMs \DIFdelbegin %DIFDELCMD < \cite{ono2013}%%% \DIFdelend \DIFaddbegin \cite{ono2014}\DIFaddend . However, ss3sim is less suited for quickly exploring \DIFdelbegin \DIFdel{arbitrary }\DIFdelend \DIFaddbegin \DIFadd{new }\DIFaddend SS3 model setups, which may rely on SS3 configurations not yet \DIFdelbegin \DIFdel{programmed into }\DIFdelend \DIFaddbegin \DIFadd{converted to work with }\DIFaddend the ss3sim package functions. \DIFdelbegin \DIFdel{Therefore, depending on the simulation study goal, }\DIFdelend \DIFaddbegin \R{B23}\DIFadd{Although it is possible to adapt arbitrary SS3 models to work with ss3sim (Text S1), }\DIFaddend other software frameworks may provide better alternatives\DIFaddbegin \DIFadd{, depending on the goal of the simulation study}\DIFaddend . One alternative \DIFdelbegin \DIFdel{framework is \emph{Fisheries Libraries in R} (\texttt{FLR}}\DIFdelend \DIFaddbegin \DIFadd{software framework is Fisheries Libraries in R (FLR}\DIFaddend ) \cite{kell2007} --- \DIFdelbegin \DIFdel{an }\DIFdelend \DIFaddbegin \DIFadd{a collection of }\DIFaddend open-source R \DIFdelbegin \DIFdel{package }\DIFdelend \DIFaddbegin \DIFadd{packages }\DIFaddend developed specifically for evaluating fisheries management strategies through simulation. Compared to ss3sim, \DIFdelbegin \DIFdel{\texttt{FLR} }\DIFdelend \DIFaddbegin \DIFadd{FLR }\DIFaddend is designed to explore broader questions regarding management strategies with flexible biological, economic, and management components \cite{hillary2009}. Thus, it is not specifically designed to explore the impact of structural differences within OMs and EMs. Another alternative stock assessment simulation testing framework is \DIFdelbegin \DIFdel{\emph{Fishery Simulation} }\DIFdelend \DIFaddbegin \DIFadd{Fishery Simulation }\DIFaddend (FS, \url{http://fisherysimulation.codeplex.com}). FS is primarily a file management tool adapted to aid in simulation testing. FS can work with stock assessment models besides SS3, make simple changes to input text files, generate \DIFdelbegin \DIFdel{random process }\DIFdelend \DIFaddbegin \DIFadd{simple random process errors }\DIFaddend (using a built-in random number generator) and observation errors (using the SS3 bootstrap option), run simulations in parallel, and collect results from output files. Thus, FS is closer to ss3sim in its scope than \DIFdelbegin \DIFdel{\texttt{FLR} }\DIFdelend \DIFaddbegin \DIFadd{FLR }\DIFaddend in that it specifically focuses on the performance of stock assessment models. \DIFdelbegin %DIFDELCMD < %DIFDELCMD < %%% \DIFdelend FS differs from ss3sim mainly in that it uses user-specified text manipulation commands (e.g.~change line 50 from 0 to 1) to alter model configurations rather than the approach of ss3sim, which uses modular functions tailored to specific purposes (e.g.~add a particular time-varying mortality trajectory to \DIFdelbegin \DIFdel{a particular }\DIFdelend \DIFaddbegin \DIFadd{an arbitrary }\DIFaddend OM). FS works well for testing arbitrary assessment models and model configurations because it does not rely on pre-built manipulation functions \cite{lee2012, piner2011, lee2011}. In contrast, FS cannot make complicated structural changes to a model setup (e.g.~adding time-varying parameters or changing the survey years), limiting its ability to \DIFdelbegin \DIFdel{to }\DIFdelend induce and test structural differences between OMs and EMs. In addition, the current version of FS is not an end-to-end package --- additional code is necessary to incorporate \DIFaddbegin \DIFadd{arbitrary }\DIFaddend process and observation error in simulation testing. Finally, although FS is also open-source, it requires the Microsoft .NET framework and is therefore only compatible with the Windows operating system. \section*{Research opportunities with ss3sim} The ss3sim package has been used so far to evaluate alternative assessment approaches when $M$ is thought to vary across time \DIFdelbegin %DIFDELCMD < \cite{johnson2013}%%% \DIFdel{, the importance of }\DIFdelend \DIFaddbegin \cite{johnson2014}\DIFadd{, the }\R{B25}\DIFadd{effect of various qualities and quantities of }\DIFaddend length- and age-composition data \DIFdelbegin %DIFDELCMD < \cite{ono2013}%%% \DIFdelend \DIFaddbegin \DIFadd{on the bias and accuracy of assessment model estimates }\cite{ono2014}\DIFaddend , and the causes of retrospective patterns in stock assessment \DIFdelbegin \DIFdel{models}\DIFdelend \DIFaddbegin \DIFadd{model estimates}\DIFaddend . Along with those studies, ss3sim makes many \DIFdelbegin \DIFdel{important }\DIFdelend \DIFaddbegin \DIFadd{relevant }\DIFaddend research opportunities easily approachable. Below we outline some examples. \emph{Time-varying model misspecification}: Ecological processes can vary through time in response to, for example, changes to fishing behaviour \cite{hilborn1992}, regime shifts \cite{vert-pre2013}, or climate change \cite{walther2002}. However, parameters such as $M$, catchability, and selectivity are commonly assumed to be time invariant and the consequences of \DIFdelbegin \DIFdel{assuming time invariance of such parameters }\DIFdelend \DIFaddbegin \DIFadd{these assumptions }\DIFaddend when facing true temporal changes has been a long-standing discussion in fisheries science \cite{royama1992, wilberg2006, fu2001}. Furthermore, although studies have tried to isolate the effects of single time-varying \DIFdelbegin \DIFdel{parameters}\DIFdelend \DIFaddbegin \DIFadd{parameter}\DIFaddend , such as $M$ \DIFdelbegin %DIFDELCMD < \cite{lee2011, jiao2012, deroba2013, johnson2013}%%% \DIFdelend \DIFaddbegin \cite{lee2011, jiao2012, deroba2013, johnson2014}\DIFaddend , few have considered the \DIFdelbegin \DIFdel{effect }\DIFdelend \DIFaddbegin \DIFadd{effects }\DIFaddend of multiple time-varying parameters and their potential interaction. \DIFaddbegin \DIFadd{ss3sim can easily turn parameter estimation on and off as well as add time-varying dynamics to the OM, making it an ideal candidate for assessing the effects of multiple time-varying parameters. }\DIFaddend \emph{Patterns in recruitment deviations}: Typically, estimation methods assume independent log-normally-distributed recruitment deviations around a spawning stock recruitment function. However, recruitment deviations are frequently auto-correlated and their variability can change through time \cite{beamish1995, pyper1998}. \DIFaddbegin \R{B26}\DIFaddend ss3sim \DIFdelbegin \DIFdel{makes it simple to incorporate }\DIFdelend \DIFaddbegin \DIFadd{facilitates exploring the effect of }\DIFaddend different recruitment deviation structures \DIFdelbegin \DIFdel{and test how they affect model performance }\DIFdelend \DIFaddbegin \DIFadd{on model performance by allowing the user to directly specify any vector of deviations}\DIFaddend . \emph{Retrospective patterns}: Retrospective patterns, in which model estimates are systematically biased with each additional year of data, are a major problem in stock assessment science \cite{mohn1999, legault2008}. Key questions are: what causes retrospective patterns and \DIFdelbegin \DIFdel{what }\DIFdelend \DIFaddbegin \DIFadd{which }\DIFaddend assessment approaches reduce \DIFdelbegin \DIFdel{retrospective patterns }\DIFdelend \DIFaddbegin \DIFadd{them }\DIFaddend \cite{legault2008}. ss3sim can run retrospective analyses as part of any simulation by adding a single argument\DIFdelbegin \DIFdel{--- }\DIFdelend \DIFaddbegin \DIFadd{: }\DIFaddend the number of retrospective years to investigate. \section*{Conclusions} The increasing complexity of modern integrated stock assessment models and expanding computing power allows for the inclusion of multiple sources of data and estimation of \DIFaddbegin \DIFadd{increasingly }\DIFaddend complex processes \cite{maunder2013}. However, \DIFdelbegin \DIFdel{the combination of complex models and large quantities of data are commonly associated with model misspecification, which can be difficult to detect based on residual patterns alone }\DIFdelend \DIFaddbegin \DIFadd{it is difficult to determine under which conditions these processes can be reliably estimated based on diagnostics such as residual patterns }\DIFaddend \cite{maunder2013}. \DIFdelbegin \DIFdel{Therefore, it is important to investigate the consequences of model misspecification. Investigating the consequences of model misspecification on the ability of assessment models to accurately and precisely estimate parameters is one important role of simulation testing }%DIFDELCMD < \cite{wilberg2006, deroba2013a, crone2013}%%% \DIFdel{. }\DIFdelend \DIFaddbegin \DIFadd{Simulation testing is an important tool because it provides an opportunity to explore model performance under specified conditions and develop a further understanding of a model's abilities. }\DIFaddend \DIFdelbegin \DIFdel{Most simulation testing work to date has used custom frameworks tailored to the particular needs of each study }%DIFDELCMD < \cite{helu2000, yin2004, magnusson2007, wetzel2011a, jiao2012, wilberg2006, deroba2013a, deroba2013, crone2013a, hurtadoferro2013}%%% \DIFdel{. Although the complexity of many studies requires a custom framework, we encourage authors to publish their }\DIFdelend \DIFaddbegin \DIFadd{We anticipate that ss3sim will facilitate the development of reliable assessment methods, applicable to age-structured stock assessment frameworks in general, that meet the requirements and assessment demands of many regional fisheries management organizations and national assessment agencies. For example, Johnson et~al.~}\cite{johnson2014} \DIFadd{used ss3sim to develop guidelines for how to model natural mortality (when it is suspected of being time varying but age invariant) across life histories and fishing patterns. As another example, Ono et~al.~}\cite{ono2014} \DIFadd{used ss3sim to identify the most informative combination of quantity, quality, and timing of data, depending on life history and stock-assessment-derived metrics of interest. General guidelines such as these, combined with simulations testing specific model configurations used by assessment agencies, are an important part of developing reliable assessment methods to provide sound scientific advice to fisheries management }\cite{deroba2014, crone2013}\DIFadd{. } \DIFadd{Custom-tailored simulation-testing software packages are an increasingly common tool in fisheries science, but their value would be extended if shared formally with the broader community. Published, open-source }\DIFaddend simulation frameworks, \DIFdelbegin \DIFdel{as we have done here, and where possible, to develop }\DIFdelend \DIFaddbegin \DIFadd{such as the initial release of ss3sim described here, allow other scientists to validate, reuse, and improve the software. We therefore encourage authors to publish }\DIFaddend their simulation frameworks \DIFaddbegin \DIFadd{and develop them }\DIFaddend in a generalized format\DIFdelbegin \DIFdel{that allows others to build on them. The initial release of ss3sim describes the basic structure used in recent studies }%DIFDELCMD < \cite{johnson2013, ono2013} %%% \DIFdel{and the current version of ss3sim could be used to address other important questions in stock assessment science. We hope }\DIFdelend \DIFaddbegin \DIFadd{, where possible. We anticipate }\DIFaddend that users will both benefit from ss3sim in its current form and extend it for their own needs, potentially contributing \DIFdelbegin \DIFdel{back }\DIFdelend to future versions. \section*{Acknowledgements} We thank the participants and mentors of the University of Washington's School of Aquatic and Fishery Sciences 2013 FISH 600 course. Discussions with these individuals were instrumental to the conceptual and technical development of ss3sim. Many participants also contributed code and are listed within specific ss3sim R functions. Participants: , , , , , , , and . Mentors: , Andr\'{e} Punt, , and . We thank , Andr\'{e} Punt, \DIFdelbegin \DIFdel{and Iewart}\DIFdelend \DIFaddbegin \DIFadd{, , , and an anonymous reviewer }\DIFaddend for comments that \DIFaddbegin \DIFadd{greatly }\DIFaddend improved our manuscript. SCA was supported by Fulbright Canada (generously hosted by Trevor \DIFdelbegin \DIFdel{A. }\DIFdelend Branch), NSERC, and a Garfield Weston Foundation/B.C. Packers Ltd. \DIFdelbegin \DIFdel{~}\DIFdelend Graduate Fellowship in Marine Sciences. \DIFdelbegin \DIFdel{KFJ and KO were partially supported by NOAA grant 423 }\DIFdelend \DIFaddbegin \DIFadd{This work was partially funded by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement No.~}\DIFaddend NA10OAR4320148\DIFdelbegin \DIFdel{. }\DIFdelend \DIFaddbegin \DIFadd{, Contribution No.~2193. }\DIFaddend This research addresses the methods component of the good practices guide to stock assessment program of the Center for the Advancement of Population Assessment Methodology (CAPAM). %\bibliography{ss3sim-ms} \DIFaddbegin \DIFaddend \begin{thebibliography}{10} \providecommand{\url}[1]{\texttt{#1}} \providecommand{\urlprefix}{URL } \expandafter\ifx\csname urlstyle\endcsname\relax \providecommand{\doi}[1]{doi:\discretionary{}{}{}#1}\else \providecommand{\doi}{doi:\discretionary{}{}{}\begingroup \urlstyle{rm}\Url}\fi \providecommand{\bibAnnoteFile}[1]{% \IfFileExists{#1}{\begin{quotation}\noindent\textsc{Key:} #1\\ \textsc{Annotation:}\ \input{#1}\end{quotation}}{}} \providecommand{\bibAnnote}[2]{% \begin{quotation}\noindent\textsc{Key:} #1\\ \textsc{Annotation:}\ #2\end{quotation}} \providecommand{\eprint}[2][]{\url{#2}} \bibitem{gulland1983} Gulland JA (1983) Fish Stock Assessment: A Manual of Basic Methods. \newblock New York: Wiley. \bibAnnoteFile{gulland1983} \bibitem{hilborn1992} , Walters C (1992) Quantitative Fisheries Stock Assessment: Choice, Dynamics, and Uncertainty. \newblock London: Chapman and Hall. \bibAnnoteFile{hilborn1992} \bibitem{hilborn1987} , Walters CJ (1987) A general model for simulation of stock and fleet dynamics in spatially heterogeneous fisheries. \newblock Can J Fish Aquat Sci 44: 1366--1369. \bibAnnoteFile{hilborn1987} \bibitem{rosenberg1994} Rosenberg AA, Restrepo VR (1994) Uncertainty and risk evaluation in stock assessment advice for {U.S.} marine fisheries. \newblock Can J Fish Aquat Sci 51: 2715--2720. \bibAnnoteFile{rosenberg1994} \bibitem{peterman2004} (2004) Possible solutions to some challenges facing fisheries scientists and managers. \newblock ICES J Mar Sci 61: 1331--1343. \bibAnnoteFile{peterman2004} \bibitem{deroba2014} , , , De~, , et~al. (2014) Simulation testing the robustness of stock assessment models to error: some results from the {ICES} strategic initiative on stock assessment methods. \newblock ICES J Mar Sci. In press. DOI: 10.1093/icesjms/fst237. \bibAnnoteFile{deroba2014} \bibitem{lee2011} , , , Methot RD (2011) Estimating natural mortality within a fisheries stock assessment model: An evaluation using simulation analysis based on twelve stock assessments. \newblock Fish Res 109: 89--94. \bibAnnoteFile{lee2011} \bibitem{jiao2012} , , O'Reilly R, Orth DJ (2012) Modelling non-stationary natural mortality in catch-at-age models. \newblock ICES J Mar Sci 69: 105--118. \bibAnnoteFile{jiao2012} \bibitem{deroba2013} , Schueller AM (2013) Performance of stock assessments with misspecified age- and time-varying natural mortality. \newblock Fish Res 146: 27--40. \bibAnnoteFile{deroba2013} \bibitem{johnson2014} , , McGilliard CR, Vert-pre KA, Anderson SC, et~al. (2014) Time-varying natural mortality in fisheries stock assessment models: Identifying a default approach. \newblock In review at ICES J Mar Sci. \bibAnnoteFile{johnson2014} \bibitem{lee2012} , Maunder MN, Piner KR, Methot RD (2012) Can steepness of the stock-recruitment relationship be estimated in fishery stock assessment models? \newblock Fish Res 125--126: 254--261. \bibAnnoteFile{lee2012} \bibitem{schirripa2009} Schirripa MJ, Goodyear CP, Methot RM (2009) Testing different methods of incorporating climate data into the assessment of {US West Coast} sablefish. \newblock ICES J Mar Sci 66: 1605-1613. \bibAnnoteFile{schirripa2009} \bibitem{magnusson2007} , (2007) What makes fisheries data informative? \newblock Fish Fish 8: 337--358. \bibAnnoteFile{magnusson2007} \bibitem{wetzel2011a} Wetzel CR, Punt AE (2011) Performance of a fisheries catch-at-age model ({Stock Synthesis}) in data-limited situations. \newblock Mar Freshwater Res 62: 927--936. \bibAnnoteFile{wetzel2011a} \bibitem{ono2014} , , , , Anderson SC, et~al. (2014) The importance of length and age composition data in statistical age-structured models for marine species. \newblock ICES J Mar Sci. In press. DOI: 10.1093/icesjms/fsu007. \bibAnnoteFile{ono2014} \bibitem{yin2004} , Sampson DB (2004) Bias and precision of estimates from an age-structured stock assessment program in relation to stock and data characteristics. \newblock N Am J Fish Manage 24: 865-879. \bibAnnoteFile{yin2004} \bibitem{methot2013} Methot RD Jr, Wetzel CR (2013) {Stock Synthesis}: A biological and statistical framework for fish stock assessment and fishery management. \newblock Fish Res 142: 86--99. \bibAnnoteFile{methot2013} \bibitem{maunder2013} Maunder MN, Punt AE (2013) A review of integrated analysis in fisheries stock assessment. \newblock Fish Res 142: 61--74. \bibAnnoteFile{maunder2013} \bibitem{cope2013} Cope JM (2013) Implementing a statistical catch-at-age model ({Stock Synthesis}) as a tool for deriving overfishing limits in data-limited situations. \newblock Fish Res 142: 3--14. \bibAnnoteFile{cope2013} \bibitem{haltuch2013} Haltuch MA, Ono K, Valero JL (2013) Status of the {U.S.} petrale sole resource in 2012. \newblock Technical report, Pacific Fishery Management Council. \bibAnnoteFile{haltuch2013} \bibitem{helu2000} Helu SL, Sampson DB, Yin Y (2000) Application of statistical model selection criteria to the {Stock Synthesis} assessment program. \newblock Can J Fish Aquat Sci 57: 1784--1793. \bibAnnoteFile{helu2000} \bibitem{crone2013a} Crone PR, Valero JL (In review) Evaluation of length vs.~age selectivity in stock assessments based on management criteria: It matters when fitting to length-composition data. \newblock Fish Res. \bibAnnoteFile{crone2013a} \bibitem{hurtadoferro2013} , Punt AE, Hill KT (2013) Use of multiple selectivity patterns as a proxy for spatial structure. \newblock Fish Res. In press. DOI: 10.1016/j.fishres.2013.10.001. \bibAnnoteFile{hurtadoferro2013} \bibitem{wilberg2006} , (2006) Performance of time-varying catchability estimators in statistical catch-at-age analysis. \newblock Can J Fish Aquat Sci 63: 2275--2285. \bibAnnoteFile{wilberg2006} \bibitem{rcoreteam2013} R Core~Team (2013) R: A Language and Environment for Statistical Computing. \newblock R Foundation for Statistical Computing, Vienna, Austria. \newblock \urlprefix\url{http://www.R-project.org/}. \bibAnnoteFile{rcoreteam2013} \bibitem{r4ss2013} , , , , , et~al. (2013) r4ss: R code for Stock Synthesis. \newblock \urlprefix\url{http://code.google.com/p/r4ss/}. \newblock R package version 1.20. \bibAnnoteFile{r4ss2013} \bibitem{fournier2012} , Skaug HJ, , , , et~al. (2012) {AD Model Builder}: Using automatic differentiation for statistical inference of highly parameterized complex nonlinear models. \newblock Optim Methods \& Soft 27: 233--249. \bibAnnoteFile{fournier2012} \bibitem{bolker2013} , , , W, , et~al. (2013) Strategies for fitting nonlinear ecological models in {R}, {AD Model Builder}, and {BUGS}. \newblock Methods Ecol Evol 4: 501--512. \bibAnnoteFile{bolker2013} \bibitem{hill2012} , Crone PR, Lo NCH, , , et~al. (2012) Assessment of the {Pacific} sardine resource in 2012 for {U.S.} management in 2013. \newblock Technical report, Pacific Fishery Management Council, 7700 NE Ambassador Place, Portland, OR 97220, USA. \bibAnnoteFile{hill2012} \bibitem{wilson2014} , , Brown CT, Chue~Hong NP, , et~al. (2014) Best practices for scientific computing. \newblock PLOS Biol 12: e1001745. \bibAnnoteFile{wilson2014} \bibitem{methot2011} , Taylor IG (2011) Adjusting for bias due to variability of estimated recruitments in fishery assessment models. \newblock Can J Fish Aquat Sci 68: 1744--1760. \bibAnnoteFile{methot2011} \bibitem{wickham2009} (2009) \textbf{ggplot2}: Elegant graphics for data analysis. \newblock New York: Springer. \newblock \urlprefix\url{http://had.co.nz/ggplot2/book}. \bibAnnoteFile{wickham2009} \bibitem{kell2007} , , , , , et~al. (2007) {FLR}: An open-source framework for the evaluation and development of management strategies. \newblock ICES J Mar Sci 64: 640--646. \bibAnnoteFile{kell2007} \bibitem{hillary2009} (2009) An introduction to {FLR} fisheries simulation tools. \newblock Aquat Living Resour 22: 225--232. \bibAnnoteFile{hillary2009} \bibitem{piner2011} , , , (2011) A simulation-based method to determine model misspecification: Examples using natural mortality and population dynamics models. \newblock Mar Coast Fish 3: 336--343. \bibAnnoteFile{piner2011} \bibitem{vert-pre2013} Vert-pre KA, Amoroso RO, , (2013) Frequency and intensity of productivity regime shifts in marine fish stocks. \newblock Proc Natl Acad Sci USA 110: 1779--1784. \bibAnnoteFile{vert-pre2013} \bibitem{walther2002} , Post E, , , , et~al. (2002) Ecological responses to recent climate change. \newblock Nature 416: 389--395. \bibAnnoteFile{walther2002} \bibitem{royama1992} Royama T (1992) Analytical Population Dynamics. \newblock London: Chapman and Hall. \bibAnnoteFile{royama1992} \bibitem{fu2001} , , (2001) Why the {Atlantic cod} (\textit{Gadus morhua}) stock off eastern {Nova Scotia} has not recovered. \newblock Can J Fish Aquat Sci 58: 1613--1623. \bibAnnoteFile{fu2001} \bibitem{beamish1995} J (1995) Climatic change and northern fish populations. \newblock Can. Spec. Publ. Fish. Aquat. Sci. No. 121. \bibAnnoteFile{beamish1995} \bibitem{pyper1998} , (1998) Comparison of methods to account for autocorrelation in correlation analyses of fish data. \newblock Can J Fish Aquat Sci 55: 2127--2140. \bibAnnoteFile{pyper1998} \bibitem{mohn1999} (1999) The retrospective problem in sequential population analysis: An investigation using cod fishery and simulated data. \newblock ICES J Mar Sci 56: 473--488. \bibAnnoteFile{mohn1999} \bibitem{legault2008} Legault CM (2008) Report of the retrospective working group. {NOAA NMFS Northeast Fisheries Science Center Reference Document} 09--01. \newblock Northeast Fisheries Science Center Reference Document 09--01, NOAA NMFS, , Massachusetts. \bibAnnoteFile{legault2008} \bibitem{crone2013} , , , , X (2013) Selectivity: Theory, estimation, and application in fishery stock assessment models. \newblock Workshop series report 1, Center for the Advancement of Population Assessment Methodology (CAPAM). \bibAnnoteFile{crone2013} \DIFaddbegin \end{thebibliography} \DIFaddend \clearpage \section*{Figure Legends} \begin{figure}[!ht] \begin{center} %\includegraphics[width=3.27in]{sim-steps.pdf} \end{center} \caption{ {\bf Flow diagram of the main steps in an ss3sim simulation carried out using \texttt{run\_ss3sim}.} Functions that are called internally are shown in a monospaced font. } \label{fig:sim-steps} \end{figure} %DIF > \clearpage \DIFaddbegin \DIFaddend \begin{figure}[!ht] \begin{center} %DIF > \includegraphics[width=4.86in]{file-structure.pdf} \DIFaddbeginFL \end{center} \caption{\DIFaddFL{ {\bf Illustration of input and output folder and file structure for an ss3sim simulation.} Folders are shown in blue, input files in orange, and output files in grey. All input and output files are in plain text format. OM refers to operating model and EM to estimation model. Case files (orange files at bottom left) combine cases (e.g.~\texttt{M0} for a given natural mortality trajectory) with species or life-history OMs and EMs (e.g.~cod-like or sardine-like). Alternatively, a user can skip setting up case files and specify the simulation cases directly in R code (see the accompanying vignette [Text S1]). }} \end{figure} %DIF > \clearpage \begin{figure}[!ht] \begin{center} \DIFaddendFL %\includegraphics[width=4.86in]{fig2-20131109.pdf} \end{center} \caption{ {\bf Example output from an ss3sim simulation.} We ran a crossed simulation in which we considered (1) the effect of fixing natural mortality ($M$) at its true value (0.2; case E0) or estimating $M$ (case E1) and (2) the effect of high survey effort ($\sigma_\mathrm{survey} = 0.1$; case D0) or low survey effort ($\sigma_\mathrm{survey} = 0.4$; case D1). Upper panels (blue) show time series of relative error in spawning stock biomass (SSB). Lower panels (grey) show the distribution of relative error across four scalar variables: depletion \DIFaddbeginFL \DIFaddFL{(the ratio of terminal year spawning biomass to unfished spawning biomass)}\DIFaddendFL , $M$, SSB at maximum sustainable yield ($\mathrm{SSB}_\mathrm{MSY}$), and fishing mortality ($F$) in the terminal year. We show the values across simulation iterations with dots and the distributions with beanplots (kernel density smoothers). } \label{fig:results} \end{figure} \clearpage \section*{Tables} \textbf{Table 1. Main ss3sim functions and a description of their purpose.} Simulations can be run through the \texttt{run\_ss3sim} function, which then calls the \texttt{change} functions. Users can control what the \texttt{change} functions do through a series of plain-text case files. For example, the case ID \texttt{D1} corresponds to the case files \texttt{lcomp1}, \texttt{agecomp1}, and \texttt{index1}, as described in the table. Users can also skip setting up case files and specify arguments to \texttt{ss3sim\_base} directly, or use the \texttt{change} functions as part of their own simulation structure (Text S1). \begin{longtable}[c]{@{}ll@{}} \hline\noalign{\medskip} \begin{minipage}[b]{0.32\columnwidth}\raggedright Function name \end{minipage} & \begin{minipage}[b]{0.57\columnwidth}\raggedright Description \end{minipage} \\\noalign{\medskip} \hline\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{run\_ss3sim} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Main high-level function to run ss3sim simulations. \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{ss3sim\_base} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Underlying base simulation function. Can also be called directly. \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_rec\_devs} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Substitutes recruitment deviations. \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_f} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Adds fishing mortality time series. (Case file and ID \texttt{F}) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_tv} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Adds time-varying features. For example, time-varying natural mortality, growth, or selectivity. (Any case file and ID, e.g.~\texttt{M}, starting with ``\texttt{function\_type; change\_tv}'') \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_index} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Controls how the fishery and survey indices are sampled. (Case file \texttt{index}, case ID \texttt{D}) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_agecomp} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Controls how age composition data are sampled. (Case file \texttt{agecomp}, case ID \texttt{D}) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_lcomp} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Controls how length composition data are sampled. (Case file \texttt{lcomp}, case ID \texttt{D}) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_retro} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Controls the number of years to discard for a retrospective analysis. (Case file and ID R) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{change\_e} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Controls which and how parameters are estimated. (Case file and ID \texttt{E}) \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{run\_bias\_ss3} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Determines the level of adjustment to ensure mean-unbiased estimates of recruitment and biomass. \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{get\_results\_scenario} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Extracts results for a single scenario. \end{minipage} \\\noalign{\medskip} \begin{minipage}[t]{0.32\columnwidth}\raggedright \texttt{get\_results\_all} \end{minipage} & \begin{minipage}[t]{0.57\columnwidth}\raggedright Extracts results for a series of scenarios. \end{minipage} \\\noalign{\medskip} \hline \end{longtable} \DIFdelbegin %DIFDELCMD < %DIFDELCMD < %%% %DIF < Please keep the abstract between 250 and 300 words %DIF < \section*{Abstract} %DIFDELCMD < %DIFDELCMD < %%% %DIF < Please keep the Author Summary between 150 and 200 words %DIF < Use first person. PLoS ONE authors please skip this step. %DIF < Author Summary not valid for PLoS ONE submissions. %DIF < \section*{Author Summary} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \section*{Introduction} %DIFDELCMD < %DIFDELCMD < %%% %DIF < Results and Discussion can be combined. %DIF < \section*{Results} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \subsection*{Subsection 1} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \subsection*{Subsection 2} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \section*{Discussion} %DIFDELCMD < %DIFDELCMD < %%% %DIF < You may title this section "Methods" or "Models". %DIF < "Models" is not a valid title for PLoS ONE authors. However, PLoS ONE %DIF < authors may use "Analysis" %DIF < \section*{Materials and Methods} %DIFDELCMD < %DIFDELCMD < %%% %DIF < Do NOT remove this, even if you are not including acknowledgments %DIF < \section*{Acknowledgments} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \section*{References} %DIF < The bibtex filename %DIF < \bibliography{template} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \section*{Figure Legends} %DIF < \begin{figure}[!ht] %DIF < \begin{center} %DIF < %\includegraphics[width=4in]{figure_name.2.eps} %DIF < \end{center} %DIF < \caption{ %DIF < {\bf Bold the first sentence.} Rest of figure 2 caption. Caption %DIF < should be left justified, as specified by the options to the caption %DIF < package. %DIF < } %DIF < \label{Figure_label} %DIF < \end{figure} %DIFDELCMD < %DIFDELCMD < %%% %DIF < \section*{Tables} %DIF < \begin{table}[!ht] %DIF < \caption{ %DIF < \bf{Table title}} %DIF < \begin{tabular}{|c|c|c|} %DIF < table information %DIF < \end{tabular} %DIF < \begin{flushleft}Table caption %DIF < \end{flushleft} %DIF < \label{tab:label} %DIF < \end{table} \DIFdelend \end{document} content/publication/lu-2013-learningcca/cite.bib0 @inproceedings{lu2013learningcca, author = {}, booktitle = {International Joint Conference on Artificial Intelligence (IJCAI}, pages = {1516--1522}, title = {Learning canonical correlations of paired tensor sets via tensor-to-vector projection}, year = {2013} } LinusCastelino/Resume0 % This template has been downloaded from: % http://www.LaTeXTemplates.com % % Original author: % () with major modifications by % Vel () % % License: % The MIT License (see included LICENSE file) \documentclass[letter]{twentysecondcv} \usepackage[none]{hyphenat} % To remove text hyphenation \cvjobtitle{Full-Stack Web Developer\\ \\M.S. in Computer Science} \cvdate{} \cvaddress{} \cvmail{} \cvnumberphone{+1 (201) 952-4216} \cvwebsite{linuscastelino.com} \cvlinkedin{linkedin.com/in/linuscastelino/} \cvgithub{github.com/LinusCastelino} \workexp{ \textbf{University at Buffalo - SUNY Research Foundation} \newline Jan. 2019 - Present \newline Research Assistant \newline Buffalo, New York \newline \newline \textbf{Salesforce Inc.} \newline May 2019 - Aug. 2019 \newline Software Engineer Intern - AMTS \newline Indianapolis, Indiana \newline \textbf{Tata Consultancy Services Pvt. Ltd.} \newline Oct. 2015 - Jul. 2018 \newline Systems Engineer \newline Mumbai, India } \Education{ \textbf{University at Buffalo, The State University of New York} \newline Master's in Computer Science \newline Class of 2020 \newline \textbf{GPA:} 3.89 \newline \newline \textbf{University of Mumbai} \newline Bachelor's in Computer Engineering \newline Class of 2015 \newline \textbf{GPA:} 3.8 } \begin{document} {\fontsize{28pt}{12pt}\selectfont\color{mainblue}\textbf{}} \\ \fontsize{12pt}{11pt} \makeprofile % Print the sidebar \section{Core Skills} \begin{tabular}{p{3cm}p{10cm}} \textbf{Programming:} & {Java, JavaScript, jQuery, C++, Python}\\ \textbf{Frameworks:} & {ReactJS, Angular, NodeJS, Oracle Web Commerce, Spring Boot, Spring WebFlux, Salesforce Suite(Apex, Aura, LWC)}\\ \textbf{Database:} & {Oracle Database, ATG RQL, MongoDB}\\ \textbf{Automation:} & {Shell Scripting , Maven, ANT, Autosys}\\ \textbf{Tools:} & {IntelliJ, VS Code, SQL Developer, Putty, WinSCP, NPM, SFDX} \vspace{5mm} \end{tabular} \section{Projects} \begin{twenty} \vspace{1mm} \underline{\large{Professional:}}\\ \project{Real Estate Management System}{May 2019 - Aug. 2019}{Salesforce Inc.}{Salesforce Platform Developer}{Worked as a Software Developer Intern in the BizTech Org at Salesforce. Primary responsibilities included developing quality code for user stories, developing corresponding unit test cases and performing peer code reviews.}{Developed a logging framework and a scheduler process to generate daily notifications for pending spend approvals.}{} \project{ToysRUs E-commerce Webstore}{Jan. 2016 - Jul. 2018}{Tata Consultancy Services Pvt. Ltd.}{Oracle ATG Web Commerce Developer}{Developed software modules for the ToysRUs E-commerce webstores. Primary responsibilities included developing code for the given user requirements, fixing bugs raised by business teams and end users, and writing automation scripts for scheduled jobs and feed monitoring.}{}{} \vspace{1mm} \underline{\large{Academic:}}\\ \project{OneDataShare}{Aug. 2018 - Present}{University at Buffalo - SUNY Research Foundation}{Research Assistant and Full Stack Developer}{OneDataShare is a research project funded by NSF aimed at developing a Managed File Transfer system that enhances high-volume data sharing across different protocols. Actively contributed to the project as a team lead, developing functional modules and researching efficient approaches to solve algorithmic problems.}{Implemented recursive folder transfers for multiple protocols and developed a ticketing system for error reporting.}{https://www.onedatashare.org/} \project{TwiFi}{Dec. 2018}{University at Buffalo}{Full-stack Developer}{Developed TwiFi - a product that captures a user query, retrieves the relevant tweets, performs sentiment analysis and displays a statistical view of the retrieved results along with faceted search options.}{Designed the project architecture and developed UI components.}{https://github.com/LinusCastelino/UB-F18-CSE535\_Proj4} \end{twenty} \section{Certifications} \begin{twenty} \award{Oracle Certified Professional Java Programmer}{Aug. 2014}{License 234706197OCPJSE6P \\ Secured 100\% score} \end{twenty} \vspace{5mm} % \section{Honors and Awards} % \begin{twenty} % \award{Honorary Trailhead Ranger}{Jul. 2019}{Salesforce Inc. - Software Engineer (AMTS)} % \award{iON Commitment Award}{Apr. 2016}{Tata Consultancy Services Pvt. Ltd. - Systems Engineer} % \award{Champions of ILP}{Feb. 2016}{Tata Consultancy Services Pvt. Ltd. - Systems Engineer} % \award{Star of Learners Group}{Dec. 2015}{Tata Consultancy Services Pvt. Ltd. - Assistant Systems Engineer} % \award{Procoders}{Mar. 2015}{St. Francis Institute of Technology - Student} % \end{twenty} \end{document} \hypertarget{dba_8php}{}\doxysection{vendor/jetbrains/phpstorm-\/stubs/dba/dba.php File Reference} \label{dba_8php}\index{vendor/jetbrains/phpstorm-\/stubs/dba/dba.php@{vendor/jetbrains/phpstorm-\/stubs/dba/dba.php}} \doxysubsection*{Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{dba_8php_a573f0971cdec2b927b2cc8378a31628f}{dba\+\_\+open}} (\$path, \$mode, \$handler=\mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}, \$\mbox{\hyperlink{gettext_8php_a846b230f82cfd260aff26f4287060106}{\+\_\+}}=\mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}) \item \mbox{\hyperlink{dba_8php_a17a72cbee059fd76a00a7ac78340f3b0}{dba\+\_\+popen}} (\$path, \$mode, \$handler=\mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}, \$\mbox{\hyperlink{gettext_8php_a846b230f82cfd260aff26f4287060106}{\+\_\+}}=\mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}) \item \mbox{\hyperlink{dba_8php_aab9d3409fe84048aefdc82910613722e}{dba\+\_\+close}} (\$handle) \item \mbox{\hyperlink{dba_8php_ae5a1f253800b1a450eda7a94de932f37}{dba\+\_\+delete}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$handle) \item \mbox{\hyperlink{dba_8php_a0ed82fec5f19e04bf7a55ed6b5a52674}{dba\+\_\+exists}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$handle) \item \mbox{\hyperlink{dba_8php_af4d6041077691c2f069275d26a211381}{dba\+\_\+fetch}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$handle) \item \mbox{\hyperlink{dba_8php_ab6a956cb65f955f08b38fdb930a3906f}{dba\+\_\+fetch}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$skip, \$handle) \item \mbox{\hyperlink{dba_8php_a82d5372d4342596dde66f5b6cd36b430}{dba\+\_\+insert}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$value, \$handle) \item \mbox{\hyperlink{dba_8php_a41545d33616d1c0825ac336d6c2dcbc2}{dba\+\_\+replace}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}, \$value, \$handle) \item \mbox{\hyperlink{dba_8php_ab46a36d00e4713785e70cfbf761717f6}{dba\+\_\+firstkey}} (\$handle) \item \mbox{\hyperlink{dba_8php_a60768204780fcd38b667a2bd4679f1a7}{dba\+\_\+nextkey}} (\$handle) \item \mbox{\hyperlink{dba_8php_afa8769571533d879d30975645b9e4744}{dba\+\_\+optimize}} (\$handle) \item \mbox{\hyperlink{dba_8php_acd38a98eb82a7341a835d8e821eafb55}{dba\+\_\+sync}} (\$handle) \item \mbox{\hyperlink{dba_8php_ac60d89e5c0e0d3eb45c6b0b3407fca2d}{dba\+\_\+handlers}} (\$full\+\_\+info=\mbox{\hyperlink{_core__d_8php_a2f585b9641adc7a08b35a997065e95b4}{false}}) \item \mbox{\hyperlink{dba_8php_a1c0a48f42b3bdc152f8ac54aa8a701be}{dba\+\_\+list}} () \item \mbox{\hyperlink{dba_8php_a597abbe37cb1f15193cb7c1129d0f5ee}{dba\+\_\+key\+\_\+split}} (\$\mbox{\hyperlink{standard__8_8php_af7fce235ef822d2f9df754383b89cd63}{key}}) \end{DoxyCompactItemize} \doxysubsection{Function Documentation} \mbox{\Hypertarget{dba_8php_aab9d3409fe84048aefdc82910613722e}\label{dba_8php_aab9d3409fe84048aefdc82910613722e}} \index{dba.php@{dba.php}!dba\_close@{dba\_close}} \index{dba\_close@{dba\_close}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_close()}{dba\_close()}} {\footnotesize\ttfamily dba\+\_\+close (\begin{DoxyParamCaption}\item[{}]{\$handle }\end{DoxyParamCaption})} Close a D\+BA database \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{resource \$handle }} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} void No value is returned. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_ae5a1f253800b1a450eda7a94de932f37}\label{dba_8php_ae5a1f253800b1a450eda7a94de932f37}} \index{dba.php@{dba.php}!dba\_delete@{dba\_delete}} \index{dba\_delete@{dba\_delete}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_delete()}{dba\_delete()}} {\footnotesize\ttfamily dba\+\_\+delete (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$handle }\end{DoxyParamCaption})} Delete D\+BA entry specified by key \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key of the entry which is deleted. \begin{DoxyParams}[1]{Parameters} resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a0ed82fec5f19e04bf7a55ed6b5a52674}\label{dba_8php_a0ed82fec5f19e04bf7a55ed6b5a52674}} \index{dba.php@{dba.php}!dba\_exists@{dba\_exists}} \index{dba\_exists@{dba\_exists}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_exists()}{dba\_exists()}} {\footnotesize\ttfamily dba\+\_\+exists (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$handle }\end{DoxyParamCaption})} Check whether key exists \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key the check is performed for. \begin{DoxyParams}[1]{Parameters} resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} if the key exists, {\bfseries{F\+A\+L\+SE}} otherwise. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_af4d6041077691c2f069275d26a211381}\label{dba_8php_af4d6041077691c2f069275d26a211381}} \index{dba.php@{dba.php}!dba\_fetch@{dba\_fetch}} \index{dba\_fetch@{dba\_fetch}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_fetch()}{dba\_fetch()}\hspace{0.1cm}{\footnotesize\ttfamily [1/2]}} {\footnotesize\ttfamily dba\+\_\+fetch (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$handle }\end{DoxyParamCaption})} Fetch data specified by key \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key the data is specified by. When working with inifiles this function accepts arrays as keys where index 0 is the group and index 1 is the value name. See\+: {\bfseries{dba\+\_\+key\+\_\+split}}. \begin{DoxyParams}[1]{Parameters} resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} string$\vert$false the associated string if the key/data pair is found, {\bfseries{F\+A\+L\+SE}} otherwise. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_ab6a956cb65f955f08b38fdb930a3906f}\label{dba_8php_ab6a956cb65f955f08b38fdb930a3906f}} \index{dba.php@{dba.php}!dba\_fetch@{dba\_fetch}} \index{dba\_fetch@{dba\_fetch}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_fetch()}{dba\_fetch()}\hspace{0.1cm}{\footnotesize\ttfamily [2/2]}} {\footnotesize\ttfamily dba\+\_\+fetch (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$skip, }\item[{}]{\$handle }\end{DoxyParamCaption})} Fetch data specified by key \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key the data is specified by. When working with inifiles this function accepts arrays as keys where index 0 is the group and index 1 is the value name. See\+: {\bfseries{dba\+\_\+key\+\_\+split}}. \begin{DoxyParams}[1]{Parameters} int & {\em \$skip} & The number of key-\/value pairs to ignore when using cdb databases. This value is ignored for all other databases which do not support multiple keys with the same name. \\ \hline resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} string$\vert$false the associated string if the key/data pair is found, {\bfseries{F\+A\+L\+SE}} otherwise. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_ab46a36d00e4713785e70cfbf761717f6}\label{dba_8php_ab46a36d00e4713785e70cfbf761717f6}} \index{dba.php@{dba.php}!dba\_firstkey@{dba\_firstkey}} \index{dba\_firstkey@{dba\_firstkey}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_firstkey()}{dba\_firstkey()}} {\footnotesize\ttfamily dba\+\_\+firstkey (\begin{DoxyParamCaption}\item[{}]{\$handle }\end{DoxyParamCaption})} Fetch first key \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{resource \$handle }} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} string the key on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_ac60d89e5c0e0d3eb45c6b0b3407fca2d}\label{dba_8php_ac60d89e5c0e0d3eb45c6b0b3407fca2d}} \index{dba.php@{dba.php}!dba\_handlers@{dba\_handlers}} \index{dba\_handlers@{dba\_handlers}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_handlers()}{dba\_handlers()}} {\footnotesize\ttfamily dba\+\_\+handlers (\begin{DoxyParamCaption}\item[{}]{\$full\+\_\+info = {\ttfamily \mbox{\hyperlink{_core__d_8php_a2f585b9641adc7a08b35a997065e95b4}{false}}} }\end{DoxyParamCaption})} List all the handlers available \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{bool \$full\+\_\+info \mbox{[}optional\mbox{]} }} Turns on/off full information display in the result. \begin{DoxyReturn}{Returns} array an array of database handlers. If {\itshape full\+\_\+info} is set to {\bfseries{T\+R\+UE}}, the array will be associative with the handlers names as keys, and their version information as value. Otherwise, the result will be an indexed array of handlers names. \end{DoxyReturn} When the internal cdb library is used you will see cdb and cdb\+\_\+make. \begin{DoxySince}{Since} 4.\+3 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a82d5372d4342596dde66f5b6cd36b430}\label{dba_8php_a82d5372d4342596dde66f5b6cd36b430}} \index{dba.php@{dba.php}!dba\_insert@{dba\_insert}} \index{dba\_insert@{dba\_insert}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_insert()}{dba\_insert()}} {\footnotesize\ttfamily dba\+\_\+insert (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$value, }\item[{}]{\$handle }\end{DoxyParamCaption})} Insert entry \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key of the entry to be inserted. If this key already exist in the database, this function will fail. Use {\bfseries{dba\+\_\+replace}} if you need to replace an existent key. \begin{DoxyParams}[1]{Parameters} string & {\em \$value} & \\ \hline \end{DoxyParams} The value to be inserted. \begin{DoxyParams}[1]{Parameters} resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a597abbe37cb1f15193cb7c1129d0f5ee}\label{dba_8php_a597abbe37cb1f15193cb7c1129d0f5ee}} \index{dba.php@{dba.php}!dba\_key\_split@{dba\_key\_split}} \index{dba\_key\_split@{dba\_key\_split}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_key\_split()}{dba\_key\_split()}} {\footnotesize\ttfamily dba\+\_\+key\+\_\+split (\begin{DoxyParamCaption}\item[{}]{\$key }\end{DoxyParamCaption})} Splits a key in string representation into array representation \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{mixed \$key }} The key in string representation. \begin{DoxyReturn}{Returns} array$\vert$false an array of the form array(0 =$>$ group, 1 =$>$ value\+\_\+name). This function will return {\bfseries{F\+A\+L\+SE}} if {\itshape key} is {\bfseries{N\+U\+LL}} or {\bfseries{F\+A\+L\+SE}}. \end{DoxyReturn} \begin{DoxySince}{Since} 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a1c0a48f42b3bdc152f8ac54aa8a701be}\label{dba_8php_a1c0a48f42b3bdc152f8ac54aa8a701be}} \index{dba.php@{dba.php}!dba\_list@{dba\_list}} \index{dba\_list@{dba\_list}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_list()}{dba\_list()}} {\footnotesize\ttfamily dba\+\_\+list (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} List all open database files \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{array An associative array, in the form resourceid =$>$ filename. 4.\+3 5.\+0 }}\mbox{\Hypertarget{dba_8php_a60768204780fcd38b667a2bd4679f1a7}\label{dba_8php_a60768204780fcd38b667a2bd4679f1a7}} \index{dba.php@{dba.php}!dba\_nextkey@{dba\_nextkey}} \index{dba\_nextkey@{dba\_nextkey}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_nextkey()}{dba\_nextkey()}} {\footnotesize\ttfamily dba\+\_\+nextkey (\begin{DoxyParamCaption}\item[{}]{\$handle }\end{DoxyParamCaption})} Fetch next key \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{resource \$handle }} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} string the key on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a573f0971cdec2b927b2cc8378a31628f}\label{dba_8php_a573f0971cdec2b927b2cc8378a31628f}} \index{dba.php@{dba.php}!dba\_open@{dba\_open}} \index{dba\_open@{dba\_open}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_open()}{dba\_open()}} {\footnotesize\ttfamily dba\+\_\+open (\begin{DoxyParamCaption}\item[{}]{\$path, }\item[{}]{\$mode, }\item[{}]{\$handler = {\ttfamily \mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}}, }\item[{}]{\$\+\_\+ = {\ttfamily \mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}} }\end{DoxyParamCaption})} Open database \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$path }} Commonly a regular path in your filesystem. \begin{DoxyParams}[1]{Parameters} string & {\em \$mode} & \\ \hline \end{DoxyParams} It is r for read access, w for read/write access to an already existing database, c for read/write access and database creation if it doesn\textquotesingle{}t currently exist, and n for create, truncate and read/write access. The database is created in B\+Tree mode, other modes (like Hash or Queue) are not supported. Additionally you can set the database lock method with the next char. Use l to lock the database with a .lck file or d to lock the databasefile itself. It is important that all of your applications do this consistently. If you want to test the access and do not want to wait for the lock you can add t as third character. When you are absolutely sure that you do not require database locking you can do so by using -\/ instead of l or d. When none of d, l or -\/ is used, dba will lock on the database file as it would with d. There can only be one writer for one database file. When you use dba on a web server and more than one request requires write operations they can only be done one after another. Also read during write is not allowed. The dba extension uses locks to prevent this. See the following table\+: \tabulinesep=1mm \begin{longtabu}spread 0pt [c]{*{0}{|X[-1]}|} \hline \end{longtabu} locking already open {\itshape mode} = \char`\"{}rl\char`\"{} {\itshape mode} = \char`\"{}rlt\char`\"{} {\itshape mode} = \char`\"{}wl\char`\"{} {\itshape mode} = \char`\"{}wlt\char`\"{} {\itshape mode} = \char`\"{}rd\char`\"{} {\itshape mode} = \char`\"{}rdt\char`\"{} {\itshape mode} = \char`\"{}wd\char`\"{} {\itshape mode} = \char`\"{}wdt\char`\"{} not open ok ok ok ok ok ok ok ok {\itshape mode} = \char`\"{}rl\char`\"{} ok ok wait false illegal illegal illegal illegal {\itshape mode} = \char`\"{}wl\char`\"{} wait false wait false illegal illegal illegal illegal {\itshape mode} = \char`\"{}rd\char`\"{} illegal illegal illegal illegal ok ok wait false {\itshape mode} = \char`\"{}wd\char`\"{} illegal illegal illegal illegal wait false wait false ok\+: the second call will be successfull. wait\+: the second call waits until {\bfseries{dba\+\_\+close}} is called for the first. false\+: the second call returns false. illegal\+: you must not mix \char`\"{}l\char`\"{} and \char`\"{}d\char`\"{} modifiers for {\itshape mode} parameter. \begin{DoxyParams}[1]{Parameters} string & {\em \$handler} & \mbox{[}optional\mbox{]} \\ \hline \end{DoxyParams} The name of the handler which shall be used for accessing {\itshape path}. It is passed all optional parameters given to {\bfseries{dba\+\_\+open}} and can act on behalf of them. \begin{DoxyParams}[1]{Parameters} mixed & {\em \$\+\_\+} & \mbox{[}optional\mbox{]} \\ \hline \end{DoxyParams} \begin{DoxyReturn}{Returns} resource a positive handle on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_afa8769571533d879d30975645b9e4744}\label{dba_8php_afa8769571533d879d30975645b9e4744}} \index{dba.php@{dba.php}!dba\_optimize@{dba\_optimize}} \index{dba\_optimize@{dba\_optimize}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_optimize()}{dba\_optimize()}} {\footnotesize\ttfamily dba\+\_\+optimize (\begin{DoxyParamCaption}\item[{}]{\$handle }\end{DoxyParamCaption})} Optimize database \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{resource \$handle }} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a17a72cbee059fd76a00a7ac78340f3b0}\label{dba_8php_a17a72cbee059fd76a00a7ac78340f3b0}} \index{dba.php@{dba.php}!dba\_popen@{dba\_popen}} \index{dba\_popen@{dba\_popen}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_popen()}{dba\_popen()}} {\footnotesize\ttfamily dba\+\_\+popen (\begin{DoxyParamCaption}\item[{}]{\$path, }\item[{}]{\$mode, }\item[{}]{\$handler = {\ttfamily \mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}}, }\item[{}]{\$\+\_\+ = {\ttfamily \mbox{\hyperlink{_core__d_8php_afa4c647ecaee0da639b8549bb3abf62b}{null}}} }\end{DoxyParamCaption})} Open database persistently \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$path }} Commonly a regular path in your filesystem. \begin{DoxyParams}[1]{Parameters} string & {\em \$mode} & \\ \hline \end{DoxyParams} It is r for read access, w for read/write access to an already existing database, c for read/write access and database creation if it doesn\textquotesingle{}t currently exist, and n for create, truncate and read/write access. \begin{DoxyParams}[1]{Parameters} string & {\em \$handler} & \mbox{[}optional\mbox{]} \\ \hline \end{DoxyParams} The name of the handler which shall be used for accessing {\itshape path}. It is passed all optional parameters given to {\bfseries{dba\+\_\+popen}} and can act on behalf of them. \begin{DoxyParams}[1]{Parameters} mixed & {\em \$\+\_\+} & \mbox{[}optional\mbox{]} \\ \hline \end{DoxyParams} \begin{DoxyReturn}{Returns} resource a positive handle on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_a41545d33616d1c0825ac336d6c2dcbc2}\label{dba_8php_a41545d33616d1c0825ac336d6c2dcbc2}} \index{dba.php@{dba.php}!dba\_replace@{dba\_replace}} \index{dba\_replace@{dba\_replace}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_replace()}{dba\_replace()}} {\footnotesize\ttfamily dba\+\_\+replace (\begin{DoxyParamCaption}\item[{}]{\$key, }\item[{}]{\$value, }\item[{}]{\$handle }\end{DoxyParamCaption})} Replace or insert entry \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{string \$key }} The key of the entry to be replaced. \begin{DoxyParams}[1]{Parameters} string & {\em \$value} & \\ \hline \end{DoxyParams} The value to be replaced. \begin{DoxyParams}[1]{Parameters} resource & {\em \$handle} & \\ \hline \end{DoxyParams} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} \mbox{\Hypertarget{dba_8php_acd38a98eb82a7341a835d8e821eafb55}\label{dba_8php_acd38a98eb82a7341a835d8e821eafb55}} \index{dba.php@{dba.php}!dba\_sync@{dba\_sync}} \index{dba\_sync@{dba\_sync}!dba.php@{dba.php}} \doxysubsubsection{\texorpdfstring{dba\_sync()}{dba\_sync()}} {\footnotesize\ttfamily dba\+\_\+sync (\begin{DoxyParamCaption}\item[{}]{\$handle }\end{DoxyParamCaption})} Synchronize database \mbox{\hyperlink{patch2_8txt_a3f9992ab8fa942ed02faa63682c3a84c}{resource \$handle }} The database handler, returned by {\bfseries{dba\+\_\+open}} or {\bfseries{dba\+\_\+popen}}. \begin{DoxyReturn}{Returns} bool {\bfseries{T\+R\+UE}} on success or {\bfseries{F\+A\+L\+SE}} on failure. \end{DoxyReturn} \begin{DoxySince}{Since} 4.\+0 5.\+0 \end{DoxySince} 0 \hypertarget{structlimits_3_01long_01double_01_4}{}\doxysection{limits$<$ long double $>$ Struct Reference} \label{structlimits_3_01long_01double_01_4}\index{limits$<$ long double $>$@{limits$<$ long double $>$}} \doxysubsection*{Static Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_a5b1194e1ff037505261fff10f95262ab}\label{structlimits_3_01long_01double_01_4_a5b1194e1ff037505261fff10f95262ab}} static long double {\bfseries min} () \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_a1e0955815aadd9e1b3b9e225606ba996}\label{structlimits_3_01long_01double_01_4_a1e0955815aadd9e1b3b9e225606ba996}} static long double {\bfseries max} () \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_a6c8a51fa51548d8ad77d336b81a36aac}\label{structlimits_3_01long_01double_01_4_a6c8a51fa51548d8ad77d336b81a36aac}} static long double {\bfseries smallest} () \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_add4e9739743af505650a32e19fde761e}\label{structlimits_3_01long_01double_01_4_add4e9739743af505650a32e19fde761e}} static long double {\bfseries epsilon} () \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_a80132dce41c232028ef6e28f6ed9da65}\label{structlimits_3_01long_01double_01_4_a80132dce41c232028ef6e28f6ed9da65}} static bool {\bfseries is\+Integral} () \item \mbox{\Hypertarget{structlimits_3_01long_01double_01_4_a4e7b0192df0329d495ed00e3bd2cb429}\label{structlimits_3_01long_01double_01_4_a4e7b0192df0329d495ed00e3bd2cb429}} static bool {\bfseries is\+Signed} () \end{DoxyCompactItemize} The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item Imath\+Limits.\+h\end{DoxyCompactItemize} watate/nusthesismain.tex %%% enable "DoubleSided" for paper-based submission, i.e. printing, and %%% remove it for the final submission, i.e. electronic thesis %\newcommand*{\DoubleSided}{} \input{_global_settings.tex} \begin{document} \renewcommand{\abstractname}{Summary} \title{Deep learning for map-less navigation: Smooth Velocity Change} \author{ } %\prevdegrees{% % B.S., Culinary University} \degree{Bachelor of Engineering} \field{Mechanical Engineering} \degreeyear{2020} \supervisor{Associate Professor } % only involve the examiners in the final submission %\examiners{% % Associate Professor \\ % Assistant Professor CRAB \\ % Professor , Dessert University} \maketitle \declaredate{27 March 2020} \declaresign{\RootPicDir /signature.jpg} % remove this if you prefer to sign physically \declarationpage \begin{frontmatter} \dedicate{To my friends from ME and USP} \input{chapters/acknowledgments.tex} \tableofcontents \input{chapters/summary.tex} \listoffigures \listofsymbols \end{frontmatter} \input{chapters/ch-intro.tex} \input{chapters/ch-background.tex} \input{chapters/ch-main.tex} \input{chapters/ch-concl.tex} \bookmarksetup{startatroot} \printbibliography[heading=bibintoc] \appendix \input{chapters/ap-worlds.tex} %\input{chapters/ap-publication.tex} % optional to include your publication list \end{document} \hypertarget{category_n_s_file_manager_07hnk__utils_08}{}\section{N\+S\+File\+Manager(hnk\+\_\+utils) カテゴリ} \label{category_n_s_file_manager_07hnk__utils_08}\index{N\+S\+File\+Manager(hnk\+\_\+utils)@{N\+S\+File\+Manager(hnk\+\_\+utils)}} \subsection*{実体メソッド} \begin{DoxyCompactItemize} \item \hypertarget{category_n_s_file_manager_07hnk__utils_08_a6cdd8a9f8304edbc418bbb0be0c6ebcd}{}(void) -\/ {\bfseries hnk\+\_\+enumerate\+Contents\+Of\+Directory\+At\+Path\+:ordered\+By\+Property\+:ascending\+:using\+Block\+:}\label{category_n_s_file_manager_07hnk__utils_08_a6cdd8a9f8304edbc418bbb0be0c6ebcd} \end{DoxyCompactItemize} このカテゴリ詳解は次のファイルから抽出されました\+:\begin{DoxyCompactItemize} \item Framework\+Manager/sample/packages/\+Project\+Package/i\+O\+S\+Sample/\+Libraries/\+Haneke/H\+N\+K\+Cache.\+m\end{DoxyCompactItemize} lrkrol/dissertation \documentclass[a4paper,11pt]{article} \usepackage[utf8]{inputenc} \usepackage{tabularx} \setlength\parindent{0pt} \begin{document} \thispagestyle{empty} \section*{Einverständnis zur kumulativen Dissertation} Ich bin mit dem Format der kumulativen Dissertation und den verwendeten Publikationen einverstanden. Die genutzten Publikationsorgane sind für das angestrebte Fach der Promotion einschlägig. \begin{tabularx}{\textwidth}{XXX} & & \vspace{2cm} \hrule \\ \end{tabularx} \end{document} % BLANK PAGE \DeclareNewLayer[foreground,contents={\parbox[b][\layerheight][c]{\layerwidth}{\centering\latofamily Esta página fue dejada en blanco intencionalmente.}}]{blankpage.fg} \DeclarePageStyleByLayers{blank}{blankpage.fg} \newcommand{\blankpage}{\newpage\null\thispagestyle{blank}\newpage}njcuk9999/neil-thesis-template % Must have the following in root tex file to make this work % % %Make Glossary % \makeglossary % \newglossarytype[nlg]{notation}{not}{ntn} % \newcommand{\notationname}{List of Symbols} % \makenotation % \makeindex % \linespread{1.6} %\notation{{name=},{sort=}{description=}} % name is the text shown as key % sort is the text sorted by (otherwise uses name) % description is the text shown to explain name \notation{{name=$\alpha$,$\delta$},{sort=ad},{description={Right Ascension and declination on the sky, measured in degrees}}} \notation{{name=\micron},{sort=micron},{description=10$^{-6}$ m}} \notation{{name=\Msun},{sort=msun},{description=Solar Mass (1.989 $\times$10$^{30}$ kg)}} \notation{{name=\Rsun},{sort=rsun},{description=Solar Radius (6.963 $\times$10$^{8}$ m)}} \notation{{name=$M_{jup}$},{sort=mjup},{description=Jupiter Mass (1.898 $\times$10$^{27}$ kg)}} \notation{{name=$R_{jup}$},{sort=rjup},{description=Jupiter Radius (6.991 $\times$10$^{7}$ m)}} \notation{{name=\Teff},{sort=teff},{description=A objects effective surface temperature in Kelvin, K}} \notation{{name=\logg},{sort=logg},{description=A objects surface gravity, where g is measured in cgs units}} \notation{{name=\EBV},{sort=ebv},{description=Colour Excess}} \notation{{name=\Av},{sort=av},{description=Visual Extinction}} \notation{{name=\Alam},{sort=alam},{description=Extinction at wavelength, $\lambda$}} \notation{{name=pc},{sort=pc},{description=parsec (3.086 $\times$10$^{16}$ m)}} \notation{{name=AU},{sort=au},{description=Astronomical Unit (1.496 $\times$10$^{11}$ m)}} \notation{{name=\AA},{sort=aa},{description=Angstrom (1.0 $\times$10$^{-10}$ m)}} \notation{{name=$h$},{sort=h},{description=Planck Constant (6.62606896(33) $\times$10$^{-34}$ m$^{2}$ kg s$^{-1}$)}} \notation{{name=$\lambda$},{sort=lam},{description=Wavelength}} \notation{{name={\em c}},{sort=c},{description=Speed of Light (3.0 x 10$^{8}$ m s$^{-1}$)}} \notation{{name=$H_0$},{sort=h0},{description=Hubble constant (72 km s$^{-1}$Mpc$^{-1}$)}} \notation{{name=$k_B$},{sort=kb},{description=Boltzmann constant (1.3806504(24) $\times$10$^{-23}$ J K$^{-1}$)}} \notation{{name=Mpc},{sort=mpc},{description=10$^6$ pc}} \notation{{name=Myr},{sort=myr},{description=10$^6$ yr}} \notation{{name=Gyr},{sort=gyr},{description=10$^9$ yr}} \notation{{name=$\sigma$},{sort=sigma},{description=The uncertainty on a measurement (unless otherwise stated)}} \notation{{name=$\beta$},{sort=beta},{description=The binary fraction}} \notation{{name=$H_X$},{sort=Hx},{description={The reduced proper motion using photometric band X (\ie $X=V$, $X=J$)}}} \notation{{name=$M_X$},{sort=Mx},{description={The absolute magnitude (apparent magnitude at 10 pc) in photometric band X (\ie $X=V$, $X=J$)}}} \notation{{name={$\mu$,$\mu_{\alpha}$,$\mu_{\delta}$}},{sort=mu},{description={The total proper motion and the proper motion components in $\alpha$ and $\delta$, masured in \arcsec yr$^{-1}$}}} \notation{{name={$J$,$H$,$K_S$}},{sort=JHK},{description={2MASS photometric bands (centred at 1.25, 1.65 and 2.16\micron)}}} \notation{{name={$g$,$r$,$i$,$z$}},{sort=gri},{description={SDSS photometric bands (centred at 0.46, 0.62, 0.66 and 0.89\micron)}}} \notation{{name={$V$}},{sort=v},{description={V band taken from the SDSS transformation}}} \notation{{name={$V_J$}},{sort=v},{description={Johnson V magnitude derived from $B_T$ and $V_T$}}} \notation{{name={$V_T$, $B_T$}},{sort=v},{description={V band taken from the Tycho-2 catalogue.}}} \notation{{name={$W1$,$W2$}},{sort=w1w2},{description={WISE photometric bands (centred at 3.4 and 4.6\micron)}}} %currently unused reil/lang.sty \lstdefinelanguage{reil}{ sensitive=true, % comments. % ; line comment comment=[l]{;}, % instructions. % ref: http://www.zynamics.com/binnavi/manual/html/reil_language.htm keywords=[1]{ ADD, SUB, MUL, DIV, MOD, BSH, AND, OR, XOR, LDM, STM, STR, BISZ, JCC, UNDEF, UNKN, NOP, add, sub, mul, div, mod, bsh, and, or, xor, ldm, stm, str, bisz, jcc, undef, unkn, nop }, % registers. % ref: http://www.nasm.us/doc/nasmdo11.html#section-11.1 keywords=[3]{ AH, AL, AX, BH, BL, BP, BPL, BX, CH, CL, CX, DH, DI, DIL, DL, DX, EAX, EBP, EBX, ECX, EDI, EDX, ESI, ESP, R10, R10B, R10D, R10W, R11, R11B, R11D, R11W, R12, R12B, R12D, R12W, R13, R13B, R13D, R13W, R14, R14B, R14D, R14W, R15, R15B, R15D, R15W, R8, R8B, R8D, R8W, R9, R9B, R9D, R9W, RAX, RBP, RBX, RCX, RDI, RDX, RSI, RSP, SI, SIL, SP, SPL, ah, al, ax, bh, bl, bp, bpl, bx, ch, cl, cx, dh, di, dil, dl, dx, eax, ebp, ebx, ecx, edi, edx, esi, esp, r10, r10b, r10d, r10w, r11, r11b, r11d, r11w, r12, r12b, r12d, r12w, r13, r13b, r13d, r13w, r14, r14b, r14d, r14w, r15, r15b, r15d, r15w, r8, r8b, r8d, r8w, r9, r9b, r9d, r9w, rax, rbp, rbx, rcx, rdi, rdx, rsi, rsp, si, sil, sp, spl }, } fermi-lat/CalRecon \documentclass[a4paper,11pt]{article} \title{Calorimeter moments analysis} \author{ (), ()} \usepackage{ifthen} \usepackage{graphicx} \usepackage{amsmath} \usepackage{hyperref} \usepackage{bbm} \usepackage[margin=2cm,bottom=3cm,top=4cm,marginparsep=0pt,marginparwidth=0pt]% {geometry} \usepackage[margin=1cm, font=small, labelfont=bf, labelsep=endash]{caption} \newcommand{\pder}[2]{\frac{\partial#1}{\partial#2}} \newcommand{\pdersec}[3]% {\ifthenelse{\equal{#2}{#3}}% {\frac{\partial^2#1}{\partial#2^2}}% {\frac{\partial^2#1}{\partial#2\partial#3}}% } \newcommand{\itm}{\mathbbm I} \newcommand{\itc}[1]{\itm_{#1}} \newcommand{\firstder}[2]{\frac{{\rm d}#1}{{\rm d}#2}} \newcommand{\secder}[2]{\frac{{\rm d}^2#1}{{\rm d}#2^2}} \newcommand{\tmax}{t_{\rm max}} \newcommand{\diff}{{\rm d}} \newcommand{\xdir}{\ensuremath{x_{\rm dir}}} \newcommand{\ydir}{\ensuremath{y_{\rm dir}}} \newcommand{\zdir}{\ensuremath{z_{\rm dir}}} \begin{document} \maketitle \abstract{These are some (more or less) random notes about the calorimeter moments analysis. The first part is a brief description of the basic code that I put together while trying to understand it. There's also some stuff dealing with possible improvements of the moments analysis, namely the measurement of the shower skeweness and the the error analysis aimed at the projection of the calorimeter clusters into the ACD. } \section{Basics of the CAL moments analysis} The code for the moments analysis basically calculates the moment of inertia tensor (using energy as the weight instead of mass) and then diagonalizes this to get the three principal axes. The basic definitions can be found in \cite{goldstein,landau} and in our case they read: \begin{align} \itc{xx} = \sum_{i=1}^n w_i(r_i^2 - x_i^2),\quad \itc{yy} &= \sum_{i=1}^n w_i(r_i^2 - y_i^2),\quad \itc{zz} = \sum_{i=1}^n w_i(r_i^2 - z_i^2)\\ \itc{xy} = -\sum_{i=1}^n w_ix_iy_i,\quad \itc{xz} &= -\sum_{i=1}^n w_ix_iz_i,\quad \itc{yz} = -\sum_{i=1}^n w_iy_iz_i \end{align} where the index $i$ runs over the $n$ hits in the calorimeter and the $w_i$ are the weights associated with the hits (essentially the energy release). In addition to the moment of inertia, the sum of weights and the coordinates of the energy centroids are also used: \begin{align} W &= \sum_{i=1}^n w_i\\ \mathbf{r}_c &= \frac{\sum_{i=1}^n w_i\mathbf{r}_i}{W} \end{align} In order to reduce the tensor of inertia to the principal axes we do have to solve the secular equation: \begin{equation} \det(\itm - \lambda{\mathbbm 1}) = \det\begin{pmatrix} \itc{xx} - \lambda & \itc{xy} & \itc{xz}\\ \itc{xy} & \itc{yy} - \lambda & \itc{yz}\\ \itc{xz} & \itc{yz} & \itc{zz} - \lambda \end{pmatrix} = 0 \end{equation} which is a cubic equation in $\lambda$ yielding the three eigenvalues. By working out the tedious algebra we can write the equation as: $$ \lambda^3 + c_2\lambda^2 + c_1\lambda + c_0 = 0 $$ where: \begin{align} c_2 &= -(\itc{xx} + \itc{yy} + \itc{zz})\\ c_1 &= \itc{xx}\itc{yy} + \itc{yy}\itc{zz} + \itc{xx}\itc{zz} - (\itc{xy}^2 + \itc{yz}^2 + \itc{xz}^2)\\ c_0 &= -\itc{xx}\itc{yy}\itc{zz} - 2\itc{xy}\itc{yz}\itc{xz} + \itc{xx}\itc{yz}^2 + \itc{yy}\itc{xz}^2 + \itc{zz}\itc{xy}^2 \end{align} If we now define a new variable $\lambda' = \lambda + c_2/3$, the previous equation becomes: $$ \lambda'^3 + a\lambda' + b = 0 $$ where: \begin{align} a &= \left(\frac{3c_1 - c_2^2}{3}\right)\\ b &= \left(\frac{27c_0 + 2c_2^2 - 9c_1c_2}{27}\right) \end{align} (the algebra is fully worked out in \cite{wolfram}). We now set: \begin{align} m &= 2\sqrt{\frac{-a}{3}}\\ \psi &= \frac{1}{3}\arccos\left(\frac{3b}{am}\right) \end{align} and, finally we get the three real solutions (guranteed by the fact that the tensor of inertia is symmetruc): \begin{align} \lambda_0 &= m\cos(\psi) - c_2/3\\ \lambda_1 &= m\cos(\psi + 2\pi/3) - c_2/3\\ \lambda_2 &= m\cos(\psi + 4\pi/3) - c_2/3 \end{align} Once we have the three eigenvalues we can work out the calcultaion of the eigenvectors $\mathbf{e}^i$ ($i = 1\ldots3$) defined by: $$ \itm\mathbf{e}^i = \lambda_i\mathbf{e}^i $$ Following these conventions, $\lambda_1$ is the largest eigenvalue and, as a consequence, $\mathbf{e}^1$ is the principal axis of the cluster. Once the three principal axis of the cluster have been found, the cluster $\chi^2$ (normalized to the number of \emph{degree of freedom}) is calculated as: \begin{equation} \chi^2 = \frac{\sum_{i=1}^n w_i d_i^2}{nW} \end{equation} where $d_i$ are the distances from each of the calorimeter hits to the axis parallel to $\mathbf{e}^1$ and passing throught the cluster centroid. Finally the some well know Merit quantities are calculated: \begin{align} \texttt{CalTransRms} &= \sqrt{\frac{|\lambda_1|}{W}}\\ \texttt{CalLongRms} &= \sqrt{\frac{|\lambda_0| + |\lambda_2|}{2W\log L}}\\ \texttt{CalLRmsAsym} &= \sqrt{\frac{|\lambda_0| - |\lambda_2|} {|\lambda_0| + |\lambda_2|}} \end{align} where $L$ is the number of radiation lengths transversed. \subsection{Outline of the iterative moments analysis} Put here some details about the iteration scheme, as they are relevant for the calculation of the skewness (i.e. you get significantly different answers at different steps). \section{The (toy) problem in 2D} I thought the problem of the diagonalization of the inertia tensor in 2D could be useful for working out the error analysis, so here is a quick look at it. The basic definitions read as: \begin{align} \itc{xx} = \sum_{i=1}^n w_i y_i^2,\quad \itc{yy} = \sum_{i=1}^n w_i x_i^2,\quad \itc{xy} = -\sum_{i=1}^n w_ix_iy_i \end{align} and the secular equation is: \begin{equation} \det(\itm - \lambda{\mathbbm 1}) = \det\begin{pmatrix} \itc{xx} - \lambda & \itc{xy}\\ \itc{xy} & \itc{yy} - \lambda \end{pmatrix} = 0 \end{equation} The eigenvalues are readily found: \begin{align} \lambda_0 &= \frac{(\itc{xx} + \itc{yy}) - \sqrt{(\itc{xx} - \itc{yy})^2 + 4\itc{xy}^2}}{2}\\ \lambda_1 &= \frac{(\itc{xx} + \itc{yy}) + \sqrt{(\itc{xx} - \itc{yy})^2 + 4\itc{xy}^2}}{2}\\ \end{align} At this point, assuming that $\lambda_0$ is the smallest eigenvalue, we're left with the problem of calculating the corresponding eigenvector $\mathbf{e}^0$, obeying the equation: $$ \itm \mathbf{e}^0 = \lambda_0\mathbf{e}^0 $$ Being the problem bi-dimensional, the two eigenvectors can be parametrized in terms of a single rotation angle $\phi$ (to be determined), representing the angle of the principal axis with respect to the original reference frame i.e.: \begin{equation} \begin{pmatrix} e^0_x\\ e^0_y \end{pmatrix} = \begin{pmatrix} \cos\phi\\ \sin\phi \end{pmatrix},\quad \begin{pmatrix} e^1_x\\ e^1_y \end{pmatrix} = \begin{pmatrix} -\sin\phi\\ \cos\phi \end{pmatrix} \end{equation} By putting everything together, from the definition of the principal axis we get: $$ \begin{pmatrix} \itc{xx} & \itc{xy}\\ \itc{xy} & \itc{yy} \end{pmatrix} \begin{pmatrix} \cos\phi\\ \sin\phi \end{pmatrix} = \lambda_1 \begin{pmatrix} \cos\phi\\ \sin\phi \end{pmatrix} $$ The first of the two equations in the system can be squared, yielding: $$ \frac{(\itc{xx} - \itc{yy})}{\itc{xy}}\tan\phi = \tan^2\phi - 1 $$ and eventually, through the trigonometric equation $$ \tan(2\phi) = \frac{2\tan\phi}{1 - \tan^2\phi} $$ we get: \begin{equation} \phi = -\frac{1}{2} \arctan \left( \frac{2\itc{xy}}{\itc{yy} - \itc{xx}} \right) \end{equation} The rotation matrix between the original system and the one defined by the principal axis has the (transpose of) the two eigenvectors as its rows: \begin{equation} S = \begin{pmatrix} e^0_x & e^0_y\\ e^1_x & e^1_y \end{pmatrix} = \begin{pmatrix} \cos\phi & \sin\phi\\ -\sin\phi & \cos\phi \end{pmatrix} \end{equation} and obviously we have: \begin{equation}\label{eq:rotation} \lambda = \begin{pmatrix} \lambda_0 & 0\\ 0 & \lambda_1 \end{pmatrix} = S\;\itm\;S^{-1} = S\;\itm\;S^{\rm T} \end{equation} \subsection{Error analysis in 2 dimensions} As a first ingredient we'll need the derivatives of the components of the tensor of inertia with respect to the coordinates and the weights: \begin{align} \pder{\itc{xx}}{x_i} &= 0,\quad \pder{\itc{xx}}{y_i} = 2w_iy_i,\quad \pder{\itc{xx}}{w_i} = y_i^2\\ \pder{\itc{yy}}{x_i} &= 2w_ix_i,\quad \pder{\itc{yy}}{y_i} = 0,\quad \pder{\itc{yy}}{w_i} = x_i^2\\ \pder{\itc{xy}}{x_i} &= -w_iy_i,\quad \pder{\itc{xy}}{y_i} = -w_ix_i,\quad \pder{\itc{xy}}{w_i} = -x_iy_i \end{align} We can then calculate the full covariance matrix of the errors using the usual formula (see \cite{pdg}, section 32.1.4 for instance). Assuming that the errors on the two positions and on the weights are mutually not correlated (i.e. their covariance matrix is diagonal), we have: \begin{equation} \Sigma_{k-l} = \sum_{i=1}^n \pder{\itc{k}}{x_i}\pder{\itc{l}}{x_i}(\Delta x_i)^2 + \pder{\itc{k}}{y_i}\pder{\itc{l}}{y_i}(\Delta y_i)^2 + \pder{\itc{k}}{w_i}\pder{\itc{l}}{w_i}(\Delta w_i)^2 \end{equation} where $k$ and $l$ run over the three (double) indexes $xx$, $yy$ and $xy$. We're ready to work out the details: \begin{align} \Sigma_{xx-xx} &= (\Delta\itc{xx})^2 = \sum_{i=1}^n \left[4w^2_iy^2_i(\Delta y_i)^2 + y_i^4(\Delta w_i)^2\right]\\ \Sigma_{yy-yy} &= (\Delta\itc{yy})^2 = \sum_{i=1}^n \left[4w^2_ix^2_i(\Delta x_i)^2 + x_i^4(\Delta w_i)^2\right]\\ \Sigma_{xy-xy} &= (\Delta\itc{xy})^2 = \sum_{i=1}^n \left[w_i^2y_i^2(\Delta x_i)^2 + w_i^2x_i^2(\Delta y_i)^2 + x_i^2y_i^2(\Delta w_i)^2 \right]\\ \Sigma_{xx-yy} &= \sum_{i=1}^n \left[ x_i^2y_i^2 (\Delta w_i)^2\right]\\ \Sigma_{xx-xy} &= -\sum_{i=1}^n \left[ 2w_i^2x_iy_i(\Delta y_i)^2 + x_iy_i^3(\Delta w_i)^2 \right]\\ \Sigma_{xy-yy} &= -\sum_{i=1}^n \left[ 2w_i^2x_iy_i(\Delta x_i)^2 + x_i^3y_i(\Delta w_i)^2 \right] \end{align} The rest of this section follows closely the prescription described in \cite{errors} for the error propagation. We can slice the $2 \times 2$ tensor of inertia and define a 4-component vector with the two columns one on top of the other (this is what we call the $vec$ operator): \begin{equation} vec(\itm) = \begin{pmatrix} \itc{xx}\\ \itc{xy}\\ \itc{xy}\\ \itc{yy} \end{pmatrix} \end{equation} That said, we can rewite the equation (\ref{eq:rotation}) using the Kronecker product of the rotation matrix $$ T = S \otimes S $$ as: \begin{equation}\label{eq:tensor_transform} vec(\lambda) = \begin{pmatrix} \lambda_0\\ 0\\ 0\\ \lambda_1 \end{pmatrix} = T vec(\itm) \end{equation} It is useful to rearrange the the elements of the $vec$ operator in such a way that the diagonal elements of the tensor come first, followed by the non-diagonal ones, getting rid of the duplicated terms. This is accomplished by introducing the matrix: \begin{equation} D = \begin{pmatrix} 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 1\\ 0 & \frac{1}{2} & \frac{1}{2} & 0 \end{pmatrix} \end{equation} which allows to define a new operator $v_d$: \begin{equation} v_d(\itm) = D vec(\itm) = \begin{pmatrix} 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 1\\ 0 & \frac{1}{2} & \frac{1}{2} & 0 \end{pmatrix} \begin{pmatrix} \itc{xx}\\ \itc{xy}\\ \itc{xy}\\ \itc{yy} \end{pmatrix} = \begin{pmatrix} \itc{xx}\\ \itc{yy}\\ \itc{xy} \end{pmatrix} \end{equation} Talking about which, the covariance matrix of $v_d(\itm)$ reads: \begin{equation} \Sigma_{v_d(\itm)} = \begin{pmatrix} \Sigma_{xx-xx} & \Sigma_{xx-yy} & \Sigma_{xx-xy}\\ \Sigma_{xx-yy} & \Sigma_{yy-yy} & \Sigma_{xy-yy}\\ \Sigma_{xx-xy} & \Sigma_{xy-yy} & \Sigma_{xy-xy} \end{pmatrix} \end{equation} in terms of the quantities we have calculated a few lines above. In order to go back from the $v_d$ to the $vec$ representation we need the so called pseudo-inverse of $D$: \begin{equation} D^+ = \begin{pmatrix} 1 & 0 & 0\\ 0 & 0 & 1\\ 0 & 0 & 1\\ 0 & 1 & 0 \end{pmatrix} \end{equation} The paper \cite{errors} is wrong to this respect (see \cite{errors_corr}). Using these definitions, equation (\ref{eq:tensor_transform}) can be rewritten as: \begin{equation} v_d(\lambda) = \begin{pmatrix} \lambda_0\\ \lambda_1\\ 0 \end{pmatrix} = D vec(\lambda) = DT vec(\itm) = DTD^+ v_d(\itm) \end{equation} If we define: \begin{equation} V = DTD^+ \end{equation} we can rewrite the previous equation as: \begin{equation} v_d(\lambda) = V v_d(\itm) \end{equation} The infinitesimal change $\diff S$ in the rotation matrix $S$ when we change the rotation angle $\phi$ by an infinitesimal amount $\diff\phi$ can be easily calculated by differentiating $S$ itself: $$ \diff S = \begin{pmatrix} -\sin\phi\;\diff\phi & \cos\phi\;\diff\phi\\ -\cos\phi\;\diff\phi & -\sin\phi\;\diff\phi\\ \end{pmatrix} $$ If we introduce the antisimmetric tensor: \begin{equation} \Omega = \begin{pmatrix} 0 & -\diff\phi\\ \diff\phi & 0 \end{pmatrix} \end{equation} we can rewrite the previous equation as: \begin{equation} \diff S = -S\Omega \end{equation} and, along the same lines, we have: \begin{equation} \diff S^{\rm T} = \Omega S^{\rm T} \end{equation} as can be verified by direct matrix multiplication. It seems a bit odd, here, to call $\Omega$ an infinitesimal quantity (I would have rather named it $\diff\Omega$), but we'll bravely follow the conventions used in the paper to avoid confusion. We now define the quantity: \begin{equation} \Omega^p = S\Omega S^{\rm T} = \begin{pmatrix} 0 & -\diff\phi\\ \diff\phi & 0\\ \end{pmatrix} = \Omega \end{equation} By putting all together we have: \begin{equation} S \diff \itm S^{\rm T} = \diff\mu = \Omega^p\lambda - \lambda\Omega^p + \diff\lambda = \begin{pmatrix} \diff\lambda_0 & (\lambda_0 - \lambda_1)\diff\phi\\ (\lambda_0 - \lambda_1)\diff\phi & \diff\lambda_1 \end{pmatrix} \end{equation} and we can write: \begin{equation} vec(\diff\mu) = G\beta \end{equation} where: \begin{equation} G = \begin{pmatrix} 1 & 0 & 0\\ 0 & 0 & (\lambda_1 - \lambda_0)\\ 0 & 0 & (\lambda_1 - \lambda_0)\\ 0 & 1 & 0 \end{pmatrix} \end{equation} and \begin{equation} \beta = \begin{pmatrix} \diff\lambda_0\\ \diff\lambda_1\\ -\diff\phi \end{pmatrix} \end{equation} (again, $\beta$ is a differential quantities, so it is a bit odd to name it without a $\diff$ in front). Further on through the paper: \begin{equation} v_d(\diff\itm) = \begin{pmatrix} d\itc{xx}\\ d\itc{yy}\\ d\itc{xy} \end{pmatrix} = D (S^{\rm T} \otimes S^{\rm T}) vec(d\mu) = D (S^{\rm T} \otimes S^{\rm T}) G\beta = F\beta \end{equation} where we have defined: \begin{equation} F = D (S^{\rm T} \otimes S^{\rm T}) G \end{equation} All we have to do is invert this equation, namely find $F^{-1}$. If we were dealing with square matrices, all we had to do would be: $$ F^{-1} = \left( D (S^{\rm T} \otimes S^{\rm T}) G \right)^{-1} = G^{-1} \left( S^{\rm T} \otimes S^{\rm T} \right)^{-1} D^{-1} = G^{-1} \left( S \otimes S \right) D^{-1} = G^{-1} T D^{-1} $$ But in fact $G$ and $D$ are rectangular matrices, so we need again the pseudo inverses. We have already the one for $D$, while for $G$, since $G^TG$ is not singular, the solution is even easier: \begin{equation} G^+ = \left( G^{\rm T}G \right)^{-1} G^{\rm T} = \begin{pmatrix} 1 & 0 & 0 & 0\\ 0 & 0 & 0 & 1\\ 0 & \frac{1}{2(\lambda_1-\lambda_0)} & \frac{1}{2(\lambda_1-\lambda_0)} & 0 \end{pmatrix} \end{equation} (this satisfies $G^+G = I_{3\times3}$, as can be easily verified by direct multiplication). Summarizing: \begin{equation} \beta = F^{-1}v_d(\diff\itm) \end{equation} and, more important, we get the full covariance matrix of the eigenvalues and rotation angle: \begin{equation} \Sigma_\beta = F^{-1} \Sigma_{v_d(\itm)} (F^{-1})^T \end{equation} Once we have the error on the rotation angle: \begin{equation} (\Delta\phi)^2 = (\Sigma_\beta)_{33} \end{equation} the covariance matrix of the two components of the principal axis is: \begin{equation} \Sigma_{\mathbf{e}^0} = \begin{pmatrix} \pder{e^0_x}{\phi}\pder{e^0_x}{\phi}(\Delta\phi)^2 & \pder{e^0_x}{\phi}\pder{e^0_y}{\phi}(\Delta\phi)^2\\ \pder{e^0_x}{\phi}\pder{e^0_y}{\phi}(\Delta\phi)^2 & \pder{e^0_y}{\phi}\pder{e^0_y}{\phi}(\Delta\phi)^2 \end{pmatrix} = \begin{pmatrix} \sin^2\phi & -\sin\phi\cos\phi\\ -\sin\phi\cos\phi & \cos^2\phi \end{pmatrix}(\Delta\phi)^2 \end{equation} \section{Error analysis in 3 dimensions} Along the same lines of the previous subsection the derivatives of the components of the inertia tensor are: \begin{align} \pder{\itc{xx}}{x_i} &= 0,\quad \pder{\itc{xx}}{y_i} = 2w_iy_i,\quad \pder{\itc{xx}}{z_i} = 2w_iz_i,\quad \pder{\itc{xx}}{w_i} = (y_i^2 + z_i^2)\\ \pder{\itc{yy}}{x_i} &= 2w_ix_i,\quad \pder{\itc{yy}}{y_i} = 0,\quad \pder{\itc{yy}}{z_i} = 2w_iz_i,\quad \pder{\itc{yy}}{w_i} = (x_i^2 + z_i^2)\\ \pder{\itc{zz}}{x_i} &= 2w_ix_i,\quad \pder{\itc{zz}}{y_i} = 2w_iy_i,\quad \pder{\itc{zz}}{z_i} = 0,\quad \pder{\itc{zz}}{w_i} = (x_i^2 + y_i^2)\\ \pder{\itc{xy}}{x_i} &= -w_iy_i,\quad \pder{\itc{xy}}{y_i} = -w_ix_i,\quad \pder{\itc{xy}}{z_i} = 0,\quad \pder{\itc{xy}}{w_i} = -x_iy_i\\ \pder{\itc{xz}}{x_i} &= -w_iz_i,\quad \pder{\itc{xz}}{y_i} = 0,\quad \pder{\itc{xz}}{z_i} = -w_ix_i,\quad \pder{\itc{xz}}{w_i} = -x_iz_i\\ \pder{\itc{yz}}{x_i} &= 0,\quad \pder{\itc{yz}}{y_i} = -w_iz_i,\quad \pder{\itc{yz}}{z_i} = -w_iy_i,\quad \pder{\itc{yz}}{w_i} = -y_iz_i \end{align} The elements of the covariance matrix are: \begin{equation} \Sigma_{k-l} = \sum_{i=1}^n \pder{\itc{k}}{x_i}\pder{\itc{l}}{x_i}(\Delta x_i)^2 + \pder{\itc{k}}{y_i}\pder{\itc{l}}{y_i}(\Delta y_i)^2 + \pder{\itc{k}}{z_i}\pder{\itc{l}}{z_i}(\Delta z_i)^2 + \pder{\itc{k}}{w_i}\pder{\itc{l}}{w_i}(\Delta w_i)^2 \end{equation} where now $k$ and $l$ run over the 6 independent (double) indexes of the indertia tensor $xx$, $yy$, $zz$, $xy$, $xz$ and $yz$. Therefore the $6\times6$ covariancewe will have $6(6+1)/2 = 21$ components: \begin{align} \Sigma_{xx-xx} &= \sum_{i=1}^n \left\{ 4w_i^2 \left[ y_i^2(\Delta y_i)^2 + z_i^2(\Delta z_i)^2 \right] + \left(y_i^2 + z_i^2\right)^2(\Delta w_i)^2 \right\}\\ \Sigma_{xx-yy} &= \sum_{i=1}^n \left\{ 4w_i^2z_i^2 (\Delta z_i)^2 + (x_i^2 + z_i^2)(y_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{xx-zz} &= \sum_{i=1}^n \left\{ 4w_i^2y_i^2 (\Delta y_i)^2 + (x_i^2 + y_i^2)(y_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{xx-xy} &= -\sum_{i=1}^n x_iy_i \left\{ 2w_i^2 (\Delta y_i)^2 + (y_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{xx-xz} &= -\sum_{i=1}^n x_iz_i \left\{ 2w_i^2 (\Delta z_i)^2 + (y_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{xx-yz} &= -\sum_{i=1}^n y_iz_i \left\{ 2w_i^2 \left[ (\Delta y_i)^2 + (\Delta z_i)^2 \right] + (y_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{yy-yy} &= \sum_{i=1}^n \left\{ 4w_i^2 \left[ x_i^2(\Delta x_i)^2 + z_i^2(\Delta z_i)^2 \right] + \left(x_i^2 + z_i^2\right)^2(\Delta w_i)^2 \right\}\\ \Sigma_{yy-zz} &= \sum_{i=1}^n \left\{ 4w_i^2x_i^2 (\Delta x_i)^2 + (x_i^2 + y_i^2)(x_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{yy-xy} &= -\sum_{i=1}^n x_iy_i \left\{ 2w_i^2 (\Delta x_i)^2 + (x_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{yy-xz} &= -\sum_{i=1}^n x_iz_i \left\{ 2w_i^2 \left[ (\Delta x_i)^2 + (\Delta z_i)^2 \right] + (x_i^2 + z_i^2)(\Delta w_i)^2 \right\} \end{align} \begin{align} \Sigma_{yy-yz} &= -\sum_{i=1}^n y_iz_i \left\{ 2w_i^2 (\Delta z_i)^2 + (x_i^2 + z_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{zz-zz} &= \sum_{i=1}^n \left\{ 4w_i^2 \left[ x_i^2(\Delta x_i)^2 + y_i^2(\Delta y_i)^2 \right] + \left(x_i^2 + y_i^2\right)^2(\Delta w_i)^2 \right\}\\ \Sigma_{zz-xy} &= -\sum_{i=1}^n x_iy_i \left\{ 2w_i^2 \left[ (\Delta x_i)^2 + (\Delta y_i)^2 \right] + (x_i^2 + y_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{zz-xz} &= -\sum_{i=1}^n x_iz_i \left\{ 2w_i^2 (\Delta x_i)^2 + (x_i^2 + y_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{zz-yz} &= -\sum_{i=1}^n y_iz_i \left\{ 2w_i^2 (\Delta y_i)^2 + (x_i^2 + y_i^2)(\Delta w_i)^2 \right\}\\ \Sigma_{xy-xy} &= \sum_{i=1}^n \left\{ w_i^2 \left[ y_i^2 (\Delta x_i)^2 + x_i^2 (\Delta y_i)^2 \right] + x_i^2y_i^2 (\Delta w_i)^2 \right\}\\ \Sigma_{xy-xz} &= \sum_{i=1}^n y_iz_i \left\{ w_i^2 (\Delta x_i)^2 + x_i^2 (\Delta w_i)^2 \right\}\\ \Sigma_{xy-yz} &= \sum_{i=1}^n x_iz_i \left\{ w_i^2 (\Delta y_i)^2 + y_i^2 (\Delta w_i)^2 \right\}\\ \Sigma_{yz-yz} &= \sum_{i=1}^n \left\{ w_i^2 \left[ z_i^2 (\Delta y_i)^2 + y_i^2 (\Delta z_i)^2 \right] + y_i^2z_i^2 (\Delta w_i)^2 \right\}\\ \Sigma_{yz-xz} &= \sum_{i=1}^n x_iy_i \left\{ w_i^2 (\Delta z_i)^2 + z_i^2 (\Delta w_i)^2 \right\}\\ \Sigma_{xz-xz} &= \sum_{i=1}^n \left\{ w_i^2 \left[ z_i^2 (\Delta x_i)^2 + x_i^2 (\Delta z_i)^2 \right] + x_i^2z_i^2 (\Delta w_i)^2 \right\} \end{align} In analogy with the two-dimensional case we define: \begin{equation} D = \begin{pmatrix} 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1\\ 0 & \frac{1}{2} & 0 & \frac{1}{2} & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & \frac{1}{2} & 0 & 0 & 0 & \frac{1}{2} & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & \frac{1}{2} & 0 & \frac{1}{2} & 0 \end{pmatrix} \end{equation} so that \begin{equation} v_d(\itm) = D vec(\itm) = \begin{pmatrix} \itc{xx}\\ \itc{yy}\\ \itc{zz}\\ \itc{xy}\\ \itc{xz}\\ \itc{yz} \end{pmatrix} \end{equation} and the $6\times6$ covariance matrix of the components of the inertia tensor is: \begin{equation} \Sigma_{v_d(\itm)} = \begin{pmatrix} \Sigma_{xx-xx} & \Sigma_{xx-yy} & \Sigma_{xx-zz} & \Sigma_{xx-xy} & \Sigma_{xx-xz} & \Sigma_{xx-yz}\\ \Sigma_{xx-yy} & \Sigma_{yy-yy} & \Sigma_{yy-zz} & \Sigma_{yy-xy} & \Sigma_{yy-xz} & \Sigma_{yy-yz}\\ \Sigma_{xx-zz} & \Sigma_{yy-zz} & \Sigma_{zz-zz} & \Sigma_{zz-xy} & \Sigma_{zz-xz} & \Sigma_{zz-yz}\\ \Sigma_{xx-xy} & \Sigma_{yy-xy} & \Sigma_{zz-xy} & \Sigma_{xy-xy} & \Sigma_{xy-xz} & \Sigma_{xy-yz}\\ \Sigma_{xx-xz} & \Sigma_{yy-xz} & \Sigma_{zz-xz} & \Sigma_{xy-xz} & \Sigma_{xz-xz} & \Sigma_{xz-yz}\\ \Sigma_{xx-yz} & \Sigma_{yy-yz} & \Sigma_{zz-yz} & \Sigma_{xy-yz} & \Sigma_{xz-yz} & \Sigma_{yz-yz}\\ \end{pmatrix} \end{equation} In the three-dimensional case $D^+$ reads: \begin{equation} D^+ = \begin{pmatrix} 1 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 1 & 0 & 0\\ 0 & 0 & 0 & 0 & 1 & 0\\ 0 & 0 & 0 & 1 & 0 & 0\\ 0 & 1 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 1\\ 0 & 0 & 0 & 0 & 1 & 0\\ 0 & 0 & 0 & 0 & 0 & 1\\ 0 & 0 & 1 & 0 & 0 & 0 \end{pmatrix} \end{equation} and the rotation matrix is obviously: \begin{equation} S = \begin{pmatrix} e^0_x & e^0_y & e^0_z\\ e^1_x & e^1_y & e^1_z\\ e^2_x & e^2_y & e^2_z \end{pmatrix} \end{equation} The last pieces of information that we need are: \begin{equation} S_p = \begin{pmatrix} 0 & S_{20} & -S_{10}\\ -S_{20} & 0 & S_{00}\\ S_{10} & -S_{00} & 0\\ 0 & S_{21} & -S_{11}\\ -S_{21} & 0 & S_{01}\\ S_{11} & -S_{01} & 0\\ 0 & S_{22} & -S_{12}\\ -S_{22} & 0 & S_{02}\\ S_{12} & -S_{02} & 0 \end{pmatrix} \end{equation} and: \begin{equation} G^+ = \begin{pmatrix} 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 1 & 0 & 0 & 0 & 0\\ 0 & 0 & 0 & 0 & 0 & 0 & 0 & 0 & 1\\ 0 & 0 & 0 & 0 & 0 & \frac{1}{2(\lambda_2 - \lambda_1)} & 0 & \frac{1}{2(\lambda_2 - \lambda_1)} & 0\\ 0 & 0 & \frac{1}{2(\lambda_0 - \lambda_2)} & 0 & 0 & 0 & \frac{1}{2(\lambda_0 - \lambda_2)} & 0 & 0\\ 0 & \frac{1}{2(\lambda_1 - \lambda_0)} & 0 & \frac{1}{2(\lambda_1 - \lambda_0)} & 0 & 0 & 0 & 0 & 0 \end{pmatrix} \end{equation} And at this point we just follow the paper (the new one, not the old one): \begin{align} F^{-1} &= G^+ (S \otimes S) D^+\\ K &= \begin{pmatrix} 0_{9\times3} & S_p\\ I_{3\times3} & 0_{3\times3} \end{pmatrix} F^{-1} \end{align} and eventually: \begin{equation} \Sigma_{vec(S), \lambda} = K \Sigma_{v_d(\itm)} K^T \end{equation} \subsection{Simplified error analysis} The main purpose in this section is to recover the expression for the covariance matrix in the $(\xdir, \ydir, \zdir)$ representation when error estimates on $\theta$ and $\phi$ are available---either on an event-by-event basis or through some phenomenological parametrization. We will assume for simplicity that the errors on $\theta$ and $\phi$ are uncorrelated. The basic transformation equations are: \begin{align} \xdir & = \sin\theta \cos\phi\\ \ydir & = \sin\theta \sin\phi\\ \zdir & = \cos\theta \end{align} and the elements of the covariance matrix, as a function of $\theta$ and $\phi$, read: \begin{equation} \Sigma_{ij} = E[C_i C_j] \end{equation} where, denoting the central values with the subscript zero: \begin{align} C_x & = (\sin\theta \cos\phi - \sin\theta_0 \cos\phi_0)\\ C_y & = (\sin\theta \sin\phi - \sin\theta_0 \sin\phi_0)\\ C_z & = (\cos\theta - \cos\theta_0) \end{align} We will take advantage of the fact that in the series expansion of the expectation values: \begin{align} E[f(\theta, \phi)] & \approx f(\theta_0, \phi_0) + \left[\pder{f(\theta, \phi)}{\theta}\right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!E[(\theta - \theta_0)] + \left[\pder{f(\theta, \phi)}{\phi}\right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!E[(\phi - \phi_0)] +\\\nonumber & \frac{1}{2}\left[\pdersec{f(\theta, \phi)}{\theta}{\theta} \right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!E[(\theta - \theta_0)^2] + \frac{1}{2}\left[\pdersec{f(\theta, \phi)}{\phi}{\phi} \right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!E[(\phi - \phi_0)^2] +\\\nonumber & \left[\pdersec{f(\theta, \phi)}{\theta}{\phi}\right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!E[(\theta - \theta_0)(\phi - \phi_0)] \end{align} the first three terms vanish for obvious reasons, and so does the last, if the errors on $\theta$ and $\phi$ are not correlated. Therefore: \begin{align} E[f(\theta, \phi)] \approx \frac{1}{2}\left[\pdersec{f(\theta, \phi)}{\theta}{\theta} \right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!\sigma^2_\theta + \frac{1}{2}\left[\pdersec{f(\theta, \phi)}{\phi}{\phi} \right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\!\!\!\sigma^2_\phi \end{align} In the following we will need: \begin{align} \nonumber \left[\pdersec{C_i(\theta, \phi)C_j(\theta, \phi)}{\theta}{\theta} \right]_{\theta_0, \phi_0} \quad {\rm and}\quad \left[\pdersec{C_i(\theta, \phi)C_j(\theta, \phi)}{\phi}{\phi} \right]_{\theta_0, \phi_0} \end{align} The basic blocks are: \begin{align} \nonumber \pder{C_x}{\theta} &= \cos\theta\cos\phi ,\quad \pder{C_x}{\phi} = -\sin\theta\sin\phi\\\nonumber \pder{C_y}{\theta} &= \cos\theta\sin\phi ,\quad \pder{C_y}{\phi} = \sin\theta\cos\phi\\\nonumber \pder{C_z}{\theta} &= -\sin\theta ,\quad \pder{C_z}{\phi} = 0\\\nonumber \pdersec{C_x}{\theta}{\theta} &= -\sin\theta\cos\phi ,\quad \pdersec{C_x}{\phi}{\phi} = -\sin\theta\cos\phi\\\nonumber \pdersec{C_y}{\theta}{\theta} &= -\sin\theta\sin\phi ,\quad \pdersec{C_y}{\phi}{\phi} = -\sin\theta\sin\phi\\\nonumber \pdersec{C_z}{\theta}{\theta} &= -\cos\theta ,\quad \pdersec{C_z}{\phi}{\phi} = 0\\\nonumber \end{align} And the actual ingredients for the covariance matrix can be easily calculated remembering that all the terms containing a $C_i$ multiplicative factor are zero when calculated in $(\theta_0, \phi_0)$: \begin{align} \left[\pdersec{C^2_x}{\theta}{\theta}\right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\! = \left[\pder{}{\theta}\left(2C_x\pder{C_x}{\theta}\right) \right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\! = \left[2\left(\pder{C_x}{\theta}\right)^2 + 2C_x\pdersec{C_x}{\theta}{\theta}\right]_{\theta_0, \phi_0} \!\!\!\!\!\!\!\! = 2\cos^2\theta_0\cos^2\phi_0 \end{align} So, along the same lines (and defining for compactness $s_{\theta_0} = \sin\theta_0$, $c_{\theta_0} = \cos\theta_0$, $s_{\phi_0} = \sin\phi_0$, $c_{\phi_0} = \cos\phi_0$): \begin{equation} %m[0, 0] = ct2*cp2*dt2 + st2*sp2*dp2 %m[1, 1] = ct2*sp2*dt2 + st2*cp2*dp2 %m[2, 2] = st2*dt2 %m[0, 1] = ct2*cp*sp*dt2 - st2*cp*sp*dp2 %m[0, 2] = -st*ct*cp*dt2 %m[1, 2] = -st*ct*sp*dt2 %m[1, 0] = m[0, 1] %m[2, 0] = m[0, 2] %m[2, 1] = m[1, 2] \Sigma = \begin{pmatrix} c_{\theta_0}^2c_{\phi_0}^2\sigma^2_\theta + s_{\theta_0}^2s_{\phi_0}^2\sigma^2_\phi & c_{\theta}^2 c_{\phi}s_{\phi}\sigma^2_\theta - s_{\theta}^2c_{\phi}s_{\phi}\sigma^2_\phi & -s_{\theta}c_{\theta}c_{\phi} \sigma^2_\theta\\ c_{\theta}^2 c_{\phi}s_{\phi}\sigma^2_\theta - s_{\theta}^2c_{\phi}s_{\phi}\sigma^2_\phi & c_{\theta_0}^2s_{\phi_0}^2\sigma^2_\theta + s_{\theta_0}^2c_{\phi_0}^2\sigma^2_\phi & -s_{\theta}c_{\theta}s_{\phi} \sigma^2_\theta\\ -s_{\theta}c_{\theta}c_{\phi} \sigma^2_\theta & -s_{\theta}c_{\theta}s_{\phi} \sigma^2_\theta & s_{\theta_0}^2\sigma^2_\theta\\ \end{pmatrix} \end{equation} \section{Shower development: basic formul\ae} The longitudinal profile of an electromagnetic shower is described by: \begin{equation} \firstder{E}{t} = E_0 p(t) = E_0 k t^\alpha e^{-bt} \end{equation} where $$ k = \frac{b^{\alpha + 1}}{\Gamma(\alpha + 1)} $$ (with this definition $p(t)$ is normalized to 1 and is therefore a probability density) and the Euler $\Gamma$ function, defined by: $$ \Gamma(\alpha) = \int_{0}^\infty t^{\alpha - 1} e^{-t} \diff t $$ satisfies the well know relation: $$ \Gamma(\alpha + 1) = \alpha \Gamma(\alpha) $$ The position of the shower maximum is given by the condition: $$ \left.\firstder{p}{t}\right|_{\tmax} = k \tmax^{\alpha - 1} e^{-b\tmax} (\alpha - b\tmax) = 0 $$ and therefore: \begin{equation} \tmax = \frac{\alpha}{b} \end{equation} The other two pieces of necessary information are the dependences of $\alpha$ and $b$ on the energy. These are given by the relations: \begin{equation} b \approx 0.5 \end{equation} and: \begin{equation} \tmax = \frac{\alpha}{b} = \ln\left(\frac{E_0}{E_c}\right) + C \end{equation} where $C=0.5$ for photons and $C=-0.5$ for electrons and $E_c$ is the critical energy for the material. \section{Longitudinal moments} Let's start from the calculation of the lowest order moments of the shower longitudinal profile around $t=0$. The first one is the mean: \begin{align} \left< t \right> &= \mu = \int_{0}^\infty t p(t) ] \diff t = k \int_{0}^\infty t^{\alpha + 1} e^{-bt} \diff t =\nonumber\\ &= \frac{b^{\alpha + 1}}{\Gamma(\alpha + 1)} \frac{\Gamma(\alpha + 2)}{b^{\alpha + 2}} = \frac{(\alpha + 1)}{b} \end{align} (i.e. the mean of the profile is exactly $1/b$ radiation lengths to the right of the shower maximum). Along the same lines: \begin{align} \left< t^2 \right> = \frac{b^{\alpha + 1}}{\Gamma(\alpha + 1)} \frac{\Gamma(\alpha + 3)}{b^{\alpha + 3}} = \frac{(\alpha + 2)(\alpha + 1)}{b^2} \end{align} and: \begin{align} \left< t^3 \right> = \frac{b^{\alpha + 1}}{\Gamma(\alpha + 1)} \frac{\Gamma(\alpha + 4)}{b^{\alpha + 4}} = \frac{(\alpha + 3)(\alpha + 2)(\alpha + 1)}{b^3} \end{align} We can apply the usual formul\ae\ for the moments $M_n$ centered around the mean (as opposed to the ones centered around 0): \begin{equation} M_2 = \sigma^2 = \left< t^2 \right> - \mu^2 = \frac{(\alpha + 1)}{b^2} \end{equation} and \begin{equation} M_3 = \left< t^3 \right> - 3\mu\sigma^2 - \mu^3 = \frac{2(\alpha + 1)}{b^3} \end{equation} The skewness $\gamma$ is given by: \begin{equation} \gamma = \frac{M_3}{\sigma^3} = \frac{2}{\sqrt{\alpha + 1}} \end{equation} Let's look at the problem from a different perspective, which will hopefully turn out to be handy in the following. Integrating by parts, we get: \begin{align*} \left< t^n \right> & = k \int_{0}^\infty t^n \cdot t^\alpha e^{-bt} \diff t = k \int_{0}^\infty t^\alpha e^{-bt} \diff\left(\frac{t^{n+1}}{n+1}\right) =\\ &= k\left.\frac{t^{n+1}}{n+1} t^\alpha e^{-bt}\right|_0^\infty - k \int_0^\infty \frac{t^{n+1}}{n+1} \left( \alpha t^{\alpha - 1}e^{-bt} - bt^\alpha e^{-bt} \right) \diff t =\\ &= \frac{kb}{n+1} \int_{0}^\infty t^{\alpha + n + 1} e^{-bt} \diff t - \frac{k\alpha}{n+1} \int_{0}^\infty t^{\alpha + n} e^{-bt} \diff t = \frac{b \left< t^{n+1} \right> - \alpha\left< t^n \right>}{n+1} \end{align*} from which it follows that: \begin{equation} \left< t^{n+1} \right> = \frac{(\alpha + n + 1)}{b}\left< t^n \right> \end{equation} For $n = 1$ we get: $$ \left< t^2 \right> = \frac{(\alpha + 2)}{b}\left< t \right> $$ or: \begin{equation} \sigma^2 = \frac{(\alpha + 2)}{b}\mu - \mu^2 \end{equation} Whereas for $n = 2$: $$ \left< t^3 \right> = \frac{(\alpha + 3)}{b}\left< t^2 \right> $$ which translates into: \begin{equation} \gamma = \frac{\mu}{\sigma^3}\left[ \frac{(\alpha + 3)(\alpha + 2)}{b^2} - 3\sigma^2 - \mu^2 \right] \end{equation} All this equations can be directly verified by plugging in the expressions for $\mu$, $\sigma$ and $\gamma$ explicitly obtained before, but the hope is to generalize them to the case in which we don't sample the entire shower (see the following section). \section{Longitudinal moments over a finite interval} We can generalize the previous relations to the case in which we only sample a finite fraction of the longitudinal shower development, say between $t_1$ and $t_2$. The formalism is essentially identical, except for the fact that now we're dealing with a probability density function over a finite interval: $$ p_{\rm f}(t) = k_{\rm f} t^\alpha e^{-bt} $$ with $k_{\rm f}$ being: $$ k_{\rm f} = \frac{1}{\int_{t_1}^{t_2} t^\alpha e^{-bt} \diff t} $$ (physically $k_{\rm f}$ is the ratio between the raw energy deposited in the calorimeter and the true energy of the particle). So now we have: \begin{equation} \left< t^{n+1} \right> = \frac{(\alpha + n + 1)}{b}\left< t^n \right> - \left.\frac{k_{\rm f}}{b} t^{(\alpha + n + 1)} e^{-bt}\right|_{t_1}^{t_2} \end{equation} and therefore: \begin{equation} \left< t^2 \right> = \frac{(\alpha + 2)}{b}\left< t \right> - \frac{k_{\rm f}}{b} \left[t_2^{(\alpha + 2)} e^{-bt_2} - t_1^{(\alpha + 2)} e^{-bt_1}\right] \end{equation} and: \begin{equation} \left< t^3 \right> = \frac{(\alpha + 3)}{b}\left< t^2 \right> - \frac{k_{\rm f}}{b} \left[t_2^{(\alpha + 3)} e^{-bt_2} - t_1^{(\alpha + 3)} e^{-bt_1}\right] \end{equation} Some more formula that might turn out to be useful for the normalization of the skewness to the expected value for electronmagnetic showers. The moments of the longitudinal distribution can be written as a function of the incomplete gamma function, defined as: \begin{equation} \gamma(\alpha, t) = \frac{1}{\Gamma(\alpha)} \int_{0}^{t} t^{\alpha - 1} e^{-t} \diff t \end{equation} from which it follows that: \begin{equation} \int_{t_1}^{t_2} t^{\alpha} e^{-bt} \diff t = \frac{\Gamma(\alpha+1)}{b^{\alpha+1}} \left( \gamma(\alpha+1, bt_2) - \gamma(\alpha+1, bt_1) \right) \end{equation} If we define: $$ {\mathcal G}(\alpha, b, t_1, t_2) = \frac{\Gamma(\alpha)}{b^{\alpha}} \left( \gamma(\alpha, bt_2) - \gamma(\alpha, bt_1) \right) $$ we have: \begin{align} \left< t^n \right> = \frac{{\mathcal G}(\alpha + n + 1, b, t_1, t_2)} {{\mathcal G}(\alpha + 1, b, t_1, t_2)} \end{align} \clearpage \appendix \emph{Caution: the stuff in the appendix is mostly crap, at this time. I'll move it into appropriate sections as soon as it's in a reasonable shape (and, of course, this does not mean that people should not take a look).} Let's go back to the basic equation for the principal eigenvector: $$ \itm\mathbf{e}^1 = \lambda_1\mathbf{e}^1 $$ Doing a full error propagation is not easy, since in this equation we do have error on the six independent components of the inertia tensor, as well as on the eigenvalue $\lambda_1$ we've just calculated. The errors on the $\itc{ij}$ are reasonably easy to calculate, starting from the errors associated with the finite dimensions of the crystals. On the other side the propagation of the errors to $\lambda_1$ is not trivial, as the expression is complicated. On top of that, these different error are not indipendent from each other, as $\lambda_1$ is calculated starting from the component of the inertia tensor. The solution to this equation is: \begin{align} e^1_x &= \frac{1}{\sqrt{1 + \frac{A^2}{B^2} + \frac{A^2}{C^2}}}\\ e^1_y &= \frac{1}{\sqrt{1 + \frac{B^2}{A^2} + \frac{B^2}{C^2}}}\\ e^1_z &= \frac{1}{\sqrt{1 + \frac{C^2}{A^2} + \frac{C^2}{B^2}}} \end{align} where: \begin{align} A &= \itc{yz}(\itc{xx} - \lambda_1) - \itc{xy}\itc{xz}\\ B &= \itc{xz}(\itc{yy} - \lambda_1) - \itc{xy}\itc{yz}\\ C &= \itc{xy}(\itc{zz} - \lambda_1) - \itc{xz}\itc{yz} \end{align} \begin{thebibliography}{100} \bibitem{goldstein}H.~Goldstein, \emph{Classical mechanics}. \bibitem{landau}L.~D.~Landau, E.~M.~Lif\^sic, \emph{Mechanics}. \bibitem{wolfram}\url{http://mathworld.wolfram.com/CubicFormula.html} \bibitem{pdg}PDG Review of Particle Physics, \emph{Physics Letters B.} (2004) {\bf 592} \bibitem{errors}T.~Soler and B.~H.~W.~van~Gelder, \emph{Geophys. J. Int.} (1991) {\bf 105}, 537--546. \bibitem{errors_corr}T.~Soler and B.~H.~W.~van~Gelder, \emph{Geophys. J. Int.} (2006) {\bf 165}, 382. \end{thebibliography} \end{document} @article{liang2021human, title={Human--Robot Collaboration in Construction: Classification and Research Trends}, author={ Kamat, and }, journal={Journal of Construction Engineering and Management}, volume={147}, number={10}, pages={03121006}, year={2021}, publisher={American Society of Civil Engineers} }\documentclass{beamer} \mode { % The Beamer class comes with a number of default slide themes % which change the colors and layouts of slides. Below this is a list % of all the themes, uncomment each in turn to see what they look like. %\usetheme{default} %\usetheme{AnnArbor} %\usetheme{Antibes} %\usetheme{Bergen} %\usetheme{Berkeley} %\usetheme{Berlin} %\usetheme{Boadilla} %\usetheme{CambridgeUS} %\usetheme{Copenhagen} %\usetheme{Darmstadt} %\usetheme{Dresden} \usetheme{Frankfurt} %\usetheme{Goettingen} %\usetheme{Hannover} %\usetheme{Ilmenau} %\usetheme{JuanLesPins} %\usetheme{Luebeck} %\usetheme{Madrid} %\usetheme{Malmoe} %\usetheme{Marburg} %\usetheme{Montpellier} %\usetheme{PaloAlto} %\usetheme{Pittsburgh} %\usetheme{Rochester} %\usetheme{Singapore} %\usetheme{Szeged} %\usetheme{Warsaw} % As well as themes, the Beamer class has a number of color themes % for any slide theme. Uncomment each of these in turn to see how it % changes the colors of your current slide theme. %\usecolortheme{albatross} %\usecolortheme{beaver} %\usecolortheme{beetle} \usecolortheme{crane} %\usecolortheme{dolphin} %\usecolortheme{dove} %\usecolortheme{fly} %\usecolortheme{lily} %\usecolortheme{orchid} %\usecolortheme{rose} %\usecolortheme{seagull} %\usecolortheme{seahorse} %\usecolortheme{whale} %\usecolortheme{wolverine} %\setbeamertemplate{footline} % To remove the footer line in all slides uncomment this line %\setbeamertemplate{footline}[page number] % To replace the footer line in all slides with a simple slide count uncomment this line %\setbeamertemplate{navigation symbols}{} % To remove the navigation symbols from the bottom of all slides uncomment this line } \usepackage{extpfeil} \usepackage{extarrows} %Allows long equation signs \usepackage{graphicx} % Allows including images \usepackage{booktabs} % Allows the use of \toprule, \midrule and \bottomrule in tables \usepackage{physics} \usepackage{tikz} \usepackage{cite} %花体字母 \usepackage{amsthm,amsmath,amssymb} \usepackage{mathrsfs} \usepackage{dutchcal} \definecolor{tianyiblue}{rgb}{135,206,250} %---------------------------------------------------------------------------------------- % TITLE PAGE %---------------------------------------------------------------------------------------- \title[VP260 RC]{VP260 Recitation Class 1} % The short title appears at the bottom of every slide, the full title is only on the title page \author{} % Your name \institute[UM-SJTU JI] % Your institution as it will appear on the bottom of every slide, may be shorthand to save space { University of Michigan - Shanghai Jiao Tong University Joint Institute\\% Your institution for the title page \medskip } \date{\today} % Date, can be changed to a custom date \begin{document} \begin{frame} \titlepage % Print the title page as the first slide \end{frame} \section{Vector Calculus} % Section title slide, unnumbered \begin{frame}{Vector Algebra} \begin{itemize} \item vector addition: $\va{v} + \va{u} $ \item scalar multiplication: $\lambda \va{v}$ \item dot product: $\va{v} \vdot \va{u}$ \item cross product: $\va{v} \cp \va{u}$ \item scalar triple product: $\va{a} \vdot (\va{b} \cp \va{c}) =\va{b} \vdot (\va{c} \cp \va{a}) = \va{c} \vdot (\va{a} \cp \va{b})$ \item vector triple product: $\va{a} \cp (\va{b} \cp \va{c}) = \va{b} (\va{a} \cdot \va{c}) - \va{c} (\va{a} \cdot \va{b})$ \end{itemize} \end{frame} \begin{frame}{Differential Calculus} \begin{itemize} \item gradient: $$\grad f = \pdv{f}{x} \vu{x} + \pdv{f}{y} \vu{y} + \pdv{f}{z} \vu{z}$$ \item divergence: $\div{\va{v}}$ \item curl: $$\curl{\va{v}} = \vu{x} \left(\pdv{v_z}{y} - \pdv{v_y}{z}\right) + \vu{y} \left( \pdv{v_x}{z} - \pdv{v_z}{x}\right) + \vu{z} \left( \pdv{v_y}{x} - \pdv{v_x}{y}\right)$$ \end{itemize} \end{frame} \begin{frame}{Stoke's formula} \begin{beamerboxesrounded}{Foundamental theorem for gradients} \begin{equation} \int_{\vb{a}}^{\vb{b}} (\grad f) \vdot \dd{\vb*{l}} = f(\vb{b}) - f(\vb{a}) \end{equation} \end{beamerboxesrounded} \begin{beamerboxesrounded}{Divergence theorem} \begin{equation} \int_V (\div \vb{v}) \dd{\tau} = \oint_{\partial V} \vb{v} \vdot \dd{\vb*{A}} \end{equation} \end{beamerboxesrounded} \begin{beamerboxesrounded}{Stoke's theorem} \begin{equation} \int_S (\curl \vb{v}) \vdot \dd{\vb{A}} = \oint_{\partial S} \vb{v} \vdot \dd{\vb{l}} \end{equation} \end{beamerboxesrounded} \end{frame} \section{Electrostatics} \begin{frame}{Charges and Laser Printing} \begin{itemize} \item positive(+), negative(-) \item electric charge: $Q$ (unit: [C]) \item charges with different signs attract each other \end{itemize} \vspace{.5em} \begin{figure}[htbp] \centering \includegraphics[width=8cm]{Images/Laser-printer} \caption{Laser Printing} \end{figure} \end{frame} \begin{frame}{Coulomb Force} \begin{beamerboxesrounded}{Coulomb Force} \begin{equation} \vec{F} = \frac{1}{4\pi\epsilon_0} \frac{\abs{q_1 q_2}}{r^2} \vu{r} \end{equation} \end{beamerboxesrounded} \begin{itemize} \item permittivity of vacuum: $\epsilon_0 = 8.85 \times 10^{-12} \ C^2/(N \cdot m^2)$ \item The direction of $\vu{r}$ is decided by charges. \end{itemize} \end{frame} \begin{frame}{Superposition Principle} \begin{columns} \begin{column}{.5\textwidth} \begin{figure}[htbp] \centering \begin{tikzpicture} \node[fill=blue!20, circle, label=above:$q_0$] (X0) at (5,5) {+}; \node[fill=yellow!50, circle, label=below:$q_1$] (X1) at (6,7) {+}; \node[fill=yellow!50, circle, label=below:$q_2$] (X2) at (3,4) {-}; \node[fill=yellow!50, circle, label=below:$q_3$] (X3) at (5,8) {+}; \node[] (X3') at (5,3) {$F_3$}; \node[label=above:$F_2$] (X2') at (4,4.5) {}; \node[label=left:$F_1$] (X1') at (3.5,2) {}; \draw[-to, line width=.1cm, green!50] (X0) -- (X3'); \draw[-to, line width=.1cm, green!50] (X0) -- (X2'); \draw[-to, line width=.1cm, green!50] (X0) -- (X1'); \end{tikzpicture} \end{figure} \end{column} \begin{column}{.5\textwidth} \begin{equation} \va{F} = \va{F_1} + \va{F_2} + \va{F_3} \end{equation} \end{column} \end{columns} \end{frame} \begin{frame}{Electric Field} \begin{beamerboxesrounded}[shadow=true]{\bf Electric Field} \begin{equation} \vec{E} = \frac{\vec{F}}{q_0} \end{equation} \end{beamerboxesrounded} \begin{itemize} \item Only valid when $q_0$ is a point charge. \item Electric field is a vector field in space. \item "real" physical entity \item unit: $V/m$ or $N/C$ \end{itemize} \end{frame} \begin{frame}{Electric Field Lines} \begin{figure}[htbp] \centering \includegraphics[width=0.3\textwidth]{Images/ele_field.png} \caption{Electrical field} \end{figure} \begin{itemize} \item $\va{E}$ is tangential to the field line. \item Field lines only intersect at point charges. \item Density of field line represents the magnitude of $\va{E}$. \end{itemize} \end{frame} \begin{frame}{Superposition Principle} \begin{columns} \begin{column}{.5\textwidth} \begin{figure}[htbp] \centering \begin{tikzpicture} \node[fill=green!50, circle, label=above:A] (X0) at (5,5) {}; \node[fill=yellow!50, circle, label=below:$q_1$] (X1) at (6,7) {+}; \node[fill=yellow!50, circle, label=below:$q_2$] (X2) at (3,4) {-}; \node[fill=yellow!50, circle, label=below:$q_3$] (X3) at (5,8) {+}; \node[] (X3') at (5,3) {$E_3$}; \node[label=above:$E_2$] (X2') at (4,4.5) {}; \node[label=left:$E_1$] (X1') at (3.5,2) {}; \draw[-to, line width=.1cm, green!50] (X0) -- (X3'); \draw[-to, line width=.1cm, green!50] (X0) -- (X2'); \draw[-to, line width=.1cm, green!50] (X0) -- (X1'); \end{tikzpicture} \end{figure} \end{column} \begin{column}{.5\textwidth} \begin{equation} \va{E} = \va{E_1} + \va{E_2} + \va{E_3} \end{equation} If we put a point charge $q_0$ at A, the Coulomb force will be, \begin{equation} \va{F} = q_0 \va{E}. \end{equation} \end{column} \end{columns} \end{frame} \begin{frame}{Continuous Charge Distributions} \begin{beamerboxesrounded}{Charge density} $\rho(\va{r}) : \mathbb{R}^3 \to \mathbb{R}$ [$C/m^3$] \end{beamerboxesrounded} \begin{itemize} \item The electric charge in a particular volume $V$ is, \begin{equation} Q = \int_V \rho(\vb{r}) \dd{\tau}. \end{equation} \item We can define line charge density $\lambda(\vb{r})$ and surface charge density $\sigma(\vb{r})$ similarly. \end{itemize} \begin{beamerboxesrounded}{Electronic field for continuous distribution} \begin{equation} \va{E} (\va{r}) = \frac{1}{4 \pi \epsilon_0 } \int \frac{\rho(\va{r'})}{\abs{\mathbcal{r}}^2} \vu{\mathbcal{r}} \dd{\tau} \end{equation} \end{beamerboxesrounded} \begin{itemize} \item $\va{\mathbcal{r}} = \va{r'} - \va{r}$ \end{itemize} \end{frame} \begin{frame}{Dirac delta} \textbf{How to consider point charge as a "continuous" distribution?} \begin{itemize} \item It should be infinity at a point and zero elsewhere. \end{itemize} \begin{beamerboxesrounded}{Dirac delta} \begin{equation} \delta(x) = \begin{cases} \infty, & x = 0\\ 0,& x \neq 0 \end{cases} \end{equation} \end{beamerboxesrounded} \begin{itemize} \item It is not a function! \item We can define it in mathematics as "generalized functions" or "distributions". \item important property: $$\int \delta(x - a) f(x) \dd{x} = f(a) $$ \end{itemize} \end{frame} \begin{frame}{Electric Dipole} \begin{figure}[htbp] \centering \begin{tikzpicture} \node[fill=blue!20, circle, label=below:-q] (N) at (0,0) {}; \node[fill=red!20, circle, label=above:+q] (P) at (1,1) {}; \draw[-to, line width=.1cm, black!50] (N) -- (P); \draw[-to, line width=.05cm, black] (-0.5,1) -- (0,1); \node[label=above:$\va{E}$] (E) at (0,1) {}; \end{tikzpicture} \end{figure} \begin{beamerboxesrounded}{Electric dipole} \begin{equation} \va{p} = q \va{l} \end{equation} \begin{equation} \va{\tau} = \va{p} \cp \va{E} \end{equation} \begin{equation} U = - \va{p} \vdot \va{E} \end{equation} \end{beamerboxesrounded} \end{frame} \begin{frame}{Gauss' Law} \begin{beamerboxesrounded}{Electric flux} \begin{equation} \Phi_E = \int_{S} \va{E} \vdot \dd{\va{A}} \end{equation} \end{beamerboxesrounded} \vspace{1em} \begin{beamerboxesrounded}{Gauss' law} \begin{equation} \oint_{S} \va{E} \cdot \dd{\va{A}} = \frac{q_{enc}}{\epsilon_0} \end{equation} \begin{equation} \div \va{E} = \frac{\rho(\vec{r})}{\epsilon_0} \end{equation} \end{beamerboxesrounded} \begin{itemize} \item The $q_{enc}$ is the total charge enclosed in the surface. \end{itemize} \end{frame} \section{Exercise} \begin{frame}{Exercise 1} Find the electric field a distance $z$ above the center of a square loop (side $a$) carrying uniform line charge $\lambda$. \begin{figure}[htbp] \centering \includegraphics{Images/ex1.jpg} \end{figure} \end{frame} \begin{frame}{Exercise 2} An infinite cylinder of radius R is charged non-uniformly with the density of charge $\rho=Ar$, where A is a positive constant. Find the electric field at any point of space. (consider both: $rR$) \end{frame} \begin{frame}{Exercise 3} Consider a uniform insulating ball with bulk density of charge $\rho < 0$. Imagine that a part of the ball has been removed, leaving an empty spherical bubble inside the ball. The distance between the center of the ball and the center of the bubble is $r_0$. Use the superposition principle to find the electric field (both magnitude and direction) inside the bubble. \begin{figure}[H] \centering \includegraphics{Images/E2.png} \end{figure} \end{frame} \begin{frame}{Exercise 4} Three infinitely large metal board carry charges $Q_1, Q_2, Q_3$ correspondingly and uniformly. What is the charge on each side of the three boards? \begin{figure}[H] \centering \begin{tikzpicture} \node[label=above:$Q_1$] (q1) at (0,4) {}; \node[label=above:$Q_2$] (q2) at (2,4) {}; \node[label=above:$Q_3$] (q3) at (4,4) {}; \draw[line width=.3cm, blue!70] (0,0) -- (0,4); \draw[line width=.3cm, blue!70] (2,0) -- (2,4); \draw[line width=.3cm, blue!70] (4,0) -- (4,4); \end{tikzpicture} \end{figure} \end{frame} \section{Appendix} \begin{frame} \begin{center} \LARGE\bf Thank for listening! \end{center} \end{frame} \begin{frame}{\bf References} \nocite{*} % Display all references regardless of if they were cited \bibliography{rc1.bib} \bibliographystyle{plain} \end{frame} \end{document} \begin{frame} \frametitle{Instant Meshes} Apresenta uma nova abordagem para refazer uma superfície em uma malha triangular isotrópica ou quad-dominante usando um operador de suavização local unificado que otimiza as orientações da borda e as posições dos vértices na malha de saída. O algoritmo produz malhas com alta isotropia, alinhando e ajustando naturalmente as arestas às características nítidas. \end{frame} \begin{frame} \frametitle{OpenMVG (Multiple View Geometry)} OpenMVG (Multiple View Geometry) "open Multiple View Geometry" é uma biblioteca de visão computacional especialmente direcionada à comunidade da Multiple View Geometry. Projetada para fornecer acesso fácil aos algorítmos clássicos de geometria computacional e auxiliar na solução de problemas com precisão. \end{frame} KPO-2020-2021/zad5_3-Patidzon \hypertarget{dir_631bcb342c210cbc94ad97e2e86b3766}{}\doxysection{/home/patryk/\+Pulpit/zad5\+\_\+3-\/\+Patidzon/build Directory Reference} \label{dir_631bcb342c210cbc94ad97e2e86b3766}\index{/home/patryk/Pulpit/zad5\_3-\/Patidzon/build Directory Reference@{/home/patryk/Pulpit/zad5\_3-\/Patidzon/build Directory Reference}} \doxysubsection*{Directories} \begin{DoxyCompactItemize} \item directory \mbox{\hyperlink{dir_c61bce3557cc9f7bef32d7e5c9c33299}{C\+Make\+Files}} \item directory \mbox{\hyperlink{dir_1e36efcf4bb3ab794ea7aea863891213}{tests}} \end{DoxyCompactItemize} \doxysubsection*{Files} \begin{DoxyCompactItemize} \item file \mbox{\hyperlink{_dart_configuration_8tcl}{Dart\+Configuration.\+tcl}} \item file \mbox{\hyperlink{example_config_8h}{example\+Config.\+h}} \end{DoxyCompactItemize} \hypertarget{dir_0379a6ec9eef3d94449627e8b4297556}{}\section{includes/netflex/http Directory Reference} \label{dir_0379a6ec9eef3d94449627e8b4297556}\index{includes/netflex/http Directory Reference@{includes/netflex/http Directory Reference}} TheMisterPenguin/ProjetL2 \hypertarget{objet_8c}{}\doxysection{Référence du fichier objet.\+c} \label{objet_8c}\index{objet.c@{objet.c}} Fichier contenant toutes les fonctions concernant les objets. Graphe des dépendances par inclusion de objet.\+c\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{dd/de8/objet_8c__incl} \end{center} \end{figure} \doxysubsection*{Fonctions} \begin{DoxyCompactItemize} \item objet\+\_\+t $\ast$ \mbox{\hyperlink{objet_8c_a85e95708f68baa868d64811e0259c340}{creer\+\_\+objet}} (const int id, const char $\ast$const texture\+\_\+src, const \mbox{\hyperlink{objet_8h_a5164672bbab2547df809744bf0af3dc9}{t\+\_\+item}} type, const char $\ast$nom, const short int niveau, const int att, const int def, const int vit) \begin{DoxyCompactList}\small\item\em Créé un objet du jeu. \end{DoxyCompactList}\item void \mbox{\hyperlink{objet_8c_a0cad7222129cd23bfca9ab8d94885f39}{afficher\+\_\+objet}} (objet\+\_\+t $\ast$obj) \begin{DoxyCompactList}\small\item\em Affiche les caractéristiques d\textquotesingle{}un objet dans la console. \end{DoxyCompactList}\item void \mbox{\hyperlink{objet_8c_a9903bbbae9d8ad00d41b6e903d459877}{detruire\+\_\+objet}} (objet\+\_\+t $\ast$$\ast$obj) \begin{DoxyCompactList}\small\item\em Libère la mémoire allouée à un objet. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Description détaillée} \begin{DoxyAuthor}{Auteur} (\href{mailto:}{\texttt{ Max.\+Descomps.\+}}) \end{DoxyAuthor} \begin{DoxyVersion}{Version} 0.\+2 \end{DoxyVersion} \begin{DoxyDate}{Date} 24/02/2022 \end{DoxyDate} \begin{DoxyCopyright}{Copyright} Copyright (c) 2022 \end{DoxyCopyright} \doxysubsection{Documentation des fonctions} \mbox{\Hypertarget{objet_8c_a0cad7222129cd23bfca9ab8d94885f39}\label{objet_8c_a0cad7222129cd23bfca9ab8d94885f39}} \index{objet.c@{objet.c}!afficher\_objet@{afficher\_objet}} \index{afficher\_objet@{afficher\_objet}!objet.c@{objet.c}} \doxysubsubsection{\texorpdfstring{afficher\_objet()}{afficher\_objet()}} {\footnotesize\ttfamily void afficher\+\_\+objet (\begin{DoxyParamCaption}\item[{objet\+\_\+t $\ast$}]{obj }\end{DoxyParamCaption})} \begin{DoxyAuthor}{Auteur} \end{DoxyAuthor} \begin{DoxyParams}{Paramètres} {\em obj} & L\textquotesingle{}objet à afficher \\ \hline \end{DoxyParams} Voici le graphe des appelants de cette fonction \+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=309pt]{d2/dd1/objet_8c_a0cad7222129cd23bfca9ab8d94885f39_icgraph} \end{center} \end{figure} \mbox{\Hypertarget{objet_8c_a85e95708f68baa868d64811e0259c340}\label{objet_8c_a85e95708f68baa868d64811e0259c340}} \index{objet.c@{objet.c}!creer\_objet@{creer\_objet}} \index{creer\_objet@{creer\_objet}!objet.c@{objet.c}} \doxysubsubsection{\texorpdfstring{creer\_objet()}{creer\_objet()}} {\footnotesize\ttfamily objet\+\_\+t$\ast$ creer\+\_\+objet (\begin{DoxyParamCaption}\item[{const int}]{id, }\item[{const char $\ast$const}]{texture\+\_\+src, }\item[{const \mbox{\hyperlink{objet_8h_a5164672bbab2547df809744bf0af3dc9}{t\+\_\+item}}}]{type, }\item[{const char $\ast$}]{nom, }\item[{const short int}]{niveau, }\item[{const int}]{att, }\item[{const int}]{def, }\item[{const int}]{vit }\end{DoxyParamCaption})} \begin{DoxyAuthor}{Auteur} \end{DoxyAuthor} \begin{DoxyParams}{Paramètres} {\em id} & L\textquotesingle{}identificateur de l\textquotesingle{}objet \\ \hline {\em texture\+\_\+src} & Chemin vers l\textquotesingle{}image de l\textquotesingle{}objet \\ \hline {\em type} & Type d\textquotesingle{}objet \\ \hline {\em nom} & Nom de l\textquotesingle{}objet \\ \hline {\em niveau} & Niveau nécessaire pour s\textquotesingle{}équiper de l\textquotesingle{}objet \\ \hline {\em att} & Bonus d\textquotesingle{}attaque de l\textquotesingle{}objet \\ \hline {\em def} & Bonus de défense de l\textquotesingle{}objet \\ \hline {\em vit} & Bonus de vitesse de l\textquotesingle{}objet \\ \hline \end{DoxyParams} \begin{DoxyReturn}{Renvoie} Instance nouvellement allouée du type objet\+\_\+t ou N\+U\+LL \end{DoxyReturn} Voici le graphe des appelants de cette fonction \+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d2/dd1/objet_8c_a85e95708f68baa868d64811e0259c340_icgraph} \end{center} \end{figure} \mbox{\Hypertarget{objet_8c_a9903bbbae9d8ad00d41b6e903d459877}\label{objet_8c_a9903bbbae9d8ad00d41b6e903d459877}} \index{objet.c@{objet.c}!detruire\_objet@{detruire\_objet}} \index{detruire\_objet@{detruire\_objet}!objet.c@{objet.c}} \doxysubsubsection{\texorpdfstring{detruire\_objet()}{detruire\_objet()}} {\footnotesize\ttfamily void detruire\+\_\+objet (\begin{DoxyParamCaption}\item[{objet\+\_\+t $\ast$$\ast$}]{obj }\end{DoxyParamCaption})} \begin{DoxyAuthor}{Auteur} \end{DoxyAuthor} \begin{DoxyParams}{Paramètres} {\em obj} & L\textquotesingle{}objet à libérer \\ \hline \end{DoxyParams} Voici le graphe d\textquotesingle{}appel pour cette fonction \+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=296pt]{d2/dd1/objet_8c_a9903bbbae9d8ad00d41b6e903d459877_cgraph} \end{center} \end{figure} Voici le graphe des appelants de cette fonction \+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=312pt]{d2/dd1/objet_8c_a9903bbbae9d8ad00d41b6e903d459877_icgraph} \end{center} \end{figure} urbanjost/M_CLI2docs/doxygen_out/latex/interfacem__cli2_1_1str.tex1-10 \hypertarget{interfacem__cli2_1_1str}{}\doxysection{m\+\_\+cli2\+::str Interface Reference} \label{interfacem__cli2_1_1str}\index{m\_cli2::str@{m\_cli2::str}} \doxysubsection*{Private Member Functions} \begin{DoxyCompactItemize} \item character(len=\+:) function, allocatable \mbox{\hyperlink{interfacem__cli2_1_1str_aa674bc2d219db87c434b2e7bcdf90ea9}{msg\+\_\+scalar}} (generic0, generic1, generic2, generic3, generic4, generic5, generic6, generic7, generic8, generic9, generica, genericb, genericc, genericd, generice, genericf, genericg, generich, generici, genericj, sep) \item character(len=\+:) function, allocatable \mbox{\hyperlink{interfacem__cli2_1_1str_a9a992f68fd52d921b6c33c1979f381fb}{msg\+\_\+one}} (generic0, generic1, generic2, generic3, generic4, generic5, generic6, generic7, generic8, generic9, sep) \end{DoxyCompactItemize} \doxysubsection{Member Function/\+Subroutine Documentation} \mbox{\Hypertarget{interfacem__cli2_1_1str_a9a992f68fd52d921b6c33c1979f381fb}\label{interfacem__cli2_1_1str_a9a992f68fd52d921b6c33c1979f381fb}} \index{m\_cli2::str@{m\_cli2::str}!msg\_one@{msg\_one}} \index{msg\_one@{msg\_one}!m\_cli2::str@{m\_cli2::str}} \doxysubsubsection{\texorpdfstring{msg\_one()}{msg\_one()}} {\footnotesize\ttfamily character(len=\+:) function, allocatable m\+\_\+cli2\+::str\+::msg\+\_\+one (\begin{DoxyParamCaption}\item[{class($\ast$), dimension(\+:), intent(in)}]{generic0, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic1, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic2, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic3, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic4, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic5, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic6, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic7, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic8, }\item[{class($\ast$), dimension(\+:), intent(in), optional}]{generic9, }\item[{character(len=$\ast$), intent(in), optional}]{sep }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [private]}} \mbox{\Hypertarget{interfacem__cli2_1_1str_aa674bc2d219db87c434b2e7bcdf90ea9}\label{interfacem__cli2_1_1str_aa674bc2d219db87c434b2e7bcdf90ea9}} \index{m\_cli2::str@{m\_cli2::str}!msg\_scalar@{msg\_scalar}} \index{msg\_scalar@{msg\_scalar}!m\_cli2::str@{m\_cli2::str}} \doxysubsubsection{\texorpdfstring{msg\_scalar()}{msg\_scalar()}} {\footnotesize\ttfamily character(len=\+:) function, allocatable m\+\_\+cli2\+::str\+::msg\+\_\+scalar (\begin{DoxyParamCaption}\item[{class($\ast$), intent(in), optional}]{generic0, }\item[{class($\ast$), intent(in), optional}]{generic1, }\item[{class($\ast$), intent(in), optional}]{generic2, }\item[{class($\ast$), intent(in), optional}]{generic3, }\item[{class($\ast$), intent(in), optional}]{generic4, }\item[{class($\ast$), intent(in), optional}]{generic5, }\item[{class($\ast$), intent(in), optional}]{generic6, }\item[{class($\ast$), intent(in), optional}]{generic7, }\item[{class($\ast$), intent(in), optional}]{generic8, }\item[{class($\ast$), intent(in), optional}]{generic9, }\item[{class($\ast$), intent(in), optional}]{generica, }\item[{class($\ast$), intent(in), optional}]{genericb, }\item[{class($\ast$), intent(in), optional}]{genericc, }\item[{class($\ast$), intent(in), optional}]{genericd, }\item[{class($\ast$), intent(in), optional}]{generice, }\item[{class($\ast$), intent(in), optional}]{genericf, }\item[{class($\ast$), intent(in), optional}]{genericg, }\item[{class($\ast$), intent(in), optional}]{generich, }\item[{class($\ast$), intent(in), optional}]{generici, }\item[{class($\ast$), intent(in), optional}]{genericj, }\item[{character(len=$\ast$), intent(in), optional}]{sep }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [private]}} \hypertarget{namespacem__cli2_autotoc_md115}{}\doxysubsubsection{N\+A\+ME}\label{namespacem__cli2_autotoc_md115} str(3f) -\/ \mbox{[}M\+\_\+\+C\+L\+I2\mbox{]} converts any standard scalar type to a string (L\+I\+C\+E\+N\+SE\+:PD) \hypertarget{namespacem__cli2_autotoc_md116}{}\doxysubsubsection{S\+Y\+N\+O\+P\+S\+IS}\label{namespacem__cli2_autotoc_md116} \begin{DoxyVerb}function str(g0,g1,g2,g3,g4,g5,g6,g7,g8,g9,ga,gb,gc,gd,ge,gf,gg,gh,gi,gj,sep) class(*),intent(in),optional :: g0,g1,g2,g3,g4,g5,g6,g7,g8,g9 class(*),intent(in),optional :: ga,gb,gc,gd,ge,gf,gg,gh,gi,gj character(len=*),intent(in),optional :: sep character,len=(:),allocatable :: str \end{DoxyVerb} \hypertarget{namespacem__cli2_autotoc_md117}{}\doxysubsubsection{D\+E\+S\+C\+R\+I\+P\+T\+I\+ON}\label{namespacem__cli2_autotoc_md117} str(3f) builds a space-\/separated string from up to twenty scalar values.\hypertarget{namespacem__cli2_autotoc_md118}{}\doxysubsubsection{O\+P\+T\+I\+O\+NS}\label{namespacem__cli2_autotoc_md118} g\mbox{[}0-\/9a-\/j\mbox{]} optional value to print the value of after the message. May be of type I\+N\+T\+E\+G\+ER, L\+O\+G\+I\+C\+AL, R\+E\+AL, D\+O\+U\+B\+L\+E\+P\+R\+E\+C\+I\+S\+I\+ON, C\+O\+M\+P\+L\+EX, or C\+H\+A\+R\+A\+C\+T\+ER. Optionally, all the generic values can be single-\/dimensioned arrays. Currently, mixing scalar arguments and array arguments is not supported. sep separator to place between values. Defaults to a space. \hypertarget{namespacem__cli2_autotoc_md119}{}\doxysubsubsection{R\+E\+T\+U\+R\+NS}\label{namespacem__cli2_autotoc_md119} str description to print \hypertarget{namespacem__cli2_autotoc_md120}{}\doxysubsubsection{E\+X\+A\+M\+P\+L\+ES}\label{namespacem__cli2_autotoc_md120} Sample program\+: \begin{DoxyVerb} program demo_msg use M_CLI2, only : str implicit none character(len=:),allocatable :: pr character(len=:),allocatable :: frmt integer :: biggest pr=str('HUGE(3f) integers',huge(0),'and real',huge(0.0),'and double',huge(0.0d0)) write(*,'(a)')pr pr=str('real :',huge(0.0),0.0,12345.6789,tiny(0.0) ) write(*,'(a)')pr pr=str('doubleprecision :',huge(0.0d0),0.0d0,12345.6789d0,tiny(0.0d0) ) write(*,'(a)')pr pr=str('complex :',cmplx(huge(0.0),tiny(0.0)) ) write(*,'(a)')pr ! create a format on the fly biggest=huge(0) frmt=str('(*(i',int(log10(real(biggest))),':,1x))',sep=' ') write(*,*)'format=',frmt ! although it will often work, using str(3f) in an I/O statement is not recommended ! because if an error occurs str(3f) will try to write while part of an I/O statement ! which not all compilers can handle and is currently non-standard write(*,*)str('program will now stop') end program demo_msg \end{DoxyVerb} Output H\+U\+G\+E(3f) integers 2147483647 and real 3.\+40282347E+38 and double 1.\+7976931348623157E+308 real \+: 3.\+40282347E+38 0.\+00000000 12345.\+6787 1.\+17549435E-\/38 doubleprecision \+: 1.\+7976931348623157E+308 0.\+0000000000000000 12345.\+678900000001 2.\+2250738585072014E-\/308 complex \+: (3.\+40282347E+38,1.\+17549435E-\/38) format=($\ast$(i9\+:,1x)) program will now stop\hypertarget{namespacem__cli2_autotoc_md121}{}\doxysubsubsection{A\+U\+T\+H\+OR}\label{namespacem__cli2_autotoc_md121} \hypertarget{namespacem__cli2_autotoc_md122}{}\doxysubsubsection{L\+I\+C\+E\+N\+SE}\label{namespacem__cli2_autotoc_md122} Public Domain The documentation for this interface was generated from the following file\+:\begin{DoxyCompactItemize} \item /home/urbanjs/venus/\+V600/github/\+M\+\_\+\+C\+L\+I2/src/\mbox{\hyperlink{M__CLI2_8f90}{M\+\_\+\+C\+L\+I2.\+f90}}\end{DoxyCompactItemize} 0 @article{obolenskaya1997, title={Glutathione status of placentae from differently polluted regions of Ukraine}, author={Obolenskaya, and Tschaikovskaya, TL and Lebedeva, LM and Macewicz, LL and Didenko, LV and Decker, K}, journal={European Journal of Obstetrics \& Gynecology and Reproductive Biology}, volume={71}, number={1}, pages={23--30}, year={1997}, publisher={Elsevier} } mathigatti/orga2-tp Este trabajo nos sirvió para aplicar y desarrollar el conocimiento adquirido en la materia a un tema que nos interesaba. Al finalizarlo pudimos realizar una red neuronal perfectamente funcional en $C$ y $ASSEMBLER$ con una performance superior a la que obtenían expertos en el área hace menos de una década\footnote{\url{http://yann.lecun.com/exdb/mnist/}}, esto habría sido imposible sin cursar Organización del Computador 2. El trabajo desarrollado nos permitió aprender mas de los lenguajes de programación, herramientas de compilación, desarrollo de experimentos y demás temas que vimos en la materia. \\ \\ Al finalizar este trabajo pudimos corroborar exitosamente la hipótesis de que la utilización de SIMD resultaría en una mejora en la performance temporal de nuestro programa en ciertas partes críticas. Viendo en detalle los resultados de nuestros experimentos pudimos ver como para algunos casos las optimizaciones que realiza C fueron suficientes e incluso superiores a las mejoras que realizamos nosotros en $ASSEMBLER$ lo cual nos hizo darnos cuenta que es útil sacar el máximo provecho de las mismas antes de recaer en optimizaciones hechas a mano. De todas maneras como se pudo ver con las funciones $cost\_derivative$, $hadamard\_product$ y $matrix\_prod$, estas obtuvieron resultados considerablemente superiores a sus versiones en $C$, lo cual prueba que bajo ciertas circunstancias tiene sentido y es muy fructífero realizar este tipo de mejoras. \\ \\ Otra conclusión importante fue que a veces paralelizar al máximo genera código más complejo lo cual termina volviendo al programa más lento y difícil de mantener por lo que puede ser incluso mejor trabajar con una concurrencia de menos operaciones a la vez. \\ \\ Cómo conclusión final entendemos que bajo ciertas circunstancias, donde mejoras en tiempo son cruciales, la utilización de SIMD puede ser una herramienta fundamental, aunque al mismo tiempo hay que tener en cuenta que los tiempos de desarrollo suelen ser mayores debido al bajo nivel de $ASSEMBLER$, incluso para funciones simples como las que hicimos y como vimos con $update\_weight$ y $vector\_sum$ para algunos casos incluso pueden encontrarse resultados mediocres.\chapter{ENADE 1998} \section{\color{blue} Quest\~oes} \subsection{\color{blue} Quest\~ao 1} Seja $R$ uma regi\~ao do plano que satisfaz as condi\c c\~oes do Teorema de Green. \begin{enumerate} \item[(a)] Mostre que a \'area de $R$ \'e dada por $\displaystyle \frac1{2} \int_{\partial R} x dy - y dx$ \item[(b)] Use o item (a) para calcular a \'area da elipse de equa\c c\~oes $\begin{cases} x=a\cos(\theta) \\ y=b\sin(\theta)\end{cases}$ onde $a > 0$ e $b > 0$ s\~ao fixos, e $0 \leq \theta \leq 2 \pi$ (valor: 20,0 pontos) \end{enumerate} \paragraph{Dados/Informa\c c\~oes adicionais:} Teorema de Green: Seja $R$ uma regi\~ao do plano com interior n\~ao vazio e cuja fronteira $\partial R$ \'e formada por um n\'umero finito de curvas fechadas, simples, disjuntas e de classe $C^1$ por partes. Sejam $L(x,y)$ e $M(x,y)$ fun\c c\~oes de classe $C^1$ em $R$. Ent\~ao $\displaystyle \int\!\int_R \left(\frac{\partial M}{\partial x}-\frac{\partial L}{\partial y}\right)dx dy=\int_{\partial R} L dx + M dy$ \subsection{\color{blue} Quest\~ao 2} Resolva a equa\c c\~ao diferencial $y'''- 4y'' + 4y' = e^x$, onde $y'=\displaystyle\frac{dy}{dx}$; $y''=\displaystyle\frac{d^2 y}{dx^2}$; $y'''=\displaystyle\frac{d^3 y}{dx^3}$ (valor: 20,0 pontos) \subsection{\color{blue} Quest\~ao 3} Prove que se uma seqü\^encia de fun\c c\~oes $f_n: D \to \mathbb R, D \subset R$ converge uniformemente para $f: D \to \mathbb R$ e cada $f_n$ \'e cont\'\i nua no ponto $a \in D$, ent\~ao $f$ \'e cont\'\i nua no ponto $a$. \paragraph{Dados/Informa\c c\~oes adicionais:} Uma seqü\^encia de fun\c c\~oes $f_n: D \to \mathbb R, D \subset R$ converge uniformemente para $f: D \to \mathbb R$ se para todo $\epsilon > 0$ dado existe $n_0 \in \mathbb N$ tal que $n > n_0 \Longrightarrow |f_n(x) - f(x)| < \epsilon$ para todo $x \in D$. (valor: 20,0 pontos) \subsection{\color{blue} Quest\~ao 4} Seja $\gamma: [0,2\pi] \to \mathbb C$ a curva $\gamma (\theta) = e^{i\theta}$. Calcule $\displaystyle \int_\gamma \frac1{z-z_0} dz$ nos seguintes casos: \begin{enumerate} \item[(a)] $z_0=\displaystyle \frac1{2} (1+i)$ \item[(b)] $z_0 = 2(1 + i)$. (valor: 20,0 pontos) \end{enumerate} \subsection{\color{blue} Quest\~ao 5} Sejam $\alpha$ um n\'umero alg\'ebrico de grau $n$ e $\beta = b_0 + b_1\alpha + ... + b_{n-1}\alpha^{n-1}$ um elemento n\~ao nulo no corpo $\mathbb Q(\alpha)$, i.e., os coeficientes $b_i$ s\~ao racionais, $0 \leq i \leq n-1$, e, pelo menos, um deles \'e diferente de zero. \begin{enumerate} \item[(a)] Prove que $\displaystyle\frac1{\beta}$ \'e um polinômio em $\alpha$. \item[(a)] Racionalize a fra\c c\~ao $\displaystyle \frac1{2+\sqrt[3]{2}}$. (valor: 20,0 pontos) \end{enumerate} \section{\color{red} Solu\c c\~oes} \subsection{\color{red} Quest\~ao 1} \begin{enumerate} \item[(a)] A integral dada no enunciado nos fornece $L(x,y)=-y$ e $M(x,y)=x$. Calculando $\displaystyle \frac{\partial M}{\partial x}-\frac{\partial L}{\partial y}$ obtemos: 2. Como foi dito que a fun\c c\~ao satisfaz as condi\c c\~oes do Teorema de Green, ent\~ao a integral $\displaystyle \frac1{2} \int_{\partial R} x dy - y dx=\displaystyle \frac1{2}\int\!\int_R 2 dx dy=\int\!\int_R dx dy$, que corresponde \`a \'area da regi\~ao $R$. \item[(b)] Temos ent\~ao $\displaystyle \frac1{2} \int_{\partial R} x dy - y dx=\frac1{2} \int_{\partial R} a\cos(\theta) dy-b\sin(\theta) dx$. Mas $dy=b \cos(\theta)d\theta$ e $dx=-a \sin(\theta) d\theta$, ent\~ao a integral se torna: $$\frac1{2} \int_{\partial R} a\cos(\theta) b \cos(\theta)d\theta-b\sin(\theta) (-a \sin(\theta)) d\theta=$$ $$=\frac{ab}{2} \int_{\partial R}\cos^2(\theta)+\sin^2(\theta) d\theta=\frac{ab}{2}\int_0^{2\pi} d\theta=ab\pi$$ \end{enumerate} \subsection{\color{red} Quest\~ao 2} Fazendo a substitui\c c\~ao: $u(x)=y'(x)$ a equa\c c\~ao diferencial assume a forma $u''-4u'+4u=e^x$. A solu\c c\~ao da equa\c c\~ao caracter\'\i stica \'e: $\lambda=2$, portanto a solu\c c\~ao da equa\c c\~ao homog\^enea associada \'e $u(x)=c_1e^{2x}+c_2xe^{2x}$. Pela equa\c c\~ao n\~ao homog\^enea, uma aparente solu\c c\~ao \'e $u(x)=e^x$. De fato: $e^x-4e^x+4e^x=e^x$, portanto pelo princ\'\i pio da sobreposi\c c\~ao uma solu\c c\~ao da equa\c c\~ao diferencial \'e $u(x)=c_0e^x+c_1e^{2x}+c_2xe^{2x}$. Mas $u=y'$, ent\~ao $$y(x)=\int u(x) dx=\int c_0e^x+c_1e^{2x}+c_2xe^{2x}dx$$ Portanto a solu\c c\~ao da eq. diferencial \'e $y(x)=C_0e^x+C_1e^{2x}+ C_2 xe^{2x}+C_3$. \subsection{\color{red} Quest\~ao 3} Como a sequ\^encia de fun\c c\~oes converge para $f$, ent\~ao dado $\epsilon>0$ existe $n_o \in \mathbb N$ tal que para $n>n_0$, $|f_n(x)-f(x)|<\epsilon$. Mas cada $f_n$ \'e cont\'\i nua no ponto $a$, ou seja, para $\delta>0$, $|x-a|<\delta$ implica que $|f_n(x)-f_n(a)|<\epsilon$. Como $f_n(x)$ converge para $f(x)$ ent\~ao $|f(x)-f(a)|<\epsilon$, portanto $f$ \'e cont\'\i nua em $a$. \subsection{\color{red} Quest\~ao 4} A curva em quest\~ao \'e a circunfer\^encia de raio $1$, ent\~ao: \begin{enumerate} \item[(a)] Como $z_0=\frac1{2}(1+i)$ est\'a dentro da curva $\gamma$, pois $|z_0|=\frac{\sqrt 2}{2}<1$, podemos usar o teorema de Cauchy para as integrais complexas, assim: $$\displaystyle \int_\gamma \frac1{z-\frac1{2}(1+i)} dz=2i\pi$$ \item[(b)] Como $z_0=2(1+i)$ est\'a fora da curva $\gamma$, pois $|z_0|=2\sqrt 2 > 1$, o valor da integral \'e zero. \end{enumerate} santteegt/ucl-drl-msc \chapter{Background} \label{sec:chapterlabel2} In this chapter, we will present all the necessary background concepts as well as some related work on recommendation systems and reinforcement learning methods that help to understand the work of this dissertation project. First, we give an overview of the existing recommendation systems techniques and approaches that have been used to solve the recommendation problem. Then, we introduce the basic concepts to define a reinforcement learning task and expose the first attempts made on the field to create reinforcement agents that learn to give recommendation to users. Finally, recent work on reinforcement learning algorithms using deep learning is reviewed. \section{Recommender Systems} In recent years, there has been growing focus on the study of Recommender Systems (RSs), where different techniques have been published in order to to address the problem of information overload on the Internet. Consequently, the evolution of RS and the web usually go hand-in hand. Resnick and Varian \cite{resnick1997recommender} defined RS as systems that help users limit their search by supplying a list of items that might interest them. Such systems has found an active application area for diverse Information Retrieval, Web Mining and Machine Learning techniques that help to solve typical recommendation problems. Additionally, different RS mechanisms have been categorized depending on the way they analyze the data sources to find affinities between users and items. Figure \ref{fig:iomatrix} shows the most general setting in which recommender systems are studied. Having a rating matrix of n users and m items representing the current user preferences, the task of a RS is to predict all missing ratings $r_{a,i}$ for the active user a, and then recommend the item(s) with the highest rate \cite{melville2011recommender}. Nevertheless, the user ratings matrix is typically sparse, as most users do not rate most items. Different approaches to solve this task can be categorized into the following types. \begin{figure}[t] \centering \includegraphics[scale=0.9]{images/uimatrix} \caption[Typical user rating matrix]{Typical user ratings matrix, where each cell $r_{u,i}$ corresponds to the rating of user u for item i.} \label{fig:iomatrix} \end{figure} \subsection{Collaborative Filtering} In Collaborative Filtering (CF) systems, items are recommended by exploiting the similarities amongst several users based on the feedback of previously consumed items. Usually, databases that fit into a CF approach, store the past users interactions in the form of explicit ratings (e.g., 1 to 5 range), or implicit ratings (e.g. a song played by a user, or an item bought by her). In general, CF methods are subdivided into: memory-based and model-based approaches. \begin{itemize} \item \textbf{Memory-based CF:} items are recommended using a weighted combination of ratings in a subset of users that are chosen based on their rating similarity to the target user. The most commonly used measure is the Pearson correlation coefficient \cite{resnick1994grouplens}. %that measures the extent to which there is a linear dependence between the ratings of two users. Alternatively, the similarity between the ratings of two users (represented as a vector in an m-dimensional space) can be and computed using the Cosine similarity or the adjusted Cosine similarity measure which overcomes the issue of users with different rating behavior schemes \cite{yildirim2008random}. Previous studies found that correlation \cite{breese1998empirical} and adjusted cosine similarity \cite{yildirim2008random} performs slightly better. Memory-based CF methods have been extended and improved over the years. Linden, Smith, and York \cite{linden2003amazon} proposed an \textit{Item-based} (or item-item) CF method to overcome the problem of dimensionality found when conventional memory-based algorithms cannot scale well when finding similarities between millions of users and items. This approach, which matches a user's rated items using Pearson correlation, lead to faster online systems and improved recommendations. \item \textbf{Model-based CF:} provides recommendations by using statistical models for predicting user ratings. The most widely used models are Bayesian classifiers, neural networks, fuzzy systems, genetic algorithms, latent features and matrix factorization \cite{bobadilla2013recommender}. For instance, CF methods can be turned into a classification problem, where a classifier is built for each active user representing items as features over users and available ratings as labels. However, latent factor and matrix factorization models have emerged as a state-of-the-art methodology in this class of techniques \cite{koren2009matrix}. \end{itemize} %\subsubsection*{Matrix Factorization} %\label{sec:matrixFactorization} % %To reduce the high levels of sparsity in RS databases, matrix factorization in conjunction with dimensionality reduction techniques have been used. Having a sets U of users, and a set D of items, let $\mathbf{R}$ of size $|U| \times |D|$ be the matrix that contains all the ratings that the users have assigned to the items. Assuming that we would like to discover K latent features, the task is to find two matrices $\mathbf{P}$ (a $|U| \times K$ matrix) and $\mathbf{Q}$ (a $|D| \times K$ matrix) such that their product approximates $\mathbf{R}$: % %\begin{equation} %\label{} %\mathbf{R} \approx \mathbf{P} \times \mathbf{Q^{T}} = \hat{\mathbf{R}} %\end{equation} % %Each row of $\mathbf{P}$ would represent the strength of the associations between a user and the features. Similarly, each row of $\mathbf{Q}$ would represent the strength of the associations between an item and the features. One way to obtain $\mathbf{P}$ and $\mathbf{Q}$ is to first initialize both matrices with some values, and calculate how different their product is to $\mathbf{M}$ by using gradient descent with a regularization term in order to minimize the difference between $\mathbf{R}$ and the predicted preference matrix $\mathbf{\hat{R}}$ (e.g using the Mean Square Error). The update rules and the cost function are defined as follows: % %\begin{equation} %\label{} %p^{'}_{ik}=p_{ik}+\alpha\frac{\partial}{\partial p_{ik}}e^{2}_{ij}=p_{ik}+\alpha(2e_{ij}q_{kj}-\beta p_{ik}) %\end{equation} % %\begin{equation} %\label{} %q^{'}_{kj}=q_{kj}+\alpha\frac{\partial}{\partial q_{kj}}e^{2}_{ij}=q_{kj}+\alpha(2e_{ij}p_{ik}-\beta p_{kj}) %\end{equation} % %\begin{equation} %\label{} %e^2_{ij}=(r_{ij}-\hat{r{ij}})^2=(r_{ij}-\sum_{k=1}^{K}p^{'}_{ik}q^{'}_{kj})^2 %\end{equation} % %Another model-based technique combines Latent Semantic Index (LSI) and the reduction method Singular Value Decomposition (SVD) are typically to solve the same problem[***](Using singular value decomposition approximation for collaborative filtering). Although SVD methods provide good prediction results, it is computationally very expensive, therefore it might be only efficient in static off-line settings. \subsection{Content-based RS} Content-based (CB) RS (compared to pure CF that only utilizes the user rating matrix) tries to make a better personalized recommendation by exploiting the knowledge about a user (e.g. demographic information), or the properties of items (e.g. genre of a movie). Several approaches have treated this problem as an information retrieval (IR) task, where the content associated with the user's preferences is treated as a query, and the unrated items are scored with relevance/similarity to this query \cite{balabanovic1997fab}. Alternatively, CB-RS has also been treated as a classification task using algorithms such as k-Nearest Neighbors (k-NN), decision trees, and neural networks \cite{pazzani1997learning} \subsection{Hybrid approaches} In order to leverage the strengths of both CB-RS and CF methods, several hybrid approaches have been proposed. For instance, Melville et al. presented in \cite{melville2002content} a general framework for content-boosted collaborative filtering, where content-based predictions are applied to convert a sparse user ratings matrix into a full ratings matrix, and then a CF method is used to provide recommendations. In essence, this approach has been shown to perform better than pure CF, pure CB-RS, and a linear combination of the two. \subsection{The cold-start problem} The cold-start problem is a common issue that happens in recommender systems when it is not possible to make reliable recommendations due to an initial lack of ratings \cite{bobadilla2013recommender}. There are three kinds of known cold-start problems: new community, new item and new user. However, the latter represents one of the greatest difficulties faced by an RS in operation. Since new users have not yet provided any rating in the RS, they cannot receive any personalized recommendations when using memory-based CF. Usually, when the users enter their firsts ratings they expect the RS to offer them personalized recommendations, but there are not enough ratings yet to be able to make reliable predictions. As a consequence, new users feel that the RS does not offer the service they expected. This problem is often faced using hybrid approaches such as CF-CB RS, CF-demographic based RS, or CF-social based RS \cite{bobadilla2013recommender}. However, these methods can be combined with clustering techniques over items in order to improve the prediction quality. \subsection{Trends in Recommendation Systems} Latest studies in CF showed that combining conceptual (explicit) and usage (implicit) information can improve the quality of web recommendations. Bobadilla et al. \cite{bobadilla2013recommender} presented a taxonomy for RS which unifies the current recommender methods and algorithms that can be applied to incorporate memory-based, social and content-based information into their CF system depending on the type of information available. The taxonomy depicted in Figure \ref{fig:taxonomy}, also detail at its higher levels the current evaluation methods for RS in terms of quality measures, diversity and novelty. \begin{figure}[t] \centering \includegraphics[scale=0.8]{images/taxonomyrs} \caption[Recommender Systems taxonomy]{Recommender System taxonomy. Source: \textit{Recommender systems survey}\cite{bobadilla2013recommender}} \label{fig:taxonomy} \end{figure} Moreover, Bobadilla et al. argue that one the most widely used algorithm for CF is the k-nearest neighbors (kNN) model. In general, kNN generates recommendations by executing the following tasks: (1) determine k users neighbors; (2) implement an aggregation approach with the ratings for the neighborhood in items not rated by a; and (3) select the top N recommendations from predictions obtained in step 2. On the other hand, a more robust model that combines the advantages of Support Vector Machines with factorization models was introduced by . in \cite{rendle2010factorization}. The Factorization Machine (FM) model is able to compute all interactions between variables using factorized parameters even in scenarios with huge sparsity like recommender systems. Later, . presented in \cite{rendle2012factorization} the \textit{LibFM} software package that implements three learning methods have been proposed for FMs: Stochastic Gradient Descent (SGD) \cite{rendle2010factorization}, Alternating Least-Squares (ALS)\cite{rendle2011fast} and the Markov Chain Monte Carlo (MCMC) \cite{freudenthaler2011bayesian}. Wang et al. \cite{wang2015collaborative} made a generalization of recent advances in Deep Learning \cite{bengio2013representation} from i.i.d. input and applied it to the non-i.i.d. (e.g. CF-based) input by presenting hierarchical Bayesian models that jointly performs deep representation learning for the content information and CF for the ratings matrix. Additionally, Wu Y. et al. in \cite{wu2016collaborative} proposed a Collaborative Deep Auto Encoders (CDAE) model which formulates the top-N recommendation problem using Denoising Auto-Encoders to learn the latent factors from corrupted inputs. Experiments carried out over different domains have shown that the two deep learning approaches can significantly outperform the state of the art. Therefore, recent research has started to focused on defining deep representational approaches that can lead to better recommendations. \section{Reinforcement Learning} Reinforcement learning (RL) \cite{kaelbling1996reinforcement} is a family of machine learning algorithms that optimize sequential decision making processes based on scalar evaluations or rewards. Similarly to an n-armed bandit model \cite{katehakis1987multi}, RL considers the problem as a goal-directed agent interacting with an uncertain environment, where the main objective is to perform actions that maximize the expected sum of future reward for each state in the long term. An important feature that distinguishes RL from other types of learning is that it uses training information that evaluates the actions taken and learns from its own experience, like a trial-and-error search, rather than instructs by giving the correct actions (like supervised learning algorithms). RL uses a formal framework which defines the continuous interaction in terms of states, actions, and rewards, between the \textit{agent} and an \textit{environment} \cite{sutton1998reinforcement}. At each time step $t$, the agent receives some state representation of the environment $s_t \in \mathcal{S}$, where $\mathcal{S}$ is the set of possible states. Then it selects an action $a_t \in \mathcal{A}(s_t)$ , where $\mathcal{A}(s_t)$ is the set of actions available in the observed state. One time step later, the agent receives a numerical reward $r_{t+1} \in \mathcal{R}$, and the environment turns into a new state $s_{t+1}$. Figure \ref{fig:rlenvironment} shows the whole interaction in an agent-environment interface. \begin{figure}[t] \centering \includegraphics[scale=2]{images/rlenvironment} \caption[Agent-Environment interface in a Reinforcement Learning problem]{The Agent-Environment interface in a Reinforcement Learning problem. Source: Reinforcement Learning: a Survey \cite{sutton1998reinforcement} } \label{fig:rlenvironment} \end{figure} This framework is intended to be a simple way of representing four essential features of the artificial intelligence problem: (1) a \textit{policy} (commonly stochastic) defines a mapping from the perceived states of the environment to actions to be taken when in those states; (2) a \textit{reward} function maps each state-action pair to a single number that represents how good or bad the new state is for the environment; (3) a \textit{value} function specifies the total amount of reward an agent can expect to accumulate over the future and starting from a given state. This function is considered the most important as it is used by the agent during decision-making and planning; and optionally (4) a model that mimics the behavior of the environment (transitions and rewards) and provides a way of deciding which action to perform considering possible future situations before they are actually experienced. In general, agents are categorized depending on how the reinforcement learning problem needs to be modeled. \textit{Value-based} and \textit{Policy-based} agents are solely based on the value and policy functions respectively. \textit{Actor-critic} agents use both policy and value functions; \textit{Model-free} agents use policy and/or value function but no model; and \textit{Model-based} agents have all properties mentioned above. On the other hand, The \textit{return} $R_t$ is defined as the sum of the discounted future rewards over an episode of $T$ time steps that an agent actually seeks to maximize: \begin{equation} \label{eq:return} R_t = r_{t+1} + \gamma r_{t+2} + \gamma^2 r_{t+3} + ... = \sum^{\infty}_{k=0} \gamma^k r_{t+k+1} \end{equation} where $\gamma, 0 \leq \gamma \leq 1$ is a discount factor that determines the present value of future rewards \subsection{Markov Decision Processes} A Markov Decision Process (MDP) is a model for sequential stochastic decision problems. As such, it is widely used in applications where an autonomous agent is influencing its surrounding environment through actions \cite{shani2005mdp}. It is defined by a tuple $\langle S, A, R, Pr \rangle$ representing a Markov Chain with values and decisions that follows the Markov property: \textit{a state S is Markov if and only if $\mathbb{P}[S_{t+1}|S_t] = \mathbb{P}[S_{t+1}|S_1,...S_t]$. In other words, in a MDP the future is independent of the past given the present}. Therefore, if a reinforcement learning task satisfies the Markov property and the state and action spaces are finite, then it can be modeled as a finite Markov Decision Process. A particular finite MDP is defined by its state and action sets, a reward function $R$ (equation \ref{eq:expectedRew}) that assigns a real value to each state/action pair, and a state-transition function $Pr$ (equation \ref{eq:transitionsProb}) which provides the probability of transitioning between every pair of states given each action. Altogether, these quantities completely specify the most important aspects of the dynamics of a finite MDP. \begin{equation} \label{eq:expectedRew} \mathcal{R}^a_{ss'} = \mathbb{E} {r_{t+1} | s_t=s, a_t=a, s_{t+1}=s'} \end{equation} \begin{equation} \label{eq:transitionsProb} \mathcal{P}^a_{ss'} = Pr {s_{t+1} = s' | s_t=s, a_t=a} \end{equation} Therefore, the decision-maker's goal is to find an optimal policy $\pi$, such that at each stage of the decision process, the agent needs only to obtain the current state $s$ and execute the action $a=\pi(s)$. Various exact and approximate algorithms have been proposed for estimating the optimal policy $\pi*$, such as policy iteration, value iteration, Monte-Carlo learning, temporal difference, Q-learning, SARSA, etc \cite{kaelbling1996reinforcement}\cite{sutton1998reinforcement}. \subsection{A model-free Reinforcement Learning Setup} \label{sec:model-free} A reinforcement learning setup consists of an agent interacting with an environment $E$ at discrete time steps. At each time step $t$, the agent receives an observation $s_t$, takes an action $a_t$ and receives a reward $r_t$. Additionally, the environment $E$ may be stochastic so it can be modeled as an MDP with a state space $\mathcal{S}$, action space $\mathcal{A}$, an initial state distribution $\rho(s_1)$, transition dynamics $\rho(s_{t+1}|s_t, a_t)$, and reward function $r(s_t, a_t)$. On the other hand, the agent's behavior is defined by a policy $\pi$, which maps states to a probability distribution over the actions $\pi : \mathcal{S} \rightarrow P(\mathcal{A})$. Finally, the return from a state is defined as the sum of the discounted future reward $R_t = \sum^T_{i=t} \gamma^{(i?t)}r(s_i, a_i)$. As the return depends on the actions chosen and therefore on $\pi$, it may be stochastic. The goal in reinforcement learning is to learn a policy which maximizes the expected return from a start distribution $J = \mathbb{E}_{r_i,s_i \sim E,a_i \sim \pi}[R_1]$. The action-value function is used in many RL algorithms and describes the expected return after taking an action $a_t$ in state $s_t$ and thereafter following policy $\pi$: \begin{equation} \label{eq:action-valueFcn} Q^{\pi}(s_t, a_t) = \mathbb{E}_{r_{i \geq t}, s_{i > t} \sim E, a_{i > t} \sim \pi} [R_t| s_t, a_t] \end{equation} Many approaches in RL opt to represent the action-value function as a recursive relationship using the Bellman equation (eq. \ref{eq:bellmanEq}). Nevertheless, If the target policy is deterministic we can describe it as a function $\mu: \mathcal{S} \leftarrow \mathcal{A}$ and avoid the inner expectation as shown in equation \ref{eq_Qdeterministic}. As the expectation depends only on the environment, it is possible to learn $Q^{\mu}$ off-policy, using transitions which are generated from a different stochastic behavior policy $\mu$. \begin{equation} \label{eq:bellmanEq} Q^{\pi}(s_t, a_t) = \mathbb{E}_{r_{t}, s_{t+1} \sim E} [r(s_t, a_t) + \gamma \mathbb{E}_{a_{t+1} \sim \pi} [Q^{\pi}(s_{t+1}, a_{t+1})] ] \end{equation} \begin{equation} \label{eq_Qdeterministic} Q^{\mu}(s_t, a_t) = \mathbb{E}_{r_{t}, s_{t+1} \sim E} [r(s_t, a_t) + \gamma Q^{\mu}(s_{t+1}, \mu_{t+1}) ] \end{equation} Q-learning \cite{watkins1992q}, is a commonly used off-policy algorithm which uses a \textit{greedy} policy $\mu(s) = argmax_aQ(s, a)$ in order to estimate the action that gives the maximum reward. As the algorithm considers function approximators parameterized by $\theta^Q$, it can be used as an optimizer that minimize the loss: \begin{equation} \label{} L(\theta^Q) = \mathbb{E}_{s_{t} \sim \rho^{\beta}, a_t \sim \beta, r_t \sim E} [(Q(s_t, a_t | \theta_Q) - y_t)^2] \end{equation} where $y_t = \mathbb{E}_{s'\sim \mathcal{E}} [r(s_t, a_t) + \gamma max_{a'} Q(s_{t+1}, \mu(s_{t+1}) | \theta^Q)]$. %While $y_t$ is dependent on $\theta^Q$, this is typically ignored. \subsection{Exploration/Exploitation Dilemma} One of the challenges that arise in reinforcement learning is the trade-off between exploration and exploitation \cite{kaelbling1996reinforcement}\cite{sutton1998reinforcement}. To obtain a lot of reward, a reinforcement learning agent must prefer actions that has already tried in the past and found to be effective in terms of reward. In order to discover such patterns, it needs to try actions that it has not been selected before. so the agent has to be able to exploit actions that are already known as effective, but it also needs to explore in order to make better action selections in the future. Na\"{i}ve approaches use a greedy policy $\pi(s) = argmax_aQ(s, a)$ to get the action with maximum reward as a pure exploitative model. However, an $\epsilon$-greedy policy can be considered instead to allow exploration in the environment and improve the estimation of the non-greedy action values. Therefore, the agent will select a random action with a probability $\epsilon$, and will act greedily with probability $1 - \epsilon$ . The advantage of $\epsilon$-greedy policy over greedy policy is that the former continuously explore and improve the chances of recognizing possible actions that can lead to better rewards. Nevertheless, this has still been an issue to solve in some reinforcement learning tasks, so research has continuously focused on trying different techniques to solve the exploration/exploitation trade-off that can lead to the discovery of better policies \cite{kaelbling1996reinforcement}. \section{Recommender Systems and Reinforcement Learning} Although reinforcement learning seems to have great potential for improving the recommendation problem, it has received relatively little attention and found only limited application. The first experiments on using reinforcement learning techniques were focused on developing recommendation systems for web content using implicit data from server files which stored the navigational logs of the users throughout their contents. Ten Hagen et al. \cite{ten2003exploration} considered RL approach with an exploration/exploitation trade-off for discovering unknown areas and automatically improve their recommendation policy to helping users navigate through an adaptive web site. They stated that a model-free algorithm like Q-Learning can become infeasible if the number of actions are relatively high and showed that a greedy policy can indeed lead to a suboptimal recommendation model as the use of such kind of policy function, where the \"best\" actions (e.g with higher probability) are taken, does not always improve the policy. They finally demonstrated that this problem is solved by starting the algorithm with a lot of exploration and gradually reduce the parameter $\epsilon$ as the policy is getting closer to an optimum. Results concluded that a recommender system without exploration potentially can get trapped into a local maxima. Rojanavasu et al. in \cite{rojanavasu2005new} presented a general framework for web recommendation that learns directly from customer's past behavior by applying an RL process based on the SARSA method and a $\epsilon$-greedy policy. The system was composed of two models: a global model to keep track of customers trends as a whole, and a local model to record the user's individual browsing history. In general, the model treated pages as states of the system and links within a page as actions. To predict the next state, it used a ranking system that is also separated into 2 parts. i) a global ranking system using the data from a $Q_{global}$-matrix obtained from the action-value function and an $\epsilon$-greedy policy that gave the chance for rank new items that have few clicks but may match a user's interest; and ii) a local ranking $Q_{local}$ using an inverse $\epsilon$-greedy policy. The system then finds the final total reward using $Q_{total}=Q_{local} + wQ_{global}$ where $w \in (0-1]$ is a weight hyper parameter of the model. Overall, the system provided customers the chance to explore other products than the ones they already visited. Experimental results showed that a value of $\epsilon=0.2$ preserves the balance between exploration and exploitation, meanwhile If $\epsilon < 0.2$ the system will gave small chances to users to explore new items, or otherwise may include products that do not match to the customer's interest ($\epsilon > 0.2$). On the other hand, results also prove that even if the purpose of $Q_{local}$ was to discover new products, it appeared to be less effective than $Q_{global}$. Shani et al. in \cite{shani2005mdp} argued that it is more appropriate to formulate the problem of generating recommendations as a sequential optimization problem so they described a novel approach for a commercial web site based on MDPs together with a predictive model. The MDP-based recommender system took into account the expected utility and the long-time effect of a particular recommendation. Then it suggested items whose immediate reward is lower, but leads to more \textit{profitable} rewards in the future. However, the benefits of this model are offset by the fact that the model parameters are unknown, randomly initialized and that they take considerable time to converge. So, they defined a strong initial model-based for collaborative filtering, which solves quickly, and that does not consume too much memory. Before implementing the recommender, they initialize a predictive model of user behavior using data extracted from the web site, and then used it to provide the initial parameters for the MDP. They proposed to use maximum-likelihood estimation with three major enhancements: skipping, clustering, mixture modeling, so the predictive model can be thought of as a first-order Markov chain (MC) of user dynamics in which states correspond to sequences of \textit{k} events (in this case: previous selections) representing relevant information about the users interaction. Then, the transition function described the probability that a user ,whose \textit{k} recent selections were $x_1, . . . ,x_k$ will select the item $x'$ next. Authors experimented with small values of \textit{k} ($k \in [1-5]$) in order to overcome data sparsity and MDP solution complexity issues. Moreover, enhancements on the maximum-likelihood n-gram formulation showed to be useful to find a good estimate of the transition function for the initial predictive model without suffering the problem of data sparsity and bad performance presented on other model variations. %The proposed maximum-likelihood estimation include three major enhancements. First, a form of skipping based on the observation that the occurrence of the sequence $x_1,x_2,_x3$ lends some likelihood to the sequence $x_1,x_3$. In other words, after initializing the counts of each state transition using the observed data, for any given user sequence $x_1,x_2, ...,x_n$, they added a fractional count $1/2^{( j-(i+3))}$ to the transition from $s=\langle x_i,x_{i+1},x_{i+2}\rangle$ to $s'=\langle x_{i+1},x_{i+2},x_j\rangle$, for all $i+3 < j ?n$, which acts as a diminishing probability of skipping a large number of transactions in the sequence. Equation \ref{eq:trMCSkipping} shows the updated formulation of the (normalized )maximum-likelihood estimation, where $count(s,s')$ is the fractional count associated with the transition from s to s'. % %\begin{equation} %\label{eq:trMCSkipping} %tr_{MC}^{base}(s,s')=\frac{count(s, s')}{\sum_{s'}count(s, s')} %\end{equation} % %The second enhancement exploits the similarity of sequences in a form of clustering. The idea is that the likelihood of transition from $s$ to $s'$ can be predicted by occurrences from $t$ to $s'$, where $s$ and $t$ are similar. Equation \ref{eq:simTr} define the similarity of states $s_i$ and $s_j$ where $\delta(\dot, \dot)$ is the Kronecker delta function and $s^m_i$ is the m-th item in state $s_i$. Then, the similarity count from state $s$ to $s'$ was defined using equation \ref{eq:simCountTr}. Finally, the new transition probability from $s$ to $s'$ given by equation \ref{eq:newTr} yielded the best results during evaluation. % %\begin{equation} %\label{eq:simTr} %sim(s_i, s_j)=\sum_{m=1}^{k}\delta(s_i^m, s_j^m) \cdot (m+1) %\end{equation} %\begin{equation} %\label{eq:simCountTr} %simcount(s, s')=\sum_{s_i} sim(s, s_i) \cdot tr^{base}_{MC}(s, s') %\end{equation} %\begin{equation} %\label{eq:newTr} %tr_{MC}(s, s')=\frac{1}{2}tr^{base}_{MC}(s, s') + \frac{1}{2}\frac{simcount(s, s')}{\sum_{s''}simcount(s, s'')} %\end{equation} % %Due to larger values of k lead to states that are more informative whereas smaller values of k lead to states with more statistical meaning, they added as a third enhancement a finite mixture modeling (e.g. combine a trigram, a bigram and a unigram into a single model) in a way to balance these conflicting properties by mixing k models, where the \textit{i-th} model looks at the last i transactions. In addition, they applied mixture weights during experiments as $\pi_1 = · · · = \pi_k = 1/k$ so the generation of the models for smaller values entails little computational overhead. %An evaluation on the accuracy of the predictive model was performed using real user transactions and browsing paths (from web logs) from the commercial store web site. Besides, they compared different variations of their enhancements on the MC approach with other models like i)both sequential and non-sequential form of the \textit{Microsoft Commerce Server 2000}(Heckerman et al. (2000))[***] (local distributions are probabilistic decision trees); and ii)unordered versions of MC models. Results outlined that the MC models using skipping, clustering, and mixture modeling yielded better results than any other other variation of model \textit{i}. Sequence-sensitive models outperformed the accuracy of non-sequential models, while MC models were superior to the predictor models. and the skipping enhancement was only beneficial for the transactions data set On the other hand, %the MDP-based recommender system, models the recommendation process as a sequence of states and attempts to optimize it. Here, the predictive model played an important role in the construction of the model as it provides the probability, denoted by $Pr_{pred}(x|x_1, . . . ,x_k)$, that a user purchase a particular item x given her sequence of past purchases $x_1, . . . ,x_k$. the components of the reinforcement learning model were defined as follows: a) the states are the \textit{k}-tuples of items purchased; b) actions correspond to a recommendation of an item; c) rewards depends only on the last item defining the current state, where its net profit was used; and d) the transition function as the stochastic element of the model that estimates the user's actual choice.% In brief, equation \ref{eq:trMDP} represents the transition function or the probability that a user will select item $x''$ given that item $x'$ is recommended in state $\langle x_1,x_2,x_3 \rangle$. %\begin{equation} %\label{eq:trMDP} %tr^1_{MDP}(\langle x_1,x_2,x_3 \rangle,x',\langle x_2,x_3,x'' \rangle) %\end{equation} In order to solve the MDP problem, a policy iteration algorithm was used as the defined state space presented certain characteristics that lead to fast convergence. %Indeed, tests performed on real data showed that the policy iteration converges after a few iterations. %: i) the inherent directionality that transitions showed are useful to reduce the running time of the solution algorithm; ii) the computation of an optimal policy is not sensitive to variations in the number of \textit{k} past transactions a state represent; iii) ignoring unobserved states by maintaining transition probabilities only for states where a transition actually occurred; and iv) using the independence of recommendations or Markov property, where the probability that a user buys a particular item depends only on her current state, allowing the algorithm to handle large action spaces. The transition function was carefully initialized in order to be fairly accurate when the system was first deployed to avoid the cold-start problem,causing states that are infrequently observed to be updated faster than already observed states. Moreover, when the first transition to a state is observed, its probability is initialized to 0.9. In this way, the system balances the need to explore unobserved items in order to improve its model by recommending non-optimal items occasionally until getting their actual counts. Finally, to update the model, the system used an off-line approach which keeps track of the recommendations and the user selections and build a new model at fixed time intervals (e.g. once a week). %So, to re-estimate the transition function at time $t+1$ the following counts are obtained: % %\begin{equation} %\label{eq:countIn} %c^{t+1}_{in}(s, r, s \cdot r) = c^{t}_{in}(s, r, s \cdot r) + count(s, r, s \cdot r) %\end{equation} %\begin{equation} %\label{eq:countOut} %c^{t+1}_{out}(s, r, s \cdot r) = c^{t}_{out}(s, r, s \cdot r) + count(s, s \cdot r) - count(s, r, s \cdot r) %\end{equation} %\begin{equation} %\label{eq:countTotal} %c^{t+1}_{total}(s, s \cdot r) = c^{t}_{total}(s, r, s \cdot r) + count(s, s \cdot r) %\end{equation} %\begin{equation} %\label{eq:modelProb1} %tr(s, r \in R, s \cdot r) = \frac{c^{t+1}_{in}(s, r, s \cdot r)}{c^{t+1}_{total}(s, s \cdot r)} %\end{equation} %\begin{equation} %\label{eq:modelProb2} %tr(s, r \notin R, s \cdot r) = \frac{c^{t+1}_{out}(s, r, s \cdot r)}{c^{t+1}_{total}(s, s \cdot r)} %\end{equation} %\begin{equation} %\label{eq:initIn} %c^{0}_{in}(s, r, s \cdot r) = \xi_S \cdot tr(s, r, s \cdot r) %\end{equation} %\begin{equation} %\label{eq:initOut} %c^{0}_{out}(s, r, s \cdot r) = \xi_S \cdot tr(s, r, s \cdot r) %\end{equation} %\begin{equation} %\label{eq:initTotal} %c^{0}_{total}(s, s \cdot r) = \xi_S %\end{equation} %where Equation \ref{eq:countIn} corresponds to the number of times a recommendation r was accepted in state s, equation \ref{eq:countOut} is the number of times a user select item r in state s even though it was not recommended, equation \ref{eq:countTotal} is the number of times a user pick item r while being in state s, regardless of whether it was recommended or not, and finally, equations \ref{eq:modelProb1} and \ref{eq:modelProb2} represent the actual probability that a user will select item r given state s when r was recommended or not respectively. %The transition function had to be carefully initialized in order to be fairly accurate when the system was first deployed to avoid the cold-start problem, so the at time $t=0$ were calculated using equations \ref{eq:initIn} \ref{eq:initOut} \ref{eq:initTotal}, where $\xi_s= 10 \cdot count(s)$, causing states that are infrequently observed to be updated faster than already observed states. Moreover, when the first transition to a state $s \cdot r$ is observed, its probability is initialized to 0.9 the probability of the most likely next item in state s with $\xi_s =10$. In this way, the system balances the need to explore unobserved items in order to improve its model by recommending non-optimal items occasionally until getting their actual counts. Experiments demonstrated that a deployed system using three mixture components (combined using an equal weight), with history length $k \in [1,3]$ was able to generate 28\% higher average user reward compared to pure MC model, while in terms of computational costs, the MDP-based model was built quickly and provided fastest recommendations at the price of more memory use. All in all, the off-line predictive model approach provided an adequate initial performance that overcomes the cold-start problem, however, authors avoid using some form of reinforcement learning technique as they argued that at that time, its implementation requires many calls and computations by a recommender system online, which lead to slower responses and undesirable results for the web site owner. A machine learning perspective, introduced by Taghipour et al. in \cite{taghipour2007usage}, used an off-line reinforcement learning model to solve the recommendation problem using the Q-Learning algorithm while employing concepts and techniques commonly applied in the web usage mining domain. They argued that this method is appropriate for the nature of web page recommendation problem as it provides a mechanism which is constantly learning, does not need periodic updates, can be easily adapted to changes in the website structure and new trends in users behavior. The RL problem formulation was considered as as a competition between different recommender systems to gather more points, %than a 2-player game, as in the recommender system environment, self-play, a typical technique used in training RL systems[***], cannot be used to train the system so the system needed of actual web usage data for training. together with a stochastic model with the following properties: \begin{itemize} \item \textit{states:} showing the history of pages visited by the user so far. For this case, a notion of N-Grams was adopted and a sliding window of size $w$ was set to limit the page visit sequences to a constant number and avoid large state spaces. \item \textit{actions:} consisting of a single page recommendation at each state. \item \textit{policy:} rewards actions positively if it recommends a page that will be visited in one of the consequent states \item \textit{reward} function: defined as $r(s, a) += reward(Dist(Rs', p), Time(p^v_w))$, where $Dist(Rs', p)$ is the distance of page p from the end of the recommended pages list to state $s'$, and $Time(p^v_w)$ indicates the time user has spent on the last page of the state. Finally, the $reward(Dist, Time)$ function was defined as a linear combination of both values: $reward(Dist, Time) = \alpha \times dist + \beta \times Time$ with $\alpha + \beta = 1$. \end{itemize} %As the model did not have a predetermined reward function $R(s, a)$ or a transition function $\delta(s, a)$ to give the next state, the reward was estimated by considering that each state s is formed by two sequences $V_s =\langle p^V_{s,1}, p^V_{s,2},..., p^V_{s,w} \rangle$, and $R_s =\langle p^R_{s,1}, p^R_{s,2},..., p^R_{s, n} \rangle$, indicating the sequence of visited and previously recommended pages respectively, where $p^V_{s, i}$ , indicates the ith visited page in the state and $p^R_{s, i}$ indicates the ith recommended page in the state s, Additionally, they had two considerations in the implementation of the reward function: i) only the occurrence of the last page visited in the recommended pages list in state $s'$ is used to reward the action performed in the previous sate $s$, and ii) the time the user spends on a page, assuming that the more time a user spends on a page the more interested. So, the reward function was defined as follows: %\begin{enumerate} %\item Assume $\delta(s, a) = s'$ %\item $P_R = V_{s', w} \cap R_{s'}$ %\item if $p \neq \emptyset$ %\item For page $p$ in $P_R$ %\begin{enumerate} %\item $r(s, a) += reward(Dist(Rs', p), Time(p^v_w))$ %\end{enumerate} %\end{enumerate} %where $Dist(R_i, p)$ is the distance of page p from the end of the recommended pages list and $Time(p^v_w)$ indicates the time user has spent on the last page of the state. Finally, The $reward(Dist, Time)$ function was defined as a linear combination of both values as $reward(Dist, Time) = \alpha \times dist + \beta \times Time$ with $\alpha + \beta = 1$. % %Subsequently, the definition of the learning algorithm also took into consideration the following characteristics: 1) the Q-Learning algorithm given by equation \ref{eq:QLearningalg} was proposed as an update rule and the structure to estimate how successful a prediction can be. Here, the decreasing value of $\alpha_n$ caused these values to gradually converge and decreases the impact of changing reward values as the training continues; 2) an $\epsilon$-greedy action selection was picked as it is important to add some exploration of the state space especially at the beginning of the training; 3) the algorithm followed a TD(0) off-policy learning[***] procedure, as the maximum Q value of the next state is considered when estimating the future reward; and 4) the computation of $r(s, a)$ suffered a slightly change by considering value of $Q(s',a)$ with a coefficient $\gamma$ in order to propagate the value of performing a specific action beyond the limits imposed by the $w$. As the authors mentioned in their work: "the basic idea is that when an action/recommendation is appropriate in state Si, indicating the recommended page is likely to occur in the following states, it should also be considered appropriate in state $S_{i-1}$ and the actions in that state that frequently lead to $S_i$". During the training phase, the described algorithm converged after a few thousand (between 3000 and 5000) visits of each episode or user session. % %\begin{equation} %\label{eq:QLearningalg} %Q_n(s, a )= [(1 - \alpha_n)Q_{n-1}(s, a) + \alpha_n[r(s, a) + \gamma \max_{a'} Q_{n-1}(\delta(s, a), a')] %\end{equation} %with %\begin{equation} %%\label{} %\alpha_n = \frac{1}{1 + visits_n(s, a)} %\end{equation} % %For the purpose of performing a set of experimental evaluations of the proposed model, simulated log files generated by a web traffic simulator were used to train and tune the rewarding functions. The metric used for each evaluation were Recommendation Accuracy, Coverage (similar to precision and recall metrics) and Shortcut Gain (measures how many page-visits users can save if they follow the recommendations). First of all, an experiment with different values of $w$ showed that fixed window size of 3 as recommendation history resulted on better accuracy and shortcut gain, so this value was set for the rest of system evaluations. Next, they experimented the impact of gradually increasing (in steps of 5\%) the coefficient $\alpha$ of parameter \textit{Dist} in the reward function, resulting on both higher accuracy and higher shortcut gain for values up to 15\%. This matched the natural consequence of adding a bounded size of window on recommendations history. Finally, a set of experiments also tested the system performance by increasing the coefficient $\gamma$ in the reward function, obtaining an increase of the accuracy until reaching an upper bound (around 0.20\%) where it began to drop, while the shortcut gain increased steadily up to a point where recommendations became so inaccurate. The proposed off-line RL method used a simplified formulation of the recommendation problem but it obtained much better results compared to association rules and item-based CF methods, displaying a lower rate in which its accuracy decreased. Authors concluded that the algorithm is a good candidate for solving the recommendation problem as it does not rely on any previous assumptions regarding the probability distribution of visiting a page after having visited a sequence of pages, and that the nature of the problem matches perfectly with the notion of delayed reward (known as temporal difference). Additionally, they suggested that in order to produce better results it could be possible to use a more complicated formulation of the reward function such as a neural networks, rather than a linear combination of factors. Later in \cite{taghipour2008hybrid} they exploited the previous RL framework and present a hybrid web recommendation method enriched with semantic knowledge about the usage behavior and thus obtain a more generalized solution regarding to the usage data it has. The new system used the incremental Document Conceptual Clustering \cite{godoy2006modeling} algorithm to map pages to higher level concepts, and exploit the hierarchical and conceptual document clustering to provide a semantic relationship between them in combination with the usage data in a user session. Therefore, new states consist of a sequence of concepts visited by the user, while actions are recommendation of pages that belong to a specific concept. This definition resulted in a much smaller state-action space as the size of the state space is now dependent on the number of distinct page clusters. The reward function now takes into account the content similarity of the recommended and visited pages along with the usage-based reward defined in the previous approach. While, the modified algorithm does not make predictions based on weak usage patterns as the new states represent a generalized view of many single visit sequences. Evaluation results finally showed the flexibility of the new RL approach to incorporate different sources of information in order to improve the quality of recommendations. % %Now, an action $a$ recommending a concept $c$ is rewarded if the user visits a page belonging to concept $c$ later in his browsing session. The new reward function shown in equation \ref{eq:hybridreward} takes into account the content similarity of the recommended and visited pages, where CBR represents the content-based reward of an action (which is equal to the similarity score between concepts) defined in equation \ref{}, and UBR is the usage-based reward defined in the previous approach. % %\begin{equation} %\label{eq:hybridreward} %UBR(Dist(R_{s'} , r),Time(P_{t+1})) \times CBR(r, C(P_{t+1})) %\end{equation} %\begin{equation} %\label{} %l(c) = -log \mathcal{p}(c) %\end{equation} %\begin{equation} %\label{} %Sim(c_1, c_2) = max_{a \in LCA} {l(a)} %\end{equation} % %\textbf{Improving adaptation of ubiquitous recommender systems by using reinforcement learning and collaborative filtering} proposed a learning system for Context-based Recommender System (CBRS)[***] by modeling an MDP agent which combines a hybrid Q-learning algorithm (HQL), collaborative filtering and case-based reasoning techniques in order to define a \textit{contextual} recommendation process based on different context dimensions (cognitive, social, temporal, geographic). It addressed the following problems that came out in recommender systems: a)avoid the intervention of experts as the use of the Q-learning algorithm does not need initial user?s information; b)reduce the cold start problem thanks to the ability of the Q-learning algorithm to explore the knowledge of other users in the same context by using CF; c)accelerate the learning process by mixing Q-learning with case-based reasoning techniques (reuse of cases yields to faster user satisfaction); and d)an exploration strategy allows recommendations to adapt to the user's interest evolution. % %As a result, the model was based in three main algorithms. First, a hybrid-based CF approach which combines the advantages of the memory-based (fill the missing rating values of the user-item matrix) and model-based CF (form the nearest neighbors of each item). Second, the Case based reasoning (CBR)[***] algorithm was picked as it uses knowledge of previous cases to solve new problems, by finding a similar past case and thus reusing it to solve the current situation. Finally, the Q-learning algorithm was improved by: i)reusing past cases information gathered from the CBR, and ii)giving the ability to use information from other users sharing the same interests, by extending the $\epsilon$-greedy strategy to select a random action based on the similarity of user profiles (obtained from the CF algorithm). % %The global mechanism of a context-based recommender system (CBRS) which uses the HyQL algorithm was composed mainly by the following modules: % %\textbf{sensing module:} detects time, location, cognitive and social dimensions of the context. % %\textbf{thinking module:} composed by an abstraction phase that is based on inference rules defined on temporal and space ontologies, and an aggregation phase the two dimensions. % %\textbf{reasoning module:} chooses an action based on the HyQL algorithm % %Finally, evaluations comparing the Q-learning and HyQL with respect to solving the cold start problem were performed, and consisted on testing the precision of the first 100 trials of the system starting when the user is connected to the system. In General, results showed that the precision of HyQL was greater than the precision of Q-learning, demonstrating that this approach is a good candidate to be used in the recommender system problem. However, as the system implementation depends on the context data, it needs of a hand-crafted feature creation process in order to be used on different situations. %\textbf{Exploration in Interactive Personalized Music Recommendation: A Reinforcement Learning Approach} formulated an interactive and personalized music recommendation approach as a reinforcement learning task based on the multi-armed bandit problem, where songs were treated as arms and user ratings as payoffs, and the objective function was set to maximize the cumulative reward of a targeted user over the long term. The model considered the exploration-exploitation trade-off and a Bayesian model in order to deal with both audio content and the novelty of recommendations and thus learn the user musical preferences online based on their feedback. % %The model differs from other approaches in three aspects: i)it is based on audio content; ii)the model is highly efficient as it allows easy online updates; and iii)the model is evaluated based on online real-life user interaction data and does not need a test dataset. Moreover, each recommendation serves two objectives: (1) satisfy the user?s current musical taste, and (2) obtain user feedback needed to improve future recommendations. Even though authors mentioned that an MDP formulation can model a broader range of problems than the multi-armed bandit, it requires much more data to train and is often more computationally expensive. % %Their choice to work with a the content-based approach rather CF was justified for the following reasons. First, due to the bayesian nature of their formulation, methods for matrix factorization (MF) are much more complicated than the proposed linear model. Second, MF often needs of large amount of data for training. Third, existing Bayesian MF methods are inefficient for online updating, so they do not fit in a bandit model which is updated once a new rating is obtained, Fourth, a content-based method does not suffer of the new song problem. Finally, content-based approaches capture causality within music content rather than pure correlation that CF methods offer. % %Subsequently, the interactive, personalized recommender system used a reinforcement learning task called the multi-armed bandit[***] which addressed both the exploration-exploitation trade-off and playlist generation with a single unified model. Whereas some RL models considered only a greedy strategy that does not actively seek fort user feedback, resulting in suboptimal recommendations over the long term, this model takes into account the uncertainty of both the mean and the variance of the rating distribution, allowing the recommender to explore user preferences actively rather than merely exploiting the available rating information. An approach better than the typical $\epsilon$-greedy method to solve of the multi-armed bandit problem, is the use of the Upper Confidence Bound (UCB)[***] algorithm, but it requires an explicit form of the confidence bound that is difficult to derive in the case of music recommendation. % %The personalized music rating model was focused on audio content and novelty. The former considers the overall preference of a song and is represented by the linear function $U_c = \mathbf{\theta'x}$ where $\mathbf{x}$ is the feature vector of the audio content and $\mathbf{\theta}$ is the user preference in different music features (it was considered constant over time). On the other hand, the novelty factor was defined after examining the song's repetition distribution of 1000 users? listening histories collected from the Last.fm dataset. Results showed that most of the songs a user listens to are repeats and that the frequency distribution approximately follows the Zipf?s law[***], where only a small set of songs are repeated most of the time. As a consequence, it was assumed that the novelty of a song decays immediately after it is listened to and then gradually recovers according to the function $U_n = 1 - e^{-t/s}$, where s is a recovery speed (unknown) parameter learned from user interactions and $e^{-t/s}$ is the well-established forgetting curve[***] that measures user's memory retention of a song. In other words, novel songs are those of which that at a certain time a user has little or no memory. It is important to emphasize that the definition of this factor could be different or not applicable to other recommendation contexts than the musical environment. % %Overall, the rating model, defined in equation \ref{eq:combinedRatingModel}, assumed that a rating is formed as the combination of the user?s preference of the song?s content and the dynamically changing novelty. Hence, each user was represented by a set of parameters $\Omega = \{\theta, s\}$, where $\Omega$ needs to be estimated from historical data, so uncertainty had to be taken into account. % %\begin{equation} %\label{eq:combinedRatingModel} %U = U_cU_n = \mathbf{\theta'x}(1 - e^{-t/s}) %\end{equation} % %With this in mind, the new problem formulation was solved using the Bayesian Upper Confidence Bound (Bayes-UCB)[***] algorithm. In The Bayes-UCB, the true expected payoff $U_i$ defined in equation \ref{eq:expectedRatingUCB} for arm \textit{i} is treated as a random variable and the posterior distribution $p(U_i|\mathcal{D})$ of $U_i$ given the history of payoffs $\mathcal{D_l} = \{(\mathbf{x_i}, t_i, r_i)\}^{l}_{i=1}$ is predicted using equations\ref{eq:posteriorDistrOmega} and \ref{eq:posteriorDistrU}. Finally, the Bayes-UCB recommends song $k^*$ that maximizes the quantile function $argmax_{k=1...|S|} Q(\alpha, P(U_k|D_l))$ where Q satisfies$\mathcal{P}[Uk \leq Q(\alpha, P(U_k|D_l))] = \alpha$ and $\alpha = 1 - \frac{1}{l+1}$ % %\begin{equation} %\label{eq:expectedRatingUCB} %\mathcal{E}[R_i] = U_i = \mathbf{\theta'x_i}(1 - e^{-t_i/s}) %\end{equation} %\begin{equation} %\label{eq:posteriorDistrOmega} %p(\mathbf{\Omega}|D_l) \propto p(\mathbf{\Omega}) p(\mathbf{\Omega}|D_l) %\end{equation} %\begin{equation} %\label{eq:posteriorDistrU} %p(U_k|D_l) = \int p(U_k|\mathbf{\Omega}) p(\mathbf{\Omega}|D_l) \mathbf{d\Omega} %\end{equation} % %Since equation \ref{eq:posteriorDistrOmega} does not have a closed-form solution, a Markov Chain Monte Carlo (MCMC)[***] should be used as an approximate inference algorithm. However, authors argued that its performance is very low and users could wait for up to a minute until the MC converges. Hence, they developed a Bayesian model using a piecewise-linear function approximation, as well as a variational inference algorithm to approximate posterior distribution of $\Omega$. For more details about the model definition, please refer to section 4.2 in [***](this paper). Authors mentioned that even if the model just considered audio content and novelty of music on its definition, other factors like diversity, mood or genre could be also approximated by linear functions and then added to it. All in all, the defined approximate Bayesian model was used to obtain the fixed-level quantile of $p(U_i|\mathcal{D})$ and thus estimate the expected payoff and confidence bound of the conceptual Bayes-UCB. % %Evaluations of both efficiency and effectiveness were carried out over six recommendation algorithms and models (Random, LinUCB-C (Content-based), LinUCB-CN (Content and novelty based), Bayes-UCB-CN, Bayes-UCB-CN-V (Variational Inference), and Greedy-CN) demonstrating that the proposed approaches were accurate and highly efficient. The effectiveness study used the \textit{regret}[***] metric to compare the algorithms, showing that the Bayes-UCB-based algorithms performed better than Greedy-CN due to the balance of exploration and exploitation it offers, whereas the good performance of the Bayes-UCB- CN-V indicated that the piecewise-linear approximation and variational inference were appropriately defined. Moreover, results in the efficiency study empirically proved that the developed variational inference algorithm was 100 times faster than the MCMC. Additionally, a user study and an overall evaluation of recommendation performance dropped good conclusions. Results showed that the bandit approach with the novelty factor addressed the cold-start problem improving the recommendation performance. % %To sum up, this work was considered by the authors as the first to balance exploration and exploitation based on reinforcement learning and the multi-armed bandit, and thus improve recommendation performance by addressing the cold-start problem in music recommendation without relying on additional (contextual information). An approximation to the rating model and the new probabilistic inference algorithms helped to achieve real-time recommendation performance that could be generalized to other recommenders and/or media types. Finally, authors suggested that this work could be extended to model the correlations between different users to further reduce the amount of exploration by using hierarchical Bayesian models. On the other hand, if people prefer to boost the performance of CF, they suggested to exploit the exploration/exploitation trade-off idea by using the latent features learned during the matrix factorization rather than the audio features and keep other parts of the proposed system unchanged. % %\textbf{ENHANCING COLLABORATIVE FILTERING MUSIC RECOMMENDATION BY BALANCING EXPLORATION AND EXPLOITATION} extended the work presented above by introducing exploration into the CF context. The approach used a Bayesian graphical model that takes into account the CF latent factors and novelty of recommendation, as well as a Bayesian inference algorithm to efficiently estimate the posterior rating distributions. % %Authors argued that their previous work, based on a content-based approach, had experienced the following drawbacks: (i) the personalized user rating model suffer of a semantic meaning between low-level audio features and high-level user preferences; (ii) it is difficult to determine which acoustic features are actually effective in the music recommendation scenario; and (iii) recommendation of songs under this approach lacks of variety due to most of them are acoustically similar. Therefore, they presented a memory-based CF approach using matrix factorization in order to improve recommendations. % %Recalling the definition of Matrix factorization in Section \ref{sec:matrixFactorization}, this method characterizes users and songs by vectors of latent factors $\mathbf{u_i}$ and $\mathbf{v_j}$ respectively with $i \in [1, m], j \in [1, n]$. In order to learn the latent feature vectors, the system used equation \ref{eq:matrixFactorization} and Alternating Least Squares (ALS)[***] to minimize the regularized cost function on the training set: % %\begin{equation} %\label{eq:matrixFactorization} %\sum_{(i,j) \in I}(r_{ij}-\mathbf{u^T_i}\mathbf{v_j})^2 + \lambda (\sum_{i=1}^{m}n_{ui}\| \mathbf{u_i} \|^2 + \sum_{j=1}^{n}n_{vi}\| \mathbf{v_j} \|^2) %\end{equation} %where I is the index set of all known ratings, $\lambda$ a regularization parameter, $n_{ui}$ the number of ratings by user i, and $n_{vj}$ the number of ratings of song j. However, the traditional CF approach often fails to take into consideration novelty and works greedily. % %As a result, a reinforcement learning approach for CF-based music recommendation based on the n-arm bandit problem was proposed. The user rating model considered that song?s rating is affected by two factors: \textit{CF score}, (how much a user likes the song in terms of each CF latent factor) denoted by $U_{CF}$ and defined inn terms of matrix factorization, and \textit{novelty score} (the dynamically changing novelty of the song) denoted by $U_N$. Thus, the final user rating model, given in equation \ref{eq:UratingModel}, can be defined as a combination of the two scores, where vector $\theta$ indicates the user?s preferences for different CF latent factors, $\mathbf{v}$ is the song feature vector learned by the ALS CF algorithm, t is the time elapsed since when the song was last heard, s the relative strength of the user?s memory, and $e^{?t/s}$ the well-known forgetting curve. % %\begin{equation} %\label{eq:UratingModel} %U=U_{CF}U_{N}=(\mathbf{\theta^Tv})(1-e^{-t/s}) %\end{equation} % %In the same way as in [***](previous paper), each user was associated with a pair of parameters $\ohm=(\mathbf{\theta},s)$ to be learned from the user's rating history, and their underlying ratings were considered random rather than fixed numbers (estimated using $\mathcal{E}[R_j] = U_j$). Exploration was also introduced by using both a similar Bayesian Upper Confidence Bound (UCB) sampling algorithm and Bayesian Graphical model to estimate the posterior distribution $p(U_j|\mathcal{D})$ of $U_j$ given the target user?s rating history $\mathcal{D}$, but in terms of the CF latent factors rather than audio content. Then, the song with the highest fixed-level \textit{quantile} value (equation \ref{}) of $p(U_j|\mathcal{D})$ will be recommended to the target user. % %=BEGIN=ALREADY COMMENTED %The graphical model used in [***12] and depicted in Figure \ref{}*** was adopted to estimate the posterior distribution of $U$, and which is defined as follows: % %\begin{equation} %\label{} %R|\mathbf{v},t,\mathbf{\theta},s,\sigma^2 \sim \mathcal{N}(mathbf{\theta}^T\mathbf{v}(1-e^{-t/s}), \sigma^2) \\ %\end{equation} % %\begin{equation} %\label{} %\mathbf{\theta}|\sigma^2 \sim \mathcal{N}(\mathbf{0}, a_0\sigma^2\mathbf{I}) %\end{equation} % %\begin{equation} %\label{} %s \sim \mathit{Gamma}(b_0, c_0) %\end{equation} % %\begin{equation} %\label{} %\tau=1/\sigma^2 \sim \mathit{Gamma}(d_0, c_0) %\end{equation} % %where $\mathbf{I}$ is the $\mathit{f} \times \mathit{f}$ identity matrix, $\mathcal{N}$ represents the Gaussian distribution[***] and $\textit{Gamma}$ %represents the Gamma distribution[***] with parameters shape and rate. $\mathbf{\theta}$, s, and $\tau$ are parameters, while $a_0$, $b_0$, $c_0$, $d_0$, and $e_0$ are hyperparameters of the priors % %During iteration $h+1$, the model has delivered h recommendations to the user's history $\mathcal{D}_h = \{(\mathbf{v_i},t_i,r_i)\}^h_{i=1}$ which posterior distribution can be defined using Bayes theorem as $p(\Omega|\mathcal{D}_h) \propto p(\Omega)p(\mathcal{D}_h|\Omega)$. They used their own implementation of the Gibbs sampling algorithm to speed-up convergence compared the Monte Carlo Markov Chain (MCMC)[***] algorithm to sample the user's parameters $\Omega=(\mathcal{\theta}, s)$ by sampling from a conditional distribution. Then, they used Eq. \ref{eq:expUserRatingSongj} to obtain a sample for $U_j$, and finally, the posterior Probability Density Function PDF $p(U_j,\mathcal{D})$ was approximated by the histogram of the samples of $U_j$. % %After estimating the posterior PDF for rating each song, they used a Bayes-UCB approach to recommend song $\mathit{j}^*$ that maximizes the quantile function: % %\begin{equation} %\label{} %j^*=\arg\max_{j=1,...,|S|} \mathcal{Q}(\alpha, p(U_j,D_h)) %\end{equation} % %where $\alpha=1-\frac{1}{h+1}$ and the quantile function $\mathcal{Q}$ returns a value \textit{x} such that $Pr(U_j \leq x|D_h)=\alpha$ %=END=ALREADY COMMENTED % %However, the efficiency of the convergence in the Bayesian inference of this approach was improved by developing a specific Gibbs sampling algorithm[***] as in this case it is simple to sample the ratings from the conditional distribution of $\mathbf{\theta}$, while, a Metropolis-Hastings (MH) algorithm[***] was used to draw samples of $s$. For a more detail explanation of the algorithms, please refer to sections 3.2 and 3.3 in [***](this paper). % %Efficiency experiments were carried out over to compare the developed sampling implementation and an MCMC algorithm developed in JAGS\footnote{\url{http://mcmc-jags.sourceforge.net/}}. The results showed that their proposed Gibbs sampling algorithm is hundreds of times faster than MCMC evidencing the suitability of the algorithm for online recommender systems. Additionally, an online user study which compared the effectiveness of the proposed Bayes-UCB-CF algorithm against the traditional greedy method and the previous implementation Bayes-UCB-Content[***](previous paper), proved that the cumulative average rating of new algorithm significantly outperformed the baselines. All in all, the Bayes-UCB-CF algorithm achieved a better balanced exploration/exploitation trade-off and significantly showed an improvement on its recommendation performance. % %In conclusion, this first attempt to remedy the greedy nature of CF approaches could enhance the performance of CF-based music recommendation significantly. Moreover, the reinforcement learning model seems to be applicable to other contexts different from the music environment, as well as it could be deployed together with the content-based RL model from [***](previous paper) in order to build a hybrid framework which combines the strengths of both approaches. % %=BEGIN=ALREADY COMMENTED %\textbf{Generating Music Playlists with Hierarchical Clustering and Q-Learning} %describes a system that uses reinforcement learning over hierarchically-clustered sets of songs to learn a user?s listening preferences for a Automatic Playlist Generation (APG), % %recommendation algorithms focus on discovery, whereas APG tends to be more concerned with coherence within a set of ordered tracks % %The approach we have taken is to use CB methods to inform an unsupervised machine learning algorithm (Q-learning), which over time can learn fromimplicit user feedback to generate playlists which are personalised to different listeners % %Features extracted from the audio are also used as part of this process %solution that does not rely on tagged audio files from any data source or any external services to achieve a completely personal and independent music player. % %The playermonitors the user?s actions continuously using reinforcement learn- %ing and updates its matrices based on user behaviour. Using implicit user be- haviouronly, ourplayeris able to learn user preferences providing users with a better music listening experience % % %the aim is to create playlists from individual songs and learn the proba- bilities of transitioning between them ($n^2$ transitions for a library of n songs) %the solution is to first cluster the songs, and then learn transitions between these clusters % %The solution we propose is therefore to use hierarchical clusters, in which %atree of clustersisconstructed using k-means clustering, and Q-learning (ex- plained below) is performed on nodes at multiple levels. This keeps the benefits of clustering without introducing large transition matrices or large groups of %songs to choose between. In fact, this reduces the space complexity from O(n2) to just O(n). % %**learning from user behavior % %This is known as active learning %The MDP was defined as follows: set of states S corresponds to the set of clusters, and not the individual songs %we will need one model per node in the tree, with S being only the node?s child clusters % %the model-free Q-learning was used to compute the expected utility of taking a particular action a %In the music player, the agent does actually know %S, since this is just a deterministic function in which a1 means ?transition to s1?, a2??transition to s2?, and so on. % %However, R is clearly unknown because the rewards are assigned by the user. % %Having the transition matrix P % %\begin{equation} %\label{} %P^'_{r, c} = P_{r, c} + \alpha_t \times [R_t + \gamma_a P_{c, a} - P_{r, c}] %\end{equation} % %$\apha$ and R are functions which may depend on other parameters including the current song choice % %Q-learning algorithms often iterate within ?episodes? until local convergence %is reached. This method doesn?t apply well to the music player scenario, so instead there is just one update per user action. This reflects that rather than having a single clear goal for which an optimum path must be found, we are continually trying to find good states and may continue indefinitely. So there are no absorbing states in the MDP, and its transition graph may contain loops. % %Reward calculation scheme %takes values between -1,1 based on a linear interpolation of the time the user listened the song. %The basic measure of implicit reward is listening time, %This leads us to r = ?1.0 as the greatest negative reward for a skip after 0 seconds, and r =1.0 as the greatest positive reward for not skipping at all %So a track finishing is just a special case of this general rule, and we interpolate %linearly between the two reward extremes based on how far through the track we got to before the skip %whenever a track is skipped quickly, the previous song should be treated as the current track when future rewards are calculated. % %Also, the small rewards %should only be applied for smaller playlists since % % %**conclusions %perform well in a small user study, greatly reducing the relative number of songs that a user skips. %=END=ALREADY COMMENTED A model-based RL agent for music playlist recommendation proposed by Liebman et al. \cite{liebman2015dj} modeled the preferences of both songs and song transitions as MDPs, and demonstrated to be potentially effective in the domain and particularly good to solve the cold-start problem as it is able to generate personalized song sequences within a single listening session and with no prior knowledge of the new user's preferences. The adaptive playlist generation problem was defined as an episodic MDP $\langle S, A, P, R, T \rangle$ with the following components: \begin{itemize} \item a state space S composed by the ordered sequence of songs played, $S = \{(a_1, a_2, ..., a_i)|1 \leq i \leq k; \forall j \leq i, a_j \in \mathcal{M}\}$ \item an actions space A formed by all possible candidates for being the next song to play, $a_k \in A$ which means $A = \mathcal{M}$ \item a deterministic transition function $P(s, a) = s'$ which indicates the existing transitions from state s to s' by taking action a \item an utility function R(s, a) derived from hearing song a when in state s \item $T = \{(a_1, a_2, ..., m_k)\}$: the set of playlists of length k. \end{itemize} In order to find an optimal policy $\pi*$ (and thus obtain the most pleasing sequence of songs to the listener), a model-based approach was chosen arguing that even though model-free approaches learn the value of taking an action a from state s directly, it requires a lot of data to converge, and it is considerably scarce in the domain music recommendation (as well as in other kind of applications). On the other hand, model-based formulations often requires a lot of computation to find an approximate solution to the MDP, but the trade-off of computational expense for data efficiency makes the model-based approach a good option for this problem. Since P is deterministic, only the listener's reward function R was required to be modeled. Therefore, the reward function defined as $R(s, a) = R_s(a)+R_t(s, a)$ represented the sum of two distinct components: (1) the listener's preference over songs, $R_s : A \rightarrow \mathcal{R}$, and (2) the preference over transitions from songs played to a new song, $R_t : S \times A \rightarrow \mathcal{R}$. % Hence, R was defined using equation \ref{eq:Rdecomposed}. %\begin{equation} %\label{eq:Rdecomposed} %R(s, a) = R_s(a)+R_t(s, a) %\end{equation} In essence, the model represents each song as a compacted vector of spectral auditory descriptors that capture meaningful differences in user's preferences, so the system is able to leverage knowledge using only a few transition examples to plan a future sequence of songs. %However, the framework is in principle robust and agnostic to the choice of a specific song corpus. %First, the listener reward (linear) function over songs $R_s$ generates a binary feature vector using a sparse encoding of the song descriptors as $R_s(a) = \phi_s(u) \cdot \theta_s(a)$, where $\phi_s(u)$ represents the listener pleasure for a particular set of active features. Then, the listener reward (linear) function over transitions $R_t$ obtains a sparse binary feature vector $R_t(a_i, a_j) = \phi_t(u) \cdot \theta_t(a_i, a_j)$, where $\phi_t(u)$ is a user-dependent weight vector and $\theta_t$ is a binary feature vector, with both representing transitions between 10-percentile bins of the song descriptors they share (for learnability purposes). An evaluation on the expressiveness of feature representation showed that different sequences are indeed distinguishable. As a result, the compact representation of songs was still rich enough to capture meaningful differences in user's preferences, and the system was able to leverage knowledge using a few transition examples to plan a future sequence of songs. On the other hand, the agent architecture was composed by a module for learning the listener parameters %($\phi_s$ and $\phi_t$) which performs the during initialization and learning on the fly processes; and an additional module for planning a sequence of songs in the playlist. % and thus estimate the next appropriate song to play. The initialization is divided in two parts: (1) initialization of song preferences polls the listener for her $k_s$ favorite songs in the database and updates the user's preference vector %value $\phi_s(u)$ using the feature descriptors of each of the selected items; and 2) initialization of the transition preferences by presenting different possible transitions that encapsulate the variety in the dataset, and directly asking which of a possible set of options the listener would prefer. Updates are carried out in the same manner as the initialization of song preferences. After initialization, the agent begins to play songs for the listener, waiting for her feedback (reward), and updating the user's preferences accordingly. This learning of the fly procedure is perceived as a temporal-difference update with an attenuating learning rate that balances the trust between the previous history of observations and the newly obtained signal. %After initialization, the agent begins playing songs for the listener, waiting for her feedback, and updating $\phi_s$ and $\phi_t$ accordingly. The latter is computed by using a single unified reward signal that considers the relative contributions of the song and transition rewards to set weights (equations \ref{eq:w_s} and \ref{eq:w_t}) for credit assignment (equations \ref{eq:phis} and \ref{eq:phit}). In general, the learning of the fly procedure is perceived as a temporal-difference update with an attenuating learning rate that balances the trust between the previous history of observations and the newly obtained signal. % %\begin{equation} %\label{eq:w_s} %w_s = \frac{R_s(a_i)}{R_s(a_i) + R_t{a_{i-1}, a_i}} %\end{equation} %\begin{equation} %\label{eq:w_t} %w_t = \frac{R_t(a_{i-1}, a_i)}{R_s(a_i) + R_t(a_{i-1}, a_i)} %\end{equation} %\begin{equation} %\label{eq:phis} %\phi_s = \frac{i}{i+1} \cdot \phi_s + \frac{i}{i+1} \cdot \theta_s \cdot w_s \cdot r_{incr} %\end{equation} %\begin{equation} %\label{eq:phit} %\phi_t = \frac{i}{i+1} \cdot \phi_t + \frac{i}{i+1} \cdot \theta_t \cdot w_t \cdot r_{incr} %\end{equation} %After determining the MDP reward function, Then, a Monte Carlo Tree Search (MCTS) heuristic \cite{chaslot2010monte} is used for planning. The iterative process chooses a subset of 50\% of the songs in the database with highest $R_s$ score, and then at each point, it simulates a trajectory of future songs selected at random and calculate the expected payoff of the song trajectory using $R_s$ and $R_t$. The process ends when it finds the trajectory which yields to the highest expected payoff, and finally, the first item in the trajectory is selected to be the next song to recommend. Additionally, to mitigate the problem of high complexity during re-planning under large song spaces, the agent used the canonical K-means algorithm for clustering songs according to song types and reduce the search complexity drastically. An evaluation of the cumulative reward distribution of the proposed approach showed that overall, the DJ-MC agent outperforms two baselines models (a random agent and a greedy agent), while in terms of transition reward preferences, the new system presented a small but significant boost in performance compared to a model based only on reasoning about song preferences. All in all, this approach demonstrated to be a good improvement as a reinforcement learning model for music recommendation, as it enables learning from relatively few examples, and provides quality on recommendation sequences. %\textbf{Hybrid Collaborative Filtering with Neural Networks} \section{Deep Reinforcement Learning} So far, the RL agents presented above have shown to achieve some success under the recommendation domain, nevertheless, their performance has been conditioned to the quality of the hand-crafted feature engineering made, as well as the custom definition of value functions or policy representations. Moreover, to use RL successfully in situations approaching to real-world complexity, agents usually need to derive efficient representations of the environment from high-dimensional sensory inputs, and use these to generalise past experiences to new situations. Recent advances in deep learning \cite{bengio2013representation} have made it possible, but even if it seems that this approach could fit in a RL task, its applicability to the domain may present several challenges. For instance, successful deep learning applications have required large amounts of hand-labelled training data to obtain good predictions results. Additionally, most deep learning algorithms assume the data samples to be i.i.d., while in RL an agent typically encounters with non-i.i.d. sequences of highly correlated states. On the other hand, agents must be able to learn from a scalar reward signal that is frequently sparse, noisy and delayed, while the RL data distribution changes as the algorithm learns new behaviour. However, thanks to the increasing success of adapting different deep learning techniques to different domains, Deep Reinforcement Learning (DRL) models have started to be proposed, getting promising results. Mnih et al. presented in \cite{mnih2013playing}\cite{mnih2015human} the first deep learning model which uses an end-to-end reinforcement learning approach to learn control policies directly from a high-dimensional raw input space. The model, applied to a range of Atari 2600 games, consists of a convolutional neural network and an experience replay mechanism \cite{adam2012experience} to alleviate the problems of correlated data and non-stationary distributions in typical RL problems. Moreover, the model's architecture uses only the state representation as input, and a separate output unit for each possible action (target network), corresponding to the predicted Q-values of each individual action that can be applied to the input state. This configuration allows the computation of all possible actions reward in a given state with only a single forward pass. Finally, the resulting Deep Q neural network (DQN) is trained with a variant of Q-learning which learns from raw pixels input and the underlying environment properties, and outputs a value function estimating the future rewards. On the other hand, the RL formulation models the environment $\mathcal{E}$ as a MDP composed by a state space $\mathcal{S}$, a discrete action space $\mathcal{A}=\{1...K\}$, a scalar reward function $r(s_t, a_t)$ and a transition dynamics function $\rho(s_{t+1}|s_t, a_t)$. Then, given the agent's behaviour defined by a policy $\pi$, they estimate the action-value function by using an approximator function of the Bellman equation (presented in eq. \ref{eq:bellmanEq}) as an iterative update. A Q-network function approximator with weights $\theta$ is then trained by minimising the sequence of loss functions $L_i(\theta_i) = \mathbb{E}_{s, a \sim \rho (\cdot)}[(y_i - Q(s, a; \theta_i))^2]$ where $\rho(s, a)$ is the behaviour probability distribution over sequences of states s and actions a. %\begin{equation} %\label{eq:QnetLoss} %L_i(\theta_i) = \mathbb{E}_{s, a \sim \rho (\cdot)}[(y_i - Q(s, a; \theta_i))^2] %\end{equation} %where $y_i = \mathbb{E}_{s'\sim \mathcal{E}} [r + \gamma max_{a'} Q(s', a'; \theta_{i-1})|s, a]$ is the target for iteration i and $\rho(s, a)$ is a probability distribution over sequences of states s and actions a known as behavior. Finally, to compute the full expectations in the gradient of the loss function, they considered stochastic gradient mini-batch updates of the weight parameters, and a learning algorithm which is: a)\textit{model-free}: replaces the expectations by directly using uniform samples from the behaviour distribution $\rho$, and b)\textit{off-policy}: follows an $\epsilon$-greedy strategy of the behaviour distribution that ensures an adequate exploration of the state space. Evaluation results showed that this method is able to learn how the value function evolves for a reasonably complex sequence of events. Furthermore, the proposed DRL algorithm demonstrated to perform better compared to the SARSA and Contingency \cite{bellemare2012investigating} methods as they need to incorporate significant (handcrafted) prior knowledge about the visual problem to get considerable performance, in comparison to the raw inputs that DRL use. Also, DRL achieved better performance than an expert human player in 29 out of 49 games. However, this model is only able to handle discrete low-dimensional action spaces and cannot be directly applied to tasks under the continuous domain as the iterative process of finding the action that maximises the action-value function becomes intractable. Additionally, a na\"{i}ve discretization of the action space would no be a good solution as it may throw away information about the structure of the action domain. Lillicrap et al. \cite{lillicrap2015continuous} adapted the ideas presented above to the continuous action space and defined an actor-critic, model-free approach, based on the Deterministic Policy Gradient \cite{Silver2014} (DPG) algorithm, that is able to find policies end-to-end and whose performance is competitive compared to approaches based on a planning algorithm with access to the dynamics of the domain. While the model's architecture uses similar features of DQN (network architecture, experience replay memory, and target outputs), along with batch normalization \cite{ioffe2015batch} to overcome the internal covariate shift problem in the network input layers during mini-batches, the Deep Deterministic Policy Gradient (DDPG) algorithm maintains two functions: (1)an actor function $\mu(s|\theta_{\mu})$ that obtains the current policy by deterministically mapping states to specific actions; and 2)the critic function $Q(s,a)$ learned by applying the Bellman equation. During the learning process, DDPG updates the actor by applying the policy gradient to the expected return from a start distribution $J = \mathbb{E}_{r_i, si \sim E, a_i \sim \pi}[R_1]$ and with respect to the actor parameters. Even if convergence is no longer guaranteed due to the added non-linearity of the proposed approximation function, it produces a good generalization over large state spaces. %\begin{equation} %\label{} %\begin{aligned} %\nabla_{\theta^{\mu}}J & \approx \mathbb{E}_{s_t \sim \rho^{\beta}} [\nabla_{\theta^{\mu}} Q(s, a | \theta_Q)|_{s=s_t, a=\mu(s_t|\theta_{\mu})} ] \\ %& = \mathbb{E}_{s_t \sim \rho^{\beta}} [\nabla_a Q(s, a | \theta_Q)|_{s=s_t, a=\mu(s_t)} \nabla_{\theta^{\mu}} \mu(s|\theta^{\mu})|_{s=s_t}] %\end{aligned} %\end{equation} The DDPG algorithm also implements certain improvements to the RL task. First, it introduces stability in the learning process by applying soft target updated to the target networks instead of directly copying the weights. This is performed by exponentially moving their average values using $\theta' \leftarrow \tau\theta + (1 - \tau)\theta'$ with $\tau \ll 1$. Second, batch normalization allows DDPG to be able to generalize across environments using the same hyper parameters for learning. Finally, the exploration-explotaition trade-off is managed independently from the learning algorithm by adding noise sampled from a noise process $\mathcal{N}$ to the actor policy $\mu$. % %A second improvement solved the difficulty of finding hyper-parameters that allow generalization across environments when the low dimensional feature vector observations have different physical units and the ranges. This issue is addressed by adapting a batch normalization technique proposed by [***], that minimizes the covariance shift during training by normalizing each dimension across samples in a mini-batch to have unit mean and variance. As a result, DDPG ensures that each layer receives whitened input in favor of learning effectively across many different tasks with differing types of units. % %The third improvement tries to overcome the common challenge on how to manage exploration in reinforcement learning tasks with continuous actions spaces. The exploration-explotaition trade-off was managed independently from the learning algorithm by adding noise sampled from a noise process $\mathcal{N}$ to the actor policy $\mu$ and thus create an exploration policy $\mu'(s_t) = \mu(s_t|\theta^{\mu}_t) + \mathcal{N}$. Hence, any sampling process $\mathcal{N}$ can be plugged in to suit exploration on specific environments. To evaluate the algorithm performance, authors used simulated physical environments with different levels of difficulty, and the Ornstein-Uhlenbeck \cite{bibbona2008ornstein} sampling process for adding exploration to the learning policy. Results demonstrated that even when harder tasks obtain poor Q estimates, DDPG is able to learn good policies across a variety of domains with continuous action spaces and using both low-dimensional feature vector and high-dimensional pixel inputs. Additionally, all the experiments were solved using fewer steps of experience than the DQN algorithm, showing that DDPG may be able to solve even more difficult problems. However, the DDPG algorithm requires of a large number of training episodes to find solutions and only works under environments with a not so big number of actions. Therefore, a more robust model-free approach is needed in order to tackle these limitation. Later, al. \cite{Dulac-Arnold2015} presented a new policy architecture, under the same actor-critic framework used above, that not only allows reinforcement learning methods to be applied to large-scale learning problems, and to operate efficiently with a large number of actions, but also can generalize over the action set in logarithmic time. The full policy is then trained and optimized using DDPG. As a result, they obtained a more efficient algorithm that makes both learning and acting tractable in time and allows value-based policies to use the action features to reason about previously unseen actions. The so-called \textit{Wolpertinger} architecture leverages prior information about the actions and embed them in a continuous space upon which the actor can generalize using a smooth function approximator. After the policy produces a continuous action within this space, it uses an approximate nearest neighbour search to find the set of closest discrete actions in logarithmic time to finally select the action with the highest reward. Consequently, the algorithm avoids the heavy cost of evaluating all actions mainly by defining an efficient action-generating actor, and then using the critic to refine the actor's choices for the full policy. % %The action generation function $f_{\theta^{\pi}}(s) = \hat{\mathbf{a}}$ provides a proto-action in $\mathbb{R}^n$ for a given state s. $\hat{\mathbf{a}}$ is then mapped to the discrete action set $\mathcal{A}$ by applying the k-nearest-neighbor\footnote{The size of the generated action set k is task specific, and allows for an explicit trade-off between policy quality and speed} function from equation \ref{eq:knnDDPG}, which returns the k closest actions in $\mathcal{A}$ by computing the $L_2$ distance between them. Even if $g_k$ has the same complexity as the \textit{argmax} in the Q action-value function, each step of evaluation of the $L_2$ distance is faster than a full value-function evaluation as the lookup is performed in logarithmic time. %\begin{equation} %\label{eq:knnDDPG} %g_k(\hat{\mathbf{a}}) = \arg\min^k_{a \in \mathcal{A}} | \mathbf{a} - \mathbf{\hat{\mathbf{a}}} |_2 %\end{equation} % %Following the selection of the of k closest actions to $\hat{\mathbf{a}}$, the choice of the actual action to be applied to the environment is set according to the highest-scoring action obtained according $Q_{\theta^Q}$ defined in equation \ref{eq:qDDPG}. Consequently, the full Wolpertinger policy $\pi_{\theta}$ with parameter $\theta$, representing both the parameters of the action generation element $\theta_{\pi}$ and of the critic $\theta_Q$, makes the algorithm significantly more robust to imperfections in the choice of action representation. % %\begin{equation} %\label{eq:qDDPG} %\pi_{\theta}(s) = \arg\max_{a \in g_k \circ f_{\theta^{\pi}}(s) } Q_{\theta^Q}(s, a) %\end{equation} Having the definition of the full policy as $\pi_{\theta^{\pi}} = g \circ f_{\theta^{\pi}}(s)$, where $f_{\theta^{\pi}}(s) = \hat{\mathbf{a}}$ is the proto-action generation function and $g$ the actual action function that return the set of k closest actions to $\hat{\mathbf{a}}$, the final goal of this approach is to perform policy iteration by alternatively performing policy evaluation on the current policy with Q-learning, and then improving upon the current policy by following the DDPG over $f_{\theta^{\pi}}$, considering that the effects of $g$ are a deterministic aspect of the environment. The Wolpertinger agent, evaluated on three environment classes with large actions spaces (including a simulated recommender system), showed that it can scale to real-world MDPs with large number of actions as only a subset of the full set of actions was sufficient in many tasks to converge and provide significant speedups. Nevertheless, there is still an existing issue of exploration when the agent needs to learn from scratch.ruivalmeida/novathesis %!TEX root = ../../template.tex \section{Hypothesis}% \label{sec:intro_hypothesis} As stated in Section~\ref{sec:research_questions}, the main goal of my work was to provide an answer to the question of how to design a miniaturized tomographic atmosphere monitoring system, based on \gls{DOAS}. This sentence is the most important point of the project. Every development from this point forward stems from it and is motivated by it. It also requires some deconstruction in order to understand the true scope of the matter. First, it is a miniaturized device. This means that besides all the habitual requirements (performance, function, adequacy, compatibility with the other components, safety and security) in defining components in an engineering project, one must also keep in mind the footprint of each component, and how much it weighs. Second, it is a tomographic system, which means that not only the device must be able to take line integrals from some kind of medium, it must also be able to describe a predefined trajectory, with admissible levels of geometric error, which complies to a certain projection geometry. Otherwise, one would not be able to apply a tomographic reconstruction routine to obtain the map of the target species concentrations. Finally the system is supposed to monitor the atmosphere in some way. Now, as implied in the same sentence, \gls{DOAS} is the technique that I am trying to apply in this system. But as I point out in Section~\ref{sub:doas} and with more depth in Section~\ref{sec:doas}, there are two families of \gls{DOAS}, and within them, many sub-techniques. This system which I am developing is based on the hypothesis that we can use an almost hybrid approach to \gls{DOAS}: passive \gls{DOAS} instrumental simplicity and active \gls{DOAS} retrieval simplicity. One could use a scattered sunlight measurement as a light source for a \gls{DOAS} analysis, provided distances between the two points are kept small, optical densities are low (clear atmosphere), and both spectral measurements are taken in the same angle. With this process, we would only effectively be using as projections the spectral measurements of the \gls{ROI}. This hypothesis, while not being mentioned in the library directly, has already been hinted at in several papers, namely references~\cite{Frins2006, Casaballe2017, Johansson2009}. \hypertarget{class_sebastian_bergmann_1_1_version}{}\section{Version Class Reference} \label{class_sebastian_bergmann_1_1_version}\index{Version@{Version}} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_sebastian_bergmann_1_1_version_a66f960c649f472a51d65548b542ca018}{\+\_\+\+\_\+construct}} (\$release, \$path) \item \mbox{\hyperlink{class_sebastian_bergmann_1_1_version_afa8e7a3a646144eab50188b7a805a389}{get\+Version}} () \end{DoxyCompactItemize} \subsection{Detailed Description} \begin{DoxySince}{Since} Class available since Release 1.\+0.\+0 \end{DoxySince} \subsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{class_sebastian_bergmann_1_1_version_a66f960c649f472a51d65548b542ca018}\label{class_sebastian_bergmann_1_1_version_a66f960c649f472a51d65548b542ca018}} \index{Sebastian\+Bergmann\+::\+Version@{Sebastian\+Bergmann\+::\+Version}!\+\_\+\+\_\+construct@{\+\_\+\+\_\+construct}} \index{\+\_\+\+\_\+construct@{\+\_\+\+\_\+construct}!Sebastian\+Bergmann\+::\+Version@{Sebastian\+Bergmann\+::\+Version}} \subsubsection{\texorpdfstring{\+\_\+\+\_\+construct()}{\_\_construct()}} {\footnotesize\ttfamily \+\_\+\+\_\+construct (\begin{DoxyParamCaption}\item[{}]{\$release, }\item[{}]{\$path }\end{DoxyParamCaption})} \begin{DoxyParams}[1]{Parameters} string & {\em \$release} & \\ \hline string & {\em \$path} & \\ \hline \end{DoxyParams} \subsection{Member Function Documentation} \mbox{\Hypertarget{class_sebastian_bergmann_1_1_version_afa8e7a3a646144eab50188b7a805a389}\label{class_sebastian_bergmann_1_1_version_afa8e7a3a646144eab50188b7a805a389}} \index{Sebastian\+Bergmann\+::\+Version@{Sebastian\+Bergmann\+::\+Version}!get\+Version@{get\+Version}} \index{get\+Version@{get\+Version}!Sebastian\+Bergmann\+::\+Version@{Sebastian\+Bergmann\+::\+Version}} \subsubsection{\texorpdfstring{get\+Version()}{getVersion()}} {\footnotesize\ttfamily get\+Version (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \begin{DoxyReturn}{Returns} string \end{DoxyReturn} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item /\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/sebastian/version/src/\mbox{\hyperlink{sebastian_2version_2src_2_version_8php}{Version.\+php}}\end{DoxyCompactItemize} \subsubsection{Chaves Fim de curso} % SLIDE DE CHAVES FIM DE CURSO \begin{frame} \frametitle{Chaves fim de curso} Limitação de campo de movimento de eixos, como os presentes na mesa cartesiana. Tem a capacidade de mudança de estado de conexão em circuitos, alternando o estado de aberto para fechado e vice-versa. \begin{figure} \centering \includegraphics[scale = 0.15]{figuras/chavefimdecurso} \end{figure} \end{frame} \appendix \section{License} \input{TroubleShooting} % -*-LaTeX-*- % The log below is the old log from when RCS was being used in % conjunction with this. % % $Log: master.tex,v $ % Revision 1.7 2010/03/16 00:32:56 stiber % Final revisions prior to Spring 2010. % % Revision 1.6 2007/12/25 18:01:46 stiber % Final changes for winter 2008. % % Revision 1.5 2007/03/20 23:49:49 stiber % Final cleanup for start of spring 2007. % % Revision 1.4 2006/05/13 17:08:46 stiber % Updated version number for second revision, Spring 2006. % % Revision 1.3 2006/03/27 23:32:04 stiber % Modified for Spring 2006, with corrections and incorporation of % newer LaTeX packages. % % Revision 1.2 2004/03/29 19:46:04 stiber % Updated for Spring 2004 and new textbook (DSP First). % % Revision 1.1 2004/02/19 00:19:10 stiber % Initial revision % \documentclass[12pt]{book} \newif\ifHtml \Htmlfalse \newif\ifPDF \PDFtrue \ifHtml \PDFfalse \usepackage[tex4ht,colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pdftitle={Signal Computing: Digital Signals in the Software Domain}]{hyperref} \hyperlinkfileprefix{} \else \ifPDF \usepackage[pdftex,pdftitle={Signal Computing: Digital Signals in the Software Domain},colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue,bookmarks,bookmarksopen]{hyperref} \usepackage{fullpage} \else \usepackage{hyperref} \usepackage{mathptmx,helvet,courier,fullpage} \fi \fi \usepackage{fullpage,graphicx,algorithmic,multirow,rotating} %\usepackage{html} \usepackage[chapter]{algorithm} % To get footnote text for figures on the same page \usepackage{afterpage} % For overlaying a picture % \usepackage[percent]{overpic} % Superscripts outside math mode \newcommand{\spscript}[2]{\mbox{#1}$^\mathrm{#2}$} \usepackage{makeidx} \makeindex % XMP include for Creative Commons \usepackage{xmpincl} \includexmp{metadata} % Sidebars implemented with picinpar. Use for defining % important terms, mentioning important points in the text, indicating % the current reading assignment, links to useful resources, etc. \usepackage{xcolor} \definecolor{psetback}{HTML}{CFECEC} %html is : light cyan 2 \usepackage{picinpar} \newcommand{\sidebar}[1]{{\fbox{\parbox{0.4\textwidth}{\footnotesize #1}}}} %\newcommand{\problemset}[1]{{\fbox{\parbox{0.9\textwidth}{\footnotesize #1}}}} %\newcommand{\problemset}[1]{{\colorbox{psetback}{{#1}}}} \newcommand{\problemset}[1]{\vline {\colorbox{psetback}{{\parbox{0.9\textwidth}{\footnotesize #1}}}}} % Number figures within the chapter \renewcommand{\thefigure}{\thechapter.\arabic{figure}} \renewcommand{\thetable}{\thechapter.\arabic{table}} \renewcommand{\theequation}{\thechapter-\arabic{equation}} \setcounter{tocdepth}{2} % Fancy chapter beginnings \usepackage[Lenny]{fncychap} \makeatletter \def\@chapapp{} \renewcommand{\DOCH}{} \renewcommand{\DOTI}[1]{% \settowidth{\px}{\CTV\FmTi{#1}\space} \addtolength{\px}{5pt} \settoheight{\py}{\CTV\FmTi{#1}} \addtolength{\py}{1pt} \settowidth{\mylen}{\CNoV\thechapter\space\CTV\FmTi{#1}} \addtolength{\mylen}{1pt} \settowidth{\pxx}{\CNoV\thechapter} \addtolength{\pxx}{-1pt} \settoheight{\pyy}{\CNoV\thechapter} \addtolength{\pyy}{-2pt} \setlength{\myhi}{\pyy} \addtolength{\myhi}{-1\py} \par \parbox[b]{\textwidth}{% \raggedright% \CNoV\thechapter\space\CTV\FmTi{#1}% \hskip -\px% \rule[\pyy]{\px}{\RW} \hskip -5pt% \mghrulefill{\RW}% \rule{\RW}{\pyy}\par\nobreak% \vskip -\baselineskip% \vskip -\pyy% \hskip \mylen% \mghrulefill{\RW}\par\nobreak% \vskip \pyy}% \vskip 40\p@} \renewcommand{\DOTIS}[1]{% \settowidth{\px}{\CTV\FmTi{#1}} \addtolength{\px}{1pt} \settoheight{\py}{\CTV\FmTi{#1}} \addtolength{\py}{1pt} \settowidth{\mylen}{\CNoV\thechapter\space\CTV\FmTi{#1}} \addtolength{\mylen}{1pt} \settowidth{\pxx}{\CNoV\thechapter} \addtolength{\pxx}{-1pt} \settoheight{\pyy}{\CNoV\thechapter} \addtolength{\pyy}{-2pt} \setlength{\myhi}{\pyy} \addtolength{\myhi}{-1\py} \par \parbox[b]{\textwidth}{% \rule[\py]{\RW}{\myhi}% \raggedright% \CTV\FmTi{#1}% \hskip -\px% \rule[\pyy]{\px}{\RW} \hskip -5pt% \mghrulefill{\RW}% \rule{\RW}{\pyy}\par\nobreak% \vskip -\baselineskip% \vskip -\pyy% \mghrulefill{\RW}\par\nobreak% \vskip \pyy}% \vskip 40\p@} \makeatother \renewcommand{\topfraction}{0.9} % max fraction of page for floats at top \renewcommand{\bottomfraction}{0.9} % max fraction of page for floats at bottom \renewcommand{\textfraction}{0.10} % min fraction of page for text \renewcommand{\floatpagefraction}{0.8} % min fraction of floatpage that should have floats % Fancy page headings \usepackage{fancyhdr} \setlength{\headheight}{15pt} \fancyhf{} % clear all header and footer fields \fancyfoot[L]{Signal Computing} \fancyfoot[R]{\thepage} \fancyhead[L]{\slshape\leftmark} \renewcommand{\headrulewidth}{0.4pt} \renewcommand{\footrulewidth}{0.4pt} \fancypagestyle{plain}{% \fancyhf{} % clear all header and footer fields \fancyfoot[L]{Signal Computing} \fancyfoot[R]{\thepage} \renewcommand{\headrulewidth}{0pt} \renewcommand{\footrulewidth}{0.4pt}} \pagestyle{fancy} % Math and logic \usepackage{amsmath} \newcommand{\pred}[1]{\ensuremath{\textit{#1}}} \newcommand{\const}[1]{\ensuremath{\textsf{#1}}} \newcommand{\deriv}[2]{\ensuremath{\frac{\mathrm{d} #1}{\mathrm{d} #2}}} \newcommand{\derivin}[2]{\ensuremath{{\mathrm{d} #1}/{\mathrm{d} #2}}} \newcommand{\sderiv}[2]{\ensuremath{\frac{\mathrm{d}^{2} #1}{\mathrm{d} #2^{2}}}} \newcommand{\nderiv}[3]{\ensuremath{\frac{d^{#3} #1}{d #2^{#3}}}} \newcommand{\pderiv}[2]{\ensuremath{\frac{\partial #1}{\partial #2}}} \newcommand{\spderiv}[2]{\ensuremath{\frac{\partial^2 #1}{\partial #2^{2}}}} \newcommand{\pderivin}[2]{\ensuremath{{\partial #1}/{\partial #2}}} \newcommand{\spderivin}[2]{\ensuremath{{\partial^2 #1}/{\partial #2^{2}}}} \newcommand{\transpose}{\ensuremath{^{\mathsf{T}}}} \DeclareMathOperator{\Real}{Re} \DeclareMathOperator{\Imag}{Im} \DeclareMathOperator{\sinc}{sinc} \newcommand{\commentedout}[1]{} % Variable thickness \hrulefill \makeatletter \def\vhrulefill#1{\leavevmode\leaders\hrule\@height#1\hfill \kern\z@} \makeatother \title{Signal Computing: Digital Signals in the Software Domain} % \author{} isn't being used in this document, so no attempt to make % it "right". \author{ \and \\ {\normalsize University of Washington, Bothell}\\ {\normalsize Prof. }\\ {\normalsize Southern Methodist University}} % \date{August 2016} % \includeonly{ch-fft/fft} % ch-iir/fb-filters} % ch-conv/convolution} % % % %intro,ch-physical/physical-signals,ch-computer/computer-signals,ch-fir/filt-intro} \begin{document} \begin{titlepage} % Another possibility for overlaying a picture % \usepackage[percent]{overpic} needed above % \begin{overpic}[width=\textwidth,grid,tics=10]{cover} % % \begin{overpic}[width=\textwidth]{cover} % \put (0,85) {\begin{minipage}{\textwidth} \mbox{}\\[0.5in] \begin{center} \begin{bfseries} \mbox{}\vhrulefill{2pt}\mbox{}\\[0.25in] {\Huge Signal Computing:}\\[0.25in] {\LARGE Digital Signals in the Software Domain}\\[0.25in] \mbox{}\vhrulefill{2pt}\mbox{}\\[3in] \end{bfseries} %\end{minipage}} %\put (55,-10) {\begin{minipage}{3in} \mbox{}\hfill \parbox{1.5in}{\mbox{}} \parbox{3in}{\raggedright Spring 2020\\[0.25in] \\ \\ University of Washington Bothell\\ 18115 Campus Way NE\\ Bothell, Washington 98011\\[0.25in] \\ Southern Methodist University\\ Lyle School of Engineering\\ 3145 Dyer Street\\ Dallas, TX 75205} \end{center} %\end{minipage}} %\end{overpic} % \mbox{} % \vspace*{5in} % \includegraphics[width=\textwidth]{cover} % \vspace*{-11in} \newpage \mbox{}\vspace{2in} \begin{flushright} Copyright \copyright\ 2002--2020 by Michael and and \\[1in] This material is based upon work supported by the National Science Foundation under Grant No. 0443118.\\[1in] Cover art by Eduardo Sainos.\\[1in] This work is licensed under the Creative Commons Attribution-ShareAlike 4.0 International License. To view a copy of this license, visit \url{http://creativecommons.org/licenses/by-sa/4.0/}.\\ \includegraphics[height=1.5in]{by-sa} \end{flushright} \thispagestyle{empty} \end{titlepage} \frontmatter \tableofcontents \listoffigures \listoftables \listofalgorithms \markboth{LIST OF ALGORITHMS}{} \include{intro} %\newpage %\pagenumbering{arabic} \mainmatter \include{ch-physical/physical-signals} \include{ch-computer/computer-signals} \include{ch-fir/filt-intro} \include{ch-conv/convolution} \include{ch-iir/fb-filters} % \include{chapter5/fourier} % Superseded by ch-fft/fft \include{ch-fft/fft} \include{ch-comp/compression} \include{ch-av/audio-video} \include{ch-rev/review} \appendix \include{exercise-answers} \backmatter \addcontentsline{toc}{chapter}{Index} \printindex \end{document} % LocalWords: MATLAB \begin{table}[!htbp] \footnotesize \centering \caption{\textbf{Descriptive Statistics and Correlation Analysis (NBA)}} \label{table:correlation-nba} \begin{tabularx}{0.9\textwidth}{{r@{ \ \ } p{35mm} r@{}lp{1mm} r@{}l p{5mm} r@{}l p{2mm} r@{}l p{2mm} r@{}l p{2mm} r@{}l p{2mm} r@{}l }} & \\ \hline & \\ \multicolumn{2}{c}{\textbf{ }} & \multicolumn{2}{c}{\textbf{M}} & & \multicolumn{2}{c}{\textbf{SD}} & & \multicolumn{2}{c}{\textbf{1}} & & \multicolumn{2}{c}{\textbf{2}} & & \multicolumn{2}{c}{\textbf{3}} & & \multicolumn{2}{c}{\textbf{4}} & & \\ & \\ \hline & \\ \textbf{1} & \textbf{height} & 79&.2 & & 3&.78 & & 1& & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \\ & \\ \textbf{2} & \textbf{wingspan} & 82&.6 & & 4&.49 & & &.90{$^{***}$} & & 1& & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \\ & \\ \textbf{3} & \textbf{standing.reach} & 103&.7 & & 5&.59 & & &.96{$^{***}$} & & &.95{$^{***}$} & & 1& & & \multicolumn{2}{c}{ \ \ \ \ \ } & & \\ & \\ \textbf{4} & \textbf{hand.length} & 8&.8 & & &.60 & & &.75{$^{***}$} & & &.88{$^{***}$} & & &.84{$^{***}$} & & 1& & & \\ & \\ \textbf{5} & \textbf{hand.width} & 9&.5 & & &.71 & & &.47{$^{***}$} & & &.52{$^{***}$} & & &.51{$^{***}$} & & &.65{$^{***}$} & & \\ & \\ \hline & \\ \multicolumn{19}{p{0.81\textwidth}}{ \footnotesize { \begin{hangparas}{0.75in}{1} \textbf{\underline{Notes}:} \ \ Pearson pairwise correlations are reported; \newline a two-side test was performed to report correlation significance. \end{hangparas} } } & \\ \multicolumn{19}{p{0.81\textwidth}}{ {\tiny {$^{\dagger} p < .10$} } { } {\tiny {$^{*} p < .05$} } { } {\tiny {$^{**} p < .01$} } { } {\tiny {$^{***} p < .001$} } { } } & \\ & \\ \hline \end{tabularx} \end{table} \documentclass[{../../master}]{subfiles} \graphicspath{{../../}} % 個別コンパイル時の画像パスを解決する \begin{document} \section{\textsf{adamr2\_driver}の作成} いよいよモータードライバと通信するソフトウェアドライバノード\textsf{adamr2\_driver\_node}を作成します． ノードの作成にはC++言語を使用するので，ある程度のC++の知識が必要になります． \subsection{パッケージの作成} \textsf{adamr2\_driver\_node}を開発するためのパッケージを新たに作成します． パッケージ名は\textsf{adamr2\_driver}とします． コード\ref{code:create_adamr2_driver}をROSワークスペースディレクトリで実行し，パッケージを生成します． \begin{lstlisting}[language=sh, label=code:create_adamr2_driver, caption= Create \textsf{adamr2\_driver} Package] catkin create pkg adamr2_driver \ --catkin-deps roscpp hardware_interface transmission_interface controller_manager ypspur \end{lstlisting} 依存パッケージとして\textsf{roscpp}，\textsf{hardware\_interface}，\textsf{transmission\_interface}，\textsf{controller\_manager}，\textsf{ypspur}を指定しています． パッケージのディレクトリ構成はコード\ref{code:directory_structure_of_adamr2_driver}のようになります． \textsf{roscpp}を依存パッケージに指定したので，生成されたパッケージのディレクトリにはソースコードを置くためのディレクトリが自動生成されています． このパッケージには\textsf{launch/}や\textsf{config/}ディレクトリ等は作りません． \begin{lstlisting}[label=code:directory_structure_of_adamr2_driver, caption=Directory Structure of \textsf{adamr2\_driver}] adamr2_control/ ├ include/adamr2_driver/ ├ src/ ├ package.xml └ CMakeLists.txt \end{lstlisting} \subsection{C++ノードの作成} ここからC++ノードを作成していきます． ここでは以下の3つのソースファイルから1つのノードを作成することにします． \begin{itemize} \item adamr2\_driver.h：クラス宣言のためのヘッダファイル \item adamr2\_driver.cpp：クラス定義のためのソースファイル \item adamr2\_driver\_node.cpp：ノードの本体 \end{itemize} \textsf{adamr2\_driver.h}は\textsf{include/adamr2\_driver/}ディレクトリに，\textsf{adamr2\_driver.cpp}と\textsf{adamr2\_driver\_node.cpp}は\textsf{src/}ディレクトリに置きます． \subsubsection{\textsf{adamr2\_dirver.h}の記述} まずはヘッダファイルから記述していきます． \textsf{include/adamr2\_driver/}ディレクトリに\textsf{adamr2\_driver.h}という名前のファイルを作成し，コード\ref{code:adamr2_driver_h}のように記述します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_h, caption=\textsf{adamr2\_driver.h}] #ifndef ADAMR2_ADAMR2_DRIVER_H_ #define ADAMR2_ADAMR2_DRIVER_H_ #include #include #include #include namespace adamr2 { class Adamr2Driver : public hardware_interface::RobotHW { public: Adamr2Driver(); ~Adamr2Driver(); ros::Time getTime() const { return ros::Time::now(); } ros::Duration getPeriod() const { return ros::Duration(0.01); } int open() const; void stop() const; void read(ros::Time, ros::Duration); void write(ros::Time, ros::Duration); protected: hardware_interface::JointStateInterface joint_state_interface; hardware_interface::VelocityJointInterface joint_vel_interface; double cmd_[2]; double pos_[2]; double vel_[2]; double eff_[2]; }; } // namespace adamr2 #endif \end{lstlisting} オリジナルのロボットで\textsf{ros\_control}を使用する場合は，\textsf{hardware\_interface::RobotHW}を継承したクラスを作成し，ロボットに合わせた設定や関数定義をする必要があります．\cite{2018実用ロボット開発のためのrosプログラミング} ロボットのドライバと通信を行うのが\textsf{read()}メンバ関数と\textsf{write()}メンバ関数です． \textsf{read()}関数でモータードライバからホイールエンコーダの情報を読み取り，\textsf{write()}関数でモータードライバに車輪の速度を与えます． \footnote{文献によっては\textsf{read()}で指令送信，\textsf{write()}でデータ読み取りを行う，と書かれていることがありますが，感覚的に見て\textsf{write()}で書き込み，\textsf{read()}で読み取りを行うのが道理でしょう．} \textsf{open()}メンバ関数と\textsf{stop()}メンバ関数はそれぞれYP-Spurの初期化関数と停止関数です． メンバ関数の中身の実装は\textsf{adamr2\_driver.cpp}で行います． メンバ変数の\textsf{joint\_state\_interface}及び\textsf{joint\_vel\_interface}は，URDFで定義したホイールジョイントの情報を登録するための変数です． これらの変数の初期化はクラスコンストラクタ内で行います． \textsf{cmd\_[]}メンバ変数は，コントローラが計算した車輪への速度指令値が格納される配列です．対向2輪ロボットなので配列の要素数は2つです． \textsf{pos\_[]}，\textsf{vel\_[]}，\textsf{eff\_[]}メンバ変数は，ジョイントの状態(位置，速度，トルク)を保持するための配列です． モータードライバから報告された位置，速度をこの変数に書き込み，それらの値をもとにコントローラがオドメトリ等を計算します． 尚，トルクに関しては本ロボットでは扱わないので\textsf{eff\_[]}変数が更新されることはありません． \subsubsection{\textsf{adamr2\_dirver.cpp}の記述} 次にクラスの実装を行うソースファイルを記述します． \textsf{src/}ディレクトリに\textsf{adamr2\_driver.cpp}という名前のファイルを作成し，コード\ref{code:adamr2_driver_cpp}のように記述します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_cpp, caption=\textsf{adamr2\_driver.cpp}] #include "adamr2_driver/adamr2_driver.h" #include #include namespace adamr2 { Adamr2Driver::Adamr2Driver() { // YP-Spur initialization. if (this->open() < 0) { ROS_WARN_STREAM("Error: Couldn't open spur.\n"); } pos_[0] = 0.0; pos_[1] = 0.0; vel_[0] = 0.0; vel_[1] = 0.0; eff_[0] = 0.0; eff_[1] = 0.0; cmd_[0] = 0.0; cmd_[1] = 0.0; // Joint state setting for right-wheel-joint hardware_interface::JointStateHandle state_handle_1("right_wheel_joint", &pos_[0], &vel_[0], &eff_[0]); joint_state_interface.registerHandle(state_handle_1); // Joint state setting for left-wheel-joint hardware_interface::JointStateHandle state_handle_2("left_wheel_joint", &pos_[1], &vel_[0], &eff_[0]); joint_state_interface.registerHandle(state_handle_2); registerInterface(&joint_state_interface); // Joint handle setting for right-wheel-joint hardware_interface::JointHandle vel_handle_1(joint_state_interface.getHandle("right_wheel_joint"), &cmd_[0]); joint_vel_interface.registerHandle(vel_handle_1); // Joint handle setting for left-wheel-joint hardware_interface::JointHandle vel_handle_2(joint_state_interface.getHandle("left_wheel_joint"), &cmd_[1]); joint_vel_interface.registerHandle(vel_handle_2); registerInterface(&joint_vel_interface); } Adamr2Driver::~Adamr2Driver() { this->stop(); } int Adamr2Driver::open() const { int ret = Spur_init(); // Set the maximum angular velocity and acceleration. // unit is [rad/s] and [rad/s^2] in tire axis. YP_set_wheel_vel(11.6, 11.6); YP_set_wheel_accel(19.35, 19.35); return ret; } void Adamr2Driver::stop() const { YP_wheel_vel(0, 0); Spur_stop(); ros::Duration(1).sleep(); Spur_free(); } void Adamr2Driver::read(ros::Time time, ros::Duration period) { // yp_vel[0] is right wheel velocity, yp_vel[1] is left wheel velocity. double yp_vel[2] = {0.0, 0.0}; YP_get_wheel_vel(&yp_vel[0], &yp_vel[1]); // Reverse the velocity of the right wheel. // This is due to the coordinate system of the right wheel. yp_vel[0] = -yp_vel[0]; for (unsigned int i = 0; i < 2; i++) { pos_[i] += yp_vel[i] * period.toSec(); vel_[i] = yp_vel[i]; } } void Adamr2Driver::write(ros::Time time, ros::Duration period) { YP_wheel_vel(-cmd_[0], cmd_[1]); } } // namespace adamr2 \end{lstlisting} このソースファイルでは，\textsf{adamr2\_driver.h}で宣言したクラスのメンバ関数の実装を行っています． 以降，関数の詳細な解説を行っていきます． まずはクラスコンストラクタから解説します． コード\ref{code:adamr2_driver_constructor}にクラスコンストラクタを抜粋します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_constructor, caption=Class Constructor in \textsf{adamr2\_driver.cpp}] Adamr2Driver::Adamr2Driver() { // YP-Spur initialization. if (this->open() < 0) { ROS_WARN_STREAM("Error: Couldn't open spur.\n"); } pos_[0] = 0.0; pos_[1] = 0.0; vel_[0] = 0.0; vel_[1] = 0.0; eff_[0] = 0.0; eff_[1] = 0.0; cmd_[0] = 0.0; cmd_[1] = 0.0; // Joint state setting for right-wheel-joint hardware_interface::JointStateHandle state_handle_1("right_wheel_joint", &pos_[0], &vel_[0], &eff_[0]); joint_state_interface.registerHandle(state_handle_1); // Joint state setting for left-wheel-joint hardware_interface::JointStateHandle state_handle_2("left_wheel_joint", &pos_[1], &vel_[0], &eff_[0]); joint_state_interface.registerHandle(state_handle_2); registerInterface(&joint_state_interface); // Joint handle setting for right-wheel-joint hardware_interface::JointHandle vel_handle_1(joint_state_interface.getHandle("right_wheel_joint"), &cmd_[0]); joint_vel_interface.registerHandle(vel_handle_1); // Joint handle setting for left-wheel-joint hardware_interface::JointHandle vel_handle_2(joint_state_interface.getHandle("left_wheel_joint"), &cmd_[1]); joint_vel_interface.registerHandle(vel_handle_2); registerInterface(&joint_vel_interface); } \end{lstlisting} クラスコンストラクタでは，YP-Spur，メンバ変数，ジョイントインターフェースの初期化を行っています． 重要となるのが後半のジョイントインターフェースの初期化です． URDFで定義したホイールジョイントの情報を\textsf{hardware\_interface}に登録しています． \textsf{hardware\_interface::JointStateHandle}クラスに与える第一引数は，URDFで定義したホイールジョイントの名前です． \label{sec:add_wheel_link}で定義した通り，\textsf{right\_wheel\_joint}と\textsf{left\_wheel\_joint}という名前で登録しています． 次に，\textsf{open()}関数と\textsf{stop()}関数です．コード\ref{code:adamr2_driver_open_and_stop}に抜粋します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_open_and_stop, caption=\textsf{open()} and \textsf{stop()} Member Function in \textsf{adamr2\_driver.cpp}] int Adamr2Driver::open() const { int ret = Spur_init(); // Set the maximum angular velocity and acceleration. // unit is [rad/s] and [rad/s^2] in tire axis. YP_set_wheel_vel(11.6, 11.6); YP_set_wheel_accel(19.35, 19.35); return ret; } void Adamr2Driver::stop() const { YP_wheel_vel(0, 0); Spur_stop(); ros::Duration(1).sleep(); Spur_free(); } \end{lstlisting} \textsf{open()}関数ではYP-Spurの初期化を行っています． \textsf{ypspur}ライブラリの\textsf{Spur\_init()}関数を使って初期化を行っています． また，\textsf{YP\_set\_wheel\_vel}関数及び\textsf{YP\_set\_wheel\_accel}関数を使って，各車輪の最大速度と最大加速度を設定しています． 第一引数が右車輪，第二引数が左車輪に対応する値です．どちらも同じ数値を設定します． ここで設定している値は，直径\SI{155}{mm}のホイールを付けた際にロボットの直進移動速度が\SI{0.9}{m/s}を超えないように，また直進加速度が\SI{1.5}{m/s^2}を超えないような値に設定しています． 次に\textsf{read()}関数と\textsf{write()}関数です． コード\ref{code:adamr2_driver_read_and_write}に抜粋します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_read_and_write, caption=\textsf{read()} and \textsf{write()} Member Function in \textsf{adamr2\_driver.cpp}] void Adamr2Driver::read(ros::Time time, ros::Duration period) { // yp_vel[0] is right wheel velocity, yp_vel[1] is left wheel velocity. double yp_vel[2] = {0.0, 0.0}; YP_get_wheel_vel(&yp_vel[0], &yp_vel[1]); // Reverse the velocity of the right wheel. // This is due to the coordinate system of the right wheel. yp_vel[0] = -yp_vel[0]; for (unsigned int i = 0; i < 2; i++) { pos_[i] += yp_vel[i] * period.toSec(); vel_[i] = yp_vel[i]; } } void Adamr2Driver::write(ros::Time time, ros::Duration period) { YP_wheel_vel(-cmd_[0], cmd_[1]); } \end{lstlisting} \textsf{read()}関数では\textsf{YP\_get\_wheel\_vel()}関数を使って，モータードライバから各車輪の速度を求めています． 受け取った結果を一時変数に保持し，座標系を解決してから\textsf{pos\_[]}変数と\textsf{vel\_[]}変数に格納しています． \label{sec:add_wheel_link}でも述べたように，本ロボットのホイールの座標系は，「ロボットが直進する方向を正とする」と定めています． しかしYP-Spurが返す車輪速度値はそのようにはなっていません． 具体的にいうと，右車輪の速度の正負が逆さまになって返ってきます． そのため，右車輪の速度の値の正負を反転させてメンバ変数に記録しています． YP-Spurから得られる情報は車輪速度だけですが，時間がわかっているので位置も求めることができます(あくまで推定値ですが)． \textsf{read()}関数の最後の\textsf{for}文で，ジョイントの位置と速度をメンバ変数に記録しています． \textsf{write()}関数では車輪の速度指令値をYP-Spurに与えているだけです． ここでも右車輪の値の正負は反転させなければなりません． \subsubsection{\textsf{adamr2\_dirver\_node.cpp}の記述} 最後にROSノードの本体となるソースファイルを記述します． \textsf{src/}ディレクトリに\textsf{adamr2\_driver\_node.cpp}という名前のファイルを作成し，コード\ref{code:adamr2_driver_node}のように記述します． \begin{lstlisting}[language=C++, label=code:adamr2_driver_node, caption=\textsf{adamr2\_driver\_node.cpp}] #include "adamr2_driver/adamr2_driver.h" #include #include //extern "C" { #include //} int main(int argc, char *argv[]) { ros::init(argc, argv, "adamr2_driver_node"); ros::NodeHandle nh; adamr2::Adamr2Driver driver; controller_manager::ControllerManager cm(&driver); ros::AsyncSpinner spinner(1); spinner.start(); while(ros::ok()) { ros::Time now = driver.getTime(); ros::Duration dt = driver.getPeriod(); if (YP_get_error_state() == 0) { driver.write(now, dt); cm.update(now, dt); driver.read(now, dt); } else { ROS_WARN("T-Frog driver disconnected."); driver.stop(); while (driver.open() < 0) { ROS_WARN("Try to connect T-Frog driver..."); ros::Duration(1).sleep(); } ROS_INFO("T-Frog driver connected."); } dt.sleep(); } spinner.stop(); return 0; } \end{lstlisting} \textsf{adamr2\_driver.h}で宣言したクラスのオブジェクト(\textsf{driver})を作り，\textsf{ros\_control}の枠組みに従って\textsf{while}文で制御ループを回しています． \textsf{while}ループの中で，\textsf{driver.write()}関数を呼び出して速度指令値をモータードライバに送り，\textsf{cm.update()}関数を呼び出して状態を更新，\textsf{driver.read()}関数を呼び出して車輪の現在速度を取得しています． \subsection{\textsf{CMakeLists.txt}の編集} ノードをビルドするために，\textsf{CMakeLists.txt}を編集します． コード \begin{lstlisting}[language=cmake, label=code:adamr2_driver_cmakelists, caption=\textsf{CMakeLists.txt}] cmake_minimum_required(VERSION 2.8.3) project(adamr2_driver) find_package(catkin REQUIRED COMPONENTS roscpp hardware_interface transmission_interface controller_manager ) catkin_package() find_library(ypspur_LIBRARIES ypspur) include_directories( include ${catkin_INCLUDE_DIRS} ${ypspur_INCLUDE_DIRS} ) add_executable(adamr2_driver_node src/adamr2_driver_node.cpp src/adamr2_driver.cpp ) target_link_libraries(adamr2_driver_node ${catkin_LIBRARIES} ${ypspur_LIBRARIES} ) \end{lstlisting} \subsection{パッケージのビルド} ROSノードの実行可能ファイルを生成するために，ビルドを実行します． コードを実行して，ワークスペースをビルドしましょう． \begin{lstlisting}[language=sh, label=code:build_adamr2_driver, caption=Build Workspace] catkin build \end{lstlisting} ビルドが通ったら，無事にドライバノードを作成できたことになります． \end{document}jacopok/notes \documentclass[main.tex]{subfiles} \begin{document} \section{Symmetries and conservation laws} \marginpar{Tuesday\\ 2020-3-17, \\ compiled \\ \today} Our aim is to describe the fundamental constituents of matter with a Quantum Field Theory. The method used to derive the equations of motion is a variational principle: we will find a Lagrangian density for various particles, and then apply the variational principle to find their equations of motion. A guiding principle on the description of these fundamental particles is based on using their symmetries. We have Nöether's theorem in Quantum Field Theory: from these symmetries we are able to find conserved quantities. These symmetries are described with groups, since we can compose their application; the theory describing groups is very rich. For this lecture we will base ourselves on Peskin's chapter 2 \cite[]{peskinConceptsElementaryParticle2019}. A group \(G\) is a set of elements endowed with an operation. The set of elements can be either discrete or continuous. Examples of discrete transformations are the parity transformation \(P\): \(P \vec{x} = -\vec{x}\) and the time swap \(T\): \(T x^{\mu } = (-x^{0}, \vec{x})\). Continuous symmetries, on the other hand, are parametrized by one or more continuous-valued parameters. We distinguish: \begin{enumerate} \item \textbf{spacetime} symmetries: groups which transform our coordinate system for spacetime, such as Lorentz and Poincaré transformations; \item \textbf{internal} symmetries: groups which transform a certain field, or a certain property of our quantum system. \end{enumerate} For our set to be a group, we need to be able to define an operation --- we will usually call it multiplication --- between the elements of the group, such that if \(a, b \in G\) then \(ab \in G\). Also, we must have \begin{enumerate} \item associativity: \((ab)c = a(bc)\); \item existence of the identity \(\mathbb{1}\), such that \(\mathbb{1} a = a \mathbb{1} = a\); \item existence of inverses: there exists \(a^{-1}\) such that \(a a^{-1} = a^{-1} a = \mathbb{1}\). \end{enumerate} What is of interest to us is the association of the group with a transformation which is a symmetry: this is called a \emph{representation}, which associates to each \(g \in G\) a unitary operator \(U_g\) acting on the quantum states. We ask that this representation should preserve the group structure, that is to say, \(U_{gh} = U_g U_h\) and \(U_{g^{-1}} = U_{g}^{-1}\). We call a transformation a symmetry if, after performing the transformation, the dynamics of the system do not change. For a quantum mechanical system, we are interested in the observables: these are described by operators, whose eigenvalues are the observations we make, and which in the Heisenberg picture evolve like % \begin{align} - i \hbar \dv{}{t} O(t) = [H, O(t)] \,; \end{align} % if an operator commutes with the Hamiltonian, \([H,O]=0\), then the operator's expectation value on any state is constant --- which is to say, the operator is constant. If we perform a transformation in the form % \begin{align} \ket{\psi } \rightarrow \ket{\psi'} = U \ket{\psi } \,, \end{align} % then the operators will change by % \begin{align} O \rightarrow O^{\prime } = U ^\dag O U \,. \end{align} Note that whether we have \(U ^\dag O U\) or \(U O U ^\dag\) does not matter, since we ask observables \(O\) to be Hermitian, so \(O = O ^\dag\). We know that these transformations must always be unitary, because the conservation of probability implies that we must have \(\braket{\psi }{\psi } = \const\): so, % \begin{align} U ^\dag U = \mathbb{1} \,. \end{align} This can be also stated as \(U ^\dag = U^{-1}\). So, the function associating a unitary operator \(U\) to an element \(g\) of the group is called its \emph{unitary representation}. A transformation \(G\) is a symmetry if \(\forall a \in G\) we have % \begin{align} [U(a), H] =0 \,, \end{align} % that is, the unitary representation of the group element always commutes with the Hamiltonian. If we have a state \(\ket{\psi }\) with energy \(H \ket{\psi } = E \ket{\psi }\), then the transformation commuting with the Hamiltonian means that \(\ket{\psi'} = U \ket{\psi }\) has the same energy: % \begin{align} H (U \ket{\psi }) = HU \ket{\psi } \overset{[H, U] = 0}{=} UH \ket{\psi } = U E \ket{\psi } = E \qty(u \ket{\psi }) \,, \end{align} % so the eigenvalue of \(U \ket{\psi }\) is the same as that of \(\ket{\psi }\). Now, we can move to an example, taken from Peskin \cite[eq.\ 2.38 onward]{peskinConceptsElementaryParticle2019}. Consider the discrete group \(\mathbb{Z}_{2}\), which only has the elements \(1\) and \(-1\), with the same multiplication rules as those we would have if these elements were integers. So, the group is closed with respect to multiplication. It can be easily checked that this is indeed a group based on our definition. In order for this to be of interest to us, we can consider a quantum mechanical system and find a unitary representation acting on its Hilbert space. Let us suppose we have a QM system with a basis made of two states \(\ket{\pi^{+}}\) and \(\ket{\pi^{-}}\). Let us define the \emph{charge conjugation} operator \(C\), by: % \begin{align} C \ket{\pi^{+}} = \ket{\pi^{-}} \qquad \text{and} \qquad C \ket{\pi^{-}} = \ket{\pi^{+}} \,. \end{align} So, we can find a unitary representation of \(\mathbb{Z}_{2}\) in this system: we need to define \(U(1)\) and \(U(-1)\). We define % \begin{subequations} \begin{align} U(1) = \mathbb{1} =\left[\begin{array}{cc} 1 & 0 \\ 0 & 1 \end{array}\right] \qquad \text{and} \qquad U(-1) = \sigma_{x} = \left[\begin{array}{cc} 0 & 1 \\ 1 & 0 \end{array}\right] \,, \end{align} \end{subequations} % where the matrices are to be interpreted as acting on vectors expressed to the basis \(\qty{\ket{\pi_{+}}, \ket{\pi_{-}}}\). So, we can say that our unitary representation looks like % \begin{align} \mathbb{Z}_{2} \rightarrow \qty{\mathbb{1}, C} \,. \end{align} Now, if \([C, H] =0\) (and \(\mathbb{1}\) commutes with \(H\), which is always the case) then we say that ``\(H\) has the symmetry \(\mathbb{Z}_{2}\)'': this implies that the energies of the two \(\pi_{\pm}\) particles are equal. The interesting question to determine will be whether this is actually the case for our given group. Groups can be subdivided into abelian and non-abelian ones. A group is abelian if for every \(a, b\) in \(G\) we have \(ab = ba\), or equivalently, \([a, b] =0\). It is not if this is not the case, that is, there exist \(a, b\) such that \(ab \neq ba\). The condition on the elements directly translates to a condition on the matrices of the unitary representation. If we have commuting matrices, we can simultaneously diagonalize them: for example, in the case of \(\mathbb{Z}_{2}\) we can go to a basis in which % \begin{subequations} \begin{align} C = \left[\begin{array}{cc} 1 & 0 \\ 0 & -1 \end{array}\right] \,, \end{align} \end{subequations} % specifically the states on which this matrix will act will need to be % \begin{align} \ket{\pi_{1}} = \frac{\ket{\pi^{+}} + \ket{\pi^{-}}}{\sqrt{2}} \qquad \text{and} \qquad \ket{\pi_{2}} = \frac{\ket{\pi^{+}} - \ket{\pi^{-}}}{\sqrt{2}} \,, \end{align} % since then \(C \ket{\pi_1 } = \ket{\pi_1 }\) (we write \(C = +1\)) and \(C \ket{\pi_{2}} = - \ket{\pi_2 }\) (we write \(C = -1\)). We will often use this notation, confusing operator and eigenvalue. In the case of nonabelian groups it is not in general possible to diagonalize all the matrices; we can however do a change of basis and write the matrices as a block matrix with the smallest possible blocks: % \begin{subequations} \begin{align} U_{R} = \left[\begin{array}{ccc} U_1 & 0 & 0 \\ 0 & U_2 & 0 \\ 0 & 0 & \dots \end{array}\right] \,, \end{align} \end{subequations} % where the matrices \(U_i\) are called the \textbf{irreducible unitary representations of \(G\)}. The dimension of the matrices \(U_i\) tells us the dimension of these irreducible unitary representations. \todo[inline]{Add more details on irreps --- maybe not here? They can be found in professor Rigolin's intro to groups.} Do note that some elements of a nonabelian group can commute: for example, in the rotation group we have % \begin{align} [J^{i}, J^{j}] = \epsilon^{ijk} J_{k} \,, \end{align} % so if we take \(i = j\), that is, we consider rotations along the same axis, they will commute since then the Kronecker symbol is equal to zero. \subsection{Continuous transformations: space translations} An element of the group can be written as % \begin{align} U(a) = e^{ - i a P } \,, \end{align} % where the operator \(P\), whose eigenvalue is the momentum, is called the generator of the transformation. If we consider a plane wave we can clearly see how this action works: if we start from % \begin{align} \braket{x}{p} = e^{i p x} \,, \end{align} % we can apply the operator \(U(a)\) to \(\ket{p}\), which will yield \(e^{-ipa}\) (since eigenvectors of an operator are also eigenvectors of its exponential): so we find % \begin{align} \bra{x} U(a) \ket{p} = e^{i p (x-a)} \,, \end{align} % which means that by acting with this operator we have effectively performed a translation with displacement \(a\). If our system is invariant under translations, then Nöether's theorem tells us that the momentum is conserved. In order to be a physical observable \(P\) needs to be Hermitian: \(P = P ^\dag\). So, the adjoint of the transformation \(U(a)\) is % \begin{align} U ^\dag (a) = \sum _{n} \qty(\frac{(-iaP)^{n}}{n!}) ^\dag = \sum _{n} \frac{(iaP ^\dag)^{n}}{n!} = e^{i a P ^\dag} = e^{iaP} = U^{-1}(a) \,, \end{align} % which confirms the fact that the transformation is unitary. Let us suppose that the momentum operator \(P\) commutes with the Hamiltonian: \([P,H] = 0\). Then, % \begin{align} [U(a), H] = 0 \,, \end{align} % that is, the Hamiltonian is translation-invariant. All this is to say that a constant of motion \(O\) corresponds to an operator \(O\) which commutes with the Hamiltonian. This is formalized by Nöther's theorem, which establishes the equivalence between symmetries and conservation laws: % \begin{align} [O, H] = 0 \iff [U_O, H] = 0 \,. \end{align} As an example, take the group \(G\) of 3D rotations. They depend on a continuous parameter \(\vec{\alpha}\), just like translations depended on the parameter \(a\). The rotation is written as % \begin{align} U(\vec{\alpha}) = e^{-i \vec{\alpha} \cdot \vec{J}} \,, \end{align} % where the components of the angular momentum have the following commutation relations: % \begin{align} [J^{i}, J^{j}] = i \epsilon^{ijk} J^{k} \,. \end{align} We will be able to compose the representations of rotations: % \begin{align} U(\vec{\beta}) U(\vec{\alpha}) = U(\vec{\gamma}) \,. \end{align} This space of 3D rotations is called SO(3), since every rotation corresponds to a 3x3 matrix which is a rotation matrix --- it is orthogonal and has determinant 1. Now, we seek \textbf{representations} of these rotations: so, we choose the dimension \(d\) of a quantum-mechanical vector and describe how it changes upon the action of the unitary matrices found by exponentiating certain \(d\)-dimensional generators \(J^{i}\), which must have the algebra discussed above. If we look for 1D representations of the generators \(J^{i}\) the only option we find is \(J^{i} = 0\), which means that we are not actually performing a rotation. This is because scalars commute with each other. Which states transform this way? These are scalar states, with spin 0. For 2D representations, we have % \begin{align} J^{i} =\frac{1}{2} \sigma^{i} \,, \end{align} % where the \(\sigma^{i}\) are the Pauli matrices. We can also find 3D representations, which look like % \begin{subequations} \begin{align} J^{1}= \left[\begin{array}{ccc} 0 & 0 & 0 \\ 0 & 0 & -i \\ 0 & i & 0 \end{array}\right] \qquad J^{2}= \left[\begin{array}{ccc} 0 & 0 & i \\ 0 & 0 & 0 \\ -i & 0 & 0 \end{array}\right] \qquad J^{1}= \left[\begin{array}{ccc} 0 & -i & 0 \\ i & 0 & 0 \\ 0 & 0 & 0 \end{array}\right] \, \end{align} \end{subequations} % and represent a spin 1 particle. In general, spin \(s\) corresponds to a \(2s+1\)-dimensional representation. A rotation in 2D, represented by an element of SO(2), corresponds to a phase shift, so we can say that it is equivalent to an element of U(1). This then allows us to see that SO(2) is abelian. In general, we write for a unitary \(n \times n\) representation % \begin{align} U(n) \rightarrow e^{-i \alpha^{n} t^{a}} \,, \end{align} % where the generators \(t^{a}\) are Hermitian matrices corresponding to Hermitian operators. In particular, conventionally we say that one of these is the identity: \(t^{0} = \mathbb{1}\) (which must always be included in the group, lest we lose closure). So, we omit it and say that we have \(n^2-1\) generators for the SU\((n)\) group. We shall see that each of these generators corresponds to a particle, and for the weak interaction we will have \(2^2-1 = 3\) particles, while for the strong one we will have \(3^2-1=8\). In general, if \(t^{a}\) are the generators of an abstract Lie group, we can describe the algebra of the group by % \begin{align} \qty[t^{a}, t^{b}] = i f^{abc} t^{c} \,, \end{align} % so, the commutator is decomposed into a linear combination of the generators, whose coefficients \(f^{abc}\) are called the \textbf{structure constants} of the group. \end{document}files/bibtex/mixhop.bib0 @inproceedings{mixhop, author={ AND AND AND AND AND AND AND }, title={MixHop: Higher-Order Graph Convolution Architectures via Sparsified Neighborhood Mixing}, booktitle = {International Conference on Machine Learning (ICML)}, year = {2019}, }vforkliu/pandoc-examples % change background color for inline code in % markdown files. The following code does not work well for % long text as the text will exceed the page boundary \definecolor{bgcolor}{HTML}{E0E0E0} \let\oldtexttt\texttt \renewcommand{\texttt}[1]{ \colorbox{bgcolor}{\oldtexttt{#1}} } \documentclass[12pt]{article} \usepackage{amsmath} \usepackage{latexsym} \usepackage{amsfonts} \usepackage[normalem]{ulem} \usepackage{soul} \usepackage{array} \usepackage{amssymb} \usepackage{extarrows} \usepackage{graphicx} \usepackage[backend=biber, style=numeric, sorting=none, isbn=false, doi=false, url=false, ]{biblatex}\addbibresource{bibliography.bib} \usepackage{subfig} \usepackage{wrapfig} \usepackage{wasysym} \usepackage{enumitem} \usepackage{adjustbox} \usepackage{ragged2e} \usepackage[svgnames,table]{xcolor} \usepackage{tikz} \usepackage{longtable} \usepackage{changepage} \usepackage{setspace} \usepackage{hhline} \usepackage{multicol} \usepackage{tabto} \usepackage{float} \usepackage{multirow} \usepackage{makecell} \usepackage{fancyhdr} \usepackage[toc,page]{appendix} \usepackage[hidelinks]{hyperref} \usetikzlibrary{shapes.symbols,shapes.geometric,shadows,arrows.meta} \tikzset{>={Latex[width=1.5mm,length=2mm]}} \usepackage{flowchart}\usepackage[paperheight=11.69in,paperwidth=8.27in,left=1.0in,right=1.0in,top=1.0in,bottom=1.0in,headheight=1in]{geometry} \usepackage[utf8]{inputenc} \usepackage[T1]{fontenc} \TabPositions{0.5in,1.0in,1.5in,2.0in,2.5in,3.0in,3.5in,4.0in,4.5in,5.0in,5.5in,6.0in,} \urlstyle{same} %%%%%%%%%%%% Set Depths for Sections %%%%%%%%%%%%%% % 1) Section % 1.1) SubSection % 1.1.1) SubSubSection % 1.1.1.1) Paragraph % 1.1.1.1.1) Subparagraph \setcounter{tocdepth}{5} \setcounter{secnumdepth}{5} %%%%%%%%%%%% Set Depths for Nested Lists created by \begin{enumerate} %%%%%%%%%%%%%% \setlistdepth{9} \renewlist{enumerate}{enumerate}{9} \setlist[enumerate,1]{label=\arabic*)} \setlist[enumerate,2]{label=\alph*)} \setlist[enumerate,3]{label=(\roman*)} \setlist[enumerate,4]{label=(\arabic*)} \setlist[enumerate,5]{label=(\Alph*)} \setlist[enumerate,6]{label=(\Roman*)} \setlist[enumerate,7]{label=\arabic*} \setlist[enumerate,8]{label=\alph*} \setlist[enumerate,9]{label=\roman*} \renewlist{itemize}{itemize}{9} \setlist[itemize]{label=$\cdot$} \setlist[itemize,1]{label=\textbullet} \setlist[itemize,2]{label=$\circ$} \setlist[itemize,3]{label=$\ast$} \setlist[itemize,4]{label=$\dagger$} \setlist[itemize,5]{label=$\triangleright$} \setlist[itemize,6]{label=$\bigstar$} \setlist[itemize,7]{label=$\blacklozenge$} \setlist[itemize,8]{label=$\prime$} \setlength{\topsep}{0pt}\setlength{\parindent}{0pt} %%%%%%%%%%%% This sets linespacing (verticle gap between Lines) Default=1 %%%%%%%%%%%%%% \renewcommand{\arraystretch}{1.3} \title{Algoritmos Asimétricos } \date{} %%%%%%%%%%%%%%%%%%%% Document code starts here %%%%%%%%%%%%%%%%%%%% \begin{document} \maketitle \par \par Corporación Universitaria Minuto de Dios \par \par Facultad de Ingeniería\par \vspace{\baselineskip} \section*{1\hspace*{10pt}Introducción} \addcontentsline{toc}{section}{1\hspace*{10pt}Introducción} {\fontsize{16pt}{19.2pt}\selectfont L\par}os algoritmos asimétricos utilizan diferentes claves para cifrar y descifrar datos. Un ejemplo de cifrado asimétrico es la criptografía de clave pública. La criptografía de clave pública utiliza dos claves que forman un par de claves llamadas clave pública y clave privada. La clave que cifra el texto sin formato no se puede utilizar para descifrar el texto cifrado. La clave pública cifra el texto plano y la clave privada descifra el texto cifrado.\par \vspace{\baselineskip} \textbf{Clave pública:}Proporcionada a quienes le envían datos cifrados.\par \textbf{Clave privada:} Una clave en posesión exclusiva del usuario. Cuando un mensaje de texto sin formato se cifra con la clave pública, solo el poseedor de la clave privada puede descifrar el texto cifrado. Cuando un mensaje de texto sin formato se cifra con la clave privada, cualquiera que tenga la clave pública puede descifrarlo. Existe una certeza absoluta de que el mensaje de texto sin formato se originó con el poseedor de la clave privada. Las claves asimétricas proporcionan autenticación, integridad y no repudio. También pueden respaldar la confidencialidad cuando se utilizan para la gestión de claves.\par \section*{2\hspace*{10pt}Algoritmos Asimétrico existente:} \addcontentsline{toc}{section}{2\hspace*{10pt}Algoritmos Asimétrico existente:} \subsection{Diffie-Hellman} El algoritmo de intercambio de claves Diffie-Hellman fue publicado por primera vez en 1976 por y , aunque el algoritmo había sido inventado unos años antes por la agencia de inteligencia del gobierno británico GCHQ pero se mantuvo clasificado. En 2002, sugirió que el algoritmo fuera renombrado como "El intercambio de claves Diffie-Hellman-Merkle" en reconocimiento de la contribución de Ralph Merkle a la criptografía de clave pública.\par \vspace{\baselineskip} El algoritmo de intercambio de claves Diffie-Hellman resuelve el siguiente problema: Alice y Bob quieren compartir una clave secreta para, por ejemplo, un algoritmo de clave simétrica como DES o AES , pero solo pueden comunicarse a través de un canal inseguro que es escuchado por su adversario Eva. Es decir, todos los mensajes enviados entre Alice y Bob son observados por Eve.\par En la figura siguiente se muestra un ejemplo de funcionamiento del protocolo Diffie-Hellman.\par \vspace{\baselineskip} \vspace{\baselineskip} Los valores de $``$p$"$ y $``$g$"$ son públicos y cualquier atacante puede conocerlos, pero esto no supone una vulnerabilidad. Aunque un atacante conociese dichos valores y capturara los dos mensajes enviados entre las máquinas A y B, no sería capaz de averiguar la clave secreta. A continuación se muestra la información capturada por un atacante en el escenario de la Figura 46:\par \vspace{\baselineskip} (ga mod p) = 8 $ \rightarrow $ (5a mod 23) = 8\par (gb mod p) = 19 $ \rightarrow $ (5b mod 23) = 19\par \vspace{\baselineskip} A partir de las ecuaciones anteriores, intentar calcular los valores de $``$a$"$ y $``$b$"$ es lo que se conoce como el problema del algoritmo discreto, un problema que se cree computacionalmente intratable y cuya notación es la siguiente:\par \vspace{\baselineskip} a = log discg (ga mod p) = log disc 5 (8)\par b = log discg (gb mod p) = log disc 5 (19)\par \vspace{\baselineskip} Con los valores del ejemplo sí que es posible encontrar la solución, ya que se ha escogido un número primo $``$p$"$ muy pequeño (p = 23), y se sabe que $``$a$"$ y $``$b$"$ son menores que $``$p$"$ . Por lo tanto, para obtener los valores secretos en este ejemplo, un atacante tendría que probar sólo 22 posibles valores.\par \vspace{\baselineskip} Por suerte, las implementaciones actuales del protocolo Diffie-Hellman utilizan números primos muy grandes, lo que impide a un atacante calcular los valores de $``$a$"$ y $``$b$"$ . El valor $``$g$"$ no necesita ser grande, y en la práctica su valor es 2 ó 5. En el RFC 3526 aparecen publicados los números primos que deben utilizarse. A modo de ejemplo, se facilita aquí el número primo de 1024 bytes propuesto. El valor $``$g$"$ utilizado es 2:\par \vspace{\baselineskip} p = 28192 – 28128 – 1 + 264 x ((28062 pi) + 4743158)\par \subsection{DSA} (Digital Signature Algorithm en español Algoritmo de Firma Digital) es un estándar del Gobierno Federal de los Estados Unidos de América o FIPS para firmas digitales. Fue un algoritmo propuesto por el Instituto Nacional de Normas y Tecnología de los Estados Unidos para su uso en su Estándar de Firma Digital (DSS), especificado en el FIPS 186. DSA se hizo publico el 30 de Agosto de 1991, este algoritmo como su nombre lo indica, sirve para firmar y no para cifrar informacion. Una desventaja de este algoritmo es que requiere mucho mas tiempo de computo que RSA.\par \par \vspace{\baselineskip} \subsection{Cifrado El Gamal} El procedimiento de cifrado/descifrado ElGamal se refiere a un esquema de cifrado basado en el problema matemático del logaritmo discreto. Es un algoritmo de criptografía asimétrica basado en la idea de Diffie-Hellman y que funciona de una forma parecida a este algoritmo discreto.\par El algoritmo de ElGamal puede ser utilizado tanto para generar firmas digitales como para cifrar o descifrar.\par Fue descrito por en 19841​ y se usa en software GNU Privacy Guard, versiones recientes de PGP, y otros sistemas criptográficos. Este algoritmo no está bajo ninguna patente lo que lo hace de uso libre.\par La seguridad del algoritmo se basa en la suposición que la función utilizada es de un solo sentido debido a la dificultad de calcular un logaritmo discreto.\par \par \subsection{Criptografía de curva elíptica} Es una variante de la criptografía asimétrica o de clave pública basada en las matemáticas de las curvas elípticas. Sus autores argumentan que la CCE puede ser más rápida y usar claves más cortas que los métodos antiguos —como RSA— al tiempo que proporcionan un nivel de seguridad equivalente. La utilización de curvas elípticas en criptografía fue propuesta de forma independiente por y en 1985.\par \par \vspace{\baselineskip} \par \subsection{Criptosistema de Merkle-Hellman} Fue uno de los primeros criptosistemas de llave pública y fue inventado por y en 1978.1​ Aunque sus ideas eran elegantes, y mucho más simples que RSA, no tuvo el mismo éxito que este último, debido a que MH ya fue roto,2​ y además no ofrece funcionalidades para firmar.\par \par \par \subsection{RSA} \vspace{\baselineskip} \section*{3 Objectivos de los algoritmos Asimétricos} \addcontentsline{toc}{section}{3 Objectivos de los algoritmos Asimétricos} es suministrar la dificultad máxima al proceso de desencriptar los datos sin utilizar la llave exacta garantías de seguridad de la información en el proceso que se implemente para asegurar la información que circula diariamente por ella, algo que es de suma importancia para los desarrolladores de sistemas pues de ésto depende la confiabilidad que se le ofrezca a los usuarios.\par \section*{4\hspace*{10pt}Conclusion:} \addcontentsline{toc}{section}{4\hspace*{10pt}Conclusion:} \vspace{\baselineskip} Los Algoritmos Asimétricos son uno de los métodos para poder proteger tu información, forma parte de la seguridad informática que cada usuario puede tener, en la información anterior podemos observar que la Algoritmos Asimétricos tiene su historia, y mediante esta podemos observar las diferentes opciones que esta nos otorga para poder cuidar nuestra información, sin embargo, en este laboratorio, me intereso el tema de cifrar archivos con algoritmos complejos ya que logra una mayor confidencialidad.\par \printbibliography \end{document}1-10 % $Id$ % Purpose: Dimensional units % Copyright (c) 1998--2009, % This program may be distributed and/or modified under the % conditions of the LaTeX Project Public License (LPPL), % either version 1.2, or (at your option) any later version. % LPPL: http://www.latex-project.org/lppl.txt % The original author of this software, , seeks to improve % it with your suggestions, contributions, bug-reports, and patches. % <>, surname=zender % Department of Earth System Science % University of California at Irvine % Irvine, CA 92697-3100 % See also http://physics.nist.gov/cuu/Units/units.html % Usage: % Notes on typography: % ISO: ``A measurement consisting of a number plus a dimension is an % indivisible unit, with a smaller than normal space between them, as % 5.3\,km and 62\,kg. The dimension is in an upright font''. (KoD99 p. 142.) % comp.text.tex thread 20030709: J. Fluid. Mech. convention appears to % be \; space between value and dimension, and then \, % interdimensional % spacing, e.g., 5\;cm\,s$^{-1}$ % My convention: Unit separation from value is small space ``\,'' % (KoD99 p. 142), and intra-dimensional spacing is also small space % Dependencies: upgreek.sty (required solely for \upmu) % Convention: % Lowercase ``x'' separates numerator from denominator % As in SI, uppercase letters signify prefixes greater than 10 % (except kilo = ``k'' because ``K'' stands for Kelvin) % Uppercase letters also signify atomic species, e.g., ``N'', ``S'', ``C'' % All units are expressed in lowercase letters unless lowercasing % would create an ambiguous conflict with a fundamental dimensions, % e.g., use ``G'' for Giga instead of ``g'', which means gram. % Uppercase letters in dimension commands: % B = Byte % C = cubed or Carbon % F = to the fifth (quintic) % G = Giga % K = Kelvin % M = Mega % N = Newton or Nitrogen % Q = to the fourth (quartic) % S = squared or Sulfur % T = Ton % T = Tera % Lowercase letters in dimension commands: % u = micro % x = per % Equivalencies: % Megaton MT equals Teragram Tg % Bugs: % k and K are both used for Kelvin % Usage: % \usepackage{dmn} % Dimensional units % Units with single letter symbols (which therefore do not need commands) %\newcommand{\a}{A} % [A] Electric current (Ampere) %\newcommand{\c}{C} % [C] Charge (Coulomb) %\newcommand{\h}{H} % [H] Inductance (Henry) %\newcommand{\k}{K} % [K] Thermodynamic temperature (Kelvin) %\newcommand{\s}{S} % [S] Conductivity (Siemens) %\newcommand{\t}{T} % [T] Magnetic induction (Tesla) %\newcommand{\v}{V} % [V] Electric potential (Volt) %\newcommand{\w}{W} % [W] Power (Watt) % Message printed when LaTeX called \@ifundefined{ProvidesPackage}{}{ \ProvidesPackage{dmn}[2002/09/22 v1.81 Dimensional units] } % end ProvidesPackage \csznote{ % preview.sty method of printing CVS version number with listings.sty \NeedsTeXFormat{LaTeX2e} \def\reserved@a #1#2$#3: #4${\edef#1{\reserved@c #2#4 $}} \def\reserved@c #1 #2${#1} \reserved@a\reserved@b $HeadURL$ \ifx\reserved@b\@empty \reserved@a\reserved@b CVS-$Revision$ \else \begingroup \lccode`-=`. \def\next rel-{} \edef\next{\lowercase{\endgroup \def\noexpand\reserved@b{\expandafter\next\reserved@b}}} \next \fi \reserved@a\next $Date$ \edef\next{\noexpand\ProvidesPackage{preview}% [\next\space preview-latex \reserved@b]} } % end csznote % dmn.sty depends on upgreek.sty for provision of upright Greek letters \usepackage[Symbol]{upgreek} % Upright Greek letters [Euler,Symbol,Symbolsmallscale] % Conflicting definitions \providecommand{\C}{\chm{C}}\renewcommand{\C}{\chm{C}} % [C] Temperature (Celsius) (C is also Carbon in chm.sty) % Units % \newcommand{\Tg}{Tg} % Seems to cause problems/conflicts % \newcommand{\nbr}{\#} % [nbr] Number (ordinal) (defined in psd.sty for number distributions) % \newcommand{\s}{s} % [s] Time (seconds) conflicts with TIPA? \newcommand{\ngs}{\AA} % [A] Length (Ångströms) \newcommand{\Bxs}{B\,s$^{-1}$} % [B s-1] Connection speed (bytes per second) \newcommand{\F}{F} % [F] Temperature (Fahrenheit) \newcommand{\GBxCPU}{GB\,CPU$^{-1}$} % [GB CPU-1] Computational speed (gigabytes per CPU) \newcommand{\GB}{GB} % [GB] Storage capacity (gigabytes) \newcommand{\GCCNxcmC}{GCCN\,cm$^{-3}$} \newcommand{\GHz}{GHz} % [GHz] Frequency (gigahertz) \newcommand{\GTxyr}{GT\,yr$^{-1}$} % [GT yr-1] Mass emission rate (gigatons per year) \newcommand{\Gbxs}{Gb\,s$^{-1}$} % [Gb s-1] Network bandwidth (gigabits per second) \newcommand{\Gb}{Gb} % [Gb] Storage capacity (gigabits) \newcommand{\Ghz}{GHz} % [GHz] Frequency (gigahertz) \newcommand{\GtCxyr}{GtC\,yr$^{-1}$} % [GtC yr-1] Mass emission rate (gigatons C per year) \newcommand{\GtC}{GtC} % [GtC] Mass (gigatons C) \newcommand{\Hz}{Hz} % [Hz] Frequency (Hertz) \newcommand{\J}{J} % [J] Energy \newcommand{\K}{K} % [K] Kelvin (K is also Potassium) \newcommand{\MBxs}{MB\,s$^{-1}$} % [MB s-1] Connection speed (megabytes per second) \newcommand{\MB}{MB} % [MB] Storage capacity (megabytes) \newcommand{\MHz}{MHz} \newcommand{\MSxatm}{M$^{2}$\,atm$^{-1}$} % [M2 atm-1] Molality (molar square per atmosphere) \newcommand{\MS}{M$^{2}$} % [M2] Equilibrium constant (molar square) \newcommand{\MTxyr}{MT\,yr$^{-1}$} % [MT yr-1] Mass emission rate (megatons per year) \newcommand{\MW}{MW} % [MW] Power (megawatts) \newcommand{\Mbxs}{Mb\,s$^{-1}$} % [Mb s-1] Connection speed (Megabits per second) \newcommand{\Mev}{MeV} \newcommand{\Mhz}{MHz} \newcommand{\MtC}{Mt\,C} \newcommand{\MtS}{Mt\,S} \newcommand{\MwxkOkmS}{MW\,K$^{-8}$\,km$^{-2}$} % [MW K-8 km-2] Fire Radiant Power factor \newcommand{\Mxatm}{M\,atm$^{-1}$} \newcommand{\Mya}{Mya} % [Mya] Time (millions of years ago) \newcommand{\NmSxkg}{N\,m$^{2}$\,kg$^{-1}$} \newcommand{\NxmS}{N\,m$^{-2}$} \newcommand{\Nxm}{N\,m$^{-1}$} \newcommand{\PBxyr}{PB\,yr$^{-1}$} % [PB yr-1] Storage capacity change (petabytes per year) \newcommand{\PB}{PB} % [PB] Storage capacity (petabytes) \newcommand{\Pa}{Pa} % [Pa] Pressure (Pascals) \newcommand{\TBxyr}{TB\,yr$^{-1}$} % [TB yr-1] Storage capacity change (terabytes per year) \newcommand{\TB}{TB} % [TB] Storage capacity (terabytes) \newcommand{\TgBCxyr}{Tg\,BC\,yr$^{-1}$} \newcommand{\TgBC}{Tg\,BC} \newcommand{\TgCxyr}{Tg\,C\,yr$^{-1}$} \newcommand{\TgC}{Tg\,C} \newcommand{\TgNxyr}{Tg\,N\,yr$^{-1}$} \newcommand{\TgSxyr}{Tg\,S\,yr$^{-1}$} \newcommand{\TgS}{Tg\,S} \newcommand{\Tgxyr}{Tg\,yr$^{-1}$} \newcommand{\Thz}{THz} \newcommand{\atmmSxs}{atm\,m$^{2}$\,s$^{-1}$} % [# m-2 s-1] Atomic number flux \newcommand{\atm}{atm} % [atm] Atmospheres \newcommand{\au}{AU} % [AU] Distance (Astronomical units) \newcommand{\axcmSs}{atom\,cm$^{-2}$\,s$^{-1}$} % [# cm-2 s-1] Atomic number flux \newcommand{\axmSs}{atom\,m$^{-2}$\,s$^{-1}$} % [# m-2 s-1] Atomic number flux \newcommand{\axmS}{A\,m$^{-2}$} % [A m-2] Current density (ampere per meter squared) \newcommand{\axm}{A\,m$^{-1}$} % [A m-1] Magnetic field (ampere-turn per meter) and Magnetization (ampere per meter) \newcommand{\axv}{A\,V$^{-1}$} % [A V-1] Conductivity (Amperes per Volt = Siemens) \newcommand{\bit}{b} % [b] Bits \newcommand{\bxs}{b\,s$^{-1}$} % [b s-1] Network bandwidth (bits per second) \newcommand{\bytexs}{byte\,s$^{-1}$} % [byte s-1] Connection speed (bytes per second) \newcommand{\bytxnbr}{B\,\#$^{-1}$} % [B #-1] Bytes per element \newcommand{\byt}{B} % [B] Bytes \newcommand{\cmFxmlcS}{cm$^{5}$\,molecule$^{-2}$} \newcommand{\cmSxg}{cm$^{2}$\,g$^{-1}$} % [cm2 g-1] Specific surface area \newcommand{\cmSxmlc}{cm$^{2}$\,molecule$^{-1}$} \newcommand{\cmSxs}{cm$^{2}$\,s$^{-1}$} \newcommand{\cmS}{cm$^{2}$} % [cm2] Area (square centimeters) \newcommand{\cmxatm}{cm\,atm$^{-1}$} \newcommand{\cmxa}{cm\,a$^{-1}$} \newcommand{\cmxday}{cm\,day$^{-1}$} \newcommand{\cmxm}{cm\,m$^{-1}$} \newcommand{\cmxs}{cm\,s$^{-1}$} \newcommand{\cmxyr}{cm\,yr$^{-1}$} % [cm yr-1] \newcommand{\cm}{cm} % [cm] Length (centimeters) \newcommand{\cxkm}{C\,km$^{-1}$} % [C km-1] Lapse rate (Celsius per kilometer) \newcommand{\cxmC}{C\,m$^{-3}$} % [C m-3] Charge density (Coulomb per cubic meter) \newcommand{\cxmS}{C\,m$^{-2}$} % [C m-2] Electric displacement (Coulomb per square meter) or Electric polarization (Coulomb-meters per cubic meter) \newcommand{\cxvm}{C\,V$^{-1}$\,m$^{-1}$} % [C V-1 m-1] Susceptibility (Coulomb per Volt-meter) \newcommand{\dayxmth}{day\,mo$^{-1}$} % [day mth-1] Incidence (days per month) \newcommand{\dgrcxkm}{\ensuremath{^{\circ}\mbox{C\,km}^{-1}}} % [C km-1] Temperature lapse rate (degrees celsius per kilometer) \newcommand{\dgrc}{\ensuremath{^{\circ}\mbox{C}}} % [C] Degrees Celsius \newcommand{\dgre}{\ensuremath{^{\circ}\mbox{E}}} % [dgr] Degrees East \newcommand{\dgrf}{\ensuremath{^{\circ}\mbox{F}}} % [F] Degrees Fahrenheit \newcommand{\dgrkxkm}{\ensuremath{\mbox{K\,km}^{-1}}} % [K km-1] Temperature lapse rate (Kelvin per kilometer) \newcommand{\dgrk}{\ensuremath{\mbox{K}}} % [K] Kelvin (no degree symbol) \newcommand{\dgrn}{\ensuremath{^{\circ}\mbox{N}}} % [dgr] Degrees North \newcommand{\dgrs}{\ensuremath{^{\circ}\mbox{S}}} % [dgr] Degrees South \newcommand{\dgrw}{\ensuremath{^{\circ}\mbox{W}}} % [dgr] Degrees West \newcommand{\dgr}{\ensuremath{^{\circ}}} % [dgr] Degrees \newcommand{\dxcmS}{dyn\,cm$^{-2}$} % [dyn cm-2] Momentum flux (dynes per square centimeter) \newcommand{\dxcm}{dyn\,cm$^{-1}$} \newcommand{\excmS}{erg\,cm$^{-2}$} \newcommand{\fmolxmSs}{fmol\,m$^{-2}$\,s$^{-1}$} % [fmol m-2 s-1] Flux (femptomoles per square meter per second) \newcommand{\frqmSxmol}{Hz\,m$^{2}$\,mol$^{-1}$} \newcommand{\ft}{ft} % [ft] Length \newcommand{\fxm}{F\,m$^{-1}$} % [F m-1] Permittivity (Farad per meter) \newcommand{\gCOdxmSs}{g\,\COd\,m$^{-2}$\,s$^{-1}$} \newcommand{\gCxmS}{g\,C\,m$^{-2}$} \newcommand{\gxcmC}{g\,cm$^{-3}$} \newcommand{\gxcmSka}{g\,cm$^{-2}$\,ka$^{-1}$} \newcommand{\gxkgK}{g\,kg$^{-1}$\,K$^{-1}$} \newcommand{\gxkgkm}{g\,kg$^{-1}$\,km$^{-1}$} \newcommand{\gxkg}{g\,kg$^{-1}$} \newcommand{\gxl}{g\,$\ell^{-1}$} \newcommand{\gxmC}{g\,m$^{-3}$} \newcommand{\gxmSgcmt}{g\,m$^{-2}$\,(20\,min)$^{-1}$} \newcommand{\gxmShr}{g\,m$^{-2}$\,hr$^{-1}$} \newcommand{\gxmSs}{g\,m$^{-2}$\,s$^{-1}$} \newcommand{\gxmSyr}{g\,m$^{-2}$\,yr$^{-1}$} \newcommand{\gxmS}{g\,m$^{-2}$} \newcommand{\gxmol}{g\,mol$^{-1}$} \newcommand{\hpa}{hPa} % [hPa] Pressure \newcommand{\hr}{hr} \newcommand{\hxm}{H\,m$^{-1}$} % [H m-1] Permeability (Henry per meter) \newcommand{\hzk}{Hz\,K} \newcommand{\hzxk}{Hz\,K$^{-1}$} \newcommand{\hz}{Hz} \newcommand{\inches}{in.} % [in] Length (inches) (\in is a math mode set symbol) \newcommand{\inxyr}{in.\,yr$^{-1}$} % [in yr-1] \newcommand{\js}{J\,s} \newcommand{\jxK}{J\,K$^{-1}$} \newcommand{\jxMev}{J\,MeV$^{-1}$} \newcommand{\jxgC}{J\,(g\,C)$^{-1}$} % [J (g C)-1] Energy per carbon (joules per gram carbon) \newcommand{\jxkgK}{J\,kg$^{-1}$\,K$^{-1}$} \newcommand{\jxkgs}{J\,kg$^{-1}$\,s$^{-1}$} \newcommand{\jxkg}{J\,kg$^{-1}$} \newcommand{\jxmCK}{J\,m$^{-3}$\,K$^{-1}$} \newcommand{\jxmChz}{J\,m$^{-3}$\,Hz$^{-1}$} \newcommand{\jxmCm}{J\,m$^{-3}$\,m$^{-1}$} \newcommand{\jxmCs}{J\,m$^{-3}$\,s$^{-1}$} \newcommand{\jxmC}{J\,m$^{-3}$} \newcommand{\jxmSmon}{J\,m$^{-2}$\,mon$^{-1}$} \newcommand{\jxmSshz}{J\,m$^{-2}$\,s$^{-1}$\,Hz$^{-1}$} \newcommand{\jxmSssrm}{J\,m$^{-2}$\,s$^{-1}$\,sr$^{-1}$\,m$^{-1}$} \newcommand{\jxmSssrwvn}{J\,m$^{-2}$\,s$^{-1}$\,sr$^{-1}$\,(cm$^{-1}$)$^{-1}$} \newcommand{\jxmS}{J\,m$^{-2}$} \newcommand{\jxmks}{J\,m$^{-1}$\,K$^{-1}$\,s$^{-1}$} \newcommand{\jxmolK}{J\,mol$^{-1}$\,K$^{-1}$} \newcommand{\jxm}{J\,m$^{-1}$} \newcommand{\jxs}{J\,s$^{-1}$} \newcommand{\kBxs}{kB\,s$^{-1}$} % [kB s-1] Connection speed (kilobytes per second) \newcommand{\kB}{kB} % [kB] Storage capacity (kilobytes) \newcommand{\ka}{ka} % [yr] Millenia \newcommand{\kbxs}{kb\,s$^{-1}$} % [kb s-1] Connection speed (kilobits per second) \newcommand{\kcalxmol}{kcal\,mol$^{-1}$} \newcommand{\kgCxmSs}{kg\,C\,m$^{-2}$\,s$^{-1}$} \newcommand{\kgSxmF}{kg$^{2}$\,m$^{-5}$} \newcommand{\kgSxmSix}{kg$^{2}$\,m$^{-6}$} \newcommand{\kgsSxmF}{kg\,s$^{2}$\,m$^{-5}$} \newcommand{\kgxcmS}{kg\,cm$^{-2}$} \newcommand{\kgxhayr}{kg\,ha$^{-1}$\,yr$^{-1}$} \newcommand{\kgxkgs}{kg\,kg$^{-1}$\,s$^{-1}$} \newcommand{\kgxkg}{kg\,kg$^{-1}$} \newcommand{\kgxkmSyr}{kg\,km$^{-2}$\,yr$^{-1}$} \newcommand{\kgxmCk}{kg\,m$^{-3}$\,K$^{-1}$} \newcommand{\kgxmCm}{kg\,m$^{-3}$\,m$^{-1}$} % [kg m-3 m-1] Mass distribution \newcommand{\kgxmCs}{kg\,m$^{-3}$\,s$^{-1}$} \newcommand{\kgxmC}{kg\,m$^{-3}$} \newcommand{\kgxmSsk}{kg\,m$^{-2}$\,s$^{-1}$\,K$^{-1}$} \newcommand{\kgxmSsm}{kg\,m$^{-2}$\,s$^{-1}$\,m$^{-1}$} \newcommand{\kgxmSs}{kg\,m$^{-2}$\,s$^{-1}$} \newcommand{\kgxmS}{kg\,m$^{-2}$} \newcommand{\kgxmi}{kg\,mi$^{-1}$} % [kg mi-1] Mass per distance \newcommand{\kgxmol}{kg\,mol$^{-1}$} \newcommand{\kgxmsS}{kg\,m$^{-1}$\,s$^{-2}$} \newcommand{\kgxms}{kg\,m$^{-1}$\,s$^{-1}$} \newcommand{\kgxnbr}{kg\,\#$^{-1}$} \newcommand{\kgxs}{kg\,s$^{-1}$} \newcommand{\kg}{kg} \newcommand{\kmC}{km$^{3}$} \newcommand{\kmS}{km$^{2}$} \newcommand{\kmxhr}{km\,hr$^{-1}$} \newcommand{\kmxs}{K\,m\,s$^{-1}$} % [K m s-1] Kinematic heat flux \newcommand{\km}{km} \newcommand{\ktxyr}{kt\,yr$^{-1}$} % [kt yr-1] Flux (kilotons per year) \newcommand{\kxday}{K\,day$^{-1}$} \newcommand{\kxd}{K\,d$^{-1}$} \newcommand{\kxhr}{K\,hr$^{-1}$} \newcommand{\kxkm}{K\,km$^{-1}$} \newcommand{\kxm}{K\,m$^{-1}$} \newcommand{\kxs}{K\,s$^{-1}$} \newcommand{\kxwxmS}{K\,(W\,m$^{-2}$)$^{-1}$} \newcommand{\kxyr}{K\,yr$^{-1}$} \newcommand{\kya}{kya} % [yr] Millenia ago \newcommand{\lbxinS}{lb\,in$^{-2}$} \newcommand{\ltr}{\ensuremath{\mathrm{L}}} % [L] Liter (L is taken by something else) fxm: Should be \ell? \newcommand{\mCxkgsS}{m$^{3}$\,kg$^{-1}$\,s$^{-2}$} \newcommand{\mCxkg}{m$^{3}$\,kg$^{-1}$} \newcommand{\mCxmCm}{m$^{3}$\,m$^{-3}$\,m$^{-1}$} % [m3 m-2] Volume distribution \newcommand{\mCxmC}{m$^{3}$\,m$^{-3}$} \newcommand{\mCxmSm}{m$^{3}$\,m$^{-2}$\,m$^{-1}$} % [m3 m-2] Volume distribution, columnar \newcommand{\mCxmSsm}{m$^{3}$\,m$^{-2}$\,s$^{-1}$\,m$^{-1}$} \newcommand{\mCxmSs}{m$^{3}$\,m$^{-2}$\,s$^{-1}$} \newcommand{\mCxmS}{m$^{3}$\,m$^{-2}$} % [m3 m-2] Volume path or logartithmic columnar volume distribution or, confusingly, volume distribution in "condensed" units \newcommand{\mCxmm}{m$^{3}$\,mm$^{-1}$} \newcommand{\mCxmol}{m$^{3}$\,mol$^{-1}$} \newcommand{\mCxnbr}{m$^{3}$\,\#$^{-1}$} \newcommand{\mCxs}{m$^{3}$\,s$^{-1}$} \newcommand{\mC}{m$^{3}$} \newcommand{\mSs}{m$^{2}$\,s} % [m2 s-1] Diffusivity \newcommand{\mSxg}{m$^{2}$\,g$^{-1}$} % [m2 g-1] Specific surface area \newcommand{\mSxkgxfrq}{m$^{2}$\,kg$^{-1}$\,s$^{-1}$} \newcommand{\mSxkgxhz}{m$^{2}$\,kg$^{-1}$\,Hz} \newcommand{\mSxkgxm}{m$^{2}$\,kg$^{-1}$\,m} \newcommand{\mSxkgxwvn}{m$^{2}$\,kg$^{-1}$\,cm$^{-1}$} \newcommand{\mSxkgxxwvn}{m$^{2}$\,kg$^{-1}$\,(cm$^{-1}$)$^{-1}$} \newcommand{\mSxkg}{m$^{2}$\,kg$^{-1}$} % [m2 kg-1] Specific surface area \newcommand{\mSxmC}{m$^{2}$\,m$^{-3}$} \newcommand{\mSxmS}{m$^{2}$\,m$^{-2}$} \newcommand{\mSxmlc}{m$^{2}$\,molecule$^{-1}$} \newcommand{\mSxmol}{m$^{2}$\,mol$^{-1}$} \newcommand{\mSxnbr}{m$^{2}$\,\#$^{-1}$} \newcommand{\mSxsS}{m$^{2}$\,s$^{-2}$} \newcommand{\mSxs}{m$^{2}$\,s$^{-1}$} \newcommand{\mS}{m$^{2}$} \newcommand{\mbxday}{mm\,d$^{-1}$} \newcommand{\mbxh}{mb\,hr$^{-1}$} \newcommand{\mbxkm}{mb\,km$^{-1}$} \newcommand{\mbxs}{mb\,s$^{-1}$} \newcommand{\mb}{mb} \newcommand{\mgxcmSka}{mg\,cm$^{-2}$\,ka$^{-1}$} \newcommand{\mgxkg}{mg\,kg$^{-1}$} \newcommand{\mgxl}{mg\,l$^{-1}$} % [mg l-1] Mass concentration \newcommand{\mgxmCum}{mg\,m$^{-3}\,\upmu$m$^{-1}$} \newcommand{\mgxmC}{mg\,m$^{-3}$} \newcommand{\mgxmS}{mg\,m$^{-2}$} \newcommand{\mg}{mg} \newcommand{\mhoxm}{mho\,x$^{-1}$} % [mho m-1] Conductivity \newcommand{\mixhr}{mi.\,hr$^{-1}$} \newcommand{\mlcSxcmF}{molecule$^{2}$\,cm$^{-5}$} \newcommand{\mlcSxcmS}{molecule$^{2}$\,cm$^{-6}$} \newcommand{\mlcSxmF}{molecule$^{2}$\,m$^{-5}$} \newcommand{\mlcSxmSix}{molecule$^{2}$\,m$^{-6}$} \newcommand{\mlcxcmC}{molecule\,cm$^{-3}$} \newcommand{\mlcxcmS}{molecule\,cm$^{-2}$} \newcommand{\mlcxmC}{molecule\,m$^{-3}$} \newcommand{\mlcxmSs}{molecule\,m$^{-2}$\,s$^{-1}$} \newcommand{\mlcxmS}{molecule\,m$^{-2}$} \newcommand{\mlcxmlc}{molecule\,molecule$^{-1}$} \newcommand{\mlcxmol}{molecule\,mol$^{-1}$} \newcommand{\mlcxs}{molecule\,s$^{-1}$} \newcommand{\mmolxmC}{mmol\,m$^{-3}$} % [mmol m-3] Concentration (millimoles per cubic meter) (same as [nmol l-1]) \newcommand{\mmxday}{mm\,d$^{-1}$} % [mm day-1] Rain rate (millimeters per day) \newcommand{\mmxhr}{mm\,hr$^{-1}$} \newcommand{\mmxmth}{mm\,mo$^{-1}$} \newcommand{\mm}{mm} % [mm] Length (millimeters) \newcommand{\molSxlS}{mol$^{2}$\,L$^{-2}$} \newcommand{\molxatm}{mol\,atm$^{-1}$} \newcommand{\molxkg}{mol\,kg$^{-1}$} \newcommand{\molxlatm}{mol\,L$^{-1}$\,atm$^{-1}$} \newcommand{\molxl}{mol\,L$^{-1}$} % [mol l-1 = M] Concentration (moles per liter) \newcommand{\molxmCp}{mol\,m$^{-3}$\,Pa$^{-1}$} \newcommand{\molxmC}{mol\,m$^{-3}$} \newcommand{\molxmol}{mol\,mol$^{-1}$} \newcommand{\mol}{mol} \newcommand{\mos}{mos} % [mo] Time (months) \newcommand{\mpg}{mi\,gal$^{-1}$} % [mi gal-1] Distance per volume \newcommand{\mph}{mi\,hr$^{-1}$} % [mi hr-1] Speed \newcommand{\mtr}{m} % [m] Length (meters) \newcommand{\mxmC}{m\,m$^{-3}$} \newcommand{\mxnbr}{m\,\#$^{-1}$} \newcommand{\mxsS}{m\,s$^{-2}$} \newcommand{\mxsm}{m\,s$^{-1}$\,m$^{-1}$} \newcommand{\mxs}{m\,s$^{-1}$} \newcommand{\m}{m} % [m] Length (meters) \newcommand{\nbrxcmC}{\#\,cm$^{-3}$} \newcommand{\nbrxkg}{\#\,kg$^{-1}$} \newcommand{\nbrxl}{\#\,L$^{-1}$} % [# l-1] Concentration (number per liter) \newcommand{\nbrxmCm}{\#\,m$^{-3}$\,m$^{-1}$} \newcommand{\nbrxmC}{\#\,m$^{-3}$} \newcommand{\nbrxmSsm}{\#\,m$^{-2}$\,s$^{-1}$\,m$^{-1}$} \newcommand{\nbrxmSsum}{\#\,m$^{-2}$\,s$^{-1}$\,$\upmu$m$^{-1}$} \newcommand{\nbrxmSs}{\#\,m$^{-2}$\,s$^{-1}$} \newcommand{\nbrxmS}{\#\,m$^{-2}$} \newcommand{\nbrxmol}{\#\,mol$^{-1}$} \newcommand{\nbrxmthfz}{\#\,mo$^{-1}$\,(100,000)$^{-1}$} % [# mth-1 100000-1] (five zeroes!) \newcommand{\nbrxmth}{\#\,mo$^{-1}$} \newcommand{\nbrxs}{\#\,s$^{-1}$} \newcommand{\nbrxum}{\#\,$\upmu$m$^{-1}$} % [# um-1] Quadrature density \newcommand{\nbrxyrfz}{\#\,yr$^{-1}$\,(100,000)$^{-1}$} % [# yr-1 100000-1] (five zeroes!) \newcommand{\nbrxyr}{\#\,yr$^{-1}$} \newcommand{\ngxg}{ng\,g$^{-1}$} \newcommand{\ngxkgs}{ng\,kg$^{-1}$\,s$^{-1}$} \newcommand{\ngxmC}{ng\,m$^{-3}$} \newcommand{\ngxmSs}{ng\,m$^{-2}$\,s$^{-1}$} \newcommand{\nmolxl}{nmol\,L$^{-1}$} % [nmol l-1 = nM] Concentration (nanomoles per liter) (same as [mmol m-3]) (same as nM) \newcommand{\nmolxmSd}{nmol\,m$^{-2}$\,d$^{-1}$} % [nmol m-2 d-1] Flux (nanomoles per square meter per day) \newcommand{\nm}{nm} \newcommand{\nxmCs}{\#\,m$^{-3}$\,s$^{-1}$} \newcommand{\nxmS}{N\,m$^{-2}$} % [N m-2] Momentum flux (Newtons per square meter) \newcommand{\nxm}{N\,m$^{-1}$} \newcommand{\ohm}{\Omega} % [ohm] Resistance \newcommand{\pctxdgrc}{\%\,\ensuremath{^{\circ}\mbox{C}^{-1}}} % [% C-1] \newcommand{\pctxdgrf}{\%\,\ensuremath{^{\circ}\mbox{F}^{-1}}} % [% F-1] \newcommand{\pctxyr}{\%\,yr$^{-1}$} % [% yr-1] \newcommand{\pgxkgs}{pg\,kg$^{-1}$\,s$^{-1}$} \newcommand{\pgxmSs}{pg\,m$^{-2}$\,s$^{-1}$} \newcommand{\phtxmSshz}{\#\,m$^{-2}$\,s$^{-1}$\,Hz$^{-1}$} \newcommand{\phtxmSsm}{\#\,m$^{-2}$\,s$^{-1}$\,m$^{-1}$} \newcommand{\phtxmSsum}{\#\,m$^{-2}$\,s$^{-1}$\,$\upmu$m$^{-1}$} \newcommand{\pmolxmSs}{pmol\,m$^{-2}$\,s$^{-1}$} % [pmol m-2 s-1] Flux (picomoles per square meter per second) \newcommand{\ppbm}{ppbm} % [ppbm] Mass mixing ratio \newcommand{\ppbv}{ppbv} % [ppbv] Volumetric mixing ratio \newcommand{\ppb}{ppb} % [ppb] Volumetric mixing ratio \newcommand{\ppmm}{ppmm} % [ppmm] Mass mixing ratio \newcommand{\ppmv}{ppmv} % [ppmv] Volumetric mixing ratio \newcommand{\ppm}{ppm} % [ppm] Volumetric mixing ratio \newcommand{\pxk}{Pa\,K$^{-1}$} \newcommand{\pxm}{Pa\,m$^{-1}$} \newcommand{\scn}{s} % [s] Time (seconds) \newcommand{\sr}{sr} \newcommand{\sxcm}{s\,cm$^{-1}$} \newcommand{\sxm}{s\,m$^{-1}$} \newcommand{\ugxg}{$\upmu$g\,g$^{-1}$} \newcommand{\ugxkgs}{$\upmu$g\,kg$^{-1}$\,s$^{-1}$} \newcommand{\ugxkg}{$\upmu$g\,kg$^{-1}$} \newcommand{\ugxmC}{$\upmu$g\,m$^{-3}$} \newcommand{\ugxmSd}{$\upmu$g\,m$^{-2}$\,d$^{-1}$} \newcommand{\ugxmSs}{$\upmu$g\,m$^{-2}$\,s$^{-1}$} \newcommand{\umCxumSum}{$\upmu$m$^{3}$\,$\upmu$m$^{-3}$\,$\upmu$m$^{-1}$} % [um3 um-3 um-1] Volume distribution \newcommand{\umCxumS}{$\upmu$m$^{3}$\,$\upmu$m$^{-2}$} % [um3 um-2] Volume path or logartithmic columnar volume distribution or, confusingly, volume distribution in "condensed" units \newcommand{\umk}{$\upmu$m\,K} \newcommand{\umolxl}{$\upmu$mol\,L$^{-1}$} % [umol l-1] Concentration (micromoles per liter) \newcommand{\umolxmC}{$\upmu$mol\,m$^{-3}$} % [umol m-3] Number concentration (micromoles per cubic meter) \newcommand{\umolxmS}{$\upmu$mol\,m$^{-2}$} % [umol m-2] Number path (micromoles per square meter) \newcommand{\um}{\ensuremath{\upmu\mbox{m}}} \newcommand{\vxm}{V\,m$^{-1}$} % [V m-1] Electric field (Volt per meter) \newcommand{\wvlmSxmol}{m\,m$^{2}$\,mol$^{-1}$} \newcommand{\wvncmSxmlcatm}{cm$^{-1}$\,cm$^{2}$\,molecule$^{-1}$\,atm$^{-1}$} \newcommand{\wvncmSxmlc}{cm$^{-1}$\,cm$^{2}$\,molecule$^{-1}$} \newcommand{\wvnmSxmlc}{cm$^{-1}$\,m$^{2}$\,molecule$^{-1}$} \newcommand{\wvnmSxmol}{cm$^{-1}$\,m$^{2}$\,mol$^{-1}$} \newcommand{\wvnxatm}{cm$^{-1}$\,atm$^{-1}$} \newcommand{\wxkg}{W\,kg$^{-1}$} \newcommand{\wxmChz}{W\,m$^{-3}$\,Hz$^{-1}$} \newcommand{\wxmC}{W\,m$^{-3}$} \newcommand{\wxmScm}{W\,m$^{-2}$\,(cm$^{-1}$)$^{-1}$} \newcommand{\wxmShzsr}{W\,m$^{-2}$\,Hz$^{-1}$\,sr$^{-1}$} \newcommand{\wxmShz}{W\,m$^{-2}$\,Hz$^{-1}$} \newcommand{\wxmSkQ}{W\,m$^{-2}$\,K$^{-4}$} \newcommand{\wxmSk}{W\,m$^{-2}$\,K$^{-1}$} \newcommand{\wxmSmsr}{W\,m$^{-2}$\,m$^{-1}$\,sr$^{-1}$} \newcommand{\wxmSm}{W\,m$^{-2}$\,m$^{-1}$} \newcommand{\wxmSnmsr}{W\,m$^{-2}$\,nm$^{-1}$\,sr$^{-1}$} \newcommand{\wxmSnm}{W\,m$^{-2}$\,nm$^{-1}$} \newcommand{\wxmSsrfrq}{W\,m$^{-2}$\,sr$^{-1}$\,Hz$^{-1}$} \newcommand{\wxmSsrhz}{W\,m$^{-2}$\,sr$^{-1}$\,Hz$^{-1}$} \newcommand{\wxmSsrm}{W\,m$^{-2}$\,sr$^{-1}$\,m$^{-1}$} \newcommand{\wxmSsrs}{W\,m$^{-2}$\,sr$^{-1}$\,s$^{-1}$} \newcommand{\wxmSsrwvn}{W\,m$^{-2}$\,sr$^{-1}$\,(cm$^{-1}$)$^{-1}$} \newcommand{\wxmSsrxcm}{W\,m$^{-2}$\,sr$^{-1}$\,(cm$^{-1}$)$^{-1}$} \newcommand{\wxmSsrxm}{W\,m$^{-2}$\,sr$^{-1}$\,(m$^{-1}$)$^{-1}$} \newcommand{\wxmSsr}{W\,m$^{-2}$\,sr$^{-1}$} \newcommand{\wxmSumsr}{W\,m$^{-2}$\,$\upmu$m$^{-1}$\,sr$^{-1}$} \newcommand{\wxmSum}{W\,m$^{-2}$\,$\upmu$m$^{-1}$} \newcommand{\wxmS}{W\,m$^{-2}$} % [W m-2] Heat flux (Watts per square meter) \newcommand{\wxmk}{W\,m$^{-1}$\,K$^{-1}$} \newcommand{\xTB}{TB$^{-1}$} % per Terabyte \newcommand{\xcmC}{cm$^{-3}$} \newcommand{\xcmS}{cm$^{-2}$} \newcommand{\xcm}{\ensuremath{\mbox{cm}^{-1}}} \newcommand{\xfrq}{(s$^{-1}$)$^{-1}$} \newcommand{\xft}{ft$^{-1}$} \newcommand{\xhz}{Hz$^{-1}$} \newcommand{\xkS}{K$^{-2}$} \newcommand{\xkm}{km$^{-1}$} \newcommand{\xk}{K$^{-1}$} \newcommand{\xmCs}{m$^{-3}$\,s$^{-1}$} \newcommand{\xmCum}{m$^{-3}\,\upmu$m} \newcommand{\xmC}{m$^{-3}$} \newcommand{\xmSs}{m$^{-2}$\,s$^{-1}$} \newcommand{\xmS}{m$^{-2}$} \newcommand{\xmm}{mm$^{-1}$} \newcommand{\xmthfz}{mo$^{-1}$\,(100,000)$^{-1}$} % [mth-1 100000-1] (five zeroes!) \newcommand{\xmth}{mo$^{-1}$} % [mth-1] \newcommand{\xm}{m$^{-1}$} \newcommand{\xnbr}{\#$^{-1}$} \newcommand{\xsS}{s$^{-2}$} \newcommand{\xshz}{s$^{-1}$\,Hz$^{-1}$} \newcommand{\xsm}{s$^{-1}$\,m$^{-1}$} \newcommand{\xsr}{sr$^{-1}$} \newcommand{\xs}{s$^{-1}$} \newcommand{\xum}{$\upmu$m$^{-1}$} % [um-1] Spectral density \newcommand{\xwvn}{(cm$^{-1}$)$^{-1}$} \newcommand{\xyrfz}{yr$^{-1}$\,(100,000)$^{-1}$} % [yr-1 100000-1] (five zeroes!) \newcommand{\xyr}{yr$^{-1}$} % [yr-1] \newcommand{\yrxwk}{yr\,wk$^{-1}$} \newcommand{\yr}{yr} % Derived units \newcommand{\dgrcxwxmS}{\ensuremath{^{\circ}\mbox{C\,(W\,m$^{-2}$)}^{-1}}} % [C (W m-2)-1] Temperature sensitivity to forcing (degrees celsius per Watt per square meter) graeme-a-stewart/cpluspluscoursetalk/C++Course.tex \documentclass[compress]{beamer} \usetheme{Warsaw} \useoutertheme{split} %\includeonly{python/python} \input{setup} \newboolean{onlybasics} \setboolean{onlybasics}{false} \begin{document} \showboxdepth=\maxdimen \showboxbreadth=\maxdimen \begin{frame} \titlepage \end{frame} \begin{frame} \frametitle{Foreword} \begin{block}{What this course is not} \begin{itemize} \item It is not for absolute beginners \item It is not for experts \item It is not complete at all (would need 3 weeks...) \begin{itemize} \item although is it already too long for the time we have \item \inserttotalframenumber{} slides, \insertpresentationendpage{} pages, 10s of exercises... \end{itemize} \end{itemize} \end{block} \begin{block}{How I see it} \begin{description} \item[Adaptative] pick what you want \item[Interactive] tell me what to skip/insist on \item[Practical] let's spend time on real code \end{description} \end{block} \begin{block}{Where to find latest version ?} \begin{itemize} \item pdf format at {\small \url{http://cern.ch/sponce/C++Course}} \item full sources at {\scriptsize \url{https://github.com/hsf-training/cpluspluscourse}} \end{itemize} \end{block} \end{frame} \begin{frame} \frametitle{More courses} \begin{block}{The HSF Software Training Center} A set of course modules on more software engineering aspects prepared from within the HEP community \begin{itemize} \item Unix shell \item Python \item Version control (git, gitlab, github) \item ... \end{itemize} {\small \url{https://hepsoftwarefoundation.org/training/curriculum.html}} \end{block} \end{frame} \begin{frame} \frametitle{Outline} \begin{multicols}{2} \tableofcontents[sectionstyle=show,subsectionstyle=hide] \end{multicols} \end{frame} \begin{frame} \frametitle{Detailed outline} %\vspace{-0.5cm} \begin{scriptsize} \begin{multicols}{3} \tableofcontents[sectionstyle=show,subsectionstyle=show] \end{multicols} \end{scriptsize} \end{frame} \include{introduction/introduction} \include{basicconcepts/basicconcepts} % basic version has subset of tools at this stage \ifthenelse{\boolean{onlybasics}}{ \section[Tool]{Useful tools} \input{tools/editors} \input{tools/vcs} \input{tools/formatting} \input{tools/compiling} \input{tools/debugging} }{} % basic version only keeps a subset of these 2 chapters \ifthenelse{\boolean{onlybasics}}{ \section[OO]{Object orientation (OO)} \input{objectorientation/objectsclasses} \input{objectorientation/inheritance} \input{objectorientation/constructors} \input{objectorientation/static} \input{objectorientation/allocations} \input{objectorientation/advancedoo} \input{objectorientation/operators} \input{objectorientation/functors} \section[More]{Core modern \cpp} \input{morelanguage/constness} \input{morelanguage/exceptions} \input{morelanguage/templates} \input{morelanguage/stl} \input{morelanguage/lambda} \input{morelanguage/raii} } { \include{objectorientation/objectorientation} \include{morelanguage/morelanguage} } % do not include these chapters in basic version \ifthenelse{\boolean{onlybasics}}{}{ \include{expert/expert} \include{tools/tools} \include{concurrency/concurrency} \include{python/python} } \begin{frame} \frametitle{This is the end} \begin{center} \Huge Questions ?\\ \vspace{.5cm} \tiny \href{https://github.com/hsf-training/cpluspluscourse}{https://github.com/hsf-training/cpluspluscourse}\\ \tiny \href{http://cern.ch/sponce/C++Course}{http://cern.ch/sponce/C++Course} \end{center} \end{frame} \end{document} \documentclass[]{article} \usepackage{simplemath/core} \usepackage{simplemath/assignment} % for math-inline and tab support in verbatim mode % Verbatim, PVerb \usepackage{examplep} \usepackage{fancyvrb} \newcommand{\explain}{\qquad &} \def\explained#1{#1 \qquad & \PVerb{#1}} \begin{document} \section{Mathe-Umgebung} \paragraph{Inline} Hallo $f = a$. \begin{Verbatim} Hallo $f = a$. \end{Verbatim} \paragraph{Mehrzeilig} \begin{Eq*} \sep f = g & y \\ \Implies \sep a = b + c & x \\ \end{Eq*} \begin{Verbatim} \begin{Eq*} \sep f = g & y \\ \Implies \sep a = b + c & x \\ \end{Eq*} \end{Verbatim} \paragraph{Mehrzeilig benannt} \begin{Eq} \sep f = a \\ \Implies \sep a = b \\ \end{Eq} \begin{Verbatim} \begin{Eq} \sep f = a \\ \Implies \sep a = b \\ \end{Eq} \end{Verbatim} \section{Symbole} \begin{row} \begin{col}{0.2} \begin{Eq*} \explained{\alpha} \\ \explained{\beta} \\ \explained{\chi} \\ \explained{\delta} \\ \explained{\epsilon} \\ \explained{\eta} \\ \explained{\gamma} \\ \explained{\iota} \\ \explained{\kappa} \\ \end{Eq*} \end{col} \begin{col}{0.2} \begin{Eq*} \explained{\lambda} \\ \explained{\mu} \\ \explained{\nu} \\ \explained{o} \\ \explained{\omega} \\ \explained{\phi} \\ \explained{\pi} \\ \explained{\psi} \\ \explained{\rho} \\ \end{Eq*} \end{col} \begin{col}{0.2} \begin{Eq*} \explained{\sigma} \\ \explained{\tau} \\ \explained{\theta} \\ \explained{\upsilon} \\ \explained{\xi} \\ \explained{\zeta} \\ \explained{\digamma} \\ \explained{\varepsilon} \\ \explained{\varkappa} \\ \end{Eq*} \end{col} \begin{col}{0.2} \begin{Eq*} \explained{\varphi} \\ \explained{\varpi} \\ \explained{\varrho} \\ \explained{\varsigma} \\ \explained{\vartheta} \\ \explained{\Delta} \\ \explained{\Gamma} \\ \explained{\Lambda} \\ \explained{\Omega} \\ \end{Eq*} \end{col} \begin{col}{0.2} \begin{Eq*} \explained{\Phi} \\ \explained{\Pi} \\ \explained{\Psi} \\ \explained{\Sigma} \\ \explained{\Theta} \\ \explained{\Upsilon} \\ \explained{\Xi} \\ \end{Eq*} \end{col} \end{row} \begin{row} \begin{col}{0.5} \paragraph{Meta-Logik} \begin{Eq*} \explained{\Implies} \\ \explained{\RImplies} \\ \explained{\Iff} \\ \end{Eq*} \end{col} \begin{col}{0.5} \paragraph{Universen} \begin{Eq*} \explained{\UR} \\ \explained{\UN} \\ \explained{\UZ} \\ \explained{\UQ} \\ \explained{\UC} \\ \explained{\UB} \end{Eq*} \end{col} \end{row} \begin{row} \begin{col}{0.5} \paragraph{Logic} \begin{Eq*} \explained{\Land_x x \land y} \\ \explained{\Lor_x x \lor y} \\ \explained{\lnot x} \\ \explained{x \limplies y} \\ \explained{x \lrightimplies y} \\ \explained{x \liff y} \\ \explained{x \lxor y} \\ \explained{\forall g: g} \\ \explained{\exists g: g} \\ \end{Eq*} \end{col} \begin{col}{0.5} \paragraph{Mengen} \begin{Eq*} \explained{\emptyset} \\ \explained{x \in A} \\ \explained{x \notin A} \\ \explained{\Setunion_x x \setunion y}\\ \explained{\Setintersect_x x \setintersect y} \\ \explained{a \subset b} \\ \explained{a \subseteq b} \\ \explained{a \subsetneq b} \\ \explained{\setsize{A}} \\ \explained{C = \{ a \in A \mid a \notin B \}} \\ \explained{\partial A} \\ \explained{\bar A} \\ \explained{A_n = \{ 1 \dots n \}} \\ \end{Eq*} \end{col} \end{row} \begin{row} \begin{col}{0.5} \paragraph{Functions} \begin{Eq*} \explained{x \to y} \\ \explained{x \mapsto y}\\ \explained{f \circ g} \\ \explained{f \ast g} \\ \explained{\hat{f}} \\ \end{Eq*} \end{col} \begin{col}{0.5} \paragraph{Vergleiche} \begin{Eq*} \explained{a = b} \\ \explained{a < b} \\ \explained{a > b} \\ \explained{a \leq b} \\ \explained{a \geq b} \\ \explained{a \neq b} \\ \explained{a \equiv b} \\ \explained{a \approx b} \\ \explained{a \sim b} \\ \end{Eq*} \end{col} \end{row} \begin{row} \begin{col}{0.5} \paragraph{Arithmetik} \begin{Eq*} \explained{\pm a} \\ \explained{\lfloor a \rfloor} \\ \explained{\lceil a \rceil} \\ \explained{\sqrt{a + b}} \\ \explained{\sqrt[3]{a + b}} \\ \explained{x \cdot y} \\ \explained{\sum_{x \in X} a + x} \\ \explained{\sum_{i = x}^y a + i} \\ \explained{\prod_{x \ in X} a + i} \\ \explained{\min(a, b)} \\ \explained{\max(a, b)} \\ \end{Eq*} \end{col} \begin{col}{0.5} \paragraph{Vectorräume} \begin{Eq*} \explained{x \times y} \\ \explained{\Vector{1 \\ 2 \\ 3}}\\ \explained{\begin{Matrix} 1 & 2 \\ 3 & 4 \end{Matrix}} \\ \explained{\begin{Matrix} 1 & \dots \\ \vdots & b \end{Matrix}} \\ \explained{\Det(x)} \\ \explained{A + B} \\ \explained{A * B} \\ \explained{A \oplus B} \\ \explained{A \otimes B} \\ \explained{A / B} \\ \explained{A^\perp} \\ \explained{\langle A \rangle} \\ \explained{\dim(A)} \end{Eq*} \end{col} \end{row} \begin{row} \begin{col}{0.5} \paragraph{Lina \& AZ} \begin{Eq*} \explained{a \mod b} \\ \explained{a \mid b} \\ \explained{a \nmid b} \\ \explained{a \parallel b} \\ \explained{a \perp b} \\ \explained{\ggT(x, y)} \\ \explained{\kgV(x, y)} \\ \explained{\big[ x \big] } \\ \explained{\neutral} \\ \explained{\normdevider} \\ \explained{\normdevidereq} \\ \explained{\rnormdevider} \\ \explained{\rnormdevidereq} \\ \explained{\lnormdevider} \\ \explained{\lnormdevidereq} \\ \end{Eq*} \end{col} \begin{col}{0.5} \paragraph{Ana} \begin{Eq*} \explained{\dd x} \\ \explained{\frac{\dd f}{\dd x}} \\ \explained{\frac{\partial f}{\partial x}} \\ \explained{\int x \dd x} \\ \explained{\int_0^\infty x \dd x} \\ \explained{\big[ x \big]_0^y} \\ \explained{\lim_{x \nearrow a} f(x)} \\ \explained{\lim_{x \searrow a} f(x)} \\ \explained{\lim_{x \to a} f(x)} \\ \explained{f^\prime} \\ \explained{f^{\prime\prime}} \\ \explained{\dot f} \\ \explained{\ddot f} \\ \explained{\nabla f} \end{Eq*} \end{col} \end{row} \section{Layout} \begin{Eq*} \explained{f(x) = \begin{cases} 1 & x = 0 \\ 0 & \text{sonst} \end{cases} } \\ \end{Eq*} \section{Weiteres} \def\bin{\newbinaryop{bin}} \def\fn{\newfunc{fn}} \paragraph{Einen eigenen Binär-Operator definieren} Am Anfang des Dokuments definieren: \begin{Eq*} \PVerb{\def\bin{\newbinaryop{bin}}} \end{Eq*} Dann kann dieser wie folgt genutzt werden: \begin{Eq*} \explained{a \bin b} \\ \end{Eq*} \paragraph{Einen eigenen Funktion definieren} Am Anfang des Dokuments definieren: \begin{Eq*} \PVerb{\def\fn{\newfunc{fn}}} \end{Eq*} Dann kann dieser wie folgt genutzt werden: \begin{Eq*} \explained{\fn(a, b)} \\ \end{Eq*} \end{document} nanxstats/rankv @article{dumouchel1999, title={Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system}, author={}, journal={The American Statistician}, volume={53}, number={3}, pages={177--190}, year={1999} } @article{evans2001, title={Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports}, author={ and and }, journal={Pharmacoepidemiology and drug safety}, volume={10}, number={6}, pages={483--486}, year={2001} } @article{van2002comparison, title={A comparison of measures of disproportionality for signal detection in spontaneous reporting systems for adverse drug reactions}, author={, and Leufkens, and Lindquist, , }, journal={Pharmacoepidemiology and drug safety}, volume={11}, number={1}, pages={3--10}, year={2002} } @article{bate1998, title={A Bayesian neural network method for adverse drug reaction signal generation}, author={ Lindquist, Edwards, Olsson, Orre, Roland and Lansner, Anders and , }, journal={European journal of clinical pharmacology}, volume={54}, number={4}, pages={315--321}, year={1998} } @article{noren2006, title={Extending the methods used to screen the WHO drug safety database towards analysis of complex associations and improved accuracy for rare events}, author={Nor{\'e}n, and and and }, journal={Statistics in medicine}, volume={25}, number={21}, pages={3740--3757}, year={2006} } @article{canida2017, title={openEBGM: An R Implementation of the Gamma-Poisson Shrinker Data Mining Model}, author={ and }, journal={The R Journal}, volume={9}, number={2}, pages={499--519}, year={2017} } @article{ahmed2010, title={False discovery rate estimation for frequentist pharmacovigilance signal detection methods}, author={ }, journal={Biometrics}, volume={66}, number={1}, pages={301--309}, year={2010} } @article{pihur2007, title={Weighted rank aggregation of cluster validation measures: a monte carlo cross-entropy approach}, author={ and and }, journal={Bioinformatics}, volume={23}, number={13}, pages={1607--1615}, year={2007} } chapters/07-usage.tex \chapter{Provoz} \label{ch:provoz} Následující kapitola shrne možnosti realizovaného rozšíření pro koordinaci IoT, popíše, jaké je použití pro provoz sítě Internetu věcí a poté ověří realizované prostředky pro koordinaci IoT na produkčním příkladu. \section{Nasazení a provoz systému}\label{sec:nasazení-a-provoz-systému} Pro běh systému s~realizovaným rozšířením bylo nutné zajistit následující prostředky: \begin{itemize} \item \textbf{Server} \\ Pro běh požadovaných komponent celého systému je požadován server, ke kterému se budou schopny uzly pomocí bezdrátové sítě připojit a bude schopen provozovat běhové prostředí Node.js -- v~použitém příkladu se bude jednat o~VPS s~operačním systémem Debian. \item \textbf{Běhové prostředí pro nástroj Node-RED} \\ Vzhledem k~ekosystému nástroje Node-RED byl zvolen nástroj PM2\footnote{\url{http://pm2.keymetrics.io/}}, který pro aplikace v~programovacím jazyce Javascript zajišťuje správu jejich běhu, tj. monitorování použitých systémových prostředků, jejich programové smyčky událostí či paralelní běh -- pro správce systému poskytuje možnost Node-RED spustit. \item \textbf{Broker MQTT} \\ Jako broker byla zvolena implementace s~názvem \emph{Eclipse Mosquitto}\footnote{\url{https://mosquitto.org/}} -- jedná se o~nástroj s~otevřeným zdrojovým kódem od společnosti Eclipse. Broker dle dokumentace podporuje všechny z~požadovaných vlastností (parametr zprávy \uv{retain}, tři úrovně QoS, zprávu poslední vůle atd.) a je dostupný jako aplikační balíček pro použitý operační systém. \item \textbf{Moduly ESP32} \\ Pro produkční použití byly použity moduly ESP32 s~deskou plošných spojů obsahující stabilizátor napětí, převodník UART-USB a lištu pinů umožňující zapojení do nepájivého pole -- jedná se o~moduly s~produktovým názvem \uv{ESP32-DevKitC}\footnote{\url{https://www.espressif.com/en/products/hardware/esp32-devkitc/overview}} přímo od firmy Espressif Systems. \end{itemize} Všechny výše uvedené prostředky jsem použil pro sestavení vlastního lokálního Internetu věcí, na kterém jsem následně prováděl experimenty a ověřoval funkci realizovaných prostředků. Postup spuštění tohoto systému byl následující: \begin{enumerate} \item \textbf{Instalace firmwaru na uzel} \\ Pro realizované programové vybavení pro uzly ESP32 existuje také instalátor v~podobě předpisu pro nástroj \texttt{make} -- instalace je složena ze tří hlavních kroků (více k~tomuto postupu je uvedeno v~příloze~\ref{ch:instalator}). Prvním krokem je zavedení samotného interpretu jazyka MicroPython, následuje instalace externích knihoven jakožto závislostí realizovaného firmwaru a následně samotný kód firmwaru. Po ověření funkce firmwaru je posledním krokem zapnutí automatického zavedení firmwaru při startu -- uzel je v~tu chvíli připraven pro provoz čistě přes internetové připojení. Výstupem instalace je, kromě pro provoz připraveného uzlu, unikátní identifikátor uzlu, který bude následně použit pro zacílení ze strany nástroje Node-RED. \item \textbf{Instalace Node-RED rozšíření} \\ Pro instalaci vlastního rozšíření do tohoto nástroje je nutné dodržet požadavky popsané v~\ref{sec:node-red-rozsireni} -- realizované rozšíření všechny tyto formální požadavky splňuje, je tedy možné jej nainstalovat do jmenného prostoru knihoven nástroje Node-RED, například pomocí nástroje \texttt{npm}. Ve chvíli startu si tento nástroj knihovny poskytující bloky zaregistruje do galerie bloků, které následně nabízí uživateli. \item \textbf{Příprava sítě} \\ Dalším krokem je příprava samotné sítě v~nástroji Node-RED, jedná se postupně o~výběr bloků pro síť (senzory, rozhodovací prvky, výstupní prvky), jejich nastavení (použité piny na uzlu, podmínky, rozmístění prvků v~GUI) včetně konfiguračních bloků (broker MQTT, identifikace uzlů získané při instalaci) a na závěr jejich propojení do sítě pomocí datových spojů. Příklad sítě, která s~pomocí senzoru DHT měří teplotu a vlhkost místnosti, kterou exportuje do uživatelského rozhraní, je vyobrazen na obrázku~\ref{fig:node-red-production-1} Proces přípravy sítě je zakončen jejím nasazením -- editor zašle konfiguraci do běhové části nástroje, jenž provede reinicializaci sítě a jejích bloků. \end{enumerate} \section{Scénáře užití}\label{sec:scenare-uziti} Pro ověření funkce systému byly použity dva případy užití, které budou představeny v~následujících kapitolách -- vždy textový popis případu užití, zobrazení implementace sítě v~nástroji Node-RED a ukázka ze sestaveného uživatelského rozhraní. \subsection{Měření prostředí v~místnosti s~výstupem do uživatelského rozhraní}\label{subsec:scenar-1} Síť prvního scénáře obsahuje vstupní blok reprezentující senzor typu DHT22 měřící teplotu a vlhkost okolního prostředí. Zprávy z~tohoto senzoru odečítá nasazená aplikace na uzlu s~intervalem 30 sekund a skrz firmware a broker MQTT je distribuuje do sítě nástroje Node-RED, kde jsou dále zpracovávány -- dojde k~vyextrahování hodnoty teploty a vlhkosti ze zprávy, zobrazení aktuální hodnoty v~textovém výstupu uživatelského rozhraní a promítnutí časového aritmetického průměru hodnoty za 15 minut do grafu. Implementace sítě pro tento scénář je zobrazena v~obrázku~\ref{fig:node-red-production-1}, odpovídající část uživatelského rozhraní je poté vyobrazena na obrázku~\ref{fig:node-red-production-1-ui}. \begin{figure} \centering \includegraphics[width=\textwidth]{figures/fis-flow-1.png} \caption{Implementace sítě pro scénář \textit{Měření prostředí v~místnosti s~výstupem do uživatelského rozhraní} -- vlevo se nachází vstupní blok reprezentující uzel s~připojeným senzorem typu DHT22. Výstup tohoto bloku dále směřuje do bloků exportujících konkrétní hodnoty ze zprávy (vstupem je datový typ \ic{Object}, výstupem \ic{float}), ze kterých již dochází k~zobrazení a vyhodnocení časového aritmetického průměru pro graf -- pro snažší ladění je zde přidán blok typu \texttt{debug}, který exportuje aktuální teplotu do ladícího panelu nástroje. Pro signalizaci času poslední aktualizace slouží soustava bloků pro nastavení aktuálního času při přijetí zprávy a jeho konverze na podobu do uživatelského rozhraní.} \label{fig:node-red-production-1} \end{figure} \begin{figure} \centering \includegraphics[width=\textwidth]{figures/fis-flow-1-ui.png} \caption{Ukázka z~uživatelského rozhraní pro scénář \textit{Měření prostředí v~místnosti s~výstupem do uživatelského rozhraní} -- kromě naposledy změřených hodnot se zde nachází grafy průběhů měřených hodnot a čas poslední aktualizace.} \label{fig:node-red-production-1-ui} \end{figure} \subsection{Zobrazení aktuálního času a teploty na displeji}\label{subsec:scenar-2} Druhý testovací scénář obsahuje displej sestavený z~LED diod technologie Neopixel, na který se zobrazuje aktuální čas a aktuální teplota v~měřené místnosti z~prvního scénáře -- čas je vždy zobrazen po dobu $X$ a následně je po dobu $Y$ zobrazena teplota. Tyto hodnoty jsou uživatelem stanoveny z~uživatelského rozhraní -- v~implementaci sítě tedy běží čítač, který na základě předaných vstupů definuje výstup do displeje. Speciálním případem je zadání statického textu do rozhraní, tento text je poté zobrazen bez vlivů čítače -- ve všech případech lze použít vstup pro výběr požadované barvy displeje. Ukázka z~implementace této sítě se nechází na obrázku~\ref{fig:node-red-production-2}, sestavené uživatelské rozhraní poté na obrázku~\ref{fig:node-red-production-2-ui}. \begin{figure} \centering \includegraphics[width=.7\textwidth]{figures/fis-flow-2.png} \caption{Implementace sítě pro scénář \textit{Zobrazení aktuálního času a teploty na displeji} -- funkce je založena na dvou částech. Spodní část je určena k~ukládání aktuální stavu nastavení, tedy testického textu, intervalů zobrazení a pomocí tunelového spojení i aktuální teploty -- využito je k~tomu vlastnosti nástroje umožňující uložení informace v~rámci jedné sítě. Horní část poté obsahuje časovač, který invokuje vlastní implementaci rozhodovací logiky, jejíž výstupem je zpráva směrovaná do blok displeje (a pro snažší ladění i do postranního panelu).} \label{fig:node-red-production-2} \end{figure} \begin{figure} \centering \includegraphics[width=.5\textwidth]{figures/fis-flow-2-ui.png} \caption{Ukázka z~uživatelského rozhraní pro scénář \textit{Měření prostředí v~místnosti s~výstupem do uživatelského rozhraní} -- kromě naposled změřených hodnot se zde nachází grafy průběhů měřených hodnot a čas poslední aktualizace.} \label{fig:node-red-production-2-ui} \end{figure} \section{Vyhodnocení provozu scénářů}\label{sec:vyhodnocení} Oba scénáře byly po dobu sedmi dnů provozovány paralelně na dvou samostatných uzlech. Během této doby se realizovanému \textbf{systému plně dařilo vykonávat zadanou funkci} -- firmware uzlu se vypořádal s~lokálně slabým a nestabilním signálem bezdrátové sítě pomocí procesu znovupřipojení k~internetu a brokeru. Výsledkem bylo obnovené spojení, díky kterému obdržel uzel od brokeru poslední konfigurační data a data pro aplikace. Z~těchto dat zreprodukoval vnitřní stav a výstupy či vstupy -- zobrazený text v~případě displeje či naplánované měření v~případě senzoru. Toto chování bylo autorem zjištěno pomocí manuálního připojení se na konzoli uzlů přímo v~místě. V~této době bylo také odpozorováno několik automatických restartů uzlů, příčin bylo postupně několik -- nedostatek paměti při zavedení většího množství aplikací na uzel během testování, zvýšený výkonový odběr uzlu v~případě displeje bez externího zdroje a tím i zapříčiněný pád operačního systému pro nedostatek energie v~procesoru nebo stav, kdy dojde k~uzavření spojení MQTT ze strany brokeru, a tím i automatickému restartu uzlu. \textit{Oba scénáře používaly každý samostatně svůj vlastní fyzický uzel -- pro navržený systém by ovšem nebyl problém provozovat oba scénáře na jednom uzlu vedle sebe, jak bylo autorem ověřeno během experimentů.}\subsubsection{Camera Configuration} \label{subsubsec:camera_configuration} The camera configuration should be the same during data collection and deployment. For this reason, the optimal parameters have been carefully selected. Table \ref{tab:config} lists the important configurations and their respective values. \begin{table}[hb] \caption{Camera configuration} \label{tab:config} \centering \begin{tabular}{lll} \toprule \textbf{Parameter} & \textbf{Value} & \textbf{Description} \\ \midrule BalanceWhiteAuto & \texttt{Once} & Perform the white balance once during setup \\ Gain & 4 & For the max. brightness, the max. gain is used \\ ExposureTime & \SI{250}{\micro\second} & A short exposure time of \SI{250}{\micro\second} is used \\ AcquisitionFrameRateEnable & \texttt{true} & Enable fixed frame rate \\ AcquisitionFrameRate & \SI{200}{fps} & Set fixed frame rate to \SI{200}{fps} \\ TriggerMode & \texttt{Off} & Use no trigger (free run) \\ PixelFormat & \texttt{BayerRG8} & Use the raw Baumer \texttt{BayerRG8} pixel format \\ \bottomrule \end{tabular} \end{table} Molinatoxx/starter-academic @article{azari2018key, author = { Sallouha, , Rajendran, Vinogradov, }, journal = {IEEE Communications Magazine}, number = {1}, pages = {51--57}, publisher = {IEEE}, title = {Key technologies and system trade-offs for detection and localization of amateur drones}, volume = {56}, year = {2018} } 1-10 @article{2017yang1, abstract = {The research and development of low dimensional materials have accelerated at an enormous pace. Since the discovery of carbon nanotubes (CNTs) by Iijima in 1991, there have been extensive research efforts on the synthesis, physics, electronics, chemistry and applications of one-dimensional (1D) materials, including nanotubes, nanowires, nanorods, nanofibers, and nanobelts. Recent advances in the modeling of phenomena during nanofabrication along with new methods and mechanics of controllable synthesis of 1D materials have paved the road for applications with increasing social impact, especially in energy generation and storage, bio-detection, wearable devices, specialized coatings, and environment monitoring.}, author = { and and and and }, date = {2018-07-01}, doi = {10.1016/j.nanoso.2017.10.007}, journal = {Nano-Structures & Nano-Objects}, keywords = {graphene, nanomanufacturing}, pages = {61-62}, publisher = {Elsevier}, pubstate = {published}, title = {A Special Issue on Modeling and nanofabrication of 1D and 2D materials}, tppubtype = {article}, url = {https://doi.org/10.1016/j.nanoso.2017.10.007}, volume = {15}, year = {2018} } 1-10 @InBook{Samborski, Title = {An simple epiost os configurations(asm, LaTeX, Autocode)}, Author = {}, Chapter = {10}, Pages = {13}, Publisher = {epiost}, Year = {2022}, Owner = {Tu Do}, Timestamp = {2022.21.03} } \documentclass[10pt, letterpaper]{amsart} \usepackage{tikz} \usetikzlibrary{shapes.geometric, arrows} \begin{document} \tikzstyle{startstop} = [rectangle, rounded corners, minimum width=3cm, minimum height=1cm,text centered, draw=black, fill=red!30] \tikzstyle{io} = [trapezium, trapezium left angle=70, trapezium right angle=110, minimum width=3cm, minimum height=1cm, text centered, draw=black, fill=blue!30] \tikzstyle{process} = [rectangle, minimum width=3cm, minimum height=1cm, text centered, draw=black, fill=orange!30] \tikzstyle{decision} = [diamond, minimum width=3cm, minimum height=1cm, text centered, draw=black, fill=green!30] \tikzstyle{arrow} = [thick,->,>=stealth] \resizebox{9cm}{!}{ \begin{tikzpicture}[node distance=2cm][scale=0.5] \node (OLS) [startstop] {Run OLS regression}; \node (LM) [io, below of = OLS, align=center] {LM diagnostics\\LM-Error\\LM-Lag}; \node (Significance) [decision, below of = LM, yshift=-0.9cm, align=center] {Check\\significance}; \node (insignificant) [process, left of = Significance, xshift=-2.5cm, align=center] {Neither LM-Lag nor\\LM-Error significant}; \node (keep) [startstop, below of = insignificant, align=center] {Stop and keep\\OLS results}; \node (significant) [process, right of = Significance, xshift=2.5cm, align=center] {One or both \\LM-Lag and LM-Error\\significant}; \node (spatial) [process, below of = significant, align=center] {Indicates spatial\\autocorrelation}; \node (robust) [io, below of = spatial, align=center] {Robust LM diagnostics\\Robust LM-Error\\Robust LM-Lag}; \node (Significance2) [decision, below of = robust, yshift=-0.9cm, align=center] {Check\\significance}; \node (robusterror) [process, right of = Significance2, xshift=2.5cm, align=center] {Robust LM-Error}; \node (runerror) [startstop, below of = robusterror, align=center] {Run Spatial Error model}; \node (robustlag) [process, left of = Significance2, xshift=-2.5cm, align=center] {Robust LM-Lag}; \node (runlag) [startstop, below of = robustlag, align=center] {Run Spatial Lag model}; % \draw [arrow] (OLS) -- (LM); \draw [arrow] (LM) -- (Significance); \draw [arrow] (Significance) -- (insignificant); \draw [arrow] (insignificant) -- (keep); \draw [arrow] (Significance) -- (significant); \draw [arrow] (significant) -- (spatial); \draw [arrow] (spatial) -- (robust); \draw [arrow] (robust) -- (Significance2); \draw [arrow] (Significance2) -- (robusterror); \draw [arrow] (robusterror) -- (runerror); \draw [arrow] (Significance2) -- (robustlag); \draw [arrow] (robustlag) -- (runlag); \end{tikzpicture}} \end{document} Doxygen/latex/hwlib-doxygen-#0060-char-io_8hpp.tex \hypertarget{hwlib-doxygen-#0060-char-io_8hpp}{}\section{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0060-\/char-\/io.hpp File Reference} \label{hwlib-doxygen-#0060-char-io_8hpp}\index{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0060-\/char-\/io.\+hpp@{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0060-\/char-\/io.\+hpp}} @article{abrahamDentalCalculus2005, title = {A Case Study of Elemental and Structural Composition of Dental Calculus during Several Stages of Maturation Using {{SRXRF}}}, author = {. and .}, date = {2005}, journaltitle = {Journal of Biomedical Materials Research Part A}, volume = {75A}, number = {3}, pages = {623--628}, issn = {1552-4965}, doi = {10.1002/jbm.a.30484}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/jbm.a.30484}, urldate = {2020-09-17}, abstract = {This work presents a study of elemental composition and preponderant structure of human dental calculus, as they mature in the mouth for a period of 1 year. The synchrotron radiation X-ray fluorescence technique using a white beam was employed as an analytical method. The set of samples were extracted from different dental elements of the same subject, who did not require any other clinical care. By means of analyzing the Ca/P molar ratio an estimation of the main crystallographic structure was attained, by simple comparison with stoichiometric values of the several crystalline structures that compose the calculus. The results showed a slowly progressive transformation of the initial crystalline structures (brushite) into more stable structures (hydroxyapatite), passing through octacalcium phosphate and whitlockite. The concentrations of mayor components (Ca and P) as a function of time followed a sigmoid curve. The analysis of trace element concentratios versus time indicated a null or small correlation of concentration values with the kinetics of the crystallization process. © 2005 Wiley Periodicals, Inc. J Biomed Mater Res, 2005}, langid = {english}, keywords = {dental calculus,maturation,SRXRF}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/jbm.a.30484}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/HW9B7KA5/Abraham et al. - 2005 - A case study of elemental and structural compositi.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/79IB6I9R/jbm.a.html} } @book{adamsHitchhikersGuide2002, title = {The {{Hitchhiker}}'s {{Guide}} to the {{Galaxy}}}, author = {}, date = {2002}, publisher = {{Picador}}, location = {{London}} } @book{adamsRestaurantEnd2002, title = {The {{Restaurant}} at the {{End}} of the {{Universe}}}, author = {}, date = {2002}, publisher = {{Picador}}, location = {{London}} } @article{adlerSequencingAncient2013, title = {Sequencing Ancient Calcified Dental Plaque Shows Changes in Oral Microbiota with Dietary Shifts of the {{Neolithic}} and {{Industrial}} Revolutions}, author = {. and . and . and . and . and . and . and . and Soltysiak, A. and . and . and .}, date = {2013}, journaltitle = {Nature Genetics}, shortjournal = {Nat Genet}, volume = {45}, number = {4}, pages = {450--5, 455e1}, issn = {1546-1718 (Electronic) 1061-4036 (Linking)}, doi = {10.1038/ng.2536}, abstract = {The importance of commensal microbes for human health is increasingly recognized, yet the impacts of evolutionary changes in human diet and culture on commensal microbiota remain almost unknown. Two of the greatest dietary shifts in human evolution involved the adoption of carbohydrate-rich Neolithic (farming) diets (beginning approximately 10,000 years before the present) and the more recent advent of industrially processed flour and sugar (in approximately 1850). Here, we show that calcified dental plaque (dental calculus) on ancient teeth preserves a detailed genetic record throughout this period. Data from 34 early European skeletons indicate that the transition from hunter-gatherer to farming shifted the oral microbial community to a disease-associated configuration. The composition of oral microbiota remained unexpectedly constant between Neolithic and medieval times, after which (the now ubiquitous) cariogenic bacteria became dominant, apparently during the Industrial Revolution. Modern oral microbiotic ecosystems are markedly less diverse than historic populations, which might be contributing to chronic oral (and other) disease in postindustrial lifestyles.}, pmcid = {PMC3996550}, keywords = {*Archaeology,*Diet,*Industry,Biological Evolution,Dental Plaque/*genetics/microbiology,High-Throughput Nucleotide Sequencing,Humans,Metagenome/*genetics,Mouth Mucosa/*microbiology/pathology}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/Q3U8NHDE/Adler et al. - 2013 - Sequencing ancient calcified dental plaque shows c.pdf} } @article{akcaliDentalCalculus2018, title = {Dental Calculus: The Calcified Biofilm and Its Role in Disease Development}, shorttitle = {Dental Calculus}, author = { Lang, .}, date = {2018}, journaltitle = {Periodontology 2000}, volume = {76}, number = {1}, pages = {109--115}, issn = {1600-0757}, doi = {10.1111/prd.12151}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/prd.12151}, urldate = {2021-06-07}, abstract = {Dental calculus represents the first fossilized record of bacterial communities as a testimony of evolutionary biology. The development of dental calculus is a dynamic process that starts with a nonmineralized biofilm which eventually calcifies. Nonmineralized dental biofilm entraps particles from the oral cavity, including large amounts of oral bacteria, human proteins, viruses and food remnants, and preserves their DNA. The process of mineralization involves metabolic activities of the bacterial colonies and strengthens the attachment of nonmineralized biofilms to the tooth surface. From a clinical point of view, dental calculus always harbors a living, nonmineralized biofilm, jeopardizing the integrity of the dento-gingival or implanto-mucosal unit. This narrative review presents a brief historical overview of dental calculus formation and its clinical relevance in modern periodontal practice.}, langid = {english}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/prd.12151}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/BWR9G623/Akcalı and Lang - 2018 - Dental calculus the calcified biofilm and its rol.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/IH87S6DM/prd.html} } @article{armitageExtractionIdentification1975, title = {The {{Extraction}} and {{Identification}} of {{Opal Phytoliths}} from the {{Teeth}} of {{Ungulates}}}, author = {.}, date = {1975}, journaltitle = {Journal of Archaeological Science}, shortjournal = {J Archaeol Sci}, volume = {2}, pages = {187--197}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/JUY4FMBH/Armitage - 1975 - The Extraction and Identification of Opal Phytolit.pdf} } @book{aufderheidePaleopathology1998, title = {The {{Cambridge}} Encyclopedia of Human Paleopathology}, author = { and }, date = {1998}, volume = {478}, publisher = {{Cambridge University Press Cambridge}} } @article{balajiUnusualPresentation2019, title = {An Unusual Presentation of Dental Calculus}, author = { and }, date = {2019}, journaltitle = {Journal of Indian Society of Periodontology}, shortjournal = {J Indian Soc Periodontol}, volume = {23}, number = {5}, eprint = {31543623}, eprinttype = {pmid}, pages = {484--486}, issn = {0972-124X}, doi = {10.4103/jisp.jisp_680_18}, url = {https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6737844/}, urldate = {2022-02-28}, abstract = {Calculus is a mineralized bacterial plaque that is formed on natural teeth surfaces where there is constant supply of saliva. Dental calculus is commonly seen over the buccal surfaces of maxillary molars and lingual surfaces of mandibular anterior teeth where the salivary duct opens into the oral cavity. This case report presents an unusual presentation of a large hard calcified mass in the left side of retromolar region associated with partially erupted tooth; hard mass was excised and examined histochemically which suggested the presence of calculus. Elimination of such nidus shall prevent formation of such calculus in such unusual position. This can also be achieved with proper oral hygiene measures.}, pmcid = {PMC6737844} } @article{bjarnsholtVivoBiofilm2013, title = {The in Vivo Biofilm}, author = {Bjarnsholt, , , , . and Moser, , Jensen, Høiby, Niels}, date = {2013-09}, journaltitle = {Trends in Microbiology}, shortjournal = {Trends in Microbiology}, volume = {21}, number = {9}, pages = {466--474}, issn = {0966842X}, doi = {10.1016/j.tim.2013.06.002}, url = {https://linkinghub.elsevier.com/retrieve/pii/S0966842X1300111X}, urldate = {2020-06-18}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/6HNDI4KR/Bjarnsholt et al. - 2013 - The in vivo biofilm.pdf} } @book{brothwellDiggingBones1981, title = {Digging up {{Bones}}: {{The}} Excavation, Treatment and Study of Human Skeletal Remains}, author = {}, date = {1981}, edition = {3rd}, publisher = {{British Museum (Natural History)}}, location = {{London}} } @article{bucchiComparisonsMethods2019, title = {Comparisons between Methods for Analyzing Dental Calculus Samples from {{El Mirador}} Cave ({{Sierra}} de {{Atapuerca}}, {{Spain}})}, author = { }, date = {2019-11-01}, journaltitle = {Archaeological and Anthropological Sciences}, shortjournal = {Archaeol Anthropol Sci}, volume = {11}, number = {11}, pages = {6305--6314}, issn = {1866-9565}, doi = {10.1007/s12520-019-00919-z}, url = {https://doi.org/10.1007/s12520-019-00919-z}, urldate = {2022-03-18}, abstract = {Microremains entrapped in dental calculus are being used as a source of information to address a number of archeological questions. However, current laboratory procedures may affect the recovery of microremains and this issue has not been thoroughly investigated. This study involved the analysis of dental calculus from five Chalcolithic individuals from El Mirador cave (Sierra de Atapuerca, Spain) from a methodological perspective. Two sample processing protocols published in the archeological literature were used for this purpose, and results were compared to the El Mirador archaeobotanical record published elsewhere. Analyzed as a whole, the microremains found in the dental calculus samples are consistent with a population immersed in a farming economy, although they are not representative of the richness of the archaeobotanical record of the site. Furthermore, the two protocols delivered different results, in terms of the number of microremains identified, the time required for analysis, and associated contamination problems. This data indicates that the method selected may affect the results. We recommend further research using a larger sample set to fully understand how methodological factors affect the preservation and observation of microremains embedded in dental calculus. We also call for a discussion on the role of dental calculus in archeological research.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/75CLPZHH/Bucchi et al. - 2019 - Comparisons between methods for analyzing dental c.pdf} } @article{buckleyDentalCalculus2014, title = {Dental {{Calculus Reveals Unique Insights}} into {{Food Items}}, {{Cooking}} and {{Plant Processing}} in {{Prehistoric Central Sudan}}}, author = { Usai, , }, date = {2014-07-16}, journaltitle = {PLOS ONE}, shortjournal = {PLOS ONE}, volume = {9}, number = {7}, pages = {e100808}, publisher = {{Public Library of Science}}, issn = {1932-6203}, doi = {10.1371/journal.pone.0100808}, url = {https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0100808}, urldate = {2020-06-22}, abstract = {Accessing information on plant consumption before the adoption of agriculture is challenging. However, there is growing evidence for use of locally available wild plants from an increasing number of pre-agrarian sites, suggesting broad ecological knowledge. The extraction of chemical compounds and microfossils from dental calculus removed from ancient teeth offers an entirely new perspective on dietary reconstruction, as it provides empirical results on material that is already in the mouth. Here we present a suite of results from the multi-period Central Sudanese site of Al Khiday. We demonstrate the ingestion in both pre-agricultural and agricultural periods of Cyperus rotundus tubers. This plant is a good source of carbohydrates and has many useful medicinal and aromatic qualities, though today it is considered to be the world's most costly weed. Its ability to inhibit Streptococcus mutans may have contributed to the unexpectedly low level of caries found in the agricultural population. Other evidence extracted from the dental calculus includes smoke inhalation, dry (roasting) and wet (heating in water) cooking, a second plant possibly from the Triticaceae tribe and plant fibres suggestive of raw material preparation through chewing.}, langid = {english}, keywords = {Calculus,Microfossils,Neolithic period,Paleobotany,Paleozoology,Starches,Teeth,Tubers}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/8CQB3LHA/pone.0100808.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/HPUWM829/Buckley et al. - 2014 - Dental Calculus Reveals Unique Insights into Food .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/NPK7DY4J/article.html} } @article{chenStarchGrains2021, title = {Starch Grains from Human Teeth Reveal the Plant Consumption of Proto-{{Shang}} People (c. 2000–1600 {{BC}}) from {{Nancheng}} Site, {{Hebei}}, {{China}}}, author = { .}, date = {2021-08-20}, journaltitle = {Archaeological and Anthropological Sciences}, shortjournal = {Archaeol Anthropol Sci}, volume = {13}, number = {9}, pages = {153}, issn = {1866-9565}, doi = {10.1007/s12520-021-01416-y}, url = {https://doi.org/10.1007/s12520-021-01416-y}, urldate = {2021-09-26}, abstract = {The founding processes of the first state of ancient China with a known written record, the Shang dynasty (3600–3046~cal BP), have been poorly understood. Recent discoveries of a host of archaeological sites dating to the proto-Shang culture (4000–3600~cal BP) have helped elucidate the transition to the Shang culture. Nevertheless, there are few investigations about the mode of subsistence and economy of the proto-Shang culture, and how this might have shaped the transition to statehood. In this present study, we analyzed the starch grains preserved in dental calculus and teeth surfaces from 16 samples from the site of Nancheng in order to gain a better understanding of the subsistence strategy and plant consumption of proto-Shang people. We also performed experiments to test how different cooking methods may lead to size changes in the starches of four Poaceae plants, in order to identify the processing methods used by the proto-Shang people. The results indicate that Triticum aestivum, Coix lacryma-jobi, Setaria italica and some yet-unidentified roots and tubers were consumed by these individuals. These data indicate a broader spectrum of plant consumption than that seen by previous archaeobotanical and stable isotope analyses. Such a broad spectrum of plant consumption provided a substantial economic base for proto-Shang people and might be one of the factors supporting the subsequent development of the Shang state culture.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/MEA4AWE6/Chen et al. - 2021 - Starch grains from human teeth reveal the plant co.pdf} } @article{ciochonOpalPhytoliths1990, title = {Opal Phytoliths Found on the Teeth of the Extinct Ape {{Gigantopithecus}} Blacki: Implications for Paleodietary Studies.}, shorttitle = {Opal Phytoliths Found on the Teeth of the Extinct Ape {{Gigantopithecus}} Blacki}, author = { and and }, date = {1990-10}, journaltitle = {Proceedings of the National Academy of Sciences}, volume = {87}, number = {20}, pages = {8120--8124}, publisher = {{Proceedings of the National Academy of Sciences}}, doi = {10.1073/pnas.87.20.8120}, url = {https://www.pnas.org/doi/10.1073/pnas.87.20.8120}, urldate = {2022-04-04}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/PXDMQLN3/Ciochon et al. - 1990 - Opal phytoliths found on the teeth of the extinct .pdf} } @article{collinsHomelessDental2007, title = {Homeless in {{North}} and {{West Belfast}}: An Oral Health Needs Assessment}, shorttitle = {Homeless in {{North}} and {{West Belfast}}}, author = {. and .}, date = {2007-06}, journaltitle = {British Dental Journal}, shortjournal = {Br Dent J}, volume = {202}, number = {12}, pages = {E31-E31}, publisher = {{Nature Publishing Group}}, issn = {1476-5373}, doi = {10.1038/bdj.2007.473}, url = {https://www.nature.com/articles/bdj.2007.473}, urldate = {2022-02-25}, abstract = {This is the first assessment of the oral health needs of single homeless adults in North and West Belfast.The study population had greater experience of dental caries and periodontal disease compared with adults in Northern Ireland (NI) who took part in the 1998 Adult Dental Health Survey.The widespread use of smoking and use and abuse of alcohol marked the study population as a high-risk group for oral cancer, with the increased risk calculated as 95 times greater than the NI population as a whole.Homeless people are not a homogeneous population, therefore social exclusion and psycho-social functioning should be considered when planning appropriate oral healthcare services for this diverse client group.}, issue = {12}, langid = {english}, keywords = {Dentistry}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/VWD7CA2W/Collins and Freeman - 2007 - Homeless in North and West Belfast an oral health.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/SG2NPBAT/bdj.2007.html} } @article{costertonBacterialBiofilmsNature1987, title = {Bacterial {{Biofilms}} in {{Nature}} and {{Disease}}}, author = { and Cheng, Geesey, and Nickel, and Dasgupta, M and }, date = {1987-10}, journaltitle = {Annual Review of Microbiology}, shortjournal = {Annu. Rev. Microbiol.}, volume = {41}, number = {1}, pages = {435--464}, issn = {0066-4227, 1545-3251}, doi = {10.1146/annurev.mi.41.100187.002251}, url = {https://www.annualreviews.org/doi/10.1146/annurev.mi.41.100187.002251}, urldate = {2022-02-04}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/VBKWCWSE/annurev.mi.41.100187.pdf} } @article{costertonMicrobialBiofilms1995, title = {Microbial {{Biofilms}}}, author = { Lewandowski, Caldwell, . and Korber, . and Lappin-Scott, .}, date = {1995-10}, journaltitle = {Annual Review of Microbiology}, shortjournal = {Annu. Rev. Microbiol.}, volume = {49}, number = {1}, pages = {711--745}, issn = {0066-4227, 1545-3251}, doi = {10.1146/annurev.mi.49.100195.003431}, url = {https://www.annualreviews.org/doi/10.1146/annurev.mi.49.100195.003431}, urldate = {2022-03-08}, langid = {english} } @online{daaDentalAnthropology, title = {Dental {{Anthropology Association}}}, url = {http://www.dentalanthropology.org}, urldate = {2022-03-14}, abstract = {Dental Anthropology Association}, langid = {american}, organization = {{Dental Anthropology Association}}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/H4JA2Y9Q/www.dentalanthropology.org.html} } @article{damenSilicicAcid1989, title = {The {{Effect}} of {{Silicic Acid}} on {{Calcium Phosphate Precipitation}}}, author = {. and .}, date = {1989-09-01}, journaltitle = {Journal of Dental Research}, shortjournal = {J Dent Res}, volume = {68}, number = {9}, pages = {1355--1359}, publisher = {{SAGE Publications Inc}}, issn = {0022-0345}, doi = {10.1177/00220345890680091301}, url = {https://doi.org/10.1177/00220345890680091301}, urldate = {2020-05-18}, abstract = {So that a possible involvement in the mineralization of dental plaque could be investigated, the effects of silicic acid on calcium phosphate precipitation were assessed in vitro. By measuring the decrease in Ca2+ concentration (by means of ion-selective electrodes), we determined both spontaneous precipitation and seeded crystal growth from solutions that contained 1 mmol/L calcium, 7.5 mmol/L phosphate, 50 mmol/L Hepes pH 7.2, and various amounts of silicic acid. Polymerized silicic acid, but not its monomer, was found both to cause a 60\% reduction in the lag period that precedes spontaneous precipitation and to enhance the growth rate of seeded hydroxyapatite crystals. Silica suspensions showed effects similar to those of polysilicic acid. In all cases, the precipitated material was found to be hydroxyapatite. Whereas seeded brushite crystals grew slowly without silicic acid, hydroxyapatite was the only mineral detected after crystal growth in the presence of silicic acid. Apparently, polysilicic acid acted as a substrate for hydroxyapatite nucleation, inducing secondary nuclei on both hydroxyapatite and brushite crystals. The finding that polysilicic acid could overcome part of the inhibitory effect of a phosphoprotein on calcium phosphate precipitation gave additional support for the idea that polysilicic acid and silica may promote the formation of dental calculus.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/JXLEQWHV/Damen and Ten Cate - 1989 - The Effect of Silicic Acid on Calcium Phosphate Pr.pdf} } @article{delafuenteDNAHuman2013, title = {{{DNA From Human Ancient Bacteria}}: {{A}} Novel Source of Genetic Evidence from Archaeological Dental Calculus}, shorttitle = {{{DNA FROM HUMAN ANCIENT BACTERIA}}}, author = {. and . and .}, date = {2013-08}, journaltitle = {Archaeometry}, volume = {55}, number = {4}, pages = {767--778}, issn = {0003813X}, doi = {10.1111/j.1475-4754.2012.00707.x}, url = {https://onlinelibrary.wiley.com/doi/10.1111/j.1475-4754.2012.00707.x}, urldate = {2021-07-09}, abstract = {We report a molecular methodology to obtain and analyse ancient bacterial DNA from archaeological dental calculus. Recent and archaeological DNA samples, as old as 4000 BP, were successfully extracted and amplified with species-specific PCR primers. We propose this approach in order to: detect the presence of specific bacterial species infecting past human populations; compare the composition of ancient oral microbiomes among human populations; and analyse the genetic variability and covariation of bacteria and human host populations. Genomic analysis of bacteria from dental calculus is a promising source of evidence for palaeopathological and micro-evolutionary studies, focused either on micro-organisms or their human hosts.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/UWSNJ9FN/De La Fuente et al. - 2013 - DNA FROM HUMAN ANCIENT BACTERIA A NOVEL SOURCE OF.pdf} } @article{dibdinOralUrea1998, title = {A {{Mathematical Model}} of the {{Influence}} of {{Salivary Urea}} on the {{pH}} of {{Fasted Dental Plaque}} and on the {{Changes Occurring}} during a {{Cariogenic Challenge}}}, author = {. and .}, date = {1998}, journaltitle = {Caries Research}, shortjournal = {Caries Res}, volume = {32}, number = {1}, pages = {70--74}, issn = {0008-6568, 1421-976X}, doi = {10.1159/000016432}, url = {https://www.karger.com/Article/FullText/16432}, urldate = {2021-03-17}, abstract = {Urea diffusing from saliva into dental plaque is converted to ammonia and carbon dioxide by bacterial ureases. The influence of normal salivary urea levels on the pH of fasted plaque and on the depth and duration of a Stephan curve is uncertain. A numerical model which simulates a cariogenic challenge (a 10\% sucrose rinse alone or one followed by use of chewing-gum with or without sugar) was modified to include salivary urea levels from 0 to 30 mmol/l. It incorporated: site-dependent exchange between bulk saliva and plaque surfaces via a salivary film; sugar and urea diffusion into plaque; pH-dependent rates of acid formation and urea breakdown; diffusion and dissociation of end-products and other buffers (acetate, lactate, phosphate, ammonia and carbonate); diffusion of protons and other ions; equilibration with fixed and mobile buffers; and chargecoupling between ionic flows. The Km (2.12 mmol/l) and Vmax (0.11 µmol urea/ min/mg dry weight) values for urease activity and the pH dependence of Vmax were taken from the literature. From the results, it is predicted that urea concentrations normally present in saliva (3–5 mmol/l) will increase the pH at the base of a 0.5-mm-thick fasted plaque by up to 1 pH unit, and raise the pH minimum after a sucrose rinse or sugar-containing chewing-gum by at least half a pH unit. The results suggest that plaque cariogenicity may be inversely related to salivary urea concentrations, not only when the latter are elevated because of disease, but even when they are in the normal range.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/VVQYDV7Z/Dibdin and Dawes - 1998 - A Mathematical Model of the Influence of Salivary .pdf} } @article{dobneyMethodEvaluating1987, title = {A Method for Evaluating the Amount of Dental Calculus on Teeth from Archaeological Sites}, author = { and }, date = {1987-07-01}, journaltitle = {Journal of Archaeological Science}, shortjournal = {J Archaeol Sci}, volume = {14}, number = {4}, pages = {343--351}, issn = {0305-4403}, doi = {10.1016/0305-4403(87)90024-0}, url = {http://www.sciencedirect.com/science/article/pii/0305440387900240}, urldate = {2020-05-15}, abstract = {A series of archaeological teeth, both human and animal, were scanned with a view to establishing a quick and reliable method of recording the extent and severity of dental calculus. The method involved allocating each dentition or tooth a general score, one of five grades: 0 (normal) to 4 (gross). Each tooth was then divided into morphological zones, depending upon species, and the percentage covering of calculus for each zone was recorded. Thickness was also assessed using a five-grade scoring system and was based on radiographs of affected teeth. This gave an idea of the severity of the calculus deposit.}, langid = {english}, keywords = {buttress,dental calculus,grade,infolding,involution,scoring,sub-gingival,supra-gingival}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/92WJMSQP/Dobney and Brothwell - 1987 - A method for evaluating the amount of dental calcu.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/XZH4B633/0305440387900240.html} } @article{drewettExcavationOval1975, title = {The {{Excavation}} of an {{Oval Burial Mound}} of the {{Third Millennium}} Be at {{Alfriston}}, {{East Sussex}}, 1974}, author = {}, date = {1975}, pages = {38}, abstract = {The small oval burial mound at Alfriston, East Sussex, being one of only twelve certain burial structures of the 3rd millennium be in Sussex, was totally excavated in 1974 prior to its final obliteration by ploughing. The barrow was found to consist of a simple dump mound derived from material out of flanking ditches. It covered a single burial pit containing the crouched skeleton of a young female. Information concerning the post-Neolithic land use of Alfriston Down was obtained from the ditch silts and expanded by an intensive field survey.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/YCJNL5VP/Drewett - The Excavation of an Oval Burial Mound of the Thir.pdf} } @article{dudgeonDietGeography2014, title = {Diet, {{Geography}} and {{Drinking Water}} in {{Polynesia}}: {{Microfossil Research}} from {{Archaeological Human Dental Calculus}}, {{Rapa Nui}} ({{Easter Island}})}, shorttitle = {Diet, {{Geography}} and {{Drinking Water}} in {{Polynesia}}}, author = {. and Tromp, Monica}, date = {2014}, journaltitle = {International Journal of Osteoarchaeology}, volume = {24}, number = {5}, pages = {634--648}, issn = {1099-1212}, doi = {10.1002/oa.2249}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/oa.2249}, urldate = {2021-03-15}, abstract = {Microfossil analysis of human dental calculus provides consumption-specific and archaeologically relevant data for evaluating diet and subsistence in past populations. Calculus was extracted from 114 teeth representing 104 unique individuals from a late 16th to early 18th century skeletal series on Rapa Nui (Easter Island) to address questions of human–environment interactions and possible dietary preference. Scanning electron microscopy was used in lieu of optical microscopy for its superior depth of field and resolution of surface detail. The calculus microfossil recovery produced 16,377 total biogenic silica microfossils: 4733 phytoliths and 11,644 diatoms. The majority of phytoliths correspond with the Arecaceae or palm family (n = 4,456) and the minority corresponds to the Poaceae or grass family (n = 277). Because of the relatively large sample size, we were able to test hypotheses related to age cohort, sex, dental element and geographic region. Results indicate no significant difference in phytolith or diatom recovery based on age cohort or sex. The high frequency and proportion of Arecaceae phytoliths found in calculus extracted from the anterior dentition suggests consumption of soft or cooked foods containing palm phytoliths and the high frequency of diatoms recovered from the southern part of the island argue for different sources of drinking water. Copyright © 2012 \& Sons, Ltd.}, langid = {english}, keywords = {dental calculus,diatoms,Easter Island,microfossils,phytoliths,Polynesia,Rapa Nui}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/oa.2249}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/MQTLSTTU/Dudgeon and Tromp - 2014 - Diet, Geography and Drinking Water in Polynesia M.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/STQ72PEW/oa.html} } @article{fagernasMicrobialBiogeography2021, title = {Understanding the Microbial Biogeography of Ancient Human Dentitions to Guide Study Design and Interpretation}, author = {, , and }, date = {2021-01-01}, journaltitle = {bioRxiv}, pages = {2021.08.16.456492}, doi = {10.1101/2021.08.16.456492}, url = {http://biorxiv.org/content/early/2021/08/16/2021.08.16.456492.abstract}, abstract = {The oral cavity is a heterogeneous environment, varying in factors such as pH, oxygen levels, and salivary flow. These factors affect the microbial community composition and distribution of species in dental plaque, but it is not known how well these patterns are reflected in archaeological dental calculus. In most archaeological studies, a single sample of dental calculus is studied per individual and is assumed to represent the entire oral cavity. However, it is not known if this sampling strategy introduces biases into studies of the ancient oral microbiome. Here, we present the results of a shotgun metagenomic study of a dense sampling of dental calculus from four Chalcolithic individuals from the southeast Iberian peninsula (ca. 4500-5000 BP). Inter-individual differences in microbial composition are found to be much larger than intra-individual differences, indicating that a single sample can indeed represent an individual in most cases. However, there are minor spatial patterns in species distribution within the oral cavity that should be taken into account when designing a study or interpreting results. Finally, we show that plant DNA identified in the samples may be of environmental origin, showing the importance of including environmental controls or several lines of biomolecular evidence.Competing Interest StatementThe authors have declared no competing interest.}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/TDQEBQIB/Fagernäs et al. - 2021 - Understanding the microbial biogeography of ancien.pdf} } @online{fdiOralHealth, title = {{{FDI}}’s Definition of Oral Health | {{FDI}}}, url = {https://www.fdiworlddental.org/fdis-definition-oral-health}, urldate = {2022-03-14}, organization = {{FDI World Dental Federation}}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/QV4DXVCE/fdis-definition-oral-health.html} } @article{fehrVitroCalculus1960, title = {In Vitro Calculus Formation}, author = {, .}, date = {1960-09/1960-10}, journaltitle = {J Dent Res}, volume = {39}, pages = {1041--8}, issn = {0022-0345 (Print) 0022-0345 (Linking)}, doi = {10.1177/00220345600390050801}, keywords = {*Calculi,*Teeth,*Tooth,Humans,In Vitro Techniques} } @article{fiorinCombiningDental2021, title = {Combining Dental Calculus with Isotope Analysis in the {{Alps}}: {{New}} Evidence from the {{Roman}} and Medieval Cemeteries of {{Lamon}}, Northern {{Italy}}}, shorttitle = {Combining Dental Calculus with Isotope Analysis in the {{Alps}}}, author = { and Moore, Joanna and Montgomery, Janet and and and }, date = {2021-11-26}, journaltitle = {Quaternary International}, shortjournal = {Quaternary International}, issn = {1040-6182}, doi = {10.1016/j.quaint.2021.11.022}, url = {https://www.sciencedirect.com/science/article/pii/S1040618221005632}, urldate = {2021-12-02}, abstract = {This study presents the results of complementary isotopic and dental calculus analyses of a number of individuals buried in two cemeteries of Roman and medieval chronology in Lamon, northern Italy. Eleven individuals from the Roman cemetery of San Donato and six from the medieval cemetery of San Pietro are presented and discussed. The results suggest a distinctive stability of the two populations, with most of the analysed individuals showing a local or regional origin. Carbon and nitrogen isotopes are indicative of a diet based on a mixed C3/C4 plant consumption and rich in animal proteins, with no significant difference between the Roman and the medieval populations. The consumption of C4 plants, more resilient to the Alpine climate, is consistently documented both by isotopes and dental calculus. Dental calculus results permit the characterisation of the typology of the crop consumed, namely millet, barley/wheat and legumes and may also suggest differing cooking processes between the Roman and the medieval periods. Phytoliths, vascular elements, fungal spores and animal remains from dental calculus provide new insights into the diet of the analysed individuals but also, hypothetically, into possible medicinal treatments. The presence of birds such as fowls and ducks in the medieval diet of some individuals from San Pietro has also emerged. Overall, the results of this study open a new window into the biographies of the individuals analysed, their diet, mobility, habits, and environment, thus stimulating further and more systematic investigation on the populations occupying an Alpine sector which is still poorly understood from an archaeological perspective.}, langid = {english}, keywords = {Alps,Dental calculus,Isotopes,Middle ages,Roman period}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/IRM8NVFQ/Fiorin et al. - 2021 - Combining dental calculus with isotope analysis in.pdf} } @article{flemmingBiofilmsEmergent2016, title = {Biofilms: An Emergent Form of Bacterial Life}, shorttitle = {Biofilms}, author = { Wingender, Szewzyk, , , . and Kjelleberg, Staffan}, date = {2016-09}, journaltitle = {Nature Reviews Microbiology}, shortjournal = {Nat Rev Microbiol}, volume = {14}, number = {9}, pages = {563--575}, issn = {1740-1526, 1740-1534}, doi = {10.1038/nrmicro.2016.94}, url = {http://www.nature.com/articles/nrmicro.2016.94}, urldate = {2022-03-16}, abstract = {Bacterial biofilms are formed by communities that are embedded in a self-produced matrix of extracellular polymeric substances (EPS). Importantly, bacteria in biofilms exhibit a set of ‘emergent properties’ that differ substantially from free-living bacterial cells. In this Review, we consider the fundamental role of the biofilm matrix in establishing the emergent properties of biofilms, describing how the characteristic features of biofilms — such as social cooperation, resource capture and enhanced survival of exposure to antimicrobials — all rely on the structural and functional properties of the matrix. Finally, we highlight the value of an ecological perspective in the study of the emergent properties of biofilms, which enables an appreciation of the ecological success of biofilms as habitat formers and, more generally, as a bacterial lifestyle.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/5PHFJH6B/Flemming et al. - 2016 - Biofilms an emergent form of bacterial life.pdf} } @article{foxPhytolithCalculus1996, title = {Phytolith Analysis on Dental Calculus, Enamel Surface, and Burial Soil: {{Information}} about Diet and Paleoenvironment}, author = {Fox, and and .}, date = {1996}, journaltitle = {American Journal of Physical Anthropology}, volume = {101}, number = {1}, pages = {101--113}, issn = {1096-8644}, doi = {10.1002/(SICI)1096-8644(199609)101:1<101::AID-AJPA7>3.0.CO;2-Y}, abstract = {Silica phytoliths (microscopic remains originating in plant tissues) have been identified on the enamel surface and dental calculus of a sample of teeth selected from well preserved skeletons from a Late Roman necropolis in Tarragona (Spain). Phytoliths were observed by scanning electron microscopy (SEM) and their siliceous nature was confirmed by X-ray microanalysis. The phytoliths were compared to those of soil samples from both the areas of the tombs corresponding to the abdomen and the periphery of the skeletons, and were classified taxonomically by comparison with a large collection of silica particles from modern plants in the Mediterranean area. Most of the phytoliths identified on the enamel and the dental calculus belong to the family of Poaceae, while the phytoliths from the abdominal area belong to Poaceae, Leguminosae, Cyperaceae, and Chenopodiaceae. Results are concordant with archaeological, ecological, and historical data from the same site, and with the human Mediterranean diet. If done properly, the study of phytoliths can provide direct information about the vegetable diet of past human populations, and could be applied to the study of human fossils. © 1996 Wiley-Liss, Inc.}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/KV5KXBW5/Fox et al. - 1996 - Phytolith analysis on dental calculus, enamel surf.pdf} } @article{greeneQuantifyingCalculus2005, title = {Quantifying Calculus: {{A}} Suggested New Approach for Recording an Important Indicator of Diet and Dental Health}, author = {. and .}, date = {2005}, journaltitle = {HOMO - Journal of Comparative Human Biology}, volume = {56}, number = {2}, pages = {119--132}, issn = {0018442X}, doi = {10.1016/j.jchb.2005.02.002}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/GHRENJRC/Greene et al. - 2005 - Quantifying calculus A suggested new approach for.pdf} } @article{greeneSimplifiedOral1964, title = {The {{Simplified Oral Hygiene Index}}}, author = {Greene, . and Vermillion, .}, date = {1964-01-01}, journaltitle = {The Journal of the American Dental Association}, shortjournal = {The Journal of the American Dental Association}, volume = {68}, number = {1}, pages = {7--13}, issn = {0002-8177}, doi = {10.14219/jada.archive.1964.0034}, url = {https://www.sciencedirect.com/science/article/pii/S0002817764810047}, urldate = {2022-04-08}, abstract = {Delayed-onset infection is defined as infectious swelling and trismus accompanied by pain or the presence of suppuration starting approximately 30 days after surgery. This study aimed to describe the occurrence and potential predisposing factors of delayed-onset infection. A retrospective case-control study of 223 lower third molar surgeries was performed. Participants were selected from among 1102 outpatients who underwent surgery between January 2013 and June 2018 at Semmelweis University. The inclusion criterion for the case group was inflammation of the operated area after suture removal. Patients in the control group were healthy nonsmokers {$<$}26 years old who healed without complication. Statistical analysis was performed using the Shapiro-Wilk test, the Mann-Whitney U test, and Fisher's exact test. Complications occurred only in patients {$<$}26 years old approximately 29.5 days after surgery. A significantly higher risk was observed for younger age, total soft tissue coverage, deeper impaction, lower Nolla stage (P {$<$} .001), mesioangular direction (P = .002), and full bone coverage (P {$<$} .05). Distal space was inversely correlated with complications (P {$<$} .001). Lower Nolla stage, total soft tissue coverage, lack of distal space, deeper impaction, or mesioangular tilt may promote delayed-onset infection. Follow-up of at-risk patients and the maintenance of oral hygiene are recommended. To investigate the effect of mogroside, palatinose, erythritol, and xylitol on the dental plaque pH of children and to compare the plaque pH change between caries-active and caries-free children caused by these sweeteners. Thirty-six children (mean age 6.2 ± 2.9-year-old), caries-active and caries-free, were included. After refraining from practicing oral hygiene, the accessible plaque was collected and equally divided for challenging with 6 different solutions: mogroside, palatinose, erythritol, xylitol, 10\% sucrose, and deionized water. The pH of each solution was measured using a digital pH meter at 0, 5, 10, 15, 20, 25, and 30 min. Mogroside, erythritol, xylitol, and water did not significantly lower the dental plaque pH, however, palatinose reduced dental plaque pH comparable to sucrose (p {$<$} 0.05). Comparing the caries-active and caries-free groups, only sucrose produced significantly different pH value at min 5 and 10. Mogroside, erythritol, and xylitol did not lower the dental plaque pH in caries-active or caries-free children. However, palatinose affected the dental plaque pH similar to sucrose.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/IA6STUM6/S0002817764810047.html} } @article{hardyNeanderthalMedics2012, title = {Neanderthal Medics? {{Evidence}} for Food, Cooking, and Medicinal Plants Entrapped in Dental Calculus}, shorttitle = {Neanderthal Medics?}, author = {, . and Estalrrich, Almudena and Brothwell, , , , , , , , , , , and }, options = {useprefix=true}, date = {2012-08-01}, journaltitle = {Naturwissenschaften}, shortjournal = {Naturwissenschaften}, volume = {99}, number = {8}, pages = {617--626}, issn = {1432-1904}, doi = {10.1007/s00114-012-0942-0}, url = {https://doi.org/10.1007/s00114-012-0942-0}, urldate = {2021-07-16}, abstract = {Neanderthals disappeared sometime between 30,000 and 24,000~years ago. Until recently, Neanderthals were understood to have been predominantly meat-eaters; however, a growing body of evidence suggests their diet also included plants. We present the results of a study, in which sequential thermal desorption-gas chromatography-mass spectrometry (TD-GC-MS) and pyrolysis-gas chromatography-mass spectrometry (Py-GC-MS) were combined with morphological analysis of plant microfossils, to identify material entrapped in dental calculus from five Neanderthal individuals from the north Spanish site of El Sidrón. Our results provide the first molecular evidence for inhalation of wood-fire smoke and bitumen or oil shale and ingestion of a range of cooked plant foods. We also offer the first evidence for the use of medicinal plants by a Neanderthal individual. The varied use of plants that we have identified suggests that the Neanderthal occupants of El Sidrón had a sophisticated knowledge of their natural surroundings which included the ability to select and use certain plants.}, langid = {english} } @article{hardyStarchGranulesDental2009, title = {Starch Granules, Dental Calculus and New Perspectives on Ancient Diet}, author = { , Kirkham, , Collins, Matthew}, date = {2009}, journaltitle = {Journal of Archaeological Science}, volume = {36}, number = {2}, pages = {248--255}, issn = {03054403}, doi = {10.1016/j.jas.2008.09.015} } @article{hendyProteomicCalculus2018, title = {Proteomic Evidence of Dietary Sources in Ancient Dental Calculus}, author = { , Collins, Fiddyment, Fischer, Hagan, , Holst, , Klaus, , , , , , , .}, date = {2018-07-18}, journaltitle = {Proceedings. Biological Sciences}, shortjournal = {Proc Biol Sci}, volume = {285}, number = {1883}, eprint = {30051838}, eprinttype = {pmid}, pages = {20180977}, issn = {1471-2954}, doi = {10.1098/rspb.2018.0977}, abstract = {Archaeological dental calculus has emerged as a rich source of ancient biomolecules, including proteins. Previous analyses of proteins extracted from ancient dental calculus revealed the presence of the dietary milk protein β-lactoglobulin, providing direct evidence of dairy consumption in the archaeological record. However, the potential for calculus to preserve other food-related proteins has not yet been systematically explored. Here we analyse shotgun metaproteomic data from 100 archaeological dental calculus samples ranging from the Iron Age to the post-medieval period (eighth century BC to nineteenth century AD) in England, as well as 14 dental calculus samples from contemporary dental patients and recently deceased individuals, to characterize the range and extent of dietary proteins preserved in dental calculus. In addition to milk proteins, we detect proteomic evidence of foodstuffs such as cereals and plant products, as well as the digestive enzyme salivary amylase. We discuss the importance of optimized protein extraction methods, data analysis approaches and authentication strategies in the identification of dietary proteins from archaeological dental calculus. This study demonstrates that proteomic approaches can robustly identify foodstuffs in the archaeological record that are typically under-represented due to their poor macroscopic preservation.}, langid = {english}, pmcid = {PMC6083251}, keywords = {Archaeology,dental calculus,Dental Calculus,Diet,dietary reconstruction,DNA; Ancient,England,History; 15th Century,History; 16th Century,History; 17th Century,History; 18th Century,History; 19th Century,History; Ancient,History; Medieval,mass spectrometry,Proteome,proteomics}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/9DS6ICZC/Hendy et al. - 2018 - Proteomic evidence of dietary sources in ancient d.pdf} } @article{henryCalculusSyria2008, title = {Using Plant Microfossils from Dental Calculus to Recover Human Diet: A Case Study from {{Tell}} al-{{Raqā}}'i, {{Syria}}}, author = {Henry, . and Piperno, .}, date = {2008}, journaltitle = {Journal of Archaeological Science}, volume = {35}, number = {7}, pages = {1943--1950}, issn = {03054403}, doi = {10.1016/j.jas.2007.12.005}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/S4T9XT39/Henry and Piperno - 2008 - Using plant microfossils from dental calculus to r.pdf} } @article{henryDietAustralopithecus2012, title = {The Diet of {{Australopithecus}} Sediba}, author = {Henry, . and Ungar, . and Passey, . and Sponheimer, Rossouw, Bamford, Sandberg, , }, options = {useprefix=true}, date = {2012-07}, journaltitle = {Nature}, volume = {487}, number = {7405}, pages = {90--93}, publisher = {{Nature Publishing Group}}, issn = {1476-4687}, doi = {10.1038/nature11185}, url = {https://www.nature.com/articles/nature11185}, urldate = {2022-04-20}, abstract = {Phytolith, stable carbon isotope, and dental microwear texture data for two individuals of Au. sediba, 2-million-year-old hominins from South Africa, show that they consumed a mostly C3 diet that probably included harder foods, and both dicotyledons (for example, tree leaves, fruits, and wood or bark) and monocotyledons (for example, grasses and sedges); this diet contrasts with previously described diets of other early hominin species.}, issue = {7405}, langid = {english}, keywords = {Archaeology}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/2564IA7Y/Henry et al. - 2012 - The diet of Australopithecus sediba.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/A6JUUSMQ/nature11185.html} } @article{henryNeanderthalCalculus2014, title = {Plant Foods and the Dietary Ecology of {{Neanderthals}} and Early Modern Humans}, author = { Brooks, Piperno, .}, date = {2014-04-01}, journaltitle = {Journal of Human Evolution}, shortjournal = {Journal of Human Evolution}, volume = {69}, pages = {44--54}, issn = {0047-2484}, doi = {10.1016/j.jhevol.2013.12.014}, url = {https://www.sciencedirect.com/science/article/pii/S0047248414000189}, urldate = {2021-06-25}, abstract = {One of the most important challenges in anthropology is understanding the disappearance of Neanderthals. Previous research suggests that Neanderthals had a narrower diet than early modern humans, in part because they lacked various social and technological advances that lead to greater dietary variety, such as a sexual division of labor and the use of complex projectile weapons. The wider diet of early modern humans would have provided more calories and nutrients, increasing fertility, decreasing mortality and supporting large population sizes, allowing them to out-compete Neanderthals. However, this model for Neanderthal dietary behavior is based on analysis of animal remains, stable isotopes, and other methods that provide evidence only of animal food in the diet. This model does not take into account the potential role of plant food. Here we present results from the first broad comparison of plant foods in the diets of Neanderthals and early modern humans from several populations in Europe, the Near East, and Africa. Our data comes from the analysis of plant microremains (starch grains and phytoliths) in dental calculus and on stone tools. Our results suggest that both species consumed a similarly wide array of plant foods, including foods that are often considered low-ranked, like underground storage organs and grass seeds. Plants were consumed across the entire range of individuals and sites we examined, and none of the expected predictors of variation (species, geographic region, or associated stone tool technology) had a strong influence on the number of plant species consumed. Our data suggest that Neanderthal dietary ecology was more complex than previously thought. This implies that the relationship between Neanderthal technology, social behavior, and food acquisition strategies must be better explored.}, langid = {english}, keywords = {Dental calculus,Microfossil,Microremain,Neanderthal diet,Phytolith,Starch grain}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/8L2EV6WC/Henry et al. - 2014 - Plant foods and the dietary ecology of Neanderthal.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/234UEDSM/S0047248414000189.html} } @article{hidakaDietCalculus2007, title = {An in Vitro Study of the Effect of Some Dietary Components on Calculus Formation: Regulation of Calcium Phosphate Precipitation}, author = {. and .}, date = {2007-05}, journaltitle = {Oral Diseases}, shortjournal = {Oral Dis}, volume = {13}, number = {3}, pages = {296--302}, issn = {1354-523X (Print) 1354-523X (Linking)}, doi = {10.1111/j.1601-0825.2006.01283.x}, abstract = {OBJECTIVE: We studied the effects of food components on the in vitro formation of calcium phosphate precipitates. MATERIALS AND METHODS: The effects of food components, such as starch, soybean flour, fish meal, rapeseed oil, and coconut oil, on calcium phosphate precipitation were studied using a pH drop method. RESULTS: Although the addition of starch had no effect on the rate of precipitation of amorphous calcium phosphate (ACP), it increased both the rate of transformation of ACP to hydroxyapatite (HAP) and the induction time (i.e. time for the initiation of transformation of ACP to HAP to occur); this was irrespective of the heat treatment of the starch. Amylopectin (insoluble constituent of starch) was effective in increasing the rate of HAP transformation, but amylose (soluble constituent of starch) was not. Oil specimen obtained from rapeseed (400 microl ml(-1)) increased the entire reaction of calcium phosphate precipitation, but that from coconut did not. Protein food, such as soybean flour and fish meal, decreased the rate of transformation of ACP to HAP and increased the induction time, while they had no effect on the rate of ACP precipitation. CONCLUSION: These results suggest that carbohydrate and oil (both are staple diets for the humans) enhance oral calcification (dental calculus formation or re-mineralization of tooth enamel), while side dishes of protein food would decrease it.}, keywords = {Calcium Phosphates/chemistry,Chemical Precipitation,Dental Calculus/*metabolism,Dietary Carbohydrates/*metabolism,Dietary Fats/*metabolism,Dietary Proteins/adverse effects/*metabolism,Fish Products/adverse effects,Hydrogen-Ion Concentration,Hydroxyapatites/metabolism,Plant Oils/metabolism,Soybeans/adverse effects/metabolism,Starch/metabolism,Tooth Calcification/*physiology,Tooth Remineralization}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/72B4A5QV/Hidaka and Oishi - 2007 - An in vitro study of the effect of some dietary co.pdf} } @article{hidakaStarchRole2008, title = {The {{Possible Role}} of {{Starch}} in {{Oral Calcification}}: {{The In Vitro Formation}} of {{Hydroxyapatite}} Is {{Regulated}} by a {{Combination}} of {{Protein}} and {{Mineral Content}} in {{Dietary Starch Flour}}}, shorttitle = {The {{Possible Role}} of {{Starch}} in {{Oral Calcification}}}, author = {Hidaka, , , , Akiko}, date = {2008-04-18}, journaltitle = {The Open Food Science Journal}, shortjournal = {TOFSJ}, volume = {2}, number = {1}, pages = {10--22}, issn = {18742564}, doi = {10.2174/1874256400802010010}, url = {http://benthamopen.com/ABSTRACT/TOFSJ-2-10}, urldate = {2022-02-28}, abstract = {The effects of twelve kinds of dietary starch flour, i.e. rice (non-glutinous and glutinous), wheat (soft, medium, and hard), barley (roasted), buckwheat (inner layer and straight), corn, sweet potato, kudzu, and tapioca on in vitro calcium phosphate precipitation were investigated using the pH drop method. The induction time was elongated by the addition of all of the kinds of flour. Although the rate of amorphous calcium phosphate (ACP) formation was not affected, the rate of transformation of ACP to hydroxyapatite (HAP) was either stimulated or inhibited by the different types of flour. The following observations were made: (1) When the lipid content below 0.2\% (w/w), any of the types of starch had a stimulatory effect on the transformation of ACP to HAP. (2) When the lipid content around or above 1.0\% (w/w), the value of the product of (protein content) (mineral content) seems to determine the effect of starch.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/46B2249C/Hidaka et al. - 2008 - The Possible Role of Starch in Oral Calcification.pdf} } @book{hillsonDentalAnthropology1996, title = {Dental {{Anthropology}}}, author = {}, date = {1996}, publisher = {{Cambridge University Press}}, location = {{Cambridge}}, isbn = {978-1-107-07826-0} } @incollection{irishIntroductionDental2015, title = {Introduction to {{Dental Anthropology}}}, booktitle = {A {{Companion}} to {{Dental Anthropology}}}, author = {. and }, editor = {Irish, . and }, date = {2015}, pages = {3--6}, publisher = {{John Wiley \& Sons, Ltd}}, location = {{Chichester}}, doi = {10.1002/9781118845486.ch18}, url = {http://onlinelibrary.wiley.com/doi/abs/10.1002/9781118845486.ch18}, urldate = {2021-04-21}, isbn = {978-1-118-84548-6}, langid = {english}, keywords = {Arizona State University Dental Anthropology System,distance statistics,genetic relatedness,P henetic affinities,qualitative and quantitative analyses}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/9781118845486.ch18} } @article{jepsenCalculusRemoval2011, title = {Calculus Removal and the Prevention of Its Formation}, author = { and and }, date = {2011}, journaltitle = {Periodontology 2000}, volume = {55}, number = {1}, pages = {167--188}, issn = {1600-0757}, doi = {10.1111/j.1600-0757.2010.00382.x}, url = {http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1600-0757.2010.00382.x}, urldate = {2020-06-24}, langid = {english}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1600-0757.2010.00382.x}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/TJQPWTJW/j.1600-0757.2010.00382.x.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/XD4LBVCE/Jepsen et al. - 2011 - Calculus removal and the prevention of its formati.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/NJ9DXX75/j.1600-0757.2010.00382.html} } @article{jinSupragingivalCalculus2002, title = {Supragingival {{Calculus}}: {{Formation}} and {{Control}}:}, shorttitle = {Supragingival {{Calculus}}}, author = { and }, date = {2002}, journaltitle = {Critical Reviews in Oral Biology \& Medicine}, publisher = {{SAGE Publications}}, doi = {10.1177/154411130201300506}, url = {https://journals.sagepub.com/doi/10.1177/154411130201300506}, urldate = {2020-05-13}, abstract = {Dental calculus is composed of inorganic components and organic matrix. Brushite, dicalcium phosphate dihydrate, octacalcium phosphate, hydroxyapatite, and whit...}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/2C2XJGP7/154411130201300506.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/8RTJSIRS/Jin and Yip - 2016 - Supragingival Calculus Formation and Control.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/8J2J92ND/154411130201300506.html} } @incollection{kinastonOrtnerDentition2019, title = {The {{Dentition}}: {{Development}}, {{Disturbances}}, {{Disease}}, {{Diet}}, and {{Chemistry}}}, shorttitle = {Chapter 21 - {{The Dentition}}}, booktitle = {Ortner's {{Identification}} of {{Pathological Conditions}} in {{Human Skeletal Remains}} ({{Third Edition}})}, author = { Willis, Miszkiewicz, . and Tromp, Monica and Oxenham, .}, editor = {.}, date = {2019-01-01}, pages = {749--797}, publisher = {{Academic Press}}, location = {{San Diego}}, doi = {10.1016/B978-0-12-809738-0.00021-1}, url = {https://www.sciencedirect.com/science/article/pii/B9780128097380000211}, urldate = {2022-02-23}, abstract = {The chief focus of this chapter is the dentition, with reviews of associated structures where relevant. Dental development, including the dentin and enamel, is discussed in the first section, followed by disturbances in the dentin and enamel in the second section. The third section looks at the identification of oral disease, including caries, alveolar lesions, antemortem tooth loss (AMTL), and periodontal disease. The fourth section focuses on interpreting oral health, particularly in the context of sex differences and major demographic transitions. The fifth discusses dental chemistry in terms of paleodietary reconstruction, breastfeeding and weaning, stress and disease, and finally mobility and migration. The final section examines dental calculus in the context of microparticles and then chemical analyses of calculus, with aDNA and protein analyses of dental calculus also reviewed.}, isbn = {978-0-12-809738-0}, langid = {english}, keywords = {aDNA,alveolar lesions,antemortem tooth loss,breastfeeding and weaning,caries,dental calculus,dental chemistry,dental development,dentin,Dentition,enamel,interpreting oral health,migration,mobility,paleodietary reconstruction,periodontal disease}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/T3NAQGT4/B9780128097380000211.html} } @article{leonardDentalCalculus2015, title = {Plant Microremains in Dental Calculus as a Record of Plant Consumption: {{A}} Test with {{Twe}} Forager-Horticulturalists}, shorttitle = {Plant Microremains in Dental Calculus as a Record of Plant Consumption}, author = { O'Connell, Henry, .}, date = {2015-06-01}, journaltitle = {Journal of Archaeological Science: Reports}, shortjournal = {J Archaeol Sci Rep}, volume = {2}, pages = {449--457}, issn = {2352-409X}, doi = {10.1016/j.jasrep.2015.03.009}, url = {http://www.sciencedirect.com/science/article/pii/S2352409X1500036X}, urldate = {2021-01-19}, abstract = {Starch granules and phytoliths trapped in dental calculus preserve a record of plant consumption. Analysis of these microscopic plant remains has increased in popularity in recent years, providing information on diet that complements dental microwear and stable isotope studies. However, it is unclear how accurately these microremains reflect plant consumption. This study examines how well starch granules and phytoliths in dental calculus from a living population (the Twe) with a well-documented diet capture the range and intensity of plant consumption. We find that plant microremains are a poor predictor of plant consumption on an individual level, but may provide a good signal of plant consumption across a population, as well as evidence for plant processing in the mouth. This is the first study to test how well plant microremains in dental calculus reflect plant consumption in a population with a known diet. Results from this project have implications for interpreting plant microremain data from archaeological dental calculus samples.}, langid = {english}, keywords = {Dental calculus,Foragers,Phytolith,Plant foods,Starch}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/F8UX9BFB/1-s2.0-S2352409X1500036X-mmc2.xlsx;/mnt/hogwarts/Uni/Literature/Zotero/storage/FBD4BHF8/Leonard et al. - 2015 - Plant microremains in dental calculus as a record .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/YYTU7FN9/1-s2.0-S2352409X1500036X-mmc1.docx;/mnt/hogwarts/Uni/Literature/Zotero/storage/KZLJ8KKK/S2352409X1500036X.html} } @article{lieverseDentalHealth2007, title = {Dental Health Indicators of Hunter–Gatherer Adaptation and Cultural Change in {{Siberia}}'s {{Cis-Baikal}}}, author = {Lieverse, Link, Bazaliiskiy, Goriunova, , .}, date = {2007}, journaltitle = {American Journal of Physical Anthropology}, volume = {134}, number = {3}, pages = {323--339}, issn = {1096-8644}, doi = {10.1002/ajpa.20672}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/ajpa.20672}, urldate = {2021-02-24}, abstract = {This investigation of the Cis-Baikal dental record focuses on health and lifestyle reconstruction of the region's mid-Holocene foragers, with particular interest in an apparent fifth millennium BC biocultural hiatus. The four cemetery populations considered represent two distinct biological and cultural groups separated by an apparent 700-year hiatus: the late Mesolithic-early Neolithic Kitoi culture (6800–4900 BC) and the middle Neolithic-early Bronze Age Serovo–Glaskovo cultural complex (4200–1000 BC). Research focuses on the frequency and severity of seven dental health indicators: enamel hypoplasia, caries, alveolar defects, periodontitis, antemortem tooth loss, dental calculus, and dental attrition. Together, these seven indicators provide a basis not only for better understanding mid-Holocene lifeways in the Cis-Baikal but also for independently assessing the relative effectiveness of the different adaptive strategies employed by pre- and posthiatus peoples. Results reveal some discrepancies between the Kitoi and Serovo–Glaskovo, specifically in their relative vulnerability to physiological stress, providing evidence to support previous interpretations of their distinct adaptive regimes (namely the narrower resource base and decreased mobility of the former). Results also suggest that some of the differences observed among the four sites may reflect geographical or environmental factors rather than simply cultural ones. However, despite these distinctions, the overriding trend appears to be one of general continuity, social equality, and good health among all mid-Holocene occupants of the Cis-Baikal, pre- and posthiatus alike. Am J Phys Anthropol 2007. © 2007 Wiley-Liss, Inc.}, langid = {english}, keywords = {alveolar defect,antemortem tooth loss,calculus,caries,dental attrition,enamel hypoplasia,forager,Middle Holocene,periodontitis}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajpa.20672}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/LCX4HYAB/Lieverse et al. - 2007 - Dental health indicators of hunter–gatherer adapta.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/NDU8PANJ/ajpa.html} } @article{lieverseDietAetiology1999, title = {Diet and the Aetiology of Dental Calculus}, author = {Lieverse, .}, date = {1999}, journaltitle = {International Journal of Osteoarchaeology}, volume = {9}, number = {4}, pages = {219--232}, issn = {1099-1212}, doi = {10.1002/(SICI)1099-1212(199907/08)9:4<219::AID-OA475>3.0.CO;2-V}, url = {http://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291099-1212%28199907/08%299%3A4%3C219%3A%3AAID-OA475%3E3.0.CO%3B2-V}, urldate = {2020-05-23}, abstract = {The aetiology of dental calculus formation is not fully understood, but it is known that a number of factors play a role. Generally, anthropologists have overlooked the role of other causative factors in the formation of dental calculus, attributing it almost exclusively to diet, particularly protein consumption. Anthropologists have also oversimplified the role of diet in the formation of dental calculus. This may be due to a general paucity on research on dietary effects on calculus formation, as well as a lack of integration between anthropological and non-anthropological data. Copyright © 1999 \& Sons, Ltd.}, langid = {english}, keywords = {dental calculus,dental plaque,mineralization,tartar}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/\%28SICI\%291099-1212\%28199907/08\%299\%3A4\%3C219\%3A\%3AAID-OA475\%3E3.0.CO\%3B2-V}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/YBEUBSVV/Lieverse - 1999 - Diet and the aetiology of dental calculus.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/KAE8GKCR/08)94219AID-OA4753.0.html} } @article{marshDentalPlaque2005, title = {Dental Plaque: Biological Significance of a Biofilm and Community Life-Style}, shorttitle = {Dental Plaque}, author = {.}, date = {2005}, journaltitle = {Journal of Clinical Periodontology}, volume = {32}, number = {s6}, pages = {7--15}, issn = {1600-051X}, doi = {10.1111/j.1600-051X.2005.00790.x}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1600-051X.2005.00790.x}, urldate = {2022-03-08}, abstract = {Background: Most microorganisms in nature attach to surfaces and form matrix-embedded biofilms. Biofilms are highly structured and spatially organized, and are often composed of consortia of interacting microorganisms, termed microbial communities, the properties of which are more than the sum of the component species. Microbial gene expression alters markedly in biofilms; organisms communicate by gene transfer and by secretion of diffusible signalling molecules. Cells in biofilms are less susceptible to antimicrobial agents. Aim and Materials \& Methods: To comprehensively review the literature to determine whether dental plaque displays properties consistent with those of a typical biofilm and microbial community. Results: Novel microscopic and molecular techniques have demonstrated that plaque has a structured architecture with an extracellular matrix, and a diverse composition (around 50\% of cells are unculturable). The constituent species communicate by gene transfer, by secreted peptides (Gram-positive bacteria) and autoinducer-2 (Gram-positive and Gram-negative bacteria). These organisms are functionally organized for increased metabolic efficiency, greater resistance to stress and for enhanced virulence. Plaque formation has direct and indirect effects on gene expression. Conclusion: Dental plaque displays properties that are typical of biofilms and microbial communities in general, a clinical consequence of which is a reduced susceptibility to antimicrobial agents as well as pathogenic synergism.}, langid = {english}, keywords = {antimicrobial resistance,biofilm,cell signalling,dental plaque,ecology,gene expression,gene transfer,microbial community,review}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1600-051X.2005.00790.x}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/MF3GLQGC/Marsh - 2005 - Dental plaque biological significance of a biofil.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/VV7VBN7Q/j.1600-051X.2005.00790.html} } @article{marshDentalPlaque2006, title = {Dental Plaque as a Biofilm and a Microbial Community – Implications for Health and Disease}, author = {}, date = {2006-06}, journaltitle = {BMC Oral Health}, shortjournal = {BMC Oral Health}, volume = {6}, number = {S1}, pages = {S14}, issn = {1472-6831}, doi = {10.1186/1472-6831-6-S1-S14}, url = {https://bmcoralhealth.biomedcentral.com/articles/10.1186/1472-6831-6-S1-S14}, urldate = {2020-06-17}, abstract = {Dental plaque is a structurally- and functionally-organized biofilm. Plaque forms in an ordered way and has a diverse microbial composition that, in health, remains relatively stable over time (microbial homeostasis). The predominant species from diseased sites are different from those found in healthy sites, although the putative pathogens can often be detected in low numbers at normal sites. In dental caries, there is a shift toward community dominance by acidogenic and acidtolerating species such as mutans streptococci and lactobacilli, although other species with relevant traits may be involved. Strategies to control caries could include inhibition of biofilm development (e.g. prevention of attachment of cariogenic bacteria, manipulation of cell signaling mechanisms, delivery of effective antimicrobials, etc.), or enhancement of the host defenses. Additionally, these more conventional approaches could be augmented by interference with the factors that enable the cariogenic bacteria to escape from the normal homeostatic mechanisms that restrict their growth in plaque and out compete the organisms associated with health. Evidence suggests that regular conditions of low pH in plaque select for mutans streptococci and lactobacilli. Therefore, the suppression of sugar catabolism and acid production by the use of metabolic inhibitors and nonfermentable artificial sweeteners in snacks, or the stimulation of saliva flow, could assist in the maintenance of homeostasis in plaque. Arguments will be presented that an appreciation of ecological principles will enable a more holistic approach to be taken in caries control.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/D6QHAXU6/Marsh - 2006 - Dental plaque as a biofilm and a microbial communi.pdf} } @incollection{marshDentalPlaque2016, title = {Dental {{Plaque}}}, booktitle = {Marsh and {{Martin}}'s {{Oral Microbiology}}}, author = {. and Lewis, . and and Williams, . and }, date = {2016-05-13}, edition = {6th Edition}, pages = {81--111}, publisher = {{Elsevier Health Sciences}}, abstract = {Now expanded with the latest information of relevance to current dental practice, Oral Microbiology retains its unique ecological approach to the subject which helps the reader determine whether an organism will have a pathogenic or commensal relationship at a given site. In the new edition, greater emphasis is placed on the role of current molecular biology techniques in the understanding of oral microbes. The book also provides insight into current therapeutic and prophylactic antibiotic use, infection control, and the relationships between oral and general health. Oral Microbiology provides comprehensive coverage of the subject which will be essential to readers with a specific interest in dentistry as well as those with a more general interest in host-microbe interactions and in microbial ecology. The book is suitable for undergraduate and postgraduate dental students, research workers, and a wide range of clinical dental professionals. Full coverage of the latest molecular biology techniques which have revolutionized our knowledge of oral microbes Exploration of the biological and clinical significance of the oral microflora in the form of a biofilm on dental and mucosal surfaces Contemporary views on therapeutic and prophylactic antibiotic use, infection control, and the relationships between oral and general health}, isbn = {978-0-7020-6174-5}, langid = {english}, keywords = {Medical / Dentistry / General,Medical / Infectious Diseases} } @article{marshPhysiologicalApproaches1997, title = {Physiological {{Approaches}} to the {{Control}} of {{Oral Biofilms}}}, author = {Marsh, . and Bradshaw, .}, date = {1997-04-01}, journaltitle = {Advances in Dental Research}, shortjournal = {Adv Dent Res.}, volume = {11}, number = {1}, pages = {176--185}, publisher = {{SAGE Publications Inc}}, issn = {0895-9374}, doi = {10.1177/08959374970110010901}, url = {https://doi.org/10.1177/08959374970110010901}, urldate = {2021-03-17}, abstract = {Evidence that physiological strategies may be potential routes for oral biofilm control has come from (i) observations of the variations in the intra-oral distribution of members of the resident oral microflora, (ii) changes in plaque composition in health and disease, and (iii) data from laboratory model systems. Key physiological factors that were identified as significant in modulating the microflora included the local pH, redox potential (Eh), and nutrient availability. Increases in mutans streptococci and lactobacilli occur at sites with caries; growth of these species is selectively enhanced at low pH. In contrast, periodontal diseases are associated with plaque accumulation, followed by an inflammatory host response. The increases in Gram-negative, proteolytic, and obligately anaerobic bacteria reflect a low redox potential and a change in nutrient status due to the increased flow of gingival crevicular fluid (GCF). Consequently, physiological strategies for oral biofilm control should focus on reducing the frequency of low pH in plaque by (i) inhibiting acid production, (ii) using sugar substitutes, and (iii) promoting alkali generation from arginine or urea supplements. Similarly, strategies to make the pocket environment less favorable to periodontopathogens include (i) anti-inflammatory agents to reduce the flow of (and hence nutrient supply by) GCF, (ii) bacterial protease inhibitors, and (iii) redox agents to raise the Ehh locally. Most laboratory and clinical findings support the concept of physiological control. However, some data suggest that the ordered structure and metabolically interactive organization of mature dental plaque could generate a community with a high level of homeostasis that is relatively resistant to deliberate external manipulation.}, langid = {english}, keywords = {antimicrobial agents,arginine,biofilm,ecology,fluoride,homeostasis.,pH,redox potential,urea}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/646GAGB9/08959374970110010901.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/RHXNUA57/Marsh and Bradshaw - 1997 - Physiological Approaches to the Control of Oral Bi.pdf} } @article{mickleburghNewInsights2012, title = {New Insights into the Consumption of Maize and Other Food Plants in the Pre-{{Columbian Caribbean}} from Starch Grains Trapped in Human Dental Calculus}, author = {Mickleburgh, . and Pagán-Jiménez, .}, date = {2012}, journaltitle = {Journal of Archaeological Science}, volume = {39}, number = {7}, pages = {2468--2478}, issn = {03054403}, doi = {10.1016/j.jas.2012.02.020} } @article{middletonOpalPhytoliths1994, title = {Extraction of {{Opal Phytoliths}} from {{Herbivore Dental Calculus}}}, author = {Middleton, . and }, date = {1994-07-01}, journaltitle = {Journal of Archaeological Science}, shortjournal = {J Archaeol Sci}, volume = {21}, number = {4}, pages = {469--473}, issn = {0305-4403}, doi = {10.1006/jasc.1994.1046}, url = {http://www.sciencedirect.com/science/article/pii/S0305440384710466}, urldate = {2020-05-27}, abstract = {Opal phytoliths can easily be extracted from herbivore dental calculus and examined to gain data for the reconstruction of prehistoric herbivore diet. The extraction process is a simple, quick and inexpensive three-step procedure. It consists of one or two distilled water washes to control and assess contamination and a final wash with dilute hydrochloric acid. A pilot study conducted on barnyard animals (cow, sheep and pig) from the American Colonial period site of Hampton, Virginia, demonstrates that dental calculus phytolith assemblages can provide data on herbivore diet that can be used to reconstruct livestock management practices and ecological change.}, langid = {english}, keywords = {Dental Calculus,Palaeodiet,Phytoliths}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/FQ77HW3N/Middleton and Rovner - 1994 - Extraction of Opal Phytoliths from Herbivore Denta.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/FX8A4CMT/S0305440384710466.html} } @article{middletonVitroCalculus1965, title = {Human Salivary Proteins and Artificial Calculus Formation in Vitro}, author = {.}, date = {1965-03}, journaltitle = {Archives of Oral Biology}, shortjournal = {Archives of Oral Biology}, volume = {10}, number = {2}, pages = {227--235}, issn = {00039969}, doi = {10.1016/0003-9969(65)90024-5}, url = {https://linkinghub.elsevier.com/retrieve/pii/0003996965900245}, urldate = {2021-03-04}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/W94ZX97L/Mlddleton - 1965 - Human salivary proteins and artificial calculus fo.pdf} } @article{nikitkovaStarchBiofilms2013, title = {Taking the {{Starch}} out of {{Oral Biofilm Formation}}: {{Molecular Basis}} and {{Functional Significance}} of {{Salivary}} α-{{Amylase Binding}} to {{Oral Streptococci}}}, shorttitle = {Taking the {{Starch}} out of {{Oral Biofilm Formation}}}, author = {Nikitkova, . and Haase, . and Scannapieco, .}, date = {2013-01-15}, journaltitle = {Applied and Environmental Microbiology}, shortjournal = {Appl. Environ. Microbiol.}, volume = {79}, number = {2}, eprint = {23144140}, eprinttype = {pmid}, pages = {416--423}, publisher = {{American Society for Microbiology}}, issn = {0099-2240, 1098-5336}, doi = {10.1128/AEM.02581-12}, url = {https://aem.asm.org/content/79/2/416}, urldate = {2021-04-03}, abstract = {α-Amylase-binding streptococci (ABS) are a heterogeneous group of commensal oral bacterial species that comprise a significant proportion of dental plaque microfloras. Salivary α-amylase, one of the most abundant proteins in human saliva, binds to the surface of these bacteria via specific surface-exposed α-amylase-binding proteins. The functional significance of α-amylase-binding proteins in oral colonization by streptococci is important for understanding how salivary components influence oral biofilm formation by these important dental plaque species. This review summarizes the results of an extensive series of studies that have sought to define the molecular basis for α-amylase binding to the surface of the bacterium as well as the biological significance of this phenomenon in dental plaque biofilm formation.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/2XTQBQVK/Nikitkova et al. - 2013 - Taking the Starch out of Oral Biofilm Formation M.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/7CXKNEX3/416.html} } @book{ortnerIdentificationPathological2003, title = {Identification of {{Pathological Conditions}} in {{Human Skeletal Remains}}}, author = {.}, date = {2003}, publisher = {{Academic Press}}, location = {{London}} } @article{petersonViscoelasticityBiofilms2015, title = {Viscoelasticity of Biofilms and Their Recalcitrance to Mechanical and Chemical Challenges}, author = {Peterson, . and He, Yan and Ren, Yijin and Zerdoum, Aidan and Libera, . and Sharma, . and , Arie-Jan and and and , . and Busscher, .}, options = {useprefix=true}, date = {2015-03-01}, journaltitle = {FEMS Microbiology Reviews}, volume = {39}, number = {2}, pages = {234--245}, issn = {1574-6976}, doi = {10.1093/femsre/fuu008}, url = {http://academic.oup.com/femsre/article/39/2/234/635745/Viscoelasticity-of-biofilms-and-their}, urldate = {2020-11-04}, langid = {english}, keywords = {biofilm,detachment,extracellular polymeric substances,structure}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/248Q8KCT/Peterson et al. - 2015 - Viscoelasticity of biofilms and their recalcitranc.pdf} } @article{pilloudOutliningDefinition2019, title = {Outlining a {{Definition}} of {{Oral Health}} within the {{Study}} of {{Human Skeletal Remains}}: {{Defining Oral Health}}}, shorttitle = {Outlining a {{Definition}} of {{Oral Health}} within the {{Study}} of {{Human Skeletal Remains}}}, author = {. and .}, date = {2019-07-19}, journaltitle = {Dental Anthropology Journal}, volume = {32}, number = {2}, pages = {3--11}, issn = {1096-9411}, doi = {10.26575/daj.v32i2.297}, url = {https://journal.dentalanthropology.org}, urldate = {2021-12-08}, abstract = {The term oral health is regularly used in bioarchaeological research to discuss a myriad of pathological conditions of the oral cavity. However, there is very little consensus on what conditions should be included in such a study, and some of the conditions are at odds with those in the clinical literature. \ In this manuscript, we outline the clinical definition of oral health and develop a strategy in which bioarchaeology can address this type of research.\  We argue that the terms dental disease and/or pathological conditions of the oral cavity should be used in lieu of oral health.\  Various conditions that can be included in such research are outlined.\  Finally, definitions, clinical etiologies, and recording schema for these conditions are discussed as relevant to bioarchaeological studies.}, issue = {2}, langid = {english}, keywords = {antemortem tooth loss,calculus,dental caries,hypercementosis,linear enamel hypoplasia,periapical lesions,periodontal disease}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/6JTDYDUN/Pilloud and Fancher - 2019 - Outlining a Definition of Oral Health within the S.pdf} } @article{pipernoStarchGrains2008, title = {Starch Grains on Human Teeth Reveal Early Broad Crop Diet in Northern {{Peru}}}, author = {. and .}, date = {2008-12-16}, journaltitle = {Proceedings of the National Academy of Sciences}, shortjournal = {Proceedings of the National Academy of Sciences}, volume = {105}, number = {50}, pages = {19622--19627}, issn = {0027-8424, 1091-6490}, doi = {10.1073/pnas.0808752105}, url = {http://www.pnas.org/cgi/doi/10.1073/pnas.0808752105}, urldate = {2021-07-19}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/K9NM4C93/Piperno and Dillehay - 2008 - Starch grains on human teeth reveal early broad cr.pdf} } @article{powerChimpCalculus2015, title = {Dental Calculus Evidence of {{Tai Forest Chimpanzee}} Plant Consumption and Life History Transitions}, author = {. and .}, date = {2015-10-19}, journaltitle = {Scientific Reports}, shortjournal = {Sci Rep}, volume = {5}, pages = {15161}, issn = {2045-2322 (Electronic) 2045-2322 (Linking)}, doi = {10.1038/srep15161}, abstract = {Dental calculus (calcified dental plaque) is a source of multiple types of data on life history. Recent research has targeted the plant microremains preserved in this mineralised deposit as a source of dietary and health information for recent and past populations. However, it is unclear to what extent we can interpret behaviour from microremains. Few studies to date have directly compared the microremain record from dental calculus to dietary records, and none with long-term observation dietary records, thus limiting how we can interpret diet, food acquisition and behaviour. Here we present a high-resolution analysis of calculus microremains from wild chimpanzees (Pan troglodytes verus) of Tai National Park, Cote d'Ivoire. We test microremain assemblages against more than two decades of field behavioural observations to establish the ability of calculus to capture the composition of diet. Our results show that some microremain classes accumulate as long-lived dietary markers. Phytolith abundance in calculus can reflect the proportions of plants in the diet, yet this pattern is not true for starches. We also report microremains can record information about other dietary behaviours, such as the age of weaning and learned food processing techniques like nut-cracking.}, pmcid = {PMC4611876}, keywords = {*Dental Calculus,*Herbivory,*Pan troglodytes,Animal Feed,Animals,Behavior; Animal,Cote d'Ivoire,Female,Male,Models; Theoretical}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/FLN2SC5G/Power et al. - 2015 - Dental calculus evidence of Tai Forest Chimpanzee .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/GQC34VD5/srep15161.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/T4FAS23N/41598_2015_BFsrep15161_MOESM2_ESM.xls} } @article{powerRepresentativenessDental2021, title = {The Representativeness of the Dental Calculus Dietary Record: Insights from {{Taï}} Chimpanzee Faecal Phytoliths}, shorttitle = {The Representativeness of the Dental Calculus Dietary Record}, author = {Power, . and , }, date = {2021-05-29}, journaltitle = {Archaeological and Anthropological Sciences}, shortjournal = {Archaeol Anthropol Sci}, volume = {13}, number = {6}, pages = {104}, issn = {1866-9565}, doi = {10.1007/s12520-021-01342-z}, url = {https://doi.org/10.1007/s12520-021-01342-z}, urldate = {2021-06-01}, abstract = {In recent years, new applications of microremain dietary analysis using dental calculus as a source of dietary data on ancient human subsistence and behaviours have accelerated. The dental calculus of contemporary human and non-human populations with known diets have been used as reference datasets, including the chimpanzees of Taï National Park (Côte d'Ivoire), but explaining the preservation mechanism involved is challenged by our incomplete knowledge of the microremain content within the diets of these reference populations and our rudimentary information on microremain incorporation into dental calculus. Here, we analyse phytoliths in faecal samples to assess to what extent plant phytoliths of a diet are reflected in the dental calculus as well as in the egested faeces. In this study, we identify and document the faecal phytolith assemblages as an indicator of plant consumption in two Western chimpanzees of the Taï National Park (Côte d'Ivoire) before (wet season), during (dry season) and after (dry season) a dust-rich period. Moreover, observational dietary records of these two individuals were compiled to improve the interpretability of this dental calculus phytolith dataset. The faecal phytolith assemblages vary significantly across samples in terms of abundance and diversity. The most common phytolith morphotypes were eudicot plates, single-cell and multi-cell tracheids, monocot rugulose and echinate spheroids and, to a lesser extent, unspecified thick and thin elongates. High loads of grit and other micro-remains (e.g. diatoms) are found during the dry period. Using observational dietary records as a starting point and our faecal results as a terminus, we consider how dental calculus can accumulate phytoliths. Our findings enable identification of the phytolith morphotypes that are under-represented in dental calculus, which is highly informative for future dental calculus research strategies.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/TKTTRR9R/Power et al. - 2021 - The representativeness of the dental calculus diet.pdf} } @article{radiniDirtyTeeth2022, title = {Beyond Dirty Teeth: {{Integrating}} Dental Calculus Studies with Osteoarchaeological Parameters}, shorttitle = {Beyond Dirty Teeth}, author = { and }, date = {2022-03-18}, journaltitle = {Quaternary International}, shortjournal = {Quaternary International}, issn = {1040-6182}, doi = {10.1016/j.quaint.2022.03.003}, url = {https://www.sciencedirect.com/science/article/pii/S1040618222000726}, urldate = {2022-03-18}, abstract = {The study of ancient human dental calculus (mineralized dental plaque, also known as tartar) is becoming increasingly important in osteoarchaeology, human palaeoecology and environmental archaeology. Microremains of different origin (e.g. starch granules, pollen, phytoliths, feather barbules) as well as biomolecules and chemical compounds retrieved from its mineral matrix may represent an important link between past humans and their physical, biological and social environment, but they are rarely fully linked to the evidence from skeletal remains. This paper critically reviews the lines of evidence retrieved from dental calculus in relation to osteoarchaeological parameters, employing macroscopic, microscopic and biomolecular approaches, assessing synergy potential and limitations. The scope of this paper is also to contribute to the building of a much needed theoretical framework in this emerging subfield.}, langid = {english}, keywords = {Dental calculus,Environment,Life history,Osteoarchaeology}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/VHKP55S8/Radini and Nikita - 2022 - Beyond dirty teeth Integrating dental calculus st.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/RP8DC6DD/S1040618222000726.html} } @article{radiniFoodPathways2017, title = {Beyond Food: {{The}} Multiple Pathways for Inclusion of Materials into Ancient Dental Calculus}, author = {. and . and . and . and .}, date = {2017-01}, journaltitle = {American Journal of Physical Anthropology}, shortjournal = {Am J Phys Anthropol}, volume = {162}, pages = {71--83}, issn = {1096-8644 (Electronic) 0002-9483 (Linking)}, doi = {10.1002/ajpa.23147}, abstract = {Dental calculus (mineralized dental plaque) was first recognised as a potentially useful archaeological deposit in the 1970s, though interest in human dental calculus as a resource material has increased sharply in the past few years. The majority of recent research has focused on the retrieval of plant microfossils embedded in its matrix and interpretation of these finds as largely the result of deliberate consumption of plant-derived food. However, while most of the material described in published works does represent food, dental calculus is in fact a "depositional environment" as material can enter the mouth from a range of sources. In this respect, it therefore represents an archaeological deposit that can also contain extensive non-dietary debris. This can comprise a wide variety of cultural and environmental material which reaches the mouth and can become embedded in dental calculus through alternative pathways. Here, we explore the human behaviors and activities besides eating that can generate a flux of particles into the human mouth, the broad range of additional cultural and environmental information that can be obtained through the analysis and contextualisation of this material, and the implications of the additional pathways by which material can become embedded in dental calculus.}, keywords = {*Dental Calculus,Archaeology,dental calculus,diet,Diet/*history,Environment,Food/*history,Fossils,History; Ancient,Humans,microfossils}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/8NXTX9PF/Radini et al. - 2017 - Beyond food The multiple pathways for inclusion o.pdf} } @incollection{robertsDentalDisease2007, title = {Dental {{Disease}}}, booktitle = {The {{Archaeology}} of {{Disease}}}, author = { and }, date = {2007}, edition = {3rd Edition}, pages = {63--83}, publisher = {{Cornell University Press}} } @article{sagneStudiesPeriodontal1977, title = {Studies of the {{Periodontal Status}} of a {{Medieval Population}}}, author = {. and .}, date = {1977-01}, journaltitle = {Dentomaxillofacial Radiology}, shortjournal = {Dentomaxillofacial Radiology}, volume = {6}, number = {1}, pages = {46--52}, issn = {0250-832X, 1476-542X}, doi = {10.1259/dmfr.1977.0006}, url = {http://www.birpublications.org/doi/10.1259/dmfr.1977.0006}, urldate = {2022-04-04}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/5PNC6AFQ/Sagne and Olsson - 1977 - Studies of the Periodontal Status of a Medieval Po.pdf} } @incollection{scottBriefHistoryDental2015, title = {A {{Brief History}} of {{Dental Anthropology}}}, booktitle = {A {{Companion}} to {{Dental Anthropology}}}, author = {}, editor = {. and }, date = {2015}, pages = {7--17}, publisher = {{John Wiley \& Sons, Ltd}}, location = {{Chichester}}, doi = {10.1002/9781118845486.ch18}, url = {http://onlinelibrary.wiley.com/doi/abs/10.1002/9781118845486.ch18}, urldate = {2021-04-21}, isbn = {978-1-118-84548-6}, langid = {english}, keywords = {Arizona State University Dental Anthropology System,distance statistics,genetic relatedness,P henetic affinities,qualitative and quantitative analyses}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/9781118845486.ch18} } @article{sissonsMultistationPlaque1991, title = {A {{Multi-station Dental Plaque Microcosm}} ({{Artificial Mouth}}) for the {{Study}} of {{Plaque Growth}}, {{Metabolism}}, {{pH}}, and {{Mineralization}}:}, shorttitle = {A {{Multi-station Dental Plaque Microcosm}} ({{Artificial Mouth}}) for the {{Study}} of {{Plaque Growth}}, {{Metabolism}}, {{pH}}, and {{Mineralization}}}, author = {. and . and . and .}, date = {1991}, journaltitle = {Journal of Dental Research}, shortjournal = {J Dent Res}, publisher = {{SAGE PublicationsSage CA: Los Angeles, CA}}, doi = {10.1177/00220345910700110301}, url = {https://journals.sagepub.com/doi/10.1177/00220345910700110301}, urldate = {2020-05-15}, abstract = {A plaque growth chamber was developed for long-term growth of five separate plaques from the same plaque or saliva sample under identical conditions of temperat...}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/223FAUK9/Sissons et al. - 1991 - A Multi-station Dental Plaque Microcosm (Artificia.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/42ZPNDRU/00220345910700110301.html} } @article{sotoStructuralCharacterization2019, title = {Structural Characterization and Decontamination of Dental Calculus for Ancient Starch Research}, author = {, Jamie and Clarke, Siobhán and Crowther, Alison and Covelli, Danielle and Favreau, Julien and Itambu, Makarius and Larter, Steve and Lee, Patrick and Lozano, Marina and Maley, Jason and Mwambwiga, Aloyce and Patalano, Robert and Sammynaiken, Ramaswami and Vergès, . and Zhu, Jianfeng and Mercader, Julio}, date = {2019-09-01}, journaltitle = {Archaeological and Anthropological Sciences}, shortjournal = {Archaeol Anthropol Sci}, volume = {11}, number = {9}, pages = {4847--4872}, issn = {1866-9565}, doi = {10.1007/s12520-019-00830-7}, url = {https://doi.org/10.1007/s12520-019-00830-7}, urldate = {2022-04-26}, abstract = {Ancient dental calculus research currently relies on destructive techniques whereby archeological specimens are broken down to determine their contents. Two strategies that could partly remediate a permanent loss of the original sample and enhance future analysis and reproducibility include (1) structural surface characterization through spectroscopy along with crystallographic and spectroscopic analysis of its molecular structure, and (2) surface decontamination protocols in which the efficacy of cleaning dental calculus prior to extraction is demonstrated. Dental calculus provides ancient starch research a niche where granules may be adsorbed to minerals, coated, overgrown, entrapped, and/or protected from chemical degradation. While encapsulation offers protection from degradation, it does not shield the sample’s surface from contamination. The most common approach to retrieving microbotanical particles from archeological calculus has been the direct decalcification of the sample, after a cleaning stage variously consisting of immersion in water, acids, and mechanical dislodgment via gas, sonication, and/or toothbrushes. Little is known about the efficiency of these methods for a complete removal of sediment/soil and unrelated microbotanical matter. In this paper, controlled laboratory experimentation leads to chemical structural characterization and a decontamination protocol to eradicate starch granules. Several concentrations of acids, bases, and enzymes were tested at intervals to understand their potential to gelatinize and fully destroy starch granules; arriving at a procedure that effectively eradicates modern starch prior to dissolution without damaging the matrix or entrapped starch microremains. This is the first attempt at creating synthetic calculus to understand and systematically test effective decontamination protocols for ancient starch research.}, langid = {english}, keywords = {Ancient dental calculus,Ancient starch research,Decontamination prior to decalcification,P-XRD,Raman,Starch contamination,Structural chemical characterization,XPS}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/DHK3F7JV/Soto et al. - 2019 - Structural characterization and decontamination of.pdf} } @incollection{squierOralMucosa1998, title = {Oral {{Mucosa}}}, booktitle = {Oral {{Histology}}: {{Development}}, {{Structure}}, and {{Function}}}, author = {. and .}, editor = {.}, date = {1998}, edition = {5}, pages = {345--385}, publisher = {{Mosby}}, location = {{London}}, isbn = {0-8151-2952-1}, langid = {english} } @article{tanBacterialViability2004, title = {Study of {{Bacterial Viability}} within {{Human Supragingival Dental Calculus}}}, author = {. and , , .}, date = {2004}, journaltitle = {Journal of Periodontology}, volume = {75}, number = {1}, pages = {23--29}, issn = {1943-3670}, doi = {10.1902/jop.2004.75.1.23}, url = {http://aap.onlinelibrary.wiley.com/doi/abs/10.1902/jop.2004.75.1.23}, urldate = {2020-06-24}, abstract = {Background: There is evidence that supragingival calculus contains unmineralized channels and lacunae. The purpose of this study was to investigate the viability of bacteria within these areas. Methods: Supragingival calculus harvested from patients with moderate to severe chronic periodontitis was immediately frozen to –70°C. Six samples were cryosectioned, stained with a bacterial viability kit, and examined with fluorescence microscopy. Controls comprised heat treatment of cryosections prior to staining. Four additional samples were stained and examined whole a confocal laser scanning microscope (CLSM). Nine additional samples were prepared for bacterial culture, after initial irradiation with ultraviolet light to kill viable organisms on the covering plaque layer. Test samples were crushed to expose internal bacteria, while two controls were used without crushing. Results: Viable bacteria, as identified using the bacterial viability stain, were found within cavities/lacunae in supragingival calculus cryosections. Similar results were obtained from whole calculus samples using CLSM. Of the nine experimental samples where bacterial culture was attempted, five provided positive bacterial culture under both aerobic and anaerobic conditions; one showed positive growth under aerobic conditions only; while one showed no bacterial growth. The controls showed no bacterial growth. Conclusions: From this study, it appears that viable aerobic and anaerobic bacteria may be present within supragingival calculus, specifically within the internal channels and lacunae. Clinically, this may be important, since incomplete removal of supragingival calculus may expose these reservoirs of possible pathogenic bacteria and be a factor in the recurrence of periodontal diseases after treatment. J Periodontol 2004;75:23-29.}, langid = {english}, keywords = {aerobic/growth,anaerobic/growth,bacteria,Bacteria,dental calculus,supragingival}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1902/jop.2004.75.1.23}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/A86UBYQB/Tan et al. - 2004 - Study of Bacterial Viability within Human Supragin.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/GV7CZV99/jop.2004.75.1.23.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/UPQBS8UE/jop.2004.75.1.html} } @article{tanCalculusUltrastructure2004, title = {A Preliminary Investigation into the Ultrastructure of Dental Calculus and Associated Bacteria}, author = {Tan, . and Gillam, . and . and .}, date = {2004}, journaltitle = {Journal of Clinical Periodontology}, shortjournal = {J Clin Periodontol}, volume = {31}, number = {5}, pages = {364--369}, issn = {1600-051X}, doi = {10.1111/j.1600-051X.2004.00484.x}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1600-051X.2004.00484.x}, urldate = {2020-06-18}, abstract = {Introduction: Though dental calculus is generally recognised as comprising mineralised bacteria, areas of non-mineralised bacteria may be present. Aim: To investigate the ultrastructure of non-decalcified young and mature supragingival calculus and subgingival calculus, and the possible presence of internal viable bacteria. Materials and methods: Supragingival calculus was harvested from five patients, 9–10 weeks after scaling and root debridement. Five samples of mature supragingival and subgingival calculus were taken from patients presenting with adult periodontitis. Specimens were fixed and embedded for transmission electron microscopy. Results: The ultrastructure of young and mature supragingival calculus was similar with various large and small crystal types. Non-mineralised channels were observed extending into the calculus, often joining extensive lacunae, both containing intact non-mineralised coccoid and rod-shaped microorganisms. Subgingival calculus possessed more uniform mineralisation without non-mineralised channels and lacunae. Conclusion: Supragingival calculus contains non-mineralised areas which contain bacteria and other debris. The viability of the bacteria, and their identification could not be determined in this preliminary investigation. As viable bacteria within these lacunae may provide a source of re-infection, further work needs to be done to identify the bacteria in the lacunae, and to determine their viability.}, langid = {english}, keywords = {bacteria,calculus,plaque,subgingival,supragingival}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/VIZ82QJA/Tan et al. - 2004 - A preliminary investigation into the ultrastructur.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/PTJGMC84/Tan et al. - 2004 - A preliminary investigation into the ultrastructur.html} } @book{townsendDentalAnthropology2012, title = {New {{Directions}} in {{Dental Anthropology}}: {{Paradigms}}, {{Methodologies}} and {{Outcomes}}}, shorttitle = {New {{Directions}} in {{Dental Anthropology}}}, editor = { and }, date = {2012}, publisher = {{The University of Adelaide Press}}, doi = {10.1017/9780987171870}, url = {https://www.cambridge.org/core/books/new-directions-in-dental-anthropology/7D25DC7D6C4345F71FE7028D5DAC5EBF}, urldate = {2022-02-04}, abstract = {In recent years, migration, as well as increases and decreases in the size of different human populations, have been evident as a result of globalisation. Dental features are also changing associated with changes in nutritional status, different economic or social circumstances, and intermarriage between peoples. Dental anthropological studies have explored these changes with the use of advanced techniques and refined methodologies. New paradigms are also evolving in the field of dental anthropology.}, isbn = {978-0-9871718-8-7}, keywords = {dental anthropology}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/4EP9EZG4/Townsend et al. - 2012 - New Directions in Dental Anthropology Paradigms, .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/CC7RZN8X/7D25DC7D6C4345F71FE7028D5DAC5EBF.html} } @article{trompEDTACalculus2017, title = {{{EDTA}} Decalcification of Dental Calculus as an Alternate Means of Microparticle Extraction from Archaeological Samples}, author = { }, date = {2017-08-01}, journaltitle = {Journal of Archaeological Science: Reports}, shortjournal = {Journal of Archaeological Science: Reports}, volume = {14}, pages = {461--466}, issn = {2352-409X}, doi = {10.1016/j.jasrep.2017.06.035}, url = {http://www.sciencedirect.com/science/article/pii/S2352409X17300998}, urldate = {2021-01-28}, abstract = {Dental calculus studies, though becoming more common for addressing a range of archaeological questions, are still in their infancy. Dental calculus is a mineralised biofilm that forms on the surface of teeth, with the potential to encase anything that comes into contact with the mouth. Dental calculus has provided information on the lifeways of past people including diet, health, disease, trade and fabrication. To extract this microarchaeological information from dental calculus it must first be processed, either mechanically (with a mortar and pestle) or chemically (generally with hydrochloric acid or HCl), in order to free the microparticles and biomolecular structures encased within. Until now, ethylenediaminetetraacetic acid (EDTA) has not been used for microparticle extraction. Here we present data that demonstrate EDTA is an equally and potentially more effective decalcifying agent for extracting microparticles than previous methods using samples from the Southwest Pacific and Ireland. EDTA has the added benefit of being an appropriate for the first step in genomic and proteomic sample processing.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/MIUJTFJ3/Tromp et al. - 2017 - EDTA decalcification of dental calculus as an alte.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/TWF8XDL3/S2352409X17300998.html} } @article{uzelMicrobialShifts2011, title = {Microbial Shifts during Dental Biofilm Re-Development in the Absence of Oral Hygiene in Periodontal Health and Disease}, author = {. and . and . and Song, . and Socransky, . and .}, date = {2011}, journaltitle = {Journal of Clinical Periodontology}, volume = {38}, number = {7}, pages = {612--620}, issn = {1600-051X}, doi = {10.1111/j.1600-051X.2011.01730.x}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1600-051X.2011.01730.x}, urldate = {2022-03-08}, abstract = {Uzel NG, Teles FR, Teles RP, , , Socransky SS, Haffajee AD. Microbial shifts during dental biofilm re-development in the absence of oral hygiene in periodontal health and disease. J Clin Peridontol 2011; doi: 10.1111/j.1600-051X.2011.01730.x. Abstract Aim: To monitor microbial shifts during dental biofilm re-development. Materials and methods: Supra- and subgingival plaque samples were taken separately from 28 teeth in 38 healthy and 17 periodontitis subjects at baseline and immediately after tooth cleaning. Samples were taken again from seven teeth in randomly selected quadrants during 1, 2, 4 and 7 days of no oral hygiene. Samples were analysed using checkerboard DNA–DNA hybridization. Species counts were averaged within subjects at each time point. Significant differences in the counts between healthy and periodontitis subjects were determined using the Mann–Whitney test. Results: The total supra- and subgingival counts were significantly higher in periodontitis on entry and reached or exceeded the baseline values after day 2. Supragingival counts of Veillonella parvula, Fusobacterium nucleatum ss vincentii and Neisseria mucosa increased from 2 to 7 days. Subgingival counts were greater for Actinomyces, green and orange complex species. Significant differences between groups in supragingival counts occurred for 17 of 41 species at entry, 0 at day 7; for subgingival plaque, these values were 39/41 taxa at entry, 17/41 at day 7. Conclusions: Supragingival plaque re-development was similar in periodontitis and health, but subgingival species recolonization was more marked in periodontitis.}, langid = {english}, keywords = {biofilms,health,oral bacteria,periodontal,periodontitis,subgingival,supragingival}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1600-051X.2011.01730.x}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/5DNXKRHQ/Uzel et al. - 2011 - Microbial shifts during dental biofilm re-developm.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/6U6JGEL6/j.1600-051X.2011.01730.html} } @article{vandermeerschMiddlePaleolithic1994, title = {Middle {{Paleolithic Dental Bacteria From Kebara}}, {{Israel}}}, author = {. and . and . and . and . and . and .}, date = {1994}, journaltitle = {Comptes Rendus De L Academie Des Sciences Serie Ii}, volume = {319}, number = {6}, pages = {727--731}, url = {https://weizmann.esploro.exlibrisgroup.com/esploro/outputs/journalArticle/MIDDLE-PALEOLITHIC-DENTAL-BACTERIA-FROM-KEBARA/993266802803596}, urldate = {2022-04-08}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/KHBTST24/993266802803596.html} } @article{velskoMicrobialDifferences2019, title = {Microbial Differences between Dental Plaque and Historic Dental Calculus Are Related to Oral Biofilm Maturation Stage}, author = {. and , . and and Hagan, . and Frantz, . and and }, date = {2019-12}, journaltitle = {Microbiome}, shortjournal = {Microbiome}, volume = {7}, number = {1}, pages = {102}, issn = {2049-2618}, doi = {10.1186/s40168-019-0717-3}, url = {https://microbiomejournal.biomedcentral.com/articles/10.1186/s40168-019-0717-3}, urldate = {2021-08-30}, abstract = {Background: Dental calculus, calcified oral plaque biofilm, contains microbial and host biomolecules that can be used to study historic microbiome communities and host responses. Dental calculus does not typically accumulate as much today as historically, and clinical oral microbiome research studies focus primarily on living dental plaque biofilm. However, plaque and calculus reflect different conditions of the oral biofilm, and the differences in microbial characteristics between the sample types have not yet been systematically explored. Here, we compare the microbial profiles of modern dental plaque, modern dental calculus, and historic dental calculus to establish expected differences between these substrates. Results: Metagenomic data was generated from modern and historic calculus samples, and dental plaque metagenomic data was downloaded from the Human Microbiome Project. Microbial composition and functional profile were assessed. Metaproteomic data was obtained from a subset of historic calculus samples. Comparisons between microbial, protein, and metabolomic profiles revealed distinct taxonomic and metabolic functional profiles between plaque, modern calculus, and historic calculus, but not between calculus collected from healthy teeth and periodontal disease-affected teeth. Species co-exclusion was related to biofilm environment. Proteomic profiling revealed that healthy tooth samples contain low levels of bacterial virulence proteins and a robust innate immune response. Correlations between proteomic and metabolomic profiles suggest co-preservation of bacterial lipid membranes and membrane-associated proteins. Conclusions: Overall, we find that there are systematic microbial differences between plaque and calculus related to biofilm physiology, and recognizing these differences is important for accurate data interpretation in studies comparing dental plaque and calculus.}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/6ISX7GQS/Velsko et al. - 2019 - Microbial differences between dental plaque and hi.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/HREZWT64/s40168-019-0717-3.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/STQBSWRZ/s40168-019-0717-3.pdf} } @book{waldronPalaeopathology2020, title = {Palaeopathology}, author = {}, date = {2020}, publisher = {{Cambridge University Press}} } @article{warinnerEvidenceMilk2014, title = {Direct Evidence of Milk Consumption from Ancient Human Dental Calculus}, author = { . and . and . and . and . and . and . and . and . and Fotakis, A. and Christensen, . and . and Liebert, A. and Montalva, N. and Fiddyment, S. and . and Mackie, M. and Canci, A. and . and . and . and .}, date = {2014-11-27}, journaltitle = {Scientific Reports}, shortjournal = {Sci Rep}, volume = {4}, pages = {7104}, issn = {2045-2322 (Electronic) 2045-2322 (Linking)}, doi = {10.1038/srep07104}, abstract = {Milk is a major food of global economic importance, and its consumption is regarded as a classic example of gene-culture evolution. Humans have exploited animal milk as a food resource for at least 8500 years, but the origins, spread, and scale of dairying remain poorly understood. Indirect lines of evidence, such as lipid isotopic ratios of pottery residues, faunal mortality profiles, and lactase persistence allele frequencies, provide a partial picture of this process; however, in order to understand how, where, and when humans consumed milk products, it is necessary to link evidence of consumption directly to individuals and their dairy livestock. Here we report the first direct evidence of milk consumption, the whey protein beta-lactoglobulin (BLG), preserved in human dental calculus from the Bronze Age (ca. 3000 BCE) to the present day. Using protein tandem mass spectrometry, we demonstrate that BLG is a species-specific biomarker of dairy consumption, and we identify individuals consuming cattle, sheep, and goat milk products in the archaeological record. We then apply this method to human dental calculus from Greenland's medieval Norse colonies, and report a decline of this biomarker leading up to the abandonment of the Norse Greenland colonies in the 15(th) century CE.}, pmcid = {PMC4245811}, keywords = {Animals,Archaeology,Biological Evolution,Cattle,Dairy Products,Dental Calculus/*metabolism,Humans,Lactoglobulins/metabolism,Milk/*metabolism,Sheep,Tandem Mass Spectrometry}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/FBL7IB5L/Warinner et al. - 2014 - Direct evidence of milk consumption from ancient h.pdf} } @article{warinnerNewEra2015, title = {A New Era in Palaeomicrobiology: Prospects for Ancient Dental Calculus as a Long-Term Record of the Human Oral Microbiome}, shorttitle = {A New Era in Palaeomicrobiology}, author = {, Collins, .}, date = {2015-01-19}, journaltitle = {Philosophical Transactions of the Royal Society B: Biological Sciences}, volume = {370}, number = {1660}, pages = {20130376}, publisher = {{Royal Society}}, doi = {10.1098/rstb.2013.0376}, url = {https://royalsocietypublishing.org/doi/10.1098/rstb.2013.0376}, urldate = {2022-04-26}, abstract = {The field of palaeomicrobiology is dramatically expanding thanks to recent advances in high-throughput biomolecular sequencing, which allows unprecedented access to the evolutionary history and ecology of human-associated and environmental microbes. Recently, human dental calculus has been shown to be an abundant, nearly ubiquitous, and long-term reservoir of the ancient oral microbiome, preserving not only microbial and host biomolecules but also dietary and environmental debris. Modern investigations of native human microbiota have demonstrated that the human microbiome plays a central role in health and chronic disease, raising questions about changes in microbial ecology, diversity and function through time. This paper explores the current state of ancient oral microbiome research and discusses successful applications, methodological challenges and future possibilities in elucidating the intimate evolutionary relationship between humans and their microbes.}, keywords = {ancient DNA,dental calculus,metagenomics,metaproteomics,oral microbiome,palaeomicrobiology}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/4BWQFDLC/Warinner et al. - 2015 - A new era in palaeomicrobiology prospects for anc.pdf} } @article{warinnerPathogensHost2014, title = {Pathogens and Host Immunity in the Ancient Human Oral Cavity}, author = { . and .}, options = {useprefix=true}, date = {2014-04}, journaltitle = {Nature Genetics}, shortjournal = {Nat Genet}, volume = {46}, number = {4}, pages = {336--44}, issn = {1546-1718 (Electronic) 1061-4036 (Linking)}, doi = {10.1038/ng.2906}, abstract = {Calcified dental plaque (dental calculus) preserves for millennia and entraps biomolecules from all domains of life and viruses. We report the first, to our knowledge, high-resolution taxonomic and protein functional characterization of the ancient oral microbiome and demonstrate that the oral cavity has long served as a reservoir for bacteria implicated in both local and systemic disease. We characterize (i) the ancient oral microbiome in a diseased state, (ii) 40 opportunistic pathogens, (iii) ancient human-associated putative antibiotic resistance genes, (iv) a genome reconstruction of the periodontal pathogen Tannerella forsythia, (v) 239 bacterial and 43 human proteins, allowing confirmation of a long-term association between host immune factors, 'red complex' pathogens and periodontal disease, and (vi) DNA sequences matching dietary sources. Directly datable and nearly ubiquitous, dental calculus permits the simultaneous investigation of pathogen activity, host immunity and diet, thereby extending direct investigation of common diseases into the human evolutionary past.}, pmcid = {PMC3969750}, keywords = {Archaeology,Bacteroidetes/*genetics,Base Sequence,Dental Calculus/history/*microbiology,Food Analysis,Genome; Bacterial/*genetics,Germany,History; Medieval,Humans,Microbiota/*genetics,Molecular Sequence Data,Mouth/immunology/*microbiology,Phylogeny,Proteome/*genetics,RNA; Ribosomal; 16S/genetics,Sequence Analysis; DNA,Tandem Mass Spectrometry}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/5T7XVVSK/ng.2906.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/AJA7UHZQ/Warinner et al. - 2014 - Pathogens and host immunity in the ancient human o.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/BKERMNFL/NIHMS566378-supplement-1.pdf} } @book{whiteBoneManual2005, title = {The {{Human Bone Manual}}}, author = {White, . and Folkens, .}, date = {2005-10-13}, edition = {1st edition}, publisher = {{Academic Press}}, location = {{Amsterdam ; Boston}}, isbn = {978-0-12-088467-4}, langid = {english}, pagetotal = {488} } @article{whiteDentalCalculus1997, title = {Dental Calculus: Recent Insights into Occurrence, Formation, Prevention, Removal and Oral Health Effects of Supragingival and Subgingival Deposits}, shorttitle = {Dental Calculus}, author = {White, .}, date = {1997-10}, journaltitle = {European Journal of Oral Sciences}, shortjournal = {Eur J Oral Sci}, volume = {105}, number = {5}, pages = {508--522}, issn = {0909-8836, 1600-0722}, doi = {10.1111/j.1600-0722.1997.tb00238.x}, url = {https://onlinelibrary.wiley.com/doi/10.1111/j.1600-0722.1997.tb00238.x}, urldate = {2021-07-09}, langid = {english}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/Z8S65LMX/White - 1997 - Dental calculus recent insights into occurrence, .pdf} } @book{whiteHumanOsteology2011, title = {Human {{Osteology}}}, author = {White, . and Black, . and Folkens, .}, date = {2011-03-16}, edition = {3rd edition}, publisher = {{Academic Press}}, location = {{San Diego, Calif}}, isbn = {978-0-12-374134-9}, langid = {english}, pagetotal = {688} } @online{whoOralHealth, title = {Oral Health}, url = {https://www.who.int/news-room/fact-sheets/detail/oral-health}, urldate = {2022-03-14}, langid = {english}, organization = {{World Health Organization}}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/ZTE6S332/oral-health.html} } @article{wongCalciumPhosphate2002, title = {Calcium Phosphate Deposition in Human Dental Plaque Microcosm Biofilms Induced by a Ureolytic {{pH-rise}} Procedure}, author = { and }, date = {2002-11-01}, journaltitle = {Archives of Oral Biology}, shortjournal = {Archives of Oral Biology}, volume = {47}, number = {11}, pages = {779--790}, issn = {0003-9969}, doi = {10.1016/S0003-9969(02)00114-0}, url = {https://www.sciencedirect.com/science/article/pii/S0003996902001140}, urldate = {2021-03-17}, abstract = {The objectives were to develop and characterize a procedure based on a ureolytic pH rise to deposit calcium phosphate into microcosm dental plaque biofilms and to test the importance of the plaque pH range. Plaque biofilms were cultured in a multiplaque culture system (‘artificial mouth’) with a continuous supply of a simulated oral fluid (basal medium mucin; BMM) with 146mmol/l (5\% w/v) sucrose periodically applied over 6min every 8h. After initial plaque growth, the biofilms were periodically exposed for up to 16 days to 6-min applications of calcium phosphate monofluorophosphate urea (CPMU) solution containing 20mmol/l CaCl2, 12mmol/l NaH2PO4, 5mmol/l monofluorophosphate and 500mmol/l urea (pH 5.0). Three application regimes were examined, one included a sucrose-induced acidic pH fluctuation. Plaque hydrolysis of the urea in CPMU caused the pH to rise to between 8.2 and 8.8, depositing fluoridated and carbonated calcium phosphates, and possibly some calcium carbonate, into the plaque. Calcium, phosphate and fluoride deposition was rapid for about 4 days and then slowed. After 10 days’ treatment under standard conditions (BMM containing 1mmol/l urea and 1mmol/l arginine), plaque calcium and phosphate concentrations had increased up to 50-fold and 10-fold to approximately 2–4 and 1–2mmol/g plaque protein, respectively. The calcium, phosphate and fluoride content increased steadily. Calcium phosphate deposition was proportional to the plaque resting pH, increasing over four-fold when the BMM urea concentration was increased from 0 to 20mmol/l, which raised the resting pH from 6.4 to 7.2 and yielded a mean plaque calcium concentration of 14.3mmol/g protein, one subsample reaching 20.8mmol/g protein. Supplementation of BMM with 20\% human serum inhibited deposition. These results support the hypothesis that an alkaline pH in plaque is critical in promoting plaque mineralization and that mineral deposition is modulated by serum. These factors are likely to be important in regulating calculus formation.}, langid = {english}, keywords = {Biofilm,Dental calculus,Dental plaque,Mineralization,Plaque pH,Urea}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/YYLE2C29/Wong et al. - 2002 - Calcium phosphate deposition in human dental plaqu.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/4IK6DGEP/S0003996902001140.html} } @article{wrightAdvancingRefining2021, title = {Advancing and Refining Archaeological Dental Calculus Research Using Multiomic Frameworks}, author = {. and and .}, date = {2021-01-01}, journaltitle = {STAR: Science \& Technology of Archaeological Research}, volume = {7}, number = {1}, pages = {13--30}, publisher = {{Routledge}}, issn = {null}, doi = {10.1080/20548923.2021.1882122}, url = {https://doi.org/10.1080/20548923.2021.1882122}, urldate = {2022-04-08}, abstract = {Dental calculus (calcified dental plaque) is a cross-cultural biological matrix that is emerging as a critical source of information for anthropologists and oral health professionals. It contains a multitude of diverse biomolecules, providing information about an individual’s culture, diet, ancestry, and health. Most researchers who study archaeological dental calculus use genomic or proteomic approaches, although a wide range of other techniques are now available. However, few studies have utilized efficient multiomic protocols. This lack of integration is problematic, as such approaches in other fields have proven to improve results and strengthen interpretations. Our review discusses three multiomic approaches: (1) interactions between the metaproteome and metagenome; (2) relationships between the host genome and oral metagenome; and, (3) associations between the epigenome and metagenome. We draw from multiomic studies on soil, plant, gut, and modern oral microbiomes to demonstrate how such integration can provide insights that are not attainable with single-omic approaches.}, keywords = {ancient DNA,Dental calculus,multiomics,oral microbiome}, annotation = {\_eprint: https://doi.org/10.1080/20548923.2021.1882122}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/74LHI835/Wright et al. - 2021 - Advancing and refining archaeological dental calcu.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/Y6ZB5GSC/20548923.2021.html} } @article{yatesOralMicrobiome2021, title = {The Evolution and Changing Ecology of the {{African}} Hominid Oral Microbiome}, author = {, . and Velsko, . and Aron, Franziska and Posth, Cosimo and Hofman, . and Austin, . and Parker, . and Mann, . and Nägele, Kathrin and Arthur, and Arthur, . and Bauer, . and Crevecoeur, Isabelle and Cupillard, Christophe and Curtis, . and , and . , . and Escrivá, and and Gibbon, . and Morales, and and and Henry, . and and and and Peresani, Marco and Moroder, and and , , , , , , , , . and Sankaranarayanan, , , , , , , , Christina}, date = {2021-05-18}, journaltitle = {Proceedings of the National Academy of Sciences}, shortjournal = {PNAS}, volume = {118}, number = {20}, eprint = {33972424}, eprinttype = {pmid}, publisher = {{National Academy of Sciences}}, issn = {0027-8424, 1091-6490}, doi = {10.1073/pnas.2021655118}, url = {https://www.pnas.org/content/118/20/e2021655118}, urldate = {2021-05-12}, abstract = {The oral microbiome plays key roles in human biology, health, and disease, but little is known about the global diversity, variation, or evolution of this microbial community. To better understand the evolution and changing ecology of the human oral microbiome, we analyzed 124 dental biofilm metagenomes from humans, including Neanderthals and Late Pleistocene to present-day modern humans, chimpanzees, and gorillas, as well as New World howler monkeys for comparison. We find that a core microbiome of primarily biofilm structural taxa has been maintained throughout African hominid evolution, and these microbial groups are also shared with howler monkeys, suggesting that they have been important oral members since before the catarrhine–platyrrhine split ca. 40 Mya. However, community structure and individual microbial phylogenies do not closely reflect host relationships, and the dental biofilms of Homo and chimpanzees are distinguished by major taxonomic and functional differences. Reconstructing oral metagenomes from up to 100 thousand years ago, we show that the microbial profiles of both Neanderthals and modern humans are highly similar, sharing functional adaptations in nutrient metabolism. These include an apparent Homo-specific acquisition of salivary amylase-binding capability by oral streptococci, suggesting microbial coadaptation with host diet. We additionally find evidence of shared genetic diversity in the oral bacteria of Neanderthal and Upper Paleolithic modern humans that is not observed in later modern human populations. Differences in the oral microbiomes of African hominids provide insights into human evolution, the ancestral state of the human microbiome, and a temporal framework for understanding microbial health and disease.}, langid = {english}, keywords = {dental calculus,microbiome,Neanderthal,primate,salivary amylase}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/EBLFRY6K/Yates et al. - 2021 - The evolution and changing ecology of the African .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/D7LZHTMU/e2021655118.html} } @article{yaussyCalculusSurvivorship2019, title = {Calculus and Survivorship in Medieval {{London}}: {{The}} Association between Dental Disease and a Demographic Measure of General Health}, shorttitle = {Calculus and Survivorship in Medieval {{London}}}, author = {. and DeWitte, .}, date = {2019}, journaltitle = {American Journal of Physical Anthropology}, volume = {168}, number = {3}, pages = {552--565}, issn = {1096-8644}, doi = {10.1002/ajpa.23772}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/ajpa.23772}, urldate = {2020-08-03}, abstract = {Objectives Dental plaque is associated with a variety of systemic diseases and mortality risks in living populations. However, bioarchaeologists have not fully investigated the mortality risks associated with plaque (or its mineralized form, calculus) in the past. This study examines the relationship between survivorship and calculus in a medieval skeletal sample. Materials and methods Our sample (n = 1,098) from four medieval London cemeteries, c. 1000–1540 CE, includes people who died under attritional (normal) and catastrophic (famine and plague) conditions. The associations between age and the presence of dental calculus on the permanent left first mandibular molar are assessed using binary logistic regression and Kaplan–Meier survival analysis. Results The regression results indicate a significant negative relationship between age and calculus presence for individuals of all ages who died under normal mortality conditions and for adults who died under both normal and catastrophic conditions. Survival analysis reveals decreased survivorship for people of all ages with calculus under normal mortality conditions. Similarly, during conditions of catastrophic mortality, adult males with calculus suffered reduced survivorship compared to males without it, though there was no difference in survivorship between adult females with and without calculus. Conclusions These results suggest that, as in modern populations, calculus accumulation in the inhabitants of medieval London reflects a greater risk of premature death. The evaluation of calculus, a potential measure of underlying frailty, in the context of a demographic measure of general health suggests that it might provide insights into health in past populations.}, langid = {english}, keywords = {binary logistic regression,bioarchaeology,mortality,paleodemography,paleopathology,survival analysis}, annotation = {\_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/ajpa.23772}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/AT2XP34S/Yaussy and DeWitte - 2019 - Calculus and survivorship in medieval London The .pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/MD6UFKXQ/ajpa.23772.pdf;/mnt/hogwarts/Uni/Literature/Zotero/storage/MY4JLA7B/ajpa.html} } @article{zhangDentalDisease1982, title = {Dental Disease of Neolithic Age Skulls Excavated in Shaanxi Province}, author = {}, date = {1982-01}, journaltitle = {Chinese Medical Journal}, volume = {95}, number = {06}, pages = {391--396}, publisher = {{Chinese Medical Association Publishing House}}, doi = {10.5555/cmj.0366-6999.95.06.p391.01}, url = {https://medcentral.net/doi/abs/10.5555/cmj.0366-6999.95.06.p391.01}, urldate = {2022-04-04}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/Z8SKGUXY/Zhang - 1982 - Dental disease of neolithic age skulls excavated i.pdf} } @article{zijngeBiofilmArchitecture2010, title = {Oral {{Biofilm Architecture}} on {{Natural Teeth}}}, author = { , Degener, , .}, editor = {}, options = {useprefix=true}, date = {2010-02-24}, journaltitle = {PLoS ONE}, shortjournal = {PLoS ONE}, volume = {5}, number = {2}, pages = {e9321}, issn = {1932-6203}, doi = {10.1371/journal.pone.0009321}, url = {https://dx.plos.org/10.1371/journal.pone.0009321}, urldate = {2020-10-01}, abstract = {Periodontitis and caries are infectious diseases of the oral cavity in which oral biofilms play a causative role. Moreover, oral biofilms are widely studied as model systems for bacterial adhesion, biofilm development, and biofilm resistance to antibiotics, due to their widespread presence and accessibility. Despite descriptions of initial plaque formation on the tooth surface, studies on mature plaque and plaque structure below the gum are limited to landmark studies from the 1970s, without appreciating the breadth of microbial diversity in the plaque. We used fluorescent in situ hybridization to localize in vivo the most abundant species from different phyla and species associated with periodontitis on seven embedded teeth obtained from four different subjects. The data showed convincingly the dominance of Actinomyces sp., Tannerella forsythia, Fusobacterium nucleatum, Spirochaetes, and Synergistetes in subgingival plaque. The latter proved to be new with a possibly important role in host-pathogen interaction due to its localization in close proximity to immune cells. The present study identified for the first time in vivo that Lactobacillus sp. are the central cells of bacterial aggregates in subgingival plaque, and that Streptococcus sp. and the yeast Candida albicans form corncob structures in supragingival plaque. Finally, periodontal pathogens colonize already formed biofilms and form microcolonies therein. These in vivo observations on oral biofilms provide a clear vision on biofilm architecture and the spatial distribution of predominant species.}, langid = {english}, keywords = {oral biofilm,oral microbiome}, file = {/mnt/hogwarts/Uni/Literature/Zotero/storage/4KRLH48X/Zijnge et al. - 2010 - Oral Biofilm Architecture on Natural Teeth.pdf} } {\sl \textbf{Rejuvenan Global Health, Inc}} \hfill May 2015 - December 2016 \\ Software Engineer \hfill New York, NY \begin{itemize} \item Responsible for building the Rejuvenan iOS application, writing automation scripts, and contributing code to our company website and backend services. Setup a continuous integration pipeline using Travis-CI and Fastlane for our iOS projects, jobs included kicking off unit tests and new build artifacts from master. \item Added new features to a growing Ruby-on-Rails backend, which included unit and integration tests. Maintained older backend services written in Java (Spring Framework), including adding a feature to push logs to an AWS S3 bucket and writing unit tests. \end{itemize} Capstone-Skynet/Capstone-Skynet.github.io \subsubsection{Subsystem Overview} The video capture subsystem of the computing platform consists of a camera and the video transmission protocol to the processing core. For this subsystem, the hardware utilized includes a CMOS camera, a Raspberry Pi device, and a MIPI Camera Serial interface. \subsubsection{Camera Choice} The camera used to capture video is the \textit{Raspberry Pi Camera Module}. This module supports 1080p, 720p and 480p video, captured using a fixed focus lens. The camera captures live video and sends it to the PMB, which feeds the data to both the machine learning model and the base station. A major consideration regarding camera selection was the latency and throughput of the machine learning implementation on the FPGA. As the machine learning implementation is heavily constrained by the FPGA's limited resources, it is not necessary to use a camera with a high resolution and framerate. % {MIPI interface} \subsubsection{MIPI Interface and Libraries} The MIPI camera serial interface (CSI) is a high-speed protocol primarily intended for point-to-point image and video transmission between cameras and host devices. The MIPI protocol has been widely adopted for image and video transmission, and as such the camera can easily be replaced by the client in the future, if desired (fulfilling constraint \textbf{C.EX.5}). % {Camera libraries or getting the video to a 'viewable' state} The \texttt{raspicam} C++ library manages video acquisition from the camera module. The helper program uses \texttt{raspicam} instead of the native Raspberry Pi camera libraries in order to perform capture and processing in a low-level C++ environment. This allows for the low-level memory manipulation/image format conversions required throughout the flow to be performed directly on the (more powerful) PMB, rather than the PLB. % {Explored Alternatives} \subsubsection{Explored Hardware Alternatives} There were many alternatives explored for each component of the video capture subsystem. One of the possibilities explored for the camera subsystem was the purchase of a drone package with a pre-integrated camera. The implicit requirement to disassemble and reconfigure/tap-into a hard-wired system, however, would create an unmaintainable product --- making user replacement of the camera for future research improvements difficult (violating constraint \textbf{C.EX.5}). Another consideration for the subsystem involved utilizing a USB webcam --- connecting it via the Raspberry Pi's USB port. The issue with this option is the inferior configurability of the webcam compared to the Raspberry Pi camera. The reconfigurability provided by the \texttt{raspicam} library, as previously noted, made the Rasberry Pi Camera a better solution suited to the client's needs. A possible alternative explored for the camera interface was RS-232. The RS-232 interface is a commonly used, easy to integrate serial interface. It is widely used in the communications area, however its low data transfer rate limits both the resolution and frame rate of the video footage that can be captured. The highest (reliable) baud rate of the RS-232 interface is approximately 115200 bits per second --- translating to a frame rate of 0.05 FPS for 480p video footage (violating constraint \textbf{NF.CM.3}). For the live video feed transmission, other options explored included wireless HDMI, VGA and analog transmission methods. Ultimately these options required additional hardware (contributing to higher costs and system weight) and would introduce more points of failure and complexity into the finalized system, thus they were not pursued. % The rules below clean up and shorten bibliography entries according to my % personal perferences. % 1. don't show various fields such as ISBNs, etc. % \AtEveryBibitem{\clearfield{doi}} \AtEveryBibitem{\clearfield{eprint}} \AtEveryBibitem{\clearfield{isbn}} \AtEveryBibitem{\clearfield{issn}} \AtEveryBibitem{\clearfield{pagetotal}} \AtEveryBibitem{\clearfield{series}} \AtEveryBibitem{\ifboolexpr{test {\ifentrytype{electronic}} or test {\ifentrytype{online}} or test {\ifentrytype{misc}}}{}{\clearfield{url}}} \AtEveryBibitem{\clearlist{language}} \AtEveryBibitem{\clearlist{publisher}} \AtEveryBibitem{\clearname{editor}} \AtEveryBibitem{\ifentrytype{article}{\clearfield{number}}{}} \AtEveryBibitem{\ifentrytype{inproceedings}{\clearfield{volume}}{}} \DeclareFieldFormat{url}{\url{#1}} \usepackage{xpatch} \xpatchbibdriver{online} {\printtext[parens]{\usebibmacro{date}}} {\iffieldundef{year} {} {\printtext[parens]{\usebibmacro{date}}}} {} {\typeout{There was an error patching biblatex-ieee (specifically, ieee.bbx's @online driver)}} % \AtEveryBibitem{\clearfield{month}} % \AtEveryBibitem{\clearfield{note}} % \AtEveryBibitem{\clearfield{pages}} % \AtEveryBibitem{\clearlist{institution}} % \AtEveryBibitem{\clearlist{location}} \AtEveryCitekey{\UseBibitemHook} % 2. shorten various conference and journal titles to some sane length % \DeclareSourcemap{ \maps[datatype=bibtex]{ \map{ % shorten conferences \step[fieldsource=booktitle, match={.*Internet Measurement Conference.*}, replace={ACM IMC}] \step[fieldsource=booktitle, match={.*Networked Systems Design and Implementation.*}, replace={USENIX NSDI}] \step[fieldsource=booktitle, match={.*Hot Topics in Networks.*}, replace={ACM HotNets}] \step[fieldsource=booktitle, match={.*Ope..ting Systems Design .* Implementation.*}, replace={USENIX OSDI}] \step[fieldsource=booktitle, match={.*USENIX.*Annual Technical Conference.*}, replace={USENIX ATC}] \step[fieldsource=booktitle, match={.*Conference on File and Storage Technologies.*}, replace={USENIX FAST}] \step[fieldsource=booktitle, match={.*European Conference on Computer Systems.*}, replace={ACM EuroSys}] \step[fieldsource=booktitle, match={.*Symposium on Operating Systems Principles.*}, replace={ACM SOSP}] \step[fieldsource=booktitle, match={.*(USENIX Security Symposium|USENIX Conference on Security).*}, replace={USENIX Security}] \step[fieldsource=booktitle, match={.*USENIX Symposium on Internet Technologies and Systems.*}, replace={USENIX USITS}] \step[fieldsource=booktitle, match={.*Emerging Networking E.periments .nd Technologies.*}, replace={ACM CoNEXT}] \step[fieldsource=booktitle, match={.*INFOCOM.*}, replace={IEEE Infocom}] \step[fieldsource=booktitle, match={.*Mobile Computing and Networking.*}, replace={ACM MobiCom}] \step[fieldsource=booktitle, match={.*Roedunet.*}, replace={IEEE RoEduNet}] \step[fieldsource=booktitle, match={.*Symposium on Software Defined Networking Research.*}, replace={ACM SOSR}] \step[fieldsource=booktitle, match={.*Architectures? for Networking and Communications Systems.*}, replace={ACM ANCS}] \step[fieldsource=booktitle, match={.*QoS.*Mini.*Conf.*ICC.*}, replace={IEEE ICC QoS Mini-Conference}] \step[fieldsource=institution, match={.*Barcelona Supercomputing Center.*}, replace={Barcelona Supercomputing Center (BSC)}] \step[fieldsource=booktitle, match={.*System Administration and Network Engineering Conference.*}, replace={SANE}] \step[fieldsource=booktitle, match={.*Architectural Support for Programming Languages and Operating Systems.*}, replace={ACM ASPLOS}] \step[fieldsource=booktitle, match={.*Passive and Active Measurement.*}, replace={PAM}] \step[fieldsource=booktitle, match={.*E2EMON.*}, replace={IEEE/IFIP E2EMON}] \step[fieldsource=booktitle, match={.*International Conference on Autonomic Computing.*}, replace={IEEE ICAC}] \step[fieldsource=booktitle, match={.*Hot Topics in Middleboxes.*}, replace={ACM SIGCOMM HotMiddlebox}] \step[fieldsource=booktitle, match={.*Hot Topics in Management of Internet, Cloud, and Enterprise Networks.*}, replace={USENIX Hot-ICE}] \step[fieldsource=booktitle, match={.*International (ACM )?Conference on Supercomputing.*}, replace={ACM ICS}] \step[fieldsource=booktitle, match={.*Symposium on Mass Storage Systems and Technolog.*}, replace={MSST}] \step[fieldsource=booktitle, match={.*International Symposium on Microarchitecture.*}, replace={IEEE/ACM MICRO}] \step[fieldsource=booktitle, match={.*Languages and Compilers for Parallel Computing.*}, replace={LCPC}] \step[fieldsource=booktitle, match={.*Conference for High Performance Computing, Networking, Storage and Analysis.*}, replace={ACM/IEEE SC}] \step[fieldsource=booktitle, match={.*International Symposium on Computing and Networking.*}, replace={CANDAR}] \step[fieldsource=booktitle, match={.*Euro-Par.*}, replace={Euro-Par}] \step[fieldsource=booktitle, match={.*Symposium on High-Performance Interconnects.*}, replace={IEEE Hot Interconnects}] \step[fieldsource=booktitle, match={.*Conference on Microelectronic Systems Education.*}, replace={IEEE MSE}] \step[fieldsource=booktitle, match={.*International Conference on Big Data.*}, replace={IEEE Big Data}] \step[fieldsource=booktitle, match={.*Symposium on Modeling, Analysis, and Simulation.*}, replace={IEEE MASCOTS}] \step[fieldsource=booktitle, match={.*Euromicro Conference on Digital System Design.*}, replace={Euromicro DSD}] \step[fieldsource=booktitle, match={.*USENIX Winter Conference.*}, replace={USENIX Winter}] \step[fieldsource=booktitle, match={.*SIGMETRICS.*}, replace={ACM SIGMETRICS}] \step[fieldsource=booktitle, match={.*Systems and Storage Conference.*}, replace={ACM SYSTOR}] \step[fieldsource=booktitle, match={.*Joint Meeting of the European Software Engineering Conference and the ACM SIGSOFT Symposium on the Foundations of Software Engineering.*}, replace={ESEC/FSE}] \step[fieldsource=booktitle, match={.*Hot Topics in Cloud Computing.*}, replace={USENIX HotCloud}] \step[fieldsource=booktitle, match={.*Mass Storage Systems and Technologies.*}, replace={MSST}] \step[fieldsource=booktitle, match={.*International Conference on Management of Data.*}, replace={ACM SIGMOD}] \step[fieldsource=booktitle, match={.*Workshop on Performance Analysis of Workload Optimized Systems.*}, replace={FastPath Workshop}] \step[fieldsource=booktitle, match={.*Multimedia Systems Conference.*}, replace={ACM MMSys}] \step[fieldsource=booktitle, match={.*IFIP Networking Conference.*}, replace={IFIP Networking}] \step[fieldsource=booktitle, match={.*Conference on Local Computer Networks.*}, replace={IEEE LCN}] \step[fieldsource=booktitle, match={.*International Packet Video Workshop.*}, replace={Packet Video Workshop}] \step[fieldsource=booktitle, match={.*Evolution, Performance, and Interoperability of QUIC.*}, replace={ACM EPIQ}] \step[fieldsource=booktitle, match={.*International Conference on World Wide Web.*}, replace={WWW}] \step[fieldsource=booktitle, match={.*Security Standardisation Research.*}, replace={SSR}] \step[fieldsource=booktitle, match={.*Symposium on Security and Privacy.*}, replace={IEEE SSP}] \step[fieldsource=booktitle, match={.*Network and Operating Systems Support for Digital Audio and Video.*}, replace={ACM NOSSDAV}] % keep this last \step[fieldsource=booktitle, match={.*(Conference on SIGCOMM|Special Interest Group on Data Communication|Applications, Technologies, Architectures, and Protocols|Communications Architectures and Protocols).*}, replace={ACM SIGCOMM}] % shorten journals \step[fieldsource=journal, match={.*IEEE Journal on Selected Areas in Communication.*}, replace={IEEE JSAC}] \step[fieldsource=journal, match={.*Trans. Netw..*}, replace={IEEE/ACM ToN}] \step[fieldsource=journal, match={.*Transactions on Network and Service Management*}, replace={IEEE ToNSM}] \step[fieldsource=journal, match={.*Comput. Commun. Rev..*}, replace={ACM SIGCOMM CCR}] } } } % 3. Use "Proc." for @inproceedings (not "In:") and abbreviate months consistently % \DeclareLanguageMapping{american}{bib/lars-american} \DeclareLanguageMapping{british}{bib/lars-british} \DeclareLanguageMapping{english}{bib/lars-english} langsci/277 \newcommand{\First}{\gls{1}} \newcommand{\Fsg}{\gls{1}\gls{sg}} \newcommand{\Fpl}{\gls{1}\gls{pl}} \newcommand{\Second}{\gls{2}} \newcommand{\Ssg}{\gls{2}\gls{sg}} \newcommand{\Spl}{\gls{2}\gls{pl}} \newcommand{\Third}{\gls{3}} \newcommand{\Tsg}{\gls{3}\gls{sg}} \newcommand{\Tpl}{\gls{3}\gls{pl}} \newcommand{\Fosg}{\gls{4}\gls{sg}} \newcommand{\Fopl}{\gls{4}\gls{pl}} \newcommand{\Nsg}{\gls{nsg}} \newcommand{\Abl}{\gls{abl}} \newcommand{\Abs}{\gls{abs}} \newcommand{\Acc}{\gls{acc}} \newcommand{\Act}{\gls{act}} \newcommand{\Ade}{\gls{ade}} \newcommand{\Adj}{\gls{adj}} \newcommand{\Adv}{\gls{adv}} \newcommand{\Agr}{\gls{agr}} \newcommand{\All}{\gls{all}} \newcommand{\Anim}{\gls{anim}} \newcommand{\Anom}{\gls{adnom}} \newcommand{\Ant}{\gls{ant}} \newcommand{\Antip}{\gls{antip}} \newcommand{\An}{\gls{an}} \newcommand{\Aor}{\gls{aor}} \newcommand{\Appl}{\gls{appl}} \newcommand{\Apud}{\gls{apud}} \newcommand{\Art}{\gls{art}} \newcommand{\Asp}{\gls{asp}} \newcommand{\Assoc}{\gls{assoc}} \newcommand{\At}{\gls{at}} \newcommand{\Aud}{\gls{aud}} \newcommand{\Aug}{\gls{aug}} \newcommand{\Aux}{\gls{aux}} \newcommand{\Av}{\gls{av}} \newcommand{\Bb}{\gls{B}} \newcommand{\Ben}{\gls{ben}} \newcommand{\BrP}{\gls{BrP}} \newcommand{\Caus}{\gls{caus}} \newcommand{\Cci}{\gls{cci}} \newcommand{\Cf}{\gls{cf}} \newcommand{\Cj}{\gls{cj}} \newcommand{\Clf}{\gls{clf}} \newcommand{\Cl}{\gls{cl}} \newcommand{\Cmpl}{\gls{cmpl}} \newcommand{\Cmpr}{\gls{cmpr}} \newcommand{\Cnj}{\gls{cnj}} \newcommand{\Cntr}{\gls{cntr}} \newcommand{\Comp}{\gls{comp}} \newcommand{\Compl}{\gls{compl}} \newcommand{\Com}{\gls{com}} \newcommand{\Cond}{\gls{cond}} \newcommand{\Conn}{\gls{conn}} \newcommand{\Cont}{\gls{cont}} \newcommand{\Cop}{\gls{cop}} \newcommand{\Cr}{\gls{cr}} \newcommand{\Cvb}{\gls{cvb}} \newcommand{\Dat}{\gls{dat}} \newcommand{\Decl}{\gls{decl}} \newcommand{\Def}{\gls{def}} \newcommand{\Dem}{\gls{dem}} \newcommand{\Dep}{\gls{dep}} \newcommand{\Depend}{\gls{depend}} \newcommand{\Desc}{\gls{desc}} \newcommand{\Dei}{\gls{dei}} \newcommand{\Det}{\gls{det}} \newcommand{\Detr}{\gls{detr}} \newcommand{\Dim}{\gls{dim}} \newcommand{\Dir}{\gls{dir}} \newcommand{\Disp}{\gls{disp}} \newcommand{\Distr}{\gls{distr}} \newcommand{\Dist}{\gls{dist}} \newcommand{\Dj}{\gls{dj}} \newcommand{\Dom}{\gls{dom}} \newcommand{\Do}{\gls{do}} \newcommand{\Dptcl}{\gls{dptcl}} \newcommand{\Dr}{\gls{dr}} \newcommand{\Ds}{\gls{ds}} \newcommand{\Dur}{\gls{dur}} \newcommand{\Du}{\gls{du}} \newcommand{\Ela}{\gls{ela}} \newcommand{\Emph}{\gls{emph}} \newcommand{\Ep}{\gls{ep}} \newcommand{\Erg}{\gls{erg}} \newcommand{\Es}{\gls{es}} \newcommand{\EuP}{\gls{EuP}} \newcommand{\Evid}{\gls{evid}} \newcommand{\Excl}{\gls{excl}} \newcommand{\Expl}{\gls{expl}} \newcommand{\glossF}{\gls{f}} \newcommand{\Foc}{\gls{foc}} \newcommand{\Fs}{\gls{fs}} \newcommand{\Fut}{\gls{fut}} \newcommand{\Fv}{\gls{fv}} \newcommand{\Gen}{\gls{gen}} \newcommand{\Genc}{\gls{genc}} \newcommand{\Ger}{\gls{ger}} \newcommand{\Gg}{\gls{g}-grade (Muskogean)} \newcommand{\Hab}{\gls{hab}} \newcommand{\Ill}{\gls{ill}} \newcommand{\Impass}{\gls{impass}} \newcommand{\Imp}{\gls{imp}} \newcommand{\Inch}{\gls{inch}} \newcommand{\Incl}{\gls{incl}} \newcommand{\Incmpl}{\gls{incmpl}} \newcommand{\Indef}{\gls{indef}} \newcommand{\Indic}{\gls{indic}} \newcommand{\Indir}{\gls{indir}} \newcommand{\Ind}{\gls{ind}} \newcommand{\Ine}{\gls{ine}} \newcommand{\Inf}{\gls{inf}} \newcommand{\Infr}{\gls{infr}} \newcommand{\Ins}{\gls{ins}} \newcommand{\Intr}{\gls{intr}} \newcommand{\Int}{\gls{int}} \newcommand{\Inv}{\gls{inv}} \newcommand{\Invol}{\gls{invol}} \newcommand{\Io}{\gls{io}} \newcommand{\Ipfv}{\gls{ipfv}} \newcommand{\Ipf}{\gls{ipf}} \newcommand{\Irr}{\gls{irr}} \newcommand{\Iter}{\gls{iter}} \newcommand{\Jj}{\gls{J}} \newcommand{\Lat}{\gls{lat}} \newcommand{\Ll}{\gls{l}} \newcommand{\Lnk}{\gls{lnk}} \newcommand{\Loc}{\gls{loc}} \newcommand{\Lv}{\gls{lv}} \newcommand{\M}{\gls{m}} \newcommand{\Mid}{\gls{mid}} \newcommand{\Mnom}{\gls{mnom}} \newcommand{\Mr}{\gls{mr}} \newcommand{\Mut}{\gls{mut}} \newcommand{\glossN}{\gls{n}} \newcommand{\Nact}{\gls{nact}} \newcommand{\Ncm}{\gls{ncm}} \newcommand{\Neg}{\gls{neg}} \newcommand{\Nfin}{\gls{nfin}} \newcommand{\Ng}{\gls{n}-grade (Muskogean)} \newcommand{\Nmlz}{\gls{nmlz}} \newcommand{\Nom}{\gls{nom}} \newcommand{\Npst}{\gls{npst}} \newcommand{\Nth}{\gls{nth}} \newcommand{\Num}{\gls{num}} \newcommand{\Nw}{\gls{nw}} \newcommand{\Obj}{\gls{obj}} \newcommand{\Obl}{\gls{obl}} \newcommand{\Obv}{\gls{obv}} \newcommand{\Om}{\gls{om}} \newcommand{\Ov}{\gls{ov}} \newcommand{\Ppp}{\gls{ppp}} \newcommand{\Part}{\gls{part}} \newcommand{\Pass}{\gls{pass}} \newcommand{\Pauc}{\gls{pauc}} \newcommand{\Perm}{\gls{perm}} \newcommand{\Pfv}{\gls{pfv}} \newcommand{\Pg}{\gls{pg}} \newcommand{\Pip}{\gls{pip}} \newcommand{\Pl}{\gls{pl}} \newcommand{\Pn}{\gls{pn}} \newcommand{\Poss}{\gls{poss}} \newcommand{\Prdst}{\gls{prdst}} \newcommand{\Pred}{\gls{pred}} \newcommand{\Prf}{\gls{prf}} \newcommand{\Prep}{\gls{prep}} \newcommand{\Prog}{\gls{prog}} \newcommand{\Proh}{\gls{proh}} \newcommand{\Pron}{\gls{pron}} \newcommand{\Propn}{\gls{propn}} \newcommand{\Pros}{\gls{pros}} \newcommand{\Prov}{\gls{prov}} \newcommand{\Prox}{\gls{prox}} \newcommand{\Prs}{\gls{prs}} \newcommand{\Prt}{\gls{prt}} \newcommand{\Prv}{\gls{prv}} \newcommand{\Pst}{\gls{pst}} \newcommand{\Ptcl}{\gls{ptcl}} \newcommand{\Ptcp}{\gls{ptcp}} \newcommand{\Purp}{\gls{purp}} \newcommand{\Pv}{\gls{pv}} \newcommand{\glossQ}{\gls{q}} \newcommand{\Quot}{\gls{quot}} \newcommand{\Real}{\gls{real}} \newcommand{\Rci}{\gls{rci}} \newcommand{\Recp}{\gls{recp}} \newcommand{\Red}{\gls{red}} \newcommand{\Refl}{\gls{refl}} \newcommand{\Rel}{\gls{rel}} \newcommand{\Rem}{\gls{rem}} \newcommand{\Rep}{\gls{rep}} \newcommand{\Res}{\gls{res}} \newcommand{\Rg}{\gls{rg}} \newcommand{\Rm}{\gls{rm}} \newcommand{\Rpt}{\gls{rpt}} \newcommand{\Sbjv}{\gls{sbjv}} \newcommand{\Sbj}{\gls{sbj}} \newcommand{\Secun}{\gls{sec}} \newcommand{\Sg}{\gls{sg}} \newcommand{\Simp}{\gls{simp}} \newcommand{\Sim}{\gls{sim}} \newcommand{\Sm}{\gls{sm}} \newcommand{\Sr}{\gls{sr}} \newcommand{\Ss}{\gls{ss}} \newcommand{\Stat}{\gls{stat}} \newcommand{\Subj}{\gls{subj}} \newcommand{\Subl}{\gls{subl}} \newcommand{\Supe}{\gls{supe}} \newcommand{\Supl}{\gls{supl}} \newcommand{\Ta}{\gls{ta}} \newcommand{\Term}{\gls{term}} \newcommand{\Th}{\gls{th}} \newcommand{\Ti}{\gls{ti}} \newcommand{\Tns}{\gls{tns}} \newcommand{\Topic}{\gls{top}} \newcommand{\Transl}{\gls{transl}} \newcommand{\Tr}{\gls{tr}} \newcommand{\Tto}{\gls{tt}} \newcommand{\Ut}{\gls{ut}} \newcommand{\Uu}{\gls{u}} \newcommand{\Vblz}{\gls{vblz}} \newcommand{\Vm}{\gls{vm}} \newcommand{\Vn}{\gls{vn}} \newcommand{\Voc}{\gls{voc}} \newcommand{\Vol}{\gls{vol}} \newcommand{\Yn}{\gls{y.n}} \newcommand{\Aa}{\gls{a}} \newcommand{\Pp}{\gls{p}} \newcommand{\Rr}{\gls{r}} \newcommand{\Su}{\gls{s}} \newcommand{\Tt}{\gls{t}} \newcommand{\Ig}{\gls{i}} \newcommand{\Ii}{\gls{ii}} \newcommand{\Iii}{\gls{iii}} \newcommand{\Iv}{\gls{iv}} \newcommand{\SCC}{\gls{SC3}} \newcommand{\SC}{\gls{SC2}} arminnh/lab-computer-networks0 /proc/sys/net/ipv4/ip\_forward\documentclass[a4paper,10pt]{article} \usepackage[utf8]{inputenc} \usepackage[russian]{babel} \usepackage{indentfirst} \usepackage{amssymb} \usepackage{amsmath} \begin{document} \begin{enumerate} \item \textit{В чём заключается смысл метода измерения сопротивления с помощью мостика? Какое плечо моста является плечом сравнения?} Суть ~метода ~заключается в нахождении неизвестных сопротивлений через соотношения известных, которое вычисляется по правилам Киркгоффа. Плечом сравнения является плечо $R_1$. \item \textit{Запишите закон Ома для однородного участка цепи, неоднородного участка, полной цепи.} Закон Ома для однородного участка цепи: $$\mathcal{I} = \frac{U}{R}$$ Закон Ома для неоднородного участка цепи: $$\mathcal{I} = \frac{\varphi_1 - \varphi_2 + \mathcal{E}_{12}}{R}$$ Закон Ома для полной цепи: $$\mathcal{I} = \frac{\mathcal E}{R}$$ \item \textit{Запишите закон Ома в дифференциальной форме.} $$\vec{j} = \sigma\vec{E}$$ \item \textit{Сформулируйте правила Киркгоффа.} \begin{enumerate} \item Алгебраическая сумма сил токов, которые сходятся в узле, равна 0: $$\sum\limits_{k=1}^n \mathcal{I}_k = 0$$ \item В контуре алгебраическая сумма падений напряжений (произведений сил токов на сопротивления соответственных участков) равна алгебраической сумме электродвижущих сил, которые действуют в этом контуре: $$\sum\limits_{k=1}^N \mathcal{I}_kR_k = \sum\limits_{k=1}^N\mathcal{E}_k$$ \end{enumerate} \item {Какая формула связи сопротивлений моста при условии его равновесия? Выведите её.} По второму правилу Киркгоффа для acd и cbd контуров: \begin{equation}\label{a1} \begin{cases} \mathcal{I}_xR_x+\mathcal{I}_rR_r - \mathcal{I}_3R_3 = 0\\ \mathcal{I}_1R_1 - \mathcal{I}_2R_2 - \mathcal{I}_rR_r = 0 \end{cases} \end{equation} По первому правилу Киркгоффа для узлов c и d: \begin{equation}\label{a2} \begin{cases} \mathcal{I}_x-\mathcal{I}_r-\mathcal{I}_1 = 0\\ \mathcal{I}_r+\mathcal{I}_3-\mathcal{I}_2 = 0 \end{cases} \end{equation} Если $\mathcal{I}_r = 0$, то из (\ref{a1}) и (\ref{a2}): \begin{equation}\label{a3} \begin{cases} \mathcal{I}_xR_x-\mathcal{I}_3R_3 = 0\\ \mathcal{I}_1R_1 - \mathcal{I}_2R_2 = 0 \end{cases} \end{equation} \begin{equation}\label{a4} \begin{cases} \mathcal{I}_x=\mathcal{I}_1\\ \mathcal{I}_3=\mathcal{I}_2 \end{cases} \end{equation} Подставим (\ref{a4}) в (\ref{a3}): \begin{equation}\label{a5} \mathcal{I}_1R_x=\mathcal{I}_3R_3 \end{equation} \begin{equation}\label{a6} \mathcal{I}_1R_1=\mathcal{I}_3R_2 \end{equation} Поделим (\ref{a5}) на (\ref{a6}): $$R_x = \frac{R_1R_3}{R_2}$$ \item \textit{Можна ли использовать для измерения маленьких сопротивлений? Проанализируйте и обоснуйте ответ.} Нет. Это метод не можна использовать для измерения маленьких сопротивлений, т.\,к. при этих условиях через схему будет течь слишком большой ток, что приведёт к появлению значительныйы погрешностей измерений. \end{enumerate} \end{document} FuruyamaTakeshi/DLNAclinkc/std/av/wrapper/objc/CyberLink/latex/interface_c_g_xml_node.tex \hypertarget{interface_c_g_xml_node}{\section{C\-G\-Xml\-Node Class Reference} \label{interface_c_g_xml_node}\index{C\-G\-Xml\-Node@{C\-G\-Xml\-Node}} } {\ttfamily \#import $<$C\-G\-Xml\-Node.\-h$>$} Inheritance diagram for C\-G\-Xml\-Node\-:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=4.000000cm]{interface_c_g_xml_node} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hypertarget{interface_c_g_xml_node_ad90b29bc2d65ce4b349ecab18db6a3e4}{(id) -\/ {\bfseries init}}\label{interface_c_g_xml_node_ad90b29bc2d65ce4b349ecab18db6a3e4} \item \hypertarget{interface_c_g_xml_node_acee52fdfbc894dfc20d3dcc5e2ff942a}{(id) -\/ {\bfseries init\-With\-X\-M\-L\-Node\-:}}\label{interface_c_g_xml_node_acee52fdfbc894dfc20d3dcc5e2ff942a} \item \hypertarget{interface_c_g_xml_node_acd08c6d9d64e3c7dee41be31543db633}{(N\-S\-String $\ast$) -\/ {\bfseries attribute\-Value\-For\-Name\-:}}\label{interface_c_g_xml_node_acd08c6d9d64e3c7dee41be31543db633} \item \hypertarget{interface_c_g_xml_node_a655323c5db995c6f1584e00de2b0cae6}{(N\-S\-String $\ast$) -\/ {\bfseries element\-Value\-For\-Name\-:}}\label{interface_c_g_xml_node_a655323c5db995c6f1584e00de2b0cae6} \item \hypertarget{interface_c_g_xml_node_a176e7da0a7ebf7621b884af50694f3a2}{(N\-S\-String $\ast$) -\/ {\bfseries string\-Value}}\label{interface_c_g_xml_node_a176e7da0a7ebf7621b884af50694f3a2} \item \hypertarget{interface_c_g_xml_node_a1dcaecf10b29c55303235f5af3ea7a0e}{(void) -\/ {\bfseries set\-String\-Value\-:}}\label{interface_c_g_xml_node_a1dcaecf10b29c55303235f5af3ea7a0e} \item \hypertarget{interface_c_g_xml_node_a4de8210ccd53cd0efff2c2ffb236ad9d}{(void) -\/ {\bfseries set\-Attribute\-With\-Name\-:string\-Value\-:}}\label{interface_c_g_xml_node_a4de8210ccd53cd0efff2c2ffb236ad9d} \end{DoxyCompactItemize} \subsection*{Properties} \begin{DoxyCompactItemize} \item \hypertarget{interface_c_g_xml_node_abf4de5c626ff62dd5b87b9ce92e0a162}{N\-S\-X\-M\-L\-Element $\ast$ {\bfseries xml\-Node}}\label{interface_c_g_xml_node_abf4de5c626ff62dd5b87b9ce92e0a162} \item \hypertarget{interface_c_g_xml_node_abe2618247082b4f7172ee70be264bfad}{id {\bfseries user\-Info}}\label{interface_c_g_xml_node_abe2618247082b4f7172ee70be264bfad} \end{DoxyCompactItemize} \subsection{Detailed Description} The \hyperlink{interface_c_g_xml_node}{C\-G\-Xml\-Node} class is a wrapper class for N\-S\-X\-M\-L\-Node. The documentation for this class was generated from the following files\-:\begin{DoxyCompactItemize} \item C\-G\-Xml\-Node.\-h\item C\-G\-Xml\-Node.\-m\end{DoxyCompactItemize} % Szglab4 % =========================================================================== % \section{Napló} \begin{naplo} \bejegyzes {2014.03.05.~8:00~} % Kezdet {1 óra} % Időtartam {\vadam\newline \vantal\newline \vbator\newline \vtorok} % Résztvevők {Kezdeti megbeszélés, hibák áttekintése.} % Leírás \bejegyzes {2014.03.09.~10:30~} % Kezdet {4 óra} % Időtartam {Tallér} % Résztvevők {Szekvencia- és osztálydiagramok elkészítése.} \bejegyzes {2014.03.09.~17:00~} {2 óra} {\vadam} {Map és Waypoint osztályok leírása, state chartok kijavítása.} \bejegyzes {2014.03.09.~22:00~} % Kezdet {2,5 óra} % Időtartam {Tallér} % Résztvevők {Osztályok leírásának frissítése, ellenőrzése. Dokumentum összeszerkesztése.} \bejegyzes {2010.03.09.~22:30~} {2 óra} {\vadam} {Tower, Obstacle és Projectile osztályok átírása.} \bejegyzes {2010.03.09.~22:30~} {2 óra} {\vantal} {Enemy, Gem, Game osztályok frissítése.} \bejegyzes {2010.03.10.~8:30~} {0,5 óra} {\vbator} {Ellenőrzés.} \end{naplo} \hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4}{}\section{testing\+:\+:internal\+:\+:Function$<$ R(A1, A2)$>$ Struct Template Reference} \label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4}\index{testing\+::internal\+::\+Function$<$ R(\+A1, A2)$>$@{testing\+::internal\+::\+Function$<$ R(\+A1, A2)$>$}} Inheritance diagram for testing\+:\+:internal\+:\+:Function$<$ R(A1, A2)$>$\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=550pt]{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4__inherit__graph} \end{center} \end{figure} Collaboration diagram for testing\+:\+:internal\+:\+:Function$<$ R(A1, A2)$>$\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=193pt]{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4__coll__graph} \end{center} \end{figure} \subsection*{Public Types} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a025f5192252366d73aa19718bb0ea89d}\label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a025f5192252366d73aa19718bb0ea89d}} typedef A2 {\bfseries Argument2} \item \mbox{\Hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a2de00437877c29ec6cb78396928b8e3e}\label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a2de00437877c29ec6cb78396928b8e3e}} typedef \+::testing\+::tuple$<$ A1, A2 $>$ {\bfseries Argument\+Tuple} \item \mbox{\Hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_ad07042129ff6370f55a279ad12f5e80f}\label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_ad07042129ff6370f55a279ad12f5e80f}} typedef \mbox{\hyperlink{structtesting_1_1internal_1_1_matcher_tuple}{Matcher\+Tuple}}$<$ Argument\+Tuple $>$\+::type {\bfseries Argument\+Matcher\+Tuple} \item \mbox{\Hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_ada1ad22fa21c84ec3faea47ed20c1b46}\label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_ada1ad22fa21c84ec3faea47ed20c1b46}} typedef void {\bfseries Make\+Result\+Void}(A1, A2) \item \mbox{\Hypertarget{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a89033ea870fe831b13899ce36666e102}\label{structtesting_1_1internal_1_1_function_3_01_r_07_a1_00_01_a2_08_4_a89033ea870fe831b13899ce36666e102}} typedef \mbox{\hyperlink{classtesting_1_1internal_1_1_ignored_value}{Ignored\+Value}} {\bfseries Make\+Result\+Ignored\+Value}(A1, A2) \end{DoxyCompactItemize} The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item vendor/googletest/googlemock/include/gmock/internal/gmock-\/generated-\/internal-\/utils.\+h\end{DoxyCompactItemize} \newcommand{\project}[1]{\gdef\@project{#1}} \newcommand{\@project}{} \newcommand{\supervisor}[1]{\gdef\@supervisor{#1}} \newcommand{\@supervisor}{} \renewcommand\maketitle{ \begin{titlepage} ~\\ \begin{figure}[h] \includegraphics[width=0.5\textwidth]{templates/fhnwlogo.pdf} \end{figure} ~\\[1cm] \textbf{\Large \@project} ~\\[0.8cm] \textbf{\fontsize{22}{10}\selectfont \@title} ~\\[3cm] \textit{\Large \@author } \vfill \hfill \@supervisor \hspace{1cm} \noindent\hspace{-1in}\rule{19cm}{6pt}\\ \large{{\today}} \end{titlepage} } \subsection{Real-valued functions}\label{subsec:real_valued_functions} \begin{definition}\label{def:epigraph} Let \( X \) be an arbitrary set. The \term{epigraph} of the function \( f: X \to \BbbR \) is defined as \begin{equation*} \epi f \coloneqq \{ (x, r) \in X \times \BbbR \colon r \geq f(x) \}, \end{equation*} \end{definition} \noindent %Agregar siempre que de por culo la indentación \chapter{Resultados} %Descripción de los datos obtenidos. No es un análisis de la conclusión, llegará en el capitulo que viene. Es simplemente una ordenación de los datos \section{Análisis automático} A partir del análisis automático descritos en el apartado anterior. Hemos obtenido una proporción entre el dialogo de varones y mujeres. Siendo 0\% exclusivo de los varones y 100\% exclusivo de las mujeres. \begin{figure} \centering \includegraphics[scale=0.7]{Images/distribucion.jpg} \caption{ocurrencias en tramos del 10\% de proporción de diálogo por género} \end{figure} \newpage \section{Análisis manual} A partir del análisis manual hemos realizado preguntas diferentes, debido a la diferencia de las pruebas que hemos podido automatizar y las que no. Nuestra intención ha sido la de automatizar lo mas posible para poder abarcar el total de las películas de Disney y eventualmente poder mejorar la calidad de los datos. Pero nos hemos encontrado con diferentes fuentes y formatos para los mismos datos de cada película. Especialmente debido a la naturaleza intelectual de los datos y en menor medida debido a la falta de interés de normalizar estos parámetros. En el siguiente gráfico vemos una serie de películas ordenadas temporalmente con la tasa de tiempo de hombres y mujeres sin apreciarse especial inclinación a un lado u otro. Igualmente, en la suma total de tiempos se ve una desviación de la mitad, fácilmente atribuible al ruido. \begin{figure} \centering \includegraphics[scale=0.7]{Images/peliculas.png} \caption{Distribución de las líneas de diálogo por sexos y películas ordenadas por fecha de estreno} \end{figure} \begin{figure} \centering \includegraphics[scale=0.7]{Images/general.png} \caption{Distribución de las líneas de diálogo por sexos dentro de las películas analizadas} \end{figure}computations/exam2.tex1-10 \documentclass[]{article} \usepackage{caption,subcaption,graphicx,float,url,amsmath,amssymb,tocloft,wasysym,amsthm,thmtools,textcomp,listings,amsfonts,cancel} \usepackage[hidelinks]{hyperref} \usepackage[toc,acronym,nonumberlist]{glossaries} \usepackage[]{algorithm2e} \setacronymstyle{long-short} \usepackage{glossaries-extra} \graphicspath{{figs/}} \setlength{\cftsubsecindent}{0em} \setlength{\cftsecnumwidth}{3em} \setlength{\cftsubsecnumwidth}{3em} \newcommand\numberthis{\addtocounter{equation}{1}\tag{\theequation}} \newtheorem{thm}{Theorem} \newtheorem{cor}[thm]{Corollary} \setcounter{tocdepth}{1} \usepackage[toc,page]{appendix} %opening \title{ Computation in Complex Systems\\ Week 2\\ Algorithms \& Landscapes } %\makeglossaries \begin{document} \maketitle \tableofcontents %\listofalgorithms \section{Maximum independent set} \subsection{What is the maximum value of the independent set?} \begin{figure}[H] \begin{center} \caption{What is the maximum value of the independent set?} \includegraphics[width=0.6\textwidth]{mwisQ} \end{center} \end{figure} Observations: \begin{enumerate} \item A tree with $N$ nodes has $2^N-1$ non empty subsets, so a brute-force search is exponential. Let's not go there! \item Either the maximal independent set includes the root of the tree, or it doesn't. \item If the the maximal independent set doesn't include the root of the tree, each subtree must be maximal. \item If the maximal independent set includes the root of the tree, each subtree must have a set that excludes the node. \end{enumerate} So we can solve this problem as follows: \begin{enumerate} \item Either the root node, $3$ is in $M$ or it isn't. \item If $3$ is in the tree, we have to exclude {4,1,5}, which gives a score of $3+(1+2)+2+(1+1)=9$ \item If $3$ isn't in the tree, the maximum is the sum of the maxima for the 4 subtrees: \begin{enumerate} \item (4,(1,2)): since $1+2<4$, the maximum value is 4. \item (1,(2)): since $1<2$, the maximum value is 2. \item ((5),(1,1)): since $1+1<5$, the maximum value is 5. \end{enumerate} \item Putting this together, if $3$ isn't in the tree, the maximum is $4+2+5=10$ \end{enumerate} \subsection{Algorithm for maximum independent set} Algorithm \ref{alg:find:maximum:independent:set} is a description is pseudo-code, and Appendix \ref{sect:ref:implementation} provides a reference implementation in Python. \begin{algorithm}[H] \KwData{A binary tree $T$ with a value $V(n)$ assigned to each node $n$} \KwResult{A set $S\subseteq N(T) | I(S) \land \sum_{s\in S} V(s) \ge \sum_{s\in S^\prime} V(s) \forall S^\prime \subset N(T) \text{ with } I(S^\prime)$ } initialization\; Starting at the root of the tree, traverse it top down to build two data structures: a stack \emph{unprocessed}, which has the \emph{root} at the bottom, and the $leaves$ at the top; and a dictionary \emph{score}, which is initially empty, but will map each node to the value of the maximum independent set starting at that node.\; \While {$length(unprocessed)>0$}{ $node \leftarrow pop(unprocessed)$\; \eIf {leaf(node)}{ $score(node)\leftarrow value(node)$ \;}{ $val_{out} \leftarrow \sum_{\forall child(node)} score(child)$\; $val_{in} \leftarrow value(node) + \sum_{\forall grandchild(node)} value(grandchild)$\; $score(node)\leftarrow max(val_{out},val_{in})$\; } } \caption{Find the maximum independent set from a tree}\label{alg:find:maximum:independent:set} \end{algorithm} \section{Reductions \& Translations} Table \ref{table:transform} illustrates the transformation of ASTRO to START. It exploits the common subsequence S-T-R to produce a minimum set of transformations. \begin{table}[H] \begin{center} \caption{Transforming ASTRO to START}\label{table:transform} \begin{tabular}{|c|c|c|c|c|c|c|l|l|} \hline \textbf{\cancel{A}}&S&T&&R&O&&delete&$\rightarrow$\\\hline &\textbf{S}&T&&R&O&&match&$\searrow$\\\hline &S&\textbf{T}&&R&O&&match&$\searrow$\\\hline &S&T&\textbf{A}&R&O&&insert&$\downarrow$\\\hline &S&T&A&\textbf{R}&O&&match&$\searrow$\\\hline &S&T&A&R&\textbf{\cancel{O}}&&delete&$\rightarrow$\\\hline &S&T&A&R&&\textbf{T}&insert&$\downarrow$\\\hline \end{tabular} \end{center} \end{table} \subsection{Assign Weights} Figure \ref{fig:Shortest:path} shows the path that gives rise to the transformations in Table \ref{table:transform}. In order to make this the shortest path we need the following rules: \begin{itemize} \item If two symbols are equal, accept them and mover diagonally; \item If two symbols don't match, insert or delete; disfavour mismatches (point mutations). \end{itemize} The simplest way to achieve this is with the weights shown in Figure \ref{fig:weights}. \begin{figure}[H] \begin{center} \caption{Weights}\label{fig:weights} \includegraphics[width=\textwidth]{q2-weights} \end{center} \end{figure} \subsection{Shortest path} Copying the weights from Figure \ref{fig:weights} we get Figure \ref{fig:Shortest:path}. The shorted path has length $1+0+0+1+0+1+1=4$. \begin{figure}[H] \begin{center} \caption{Shortest path}\label{fig:Shortest:path} \includegraphics[width=0.6\textwidth]{q2-path} \end{center} \end{figure} % glossary %\printglossaries % bibliography go here %\bibliographystyle{unsrt} %\addcontentsline{toc}{section}{Bibliography} %\bibliography{origins,wikipedia} \begin{appendices} \section{Reference Implementation for Algorithm \ref{alg:find:maximum:independent:set}}\label{sect:ref:implementation} One way to implement Algorithm \ref{alg:find:maximum:independent:set} is with the following code. \lstinputlisting[language=Python]{mis.py} \end{appendices} \end{document} jneem/bibtex-rs @article{key, title = "This title contains an {unopened} brace}.", } DecDury/SMACC2_Documentation \hypertarget{ManualTracing_8md}{}\doxysection{tracing/\+Manual\+Tracing.md File Reference} \label{ManualTracing_8md}\index{tracing/ManualTracing.md@{tracing/ManualTracing.md}} uliska/lualilyglyphs %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % This file is part of the 'lilyglyphs' LaTeX package. % % ========== % % % % https://github.com/openlilylib/lilyglyphs % % http://www.openlilylib.org/lilyglyphs % % % % Copyright 2012-2020 and others, % % % % 'lilyglyphs' is free software: you can redistribute it and/or modify % % it under the terms of the LaTeX Project Public License, either % % version 1.3 of this license or (at your option) any later version. % % You may find the latest version of this license at % % http://www.latex-project.org/lppl.txt % % more information on % % http://latex-project.org/lppl/ % % and version 1.3 or later is part of all distributions of LaTeX % % version 2005/12/01 or later. % % % % This work has the LPPL maintenance status 'maintained'. % % The Current Maintainer of this work is Urs Liska (see above). % % % % This work consists of the files listed in the file 'manifest.txt' % % which can be found in the 'license' directory. % % % % This program is distributed in the hope that it will be useful, % % but WITHOUT ANY WARRANTY; without even the implied warranty of % % MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{lilyglyphsManualFonts} \RequirePackage{fontspec} % Of course we need lilyglyphs itself :-) \RequirePackage{lilyglyphs} % fontspec base settings \setmainfont[% Ligatures=TeX, Numbers=OldStyle]{Linux Libertine O} \setsansfont[Scale=MatchLowercase, ItalicFont=LinBiolinumOI, BoldFont=LinBiolinumOB, Ligatures=TeX]{LinBiolinumO} \setmonofont[Scale=MatchLowercase]{DejaVu Sans Mono} \linespread{1.05} \hypertarget{dir_92709420fde8ca446636ff7c23065e8b}{\section{/home/zvebabi/\+Documents/workspace/\+L\+E\+D\+Microsensor\+\_\+project0/analizer/\+Release Directory Reference} \label{dir_92709420fde8ca446636ff7c23065e8b}\index{/home/zvebabi/\+Documents/workspace/\+L\+E\+D\+Microsensor\+\_\+project0/analizer/\+Release Directory Reference@{/home/zvebabi/\+Documents/workspace/\+L\+E\+D\+Microsensor\+\_\+project0/analizer/\+Release Directory Reference}} } \subsection*{Files} \begin{DoxyCompactItemize} \item file \hyperlink{analizer_8d}{analizer.\+d} \item file \hyperlink{_s_p_i_8d}{S\+P\+I.\+d} \end{DoxyCompactItemize} tables/table2-1.tex % -*- coding: utf-8; -*- \begin{table}[H] \centering \begin{tabular}{l>{\raggedright}p{4in}} \toprule Elemento & \centering{}Descripción\tabularnewline\midrule $C$ & es el conjunto de todas las configuraciones posibles del entorno, es decir, los obstáculos y las configuraciones libres.\tabularnewline $C_{free}$ & Subconjunto de C, configuraciones libres de obstáculos existentes.\tabularnewline $R$ & Indicador de ponderación de proximidad al punto deseado. La distancia euclidiana es la más utilizada.\tabularnewline $q_{init}$ & Configuración inicial\tabularnewline $q_{fin}$ & Configuración que se desea alcanzar\tabularnewline $q_{rand}$ & Configuración aleatoria dentro del espacio $C_{free}$ \tabularnewline $q_{near}$ & Es la conguración mas proxima a $q_{rand}$ , en el sentido denido por R, de entre las existentes en un árbol. Se evalúa con el indicador de ponderación.\tabularnewline $q_{new}$ & Configuración a añadir al árbol.\tabularnewline $e$ & Longitud de segmento de crecimiento. Geométricamente, es la distancia entre un punto del árbol y el siguiente con el que esta conectado.\tabularnewline $Arbol$ & Estructura de datos.\tabularnewline \bottomrule \end{tabular} \caption{Elementos de una RRT} \label{Flo:cnomen} \end{table} quarterly_stuff/2020Q2/chapters/deentanglement_by_divergence.tex \chapter{De-Entanglement by Divergence} \section{Visual Question Answering} Visual Question Answering (VQA) involves answering a natural language query about an image. Questions can be arbitrary and they encompass many sub-problems in computer vision: (1) Object recognition %- What is in the given image? (2) Object detection %- Are there any cats in the image? (3) Attribute classification %- What color is the cat if present ? (4) Scene classification %- Is it raining? (5) Counting. % - How many dogs are in the image? VQA is characterized by wide ranging applications from helping visually impaired people through human machine interaction. It has the potential to serve as an effective media content retrieval framework. %VQA is a challenging task since it requires reasoning across multiple modalities. It involves additional challenges of knowledge representation and reasoning both within and across them. For example,given a picture of cat, the question ‘Is the cat black in color?’ can be answered using the visual modality once the model understands that it is to check the color of cat. The information about what to look for comes from the text modality. So in the general setting, the VQA model has to combine the information from both the modalities and reason over this combined representation. A primary form of implementing a VQA system would be to use a bucketing approach: by learning image and text features and fusing them to get an answer. In recent years, there have been several extensions to the trivial approach mentioned above \cite{fukui2016multimodal}; \cite{lu2016hierarchical}; \cite{yang2016stacked}; \cite{lu2015deeper} claim to learn good representations of abstract concepts needed to answer questions. However, it has been shown \cite{agrawal2017c} that most of the approaches capture surface level correlations and fail to handle unseen novel combinations during test time. %Recently, the ability of VQA models to handle compositionality has been greatly sought after. Few models address the inferencing and reasoning abilities of VQA models that is provided by the properties of compositionality. In this work, we investigate approaches to improve compositionality in VQA, where we explicitly focus on learning compositionality between concepts and objects. Language and vision are inherently composite in nature. For example different questions share substructure viz \textit{Where is the dog?} and \textit{Where is the cat?} Similarly images share abstract concepts and attributes viz \textit{green pillow} and \textit{green light}. Hence it is vital not only to focus on understanding the information present across both these modalities, but also to model the abstract relationships so as to capture the unseen compositions of seen concepts at test time. Achieving this would then allow the model to generalize better by learning an inference procedure, resulting in true success on this task. In this work, we propose \textbf{\textit{JUPITER}} - \textbf{JU}stification via \textbf{P}ointwise combination of \textbf{I}mage and \textbf{T}ext based on \textbf{E}xpected \textbf{R}ewards, is built on top of the Neural Module Networks \cite{HuARDS17}. This is motivated from our hypothesis that generating captions can provide additional information to improve VQA. Additionally, JUPITER uses Reward Augmented Maximum Likelihood \cite{RAML}, which is improves caption generation. \section{Related Work} \noindent\textbf{Visual Question Answering}: \cite{KazemiE17} provided a strong baseline for VQA using a simple CNN-LSTM architecture, and achieved 64.6\% on the VQA 1.0 Openended QA challenge. This further proved that the dataset is biased. \cite{AishAgrawal17} introduced grounding to prevent the model from memorizing this bias. Similarly, \cite{li2018zero} used a zero-shot training approach to improve the generalizabilty of the model, and prevent the model to learn the bias. However, recently \cite{AgrawalKBP17} showed that most models degrade in performance when tested on unseen samples. In this work, we aim to tackle this lack of generalizability. \\ \noindent\textbf{Neural Module Networks}: To the best of our knowledge, the work by authors in \cite{HuARDS17} and \cite{deepmodulenets} is the only work so far that explicitly uses a divide and conquer approach for compositionality. Natural language questions are best answered when broken down into their subparts. The authors use a similar intution and propose a modular architecture. This approach first parses the natural language question into linguistic components. Second, each component is assigned to a sub-module that solves a single task. Lastly, these modules are then composed into an appropriate layout that predicts an answer for each training example. Such a dynamic network not only helps learning object-object relationships well via compositionally, but also improves the reasoning abilities of the model. % Our approach \iffalse \noindent\textbf{Multimodal Compact Bilinear Pooling }: \cite{fukui2016multimodal} propose to alter the fusion technique using the multimodal compact bilinear (MCB) pooling that calculates the outer product between two vectors thereby allowing for a multiplicative interaction between all elements of both vectors. The image and the text representations are randomly projected to higher dimensions and then using element-wise product in Fast Fourier Transform (FFT) space to convolve both vectors efficiently. A soft attention is done on the MCB to incorporate spatial attention to help identify important regions in the image. They also use MCB to encode variable length answers and then get a joint represenation over the answers and the image features. \fi \\ \noindent\textbf{Multitask Learning}: There have been number of works that explore multitask learning as an approach to joint learning of vision and language tasks. In one such work \cite{JustinJohnson2018}, authors learn related regions of the image by simultaneously training three different semantic tasks - scene graph generation, object detection, and image captioning. A multi-task learning architecture was also proposed by \cite{zhao2018multi} for image captioning where they enable sharing of a CNN encoder and an LSTM decoder between object classification task and the syntax generation tasks. \cite{ruder2017overview, lin2018multi} show mutlitask learning reduces overfitting in limited-resource settings, and can learn representations to improve downstream (part-of-speech tagging and name-entity recognition) tasks. Our purpose of joint training in multitask learning is to provide regularization on the learned features for VQA, with an added benefit of achieving better performance on the auxiliary task (of generating captions). \noindent\textbf{Incorporating additional knowledge}: In \cite{chandu2018textually} authors show that incorporating captions helps resolve some ambiguities in visual question answering. In \cite{aditya2018explicit} authors first obtain captions and then use them for improving VQA via the framework of predicate logic. In \cite{wu2016ask} authors learn attributes from an image using an image labeling and then query using an external knowledge base. \section{JUPITER - Justification by Pointwise combination of Image and Text based on Expected Rewards} The key motivation of this approach [depicted in Figure: \ref{fig:jupiter}] was to manipulating the loss function to account for captions. We hyothesize that explicitly accounting for captions in the loss function will affect the downstream VQA predictions. Figure \ref{fig:jupiter} shows the framework architecture and functioning. %In this approach, we first present the systems we built that involve manipulating the loss function. At a higher level, we want to modify/augment the loss function utilizing captions thereby affecting the answers that get generated by the system. %\subsubsection{Model Description} \subsubsection{Model Description} Our model uses Neural Module Networks (NMN), along with multiple proposed extensions. More specifically, we use the following extensions: %We have investigated the following settings to accomplish this: \begin{itemize} \item \textit{Multitask Learning}: We modify the decoder to perform multiple tasks namely, caption generation and VQA. We use the attention grid generated by \textit{`Find'} module in the NMN, the encoded question layout, and the input image to generate captions in an auto regressive. Our hypothesis is that using this conditioning, we can force the model to generate attention grid that is suitable to both downstream tasks, in turn improving VQA performance. %\textbf{cite work where MTL improves both tasks.} %In this setting we aim to use caption generation as a secondary task in VQA. For this, we use the attention grid generated by the `Find' module in the module network. Once we obtain the attention grid, we generate captions in an auto regressive fashion conditioned on the attention grid and the original image. The intuition behind this setting is to force the model to generate an attention grid that is suitable to both generate answer to the question as well as caption the image. \item \textit{Conditional Generation}: As opposed to multitask learning approach, in this extension we explicitly provide the generated captions as input to VQA decoder. More specifically, we train the model to first generate a relevant image caption using previously defined setup. Next we condition the answer decoder on the generated caption. The intuition is that providing the model with information more explicitly will help to predict answers based on this information. %In this setting we try to use captions more explicitly compared to the implicit approach of multitask learning. Specifically, in this setting we train the model to generate both the caption to the image as well as answer to a question about the image. Having said that, in the current setting the answer is generated conditioned on the generated captions. The intuition behind this approach is that the model is now more explicitly forced to utilize the information from captions while answering the question. \item \textit{Re-weighting}: In this extension, we re-weight the answer hypothesis using the generated caption. We hypothesize that this will help the model to disambiguate between answer logits that have maximum entropy. %In this setting we adjust the logits corresponding to the answers generated by the NMN using captions. The intuition behind this approach is to help the model disambiguate between confusable classes. For instance, if the model is confused between, say, glasses and eye glasses[gender is a better example], reweighting by captions might help the model discern the answer category. \item \textit{M-Hybrid and C-Hybrid}: In order to harvest complimentary benefits from our primary extensions, we also implemented two hybrid systems. M-Hybrid extension combined multitask learning and re-weighting approach, and the C-Hybrid extension combined conditional generation and re-weighting approach.% We have also investigated combining Multi task Learning, Reweighting settings as M Hybrid and Conditional Generation, Reweighting settings as C Hybrid systems \item \textit{Reinforcement Learning}: This extension uses Reward Augmented Maximum Likelihood (RAML) as opposed to Maximum Likelihood (MLE) for generating captions. The intuition for this extension was to enable the agent to generate captions that will help the model to answer the given question. More specifically, the agent at each caption generation step can perform one of the two tasks: (1) Generate next word for the captions or (2) Answer the question based on caption generated so far. The agent is rewarded based on VQA accuracy. Since training with REINFORCE is known to be unstable, we use a baseline wherein we generate answers based on the final hidden state of a deocder trained using MLE. % Maximum likelihood trained decoder. %This setting is built on top of the C-Hybrid system. In this setting, we use Reward Augmented Maximum Likelihood to obtain captions instead of Maximum Likelihood. The intuition behind this system was to enable the agent generate captions that help it answer the question about given image. To train the model in this setting, we pose the problem of VQA as a reinforcement learning problem. Specifically, the agent has one of the two tasks: (1) Generate next logits for the captions or (2) Answer the question based on the already generated captions. The agent is rewarded based on the accuracy of VQA. Since training using REINFORCE is known to be unstable, we use a baseline wherein we generate answer based on the ultimate hidden state of a Maximum likelihood trained decoder. \end{itemize} \begin{figure}[h] \centering \includegraphics[scale=0.44, frame]{images/jupiter.png} \caption{Justification by Pointwise combination of Image and Text based on Expected Rewards} \label{fig:jupiter} \end{figure} \subsubsection{Learning} We denote input question as \textit{Q} and input image as \textit{I}. \textit{L*} denotes the gold layout for \textit{Q} and \textit{C*} is gold caption for \textit{I}. We denote \textit{L} as the layout generated by NMN for \textit{Q}. \textit{C} is caption generated from JUPITER. We denote answer classes by \textit{y} and the correct answer class by \textit{y*}. \textit{T} is the training data samples of type (\textit{I, Q, y*}). Next, we describe the objective function for each extension in detail. \begin{itemize} \item \textit{Multitask Learning}: We use a two-part objective function for multitask learning. The first part is generating captions from the input and the second is generating answer logits from the input and the generated NMN layout. \\ \begin{equation} L(\theta) = \sum_{(I, Q, y*) \in T} logP_\theta(y | I, Q, L) + logP_\theta(C | I, Q) \end{equation} \item \textit{Conditional Generation}: This extension uses a similar objective function. However, we generate answer logits from the input, generated NMN layout as well as the generated captions. \begin{equation} L(\theta) = \sum_{(I, Q, y*) \in T} logP_\theta(y | I, Q, L, C) + logP_\theta(C | I, Q) \end{equation} \item \textit{Re-weighting}: This extension uses a similar objective as conditioned generation. Further, for re-weighting we define new answer logits \textit{y'}. \begin{equation} y' = C_{T} y \end{equation} where, C$_{T}$ is the final hidden state of generated caption, and y is the previous answer logits. The updated objective function is: \begin{equation} L(\theta) = \sum_{(I, Q, y*) \in T} logP_\theta(y' | I, Q, L, C) + logP_\theta(C | I, Q) \end{equation} \item \textit{Reinforcement Learning}: The agent transitions between generating next word in the caption and generating final answer. The agent receives minibatch VQA accuracy as its reward. The Baseline we use to stabilize the training and the expected reward of our agent respectively are expressed as \begin{equation} L_{baseline}(\theta) = \sum_{(I, Q, y*) \in T} logP_\theta(y | I, Q, L, C) \end{equation} \iffalse \begin{equation} Reward = \sum_{n=0}^{N} ( r_n) \end{equation} where N is the length of captions. \fi \end{itemize} We use cross-entropy loss to train the model. We jointly train our captions module in JUPITER alongside NMN, which learns a question layout \textit{L}. \section{Dataset and Input Modalities} VQA dataset by \cite{AntolALMBZP15} has 265016 images, 614163 questions. The dataset consists of 82,783 training, 40,504 validation, and 40,775 test images. Each image has 3 questions on average and 10 ground truth answers. Questions as well as answers are open ended, accounting for a more real-world scenario. The questions are rich in a way, as the require the model to have complex reasoning and understanding abilities. \section{Results and Discussion} \iffalse When presenting your results, try to follow the same sequence as you specified in the introduction of the Experimental Setup. If you stated 3 research questions, then it would be best to have 3 sub-sections in the Results and Discussion, one for each research question. Present in tables and/or figures your experimental results. It is not enough just to list the results you need to discuss them – what do they mean, what implications they have, how should they be interpreted in the broader context. If you want to make your results more convincing, where possible, include statistical tests to demonstrate the statistical significance of your findings. You should also include a discussion (if possible, with examples and figures) of the failure cases of the baseline models. \fi \begin{figure} [h] \centering \includegraphics[scale=0.3]{images/captions.png} \caption{Qualitative Analysis from JUPITER: left image depicts a scenario where generating caption helped the model in selection of the right answer. Image in the center depicts a scenario where captions end up confusing the model. Image in the right most highlights an interesting scenario where the generated caption seems irrelevant.} \label{fig:jupiter_qualitative} \end{figure} In this section, we discuss the results from our proposed approaches viz. JUPITER, VENUS and MARS, and compare them against our baselines. Table \ref{results_table} consolidates the results of our experiments. To better understand the performance of these models, we report the performance across different answer categories namely, Number, Yes/No and Other. The overall best baseline model for VQA is NMN by \cite{HuARDS17}. %In this section, we discuss the results of our proposed approaches in comparison to our baseline models on VQA. %We first discuss the baseline results for each of the proposed approach domains. For a better understanding of the models performance, we provide the results by splitting the category of answers into number, yes/no and other type. The final baseline that we aim to improve over is the one set by module networks. section{Results: Baseline Models} The input to our baseline models is the image and the question. We do not use any external knowledge. Our results show that the baseline models have highest accuracy on the Yes/No questions. However, the Number type questions often require deeper understanding of the image, and so our baselines have lowest performance on them. Humans tend to have low agreement for Yes/No questions. We attribute this to question ambiguity or missing information in the image.It has to be noted that our implementation of the NMN baseline achieves better scores compared to the open source original implementation. This can be attributed to the presence of additional modules in our implementation, specifically OR, COUNT, FILTER, and EXIST modules. %Comparing across our baselines, NMN outperforms the rest. We believe this is since the model explicitly handles compositionality, providing better generalizability. Amongst the fusion baselines, the RNN performs worse than the VED baseline. We believe this is because VED captures latent representations which are more useful for VQA. %Also, our implementation of the NMN baseline is much better than the scores on the original implementation repository. The incorporation of modules - OR, COUNT, FILTER, and EXIST as mentioned in the paper helped generate layouts which were more valid, therefore our implementation was better. %The baselines all input only the question and the image, with no addition of external knowledge. Comparing across the three baselines, we see that the NMN is the best baseline we can get for VQA. %Question Type: number Accuracy: 26.352015732546707 %Question Type: yes/no Accuracy: 64.48818031885652 %Question Type: other Accuracy: 31.55374946530223 %43.226183422213445 overall %Question Type: number Accuracy: 23.310390036053754 %Question Type: other Accuracy: 26.649336974762264 %Question Type: yes/no Accuracy: 63.936228697086314 %40.18450852590691 overall \begin{table}[!h] \centering \small %\scriptsize \begin{tabular}{@{}lllccc@{}} %\toprule % & &\textbf{Accuracy(\%)} \\ \toprule { \textbf{Model}} & {\textbf{{System}}} & {\textbf{{Input}}} & \textbf{{Number}} & \textbf{{Yes/No}} & \textbf{{Other}} \\ \midrule Human & Best & Image + Question & 83.39 & 95.77 & 72.67 \\ Human & Worst & Image + Question & 65.28 & 46.52 & 78.02 \\ \midrule NMN & Baseline (Replicated) & Image + Question & 23.31 & 63.93 & 26.65 \\ NMN & Baseline (Our implementation) & Image + Question & 26.35 & 64.49 & 31.55 \\ RNN & Baseline & Image + Question & 19.34 & 57.82 & 17.77 \\ VED & Baseline & Image + Question & 17.76 & 58.00 & 10.43 \\ \midrule RNN & MARS & Image + Question + Caption* & 23.09 & 57.88 & 18.22 \\ RNN & MARS & Question + Caption* & 21.72 & 57.95 & 21.59 \\ VED & VENUS & Image + Question + Caption* & 19.25 & 57.83 & 10.10 \\ VED & VENUS & Question + Caption* & 18.13 & 58.10 & 10.33 \\ \midrule %NMN & Conditional & Image + Question + Caption & & & \\ %NMN & Rescore & Image + Question + Caption & 27.48 & 65.8 & 32.2 \\ %NMN & Multitask & Image + Question + Caption & & & \\ NMN & M-Hybrid & Image + Question + Caption & 26.31 & 64.27 & 30.43 \\ NMN & C-Hybrid & Image + Question + Caption & 27.48 & 65.8& 32.2\\ NMN & JUPITER & Image + Question + Caption & 32.82 & 67.95 & 33.15 \\\bottomrule \end{tabular} \caption{Results from human, baselines and proposed approaches. * denotes systems that employ Gold captions} \label{results_table} \end{table} section{Results: Proposed Models} Looking at the objective evaluation results from table \ref{results_table}, it is clear that incorporating captions leads to improvements across the approaches. This result empirically validates our hypothesis related to captions: Captions help VQA. To understand the extent of this, we have also performed ablation analysis wherein we have used just captions to answer the question ignoring the input image. Surprisingly, systems built in this fashion seem to perform better than our baselines. This leads to an interesting observation: \textit{Captions seem to contain supplementary and in some cases complementary information to the images themselves}. However, we acknowledge that proving such hypothesis would require additional experimentation. For instance, it would be interesting to perform similar ablation analyses employing computationally more powerful frameworks such as attention as baselines or adding more visual information such as ground truth bounding boxes. It is also interesting to note that the proposed approaches achieve better scores compared against the \textit{worst} human performance in Yes/No category. Our approach JUPITER outperforms all other approaches across all the categories. In addition, within the models employing module networks, the system employing reinforcement learning outperforms other approaches. This is in line with our hypothesis related to Reward Augmented Maximum Likelihood and raises interesting questions related to \textit{comparison between supervised approaches such as Maximum Likelihood and their reward based reinforcement counterparts}. It would be interesting to perform a much larger scale evaluation comprehensively comparing the effectiveness of these approaches in the context of downstream tasks. In figure \ref{fig:jupiter_qualitative}, we present some scenarios that highlight the way captions get utilized for answering question about the corresponding images. \hypertarget{class_magnum_1_1_physics3}{}\section{Magnum\+:\+:Physics3 類別 參考文件} \label{class_magnum_1_1_physics3}\index{Magnum\+::\+Physics3@{Magnum\+::\+Physics3}} \hyperlink{class_magnum_1_1_physics3}{Physics3} Services. {\ttfamily \#include $<$Physics3.\+h$>$} \subsection*{複合項目} \begin{DoxyCompactItemize} \item class \hyperlink{class_magnum_1_1_physics3_1_1_collision_filter}{Collision\+Filter} \begin{DoxyCompactList}\small\item\em \hyperlink{class_magnum_1_1_physics3_1_1_collision_filter}{Collision\+Filter}. \end{DoxyCompactList}\item class \hyperlink{class_magnum_1_1_physics3_1_1_convex_cast_result}{Convex\+Cast\+Result} \item class \hyperlink{class_magnum_1_1_physics3_1_1_manager}{Manager} \item class \hyperlink{class_magnum_1_1_physics3_1_1_material}{Material} \begin{DoxyCompactList}\small\item\em \hyperlink{class_magnum_1_1_physics3_1_1_material}{Material}. \end{DoxyCompactList}\item class \hyperlink{class_magnum_1_1_physics3_1_1_ray_cast_result}{Ray\+Cast\+Result} \item class \hyperlink{class_magnum_1_1_physics3_1_1_rigid_body}{Rigid\+Body} \begin{DoxyCompactList}\small\item\em \hyperlink{class_magnum_1_1_physics3_1_1_rigid_body}{Rigid\+Body} \hyperlink{class_magnum_1_1_component}{Component}. \end{DoxyCompactList}\item struct \hyperlink{struct_magnum_1_1_physics3_1_1_service}{Service} \item class \hyperlink{class_magnum_1_1_physics3_1_1_vehicle}{Vehicle} \item class \hyperlink{class_magnum_1_1_physics3_1_1_world}{World} \end{DoxyCompactItemize} \subsection{詳細描述} \hyperlink{class_magnum_1_1_physics3}{Physics3} Services. 此類別(class) 文件是由下列檔案中產生\+:\begin{DoxyCompactItemize} \item C\+:/\+Epic\+Force\+Backup/\+Epic\+Force\+Engine/\+Epic\+Force\+Engine/\+Magnum\+Engine\+Lib/\+Magnum\+Core/Physics3.\+h\end{DoxyCompactItemize} % vim: ft=tex spelllang=de_20 Wir befinden uns an der Schwelle einer Revolution zu einer Quantentechnologie, die nicht nur auf der passiven Nutzung von Quanteneffekten, sondern auf ihrer aktiven Kontrolle beruht. An vorderster Front beinhaltet dies die Realisierung eines Quantencomputers. Das Kodieren von Informationen in Quantenzuständen als ``Qubits'' erlaubt es, Verschränkung und Quantensuperposition zu nutzen, um Rechnungen durchzuführen, die auf einem klassischen Computer unpraktikabel sind. Eine zentrale Schwierigkeit ist es dabei, Dekohärenz zu vermeiden -- der Verlust von Quanteneigenschaften aufgrund ungewollter Wechselwirkung mit der Umgebung. Diese Arbeit thematisiert die Realisierung verschränkender Zwei-Qubit-Gatter, die sowohl gegenüber Dekohärenz als auch klassischen Störeinflüssen robust sind. Sie behandelt dabei drei Aspekte: die Nutzung effizienter numerischer Methoden zur Simulation und optimaler Kontrolle offener und geschlossener Quantensysteme, die Rolle fortgeschrittener Optimierungsfunktionale zur Begünstigung von Robustheit, sowie die Anwendung dieser Techniken auf zwei führende Umsetzungen von Quantencomputern, gefangene Atome und supraleitende Schaltkreise. Nach einem Überblick über die theoretischen und numerischen Grundlagen beginnt der zentrale Teil dieser Arbeit mit der Idee einer Ensembleoptimierung, um Robustheit sowohl gegenüber klassischen Fluktuationen als auch Dekohärenz zu erreichen. Für das Beispiel eines kontrollierten Phasengatters auf gefangenen Rydberg-Atomen wird gezeigt, dass Gatter erreichbar sind, die um mindestens eine Größenordnung robuster sind als der beste bekannte analytische Ansatz. Darüber hinaus bleibt diese Ro\/bust\/heit selbst dann erhalten, wenn die Gatterdauer signifikant gegenüber der kurzmöglichsten Dauer des analytischen Gatters verkürzt wird. Supraleitenden Schaltkreise sind eine besonders vielversprechende Architektur zur Implementierung eines Quantencomputers. Ihre Flexibilität wird durch Optimierungen sowohl für diagonale als auch nicht-diagonale Gatter gezeigt. Um Robustheit gegenüber Dekohärenz zu gewährleisten, ist es essentiell, das Gatter in so kurzer Zeit wie möglich zu realisieren. Das Erreichen dieses Ziels wird durch die Optimierung hin zu einem beliebigen perfekten Verschränker erleichtert, basierend auf einer geometrischen Theorie der Zwei-Qubit-Gatter. Für das Beispiel supraleitender Qubits wird gezeigt, dass dieser Ansatz zu kürzeren Gatterzeiten, höheren Fidelitäten, sowie schnellerer Konvergenz führt, im Vergleich zur Optimierung hin zu vorausbestimmten, festen Zwei-Qubit-Gattern. Eine Optimierung im Liouville-Raum zur sauberen Berücksichtigung von Dekohärenzeffekten ist mit erheblichen numerischen Herausforderungen verbunden, da die Dimension im Vergleich zum Hilbert-Raum quadratisch wächst. Es kann allerdings gezeigt werden, dass es für ein unitäres Optimierungsziel ausreichend ist, höchstens drei Zustände anstelle der vollen Basis des Liouville-Raums zu propagieren. Sowohl für das Beispiel gefangener Rydberg-Atome also auch für supraleitende Qubits wird die erfolgreiche Optimierung von Quantengattern gezeigt, mit einem numerischen Aufwand, der weit unterhalb der bisher angenommenen Untergrenze liegt. Insgesamt zeigen die Ergebnisse dieser Arbeit zu einem umfassenden Gerüsts zur Optimierung robuster Quantengatter, und bereiten den Weg für die mögliche Realisierung eines Quantencomputers. analysis/comparison/table.tex \begin{deluxetable}{lcc} \tablecaption{Orbital parameters \label{tab:orbits}.} \tablehead{\colhead{Parameter} & \colhead{primary solution} & \colhead{alternate solution}} \startdata \cutinhead{Sampled} $P_\mathrm{inner}$ [days] & $34.879 \pm 0.001$ & $34.879 \pm 0.001$\\ $a_\mathrm{inner}$ [mas] & $4.63 \pm 0.04$ & $4.63 \pm 0.04$\\ $M_\mathrm{Ab}$ [$M_\odot$] & $0.24 \pm 0.01$ & $0.24 \pm 0.01$\\ $e_\mathrm{inner}$ & $0.63 \pm 0.01$ & $0.63 \pm 0.01$\\ $i_\mathrm{inner}$ [\degr] & $131.5 \pm 0.8$ & $48.5 \pm 0.8$\\ $\omega_\mathrm{Aa}$\tablenotemark{a} [\degr] & $81 \pm 1$ & $81 \pm 1$\\ $\Omega_\mathrm{inner}$\tablenotemark{b} [\degr] & $104 \pm 9$ & $112 \pm 9$\\ $T_{0,\mathrm{inner}}$ [JD - 2,450,000] & $2704.57 \pm 0.07$ & $2704.57 \pm 0.07$\\ $P_\mathrm{outer}$ [yrs] & $548 \pm 244$ & $555 \pm 249$\\ $M_\mathrm{B}$ [$M_\odot$] & $0.41 \pm 0.28$ & $0.40 \pm 0.28$\\ $e_\mathrm{outer}$ & $0.3 \pm 0.2$ & $0.3 \pm 0.2$\\ $i_\mathrm{outer}$ [\degr] & $139 \pm 13$ & $139 \pm 13$\\ $\omega_\mathrm{A}$\tablenotemark{a} [\degr]& \nodata\tablenotemark{c} & \nodata\tablenotemark{c}\\ $\Omega_\mathrm{outer}$\tablenotemark{b} [\degr]& \nodata\tablenotemark{c} & \nodata\tablenotemark{c}\\ $T_{0,\mathrm{outer}}$ [JD - 2,450,000]& \nodata\tablenotemark{c} & \nodata\tablenotemark{c}\\ $\varpi$ [$\arcsec$] & $27.31 \pm 0.12$ & $27.31 \pm 0.12$\\ $\sigma_\rho$ [$\arcsec$] & $0.009 \pm 0.004$ & $0.009 \pm 0.004$\\ $\sigma_\theta$ [$\degr$] & $0.024 \pm 0.008$ & $0.024 \pm 0.008$\\ $\sigma_\mathrm{CfA}$ [km s${}^{-1}$] & $3.8 \pm 0.3$ & $3.8 \pm 0.3$\\ $\sigma_\mathrm{Keck}$ [km s${}^{-1}$] & $0.8 \pm 0.1$ & $0.8 \pm 0.1$\\ $\sigma_\mathrm{FEROS}$ [km s${}^{-1}$] & $3.4 \pm 0.6$ & $3.4 \pm 0.6$\\ $\sigma_\mathrm{du\;Pont}$ [km s${}^{-1}$] & $2.1 \pm 0.4$ & $2.1 \pm 0.4$\\ (Keck - CfA) [km s${}^{-1}$] & $-1.3 \pm 0.5$ & $-1.3 \pm 0.5$\\ (FEROS - CfA) [km s${}^{-1}$] & $1.3 \pm 1.0$ & $1.3 \pm 1.0$\\ (du Pont - CfA) [km s${}^{-1}$] & $-0.2 \pm 0.7$ & $-0.2 \pm 0.7$\\ \cutinhead{Derived} $M_\mathrm{Aa}$ [$M_\odot$] & $0.29 \pm 0.01$ & $0.29 \pm 0.01$\\ $M_\mathrm{A}$ [$M_\odot$] & $0.53 \pm 0.01$ & $0.53 \pm 0.01$\\ $a_\mathrm{inner}$ [au] & $0.170 \pm 0.001$ & $0.170 \pm 0.001$\\ $a_\mathrm{outer}$ [au] & $63 \pm 18$ & $64 \pm 19$\\ $r_{p,\mathrm{outer}}$ [au] & $45 \pm 21$ & $45 \pm 21$\\ \enddata \tablenotetext{a}{The argument of periastron of the primary. $\omega_\mathrm{secondary} = \omega_\mathrm{primary} + \pi$.} \tablenotetext{b}{The ascending node is identified as the point where the secondary body crosses the sky plane \emph{receding} from the observer.} \tablenotetext{c}{Posterior is non-Gaussian; see Figure~\ref{fig:corner}} \end{deluxetable} --- --- @article{chan2021salkg, abbr={NeurIPS}, title={SalKG: Learning From Knowledge Graph Explanations for Commonsense Reasoning}, author={ and and and and and }, journal={Advances in Neural Information Processing Systems (NeurIPS),}, pdf={https://arxiv.org/pdf/2104.08793.pdf}, arxiv={2104.08793}, code={https://github.com/INK-USC/SalKG}, year={2021}, selected={true}, teaser={salkg.png} } @article{ma2021dss, abbr={AUA}, title={Dissection Gesture Sequence during Nerve Sparing Predicts Erectile Function Recovery after Robot-Assisted Radical Prostatectomy}, author={ and , ., ., Andrew}, journal={American Urological Association Annual Conference (AUA),}, html={/files/Dissection_Gesture_Abstract_v6.pdf}, year={2022 (under review)}, selected={true}, } @article{ma2021dart, abbr={AUA}, title={Dissection Assessment for Robotic Technique (DART) to Evaluate Nerve-Spare of Robot-Assisted Radical Prostatectomy}, author={ and and Xu, Jiashu and Desai, , , ., Jim and ., Andrew}, journal={American Urological Association Annual Conference (AUA),}, html={/files/DART_v5.pdf}, year={2022 (under review)}, selected={true}, } \hypertarget{structwb_1_1_survivor_data}{}\section{wb\+:\+:Survivor\+Data Struct Reference} \label{structwb_1_1_survivor_data}\index{wb\+::\+Survivor\+Data@{wb\+::\+Survivor\+Data}} Info about a still running detection. {\ttfamily \#include $<$wb\+\_\+structures.\+h$>$} \subsection*{Public Attributes} \begin{DoxyCompactItemize} \item \hypertarget{structwb_1_1_survivor_data_a332b2124ef31030c0926f39af55a9813}{}uint32 {\bfseries x}\label{structwb_1_1_survivor_data_a332b2124ef31030c0926f39af55a9813} \item \hypertarget{structwb_1_1_survivor_data_aae20a4b74ebf68b84d3fec67dc8870e8}{}uint32 {\bfseries y}\label{structwb_1_1_survivor_data_aae20a4b74ebf68b84d3fec67dc8870e8} \item \hypertarget{structwb_1_1_survivor_data_a95128ef84007dd7e5276afc4435b86bb}{}float {\bfseries response}\label{structwb_1_1_survivor_data_a95128ef84007dd7e5276afc4435b86bb} \end{DoxyCompactItemize} \subsection{Detailed Description} Info about a still running detection. Threads calculating detections get discarded, but those still running keep their data in this structure, so we can continue with computation, when we rerun detection. The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item C\+:/dev/repositories/vutbr-\/fit-\/waldboost-\/detector/src/wb\+\_\+structures.\+h\end{DoxyCompactItemize} @inproceedings{DBLP:conf/birthday/SchlatteJMTY18, author = { and and and and }, bibsource = {dblp computer science bibliography, https://dblp.org}, biburl = {https://dblp.org/rec/bib/conf/birthday/SchlatteJMTY18}, booktitle = {It's All About Coordination - Essays to Celebrate the Lifelong Scientific Achievements of Farhad Arbab}, crossref = {DBLP:conf/birthday/2018arbab}, doi = {10.1007/978-3-319-90089-6_8}, pages = {107--121}, timestamp = {Sat, 11 Aug 2018 00:57:41 +0200}, title = {Release the Beasts: When Formal Methods Meet Real World Data}, url = {https://doi.org/10.1007/978-3-319-90089-6_8}, year = {2018} } \hypertarget{forestTypes_2rerf_2unprocessedRerFNode_8h}{}\section{src/forest\+Types/rerf/unprocessed\+Rer\+F\+Node.h File Reference} \label{forestTypes_2rerf_2unprocessedRerFNode_8h}\index{src/forest\+Types/rerf/unprocessed\+Rer\+F\+Node.\+h@{src/forest\+Types/rerf/unprocessed\+Rer\+F\+Node.\+h}} {\ttfamily \#include \char`\"{}fp\+Rer\+F\+Split.\+h\char`\"{}}\newline {\ttfamily \#include $<$vector$>$}\newline {\ttfamily \#include $<$random$>$}\newline Include dependency graph for unprocessed\+Rer\+F\+Node.\+h\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{forestTypes_2rerf_2unprocessedRerFNode_8h__incl} \end{center} \end{figure} This graph shows which files directly or indirectly include this file\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=216pt]{forestTypes_2rerf_2unprocessedRerFNode_8h__dep__incl} \end{center} \end{figure} \subsection*{Classes} \begin{DoxyCompactItemize} \item class \hyperlink{classfp_1_1unprocessedRerFNode}{fp\+::unprocessed\+Rer\+F\+Node$<$ T $>$} \end{DoxyCompactItemize} \subsection*{Namespaces} \begin{DoxyCompactItemize} \item \hyperlink{namespacefp}{fp} \end{DoxyCompactItemize} @inproceedings{Budgen2008, abstract = {Background: A mapping study provides a systematic and objective procedure for identifying the nature and extent of the empirical study data that is available to answer a particular research question. Such studies can also form a useful preliminary step for PhD study. Aim: We set out to assess how effective such studies have been when used for software engineering topics, and to identify the specific challenges that they present. Method: We have conducted an informal review of a number of mapping studies in software engineering, describing their main characteristics and the forms of analysis employed. Results: We examine the experiences and outcomes from six mapping studies, of which four are published. From these we note a recurring theme about the problems of classification and a preponderance of ‘gaps' in the set of empirical studies. Conclusions: We identify our challenges as improving classification guidelines, encouraging better reporting of primary studies, and argue for identifying some 'empirical grand challenges' for software engineering as a focus for the community.}, author = {, }, booktitle = {Proceedings of PPIG}, isbn = {9783642021527}, pages = {195--204}, title = {{Using Mapping Studies in Software Engineering}}, volume = {2}, year = {2008} } @incollection{Ciolkowski2003, abstract = {A survey is an empirical research strategy for the collection of information from heterogeneous sources. In this way, survey results often exhibit a high degree of external validity. It is complementary to other empirical research strategies such as controlled experiments, which usually have their strengths in the high internal validity of the findings. While there is a growing number of (quasi-)controlled experiments reported in the software engineering literature, few results of large scale surveys have been reported there. Hence, there is still a lack of knowledge on how to use surveys in a systematic manner for software engineering empirical research. This chapter introduces a process for preparing, conducting, and analyzing a software engineering survey. The focus of the work is on questionnaire-based surveys rather than literature surveys. The survey process is driven by practical experiences from two large-scale efforts in the review and inspection area. There are two main results from this work. First, the process itself allows researchers in empirical software engineering to follow a systematic, disciplined approach. Second, the experiences from applying the process help avoid common pitfalls that endanger both the research process and its results. We report on two (descriptive) surveys on software reviews that applied the survey process, and we present our experiences, as well as models for survey effort and duration factors derived from these experiences. {\textcopyright} Springer-Verlag Berlin Heidelberg 2003.}, author = {Ciolkowski, Marcus and Laitenberger, Oliver and and }, booktitle = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, doi = {10.1007/978-3-540-45143-3_7}, isbn = {978-3-540-45143-3}, pages = {104--128}, publisher = {Springer Berlin Heidelberg}, title = {{Practical Experiences in the Design and Conduct of Surveys in Empirical Software Engineering}}, volume = {2765}, year = {2003} } @book{Creswell2018, address = {Los Angeles, CA, USA}, author = {Creswell, and Creswell, }, edition = {5th}, isbn = {9781506386706}, publisher = {SAGE Publications, Inc}, title = {{Research Design: Qualitative, Quantitative, and Mixed Methods Approaches}}, year = {2018} } @incollection{Easterbrook2008, abstract = {Selecting a research method for empirical software engineering research is problematic because the benefits and challenges to using each method are not yet well catalogued. Therefore, this chapter describes a number of empirical methods available. It examines the goals of each and analyzes the types of questions each best addresses. Theoretical stances behind the methods, practical considerations in the application of the methods and data collection are also briefly reviewed. Taken together, this information provides a suitable basis for both understanding and selecting from the variety of methods applicable to empirical software engineering.}, address = {London}, author = { and }, booktitle = {Guide to Advanced Empirical Software Engineering}, doi = {10.1007/978-1-84800-044-5_11}, keywords = {empirical,survey,theory}, pages = {285--311}, publisher = {Springer London}, title = {{Selecting Empirical Methods for Software Engineering Research}}, year = {2008} } @article{Eisenhardt1989, author = {Eisenhardt, .}, doi = {10.5465/amr.1989.4308385}, issn = {0363-7425}, journal = {Academy of Management Review}, month = {oct}, number = {4}, pages = {532--550}, title = {{Building Theories from Case Study Research}}, volume = {14}, year = {1989} } @article{Falessi2018, abstract = {{\textcopyright} 2017 Springer Science+Business Media New York[Context] Controlled experiments are an important empirical method to generate and validate theories. Many software engineering experiments are conducted with students. It is often claimed that the use of students as participants in experiments comes at the cost of low external validity while using professionals does not. [Objective] We believe a deeper understanding is needed on the external validity of software engineering experiments conducted with students or with professionals. We aim to gain insight about the pros and cons of using students and professionals in experiments. [Method] We performed an unconventional, focus group approach and a follow-up survey. First, during a session at ISERN 2014, 65 empirical researchers, including the seven authors, argued and discussed the use of students in experiments with an open mind. Afterwards, we revisited the topic and elicited experts' opinions to foster discussions. Then we derived 14 statements and asked the ISERN attendees excluding the authors, to provide their level of agreement with the statements. Finally, we analyzed the researchers' opinions and used the findings to further discuss the statements. [Results] Our survey results showed that, in general, the respondents disagreed with us about the drawbacks of professionals. We, on the contrary, strongly believe that no population (students, professionals, or others) can be deemed better than another in absolute terms. [Conclusion] Using students as participants remains a valid simplification of reality needed in laboratory contexts. It is an effective way to advance software engineering theories and technologies but, like any other aspect of study settings, should be carefully considered during the design, execution, interpretation, and reporting of an experiment. The key is to understand which developer population portion is being represented by the participants in an experiment. Thus, a proposal for describing experimental participants is put forward.}, author = { M{\"{{\"{u}}, }, doi = {10.1007/s10664-017-9523-3}, issn = {1382-3256}, journal = {Empirical Software Engineering}, keywords = {Experimentation,Generalization,Participants in experiments,Subjects of experiments,Threats to validity}, month = {feb}, number = {1}, pages = {452--489}, publisher = {Empirical Software Engineering}, title = {{Empirical software engineering experts on the use of students and professionals in experiments}}, volume = {23}, year = {2018} } @book{Feyerabend1993, address = {London, UK}, author = {}, edition = {3rd}, publisher = {Verso}, title = {{Against Method: Outline of an Anarchistic Theory of Knowledge}}, year = {1993} } @article{Garousi2019, abstract = {Context: A Multivocal Literature Review (MLR) is a form of a Systematic Literature Review (SLR) which includes the grey literature (e.g., blog posts, videos and white papers) in addition to the published (formal) literature (e.g., journal and conference papers). MLRs are useful for both researchers and practitioners since they provide summaries both the state-of-the art and –practice in a given area. MLRs are popular in other fields and have recently started to appear in software engineering (SE). As more MLR studies are conducted and reported, it is important to have a set of guidelines to ensure high quality of MLR processes and their results. Objective: There are several guidelines to conduct SLR studies in SE. However, several phases of MLRs differ from those of traditional SLRs, for instance with respect to the search process and source quality assessment. Therefore, SLR guidelines are only partially useful for conducting MLR studies. Our goal in this paper is to present guidelines on how to conduct MLR studies in SE. Method: To develop the MLR guidelines, we benefit from several inputs: (1) existing SLR guidelines in SE, (2), a literature survey of MLR guidelines and experience papers in other fields, and (3) our own experiences in conducting several MLRs in SE. We took the popular SLR guidelines of Kitchenham and Charters as the baseline and extended/adopted them to conduct MLR studies in SE. All derived guidelines are discussed in the context of an already-published MLR in SE as the running example. Results: The resulting guidelines cover all phases of conducting and reporting MLRs in SE from the planning phase, over conducting the review to the final reporting of the review. In particular, we believe that incorporating and adopting a vast set of experience-based recommendations from MLR guidelines and experience papers in other fields have enabled us to propose a set of guidelines with solid foundations. Conclusion: Having been developed on the basis of several types of experience and evidence, the provided MLR guidelines will support researchers to effectively and efficiently conduct new MLRs in any area of SE. The authors recommend the researchers to utilize these guidelines in their MLR studies and then share their lessons learned and experiences.}, author = {{\"{a}}ntyl{\"{a}}, .}, doi = {10.1016/j.infsof.2018.09.006}, issn = {09505849}, journal = {Information and Software Technology}, keywords = {Evidence-based software engineering,Grey literature,Guidelines,Literature study,Multivocal literature review,Systematic literature review,Systematic mapping study}, month = {feb}, number = {September 2018}, pages = {101--121}, publisher = {Elsevier B.V.}, title = {{Guidelines for including grey literature and conducting multivocal literature reviews in software engineering}}, volume = {106}, year = {2019} } @inproceedings{Garousi2016, abstract = {Systematic Literature Reviews (SLR) may not provide insight into the "state of the practice" in SE, as they do not typically include the "grey" (non-published) literature. A Multivocal Literature Review (MLR) is a form of a SLR which includes grey literature in addition to the published (formal) literature. Only a few MLRs have been published in SE so far. We aim at raising the awareness for MLRs in SE by addressing two research questions (RQs): (1) What types of knowledge are missed when a SLR does not include the multivocal literature in a SE field? and (2) What do we, as a community, gain when we include the multivocal literature and conduct MLRs? To answer these RQs, we sample a few example SLRs and MLRs and identify the missing and the gained knowledge due to excluding or including the grey literature. We find that (1) grey literature can give substantial benefits in certain areas of SE, and that (2) the inclusion of grey literature brings forward certain challenges as evidence in them is often experience and opinion based. Given these conflicting viewpoints, the authors are planning to prepare systematic guidelines for performing MLRs in SE.}, address = {New York, New York, USA}, author = { {\"{a}}ntyl{\"{a}}, .}, booktitle = {Proceedings of the 20th International Conference on Evaluation and Assessment in Software Engineering - EASE '16}, doi = {10.1145/2915970.2916008}, isbn = {9781450336918}, keywords = {Empirical software engineering,Grey literature,MLR,Multivocal Literature Reviews,Research methodology,SLR,Systematic literature reviews}, pages = {1--6}, publisher = {ACM Press}, title = {{The need for multivocal literature reviews in software engineering}}, volume = {01-03-June}, year = {2016} } @book{Godfrey-Smith2003, address = {Chicago, IL, USA}, author = {}, doi = {10.7208/chicago/9780226300610.001.0001}, isbn = {978-0-226-30063-4}, publisher = {University of Chicago Press}, title = {{Theory and Reality: An Introduction to the Philosophy of Science}}, year = {2003} } @article{Gregor2006, abstract = {The aim of this research essay is to examine the structural nature of theory in Information Systems. Despite the importance of theory, questions relating to its form and structure are neglected in comparison with questions relating to epistemology. The essay addresses issues of causality, explanation, prediction, and generalization that underlie an understanding of theory. A taxonomy is proposed that classifies information systems theories with respect to the manner in which four central goals are addressed: analysis, explanation, prediction, and prescription. Five interrelated types of theory are distinguished: (1) theory for analyzing, (2) theory for explaining, (3) theory for predicting, (4) theory for explaining and predicting, and (5) theory for design and action. Examples illustrate the nature of each theory type. The applicability of the taxonomy is demonstrated by classifying a sample of journal articles. The paper contributes by showing that multiple views of theory exist and by exposing the assumptions underlying different viewpoints. In addition, it is suggested that the type of theory under development can influence the choice of an epistemological approach. Support is given for the legitimacy and value of each theory type. The building of integrated bodies of theory that encompass all theory types is advocated.}, author = {}, doi = {10.2307/25148742}, issn = {02767783}, journal = {MIS Quarterly}, keywords = {Causality,Design science,Design theory,Explanation,Generalization,Information systems discipline,Interpretivist theory,Philosophy of science,Philosophy of social sciences,Prediction,Theory,Theory structure,Theory taxonomy}, number = {3}, pages = {611}, title = {{The Nature of Theory in Information Systems}}, volume = {30}, year = {2006} } @article{Hevner2007, abstract = {As a commentary to Juhani Iivari's insightful essay, I briefly analyze design science research as an embodiment of three closely related cycles of activities. The Relevance Cycle inputs requirements from the contextual envi- ronment into the research and introduces the research artifacts into environ- mental field testing. The Rigor Cycle provides grounding theories and methods along with domain experience and expertise from the foundations knowledge base into the research and adds the new knowledge generated by the research to the growing knowledge base. The central Design Cycle sup- ports a tighter loop of research activity for the construction and evaluation of design artifacts and processes. The recognition of these three cycles in a research project clearly positions and differentiates design science from other research paradigms. The commentary concludes with a claim to the pragmatic nature}, author = {Hevner, }, isbn = {0905-0167}, issn = {09050167}, journal = {Scandinavian Journal of Information Systems}, keywords = {design cycle,design science,relevance cycle,rigor cycle}, number = {2}, pages = {87--92}, title = {{A Three Cycle View of Design Science Research}}, volume = {19}, year = {2007} } @article{Hevner2004, abstract = {Two paradigms characterize much of the research in the Information Systems discipline: behavioral science and design science. The behavioral science paradigm seeks to develop and verify theories that explain or predict human or organizational behavior. The design-science paradigm seeks to extend the boundaries of human and organizational capabilities by creating new and innovative artifacts. Both paradigms are foundational to the IS discipline, positioned as it is at the confluence of people, organizations, and technology. Our objective is to describe the performance of design-science research in Information Systems via a concise conceptual framework and clear guidelines for understanding, executing, and evaluating the research. In the design-science paradigm, knowledge and understanding of a problem domain and its solution are achieved in the building and application of the designed artifact. Three recent exemplars in the research literature are used to demonstrate the application of these guidelines. We conclude with an analysis of the challenges of performing high-quality design-science research in the context of the broader IS community.}, archivePrefix = {arXiv}, arxivId = {http://dl.acm.org/citation.cfm?id=2017212.2017217}, author = { Ram}, doi = {10.2307/25148625}, eprint = {/dl.acm.org/citation.cfm?id=2017212.2017217}, issn = {02767783}, journal = {MIS Quarterly}, keywords = {Information Systems research methodologies,business environment,creativity,design artifact,design science,experimental methods,search strategies,technology infrastructure}, number = {1}, pages = {75}, primaryClass = {http:}, title = {{Design Science in Information Systems Research}}, volume = {28}, year = {2004} } @article{Host2000, abstract = {In many studies in software engineering students are used instead of professional software developers, although the objective is to draw conclusions valid for professional software developers. This paper presents a study where the difference between the two groups is evaluated. People from the two groups have individually carried out a non-trivial software engineering judgement task involving the assessment of howten different factors affect the lead-time of software development projects. It is found that the differences are only minor, and it is concluded that software engineering students may be used instead of professional software developers under certain conditions. These conditions are identified and described based on generally accepted criteria for validity evaluation of empirical studies.}, author = { }, doi = {10.1023/A:1026586415054}, issn = {1382-3256}, journal = {Empirical Software Engineering}, number = {3}, pages = {201--214}, title = {{Using students as subjects - a comparative study of students and professionals in lead-time impact assessment}}, volume = {5}, year = {2000} } @inproceedings{Hove2005, abstract = {Many phenomena related to software development are qualitative in nature. Relevant measures of such phenomena are often collected using semi-structured interviews. Such interviews involve high costs, and the quality of the collected data is related to how the interviews are conducted. Careful planning and conducting of the interviews are therefore necessary, and experiences from interview studies in software engineering should consequently be collected and analyzed to provide advice to other researchers. We have brought together experiences from 12 software engineering studies, in which a total of 280 interviews were conducted. Four areas were particularly challenging when planning and conducting these interviews; estimating the necessary effort, ensuring that the interviewer had the needed skills, ensuring good interaction between interviewer and interviewees, and using the appropriate tools and project artifacts. The paper gives advice on how to handle these areas and suggests what information about the interviews should be included when reporting studies where interviews have been used in data collection. Knowledge from other disciplines is included. By sharing experience, knowledge about the accomplishments of software engineering interviews is increased and hence, measures of high quality can be achieved}, author = {. and }, booktitle = {11th IEEE International Software Metrics Symposium (METRICS'05)}, doi = {10.1109/METRICS.2005.24}, isbn = {0-7695-2371-4}, issn = {15301435}, number = {Metrics}, pages = {23--23}, publisher = {IEEE}, title = {{Experiences from Conducting Semi-structured Interviews in Empirical Software Engineering Research}}, year = {2005} } @techreport{Jarvinen2016, abstract = {Literature reviews play an important role in a researcher's preparation of research problem. In order to find out a gap between what we already know and what we like to know, a researcher can utilize not only concepts (Webster and Watson 2002) but also classifications in performing literature review. It is assumed that a higher order framework can be unpacked into classifications. A classification consists of classes of a dimension, and classes can be concepts (variables). It will be shown that it is possible to derive some guidelines how to improve classifications for identifying research gaps in literature review.}, address = {Tampere, Finland}, author = {J{\"{}, institution = {University of Tampere, School of Information Sciences}, isbn = {978-952-03-0333-4}, issn = {1799-8158}, title = {{On lenses and their improvements for identifying research gaps in literature review}}, year = {2016} } @incollection{Jedlitschka2008, address = {London}, author = {Jedlitschka, Andreas and Ciolkowski, Marcus and }, booktitle = {Guide to Advanced Empirical Software Engineering}, doi = {10.1007/978-1-84800-044-5_8}, pages = {201--228}, publisher = {Springer London}, title = {{Reporting Experiments in Software Engineering}}, year = {2008} } @book{Johannesson2014, abstract = {This book is an introductory text on design science, intended to support both graduate students and researchers in structuring, undertaking and presenting design science work. It builds on established design science methods as well as recent work on presenting design science studies and ethical principles for design science, and also offers novel instruments for visualizing the results, both in the form of process diagrams and through a canvas format. This work focuses on design science as applied to information systems and technology, but it also includes examples from, and perspectives of, other fields of human practice. --}, address = {Cham}, author = { Perjons, Erik}, booktitle = {Springer International Publishing Switzerland}, doi = {10.1007/978-3-319-10632-8}, isbn = {978-3-319-10631-1}, pages = {197}, publisher = {Springer International Publishing}, title = {{An Introduction to Design Science}}, year = {2014} } @article{Johnson2012, author = {, Mathias and }, doi = {10.1109/MS.2012.127}, issn = {0740-7459}, journal = {IEEE Software}, keywords = {engineering,explanation,prediction,science,software engineering theory,theory}, month = {sep}, number = {5}, pages = {96--96}, publisher = {IEEE}, title = {{Where's the Theory for Software Engineering?}}, volume = {29}, year = {2012} } @techreport{Kasunic2005, abstract = {A survey can characterize the knowledge, attitudes, and behaviors of a large group of people through the study of a subset of them. However, to protect the validity of conclusions drawn from a survey, certain procedures must be followed throughout the process of designing, developing, and distributing the survey questionnaire. Surveys are used extensively by software and systems engineering organizations to provide insight into complex issues, assist with problem solving, and support effective decision making. This document presents a seven-stage, end-to-end process for conducting a survey.}, address = {Pittsburgh, PA}, author = {}, institution = {Carnegie Mellon University, Software Engineering Institute}, isbn = {0780348907}, pages = {143}, title = {{Designing an Effective Survey}}, year = {2005} } @inproceedings{Kitchenham2004a, abstract = {Our objective is to describe how software engineering might benefit from an evidence-based approach and to identify the potential difficulties associated with the approach. We compared the organisation and technical infrastructure supporting evidence-based medicine (EBM) with the situation in software engineering. We considered the impact that factors peculiar to software engineering (i.e. the skill factor and the lifecycle factor) would have on our ability to practice evidence-based software engineering (EBSE). EBSE promises a number of benefits by encouraging integration of research results with a view to supporting the needs of many different stakeholder groups. However, we do not currently have the infrastructure needed for widespread adoption of EBSE. The skill factor means software engineering experiments are vulnerable to subject and experimenter bias. The lifecycle factor means it is difficult to determine how technologies will behave once deployed. Software engineering would benefit from adopting what it can of the evidence approach provided that it deals with the specific problems that arise from the nature of software engineering.}, author = {. and .}, booktitle = {Proceedings. 26th International Conference on Software Engineering}, doi = {10.1109/ICSE.2004.1317449}, isbn = {0-7695-2163-0}, issn = {0270-5257}, pages = {273--281}, publisher = {IEEE Comput. Soc}, title = {{Evidence-based software engineering}}, year = {2004} } @article{Kitchenham2004c, abstract = {The objective of this report is to propose a guideline for systematic reviews appropriate for software engineering researchers, including PhD students. A systematic review is a means of evaluating and interpreting all available research relevant to a particular research question, topic area, or phenomenon of interest. Systematic reviews aim to present a fair evaluation of a research topic by using a trustworthy, rigorous, and auditable methodology. The guideline presented in this report was derived from three existing guidelines used by medical researchers. The guideline has been adapted to reflect the specific problems of software engineering research. The guideline covers three phases of a systematic review: planning the review, conducting the review and reporting the review. It is at a relatively high level. It does not consider the impact of question type on the review procedures, nor does it specify in detail mechanisms needed to undertake meta-analysis.}, address = {Keele, UK}, author = {}, doi = {10.1.1.122.3308}, institution = {Software Engineering Group, Department of Computer Sciene, Keele University}, isbn = {1353-7776}, issn = {13537776}, journal = {Keele, UK, Keele University}, number = {TR/SE-0401}, pages = {28}, pmid = {15046037}, title = {{Procedures for performing systematic reviews}}, volume = {33}, year = {2004} } @incollection{Kitchenham2008, address = {London}, author = {Kitchenham, . and .}, booktitle = {Guide to Advanced Empirical Software Engineering}, doi = {10.1007/978-1-84800-044-5_3}, pages = {63--92}, publisher = {Springer London}, title = {{Personal Opinion Surveys}}, year = {2008} } @techreport{Kitchenham2007, abstract = {The objective of this report is to propose comprehensive guidelines for systematic literature reviews appropriate for software engineering researchers, including PhD students. A systematic literature review is a means of evaluating and interpreting all available research relevant to a particular research question, topic area, or phenomenon of interest. Systematic reviews aim to present a fair evaluation of a research topic by using a trustworthy, rigorous, and auditable methodology. The guidelines presented in this report were derived from three existing guidelines used by medical researchers, two books produced by researchers with social science backgrounds and discussions with researchers from other disciplines who are involved in evidence-based practice. The guidelines have been adapted to reflect the specific problems of software engineering research. The guidelines cover three phases of a systematic literature review: planning the review, conducting the review and reporting the review. They provide a relatively high level description. They do not consider the impact of the research questions on the review procedures, nor do they specify in detail the mechanisms needed to perform meta-analysis.}, address = {Keele, UK}, author = { and }, booktitle = {Technical Report EBSE-2007-01}, institution = {School of Computer Science and Mathematics, Keele University}, pages = {65}, title = {{Guidelines for performing Systematic Literature reviews in Software Engineering}}, year = {2007} } @article{Kitchenham2009, abstract = {Background: In 2004 the concept of evidence-based software engineering (EBSE) was introduced at the ICSE04 conference. Aims: This study assesses the impact of systematic literature reviews (SLRs) which are the recommended EBSE method for aggregating evidence. Method: We used the standard systematic literature review method employing a manual search of 10 journals and 4 conference proceedings. Results: Of 20 relevant studies, eight addressed research trends rather than technique evaluation. Seven SLRs addressed cost estimation. The quality of SLRs was fair with only three scoring less than 2 out of 4. Conclusions: Currently, the topic areas covered by SLRs are limited. European researchers, particularly those at the Simula Laboratory appear to be the leading exponents of systematic literature reviews. The series of cost estimation SLRs demonstrate the potential value of EBSE for synthesising evidence and making it available to practitioners. ?? 2008 Elsevier B.V. All rights reserved.}, author = { and {}, , , , }, doi = {10.1016/j.infsof.2008.09.009}, isbn = {0950-5849}, issn = {09505849}, journal = {Information and Software Technology}, keywords = {Cost estimation,Evidence-based software engineering,Systematic literature review,Systematic review quality,Tertiary study}, month = {jan}, number = {1}, pages = {7--15}, title = {{Systematic literature reviews in software engineering – A systematic literature review}}, volume = {51}, year = {2009} } @book{Kuhn1970, address = {Chicago, IL, USA}, author = {.}, edition = {2nd}, isbn = {0-226-45803-2}, publisher = {University of Chicago Press}, title = {{The Structure of Scientific Revolutions}}, year = {1970} } @article{Langley1999, abstract = {In this article I describe and compare a number of alternative generic strategies for the analysis oi process data, looking at the consequences oi these strategies ior emerging theories. I evaluate the strengths and weaknesses of the strategies in terms oi their capacity to generate theory that is accurate, porsimonious. general, and useful and suggest that method and theory are inextricably intertwined, that multiple strategies are oiten advisable, and that no analysis strategy will produce theory without an uncodiiiable creative leap, however small. Finally, I argue that there is room in the organizational research literature ior more openness within the academic community toward a variety of iorms oi coupling between theory and data.}, author = {}, doi = {10.2307/259349}, issn = {03637425}, journal = {The Academy of Management Review}, month = {oct}, number = {4}, pages = {691}, title = {{Strategies for Theorizing from Process Data}}, volume = {24}, year = {1999} } @techreport{Mayring2014, address = {Klagenfurt}, author = {}, keywords = {content analysis,empirical social research,qualitative method,quantitative method,research approach,text analysis}, title = {{Qualitative content analysis: theoretical foundation, basic procedures and software solution}}, year = {2014} } @inproceedings{Molleri2016, abstract = {Background: Survey is a method of research aiming to gather data from a large population of interest. Despite being extensively used in software engineering, survey-based research faces several challenges, such as selecting a representative population sample and designing the data collection instruments. Objective: This article aims to summarize the existing guidelines, supporting instruments and recommendations on how to conduct and evaluate survey-based research. Methods: A systematic search using manual search and snowballing techniques were used to identify primary studies supporting survey research in software engineering. We used an annotated review to present the findings, describing the references of interest in the research topic. Results: The summary provides a description of 15 available articles addressing the survey methodology, based upon which we derived a set of recommendations on how to conduct survey research, and their impact in the community. Conclusion: Survey-based research in software engineering has its particular challenges, as illustrated by several articles in this review. The annotated review can contribute by raising awareness of such challenges and present the proper recommendations to overcome them.}, address = {New York, New York, USA}, author = {Molleri, }, booktitle = {Proceedings of the 10th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement - ESEM '16}, doi = {10.1145/2961111.2962619}, isbn = {9781450344272}, issn = {19493789}, pages = {1--6}, publisher = {ACM Press}, title = {{Survey Guidelines in Software Engineering}}, volume = {08-09-Sept}, year = {2016} } @article{Molleri2020a, abstract = {Context: Over the past decade Software Engineering research has seen a steady increase in survey-based studies, and there are several guidelines providing support for those willing to carry out surveys. The need for auditing survey research has been raised in the literature. Checklists have been used both to conduct and to assess different types of empirical studies, such as experiments and case studies. Objective: To operationalize the assessment of survey studies by means of a checklist. To fulfill such goal, we aim to derive a checklist from standards for survey research and further evaluate the appropriateness of the checklist in the context of software engineering research. Method: We systematically aggregated knowledge from 12 methodological studies supporting survey-based research in software engineering. We identified the key stages of the survey process and its recommended practices through thematic analysis and vote counting. We evaluated the checklist by applying it to existing surveys and analyzed the results. Thereafter, we gathered the feedback of experts (the surveys' authors) on our analysis and used the feedback to improve the survey checklist. Results: The evaluation provided insights regarding limitations of the checklist in relation to its understanding and objectivity. In particular, 19 of the 38 checklist items were improved according to the feedback received from experts. Conclusion: The proposed checklist is appropriate for auditing survey reports as well as a support tool to guide ongoing research with regard to the survey design process. A discussion on how to use the checklist and what its implications are for research practice is also provided.}, author = {Moll{\'{e}}ri, }, doi = {10.1016/j.infsof.2019.106240}, issn = {09505849}, journal = {Information and Software Technology}, keywords = {Assessment,Checklist,Methodology,Survey}, month = {mar}, pages = {106240}, title = {{An empirically evaluated checklist for surveys in software engineering}}, volume = {119}, year = {2020} } @inproceedings{Neto2019, abstract = {Background: In recent years, studies involving Grey Literature (GL) have been growing and attracting the attention of researchers in software engineering (SE). One of the sources of GL refers to content produced by professionals based on their practical experiences? Recent researches in the SE states that GL can complement areas of research that are not yet clearly defined in the scientific literature. In this context, the Multivocal Literature Review (MLR), a form of Systematic Literature Review (SLR) with the inclusion of GL, emerges. Goal: Provide preliminary work about the current research involving MLR studies? First, we investigate the motivation of the researchers to include GL in review studies; and second, we examine how GL was included in the studies. Method: A tertiary study was conducted to search MLR studies published between 2009 to April of 2019. Results: The main motivations for including GL in review studies are: lack of academic research on the topic, emerging research on this topic, and complementary evidence in the GL? Internet articles and white papers were the main sources of GL data used. Conclusions: The conducting of MLR studies is still in its early stages; we have identified only 12 secondary studies. The MLR studies were conducted using guidelines for performing SLRs. What we consider to be a threat to the validity of these studies, since guidelines to conduct SLR studies do not provide recommendations for quality analysis and synthesis of primary studies, including GL.}, author = {. and Santos, . and Endo, and Fagundes, .}, booktitle = {2019 ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM)}, doi = {10.1109/ESEM.2019.8870142}, isbn = {978-1-7281-2968-6}, month = {sep}, pages = {1--6}, publisher = {IEEE}, title = {{Multivocal literature reviews in software engineering: Preliminary findings from a tertiary study}}, volume = {2019-Septe}, year = {2019} } @book{Okasha2016, author = {}, doi = {10.1093/actrade/9780198745587.001.0001}, isbn = {9780198745587}, month = {jul}, publisher = {Oxford University Press}, title = {{Philosophy of Science: A Very Short Introduction}}, year = {2016} } @article{Peffers2007, abstract = {JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact .}, author = { Tuunanen, Rothenberger, and Chatterjee, Samir}, doi = {10.2753/MIS0742-1222240302}, issn = {0742-1222}, journal = {Journal of Management Information Systems}, month = {dec}, number = {3}, pages = {45--77}, title = {{A Design Science Research Methodology for Information Systems Research}}, volume = {24}, year = {2007} } @article{Petersen2008, abstract = {BACKGROUND: A software engineering systematic map is a defined method to build a classification scheme and structure a software engineering field of interest. The analysis of results focuses on frequencies of publications for categories within the scheme. Thereby, the coverage of the research field can be determined. Different facets of the scheme can also be combined to answer more specific research questions. OBJECTIVE: We describe how to conduct a systematic mapping study in software engineering and provide guidelines. We also compare systematic maps and systematic reviews to clarify how to chose between them. This comparison leads to a set of guidelines for systematic maps. METHOD: We have defined a systematic mapping process and applied it to complete a systematic mapping study. Furthermore, we compare systematic maps with systematic reviews by systematically analyzing existing systematic reviews. RESULTS: We describe a process for software engineering systematic mapping studies and compare it to systematic reviews. Based on this, guidelines for conducting systematic maps are defined. CONCLUSIONS: Systematic maps and reviews are different in terms of goals, breadth, validity issues and implications. Thus, they should be used complementarily and require different methods (e.g., for analysis).}, author = { Feldt, Mujtaba, Mattsson, Michael}, isbn = {0-7695-2555-5}, issn = {02181940}, journal = {EASE'08 Proceedings of the 12th international conference on Evaluation and Assessment in Software Engineering}, keywords = {evidence based software engineering,systematic mapping studies,systematic reviews}, pages = {68--77}, title = {{Systematic mapping studies in software engineering}}, year = {2008} } @article{Petersen2015, abstract = {Context Systematic mapping studies are used to structure a research area, while systematic reviews are focused on gathering and synthesizing evidence. The most recent guidelines for systematic mapping are from 2008. Since that time, many suggestions have been made of how to improve systematic literature reviews (SLRs). There is a need to evaluate how researchers conduct the process of systematic mapping and identify how the guidelines should be updated based on the lessons learned from the existing systematic maps and SLR guidelines. Objective To identify how the systematic mapping process is conducted (including search, study selection, analysis and presentation of data, etc.); to identify improvement potentials in conducting the systematic mapping process and updating the guidelines accordingly. Method We conducted a systematic mapping study of systematic maps, considering some practices of systematic review guidelines as well (in particular in relation to defining the search and to conduct a quality assessment). Results In a large number of studies multiple guidelines are used and combined, which leads to different ways in conducting mapping studies. The reason for combining guidelines was that they differed in the recommendations given. Conclusion The most frequently followed guidelines are not sufficient alone. Hence, there was a need to provide an update of how to conduct systematic mapping studies. New guidelines have been proposed consolidating existing findings.}, archivePrefix = {arXiv}, arxivId = {arXiv:1011.1669v3}, author = { Vakkalanka, }, doi = {10.1016/j.infsof.2015.03.007}, eprint = {arXiv:1011.1669v3}, isbn = {0360-1315}, issn = {09505849}, journal = {Information and Software Technology}, keywords = {Guidelines,Software engineering,Systematic mapping studies}, month = {aug}, pages = {1--18}, pmid = {25246403}, title = {{Guidelines for conducting systematic mapping studies in software engineering: An update}}, volume = {64}, year = {2015} } @article{Rainer2019, abstract = {Background: Software engineering research has a growing interest in grey literature (GL). Aim: To improve the identification of relevant and rigorous GL. Method: We develop and demonstrate heuristics to find more relevant and rigorous GL. The heuristics generate stratified samples of search and post–search datasets using a formally structured set of search keywords. Conclusion: The heuristics require further evaluation. We are developing a tool to implement the heuristics.}, author = { and }, doi = {10.1016/j.infsof.2018.10.007}, issn = {09505849}, journal = {Information and Software Technology}, keywords = {Grey literature review,Quality criteria,Reasoning,Search engines}, month = {feb}, pages = {231--233}, publisher = {Elsevier B.V.}, title = {{Heuristics for improving the rigour and relevance of grey literature searches for software engineering research}}, volume = {106}, year = {2019} } @inproceedings{Ralph2015a, abstract = {A process theory is an explanation of how an entity changes and develops. While software engineering is fundamentally concerned with how software artifacts change and develop, little research explicitly builds and empirically evaluates software engineering process theories. This lack of theory obstructs scientific consensus by focusing the academic community on methods. Methods inevitably oversimplify and over-rationalize reality, obfuscating crucial phenomena including uncertainty, problem framing and illusory requirements. Better process theories are therefore needed to ground software engineering in empirical reality. However, poor understanding of process theory issues impedes research and publication. This paper therefore attempts to clarify the nature and types of process theories, explore their development and provide specific guidance for their empirically evaluation.}, author = {}, booktitle = {2015 IEEE/ACM 37th IEEE International Conference on Software Engineering}, doi = {10.1109/ICSE.2015.25}, isbn = {978-1-4799-1934-5}, issn = {02705257}, keywords = {Case study,Field study,Process theory,Questionnaire,Research methodology}, month = {may}, pages = {20--31}, publisher = {IEEE}, title = {{Developing and Evaluating Software Engineering Process Theories}}, volume = {1}, year = {2015} } @article{Rehman2016, author = { and }, journal = {International Journal of Educational Investigations}, number = {8}, pages = {51--59}, title = {{An Introduction to Research Paradigms}}, volume = {3}, year = {2016} } @article{Runeson2009, abstract = {Case study is a suitable research methodology for software engineering research since it studies contemporary phenomena in its natural context. However, the understanding of what constitutes a case study varies, and hence the quality of the resulting studies. This paper aims at providing an introduction to case study methodology and guidelines for researchers conducting case studies and readers studying reports of such studies. The content is based on the authors' own experience from conducting and reading case studies. The terminology and guidelines are compiled from different methodology handbooks in other research domains, in particular social science and information systems, and adapted to the needs in software engineering. We present recommended practices for software engineering case studies as well as empirically derived and evaluated checklists for researchers and readers of case study research.}, archivePrefix = {arXiv}, arxivId = {arXiv:gr-qc/9809069v1}, author = { H{\"{o}}st, Martin}, doi = {10.1007/s10664-008-9102-8}, eprint = {9809069v1}, isbn = {1382325615737616}, issn = {1382-3256}, journal = {Empirical Software Engineering}, keywords = {Case study,Checklists,Guidelines,Research methodology}, month = {apr}, number = {2}, pages = {131--164}, pmid = {28843849}, primaryClass = {arXiv:gr-qc}, title = {{Guidelines for conducting and reporting case study research in software engineering}}, volume = {14}, year = {2009} } @inproceedings{Salman2015, abstract = {Background: Most of the experiments in software engineering (SE) employ students as subjects. This raises concerns about the realism of the results acquired through students and adaptability of the results to software industry. Aim: We compare students and professionals to understand how well students represent professionals as experimental subjects in SE research. Method: The comparison was made in the context of two test-driven development experiments conducted with students in an academic setting and with professionals in a software organization. We measured the code quality of several tasks implemented by both subject groups and checked whether students and professionals perform similarly in terms of code quality metrics. Results: Except for minor differences, neither of the subject groups is better than the other. Professionals produce larger, yet less complex, methods when they use their traditional development approach, whereas both subject groups perform similarly when they apply a new approach for the first time. Conclusion: Given a carefully scoped experiment on a development approach that is new to both students and professionals, similar performances are observed. Further investigation is necessary to analyze the effects of subject demographics and level of experience on the results of SE experiments.}, author = { }, booktitle = {2015 IEEE/ACM 37th IEEE International Conference on Software Engineering}, doi = {10.1109/ICSE.2015.82}, isbn = {978-1-4799-1934-5}, issn = {02705257}, keywords = {Code quality,Empirical study,Experimentation,Test-driven development}, month = {may}, pages = {666--676}, publisher = {IEEE}, title = {{Are Students Representatives of Professionals in Software Engineering Experiments?}}, volume = {1}, year = {2015} } @article{Schryen2015, abstract = {The literature review is an established research genre in many academic disciplines, including the IS discipline. Although many scholars agree that systematic literature reviews should be rigorous, few instructional texts for compiling a solid literature review, at least with regard to the IS discipline, exist. In response to this shortage, in this tutorial, I provide practical guidance for both students and researchers in the IS community who want to methodologically conduct qualitative literature reviews. The tutorial differs from other instructional texts in two regards. First, in contrast to most textbooks, I cover not only searching and synthesizing the literature but also the challenging tasks of framing the literature review, interpreting research findings, and proposing research paths. Second, I draw on other texts that provide guidelines for writing literature reviews in the IS discipline but use many examples of published literature reviews. I use an integrated example of a literature review, which guides the reader through the overall process of compiling a literature review. Keywords:}, author = {}, doi = {10.17705/1CAIS.03712}, issn = {15293181}, journal = {Communications of the Association for Information Systems}, keywords = {Literature review,Literature synthesis,Methodology,Research agenda,Research gaps,Tutorial}, pages = {286--325}, title = {{Writing Qualitative IS Literature Reviews—Guidelines for Synthesis, Interpretation, and Guidance of Research}}, volume = {37}, year = {2015} } @inproceedings{Schryen2015a, abstract = {Literature reviews play an important role in the development of knowledge. Yet, we observe a lack of theoretical underpinning of and epistemological insights into how literature reviews can contribute to knowledge creation and have actually contributed in the IS discipline. To address these theoretical and empirical research gaps, we suggest a novel epistemological model of literature reviews. This model allows us to align different contributions of literature reviews with their underlying knowledge conversions - thereby building a bridge between the previously largely unconnected fields of literature reviews and epistemology. We evaluate the appropriateness of the model by conducting an empirical analysis of 173 IS literature reviews which were published in 39 pertinent IS journals between 2000 and 2014. Based on this analysis, we derive an epistemological taxonomy of IS literature reviews, which complements previously suggested typologies.}, address = {Fort Worth, TX, USA}, author = { and Wagner, Gerit and Benlian, Alexander}, booktitle = {Proceedings of the 36th International Conference on Information Systems}, keywords = {Literature review,Research methods/methodology,Theory of knowledge,fort worth 2015,literature review,methodology,on information systems,research methods,theory of knowledge,thirty sixth international conference}, pages = {1--22}, title = {{Theory of Knowledge for Literature Reviews: An Epistemological Model , Taxonomy and Empirical Analysis of IS Literature}}, year = {2015} } @incollection{Seaman2008, abstract = {Essential Guide to Qualitative Methods in Organizational Research is an excellent resource for students and researchers in the areas of organization studies, management research and organizational psychology, bringing together in one volume the range of methods available for undertaking qualitative data collection and analysis. The volume includes 30 chapters, each focusing on a specific technique. The chapters cover traditional research methods, analysis techniques, and interventions as well as the latest developments in the field. Each chapter reviews how the method has been used in organizational research, discusses the advantages and disadvantages of using the method, and presents a case study example of the method in use. A list of further reading is supplied for those requiring additional information about a given method. The comprehensive and accessible nature of this collection will make it an essential and lasting handbook for researchers and students studying organizations.}, address = {London}, author = {Seaman, .}, booktitle = {Guide to Advanced Empirical Software Engineering}, doi = {10.1007/978-1-84800-044-5_2}, isbn = {0761948880}, issn = {07619488}, pages = {35--62}, pmid = {50}, publisher = {Springer London}, title = {{Qualitative Methods}}, year = {2008} } @article{Sharp2016, abstract = {Ethnography is a qualitative research method used to study people and cultures. It is largely adopted in disciplines outside software engineering, including different areas of computer science. Ethnography can provide an in-depth understanding of the socio-technological realities surrounding everyday software development practice, i.e., it can help to uncover not only what practitioners do, but also why they do it. Despite its potential, ethnography has not been widely adopted by empirical software engineering researchers, and receives little attention in the related literature. The main goal of this paper is to explain how empirical software engineering researchers would benefit from adopting ethnography. This is achieved by explicating four roles that ethnography can play in furthering the goals of empirical software engineering: to strengthen investigations into the social and human aspects of software engineering; to inform the design of software engineering tools; to improve method and process development; and to inform research programmes. This article introduces ethnography, explains its origin, context, strengths and weaknesses, and presents a set of dimensions that position ethnography as a useful and usable approach to empirical software engineering research. Throughout the paper, relevant examples of ethnographic studies of software practice are used to illustrate the points being made.}, author = { Dittrich, , .}, doi = {10.1109/TSE.2016.2519887}, issn = {0098-5589}, journal = {IEEE Transactions on Software Engineering}, keywords = {Design tools and techniques,computer-supported collaborative work,human factors in software design,software engineering process}, month = {aug}, number = {8}, pages = {786--804}, publisher = {IEEE}, title = {{The Role of Ethnographic Studies in Empirical Software Engineering}}, volume = {42}, year = {2016} } @book{Shull2008, abstract = {Empirical studies have become an integral element of software engineering research and practice. This unique text/reference includes chapters from some of the top international empirical software engineering researchers and focuses on the practical knowledge necessary for conducting, reporting and using empirical methods in software engineering. Part 1, Research Methods and Techniques, examines the proper use of various strategies for collecting and analysing data, and the uses for which those strategies are most appropriate. Part 2, Practical Foundations, provides a discussion of several important global issues that need to be considered from the very beginning of research planning. Finally, Knowledge Creation offers insight on using a set of disparate studies to provide useful decision support. Topics and features: Offers information across a range of techniques, methods, and qualitative and quantitative issues, providing a toolkit for the reader that is applicable across the diversity of software development contexts Presents reference material with concrete software engineering examples Provides guidance on how to design, conduct, analyse, interpret and report empirical studies, taking into account the common difficulties and challenges encountered in the field Arms researchers with the information necessary to avoid fundamental risks Tackles appropriate techniques for addressing disparate studies ensuring the relevance of empirical software engineering, and showing its practical impact Describes methods that are less often used in the field, providing less conventional but still rigorous and useful ways of collecting data Supplies detailed information on topics (such as surveys) that often contain methodological errors This broad-ranging, practical guide will prove an invaluable and useful reference for practising software engineers and researchers. In addition, it will be suitable for graduate students studying empirical methods in software development. Dr. is a senior scientist at the Fraunhofer Center for Experimental Software Engineering, Maryland, and the director of its Measurement and Knowledge Management Division. In addition, he serves as associate editor in chief of IEEE Software magazine, specializing in empirical studies. Dr. heads the Human Computer Interaction program at the National Research Council, Canada. She has been conducting empirical research in software engineering for the past 12 years. Dr. is currently research director of the software engineering group of the Simula Research Laboratory, Norway, which is ranked No. 3 in the world (out of 1400 institutions) in an evaluation in 2007 in the area of software and systems engineering.}, address = {London}, archivePrefix = {arXiv}, arxivId = {arXiv:1011.1669v3}, author = { }, booktitle = {Guide to Advanced Empirical Software Engineering}, doi = {10.1007/978-1-84800-044-5}, editor = {.}, eprint = {arXiv:1011.1669v3}, isbn = {978-1-84800-043-8}, issn = {1098-6596}, pages = {1--388}, pmid = {6565}, publisher = {Springer London}, title = {{Guide to Advanced Empirical Software Engineering}}, year = {2008} } @inproceedings{Sjoberg2007, abstract = {We present the vision that for all fields of software engineering (SE), empirical research methods should enable the development of scientific knowledge about how useful different SE technologies are for different kinds of actors, performing different kinds of activities, on different kinds of systems. It is part of the vision that such scientific knowledge will guide the development of new SE technology and is a major input to important SE decisions in industry. Major challenges to the pursuit of this vision are: more SE research should be based on the use of empirical methods; the quality, including relevance, of the studies using such methods should be increased; there should be more and better synthesis of empirical evidence; and more theories should be built and tested. Means to meet these challenges include (1) increased competence regarding how to apply and combine alternative empirical methods, (2) tighter links between academia and industry, (3) the development of common research agendas with a focus on empirical methods, and (4) more resources for empirical research.}, author = {. and }, booktitle = {Future of Software Engineering (FOSE '07)}, doi = {10.1109/FOSE.2007.30}, isbn = {0-7695-2829-5}, issn = {00985589}, month = {may}, number = {1325}, pages = {358--378}, publisher = {IEEE}, title = {{The Future of Empirical Methods in Software Engineering Research}}, volume = {SE-13}, year = {2007} } @incollection{Sonnenberg2012, abstract = {The central outcome of design science research (DSR) is prescriptive knowledge in the form of IT artifacts and recommendations. However, prescrip-tive knowledge is considered to have no truth value in itself. Given this assumption, the validity of DSR outcomes can only be assessed by means of descriptive knowledge to be obtained at the conclusion of a DSR process. This is reflected in the build-evaluate pattern of current DSR methodologies. Recog-nizing the emergent nature of IT artifacts this build-evaluate pattern, however, poses unfavorable implications regarding the achievement of rigor within a DSR project. While it is vital in DSR to prove the usefulness of an artifact a ri-gorous DSR process also requires justifying and validating the artifact design it-self even before it has been put into use. This paper proposes three principles for evaluating DSR artifacts which not only address the evaluation of an arti-fact's usefulness but also the evaluation of design decisions made to build an artifact. In particular, it is argued that by following these principles the prescrip-tive knowledge produced in DSR can be considered to have a truth-like value.}, author = { }, booktitle = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, doi = {10.1007/978-3-642-29863-9_28}, keywords = {Design science research,design theory,epistemology,evaluation}, pages = {381--397}, title = {{Evaluations in the Science of the Artificial – Reconsidering the Build-Evaluate Pattern in Design Science Research}}, year = {2012} } @inproceedings{Stol2013, abstract = {There has been a growing interest in the role of theory within Software Engineering (SE) research. For several decades, researchers within the SE research community have argued that, to become a real engineering science, SE needs to develop stronger theoretical foundations. A few authors have proposed guidelines for constructing theories, building on insights from other disciplines. However, so far, much SE research is not guided by explicit theory, nor does it produce explicit theory. In this paper we argue that SE research does, in fact, show traces of theory, which we call theory fragments. We have adapted an analytical framework from the social sciences, named the Validity Network Schema (VNS), that we use to illustrate the role of theorizing in SE research. We illustrate the use of this framework by dissecting three well known research papers, each of which has had significant impact on their respective subdisciplines. We conclude this paper by outlining a number of implications for future SE research, and show how by increasing awareness and training, development of SE theories can be improved. {\textcopyright} 2013 IEEE.}, author = { }, booktitle = {2013 2nd SEMAT Workshop on a General Theory of Software Engineering (GTSE)}, doi = {10.1109/GTSE.2013.6613863}, isbn = {978-1-4673-6273-3}, keywords = {Software engineering research,empirical research,middle-range theory,theory building,theory fragment}, month = {may}, pages = {5--14}, publisher = {IEEE}, title = {{Uncovering theories in software engineering}}, year = {2013} } @article{Venable2016, abstract = {Evaluation is a central and essential activity in conducting rigorous Design Science Research (DSR), yet there is surprisingly little guidance about designing the DSR evaluation activity beyond suggesting possible methods that could be used for evaluation. This paper extends the notable exception of the existing framework of Pries-Heje et al [11] to address this problem. The paper proposes an extended DSR evaluation framework together with a DSR evaluation design method that can guide DSR researchers in choosing an appropriate strategy for evaluation of the design artifacts and design theories that form the output from DSR. The extended DSR evaluation framework asks the DSR researcher to consider (as input to the choice of the DSR evaluation strategy) contextual factors of goals, conditions, and constraints on the DSR evaluation, e.g. the type and level of desired rigor, the type of artifact, the need to support formative development of the designed artifacts, the properties of the artifact to be evaluated, and the constraints on resources available, such as time, labor, facilities, expertise, and access to research subjects. The framework and method support matching these in the first instance to one or more DSR evaluation strategies, including the choice of ex ante (prior to artifact construction) versus ex post evaluation (after artifact construction) and naturalistic (e.g., field setting) versus artificial evaluation (e.g., laboratory setting). Based on the recommended evaluation strategy(ies), guidance is provided concerning what methodologies might be appropriate within the chosen strategy(ies).}, author = {}, doi = {10.1057/ejis.2014.36}, isbn = {978-3-642-29862-2}, issn = {0960-085X}, journal = {European Journal of Information Systems}, keywords = {design science research,evaluation method,evaluation strategy,information,research methodology,systems evaluation}, month = {jan}, number = {1}, pages = {77--89}, title = {{FEDS: a Framework for Evaluation in Design Science Research}}, volume = {25}, year = {2016} } @incollection{Venable2012, abstract = {Evaluation is a central and essential activity in conducting rigorous Design Science Research (DSR), yet there is surprisingly little guidance about designing the DSR evaluation activity beyond suggesting possible methods that could be used for evaluation. This paper extends the notable exception of the existing framework of Pries-Heje et al [11] to address this problem. The paper proposes an extended DSR evaluation framework together with a DSR evaluation design method that can guide DSR researchers in choosing an appropriate strategy for evaluation of the design artifacts and design theories that form the output from DSR. The extended DSR evaluation framework asks the DSR researcher to consider (as input to the choice of the DSR evaluation strategy) contextual factors of goals, conditions, and constraints on the DSR evaluation, e.g. the type and level of desired rigor, the type of artifact, the need to support formative development of the designed artifacts, the properties of the artifact to be evaluated, and the constraints on resources available, such as time, labor, facilities, expertise, and access to research subjects. The framework and method support matching these in the first instance to one or more DSR evaluation strategies, including the choice of ex ante (prior to artifact construction) versus ex post evaluation (after artifact construction) and naturalistic (e.g., field setting) versus artificial evaluation (e.g., laboratory setting). Based on the recommended evaluation strategy(ies), guidance is provided concerning what methodologies might be appropriate within the chosen strategy(ies).}, author = {}, booktitle = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}, doi = {10.1007/978-3-642-29863-9_31}, keywords = {Design Science Research,Evaluation Method,Evaluation Strategy,Information Systems Evaluation,Research Methodology}, pages = {423--438}, title = {{A Comprehensive Framework for Evaluation in Design Science Research}}, year = {2012} } @book{Wieringa2014, abstract = {Abstract Design scientists have to balance the demands of methodological rigor that they share with purely curiosity-driven scientists, with the demands of practical utility that they share with utility-driven engineers. Balancing these conflicting demands can be ... $\backslash$n}, address = {Berlin, Heidelberg}, author = {Wieringa, .}, doi = {10.1007/978-3-662-43839-8}, isbn = {978-3-662-43838-1}, keywords = {Design Science}, pages = {493}, publisher = {Springer Berlin Heidelberg}, title = {{Design Science Methodology for Information Systems and Software Engineering}}, year = {2014} } @incollection{Wohlin2013, abstract = {The dependence on quality software in all areas of life is what makes software engineering a key discipline for todays society. Thus, over the last few decades it has been increasingly recognized that it is particularly important to demonstrate the value of software engineering methods in real-world environments, a task which is the focus of empirical software engineering. One of the leading protagonists of this discipline worldwide is Prof. Dr. Dr. h.c. , who dedicated his entire career to empirical software engineering. For his many important contributions to the field he has received numerous awards and recognitions, including the U.S. National Science Foundations Presidential Young Investigator Award and the Cross of the Order of Merit of the Federal Republic of Germany. He is a Fellow of both the ACM and the IEEE Computer Society. This book, published in honor of his 60th birthday, is dedicated to Dieter Rombach and his contributions to software engineering in general, as well as to empirical software engineering in particular. This book presents invited contributions from a number of the most internationally renowned software engineering researchers like , , , , , , and, of course, by himself. Several key experts from the Fraunhofer IESE, the institute founded and led by , also contributed to the book. The contributions summarize some of the most important trends in software engineering today and outline a vision for the future of the field. The book is structured into three main parts. The first part focuses on the classical foundations of software engineering, such as notations, architecture, and processes, while the second addresses empirical software engineering in particular as the core field of Dieter Rombachs contributions. Finally, the third part discusses a broad vision for the future of software engineering.}, address = {Berlin, Heidelberg}, author = {}, booktitle = {Perspectives on the Future of Software Engineering}, doi = {10.1007/978-3-642-37395-4_10}, isbn = {9783642373954}, pages = {145--157}, publisher = {Springer Berlin Heidelberg}, title = {{An Evidence Profile for Software Engineering Research and Practice}}, volume = {9783642373}, year = {2013} } @inproceedings{Wohlin2014, abstract = {Background: Systematic literature studies have become common in software engineering, and hence it is important to understand how to conduct them efficiently and reliably. Objective: This paper presents guidelines for conducting literature reviews using a snowballing approach, and they are illustrated and evaluated by replicating a published systematic literature review. Method: The guidelines are based on the experience from conducting several systematic literature reviews and experimenting with different approaches. Results: The guidelines for using snowballing as a way to search for relevant literature was successfully applied to a systematic literature review. Conclusions: It is concluded that using snowballing, as a first search strategy, may very well be a good alternative to the use of database searches. Copyright 2014 ACM.}, address = {New York, New York, USA}, author = {}, booktitle = {Proceedings of the 18th International Conference on Evaluation and Assessment in Software Engineering - EASE '14}, doi = {10.1145/2601248.2601268}, isbn = {9781450324762}, keywords = {Replication,Snowball search,Snowballing,Systematic literature review,Systematic mapping studies}, pages = {1--10}, publisher = {ACM Press}, title = {{Guidelines for snowballing in systematic literature studies and a replication in software engineering}}, year = {2014} } @incollection{Wohlin2003, author = { }, booktitle = {Esernet}, doi = {10.1007/978-3-540-45143-3_2}, isbn = {3-540-40672-7}, keywords = {dblp}, pages = {7--23}, title = {{Empirical Research Methods in Software Engineering}}, volume = {2765}, year = {2003} } @book{Wohlin2012, abstract = {Empirical software engineering research can be organized in several ways, including experiments, cases studies, and surveys. Experiments sample over the variables, trying to represent all possible cases; cases studies sample from the variables, representing only the typical cases(s). Every case study or experiment should have a hypothesis to express the desired result. The experimental design is especially important because it identifies key variables and their relationships. The design uses balancing, blocking, and local control to help minimize error. Analysis techniques depend on the design, the distribution of the data, and the type of investigation being carried out. Different techniques allow us to look at variable interaction and to look at combinations of effects. Using a technique similar to a board game, we can determine when we have enough evidence to demonstrate clear relationships among variables. {\textcopyright} 1997 Academic Press Inc.}, address = {Berlin, Heidelberg}, author = { Runeson, {\"{o}}st, Ohlsson, . and Regnell, Bj{\"{{\'{e}}n, Anders}, booktitle = {Experimentation in Software Engineering}, doi = {10.1007/978-3-642-29044-2}, isbn = {978-3-642-29043-5}, pages = {1--236}, publisher = {Springer Berlin Heidelberg}, title = {{Experimentation in Software Engineering}}, volume = {9783642290}, year = {2012} } @inproceedings{Zhou2016a, abstract = {—Context: The assessment of Threats to Validity (TTVs) is critical to secure the quality of empirical studies in Software Engineering (SE). In the recent decade, Systematic Literature Review (SLR) was becoming an increasingly important empirical research method in SE as it was able to provide the strongest evidence. One of the mechanisms of insuring the level of scientific value in the findings of an SLR is to rigorously assess its validity. Hence, it is necessary to realize the status quo and issues of TTVs of SLRs in SE. Objective: This study aims to investigate the-state-of-the-practice of TTVs of the SLRs published in SE, and further support SE researchers to improve the assessment and strategies against TTVs in order to increase the quality of SLRs in SE. Method: We conducted a tertiary study by reviewing the SLRs in SE that report the assessment of TTVs. Results: We identified 316 SLRs published from 2004 to the first half of 2015, in which TTVs are discussed. The issues associated to TTVs were also summarized and categorized. Conclusion: The common TTVs related to SLR research, such as internal validity and reliability, were thoroughly discussed in most SLRs. The threats to construct validity and external validity drew less attention. Moreover, there are few strategies and tactics being reported to cope with the various TTVs.}, address = {Hamilton, New Zealand}, author = { }, booktitle = {2016 23rd Asia-Pacific Software Engineering Conference (APSEC)}, doi = {10.1109/APSEC.2016.031}, isbn = {978-1-5090-5575-3}, keywords = {Evidence-Based Software Engineering,Systematic (Literature) Review,Threats to Validity}, pages = {153--160}, publisher = {IEEE}, title = {{A Map of Threats to Validity of Systematic Literature Reviews in Software Engineering}}, year = {2016} } texlib/MyVerbatim.sty10-100 \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{MyVerbatim} [2002/03/23 v1 's verbatim facilities% ] \newlength{\bvwidth} \setlength{\bvwidth}{\textwidth} %\addtolength{\bvwidth}{\marginparsep} %\addtolength{\bvwidth}{\marginparwidth} %\addtolength{\bvwidth}{-7pt} \addtolength{\bvwidth}{-1pt} % A low quality boxed verbatim environment %\newenvironment{boxedverbatim}% % {\VerbatimEnvironment \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% % {\end{Verbatim}\end{minipage}\end{Sbox} \setlength{\fboxsep}{1mm}\noindent\fbox{\TheSbox}} % A figure-generating boxed verbatim environment % % #1: Filename % #2: Caption % #3: Label \newcommand\VFTitle{X} \newcommand\VFCaption{X} \newcommand\VFLabel{X} % \begin{Code}{The Title} % xxx % \end{Code} \newenvironment{Code}[1]{% \renewcommand\VFTitle{#1}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \bigskip\noindent\setlength{\fboxsep}{.5mm}\framebox[\bvwidth][l]{\textsf{\small\bfseries\VFTitle}}\\ \framebox[\bvwidth][l]{\TheSbox}\bigskip} % This only exists so that we can show a Code environment within one \newenvironment{CodeAlt}[1]{% \renewcommand\VFTitle{#1}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \bigskip\noindent\setlength{\fboxsep}{.5mm}\framebox[\bvwidth][l]{\textsf{\small\bfseries\VFTitle}}\\ \framebox[\bvwidth][l]{\TheSbox}\bigskip} \newenvironment{CodeFig}[3]{% \renewcommand\VFTitle{#1}% \renewcommand\VFCaption{#2}% \renewcommand\VFLabel{#3}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \setlength{\fboxsep}{.5mm}\begin{figure}[ht]\framebox[\bvwidth][l]{\textsf{\small\bfseries\VFTitle}}\\ \framebox[\bvwidth][l]{\TheSbox}\caption{\VFCaption}\label{\VFLabel}\end{figure}} % This only exists so that we can show a CodeFig environment within one \newenvironment{CodeFigAlt}[3]{% \renewcommand\VFTitle{#1}% \renewcommand\VFCaption{#2}% \renewcommand\VFLabel{#3}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \setlength{\fboxsep}{.5mm}\begin{figure}[ht]\framebox[\bvwidth][l]{\textsf{\small\bfseries\VFTitle}}\\ \framebox[\bvwidth][l]{\TheSbox}\caption{\VFCaption}\label{\VFLabel}\end{figure}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \newenvironment{CodeFigB}[2]{% \renewcommand\VFCaption{#1}% \renewcommand\VFLabel{#2}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \setlength{\fboxsep}{.5mm}\begin{figure}[ht]\framebox[\bvwidth][l]{\TheSbox}\caption{\VFCaption}\label{\VFLabel}\end{figure}} \newenvironment{CodeFigBAlt}[2]{% \renewcommand\VFCaption{#1}% \renewcommand\VFLabel{#2}% \VerbatimEnvironment% \begin{Sbox}\begin{minipage}{\bvwidth}\footnotesize\begin{Verbatim}}% {\end{Verbatim}\end{minipage}\end{Sbox} \setlength{\fboxsep}{.5mm}\begin{figure}[ht]\framebox[\bvwidth][l]{\TheSbox}\caption{\VFCaption}\label{\VFLabel}\end{figure}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % Define verbatim things each with their own type. % \newcommand\email{\begingroup \urlstyle{sf}\Url} \newcommand\filename{\begingroup \urlstyle{sf}\Url} \newcommand\code{\begingroup \urlstyle{tt}\Url} \endinput 0 \hypertarget{hwlib-doxygen-#0140-analog_8hpp}{}\section{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0140-\/analog.hpp File Reference} \label{hwlib-doxygen-#0140-analog_8hpp}\index{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0140-\/analog.\+hpp@{code/hwlib/doxyfiles/texts/hwlib-\/doxygen-\/\#0140-\/analog.\+hpp}} @InProceedings{petit:icra:2015, author = { }, title = {An integrated framework for humanoid embodiment with a BCI}, booktitle = {IEEE International Conference on Robotics and Automation}, year = {2015}, address = {Seattle (WA), USA}, month = {May 26-May 30}, url = {https://hal-lirmm.ccsd.cnrs.fr/lirmm-01222969/document}, keywords = {Navigation, Cameras, Humanoid robots, Object recognition, Joints, Robot vision systems}, doi = {10.1109/ICRA.2015.7139592}, abstract = {This paper presents a framework to embody a user (e.g. disabled persons) into a humanoid robot controlled by means of brain-computer interfaces (BCI). With our framework, the robot can interact with the environment, or assist its user. The low frequency and accuracy of the BCI commands is compensated by vision tools, such as objects recognition and mapping techniques, as well as shared-control approaches. As a result, the proposed framework offers intuitive, safe, and accurate robot navigation towards an object or a person. The generic aspect of the framework is demonstrated by two complex experiments, where the user controls the robot to serve him a drink, and to raise his own arm.} }pliniodester/PhD_thesis \chapterquote{% Time and Space\dots It is not nature which imposes them upon us, it is we who impose them upon nature because we find them convenient.}% {-- , \textit{The Value of Science} (1905)} In this chapter\footnote{The present chapter was based on the articles \cite{dester2018} and \cite{dester2021part}.}, we include the spatial position of the nodes in the analysis. % Therefore, the transmission success probability $p_s$ can no longer be an arbitrary function of the traffic, because it needs to include the spatial positions into its model. % Thus, we need a more specific model for $p_s$, preferably one that entails analytical tractability. % In the following section, we propose a model inspired in Proposition~\ref{prop:PPP_ps} and other models adopted throughout the literature. Another important concept we deal with in this chapter is the \textit{stability} of the queued packets. When a queue is not stable the number of queued packets tends to infinity \textit{almost surely} as time tends to infinity. % We are very interested in stable queues, because one of the important metrics, namely the delay, tends to infinity in unstable systems, and that is not desirable. The additional notations used in this chapter are summarized in Table~\ref{tab:symbols}. % \begin{table}[hbt] \centering \caption{Notations and symbols used in this chapter} \label{tab:symbols} \setlength{\tabcolsep}{3pt} \begin{tabular}{l l} \hline \hline \textbf{Symbol} & \textbf{Definition/explanation} \\ \hline $\alpha\in(2,\infty)$ & path loss exponent \\ $\delta\in(0,1)$ & $\triangleq 2/\alpha$ \\ $N \in \N^*$ & number of user classes \\ $\cal{C}$ & $\triangleq \{1,2,\dots,N\}$, set of classes \\ $n\in\cal{C}$ & refers to the $n$th user class \\ $p_n\in(0,1)$ & medium access probability \\ $a_n\in(0,1)$ & packet arrival rate per time slot \\ $\bm{a}\in(0,1)^N$ & $=(a_1,a_2,\dots,a_N)$ \\ $p_{s,n}\in(0,1)$ & transmission success probability \\ $\theta_n \in\R_+$ & SIR threshold for successful communication \\ $\overline{R}_n\in\R_+$ & average transmission distance \\ $D_n\in(1,\infty)$ & average packet transmission delay \\ $\bm{D}\in(1,\infty)^N$ & $=(D_1,D_2,\dots,D_N)$ \\ $P_n\in\R_+$ & transmission power \\ $\bm{P}\in\R_+^N$ & $=(P_1,P_2,\dots,P_N)$ \\ $\Phi_\TX^{(n)}$ & Poisson point process for the transmitters \\ $\lambda_n\in\R_+$ & density of $\Phi_\TX^{(n)}$ \\ $\psi_n\in\R_+$ & $\triangleq 4\,\overline{R}_n^2 \,\theta_n^{\delta} \,\pi\delta/\sin(\pi \delta)$ \\ \hline \hline \end{tabular} \end{table} \newpage % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \section{Multi-class High-mobility Bipolar Networks} \label{sec:N-class} \subsection{Main Contributions} The main contributions of the work presented in this section when compared to the literature, particularly the paper by Stamatiou and Haenggi \cite{stamatiou2010random}, are twofold. % Firstly, we have extended the analysis of stability (Definition~\ref{def:stability}) and delay in random-access wireless networks to the case of a network with an arbitrary number $N$ of user classes. As far as we know, this is the first time in the literature the stability region is found in closed form for a network with $N > 3$ different user classes sharing the same channel. % Secondly, we have expanded the analysis to show that the channel sharing mechanism in the investigated scenario can be seen as a process of partitioning a fixed and well-defined quantity into portions, each portion allotted to each user class, the size of which varying in accordance with the user class parameters. More specifically, the novelty of the results on this section are summarized below: \begin{itemize}[itemsep=3pt,parsep=3pt,topsep=3pt,partopsep=3pt] \item We propose a tractable scenario to study the performance and stability of a Poisson network with an arbitrary number $N$ of classes of users sharing the same channel; \item a simple and elegant expression relating mean delays, arrival rates, user densities, mean link distances, and bit rates of all $N$ classes is derived for the case of a stable network. This expression clearly shows that each class of user takes a well-defined portion of the available finite resource in the RF channel (Proposition~\ref{prop:identity_1}); \item a closed form solution (Definition~\ref{def:closed_form}) to the fixed-point system of equations that determine the stationary transmission success probabilities for $N$ user classes is found; \item an intuitive equation is presented relating link quality, packet arrival rate, density of users, and stationary mean delay (Proposition 1); \item we prove the necessary and sufficient conditions that determine whether a given network is stable (Theorems \ref{TH:NEC_SUFF} and \ref{TH:STABILITY}); \item we establish a simple necessary condition for stability that does not depend on the transmit powers (Corollary~\ref{cor:stab}); \item the optimum transmit powers per user class that achieve the optimum stationary mean delays for each user class (Proposition~\ref{prop:opt}) are derived; \item the optimum packet arrival rates per user class that achieve the maximum channel throughput per unit of area (Proposition~\ref{prop:eps}) are derived; \item we conclude that depending on the channel and user classes, the best strategy to maximize channel throughput is to share the channel, instead of using one single class per channel. \end{itemize} % This section is organized as follows: Section \ref{sec:SysMod} describes the model used throughout the chapter and provides some important results from the literature to be used in the following sections; % Section \ref{sec:N-users} presents the main results of the paper, i.e., necessary and sufficient conditions for stability when we have $N$ interacting user classes, and shows a simple expression for the stationary mean delay and the packet success probability; Section \ref{sec:application} applies the obtained results in two general scenarios: one scenario optimizes the transmission power of different user classes with different delay requirements sharing the same channel and the other optimizes the throughput per unit of area; Section \ref{sec:conclusion} concludes the paper. \subsection{System Model} We consider a stationary high-mobility Poisson network with density $\lambda$ on $\R^2$ and with $N$ classes of users that share the same radio frequency channel as defined in the general network model of Chapter~\ref{cap:P2_00}. % Time is slotted, $\T=\N$, and for each time slot $t \in \T$ and each user class $n \in \cal{C} \triangleq \{1,2,\dots,N\}$, we have a homogeneous Poisson point process (PPP) denoted by $\Phi_\TX^{(n)}(t)\stackrel{*}{=} \{ X_{i\,n}(t) \}_{i\in\N}$ of density $\lambda_n$ on $\R^2$, which represents the position of the sources. These PPPs are independent from each other and from the past. % Furthermore, \begin{align*} \lambda = \sum_{n\in\cal{C}} \lambda_n, \quad\text{then,}\quad \Phi_\TX(t) = \sum_{n\in\cal{C}} \Phi_\TX^{(n)}(t) \quad\text{for every $t\in\T$}. \end{align*} \begin{remark} Using the general network model of Chapter~\ref{cap:P2_00}, the set of transmitters $\cal{N}_\TX = \N\times\cal{C}$. Thus, each transmitter should be indexed by a number and a class. However, for ease of notation and since the network is stationary across the users within a class, we shall omit the number and only refer to the class for parameters and metrics. % For example, the transmission success probability $p_{s,n}$ refers to $p_{s,\N\times\{n\}}$, i.e., it refers to the set of all users from $n$th class (or a typical user of that class). \end{remark} \begin{figure}[H] \centering \if\printfig1 \includegraphics[width=0.8\textwidth]{Figures/Ch7_BipolarQueuedNetwork.pdf} \else \includegraphics[draft,width=0.8\textwidth]{Figures/Ch7_BipolarQueuedNetwork.pdf} \fi \caption{Example of a bipolar high-mobility random network with $N=3$ user classes (one for each color). The queues represent the transmitters, and the potential receivers are represented by circular shapes of the corresponding color. The purpose of the arrows is to remember that the nodes are moving. Each transmitter communicates with the closest potential receiver, as it is shown by the dashed lines. Unconnected circles represent inactive receivers. The quantities $\lambda$, $a$, $p_s$ and $\phi$ are related to density of users, rate of arrival of packets, rate of service of packets and link quality, respectively.} \label{fig:BipolarNetwork} \end{figure} Each transmitter of user class $n$ transmits with power $P_n$ constant over time, and the transmitted signal is subjected to Rayleigh short-term fading and power law path loss function $\ell(r) = r^{-\alpha}, r>0$, where $\alpha>2$ is the path loss exponent. % For each time slot the position $X_{i\,n}(t)$ of the $i$th transmitter is reallocated following the high-mobility random walk model \cite{baccelli2010stochastic}. % % The $i$th transmitter of user class $n$ communicates with a receiver located at $Y_{i\,n}(t)$. Thus, the distance between the $i$th transmitter of class $n$ and its destination is given by $R_{i\,n}(t) = || X_{i\,n}(t) - Y_{i\,n}(t) ||$. % We further assume that each transmitter is associated with a ``son'' PPP that models the locations of its potential receivers. The receiver associated with the $i$th transmitter of class $n$ is chosen as the closest point in the respective son PPP as illustrated in Figure~\ref{fig:BipolarNetwork}. As a consequence, the link distances $\{ R_{i\,n}(t) \}_t$ are iid Rayleigh random variables\footnote{The iid random variables for the link separation distance are of grave importance for the theoretical model. Otherwise, there would exist unstable queues and, consequently, the queueing network would be unstable.} \cite[Eq.~(2.35)]{kingman1992poisson}. % Rayleigh distributed link separation distance has been used in several other works investigating similar scenarios (see \cite{haenggi2013diversity}). % We denote the mean transmission distance $\E[R_{i\,n}(t)]$ simply by $\overline{R}_n$ because we have stationarity across time and across users of a given class. Then, the density of $R_{i\,n}$ can be expressed as \begin{align*} f_{R_n}(r) = \frac{\pi\,r}{2 \overline{R}_n^2} \exp\!\left[ - \frac{\pi\,r^2}{4\overline{R}_n^2}\right], \qquad r\in\R_+. \end{align*} The occupation of the buffer at each transmitter is represented by its queue length $\{ Q_{i\,n}(t) \}_t$ of infinite capacity. % The packet arrival probability at each queue is denoted by $a_n$ and the medium access probability by $p_n$. % Within each slot, the first event to take place for each transmitter with a non-empty queue is the medium access decision with probability $p_n$. If it is granted access and the signal to interference ratio (SIR) %\footnote{We assume thermal noise is negligible; refer to \cite{haenggi2012stochastic} for further details.} is greater than a threshold $\theta_n>0$, a packet is successfully transmitted and leaves the queue. Then, we have the arrival of the next packet with probability $a_n$. The last event to take place is the displacement of the transmitters and destinations. The queue lengths of the $i$th transmitter, user class $n$ are Markov chains represented by \begin{equation*} Q_{i\,n}(t+1) = (Q_{i\,n}(t) - B_{i\,n}(t))_+ + E_{i\,n}(t), \quad t \in \N, \end{equation*} where $(\cdot)_+ \triangleq \max\{\cdot,0\}$, $\{E_{i\,n}(t)\}_t$ are iid Bernoulli random variables of parameter $a_n$, i.e., $E_{i\,n}(t)\sim\mathscr{B}(a_n)$ and represents the arrival process, $ B_{i\,n}(t) = e_{i\,n}(t)\,\ind\{\text{SIR}_{i\,n}(t)>\theta_{n}\} $ represents the departure process, where $\{e_{i\,n}(t)\}_t$ are iid Bernoulli random variables of parameter $p_n$, i.e., $e_{i\,n}(t)\sim\mathscr{B}(p_n)$, and the constant $\theta_n > 0$ represents the SIR threshold for successful communication. \begin{remark} In view of the model of Chapter~\ref{cap:P2_00}, we have that the arrival point process $A_{i\,(n)}$ is a Bernoulli point process of parameter $a_n$, the access point process $T_{i\,(n)}$ is a Bernoulli point process of parameter $p_n$ and the transmissions times $T_{i\,(n)}^*$ is a thinning of $T_{i\,(n)}$ conditioned to the corresponding queue being non-empty, and the success probability model $\mathscr{S}(f) = \ind\{f(0) > 1\}$, where $f(t') = \SIR_{i\,n}(t-t')/\theta_n$. \end{remark} % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{Analysis and Results} \label{ssec:N-users} From the stationarity across users of a class and ergodicity regarding time (if the system is stable), the transmission success probability $p_{s,n}$ is the limiting probability of a successful transmission from a typical user of class $n$, i.e., $$p_{s,n} = \lim_{t\to\infty} \P(\mathrm{SIR}_{i,n}(t) > \theta_n).$$ \begin{note} Remember that, \textit{à priori}, we do not need to take $t\to\infty$, any $t$ would suffice in an ergodic process. % However, as discussed in Section~\ref{sec:poisson_network} we would have to start the system at $-\infty$ or start the system distributed according to the stationary distribution. \end{note} The stationary mean delay $D_n$ to transmit packets of class $n$ is defined as the limiting ($t\to\infty$) expected time a packet spends in the buffer and the server. % See Theorem~\ref{th:little}. % For each time slot and for each class we have an independent homogeneous PPP, which is stationary and isotropic (invariant to translation and rotation, respectively). % The following results from the literature are used in many proofs throughout the chapter. In a wireless network, let us assume that \textit{(i)} the separation distance between a given pair TX - RX is equal to $r$, \textit{(ii)} the positions of the interferers (users who will transmit packets in a given time slot) follow a PPP of density $\lambda_\mathrm{eff}$, and \textit{(iii)} every transmitter has the same transmit power. Then, the probability of a successful transmission between TX and RX is given by \cite[Sec.~III.A]{haenggi2009stochastic} % \begin{align} % \P(\mathrm{SIR} > \theta) &= \E[\euler^{-\theta r^\alpha I}] \nonumber\\ % &= \exp\left( - \pi\,\Gamma(1+\delta) \Gamma(1-\delta)\,\theta^{\delta}\,r^2\,\lambda_\mathrm{eff} \right), % \end{align} % where $\delta \triangleq 2/\alpha$ and $I \triangleq \sum_{X\in\Phi} ||X||^{-\alpha}$ is the interference received by RX normalized by the transmit power and $\Phi$ is a PPP of density $\lambda_\mathrm{eff}$, which is the effective density of active sources. % % As described in the previous section, we consider a network with $N$ classes of users. % The following proposition presents the stationary success probability and mean delay when transmitting a packet in a stable network. The results that guarantee stability are presented later in the sequence, in Theorem~\ref{TH:NEC_SUFF}. \begin{proposition} \label{prop:psk} If the network is stable, then the stationary success probability and mean delay for a typical user of class $n \in \cal{C}$ are given by \begin{align} p_{s,n} &= \left( 1 + \dfrac{\psi_n}{P_n^\delta}\, \dfrac{\sum_j P_j^\delta\,a_j\lambda_j} {1 - \sum_j \psi_j\,a_j\lambda_j} \right)^{-1},\label{eq:psn}\\ D_n &= \dfrac{1 - a_n}{p_n\,p_{s,n} - a_n}, \label{eq:Dn} \end{align} where the sums are taken over the set of user classes $\cal{C}$, ${\delta \triangleq 2/\alpha}$, and \begin{equation}\label{eq:DefPhi} \psi_n \triangleq 4\,\Gamma(1+\delta) \Gamma(1-\delta) \overline{R}_n^2 \theta_n^\delta = 4\,\overline{R}_n^2\,\theta_n^{\delta}\, \frac{\pi\delta}{\sin(\pi \delta)}. \end{equation} \end{proposition} \begin{proof} First, we need to calculate the transmission success probability for a given traffic and a given link distance. % Using the reasoning of Proposition~\ref{prop:PPP_ps} for the case of several classes of users, we have \begin{align} \label{eq:P_SIR} \P(\text{SIR}_{i\,n}(t)>\theta_n \mid R_{i\,n}) &= \E\!\left[\exp\left(-\frac{\theta\,R_{i\,n}^\alpha}{P_n} \sum_{k\in\cal{C}} P_k\,I_k\right)\right]\nonumber\\ &= \prod_{k\in\cal{C}} \E\!\left[\exp\left(-\theta\,R_{i\,n}^\alpha\frac{P_k}{P_n} I_k\right)\right] \nonumber\\ &= \prod_{k\in\cal{C}} \exp\!\left( - \pi R_{i\,n}^2\, \theta^{\delta}\,\frac{\pi\delta}{\sin(\pi\delta)} \frac{P_k^\delta}{P_n^\delta}\lambda_\mathrm{eff}^{(k)}(t)\right) \nonumber\\ &= \exp\!\left( - \pi R_{i\,n}^2\,\theta^{\delta}\,\frac{\pi\delta}{\sin(\pi\delta)} \,\sum_{k\in\cal{C}}\frac{P_k^\delta}{P_n^\delta}\lambda_\mathrm{eff}^{(k)}(t)\right), \end{align} where $I_n$ is the interference from the $n$th class normalized by the transmit power, and $\lambda_{\mathrm{eff}}^{(n)}$ is the density of the thinned Poisson point process $\Phi_\TX^{(n)}$ by the nodes that are transmitting. % It is assumed that $\{I_n\}_{n \in \cal{C}}$ is iid. As $t\to\infty$, the effective PPP density of active sources $\lambda_\mathrm{eff}^{(n)}$ for each user class $n\in\cal{C}$ converges (by hypothesis) to $\lambda_n\,p_n\,\rho_n$, where $\rho_n = a_n/(p_n\,p_{s,n})$ is the load of the queue (or the probability of having a non-empty queue), which is the ratio between the arrival rate and the service rate of packets. Thus, $\lambda_\mathrm{eff}^{(n)} = \lambda_n\,a_n/p_{s,n}$. Then, to calculate the transmission success probability $p_{s,n}$, we use \eqref{eq:P_SIR}. Thus, by deconditioning the transmission success probability on $R_{i\,n}$, we take into account that $R_{i\,n}$ is Rayleigh distributed, that is \begin{align} \label{eq:aux_psk} p_{s,n} &= \lim_{t\to\infty}\P(\text{SIR}_{i\,n}(t)>\theta_n) \nonumber\\ &= \int_0^\infty \lim_{t\to\infty} \P(\text{SIR}_{i\,n}(t)>\theta_n \mid R_{i\,n}(t) = r)\,f_{R_n}(r)\,\mathrm{d}r \nonumber\\ &= \int_0^\infty \frac{\pi r}{2 \overline{R}_n^2} \exp\!\left[ - \frac{\pi\,r^2}{4\overline{R}_n^2} \left(1+\psi_n\sum_{k\in\cal{C}}\frac{P_k^\delta}{P_n^\delta}\lambda_\mathrm{eff}^{(k)}\right)\right]\mathrm{d}r \nonumber\\ &= \left( 1 + \dfrac{\psi_n}{P_n^\delta} \sum_{k\in\cal{C}} P_k^\delta \dfrac{a_k\lambda_k}{p_{s,k}} \right)^{-1}. \end{align} This expression can be rearranged as \begin{equation} \label{eq:psk_equivalence} \dfrac{P_n^\delta}{\psi_n} \left( \dfrac{1-p_{s,n}}{p_{s,n}}\right) = \sum_{k\in\cal{C}} P_k^\delta \dfrac{a_k\lambda_k}{p_{s,k}}. \end{equation} Note that the right-hand side of \eqref{eq:psk_equivalence} does not depend on $n$. Then, for all $j\in\cal{C}$, we have% \begin{equation} \label{eq:Pi_Pk} \dfrac{P_j^\delta}{\psi_j} \left( \dfrac{1-p_{s,j}}{p_{s,j}}\right) = \dfrac{P_n^\delta}{\psi_n} \left( \dfrac{1-p_{s,n}}{p_{s,n}}\right). \end{equation} For each $j$, we can solve the above equation for $p_{s,j}$ and plug it into the sum of \eqref{eq:aux_psk}. Then, we can solve it for $p_{s,n}$, which ends the proof for the $p_{s,n}$. The buffer (plus server) is a discrete time Geo/Geo/1 queue (Definition~\ref{def:geo/geo/1}) at stationary \cite{stamatiou2010random} and the equation for the delay $D_n$ comes from Theorem~\ref{th:geo/geo/1}. \end{proof} The following theorem shows the conditions for which the network is stable, i.e., it presents the region formed by all arrival rates $\bm{a}$ that make the system stable. \begin{theorem} \label{TH:NEC_SUFF} A necessary and sufficient condition for the system network to be stable is that $\bm{a} \in \bigcup_{\nu\in\cal{V}} \cal{E}_\nu$, where $\cal{V}$ is the space of all bijective functions from $\cal{C}$ to $\cal{C}$ and $\cal{E}_\nu$ is defined below with the convention $\sum_{k=1}^0 \cdot = 0$. \begin{align} \label{eq:stab_long} \cal{E}_{\nu} \triangleq \Bigg\{ \bm{a} \in [0,1)^N \Bigm\vert~ & 0 \le \dfrac{\psi_{\nu(n)}}{P_{\nu(n)}^\delta}\dfrac{a_{\nu(n)}}{p_{\nu(n)}-a_{\nu(n)}} \nonumber\\ & < \dfrac{ 1 - \sum_{k=1}^{n-1} \psi_{\nu(k)} a_{\nu(k)} \lambda_{\nu(k)} } {\sum_{k=1}^{n-1} P_{\nu(k)}^{\delta} a_{\nu(k)}\lambda_{\nu(k)} + \sum_{k=n}^N P_{\nu(k)}^\delta p_{\nu(k)}\lambda_{\nu(k)}}, \quad n \in \cal{C} \Bigg\}. \end{align} \end{theorem} % \begin{proof} Using stochastic dominance through dominant networks \cite[Section~2.1.2]{kompella2014stable}, it is possible to derive necessary and sufficient conditions for stability. % A dominant network behaves exactly the same as the original network, except that all user classes in a subset $\cal{D}$ of $\cal{C}$ transmit dummy packets. % Thus, the dominant network have more or the same number of buffered packets as the original network \textit{almost surely}. % Then, if the dominant network is stable, the original network is stable as well. On the other hand, if the queues of the user classes in $\cal{D}$ are not empty in the original network, then this system behaves exactly like the dominant network, i.e, both systems are \emph{indistinguishable}, see Definition~\ref{def:indistinguishable} and \cite[Section~3.2]{szpankowski1994stability}. % Therefore, if the dominant network is unstable, then the original network is unstable as well. % In order to have necessary and sufficient conditions, we must perform this verification for all $\cal{D} \subset \cal{C}$. Let us start with $\cal{D}=\cal{C}$, i.e., all users transmit dummy packets. For each step of the verification, we remove the stable user class from the set $\cal{D}$. This procedure repeats until the set $\cal{D}$ becomes empty. In order to attain stability of the dominant network we must have an arrival rate smaller than the service rate (Theorem~\ref{th:loynes}). Thus, a sufficient condition for the first user class stability is, for any queue $i$ of this class (by symmetry), \begin{equation*} a_1 < p_1\,\P(\widetilde{\text{SIR}}_{i,1} > \theta_1) = p_1 \left( 1 + \dfrac{\psi_1}{P_1^\delta} \sum_{k=1}^N P_k^\delta\,p_k\lambda_k \right)^{-1}, \end{equation*} where $\widetilde{\text{SIR}}$ represents the signal-interference ratio in the dominant network and the second equality comes from the same procedure to obtain \eqref{eq:aux_psk} with $\lambda_\mathrm{eff}^{(n)} = \lambda_n$ (all users are active, since every TX transmits dummy packets). % This guarantees stability for the first user class. Let us remove it from the set $\cal{D}$. Then, we calculate the stationary success probability of the first user class $\widetilde{p}^{(1)}_{s,1}$ for this dominant network. At steady state, we have \begin{equation*} \widetilde{p}^{(1)}_{s,1} = \left( 1 + \dfrac{\psi_1}{P_1^\delta} \left( P_1^\delta\,p_1\lambda_1\dfrac{a_1}{p_1\widetilde{p}^{(1)}_{s,1}} + \sum_{k=2}^N P_k^\delta\,p_k\lambda_k \right) \right)^{-1}, \end{equation*} which can be solved for $\widetilde{p}^{(1)}_{s,1}$, \begin{equation*} \widetilde{p}^{(1)}_{s,1} = \dfrac{1 - \psi_1\,\lambda_1\,a_1}{1 + \frac{\psi_1}{P_1^\delta} \sum_{k=2}^N P_k^\delta\,p_k\lambda_k}. \end{equation*} The next step is to verify the conditions of stability for the second user class, when the first user class is at steady state. After that, we remove the second user class from the set $\cal{D}$ and calculate the stationary success probability of the two stable user classes in the dominant network. We repeat these steps until we remove all user classes, \textit{i.e}, $\cal{D} = \{\}$. We show this by induction; we suppose stability of the user classes $1,2,\dots,j-1$. Let $\cal{D} = \{j,j+1,\dots\,N\}$; the $j$th user class is stable, given that all the user classes in $\cal{C}\setminus\cal{D}$ are stable, when \begin{align} \label{eq:aux_aj} a_j &< p_j\,\P(\widetilde{\text{SIR}}_{i,j} > \theta_j) \nonumber\\ &= p_j \left( 1+\dfrac{\psi_j}{P_j^\delta} \left( \sum_{k=1}^{j-1} P_k^\delta\,\lambda_k\,\dfrac{a_k} {\widetilde{p}^{(j)}_{s,k}} + \sum_{k=j}^N P_k^\delta\,p_k\lambda_k \right) \right)^{-1}, \end{align} where $\widetilde{p}^{(j)}_{s,k}$ is the $k$th user class success probability ($1 \leq k < j$) at steady state in the dominant network at the $j$th step. To calculate this probability, we must solve the following system of equations. For $k \in \{1,2,\dots,j-1\}$ \begin{equation*} \widetilde{p}^{(j)}_{s,k} = \left( 1 + \dfrac{\psi_k}{P_k^\delta} \left( \sum_{\ell = 1}^{j-1} P_\ell^\delta\,\lambda_\ell\, \dfrac{a_\ell}{\widetilde{p}^{(j)}_{s,\ell}} + \sum_{\ell = j}^{N} P_\ell^\delta\,p_\ell\lambda_\ell \right) \right)^{-1}. \end{equation*} Using an analogous approach as the one presented in the proof of Proposition~\ref{prop:psk}, we have that for $k \in \{1,2,\dots,j-1\}$, \begin{equation*} \widetilde{p}^{(j)}_{s,k} = \left( 1 + \dfrac{\psi_k}{P_k^\delta} \dfrac{ \sum_{\ell=1}^{j-1} P_\ell^\delta\,a_{\ell}\,\lambda_\ell + \sum_{\ell=j}^{N} P_\ell^\delta\,p_\ell\lambda_\ell } { 1 - \sum_{\ell=1}^{j-1} \psi_{\ell}\,a_{\ell}\,\lambda_\ell} \right)^{-1}. \end{equation*} Comparing the last two equations, it is easy to see that \begin{align*} \sum_{\ell = 1}^{j-1} P_\ell^\delta\,\lambda_\ell\, &\dfrac{a_\ell}{\widetilde{p}^{(j)}_{s,\ell}} + \sum_{\ell = j}^{N} P_\ell^\delta\,p_\ell\,\lambda_\ell = \dfrac{ \sum_{\ell=1}^{j-1} P_\ell^\delta\,a_{\ell}\,\lambda_\ell + \sum_{\ell=j}^{N} P_\ell^\delta\,p_\ell\lambda_\ell } { 1 - \sum_{\ell=1}^{j-1} \psi_{\ell}\,a_{\ell}\,\lambda_\ell } . \end{align*} Finally, we can use this result to rewrite \eqref{eq:aux_aj} as, for all $j \in \cal{C}$, \begin{equation*} 0 \le \dfrac{\psi_j}{P_j^\delta}\,\dfrac{a_j}{p_j-a_j} < \dfrac{1 - \sum_{k=1}^{j-1} \psi_k\,a_k\,\lambda_k} {\sum_{k=1}^{j-1} P_k^\delta\,a_k\lambda_k + \sum_{k=j}^N P_k^\delta\,p_k\lambda_k}. \end{equation*} This concludes the proof since the extension for the other partitions of $\cal{C}$ is analogous. \end{proof} Theorem~\ref{TH:NEC_SUFF} requires verifying $N\!\times\!N!$ inequalities, whereas the following theorem is equivalent and it involves only $N$ inequalities. % Thus, Theorem~\ref{TH:STABILITY} presents a simpler form of verifying the conditions for stability. % Also, it relates (in the proof) the stability condition with the stationary mean delays $\bm{D}$. However, we cannot prove \ref{TH:STABILITY} without \ref{TH:NEC_SUFF}. \begin{theorem}[\textbf{Network Stability}] \label{TH:STABILITY} The system network is stable if and only of $\bm{a}\in\cal{R}$, \begin{align*} \cal{R} \triangleq \bigg\{ \bm{a}\in[0,1)^N \bigm\vert ~ a_n < p_n, \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1 - \sum_{k} \psi_k\,a_k \lambda_k} {\sum_{k} P_k^{\delta}\,a_k \lambda_k} \quad\forall n \in \cal{C} \bigg\}. % &= \bigg\{ \bm{a}\in[0,1)^N \bigm\vert ~ a_n < p_n, \\ % &\hspace{2mm} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < % \dfrac{1 - \sum_{k \neq n} \psi_k\,a_k\,\lambda_k} % {P_n^\delta\,p_n\lambda_n + \sum_{k \neq n} P_k^{\delta}\,a_k\,\lambda_k} % ~\forall n \in \cal{C} \bigg\}. \end{align*} \end{theorem} \begin{proof} The proof consists of showing that the set $\cal{R}$ is equal to the set defined in Theorem~\ref{TH:NEC_SUFF}. % First, let us prove that $\bigcup_{\nu\in\cal{V}} \cal{E}_\nu \subset \cal{R}$. % For that we suppose $\bm{a} \in \cal{E}_\nu$ and we show $\bm{a} \in \cal{R}$ for all $\nu\in\cal{V}$ by induction. % For simplicity of exposition let us take $\cal{E}_\nu$ with $\nu: n \longmapsto n$, $n\in\cal{C}$. % We assume the inequality \begin{align} \dfrac{\psi_{N-j}}{P_{N-j}^\delta}\,\dfrac{a_{N-j}}{p_{N-j}-a_{N-j}} & < \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k}{\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k}. \label{eq:th2_basecase} \end{align} % is true for all $j\in\{0,\dots,m-1\}$ and we prove that it is also true for $j=m$. % First, we have to prove the base case $m = 1$. % Since $\bm{a}\in \cal{E}_\nu$, then for $j=0$ ($n=N$) % \begin{align*} \dfrac{\psi_N}{P_N^\delta}\,\dfrac{a_N}{p_N-a_N} & < \dfrac{1 - \sum_{k=1}^{N-1} \psi_k\,a_k \lambda_k}{\sum_{k=1}^{N-1} P_k^\delta\,a_k\lambda_k + P_N^\delta\,p_N\lambda_N} \\ & = \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k + \psi_N\,a_N \lambda_N}{\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k + P_N^\delta\,(p_N-a_N)\lambda_N}. \end{align*} % Then, using simple manipulations, we can show that the above inequality is equivalent to % \begin{align*} \dfrac{\psi_N}{P_N^\delta}\,\dfrac{a_N}{p_N-a_N} & < \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k}{\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k}. \end{align*} % Thus, the base case $m = 1$ is true. % Now, for $j=m$ and $\bm{a}\in \cal{E}_\nu$, we know that \begin{align*} \dfrac{\psi_{N-m}}{P_{N-m}^\delta}\,\dfrac{a_{N-m}}{p_{N-m}-a_{N-m}} & < \dfrac{1 - \sum_{k=1}^{N-m-1} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N-m-1} P_k^\delta\,a_k\lambda_k + \sum_{k=N-m}^{N} P_k^\delta\,p_k\lambda_k} \\ & = \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k + \sum_{k=N-m}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k + \sum_{k=N-m}^{N} P_k^\delta\,(p_k-a_k)\lambda_k}. \end{align*} Through simple manipulations we can show that the above inequality is equivalent to \begin{align*} & \dfrac{\psi_{N-m}}{P_{N-m}^\delta}\,\dfrac{a_{N-m}}{p_{N-m}-a_{N-m}} < \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k + \sum_{k=N-m+1}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k + \sum_{k=N-m+1}^{N} P_k^\delta\,(p_k-a_k)\lambda_k}. \end{align*} Now, we only need to verify that \begin{align*} &\dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k + \sum_{k=N-m+1}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k + \sum_{k=N-m+1}^{N} P_k^\delta\,(p_k-a_k)\lambda_k} < \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k}. \end{align*} Again, simple manipulations lead to the equivalent inequality \begin{align*} &\dfrac{\sum_{k=N-m+1}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=N-m+1}^{N} P_k^\delta\,(p_k-a_k)\lambda_k} < \dfrac{1 - \sum_{k=1}^{N} \psi_k\,a_k \lambda_k} {\sum_{k=1}^{N} P_k^\delta\,a_k\lambda_k}, \end{align*} which is true from the base case. This can be seen by multiplying \eqref{eq:th2_basecase} by $P^\delta_{N-j}(p_{N-j}-a_{N-j})$ at both sides of the inequality and summing over $j\in\{0,\dots,m-1\}$. Thus, $\cal{E}_\nu \subset \cal{R}$ for the mapping $\nu: n \longmapsto n$. The extension for another instances of $\nu\in\cal{V}$ is analogous. This concludes the proof that $\bigcup_{\nu\in\cal{V}} \cal{E}_\nu \subset \cal{R}$. % However, we still need to prove the converse, that is ${\cal{R} \subset \bigcup_{\nu\in\cal{V}}} \cal{E}_\nu$. % Note that the set of arrival rates that makes the system stable in Theorem~\ref{TH:NEC_SUFF} requires that at least one $a_n$ ($n\in\cal{C}$) satisfies \begin{equation} \label{eq:aux_stab_first} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1}{\sum_{k=1}^N P_k^\delta\,p_k \lambda_k}. \end{equation} Let us show that $\cal{R}$ requires the same restriction by contradiction. Suppose that there exist $\bm{a}\in\cal{R}$ such that % \begin{equation} \label{eq:aux_contr0} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} \ge \dfrac{1}{\sum_{k=1}^N P_k^\delta\,p_k \lambda_k} \quad \forall n \in \cal{C}. \end{equation} % Multiplying \eqref{eq:aux_contr0} by $P_n^\delta(p_n-a_n)\lambda_n > 0$ at both sides and summing over $n\in\cal{C}$ we have % \begin{align*} \sum_{n=1}^N \psi_n a_n \lambda_n \ge \frac{\sum_{n=1}^N P_n^\delta (p_n-a_n) \lambda_n} {\sum_{k=1}^N P_k^\delta\,p_k \lambda_k}, \end{align*} % which is equivalent to % \begin{align} \label{eq:aux_case0} \left(\sum_{n=1}^N \psi_n a_n \lambda_n \right) \left(\sum_{k=1}^N P_k^\delta\,p_k \lambda_k \right) \ge \sum_{n=1}^N P_n^\delta (p_n-a_n) \lambda_n. \end{align} % Since $\bm{a}\in\cal{R}$, then \begin{align} \label{eq:aux_theo2} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1 - \sum_{k} \psi_k\,a_k \lambda_k} {\sum_{k} P_k^{\delta}\,a_k \lambda_k} \quad \forall n\in\cal{C}. \end{align} % Again, multiplying \eqref{eq:aux_theo2} by $P_n^\delta(p_n-a_n)\lambda_n > 0$ at both sides, summing over $n\in\cal{C}$ and performing some manipulations we have % \begin{align} \label{eq:aux_theo2_case0} &\left(\sum_{n=1}^N \psi_n a_n \lambda_n \right) \left(\sum_{k=1}^N P_k^\delta\,a_k \lambda_k \right) < \left(1 - \sum_{k=1}^N \psi_k\,a_k \lambda_k \right)\left(\sum_{n=1}^N P_n^\delta (p_n-a_n) \lambda_n\right). \end{align} % Then, through some manipulations on \eqref{eq:aux_case0} and \eqref{eq:aux_theo2_case0}, we have % \begin{align*} 0 &\le \sum_{n=1}^N P_n^\delta a_n \lambda_n - \left(1 - \sum_{k=1}^N \psi_k\,a_k \lambda_k \right)\left(\sum_{n=1}^N P_n^\delta p_n \lambda_n\right) < 0, \end{align*} which clearly is a contradiction, since $\cal{R}$ is a non-empty set. Thus, there exists at least one $a_n$, $n\in\cal{C}$ that satisfies \eqref{eq:aux_stab_first}. % For simplicity of exposition, let us suppose that the arrival rate $a_n$ that satisfies this restriction is from the first user class ($n=1$). The next step is to show that as in the set $\bigcup_{\nu\in\cal{V}}\cal{E}_\nu$, the set $\cal{R}$ also requires that we have at least one $a_n$, aside from $a_1$, that satisfies % \begin{equation*} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1-\psi_1\,\lambda_1\,a_1}{P_1^\delta\,a_1\lambda_1+ \sum_{k=2}^N P_k^\delta\,p_k\lambda_k}. \end{equation*} % We can also prove this by contradiction and then, for simplicity of exposition, suppose that $a_2$ is the one that satisfies this restriction. We repeat this procedure until we reach all user classes. Let us show the $j$th step for completeness, $j\in\cal{C}$. Suppose that for all $n \in \{j,j+1,\dots,N\}$, % \begin{align} \label{eq:aux_contr_j} \dfrac{\psi_n}{P_n^\delta}\dfrac{a_n}{p_n-a_n} \ge \dfrac{ 1 - \sum_{k=1}^{j-1} \psi_k a_k \lambda_k } {\sum_{k=1}^{j-1} P_k^{\delta} a_k\lambda_k + \sum_{k=j}^N P_k^\delta p_k\lambda_k}. \end{align} Multiplying \eqref{eq:aux_contr_j} by $P_n^\delta(p_n-a_n)\lambda_n > 0$ at both sides, summing over $n\in\{j,j+1,\dots,N\}$ and manipulating we have % \begin{align} \label{eq:aux_case_j} &\left(\sum_{n=j}^N \psi_n a_n \lambda_n \right)\! \left(\sum_{k=1}^N P_k^\delta\,a_k \lambda_k + \sum_{k=j}^N P_k^\delta\,(p_k-a_k) \lambda_k \right) \ge \left( 1 - \sum_{k=1}^{j-1} \psi_k a_k \lambda_k \right)\! \left( \sum_{n=j}^N P_n^\delta (p_n-a_n) \lambda_n \right). \end{align} % Once again, multiplying \eqref{eq:aux_theo2} by $P_n^\delta(p_n-a_n)\lambda_n > 0$ at both sides, summing over $n\in\{j,j+1,\dots,N\}$ and manipulating we have % \begin{align} \label{eq:aux_theo2_case_j} &\left(\sum_{n=j}^N \psi_n a_n \lambda_n \right) \left(\sum_{k=1}^N P_k^\delta\,a_k \lambda_k \right) < \left(1 - \sum_{k=1}^N \psi_k\,a_k \lambda_k \right)\left(\sum_{n=j}^N P_n^\delta (p_n-a_n) \lambda_n\right). \end{align} % Then, through some manipulations on \eqref{eq:aux_case_j} and \eqref{eq:aux_theo2_case_j}, we have % \begin{align*} 0 &\le \left(1 - \sum_{k=1}^N \psi_k\,a_k \lambda_k \right)\left(\sum_{n=j}^N P_n^\delta (p_n-a_n) \lambda_n\right) - \left(\sum_{n=j}^N \psi_n a_n \lambda_n \right) \left(\sum_{k=1}^N P_k^\delta\,a_k \lambda_k \right) < 0. \end{align*} % As expected, we have a contradiction. Then, we must have at least one $a_n$, $n\in{\{j,j+1,\dots,N\}}$ that satisfies % \begin{equation} \label{eq:aux_region_j} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1-\sum_{k=1}^{j-1} \psi_k\,a_k\,\lambda_k} {\sum_{k=1}^{j-1} P_k^\delta\,a_k\lambda_k+ \sum_{k=j}^N P_k^\delta\,p_k\lambda_k}. \end{equation} % We assume that this is satisfied by the $j$th class and in this case, $\cal{R} \subset \cal{E}_\nu$ for $\nu: n \longmapsto n$. % Without the assumption of the ordering in which \eqref{eq:aux_region_j} is satisfied, we conclude that \eqref{eq:aux_region_j} must hold for at least one permutation of $\cal{C}$. This region is exactly $\bigcup_{\nu\in\cal{V}} \cal{E}_\nu$. Therefore, ${\cal{R} \subset \bigcup_{\nu\in\cal{V}} \cal{E}_\nu}$. % Finally, ${\cal{R} = \bigcup_{\nu\in\cal{V}} \cal{E}_\nu}$. \end{proof} \begin{remark} The convoluted conditions of Theorem~\ref{TH:NEC_SUFF} are extensively reduced in Theorem~\ref{TH:STABILITY}, for which we can see that as the density of users $\lambda_n$ increases or the quantity $\psi_n$ (which is inversely related to link quality, see \eqref{eq:DefPhi}) increases for some $n\in\cal{C}$, then the stability region $\cal{R}$ decreases for all user classes. On the other hand, if the transmission power $P_n$ increases for some $n\in\cal{C}$, then the stability region $\cal{R}$ increases for the $n$th class and decreases for all other classes. Surprisingly, when the access probability $p_n$ varies, the only affected class (regarding stability region) is the $n$th class, as long $p_n>a_n$. \end{remark} Figure~\ref{fig:ThStab} shows an example of stability region for $N=3$ classes, where we have used only three non-linear inequalities instead of 18. % The following corollary establishes a simple result on stability, which is used in Section~\ref{sec:application} to propose and solve optimization problems regarding delay and throughput. \begin{corollary} \label{cor:stab} There exists a vector of transmit powers ${\bm{P} \in \R_+^N}$ such that the network is stable if and only if $\bm{a}\in\cal{S}_0$, where \begin{equation*} \cal{S}_0 \triangleq \left\lbrace \bm{a}\in[0,1)^N \bigm\vert 0 \le \sum_{n\in\cal{C}} \dfrac{\psi_n \lambda_n}{\frac{1}{a_n}-\frac{1}{p_n}} < 1 \right\rbrace. \end{equation*} \end{corollary} \begin{proof} First, let us show that $\cal{R} \subset \cal{S}_0$ for all ${\bm{P} \in \R_+^N}$. % If $\bm{a}\in\cal{R}$, then for all $n\in\cal{C}$ \begin{align*} \dfrac{\psi_n}{P_n^\delta}\,\dfrac{a_n}{p_n-a_n} < \dfrac{1 - \sum_{k} \psi_k\,a_k \lambda_k} {\sum_{k} P_k^{\delta}\,a_k \lambda_k}. \end{align*} % Multiplying both sides of the above equation by $P_n a_n \lambda_n$ and summing over all $n\in\cal{C}$ results in \vspace{-5mm} \begin{align*} \sum_{n\in\cal{C}} \psi_n \lambda_n\,\dfrac{p_n\,a_n}{p_n-a_n} < 1 \end{align*} after some manipulations. Thus, $a\in\cal{S}_0$. Now, let us show that $\cal{S}_0 \subset \cal{R}$ for some $\bm{P}\in\R_+^N$. % In particular, let us choose $P_n = \psi_n a_n/(p_n-a_n)$, $n\in\cal{C}$. Then, the inequalities that describe the region $\cal{R}$ can be rewritten as one unique inequality \begin{align*} 1 < \frac{1 - \sum_{k} \psi_k a_k \lambda_k} {\sum_k \psi_k a_k^2 \lambda_k / (p_k-a_k)} \end{align*} that does not depend on $n$ anymore. It is easy to show that this inequality is the same as the one that defines the region $\cal{S}_0$. Thus, if $\bm{a}\in\cal{S}_0$, then $\bm{a}\in\cal{R}$ for that choice of $\bm{P}$ (or any scalar multiple). This ends the proof. \end{proof} Figure~\ref{fig:CorStab} shows the region of arrival rates, according to Corollary~\ref{cor:stab}, for which it is possible to find transmit powers that make the network stable. % On the other hand, out of this region, the system is always unstable. It is worth mentioning that $\psi_n$, defined in \eqref{eq:DefPhi}, is related to the quality of the link between receiver and transmitter for class $n\in\cal{C}$; the larger is the value of $\psi_n$, the poorer is the quality of the link. % Also, the stability region $\cal{R}$ showed in Fig.~\ref{fig:ThStab} is contained in $\cal{S}_0$. This is expected since we used the same parameters for both sets and Corollary~\ref{cor:stab} considers the best case scenario, where we can choose the transmit powers $\bm{P}$ for each $\bm{a}$. \begin{figure} \centering \begin{subfigure}[t]{.45\textwidth} \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_theorem_stab.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_theorem_stab.pdf} \fi \caption{Stability region $\cal{R}$ according to Theorem~\ref{TH:STABILITY} for $p_1 = 1/3$, $p_2 = 2/3$, $p_3 = 1$, $\psi_1\lambda_1=1$, $\psi_2\lambda_2=2$, $\psi_3\lambda_3=3$, $\psi_1/P_1 = 1/3$, $\psi_2/P_2 = 1/2$, $\psi_3/P_3= 1$.} \label{fig:ThStab} \end{subfigure}% \begin{subfigure}{.05\textwidth} \hspace{.05\textwidth} \end{subfigure}% \begin{subfigure}[t]{.45\textwidth} \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_corollary_stab.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_corollary_stab.pdf} \fi \caption{Maximum stability region $\cal{S}_0$ according to Corollary~\ref{cor:stab} for $p_1 = 1/3$, $p_2 = 2/3$, $p_3 = 1$, $\psi_1\lambda_1=1$, $\psi_2\lambda_2=2$, $\psi_3\lambda_3=3$.} \label{fig:CorStab} \end{subfigure} \caption{} \end{figure} From now on, we assume that whenever there is a packet in the buffer, the corresponding transmitter attempts to transmit, i.e., the medium access probability $p_n=1$ for all $n\in\cal{C}$. % We discuss in Section~\ref{sec:high-mobility} the validity of this assumption and the high-mobility assumption. The motivation is that, when the access probability of all classes is equal to one, we maximize the stability region $\cal{R}$. % This is easy to see with the inequalities of Theorem~\ref{TH:STABILITY}, where the right-hand side does not depend on $p_n$ and the left-hand side decreases monotonically with $p_n$. Thus, the stability region is maximized when $p_n=1$ for all $n\in\cal{C}$. The same occurs in Corollary~\ref{cor:stab}. % This result is surprising and might be explained by the fact that we have independence between adjacent time slots and, therefore, for each time slot there is a new scenario (a new effective PPP). % Then, it makes sense to always try retransmission. This approach also minimizes the mean delay according to \eqref{eq:Dn}, since the success probability $p_{s,n}$ in \eqref{eq:psn} does not depend on the access probability in a stable network. Using Proposition \ref{prop:psk}, Proposition~\ref{prop:identity_1} is introduced, which presents an equation that relates all performance parameters independently of the transmission powers. Also, the conditions for stability are extensively simplified, see Corollary~\ref{cor:stab}. % \begin{proposition} \label{prop:identity_1} If the network is stable and $p_n=1$ for all ${n\in\cal{C}}$, then the following identities hold (at stationary state): \begin{equation}\label{eq:identity_1} \sum_{n\in\cal{C}} \psi_n\,\lambda_n\,\dfrac{D_n}{D_n-1}\,\dfrac{a_n}{1-a_n} = 1, \end{equation} and \begin{equation*} \frac{\psi_j}{P_j^\delta} \left( \frac{D_j}{D_j-1}\,\frac{1}{1-a_j} - 1\right) = \frac{\psi_k}{P_k^\delta} \left( \frac{D_k}{D_k-1}\, \frac{1}{1-a_k} - 1 \right) \quad \forall\,j,k\in\cal{C}. \end{equation*} \end{proposition} \begin{proof} We start with the terms of the sum, \begin{align*} \psi_n\,\lambda_n\,\dfrac{D_n}{D_n-1}\,\dfrac{a_n}{1-a_n} &\stackrel{\text{(i)}}{=} \psi_n\lambda_n\dfrac{a_n}{1-p_{s,n}}\\ &= P_n^\delta \dfrac{\lambda_n\,a_n}{p_{s,n}} \left( \dfrac{\psi_n}{P_n^\delta} \dfrac{p_{s,n}}{1-p_{s,n}} \right)\\ &\stackrel{\text{(ii)}}{=} \frac{ P_n^\delta\frac{\lambda_n\,a_n}{p_{s,n}} } { \sum_j P_j^\delta\frac{\lambda_j\,a_j}{p_{s,j}} }, \end{align*} where (i) comes from \eqref{eq:Dn} with $p_n=1$ and (ii) comes from \eqref{eq:psk_equivalence}. Summing over $\cal{C}$ ends the proof of the first identity. % For the second relation of Proposition~\ref{prop:identity_1}, we use \eqref{eq:Dn} once again to find \begin{equation*} \dfrac{\psi_n}{P_n^\delta} \left( \dfrac{D_n}{D_n-1}\,\dfrac{1}{1-a_n} - 1\right) = \dfrac{\psi_n}{P_n^\delta} \dfrac{p_{s,n}}{1-p_{s,n}}. \end{equation*} Comparing this expression with \eqref{eq:Pi_Pk} ends the proof. \end{proof} Proposition~\ref{prop:identity_1} is an elegant form to see that a channel is a limited resource regarding traffic intensity and delay. Let us rewrite the identity \eqref{eq:identity_1} in terms of physical parameters, \begin{equation} \label{eq:physical} \sum_{n=1}^N 4\,\lambda_n\,\overline{R}_n^2\,\theta_n^{2/\alpha}\,\dfrac{D_n}{D_n-1}\,\dfrac{a_n}{1-a_n} = \dfrac{\sin(2\pi/\alpha)}{2\pi/\alpha}. \end{equation} % Note that $\frac{a_n}{1-a_n}$ and $\frac{\sin(2\pi/\alpha)}{2\pi/\alpha}$ are monotonic increasing functions and $\frac{D_n}{D_n-1}$ is a monotonic decreasing function. The right hand-side of \eqref{eq:physical} can be seen as the amount of resource available to all users of the channel. % Larger $\alpha$ results in higher $\frac{\sin(2\pi/\alpha)}{2\pi/\alpha}$, meaning that a larger amount of resource is available to users. This can be explained by recalling that a larger path loss exponent leads to stronger isolation among links sharing the channel and, consequently, more users can be accommodated in the network. % Therefore, the larger the path loss exponent $\alpha$, the larger (smaller) the terms $\lambda_n$, $\overline{R}_n$, $\theta_n$, $a_n$ ($D_n$) can be. The identity \eqref{eq:physical} also tells us that the $n$th class of user takes a well-defined portion of the amount of resource available in the network, which is given by the $n$th term in the summation. % This means that the values of $\lambda_n$, $\overline{R}_n$, $\theta_n$, $a_n$ and $D_n$ for a given class $n$ can be adjusted, while keeping the quantity $\lambda_n\,\overline{R}_n^2\,\theta_n^{2/\alpha}\,\frac{D_n}{D_n-1}\,\frac{a_n}{1-a_n}$ unchanged. % For instance, we can make a direct exchange between decreasing the delay $D_n$ and decreasing the arrival rate of packets $a_n$ (by controlling the ratio of transmit power levels), such that the term $\frac{D_n}{D_n-1}\,\frac{a_n}{1-a_n}$ remains constant; or else, increase the arrival rate of packets and decrease the density of users, such that the term $\lambda_n\,\frac{a_n}{1-a_n}$ remains constant. % Therefore, Proposition \ref{prop:identity_1} reveals, through a simple expression, the interplay among traffic intensity, mean delay, density of users, link distance, and outage probability, when the network is stable. \begin{remark} Corollary~\ref{cor:stab} and Proposition~\ref{prop:identity_1} are the simplest and the most meaningful results of the present chapter, as they translate the behavior of the network in simple equations, which do not directly depend on the transmission powers. \end{remark} \begin{figure} \centering \begin{subfigure}[t]{.45\textwidth} \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_Delay_equal_a.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_Delay_equal_a.pdf} \fi \caption{Optimum delays} \label{fig:Opt_Delay_equal} \end{subfigure}% \begin{subfigure}{.05\textwidth} \hspace{.05\textwidth} \end{subfigure}% \begin{subfigure}[t]{.45\textwidth} \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_Power_equal_a.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_Power_equal_a.pdf} \fi \caption{Optimum transmit powers} \label{fig:Opt_Power_equal} \end{subfigure} \caption{These figures represent the optimization of a 3-class network with the following parameters: $a_1=a_2=a_3=a$, $\psi_1\,\lambda_1 = 0.1$, $\psi_2\,\lambda_2 = 0.3$, $\psi_3\,\lambda_3 = 0.6$ and $c_1 = \frac{10}{16},~ c_2 = \frac{5}{16},~ c_3 = \frac{1}{16}$.} \label{fig:opt} \end{figure} % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{Interpretation and Application} \label{sec:application} In this section, we solve two optimization problems using the proposed formulation applied to scenarios of different classes of terminals sharing a radio channel. % -------------------------------- % \subsubsection{Delay Optimization} \label{ssec:opt_delay} Let us consider the scenario with $N$ classes sharing a channel. Each class may represent a particular user application, with each application having a different delay requirement in the network. % Let us suppose we are interested in adjusting the transmit power of each user class, such that the weighted average delay among all classes is minimized. % This problem is addressed as follows. For fixed arrival rates of vector $\bm{a}$ that satisfies Corollary~\ref{cor:stab}, i.e., for $\bm{a}\in\cal{S}_0$, let us minimize the delays $\bm{D}$ by changing the ratio between the transmit powers $\bm{P}$. % Each user class requires a different response time, then we weight the optimization problem with the vector $(c_1,c_2,\dots,c_N) \in \R_+^N$. The larger the coefficient of a class, the smaller the resulting mean delay to deliver packets for that class. Then, we have \begin{equation} \label{eq:opt_prob} \min_{\bm{P}\in\R_+^N} \sum_{n\in\cal{C}} c_n D_n = \min_{\bm{P}\in\R_+^N} \sum_{n\in\cal{C}} \frac{c_n\,(1-a_n)} {\left( 1 + \frac{\psi_n}{P_n^\delta} \frac{\sum_j P_j^\delta\,a_j\lambda_j} {1 - \sum_j \psi_j\,a_j\lambda_j} \right)^{\hspace{-1mm}-1}\!\!\! - a_n}, \end{equation} where $D_n$ is given by Proposition~\ref{prop:psk}. Note that as thermal noise is not considered in our model, we have a degree of freedom for the optimum solution $\bm{P}^*$, which agrees with the formulation in \eqref{eq:opt_prob}. \begin{proposition} \label{prop:opt} The minimum of the optimization problem \eqref{eq:opt_prob} is attained by \begin{equation} \label{eq:opt_D} {P_n^*}^\delta = \dfrac{\beta}{\lambda_n a_n} \left( \frac{a_n\,\cal{A}_n} {1 - \sum_{k} \cal{A}_k } + \frac{\sqrt{c_n\,\cal{A}_n}} { \sum_{k} \sqrt{c_k\,\cal{A}_k } } \right), \quad n\in\cal{C}, \end{equation} where $\beta$ is any positive real constant, $\cal{A}_n \triangleq \psi_n\,\lambda_n\,\frac{a_n}{1-a_n}$ and the sums are over $\cal{C}$. \end{proposition} \begin{proof} Since we have one degree of freedom for the solution $\bm{P}^*$, let us set $\sum_j P_j^\delta\,a_j\lambda_j = 1 - \sum_j \psi_j\,a_j\lambda_j$ to extensively simplify the algebraic manipulations. Then, we use the Karush-Kuhn-Tucker conditions \cite[Section~3.3.1]{bertsekas1999nonlinear} in the \emph{Lagrangian} function \begin{equation*} \begin{split} \cal{L}(\bm{P},\mu) &= \sum_{n\in\cal{C}} \frac{c_n (1-a_n)} {\left( 1 + \frac{\psi_n}{P_n^\delta}\right)^{-1}\! - a_n} + \mu\left[ \sum_{j\in\cal{C}} P_j^\delta\,a_j\lambda_j - \left(1 - \sum_{j\in\cal{C}} \psi_j\,a_j\lambda_j\right) \right], \end{split} \end{equation*} where $\mu\in\R$ is the \emph{Lagrange} multiplier. % The objective function is strictly convex (the Hessian is a diagonal matrix with positive eigenvalues) and the feasible region is a hyperplane, therefore the solution is the global optimum. % Now, we return to the original problem that does not have the artificial constraint. Thus, we multiply the solution by an arbitrary constant $\beta>0$ to obtain the general solution. \end{proof} It is interesting to note that ${\sum_k\cal{A}_k<1}$ by Corollary~\ref{cor:stab}. Therefore, ${P_n^*}^\delta$ is always a positive quantity. % Also, if $c_n = \cal{A}_n$ for all $n\in\cal{C}$, then the optimum delays are all equal and given by ${D_1 = D_2 = \cdots = D_N = \left( 1 - \sum_k \cal{A}_k \right)^{-1}}$. Thus, we can always choose transmit powers, such that we have the same mean delay for all classes! As an example, let us consider a 3-class network, where Class 1 has a more restrictive delay requirement than Class 2, which is more restrictive than Class 3. We consider that all classes have the same arrival rate of packets, i.e., $a_1=a_2=a_3=a$. Figure~\ref{fig:Opt_Delay_equal} shows the expected waiting time of a packet before a successful transmission, which is $D_n^*-1$, ($n = 1,2,3$) since a transmission takes exactly one time slot. As expected, the optimization resulted in monotonic increasing functions and $D_1^* < D_2^* < D_3^*$ for all $a$. Figure~\ref{fig:Opt_Power_equal} shows the normalized\footnote{Whenever we refer to normalized $b_n {P_n^*}^\delta$, it means that we choose $\beta$ in Proposition~\ref{prop:opt} such that $\sum_k b_k {P_k^*}^\delta = 1$ and $b_n$ is any term that depends on $n$, for example $b_n = \lambda_n$ or $b_n = a_n \lambda_n$.} transmit powers per unit of area $\lambda_n {P_n^*}^\delta$ as a function of $a$. In this case, we do not have a clear hierarchy among the transmit powers, as it depends on the traffic intensity. For $n\in\cal{C}$, if the network is close to saturation, i.e., $\sum_k\cal{A}_k$ tends to 1, then the normalized $\lambda_n {P_n^*}^\delta$ approaches $\cal{A}_n/\sum_k\cal{A}_k$ and, at first order, it does not depend on the coefficients $c_1, c_2, \dots, c_N$. On the other hand, if the network is at low traffic, i.e., $\sum_k\cal{A}_k$ tends to 0, then the normalized $a_n \lambda_n {P_n^*}^\delta$ approaches $\sqrt{c_n\cal{A}_n}/\sum_k\sqrt{c_k\cal{A}_k}$. % -------------------------------- % \subsubsection{Throughput Optimization} Now, let us maximize the total throughput of the channel per unit of area with the constraint that the system is stable. % Since each transmitter performs retransmissions until the packet is correctly received by the intended receiver, then all packets are successfully transmitted (eventually) in a stable system. Thus, the throughput per transmitter is given by the packet arrival rate $a_n$. % The density of users per unit of area is given by $\lambda_n$, then the throughput of the $n$th user class per unit of area is simply $\mathscr{T}_n = \lambda_n\,a_n$ and the throughput per unit of area of the entire system $\mathscr{T}$ is the sum of the throughput for all classes $n\in\cal{C}$. Using Corollary~\ref{cor:stab}, we can formulate the optimization problem as ${\displaystyle\max_{\bm{a}\in\cal{S}_0}} \sum_{n} \lambda_n\,a_n$. % However, the strict inequality in the region $\cal{S}_0$ of Corollary~\ref{cor:stab} results in an optimization problem that is not well-posed. In this case, if the optimum solution lies in the boundary of the feasible region, then the solution does not exist. % To circumvent this problem, we propose a new region $\cal{S}_\epsilon\subset\cal{S}_0$ by adding an arbitrarily small parameter $\epsilon \in (0,1)$ in the inequality, i.e., the new region is given by \begin{equation*} \cal{S}_{\epsilon} = \left\lbrace \bm{a}\in[0,1)^N \bigm\vert \sum_{n\in\cal{C}} \psi_n\,\lambda_n\,\dfrac{a_n}{1-a_n} \le 1-\epsilon \right\rbrace. \end{equation*} % As the parameter $\epsilon$ increases, the system becomes less sensitive to perturbations\footnote{When the system parameters suffer a sufficiently small change, the system remains stable.}. Now, the optimization problem is posed as \begin{equation} \label{eq:opt_thr} \max_{\bm{a}\in\cal{S}_\epsilon} \mathscr{T} = \max_{\bm{a}\in\cal{S}_\epsilon} \sum_{n\in\cal{C}} \lambda_n\,a_n. \end{equation} % The following proposition presents the solution to the optimization, i.e., the optimum arrival rates $\bm{a}^*$ that maximize the throughput per unit of area and maintain the system stable. \begin{proposition} \label{prop:eps} If \begin{equation} \label{eq:opt_req} \sum_{k\in\cal{C}}\lambda_k\psi_k\left(\sqrt{\frac{\max_n \psi_n}{\psi_k}}-1\right)<1-\epsilon, \end{equation} then the solution of \eqref{eq:opt_thr} is attained by \begin{equation} a_n^* = 1 - \frac{\sum_k \lambda_k \sqrt{\psi_n\psi_k}}{1-\epsilon+\sum_k\lambda_k\psi_k}, \quad n\in\cal{C}, \end{equation} where the sums are over $\cal{C}$. If the inequality \eqref{eq:opt_req} is not satisfied, then the $m$th class is excluded, where $m = \arg\max_n \psi_n$, and the inequality is checked again. \end{proposition} \begin{proof} It is a direct application of the Karush-Kuhn-Tucker conditions \cite[Section~3.3.1]{bertsekas1999nonlinear} in the \emph{Lagrangian} function \begin{align*} \cal{L}(\bm{a},\mu) = \sum_{n\in\cal{C}} \lambda_n a_n + \mu \left[ \sum_{k\in\cal{C}}\psi_k\lambda_k\frac{a_k}{1-a_k} - (1-\epsilon)\right], \end{align*} where $\mu\in\R$ is the Lagrange multiplier associated with the constraint of stability. Equation~\eqref{eq:opt_req} guarantees that the solution $\bm{a}^*\in[0,1)^N$. % The objective function is convex (affine function) and the region $\cal{S}_\epsilon$ is strictly convex because the Hessian of the function that defines the region is a diagonal matrix with negative eigenvalues. Therefore the presented solution is the global optimum and it is unique. \end{proof} In the optimization \eqref{eq:opt_thr}, we still have freedom to choose the transmit powers $\bm{P}$, as long the network remains stable. The best way of choosing $\bm{P}$ is by minimizing the delays, which we have already done in Subsection~\ref{ssec:opt_delay}, Proposition~\ref{prop:opt}, where the arrival rates $\bm{a} \in \cal{S}_\epsilon \subset \cal{S}_0$ are fixed. When the optimization is performed in this sequence (maximization of throughput, then minimization of delay), we have the optimum throughput (per unit of area) and the optimum delays for the optimum configuration of arrival rates. %% Later in this section, we illustrate this procedure with a numerical example. In order to solve the optimization problem \eqref{eq:opt_thr} we did not have to handle with the transmit powers $\bm{P}$ directly, which would make the solution and the problem formulation more cumbersome. This shows the usefulness of Corollary~\ref{cor:stab}. \begin{figure} \centering \begin{subfigure}[t]{.45\textwidth} \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_a_eps.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_a_eps.pdf} \fi \caption{Optimum arrival rates}% \label{fig:Opt_a_eps} \end{subfigure}% \begin{subfigure}{.05\textwidth} \hspace{.05\textwidth} \end{subfigure}% \begin{subfigure}[t]{.45\textwidth}%% \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_lam_a_eps.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_lam_a_eps.pdf} \fi \caption{Optimum throughput. The dashed curve corresponds to a scenario with only the best performing class} \label{fig:Opt_lam_a_eps} \end{subfigure} % \begin{subfigure}[t]{.45\textwidth}%%% \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_D_eps.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_D_eps.pdf} \fi \caption{Optimum delays} \label{fig:Opt_D_eps} \end{subfigure}% \begin{subfigure}{.05\textwidth} \hspace{.05\textwidth} \end{subfigure}% \begin{subfigure}[t]{.45\textwidth}%%%% \centering \if\printfig1 \includegraphics[width=\textwidth]{Figures/Ch7_Opt_P_eps.pdf} \else \includegraphics[draft,width=\textwidth]{Figures/Ch7_Opt_P_eps.pdf} \fi \caption{Optimum transmit powers} \label{fig:Opt_P_eps} \end{subfigure} \caption{These figures represent the optimization of the throughput and mean delay of a 3-class network with the parameters given in Table~\ref{tab:param}.} \label{fig:opt_eps} \end{figure} Let us illustrate the throughput optimization problem with a system for which the parameters are shown in Table~\ref{tab:param}. Figure~\ref{fig:Opt_a_eps} shows the optimum arrival rates $\bm{a^*}$ that maximizes the throughput per unit of area. % \begin{table}[hbt] \centering \caption{Network parameters for Fig.~\ref{fig:opt_eps}} \begin{tabular}{l l} \hline \hline \textbf{Parameters} & \textbf{Values} \\ \hline $(\lambda_1,\lambda_2,\lambda_3)$ & $=(1,2,3)$ \\ $(\psi_1,\psi_2,\psi_3)$ & $=(0.3,0.5,0.4)$ \\ $(c_1,c_2,c_3)$ & $=(\frac{1}{3},\frac{1}{3},\frac{1}{3})$ \\ \hline \hline \end{tabular} \label{tab:param} % \vspace*{-\baselineskip} \end{table} % It is quite interesting that the optimum solution is not (necessarily) exclusively activating the class with the best link quality (i.e., the class with the smallest $\psi$, which is Class 1 in this example). In Fig.~\ref{fig:Opt_lam_a_eps} it is shown the optimum throughput for each class and the total throughput of the system. % For comparison, we plotted a dashed curve representing the total throughput if we only use the best performing user class, regarding throughput. The dashed curve is below the optimum total throughput for all $\epsilon$. Therefore, the best solution is always a combination of all user classes, as long as \eqref{eq:opt_req} is satisfied. % On the other hand, if this equation is not satisfied, it means that there is at least one user class with a bad link quality, such that it is better (regarding throughput efficiency) to reallocate this user class to another channel. Now that we have, for each $\epsilon$, the arrival rate configuration $\bm{a}^*(\epsilon)$ which gives the maximum throughput, we can use Proposition~\ref{prop:opt} to find the best configuration of transmit powers $\bm{P}^*(\epsilon)$ that minimizes the sum of the mean delays for each optimum configuration of arrival rates $\bm{a}^*(\epsilon)$. % Figure~\ref{fig:Opt_D_eps} shows the result of this optimization, which is a direct application of \eqref{eq:opt_D}. % It is worth noting that as we increase $\epsilon$ the system is farther from instability, which corresponds to having a smaller delay to transmit packets, as we can see in Fig.~\ref{fig:Opt_D_eps}, and a smaller throughput, as shown in Fig.~\ref{fig:Opt_lam_a_eps}. % Figure~\ref{fig:Opt_P_eps} shows the optimum distribution of power per unit of area required by each user class. Notice that the first user class, which has the best link quality, uses the smallest power per unit of area. However, this behavior is more intricate; it also depends on the density of users of the corresponding class. Notice, for example, the inversion between user classes 2 and 3 as we increase $\epsilon$ in Fig.~\ref{fig:Opt_P_eps}. Another interesting and direct result from Corollary~\ref{cor:stab} is to provide an upper bound for the total throughput per unit of area $\mathscr{T}$ in a stable system, which is given by $1/(\min_n\psi_n)$. % This is not a tight bound, however it is interesting on its own, due to its simplicity and the fact that it does not depend on the density of users. The proof follows \begin{align} \mathscr{T} = \sum_{k\in\cal{C}} a_k\lambda_k &\le \frac{1}{\min_{n}\psi_n} \sum_{k\in\cal{C}} \psi_k\,a_k\,\lambda_k \nonumber\\ &< \frac{1}{\min_{n}\psi_n} \sum_{k\in\cal{C}}\psi_k\,\lambda_k\,\frac{a_k}{1-a_k} \nonumber\\ &< \left(\textstyle\min_{n}\psi_n \right)^{-1}, \end{align} where the last inequality comes from Corollary~\ref{cor:stab}. % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \section{On the High-mobility Assumption} \label{sec:high-mobility} In this section, we address the high-mobility assumption, which may not be realistic in real wireless networks, since the mobility of transmitters does not change drastically between adjacent time slots. Therefore, the independence assumption would not hold. % Nevertheless, in a stable wireless network that has a small packet arrival rate $a$ per user or a small access probability $p$, the correlation might be sufficiently small such that the independence (high-mobility) assumption is reasonable. In \cite{haenggi2013diversity} the authors show that if the access probability $p$ is sufficiently small, then the independence assumption provides a good approximation. When the packet arrival rate $a$ or the access probability $p$ are small, the typical user sees a significantly different PPP of transmitters for each time slot, which justifies the independence (high-mobility) assumption. We verified this claim through simulations and the result is shown in Fig.~\ref{fig:high-mobility}, where it is used one user class with ${\lambda c \overline{R}^2 = \pi/4}$, ${\alpha = 3}$ and ${\theta = 1}$. % \begin{figure} \centering \if\printfig1 \includegraphics[width=0.55\textwidth]{Figures/Ch7_high-mobility.pdf} \else \includegraphics[draft,width=0.75\textwidth]{Figures/Ch7_high-mobility.pdf} \fi \caption{Queue load $\rho$ as a function of the access probability $p$. Simulation results with a static network are presented in marks and the theoretical results with the high-mobility assumption are presented in curves.} \label{fig:high-mobility} \end{figure} The mean load of the queues $\rho$, which is equivalent to the percentage of queues with packets to transmit, are plotted as a function of the access probability $p$ for several values of arrival rate $a$. As expected, for small values of $a$ or $p$, the theoretical model presents good estimations of the average queue load. % It is important to emphasize that we did not plot the mean delay $D$, because in a static PPP there might exist a set of unstable users, whose queues and delays tend to infinity. This would raise the average delay to infinity too. Then, we chose to plot the mean load $\rho$, which is equal to 1 for unstable users and does not tend to infinity opposed to the mean delay $D$. To establish Proposition~\ref{prop:identity_1}, we supposed that the access probability $p$ is equal to one for all users. In the context of high-mobility, this approach makes sense, since the typical user sees a different interference scenario for each time slot. Thus, it is reasonable to attempt a retransmission every time slot until the packet is successfully transmitted. This also minimizes the mean delay $D$, which is in accordance with \eqref{eq:Dn}, as the transmission success probability $p_{s,n}$ does not depend on the access probability $p_n$ in a stable network. There is another scenario, which does not require high-mobility to achieve spatial independence between adjacent time slots. This scenario is a network that uses the frequency-hopping scheme over a set of channels \cite{tse2005fundamentals}. For each time slot, there is a different PPP pattern, since the transmitting nodes select with equal probability one channel to transmit. Thus, the spatial correlation between time slots decreases with the number of channels available for selection. % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \section{Bandwidth Partitioning on High-mobility Bipolar Networks} \label{sec:bandwidth} In this section, we consider the model presented in Section~\ref{sec:N-class} with only two classes of users, i.e., $N=2$. % However, differently from our previous models, the available frequency band is partitioned among users of one of the classes, such that users of this class can access only a fraction of the original bandwidth available to the network at a time. % The other class continues using the whole frequency band available. % The motivation for such bandwidth partitioning is to reduce the interference among users, leading to a higher network capacity. The work presented in this section generalizes the paper by Jindal et al. in \cite{jindal2008bandwidth} in the sense that it (\textit{i}) considers two user classes and (\textit{ii}) investigates the effects of bandwidth partitioning on the mean \textit{queuing} delay. % By considering two classes of users, we are able to study the performance of a network in which users of one class are allowed to transmit on a fraction of the channel accessed by users of the other class. Also, by studying the delay, we are able to have a better understanding of the effects of bandwidth partitioning on network performance. % % The present work also extends the study presented in \cite{dester2018} by using bandwidth partitioning on one of the user classes. % More specifically, we recover some results from \cite{jindal2008bandwidth} and \cite{dester2018} by setting the arrival rate of Class 2 $a_2 = 0$ and by setting the number of band partitions $M=1$, respectively. We used the wireless network formulation proposed in Section~\ref{sec:N-class} to achieve a tractable framework for the analysis of the bandwidth partitioning. The main contributions presented in this section can be summarized as follows: \begin{itemize} \item We have found a pair of simple equations that relate the main parameters of the network and guarantees network stability (Theorem~\ref{th:identity}); \item A transcendental equation to find the optimum Class 1 spectral efficiency is derived, along with the optimum number of partitions (Theorem~\ref{th:optimum_eta}); \item The first-order expansion of the optimum spectral efficiency around a small use of the channel by Class 2 (Eq.~\eqref{eq:eta_expansion}); \item We show that the bandwidth partitioning strategy is more effective when the path loss exponent is not small and the performance requirements of Class 2, the one whose users access the whole bandwidth, is not high in respect to the maximum performance attainable (Fig.~\ref{fig:a_ratio}). \end{itemize} % The paper is organized as follows. In Section~\ref{sec:sysmod_M} the system model is presented and in Section~\ref{sec:stab} we derive the stability conditions and expressions for stationary transmission success probability and mean delay. Using these expressions, we present in Section~\ref{sec:opt} the optimization problem involving bandwidth partitioning and proves the existence and uniqueness of the solution. Section~\ref{sec:num} provides some numerical examples of bandwidth optimization. %, also including a case involving cellular and D2D. % Section~\ref{sec:conc} concludes the paper. % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{System Model} \label{sec:sysmod_M} We consider the network model presented in Section~\ref{sec:N-class} with two user classes (namely Class 1 and Class 2), and only one and important difference that users of Class 1 access a fraction of the bandwidth during each transmission. %The high-mobility model, which is used in several works (see, for instance \cite{jindal2008bandwidth}, \cite{baccelli2010stochastic}, \cite{stamatiou2010random}, \cite{dester2018}), assumes that users occupy a different point in space for each time slot. %The communication protocol is the slotted Aloha, i.e., for each time slot, if a source (transmitter) has packets to transmit, it will transmit one packet with a fixed probability, independently from other sources and the past. The position of source $i \in \N$ of class $n\in\{1,2\}$ at time $t\in\N$ is denoted by $X_{i,n}(t) \in \R^2$. We assume that $\{X_{i,n}(t)\}_i\subset \R^2$ is a marked homogeneous Poisson point process (PPP) of density $\lambda_n$. These PPPs are independent across classes and time slots. The transmit power of a source of class $n$ is denoted by $P_n$, assumed to be constant over time slots and the same for all sources within a class. % % Let $Y_{i,n}(t)\in\R^2$ be the position of the destination terminal with which the $i$th source of class $n$ communicates. The distribution of $Y_{i,n}(t)$ is such that the location of the destination terminal is at a random distance $R_{i,n}(t) = ||X_{i,n}(t) - Y_{i,n}(t)||$ from the corresponding transmitter in a uniformly random direction, where $||\cdot||$ is the euclidean norm. We assume that $R_{i,n}(t)$ is iid across time slots and users, and it follows a Rayleigh distribution of mean $\overline{R_n}$. % % This assumption contributes to the tractability of the model and it was also used in \cite{dester2018, lin2014spectrum, di2014stochastic}. In this context, a reasonable assumption is to consider an interference-limited network, i.e., the noise is negligible. % It should be noted that the destination terminals are not part of the Poisson Point Processes that model the positions of the sources. % Each source has a buffer of infinite capacity for arriving packets. The total number of packets in the $i$th source of class $n$ at time $t$ is denoted by the queue length $Q_{i,n}(t)$. If the queue of a given source is not empty, then the source tries to transmit a packet with probability $p_n$ (medium access probability) following the first-come-first-serve % (FCFS) discipline and the transmission takes exactly one time slot. % % % Thus, each queue length $Q_{i,n}(t)$ forms a Markov chain (for more details, see \cite{dester2018}). % % % The packet is successfully transmitted when the signal-to-interference ratio (SIR) of the received signal is greater than a threshold $\theta_n$ (capture effect). In this case, the receiver sends an acknowledge through an error-free channel and the packet leaves the queue. Packets arrive at the queue according to an iid Bernoulli distribution of parameter $a_n$. % % % Chronologically, within each time slot, we have the transmission of packets, then the arrival of packets, and, at the end of the time slot, the displacement of the sources occurs to form a new and independent PPP realization. % The SIR experienced by the typical user from class $n$ is given by $\mathrm{SIR}_{i,n}= P_n h_{i,n} R_{i,n}^{-\alpha}/I$, where $\alpha>2$ is the path-loss exponent, the Rayleigh fading effect is represented by the fading coefficient $h_{i,n}$, which is an iid (with respect to time and users) exponentially distributed random variable of unit mean and remains constant during the time slot, and $I$ is the aggregate interference from other sources. It is known that $\mathrm{SIR}_{i,n}$ is an iid random variable with respect to time \cite{baccelli2010stochastic}. More specifically, the frequency band of bandwidth $B_0$ is shared by users of both classes. For Class 1 transmissions, this frequency band is divided into $M$ partitions (sub-bands) of bandwidth $B_0/M$, and each Class 1 user transmits over one randomly selected sub-band. Therefore, the density of sources for each partition becomes $\lambda_1/M$. On the other hand, users of Class 2 transmit over the whole available bandwidth $B_0$. % Note that front-end receiver filters of the first class destinations will capture $1/M$ of the power from transmissions of users of the second class and do not capture anything from other partitions of the first class. On the other hand, the second class destinations capture all the power coming from sources of both classes. The transmission rate $R_T$ for the first-class users is fixed and, as in \cite{jindal2008bandwidth}, the SIR threshold\footnote{The transmission successfully happens if and only if the SIR is greater than the threshold.} $\theta_1$ comes from the following equation: ${R_T = (B_0/M)\,\ln(1+\theta_1)}$. Then, ${\theta_1 = \euler^{M\eta_0}-1}$, where ${\eta_0 \triangleq R_T/B_0}$ is the spectral efficiency without partitioning the frequency band. Note that the transmission rate for the first class of users remains unchanged with bandwidth partitioning. % Both user classes use the same bandwidth and the goal is to find the optimum number of partitions $M$ for the first user class. % Table~\ref{tab:symbols} summarizes the notation. % \begin{table}[hbt] % \centering % \caption{Symbols and Definitions} % \begin{tabular}{ll} % \toprule % \textbf{Symbol} & \textbf{Definitions} \\ % \midrule % $\alpha\in(2,\infty)$ & Path loss exponent \\ % $\delta\in(0,1)$ & $\triangleq 2/\alpha$ \\ % $M \in \N$ & Number of partitions \\ % $\eta_0 \in \R_+$ & Spectral efficiency without bandwidth partitioning \\ % $\eta \in \R_+$ & $\triangleq M \eta_0 $, spectral efficiency of Class 1 \\ % $p_n\in[0,1]$ & Medium access probability \\ % $a_n\in[0,1]$ & Packet arrival rate \\ % $p_{s,n}\in[0,1]$ & Stationary transmission success probability \\ % $\theta_n\in \R_+$ & SIR threshold for successful transmission \\ % $D_n\in(1,\infty)$ & Average packet transmission delay \\ % $\overline{R}_n\in \R_+$& Mean source-destination separation distance \\ % $P_n\in \R_+$ & Transmit power \\ % $\lambda_n\in \R_+$ & User density of class $n$\\ % $\psi_n\in \R_+$ & $\triangleq 4\,\Gamma(1+\delta)\, % \Gamma(1-\delta)\,\overline{R}_n^2\, % \theta_n^{\delta}$~~(link quality)\\ % % $||\cdot||$ & Euclidean norm \\ % % $\ind_A(x)$ & indicator function\\ % $\Psi_n\in \R_+$ ~& $\triangleq \psi_n\lambda_n \left(\frac{D_n}{D_n-1}\right) \left(\frac{a_n}{1-a_n}\right)$~~(resource utilization)\\ % \bottomrule % \end{tabular} % \label{tab:symbols} % % \vspace*{-\baselineskip} % \end{table} % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{Stability Conditions and Stationary Analysis} \label{sec:stab} In this section, we derive conditions for stability and carry a stationary analysis of the network, which leads to expressions relating network parameters and performance metrics. These expressions are used in Section \ref{sec:opt} to formulate an optimization problem to maximize the network performance when bandwidth partitioning is employed. % In order to derive the stability conditions, we first determine the successful transmission probability of a typical user from class $n\in\{1,2\}$, given the \textit{effective density of active sources} $\lambda^\mathrm{eff}_k(t)$ for each class $k\in\{1,2\}$ in a time slot $t\in\Z_+$. % % % The effective density of active sources of a class is the PPP density of the sources with non-empty queues and allowed by the medium access control technique to transmit a packet. % % % This access control is modeled in this work by the medium access probability $p_n$. One can calculate the probability of successful transmission by deconditioning the general expression for the successful probability given in \cite[Eq.~(9)]{haenggi2009stochastic} for the distance $R$, when $R$ follows a Rayleigh distribution. This was done in \cite[Proposition~1]{dester2018}. Then, the successful transmission probability is given by From the proof of Proposition~\ref{prop:psk} in Eq.~\eqref{eq:P_SIR}, for two classes of users, we can write that% \begin{equation} \label{eq:P_SIR_M} \P(\mathrm{SIR}_{i,n}(t) > \theta_n) = \nu_n\!\left( P_1^\delta\,\lambda_\mathrm{eff}^{(1)}(t) + P_2^\delta\,\lambda_\mathrm{eff}^{(2)}(t) \right), \quad t\in\N, \end{equation} where the function ${\nu_n: \R_+ \longrightarrow \R_+}$ is defined as \begin{equation}\label{defNu} \nu_n(x) \triangleq \left(1+\frac{\psi_n}{P_n^\delta}x\right)^{-1}, \quad x\in\R_+, n\in\{1,2\}. \end{equation} A necessary and sufficient condition for the stability of the buffers for both user classes is given by the following proposition. % % A system is stable whenever the Markov chain describing the system admits a proper limit distribution when $t\to\infty$ \cite{szpankowski1994stability}. % Stability region refers to the set of arrival rates for which the system is stable. \begin{proposition} \label{prop:stability} The system network is stable if and only if $(a_1,a_2) \in \mathcal{E}_1 \cup \mathcal{E}_2$, where \begin{align*} \mathcal{E}_1 &= \left\lbrace (a_1,a_2)\in [0,1]^2 \mid a_1 < p_1\,\nu_1\!\left( P_1^\delta\tfrac{\lambda_1}{M}p_1 +\tfrac{P_2^\delta}{M^\delta}\lambda_2 p_2\right), a_2 < p_2\,\nu_2\!\left( P_1^\delta\lambda_1 \tfrac{a_1}{\widetilde{p}_{s,1}} + P_2^\delta\lambda_2 p_2 \right) \right\rbrace,\\ \mathcal{E}_2 &= \left\lbrace (a_1,a_2)\in [0,1]^2 \mid a_2 < p_2\,\nu_2\!\left( P_1^\delta\lambda_1 p_1 +P_2^\delta\lambda_2p_2\right), a_1 < p_1\,\nu_1\!\left(P_1^\delta \tfrac{\lambda_1}{M}p_1+\tfrac{P_2^\delta}{M^\delta} \lambda_2\tfrac{a_2}{\widetilde{p}_{s,2}} \right) \right\rbrace, \end{align*} where \begin{equation*} \widetilde{p}_{s,1} = \frac{1-\psi_1\tfrac{\lambda_1}{M}a_1} {1+\psi_1(\tfrac{P_2}{M P_1})^\delta\lambda_2p_2}, \qquad \widetilde{p}_{s,2} = \frac{1-\psi_2\lambda_2 a_2} {1+\psi_2(\tfrac{P_1}{P_2})^\delta\lambda_1 p_1}. \end{equation*} \end{proposition} % \begin{proof} Sufficient conditions for the network stability are obtained by using the concept of dominant network \cite[Section~2.1.2]{kompella2014stable}, which corresponds to a network almost identical to the original one, differing only on the fact that users of some classes of the dominant network transmit dummy packets when their queues are empty. If the dominant network is stable, then the original network is stable as well. % Let us analyze a dominant network where both classes transmit dummy packets with the correspondent medium access probability $p_1, p_2$. In this case, the effective density of active sources for the $n$th traffic class is $\lambda_\mathrm{eff}^{(n)} = \lambda_n\,p_n$ for all $t$, since we are assuming that sources transmit dummy packets when their queues are empty. % Note that we have partitioned the frequency band of the first class of users into $M$ sub-bands, i.e., the destinations of packets of this user class only receive signals within their bandwidth, i.e., signals from $1/M$ of the sources of the first user class and $1/M$ of the power of signals the sources of the second class of users. % In general, a sufficient condition for stability\footnote{The system is stable if the expected rate of incoming packets is smaller than the expected number of packets leaving the queue per time slot \cite{loynes1962stability}.} for both classes are A sufficient condition for stability is \begin{equation} a_n < p_n \, P(\mathrm{SIR}_{i,n} > \theta_n), \quad n\in\{1,2\}, \end{equation} by Loynes' theorem (Theorem~\ref{th:loynes}). Then, using Eq.~\eqref{eq:P_SIR_M}, sufficient conditions for stability for the first and the second classes are \begin{equation} \label{eq:stabClass1} a_1\theta_k)$, $k\in\{1,2\}$, is the Class $k$ coverage probability in the dominant network. Therefore, the stationary effective density of active sources from the first class is $\lambda_1^\mathrm{eff} = \lambda_1\,p_1\,\rho_1$. From Eq.~\eqref{eq:P_SIR_M}, we have the following fixed-point equation \begin{equation*} \widetilde{p}_{s,1} = p_1\,\nu_1\!\left[ P_1^\delta\tfrac{\lambda_1}{M}\tfrac{a_1}{\widetilde{p}_{s,1}} + \left(\tfrac{P_2}{M}\right)^\delta\!\lambda_2 p_2 \right], \end{equation*} which is easily solvable for $\widetilde{p}_{s,1}$. Then, we can use this value to derive a weaker stability condition for the second user class, i.e., $a_2 1$ and \begin{equation} \label{eq:defPsi} \Psi_n \triangleq \psi_n\lambda_n \left(\frac{D_n}{D_n-1}\right) \left(\frac{a_n}{1-a_n}\right) \ge 0, \qquad n\in\{1,2\}. \end{equation} % with $D_n\in(1,\infty)$, $n\in\{1,2\}$. \end{theorem} \begin{proof} If the system is stable, then the equations can be verified through Proposition~\ref{prop:stationary} and some manipulations. On the other hand, to prove that a set of parameters that satisfy the equations of the theorem implies stability, one can follow these steps: plug Eq.~\eqref{eq:theor02} into the inequalities of Proposition~\ref{prop:stability}; then use Eq.~\eqref{eq:theor01} and the fact that $\Psi_n > \psi_n\lambda_n\frac{a_n}{1-a_n}$ (since $\frac{D_n}{D_n-1}>1$) to show that the inequalities of Proposition~\ref{prop:stability} are satisfied. Then, the system is stable. \end{proof} \begin{remark} \label{rmk:Th.Stb.} From Theorem~\ref{th:identity}, $\frac{\Psi_1}{M} + \Psi_2 \le 1$. This follows directly from equations \eqref{eq:theor01} and \eqref{eq:defPsi}. % Furthermore, for fixed $M$, the quantity $\Psi_1$ decreases if $\Psi_2$ increases. For example, let the parameters $M,\psi_1,\psi_2,a_1,a_2,D_1$, and $D_2$ be fixed; if we wish to increase the density of users $\lambda_1$ of Class 1 and maintain the aforementioned parameters constant, we necessarily need to decrease the density $\lambda_2$ of Class 2. % Note also that $\Psi_n$ increases with $a_n$ and decreases with $D_n$, such that $\Psi_n$ can be seen as a comprehensive measure of the performance of users of class $n \in \{1,2\}$. \end{remark} \begin{remark} \label{rmk:FreePowers} In a network where we can freely choose the transmit power ratio $P_1/P_2$, at first we only need to satisfy Eq.~\eqref{eq:theor01} when specifying system parameters ($M,\psi_1,\psi_2,\lambda_1$,$\lambda_2$) and performance parameters ($a_1,a_2,D_1$,$D_2$). After that, we use Eq.~\eqref{eq:theor02} to specify the transmit power levels $P_1$ and $P_2$. \end{remark} Theorem~\ref{th:identity} considers $p_1=p_2=1$, which minimizes the mean delay for both classes (see Proposition~\ref{prop:stationary}) and maximizes the stability region (see Remark~\ref{rem:stab}). Also, if $M=1$, we recover the result in Proposition~\ref{prop:identity_1}, i.e., $\sum_n \Psi_n = 1$. It is worth noticing the simple form through which the performance parameters $a_1,a_2,D_1$, and $D_2$ of both user classes and system stability are related. % \textcolor{blue}{If the system does not have constraints regarding the ratio of the transmit powers, we may only use equation \eqref{eq:theor01} of Theorem~\ref{th:identity}, since it is always possible to satisfy equation \eqref{eq:theor02} in this case. Precisamos explicar melhor essa ultima sentenca} % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{Optimum Bandwidth Partition} \label{sec:opt} Let us suppose we want to maximize the performance of Class 1 users when bandwidth partitioning is employed. More specifically, for given fixed arrival rate $a_2$ and required mean delay $D_2$ of the second user class (i.e., fixed $\Psi_2$), we are interested in maximizing the quantity $\frac{\Psi_1}{\psi_1\lambda_1}=\frac{D_1}{D_1-1}\frac{a_1}{1-a_1}$, when $p_1=p_2=1$ (see Remark~\ref{rem:stab}), by adjusting the number of partitions $M$. Note that, if we set a maximum tolerable mean delay $D_1$, this choice of optimization leads to the maximum arrival rate $a_1$ admissible. On the other hand, if we fix $a_1$, this choice of optimization leads to the minimum $D_1$ achievable. From Eq. \eqref{eq:theor01} of Theorem~\ref{th:identity} and recalling that $\psi_1 = 4\,\Gamma(1+\delta) \Gamma(1-\delta) \overline{R}_1^2 \theta_1^\delta$, where $\theta_1 = \euler^{M\eta_0} - 1$, one can show with simple manipulations of the ratio $\frac{\Psi_1}{\psi_1\lambda_1}$ that maximizing the quantity $\frac{D_1}{D_1-1}\frac{a_1}{1-a_1}$ is equivalent to the following optimization problem: % \begin{equation} \label{eq:opt_M} M^* = \argmax_{M\in\N} \dfrac{M\,(\euler^{M\eta_0}-1)^{-\delta}} {1+(M^{1-\delta}-1)\,\Psi_2}. \end{equation} % Our interest lies in finding the optimum number of bandwidth partitions $M^*$ for the first user class. After finding the optimum number of partitions $M^*$, it is necessary to adjust the transmit power levels to satisfy Theorem~\ref{th:identity}. In this sense, we are also choosing the optimum transmit power ratio $P_1/P_2$ (see Remark~\ref{rmk:FreePowers}). Let us now relax the constraint that $M$ must be integer to find a closed form equation for the optimization problem \eqref{eq:opt_M}, analogously to \cite{jindal2008bandwidth}. Note that after the bandwidth partition, the spectral efficiency is given by $\eta = M\eta_0$. Now, without the constraint that $M$ must be integer, we have a new goal, which is to find the optimum spectral efficiency $\eta^*$. An equivalent relaxed problem of \eqref{eq:opt_M} is \begin{equation} \label{eq:opt_eta} \eta^* = \argmin_{\eta\in\R_+} \left(\euler^\eta-1\right)^\delta\, \left( \dfrac{1}{\eta} + \dfrac{\beta}{\eta^\delta}\right), \end{equation} where \begin{equation} \beta \triangleq \frac{\Psi_2}{1-\Psi_2}\eta_0^{-(1-\delta)} \ge 0. \end{equation} This new formulation can be obtained by taking the multiplicative inverse of the objective function in \eqref{eq:opt_M}, followed by some manipulations. The following theorem guarantees the existence and uniqueness of the optimum spectral efficiency and shows how to find it. \begin{theorem} \label{th:optimum_eta} The optimum spectral efficiency $\eta^*$ is given by the unique positive solution of the following equation \begin{equation}\label{eq:Theor02} % \dfrac{h(\eta^*) - \delta}{1 - h(\eta^*)} = % \delta\,\beta\,{\eta^*}^{1-\delta}, \big(1-h(\eta^*)\big) \big(1+\delta\,\beta\,{\eta^*}^{1-\delta}\big) = 1-\delta, \end{equation} where $h(\eta) \triangleq (1-\euler^{-\eta})/\eta$. Furthermore, $\eta^*$ decreases monotonically with respect to $\beta$. \end{theorem} \begin{proof} Let us prove that the objective function \eqref{eq:opt_eta} is strictly convex in the region of interest. Since the sum of two strictly convex functions is strictly convex, then it is enough to show that ${[(\euler^\eta-1)/\eta]^\delta}$ and ${(\euler^\eta-1)^\delta/\eta}$ are convex with respect to $\eta > 0$. Throughout the proof we need the inequalities \begin{equation} \label{eq:aux_ineq} 0 < \euler^{-\eta} < h(\eta)^2 < h(\eta) < 1, \end{equation} which are valid when $\eta > 0$, and are easily proved using series expansion. First, let us show that $ {\frac{\partial^2}{\partial \eta^2} \left(\frac{\euler^\eta-1} {\eta}\right)^\delta > 0}. $ For $\eta > 0$ and $\delta > 0$ and after taking the derivatives with respect to $\eta$, this inequality can be written as $ {\delta\left( 1 - h(\eta) \right)^2 > \euler^{-\eta} - h(\eta)^2}, $ which is true by \eqref{eq:aux_ineq}. Therefore, the function ${[(\euler^\eta-1)/\eta]^\delta}$ is strictly convex, since its second derivative is positive. Now, let us show that $ \frac{\partial^2}{\partial \eta^2} \frac{(\euler^\eta-1)^\delta}{\eta} > 0. $ Again, for $\eta > 0$ and $\delta > 0$, this inequality can be written as \begin{equation}\label{eq:der2eq} \delta^2 - \left(2\,h(\eta) + \euler^{-\eta} \right)\,\delta + 2\,h(\eta)^2 > 0. \end{equation} From \eqref{eq:aux_ineq}, the inequality \eqref{eq:der2eq} is satisfied if $\delta \notin [1,2]$. Since $\alpha >2$, then $\delta\in(0,1)$, and this is enough to prove strict convexity for the function ${(\euler^\eta-1)^\delta/\eta}$ as well. Therefore, the function to be optimized is strictly convex and differentiable, i.e., if there exists a point where the derivative vanishes, then this point is unique and it is the global minimum. It is easy to verify that this strictly convex function is arbitrarily large when $\eta$ approaches 0 or $\infty$, then the global minimum exists and $\eta^*\in(0,\infty)$. Now, manipulating the equation \begin{equation*} \frac{\partial}{\partial\eta} \left[ \left( 1 + \beta\,\eta^{1-\delta}\right) \frac{\left(\euler^\eta-1\right)^\delta}{\eta}\right] = 0 \end{equation*} we obtain \eqref{eq:Theor02}, concluding the proof. The proof that $\eta^*$ decreases monotonically with respect to $\beta$ is immediate, since $h$ is a monotonically decreasing function. \end{proof} Figure~\ref{fig:optimum_partition} shows the optimum spectral efficiency $\eta^*$ of the first class as a function of $\beta$ for some values of the path loss exponent $\alpha$. \begin{figure}[!t] \centering \includegraphics[]{./Figures/Ch7_optimum_partition.pdf} \caption{Optimum spectral efficiency $\eta^*$ of the first class as a function of the parameter $\beta$, which is an increasing function of the parameter $\Psi_2$ of the second user class.} \label{fig:optimum_partition} \end{figure} Recall that $\beta$ is an increasing function of $\Psi_2$, which, in turn, is a comprehensive measure of the performance requirement of Class 2 users. As expected from the previous discussion, $\eta^*$ decreases as the performance requirement of the second class becomes more stringent, i.e., when $\Psi_2$ increases. This behavior of $\eta^*$ can be explained by recalling that higher spectral efficiency requires higher SIR threshold $\theta_1$, causing stronger interference to users of the second class. % This increased interference reduces the maximum admissible rate $a_2$ and increases the minimum achievable delay $D_2$. Since $\Psi_2$ is kept fixed in this optimization problem, $\eta^*$ must be limited. In fact, from Eq. \eqref{eq:theor01} we can see that if $\Psi_2$ increases, then $\Psi_1$ must be reduced in order to keep the system stable. % The propagation environment also affects the optimum spectrum efficiency. Environments with larger path loss exponent $\alpha$ reduce the interference among users, improving the channel quality and allowing for the use of higher spectral efficiency transmission schemes. Finally, note that without the second user class ($\Psi_2 = 0$), we have $\beta=0$ and the equation for $\eta^*$ in Theorem~\ref{th:optimum_eta} has a closed form solution, which is \begin{equation}\label{eq:opt_eff} \eta^*(\beta=0) = \tfrac{\alpha}{2} + W_0\big(-\tfrac{\alpha}{2}\,\euler^{-\alpha/2} \big), \end{equation} where $W_0$ is the principal branch of the Lambert-$W$ function, which is defined as the solution on $[-1,\infty)$ of the equation $W_0(x)\,\euler^{W_0(x)} = x$, for $x \ge -1/\euler$. % The above result is consistent with \cite[Theorem~2]{jindal2008bandwidth}, where the optimization is over the traffic density achievable and the system consists of a single user class. Expression \eqref{eq:opt_eff} was also obtained by Haenggi in \cite{Haenggi_Lambert}. As discussed in \cite{jindal2008bandwidth}, the quantity $\delta \eta^*(0) \to 1$ as $\delta \to 0$ (i.e. $\alpha\to\infty$), which means that $\eta^*(0)$ grows asymptotically as $\alpha/2$ when the path loss exponent $\alpha$ tends to infinity. % However, for a given $\beta>0$, the same does not hold true. We can show that $\delta \eta^*(\beta) \to 0$ as $\delta \to 0$. In fact, we can be more precise and show through \eqref{eq:Theor02} that if $\beta>0$, then $\sqrt{\delta} \eta^* \to 1/\sqrt{\beta}$ as $\delta\to 0$, which means that $\eta^*$ grows asymptotically as $\sqrt{\alpha/2\beta}$ when the path loss exponent $\alpha$ tends to infinity. % Thus, by introducing interference from another class in the system ($\beta>0$, i.e. $\Psi_2>0$), we change the asymptotic behavior of the optimal spectral efficiency from linear growth to squared root growth with respect to the path loss exponent $\alpha$. This asymptotic result can be seen in Figure~\ref{fig:optimum_eta_delta}, where we plot the optimal spectral efficiency $\eta^*$ adjusted by the multiplicative factor $\sqrt{\beta\delta}$, which corresponds to the reciprocal of the asymptote as $\delta$ tends to $0$ (that is why all curves converge to $1$ at $\delta = 0$). % Another simple asymptote of $\eta^*$ is given by $2(1-\delta)/(1+\beta)$ when $\delta$ tends to $1$ (i.e. $\alpha\to2$). \begin{figure}[!t] \centering \includegraphics[]{./Figures/Ch7_optimum_eta_delta.pdf} \caption{Optimum spectral efficiency $\eta^*$ of the first class adjusted by the multiplicative factor $\sqrt{\beta\delta}$ as a function of the parameter $\delta = 2/\alpha$ for several values of $\beta$.} \label{fig:optimum_eta_delta} \end{figure} Also, we can obtain a generalization of \eqref{eq:opt_eff} when the performance requirements of Class 2 is small (i.e., $\Psi_2 \approx 0$), which is equivalent to $\beta \approx 0$. Then, we have the following first order expansion for the optimum spectral efficiency \begin{align} \label{eq:eta_expansion} \frac{\eta^*\!(\beta)}{\eta^*\!(0)} = 1 - \frac{(1-\delta)\eta^*\!(0)^{1-\delta}}{\eta^*\!(0)+1-\frac{1}{\delta}} \beta + \cal{O}(\beta^2), \end{align} as $\beta$ tends to zero, $\eta^*\!(0)$ is given by Eq.~\eqref{eq:opt_eff} and $\cal{O}$ is the big O notation as defined in Definition~\ref{def:landau}. % \footnote{ % We say $f(x) = \cal{O}(g(x))$ as $x\to a$, if $\limsup\limits_{x\to a} \frac{|f(x)|}{g(x)} < \infty$. % }. % The above expression is found using implicit differentiation on \eqref{eq:opt_eta} and some manipulations. We can verify through \eqref{eq:eta_expansion} that the optimal spectral efficiency has a steeper (relative) decay for larger values of the path loss exponent $\alpha$, when $\alpha > 2.48$. % This means that the presence of interference from another class has a larger relative impact on the optimal spectral efficiency when $\alpha$ is large. % \begin{figure}[!t] % \centering % \input{./Plots/optimum_partition_norm.tex} % \caption{Optimum spectral efficiency $\eta^*$ of the first class as a function of the parameter $\beta$, which is an increasing function of the parameter $\Psi_2$ of the second user class.} % \label{fig:optimum_partition_norm} % \end{figure} % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % \subsection{Applications} \label{sec:num} In this section, we illustrate the analytical results through some numerical examples. Let the first and second user classes represent the opportunist and the main users of a given bandwidth $B_0$, respectively. % The opportunist users access the same frequency band as the main users and we perform bandwidth partitioning for the opportunistic access in order to improve the performance of the opportunist users for a given performance of the main users, which is equivalent to fixing the quantity $\Psi_2$. From now on, we shall use the numerical values in Table~\ref{tab:user_classes} for the numerical examples. % \begin{table}[!ht] \centering \caption{Parameter values.} \begin{tabular}{ c|l } \hline Parameters & Values \\ \hline\hline $\lambda_1$ & $\SI{0.005}{/m^2}$ \\ $\lambda_2$ & $\SI{0.002}{/m^2}$ \\ $\overline{R}_1$ & $\SI{10}{m}$ \\ $\overline{R}_2$ & $\SI{20}{m}$ \\ $\eta_0$ & $\SI{0.2}{bits/s/\hertz}$ \\ $p_1=p_2$ & $1$ \\ \hline \end{tabular} \label{tab:user_classes} \end{table} % We begin our analysis by comparing the stability regions with and without bandwidth partition. % Let the system parameters be $\lambda_1 = \SI{0.005}{/m^2}$, $\lambda_2 = \SI{0.002}{/m^2}$, $\overline{R}_1 = \SI{10}{m}$, $\overline{R}_2 = \SI{20}{m}$ and the (initial) spectral efficiency $\eta_0 = 0.2\,\ln(2)\,\si{nats/s/\hertz}$ for both classes. % Our approach to determine the stability regions is the following: we begin by using Theorem~\ref{th:optimum_eta} to find the optimum spectral efficiency $\eta^*$ of the first class for a given value of ${\Psi_2\in(0,1)}$ (see Remark~\ref{rmk:Th.Stb.}); then, we choose the number of sub-bands $M$ as the closest integer% \footnote{This is a functional thumb rule to find the optimum number of partitions $M^*$. However, the ideal approach is to verify through the objective function if the best is to round up or round down. In our plots, we used the thumb rule, since there is no visual difference.} to $\eta^*/\eta_0$ (it is worth remembering that $\Psi_2$ is fixed, so the performance requirements of Class 2 are still satisfied); finally, we use Theorem~\ref{th:identity} along with the fact that $p_1=p_2=1$ maximizes the stability region to obtain the union of the stability regions for all the transmit power ratios $P_1/P_2$ and for all medium access probabilities $p_1$ and $p_2$. The results are shown in Fig.~\ref{fig:optimum_stab_region} for $\alpha = 2.5$ and $4.5$. % \begin{figure}[!t] \centering \includegraphics[]{./Figures/Ch7_optimum_stab_region.pdf}% \caption{Stability regions of the network for different values of path loss exponent $\alpha$. The arrival rates $a_1$ and $a_2$ are the opportunist and main users, respectively; full and dashed curves correspond to the boundaries of the stability region with and without bandwidth partition, respectively. Parameter values are shown in Table~\ref{tab:user_classes}.} \label{fig:optimum_stab_region} \end{figure} % We can see that the optimization of the spectral efficiency of the first user class (through bandwidth partitioning) expands the stability region. Note that when the path loss exponent $\alpha$ is small (close to 2), the optimization does not improve significantly the system performance. % Another observation is that if the traffic from main users (Class 2) is high, i.e., it is responsible for the majority of the interference, then it makes practically no difference to perform bandwidth partitioning in the opportunist class (Class 1). On the other hand, as the traffic from opportunist users increases, the impact of the optimization also increases % \textcolor{red}{(esse melhor desempenho no canto direito inferior da região ($a_1$ alto e $a_2$ baixo) tem a ver apenas com a baixa interferência causada pelos usuários 2? Acho que poderíamos explorar esse melhor desempenho nessa região, pois é a região de interesse da ideia da partição.)}. Let us now consider another example to analyze the effects of bandwidth partitioning on the mean delay. % For this, we set the initial spectral efficiency for the opportunist class $\eta_0 = 0.2\,\ln(2)\,\si{nats/s/\hertz}$, the access probabilities $p_1=p_2=1$. % We then determine the mean delay $D_1$ of the opportunist class as a function of its arrival rate of packets $a_1$, for $\Psi_2 = 0.25, \, 0.5, \, 0.75$. Note that $\Psi_2$ can be seen here as a measure of the performance required by the main class (see Remark~\ref{rmk:Th.Stb.}). The procedure to find $M^*$ is analogous to the one used in the last example. The results are shown in Fig.~\ref{fig:optimum_delay}. % \begin{figure}[!t] \centering \includegraphics[]{./Figures/Ch7_optimum_delay.pdf}% \caption{Opportunist (first class) mean delay $D_1$ as a function of opportunist arrival rate $a_1$, for different values of the main user performance requirement $\Psi_2$. Full and dashed curves correspond to the cases with and without bandwidth partition, respectively. Parameter values are shown in Table~\ref{tab:user_classes}.} \label{fig:optimum_delay} \end{figure} % When the performance requirements of the main users are modest (small $\Psi_2$), bandwidth partitioning allows for a more efficient use of the channel by opportunist users, i.e., a larger arrival rate $a_1$ is possible, or a smaller mean delay is achieved. On the other hand, and as expected from the previous analysis, we can also see that when the requirements of main users become more stringent (larger $\Psi_2$), then less noticeable is the improvement from the bandwidth partitioning. % \begin{figure}[!t] \centering \includegraphics[width=0.7\columnwidth]{./Figures/Ch7_a_ratio.pdf}% \caption{Maximum throughput relative improvement of Class 1 after band partition. Parameter values are shown in Table~\ref{tab:user_classes}.} \label{fig:a_ratio} \end{figure} % \begin{figure}[!t] % \centering % \input{./Plots/a_ratio_v2.tex} % \caption{Maximum throughput improvement of Class 1 after band partition. Parameter values are shown in Table~\ref{tab:user_classes}. \textcolor{red}{Daria para colocar ``curvas de nível'', relativas a alguns valores de ganho?}} % \label{fig:a_ratio_v2} % \end{figure} Figure~\ref{fig:a_ratio} shows the relative improvement of the maximum throughput after band partition of Class 1 relative to the maximum throughput before band partition, for all possible values of $\Psi_2$ and $\delta$. For example, in the yellow region, we have almost doubled the throughput ($\approx 100\%$ improvement) after implementing the optimum number of band partitions. As we have previously seen through figures \ref{fig:optimum_stab_region} and \ref{fig:optimum_delay}, the bandwidth partitioning strategy of the first user class improves significantly the performance of Class 1 users as the Class 2 performance requirements $\Psi_2$ is small and the path loss exponent $\alpha$ is large (which is equivalent to small $\delta = 2/\alpha$). On the other hand, when the path loss exponent $\alpha$ is small or the performance requirements of the main (Class 2) users $\Psi_2$ is stringent, there is almost no gain in performing bandwidth partitioning of the opportunistic class (Class 1). % % % % % % % % % % % % % % \section{Summary} \label{sec:summ_P2_03} In this chapter, we derived necessary and sufficient conditions for stability in a network with $N$ user classes; we also provided simple closed form expressions for the packet success probability and mean delay. The advantage of using this model as a base to model other network effects is its analytic tractability. % As an example, we were able to derive simple conditions to verify the stability of an interference-limited network with undetermined transmit powers using Corollary~\ref{cor:stab}. % We also solved (analytically and in closed form) two optimization problems regarding the minimization of the delays in a network and maximization of the total throughput per unit of area. % An interesting insight from the optimization problems is that the best solution to maximize the throughput of a channel is not necessarily using solely the user class with the best link quality, i.e., a mix with other user classes may result in better use of the channel. % % All in all, this paper provides a simple way to evaluate the existing trade-offs involved in the design of wireless networks when different classes of nodes co-exist. Further, we showed that, under certain conditions, bandwidth partitioning can significantly improve the performance of the network (under stability condition) by means of higher allowable arrival rates or lower mean delays. This is the first time the problem of bandwidth partitioning is treated analytically for two interacting user classes, for which arriving packets are queued.src/geometry/2d-geometry/union-of-circles.tex \lstinputlisting{src/geometry/2d-geometry/union-of-circles.cpp} 1-10 \section{Administration manual}\label{sec:adminmanual} The administrator\footnote{In the application developed for test the admin credentials are: \code{username: admin}, \code{passoword: password}} has more options in the main menu: \begin{verbatim} Select an action: 1) Browse strategies 2) Upload a new strategy 3) Browse users 4) Create user 5) Browse data sources 0) Log-Out Enter number: \end{verbatim} \subsection{Browse users} When you select the ``Browse users'' menu entry, the application will ask for a username. You can write it partially or leave the field empty to search for all users. After that, you can select a user to \standout{delete} it. \subsection{Browse data source} When you select the ``Browse data sources'' menu entry, the application will list all the data sources available in the system. You can select one of them in order to change its configuration: \begin{verbatim} Select a Data Source: 1) COINBASE 2) BINANCE 0) Go back Enter number: 1 COINBASE | select an action: 1) View details 2) Disable Data Source 3) Browse markets 4) Delete data 0) Go back Enter number: \end{verbatim} You can \standout{disable} the source entirely (users will not be able to select it anymore to run strategies); \standout{view and edit the markets} provided by the data source; \standout{delete} the downloaded data for all the markets of the data source. \subsubsection{Browse markets} When you select the ``Browse markets'' menu entry, the application will ask to search by market name. Then you can select a market and edit its configuration: \begin{verbatim} COINBASE:BTC/EUR | Select an action: 1) View details 2) Change granularity 3) Disable selectability 4) Enable data sync 5) Delete data 0) Go back Enter number: \end{verbatim} You can \standout{view the details} of the market; change the minimum \standout{granularity} that the user can select when running a strategy on the market; \standout{enable/disable} the market for the selection by the users when running a strategy and the synchronization of data with the remote data source; \standout{delete} the downloaded data of the market. \subsubsection{Delete data} When you select the ``Delete data'' menu entry, the application will ask for a date. If you do not specify a date, all the data will be deleted. Otherwise, only the data older that the specified date will be deleted. The date must be in the format \code{yyyy-mm-dd hh:mm} \exgratia{\code{2017-03-22 12:15}}. @article{carrat_decreased_2017, abstract = {Genetic variants near ARAP1 (CENTD2) and STARD10 influence type 2 diabetes (T2D) risk. The risk alleles impair glucose-induced insulin secretion and, paradoxically but characteristically, are associated with decreased proinsulin:insulin ratios, indicating improved proinsulin conversion. Neither the identity of the causal variants nor the gene(s) through which risk is conferred have been firmly established. Whereas ARAP1 encodes a GTPase activating protein, STARD10 is a member of the steroidogenic acute regulatory protein (StAR)-related lipid transfer protein family. By integrating genetic fine-mapping and epigenomic annotation data and performing promoter-reporter and chromatin conformational capture (3C) studies in β cell lines, we localize the causal variant(s) at this locus to a 5 kb region that overlaps a stretch-enhancer active in islets. This region contains several highly correlated T2D-risk variants, including the rs140130268 indel. Expression QTL analysis of islet transcriptomes from three independent subject groups demonstrated that T2D-risk allele carriers displayed reduced levels of STARD10 mRNA, with no concomitant change in ARAP1 mRNA levels. Correspondingly, β-cell-selective deletion of StarD10 in mice led to impaired glucose-stimulated Ca2+ dynamics and insulin secretion and recapitulated the pattern of improved proinsulin processing observed at the human GWAS signal. Conversely, overexpression of StarD10 in the adult β cell improved glucose tolerance in high fat-fed animals. In contrast, manipulation of Arap1 in β cells had no impact on insulin secretion or proinsulin conversion in mice. This convergence of human and murine data provides compelling evidence that the T2D risk associated with variation at this locus is mediated through reduction in STARD10 expression in the β cell.}, author = {. and , , , . and , , Afshan and Falchi, Mario and Thurner, Matthias and Canouil, Mickaël and Pattou, Francois and Leclerc, Isabelle and Pullen, . and Cane, . and Prabhala, Priyanka and Greenwald, William and Schulte, Anke and Marchetti, Piero and Ibberson, Mark and MacDonald, . and , , . and Froguel, , , , .}, copyright = {All rights reserved}, doi = {10.1016/j.ajhg.2017.01.011}, issn = {1537-6605}, journal = {American Journal of Human Genetics}, keywords = {Alleles, Humans, GWAS, Adaptor Proteins, Signal Transducing, Animals, ARAP1, Carrier Proteins, Cloning, Molecular, diabetes, Diabetes Mellitus, Type 2, Gene Expression Regulation, Genetic Variation, genetics, GTPase-Activating Proteins, Homeostasis, insulin, Insulin, Insulin-Secreting Cells, islet, Liver, Mice, mouse, Phosphoproteins, Proinsulin, Quantitative Trait Loci, secretion, STARD10, Transcriptome}, language = {eng}, month = {February}, number = {2}, pages = {238--256}, pmcid = {PMC5294761}, pmid = {28132686}, title = {Decreased STARD10 Expression Is Associated with Defective Insulin Secretion in Humans and Mice}, volume = {100}, year = {2017} } \section{Mathematical Appendix}\label{appx: b} To simplify notation for \autoref{appx: b}, I denote the continuation value at budget $b$ by: \begin{equation*} \begin{aligned} &K_{b}:=\alpha \,\mathbb{E}_{\theta}\left[V_w(\theta', b)\right] \end{aligned} \end{equation*} \subsection{Proof for \autoref{prop:piecewiseV} and \autoref{cor:optpolicy}} \begin{proof} Fix some $b\in\mathcal{B}_w$ and, starting from \autoref{eq:full bellman}, consider the following: \begin{equation*} \begin{aligned} V_w(\theta,b) \;&=\;\max\left\{\,\overline{\mu} \, u(\theta) +\alpha \,\mathbb{E}_\theta \Big[V_w(\theta', b-1)\Big]\,,\; \alpha\,\mathbb{E}_\theta \Big[ V_w(\theta', b)\Big]\,\right\}\\ &=\; \max\left\{\,\overline{\mu} \, u(\theta) + K_{b-1} \,,\; K_b \,\right\}\\ &=\; K_{b-1} + \max\left\{\,\overline{\mu} \, u(\theta) \,,\; K_b - K_{b-1}\,\right\} \end{aligned} \end{equation*} First, note that the difference between any two consecutive continuation values $K_b$ and $K_{b-1}$ must necesarily lie between 0 and $\overline{\mu}u(1)$. This is true since the value function denotes the expected lifetime sum of discounted payoffs, and an additional right-swipe can provide an agent with, at most, an additional expected payoff of $\overline{\mu}u(1)$ and, at least, an additional payoff of $0$. Furthermore, since $u(\theta)$ is, by assumption, continuous and increasing over $\Theta$ (and we assume $\overline\mu>0$ to prune out degenerate equilibria), then, by the Intermediate Value Theorem, there exists a unique root, $\widetilde\omega_b$, satisfying: \begin{equation*} \overline\mu u(\widetilde\omega_b) = K_b-K_{b-1} \end{equation*} Consider now two cases. First, if $\theta\leq\widetilde\omega_b$, then: \begin{equation*} \begin{aligned} V_w(\theta,b) \;&=\; K_{b-1} + \max\left\{\,\overline{\mu} \, u(\theta) \,,\; K_b - K_{b-1}\,\right\}\\ &=\; K_{b-1} + K_b - K_{b-1}\\ &=\; K_b. \end{aligned} \end{equation*} Analogously, if $\theta\leq\widetilde\omega_b$, then: \begin{equation*} V_w(\theta,b) = \overline{\mu} \, u(\theta) + K_{b-1}. \end{equation*} Thus, by considering the above function over the intervals $[0, \widetilde\omega_b]$ and $[\widetilde\omega_b, 1]$ separately, and substituting back the expressions for $K_b, K_{b-1}$, we conclude that: \begin{equation*} \begin{split} V_w(\theta,b)=\begin{cases} \overline\mu u(\theta) +\alpha \,\mathbb{E}_{\theta}\Big[V_w(\theta', b-1)\Big],& \theta \geq \widetilde \omega_b \\[10pt] \alpha \,\mathbb{E}_{\theta}\Big[V_w(\theta', b)\Big],& \theta\leq\widetilde \omega_b \end{cases} \end{split} \end{equation*} Furthermore, \autoref{cor:optpolicy} follows trivially from the above by considering a cutoff policy over the above intervals such that $V_w(\theta,b)$ is attained. \begin{comment} When a woman with budget $b$ is presented a candidate with attractiveness $\theta \geq \widetilde\omega_b$ and she swipes right, her expected lifetime sum of discounted payoffs is: \begin{equation*} \begin{split} \overline\mu u(\theta) +\alpha \,\mathbb{E}_{\theta}\Big[V_w(\theta', b-1)\Big]\\ = V_w(\theta,b) \end{split} \end{equation*} Alternatively, when presented a candidate with attractiveness $\theta<\widetilde\omega_b$ and she swipes left: \begin{equation*} \begin{split} \alpha \,\mathbb{E}_{\theta}\Big[V_w(\theta', b)\Big]\\ = V_w(\theta,b) \end{split} \end{equation*} \end{comment} \end{proof} \subsection{Proof for \autoref{prop:recurrence relation}} \begin{proof} Fix some $b\in\mathcal{B}_w$ and consider the result presented by \autoref{prop:piecewiseV}, which guarantees the existence and uniqueness of some $\widetilde \omega_b$ satisfying: \begin{align} \begin{split}\label{eq:A.1} V_w(\theta, b)&=\begin{cases} \overline\mu u(\theta) + K_{b-1},& \theta> \widetilde \omega_b \\ K_b,& \theta\leq\widetilde \omega_b \end{cases} \end{split}\\ \begin{split}\label{eq:A.2} \overline\mu u(\widetilde\omega_b) &= K_b-K_{b-1} \end{split} \end{align} Starting out with \autoref{eq:A.2} and expanding out the expectation operator, we can use \eqref{eq:A.1} to substitute in the piecewise definitions of $V_w(\theta,b)$ over the appropriate intervals: \begin{equation}\label{eq:A.3} \begin{split} \overline\mu u(\widetilde\omega_b) &= K_b-K_{b-1}\\ &= \alpha \,\int^1_0 V_w(\theta',b)-V_w(\theta',b-1)\,dF_m(\theta')\\ &=\alpha \int^{\widetilde\omega_b}_0 K_b\,dF_m(\theta') \;+\; \alpha \int^1_{\widetilde\omega_b}\,\overline\mu u(\theta') + K_{b-1}\,dF_m(\theta')\\ & \quad -\,\alpha \int^{\widetilde\omega_{b-1}}_0 K_{b-1}\,dF_m(\theta') \;-\; \alpha \int^1_{\widetilde\omega_{b-1}} \overline\mu u(\theta') + K_{b-2}\,dF_m(\theta') \end{split} \end{equation} Furthermore, \autoref{eq:A.2} implies that: $$ \overline\mu u(\widetilde\omega_b) +K_{b-1}= K_b $$ $$ \overline\mu u(\widetilde\omega_{b-1}) +K_{b-2}=K_{b-1} $$ Then, by substituting these expressions into \eqref{eq:A.3}, we arrive at \eqref{eq:A.4}: \begin{equation}\label{eq:A.4} \begin{split} \overline\mu u(\widetilde\omega_b) &=\alpha \int^{\widetilde\omega_b}_0 \overline\mu u(\widetilde\omega_b) +K_{b-1}\,dF_m(\theta') \;+\; \alpha \int^1_{\widetilde\omega_b} \,\overline\mu u(\theta') + K_{b-1}\,dF_m(\theta')\\ & \quad -\,\alpha \int^{\widetilde\omega_{b-1}}_0 K_{b-1}\,dF_m(\theta') \;-\; \alpha \int^1_{\widetilde\omega_{b-1}} \overline\mu u(\theta') + K_{b-1}-\overline\mu u(\widetilde\omega_{b-1})\,dF_m(\theta') \end{split} \end{equation} With some algebra, this simplifies down to the recurrence relation in \autoref{eq:recurrence relation}: \begin{equation} u(\widetilde\omega_b)=\alpha u(\widetilde\omega_b)F_m(\widetilde\omega_b) \;+\; \alpha u(\widetilde\omega_{b-1})\Big[1 - F_m(\widetilde\omega_{b-1})\Big] \;+\; \alpha\int^{\widetilde\omega_{b-1}}_{\widetilde\omega_b} u(\theta') \,dF_m(\theta') \end{equation} Furthermore, to obtain the initial condition for the above, note that the right-swiping budget constraint imposes $V_w(\theta,0)=0, \forall b\in \mathcal{B}_w$. Then, \eqref{eq:A.1} and \eqref{eq:A.2} simplify to: \begin{align} \begin{split}\label{eq:A.5} V_w(\theta, 1)&=\begin{cases} \overline\mu u(\theta),& \theta> \widetilde \omega_1 \\ K_1,& \theta\leq\widetilde \omega_1 \end{cases} \end{split}\\ \begin{split}\label{eq:A.6} \overline\mu u(\widetilde\omega_1) &= K_1 \end{split} \end{align} Beginning with \autoref{eq:A.6}, we simplify until arriving at \autoref{eq:initial condition}: \begin{equation*} \begin{split} \overline\mu u(\widetilde\omega_1) &= \alpha \, \mathbb{E}_\theta\Big[\,V_w(\theta',1)\,\Big]\\ &= \alpha \,\int^{\widetilde\omega_1}_0\,K_1\,dF_m(\theta') + \alpha \,\int_{\widetilde\omega_1}^1 \overline\mu u(\theta')\,dF_m(\theta')\\ &= \alpha \,\int^{\widetilde\omega_1}_0 \overline\mu u(\widetilde\omega_1)\,dF_m(\theta') + \alpha \,\int_{\widetilde\omega_1}^1 \overline\mu u(\theta')\,dF_m(\theta')\\ &= \alpha \overline\mu u(\widetilde\omega_1)F_m(\widetilde\omega_1) + \alpha \,\int_{\widetilde\omega_1}^1 \overline\mu u(\theta')\,dF_m(\theta')\\ \implies u(\widetilde\omega_1) &= \alpha u(\widetilde\omega_1)F_m(\widetilde\omega_1) + \alpha \,\int_{\widetilde\omega_1}^1 u(\theta')\,dF_m(\theta') \end{split} \end{equation*} %To conclude the proof, note that the existence and uniqueness of some $\widetilde\omega_b$ that satisfies \ref{eq:A.2} is guaranteed given the assumptions on $u(\theta)$ being continuous and strictly increasing. Since the difference between any two consecutive continuation values must lie strictly between 0 and $\overline{\mu}u(1)$, then, by the Intermediate Value Theorem, there exists one and only one root $\widetilde\omega_b$ satisfying \ref{eq:A.2} and, by extension the above reoccurrence relation. \end{proof}1-10 {% load observations rest_framework_latex %} {% with original as o %} \paragraph{Damage} {{ o.get_body_part_display }}: {{ o.get_damage_age_display }} {{ o.get_damage_type_display }} {{ o.description }} {% endwith %} @misc{rfc8955, series = {Request for Comments}, number = 8955, howpublished = {RFC 8955}, publisher = {RFC Editor}, doi = {10.17487/RFC8955}, url = {https://rfc-editor.org/rfc/rfc8955.txt}, author = { and and and and }, title = {{Dissemination of Flow Specification Rules}}, pagetotal = 36, year = 2020, month = dec, abstract = {This document defines a Border Gateway Protocol Network Layer Reachability Information (BGP NLRI) encoding format that can be used to distribute (intra-domain and inter-domain) traffic Flow Specifications for IPv4 unicast and IPv4 BGP/MPLS VPN services. This allows the routing system to propagate information regarding more specific components of the traffic aggregate defined by an IP destination prefix. It also specifies BGP Extended Community encoding formats, which can be used to propagate Traffic Filtering Actions along with the Flow Specification NLRI. Those Traffic Filtering Actions encode actions a routing system can take if the packet matches the Flow Specification. This document obsoletes both RFC 5575 and RFC 7674.}, } \documentclass[ngerman, a4paper]{article} \usepackage{sans} \usepackage{graphicx,color} \usepackage[utf8]{inputenc} \usepackage{verbatim} \usepackage{tabularx} \usepackage{multicol} \usepackage{multirow} \usepackage{wrapfig} \usepackage{fancyhdr} \usepackage{pifont} \usepackage[inline]{enumitem} \usepackage{amssymb} \usepackage[left=0.5cm,right=0.5cm]{geometry} \usepackage[ngerman]{babel} \usepackage{blindtext} \newcommand{\checkx}{\rlap{$\square$}{\raisebox{2pt}{\large\hspace{1pt}\ding{51}}}} \fancyhead{} \cfoot{\parbox[t]{13cm}{\raggedright \VAR{title} \VAR{facility['name']} \\ \copyright \VAR{inspector['organization']['name']}, \VAR{inspector['organization']['street']}, \VAR{inspector['organization']['zipCode']} \VAR{inspector['organization']['city']} \\ Prüfer: \VAR{inspector['name']}}} \lfoot{\vbox to 0.3cm{\includegraphics[height=1cm]{../logo.jpg}}} \rfoot{\today \\ Seite \thepage} \renewcommand{\headrulewidth}{0pt} \renewcommand{\headwidth}{190mm} \setlength{\headheight}{0cm} \pagestyle{fancy} \setlength{\textheight}{260mm} \setlength{\textwidth}{180mm} \setlength{\voffset}{-2cm} \setlength{\hoffset}{0.5cm} \begin{document} \fboxsep0pt \fboxrule0pt \BLOCK{ if facility['picture'] } \fbox{\vbox to 0pt{\hbox to 17cm{\hfill \fbox{\includegraphics[height=3.7cm,width=4.5cm]{\VAR{facility['picture']}}}}}} \BLOCK{ endif } {\LARGE Prüfprotokoll \VAR{title} \\} \begin{tabular}{l l} Prüfgrundlagen: & \parbox[t]{9cm}{\VAR{inspectionStandards}}\\ \end{tabular} \renewcommand{\arraystretch}{1.7} \begin{tabularx}{\textwidth}{l X l X } Spielplatz: & \VAR{facility['name']} \newline \VAR{facility['street']} \newline \VAR{facility['zipCode']} \VAR{facility['city']} & Auftraggeber: & \VAR{issuer['name']} \newline \VAR{issuer['street']} \newline \VAR{issuer['zipCode']} \VAR{issuer['city']} \\ Prüfdatum: & \VAR{inspectionDate} & Prüfer: & \VAR{inspector['organization']['name']} \newline \VAR{inspector['organization']['street']} \newline \VAR{inspector['organization']['zipCode']} \VAR{inspector['organization']['city']} \newline \VAR{inspector['firstName']} \VAR{inspector['name']} \newline \VAR{inspector['email']} \\ Weitere Teilnehmer: & \VAR{attendees} & & \\ \end{tabularx} \renewcommand{\arraystretch}{1.5} \BLOCK{ for entry in entries|sort(attribute='index') } \begin{tabularx}{\textwidth}{ |l|X|X| } \hline \multirow{\VAR{entry['flaws']|length + 1}}{*}{\VAR{loop.index - 1}.} & \VAR{entry['category']['name']}: \textbf{\VAR{entry['title']}} \newline \textit{Hersteller:} \VAR{entry['manufacturer']} \newline \textit{Baujahr:} \VAR{entry['yearBuilt']} \BLOCK{ if entry['manufactureInfoAvailable'] != 'Keine Angabe' } \newline \textit{Herstellerinformation lag vor:} \begin{enumerate*}[itemjoin={\quad},label=$\square$] \item\VAR{ '[\checkx]' if entry['manufactureInfoAvailable'] == 'Ja' else '' } Ja \item\VAR{ '[\checkx]' if entry['manufactureInfoAvailable'] == 'Nein' else '' } Nein \end{enumerate*} \BLOCK{ endif } \newline \textit{Prüfzeichen:} \VAR{entry['inspectionSigns']} \BLOCK{ if entry['easyAccess'] != 'Keine Angabe' } \newline \begin{enumerate*}[itemjoin={\quad},label=$\square$] \item\VAR{ '[\checkx]' if entry['easyAccess'] == 'Ja' else '' } leicht zugänglich (Kinder \textless 3 Jahre) \end{enumerate*} \BLOCK{ endif } & \VAR{entry['category']['inspectionStandards']} \\ \BLOCK{ for flaw in entry['flaws'] } \cline{2-3} & \multicolumn{2}{>{\hsize=\dimexpr2\hsize+2\tabcolsep+\arrayrulewidth\relax}X|}{ \BLOCK{ if flaw['picture'] } \begin{minipage}[t]{0.6\textwidth} \textit{Befund/Mangel:} \VAR{flaw['flaw']}\newline \textit{Priorität:} \VAR{flaw['priority']}\newline \textit{Bemerkung:} \VAR{flaw['notes']} \end{minipage} \begin{minipage}[t]{0.33\textwidth} \raggedleft \vspace{-\ht\strutbox}\includegraphics[height=4cm]{\VAR{flaw['picture']}} \end{minipage} \BLOCK{ else } \textit{Befund/Mangel:} \VAR{flaw['flaw']}\newline \textit{Priorität:} \VAR{flaw['priority']}\newline \textit{Bemerkung:} \VAR{flaw['notes']} \BLOCK{ endif } } \\ \BLOCK{ endfor } \hline \end{tabularx} \BLOCK{ endfor } \vspace{2.3cm} \VAR{inspector.firstName} \VAR{inspector.name} Qualifizierter Spielplatzprüfer, QSP-IH-00165 \end{document} initdb/texlive-thesis-templateforms/abstract.tex0 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter*{Kurzfassung} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % 200-250 words \noindent \bigbreak %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section*{Schlagworte} Software Entwicklung, Qualitätssicherung, Continuous Integration, Continuous Delivery samghelms/mathviz \[\mathop{L^{(\alpha)}_{n}\/}\nolimits\!\left(x^{2}\right)=\frac{2(-1)^{n}}{\pi^% {\frac{1}{2}}\mathop{\Gamma\/}\nolimits\!\left(\alpha+\tfrac{1}{2}\right)n!}\*% \int_{0}^{\infty}\int_{0}^{\pi}{(x^{2}-r^{2}+2ixr\mathop{\cos\/}\nolimits\phi)% ^{n}}\*e^{-r^{2}}r^{2\alpha+1}(\mathop{\sin\/}\nolimits\phi)^{2\alpha}d\phi dr,\]MrPike/Pike.im @inproceedings{james2019incorporating, author = {, , , , }, booktitle = {2019 International Conference on Open and Innovative Education (ICOIE 2019)}, pages = {610--621}, title = {Incorporating Pedagogical Theory into VR to Teach Civil Engineering}, year = {2019} } Assignment-1/A1_sol_LaTeX/main.tex \documentclass[a4paper]{article} \input{head} \usepackage{algorithm} \usepackage[noend]{algpseudocode} \newcommand{\algrule}[1][.2pt]{\par\vskip.5\baselineskip\hrule height #1\par\vskip.5\baselineskip} \begin{document} %------------------------------- % TITLE SECTION %------------------------------- \title{\textbf{COL351 : Analysis and Design of Algorithm \\ Assignment 1 }} \author{ (2019CS10349) \& (2019CS10407)} \date{} %------------------------------- % CONTENTS %------------------------------- \maketitle \section{Minimum Spanning Tree} Let G be an edge-weighted graph with n vertices and m edges satisfying the condition that all the edge weights in G are distinct. \begin{enumerate}[label=(\alph*)] \item Prove that G has a unique MST.\\ \newline Solution: \underline{Proof by Contradiction:}\\\\ Let T and T' be two different MSTs of G. Since both T and T' have same number of vertices and are different, there exists at least one edge that belongs to one and not the other. Let the set of such edges be called P.\\\\ Let the edge with minimum weight among these edges be $e_1$. This edge is unique since all edge weights are distinct. Without loss of generality, assume $e_1$ is in T. If $e_1$ is added to MST T', it will form a cycle C with $e_1$ as one of the edges due to the property of tree. Since, T is a tree (acyclic), there must be at least one edge $e_2$ present in cycle C but not in T because if only edges from tree T is present in cycle C, a cycle wouldn't form.\\ This $e_2$ belongs to T' and not T and thus also belongs to set P. As $e_1$ is the minimum weight edge from set P, wt($e_1$) \textless \ wt($e_2$), as all edge weights are distinct. If $e_2$ is removed from T' U \{$e_1$\}, (T' U \{$e_1$\}) \textbackslash \{$e_2$\} will still be a spanning tree as all paths passing through $e_2$ can be redirected through $e_1$. \\ Thus, (T' U \{$e_1$\}) \textbackslash \{$e_2$\} is a spanning tree with cost wt(T') + wt($e_1$) - wt($e_2$). Since wt($e_1$) \textless \ wt($e_2$), the new weight of this newly formed tree is less than T'. This is a \textbf{contradiction}, as we assumed that T' is a MST.\\\\ Thus, G has a unique MST. {\hfill\qedsymbol} \newpage \item If it is given that G has at most n + 8 edges, then design an algorithm that returns a MST of G in O(n) running time. \textbf{Observations:} \begin{enumerate}[1)] \item Since $m \leq n+8$ and number of edges in a MST is $n-1$, we need to remove atmost 9 edges without disconnecting the graph. \item Cycles can be detected using DFS which taken $O(m+n)$ time but since $m \leq n+8$ so DFS will taken $O(n)$ time. \end{enumerate} \textbf{Algorithm:} \begin{algorithm} \caption{MST Algorithm}\label{alg:mst} \begin{algorithmic}[1] \Procedure{MST}{$G$}\Comment{Returns the MST of graph G} \State $n \to$ number of vertices. \State $m \to$ number of edges and $m \leq n+8$. \While{$m \not= n-1$} \State $S \gets \textsc{DetectCycle}(G)$ \State $e \gets$ Maximum edge weight in S \State $G\gets G \setminus \{e\}$ \State $m \gets m-1$ \EndWhile\label{mstendwhile} \State \textbf{return} $G$\Comment{G is the MST of the original graph} \EndProcedure \algrule \Procedure{DetectCycle}{$G$}\Comment{Returns a set of edges that form a cycle in G} \State $S \gets \{\}$ \State $visited \gets false$ for all vertices. \State $parent \gets -1$ for all vertices. \State $start \gets -1$ \State $end \gets -1$ \For{all $v \in V$} \If{$visited(v) = false$} \State $\textsc{DFS}(v,parent(v))$ \EndIf \EndFor \State $v \gets end$ \While{$v \not= start$} \State Add $v$ to $S$ \State $v \gets parent(v)$ \EndWhile \State \Return $S$ \Comment{$S$ is the set of edges of a cycle} \EndProcedure \algrule \Procedure{DFS}{$v,parent$}\Comment{Recursive function to visit vertices} \State $visited(v) \gets true$ \For{all $u$, neighbours of $v$} \If{$u=parent$} \textbf{continue} \EndIf \If{$visited(u)=true$}\Comment{Cycle Detected} \State $end \gets u$ \State $start \gets v$ \State \Return \EndIf \State $parent(u) \gets v$ \State \textsc{DFS}($u,parent(u)$) \EndFor \EndProcedure \end{algorithmic} \end{algorithm} \textbf{Correctness of Algorithm:} \underline{\textbf{Claim 1:}} Let $C$ be any cycle in $G$, and let edge $e = (v, w)$ be the most expensive edge belonging to $C$. Then $e$ does not belong to any minimum spanning tree of $G$. \underline{\textbf{Proof:}} Proof by Contradiction \\ Delete $e$ from $T$, the MST of $G$; this partitions the vertices into two components: $S$, containing node $v$; and $V \setminus S$, containing node $w$. Now, the edge we use in place of $e$ should have one end in $S$ and the other in $V \setminus S$, so as to stitch the tree back together. We can find such an edge by following the cycle $C$. The edges of $C$ other than $e$ form, by definition, a path $P$ with one end at $v$ and the other at $w$. If we follow $P$ from $v$ to $w$, we begin in $S$ and end up in $V \setminus S$, so there is some edge $e' = (v',w')$ on $P$ that crosses from $S$ to $V \setminus S$. Now consider the set of edges $T'$ = $T \setminus \{e\} \cup \{e'\}$. We claim that $T'$ is a spanning tree. Clearly $(V,T')$ is connected, since $(V,T)$ is connected, and any path in $(V,T)$ that used the edge $e = (v,w)$ can now be “rerouted” in $(V,T')$ to follow the portion of $P$ from $v$ to $v'$, then the edge $e'$, and then the portion of $P$ from $w$ to $w'$. To see that $(V,T')$ is also acyclic, note that the only cycle in $(V,T\cup\{e'\})$ is the one composed of $e'$ and the path $P$, and this cycle is not present in $(V,T')$ due to the deletion of $e$. Moreover, since $e$ is the most expensive edge on the cycle $C$, and $e'$ belongs to $C$, it must be that $e'$ is cheaper than $e$, and hence $T'$ is cheaper than $T$, a contradiction. {\hfill\qedsymbol} \bigskip \underline{\textbf{Proof of Correctness:}}\\ Consider any edge $e = (v, w)$ removed by the algorithm. At the time that $e$ is removed it is the most expensive edge on $C$. Thus by claim 1, $e$ does not belong to any minimum spanning tree. So if we show that the output $(V,T)$ of the algorithm is a spanning tree of $G$, we will be done. Clearly $(V,T)$ is connected as only edges present in a cycle are removed by the algorithm. Since number of edges in $(V,T)$ is $n-1$ as the algorithm terminates here. Thus, $(V,T)$ is acyclic and hence it is a spanning tree. {\hfill\qedsymbol} \end{enumerate} \newpage \section{Huffman Encoding} \begin{enumerate}[(a)] \item What is the optimal binary Huffman encoding for n letters whose frequencies are the first n Fibonacci numbers? What will be the encoding of the two letters with frequency 1, in the optimal binary Huffman encoding? \\ \\ Sketch of Huffman encoding algorithm (taken from notes) \begin{enumerate}[(1)] \item Replace the two letters with least frequency by a new symbol $\tilde{a}$. \item Set freq($\tilde{a}$) = $\tilde{f}$ := $f_0$ + $f_1$. \item Solve $\tilde{F}$ = $(F \cup \tilde{f}) \setminus \{f_0, f_1\}$ and find optimal binary tree $\tilde{T}$. \item If $\tilde{a}$ is node for $\tilde{f}$, then add children $a_0$ and $a_1$ to $\tilde{a}$. \end{enumerate} Let the frequency vector be $F$ where $F[i]$ represents the $i^{th}$ Fibonacci number.\hfill \\ \\ \underline{\textbf{Claim 1:}} $ \forall n \leq N, F[n] \leq \sum_{i=1}^{n-1} F[i] < F[n+1]$ \\ \\ \underline{\textbf{Proof:}} Proof by Induction on $n$ \\ {\textit{Base Case:} \par} {\hspace{10mm} $n = 2$ \par} {\hspace{10mm} $F[2] = 1$ \par} {\hspace{10mm} $\sum_{i=1}^{1} F[i] = F[1] = 1$ \par} {\hspace{10mm} $F[3] = 2$ \par} {\hspace{10mm} Thus, \par} {\hspace{10mm} $F[2] \leq \sum_{i=1}^{1} F[i] < F[3]$ \par} {\textit{Induction Hypothesis:} \par} {\hspace{10mm} $F[n-1] \leq \sum_{i=1}^{n-2} F[i] < F[n]$ \par} {\textit{Induction Step:} \par} {\hspace{10mm} $F[n-1] \leq \sum_{i=1}^{n-2} F[i]$ \hfill (Induction Hypothesis)\par} {\hspace{10mm} $F[n-1] + F[n-1] \leq \sum_{i=1}^{n-2} F[i] + F[n-1]$ \par} {\hspace{10mm} $\sum_{i=1}^{n-1} F[i] \geq F[n-1] + F[n-1] \geq F[n-1] + F[n-2]$ \par} {\hspace{10mm} $\sum_{i=1}^{n-1} F[i] \geq F[n]$ \hfill (A) \par} {\par} {\hspace{10mm} $\sum_{i=1}^{n-2} F[i] < F[n]$ \hfill (Induction Hypothesis)\par} {\hspace{10mm} $\sum_{i=1}^{n-2} F[i] + F[n-1] < F[n] + F[n-1]$ \par} {\hspace{10mm} $\sum_{i=1}^{n-1} F[i] < F[n+1]$ \hfill (B)\par} {\hspace{10mm} From (A) and (B), $F[n] \leq \sum_{i=1}^{n-1} F[i] < F[n+1]$.\par} By PMI, $ \forall n \leq N, F[n] \leq \sum_{i=1}^{n-1} F[i] < F[n+1]$. {\hfill$\qedsymbol$}\\ \underline{\textbf{Claim 2:}} In $k^{th}$ recursive call, the least two frequency symbols are $\sum_{i=1}^{k} F[i]$ and $F[k+1]$.\\ \\ \underline{\textbf{Proof:}} Proof by Induction on $k$ \\ {\textit{Base Case:} \par} {\hspace{10mm} $k = 1$ \par} {\hspace{10mm} Initially, the least two frequencies in $F$ are $F[1]$ and $F[2]$.\par} {\hspace{10mm} So, $\sum_{i=1}^{1} F[i]$ and $F[2]$ are least two frequency symbols.\par} {\textit{Induction Hypothesis:} \par} {\hspace{10mm} $\sum_{i=1}^{k-1} F[i]$ and $F[k]$ are least two frequencies in $(k-1)^{th}$ call.\par} {\textit{Induction Step:} \par} {\hspace{10mm} Applying the algorithm on $(k-1)^{th}$ call, we get a new symbol $\tilde{a}$ with frequency $\tilde{f}$.\par} {\hspace{10mm} Freq $\tilde{f} = \sum_{i=1}^{k-1} F[i] + F[k]$ as these two are least two frequencies (by induction hypothesis).\par} {\hspace{10mm} So, the new frequency vector for next call $\tilde{F_k}$ is \par} {\begin{equation} \tilde{F_k} = [ \sum_{i=1}^{k} F[i], F[k+1], F[k+2] \dots F[N]] \end{equation}} {\hspace{10mm} Using the first claim, \par} {\begin{equation} F[k+1] \leq \sum_{i=1}^{k} F[i] < F[k+2] < F[k+3] \dots < F[n] \end{equation}} {\hspace{10mm}Thus when the algorithm is called using this frequency vector. \par} {\hspace{10mm} The least two frequencies are $\sum_{i=1}^{k} F[i]$ and $F[k+1]$.\par} Hence, by PMI, the above claim is true for all recursive calls. {\hfill$\qedsymbol$}\\ Binary Encoding of all symbols is determined using claim 2. So, in the last recursive call, the only two frequencies are $\sum_{i=1}^{n-1} F[i]$ and $F[n]$ and they are encoded as `0' and `1' respectively. $\sum_{i=1}^{n-1} F[i]$ is formed by combining $\sum_{i=1}^{n-2} F[i]$ and $F[n-1]$. Thus, encoding of $\sum_{i=1}^{n-2} F[i]$ and $F[n-1]$ are `00' and `01' respectively. After expanding all the merged symbols, the final encoding of symbols are: {\hspace{10mm} $F[n] \to$ `1'\par} {\hspace{10mm} $F[n-1] \to$ `01'\par} {\hspace{10mm} $F[n-2] \to$ `001'\par} {\hspace{10mm} $\dots$\par} {\hspace{10mm} $\dots$\par} {\hspace{10mm} $\dots$\par} {\hspace{10mm} $\dots$\par} {\hspace{10mm} $F[3] \to$ `00....001' (n-3 zeros)\par} {\hspace{10mm} $F[2] \to$ `00....001' (n-2 zeros)\par} {\hspace{10mm} $F[1] \to$ `00....000' (n-1 zeros)\par} \newpage \item Suppose you aim to compress a file with 16-bit characters such that the maximum character frequency is strictly less than twice the minimum character frequency. Prove that the compression obtained by Huffman encoding, in this case, is same as that of the ordinary fixed-length encoding.\\ \underline{\textbf{Property P:}} maximum frequency \textless \ 2*min frequency\\ \underline{\textbf{Given:}} 16 bit characters with maximum frequency \textless \ 2*min frequency\\ \underline{\textbf{Claim:}} The Huffman encoding with n bit characters is reducible to n-1 bit characters by merging 2 characters and representing it as a single character using n-1 bits, following the same property i.e, maximum frequency \textless \ 2*min frequency\\ \underline{\textbf{Proof:}}\\ Let N = $2^n$ and F = [$f_1$, $f_2$ ....., $f_N$] be the frequencies sorted in ascending order.\\ Merging the least frequencies according to Huffman algorithm.\\ $\Rightarrow \tilde{f_1}$ = $f_1$ + $f_2$\\ {Since $f_N < 2*f_1 $\par} {\hspace{10mm} $\Rightarrow f_1 + f_2 \geq 2*f_1 > f_N$ \par} {\hspace{10mm} $\Rightarrow f_1 + f_2 > f_i \ \ \forall i = 1, 2.... N$ \par} {Thus the new sorted frequency vector is [$f_3, f_4,......f_N, \tilde{f_1}$]\\ Recursively merge the least two frequency characters to form a single character of the combined frequency, N/2 times. We get the new frequency vector as,\par} {\hspace{10mm} $\tilde{F} = [\tilde{f_1}, \tilde{f_2},.....\tilde{f_{N/2}}]$\par} {\hspace{10mm} where $\tilde{f_i} = f_{i*2-1} + f_{i*2}$\par} {\hspace{10mm} max($\tilde{F}$) = $f_{N-1} + f_N$ \par} {\hspace{10mm} min($\tilde{F}$) = $f_{1} + f_2$ \par} {\hspace{10mm} Since $f_N < 2*f_1 \ and \ f_{N-1} \leq f_N < 2*f_1 \leq 2*f_2$\par} {\hspace{10mm} So, $f_N + f_{N-1} < 2f_1 + 2f_2$ \par} {\hspace{10mm} $\Rightarrow$ max($\tilde{F}$) $<$ 2min($\tilde{F}$) \par } Thus, $\tilde{F}$ represents frequency vector of $2^{N-1}$ characters or n-1 bit characters with same property.{\hfill$\qedsymbol$}\\ \underline{\textbf{Proof:}} Proof by Induction on number of bits \\ {\textit{Base Case:} \par} {\hspace{10mm} $n = 1$ \par} {\hspace{10mm} Number of characters = $2^1$\par} {\hspace{10mm} Huffman encoding is `0' and `1'\par} {\hspace{10mm} Huffman encoding is `0' and `1'\par} {\textit{Induction Hypothesis:}}\\ Huffman encoding for n-1 bit characters which follow property P is same as fixed length encoding of length n-1.\\ {\textit{Induction Step:}}\\ Using the claim proved above, n bit characters are merged to form n-1 bit characters with same property P. By induction hypothesis, these n-1 bit characters has fixed length encoding and forms a complete binary tree.\\ Expand the merged symbols in n-1 bit characters i.e, expand each leaf node present at height n-1 to get 2 children which is present at height h. These resulting nodes represent the n bit characters that were merged to form the n-1 bit Huffman tree and thus the resulting tree is the Huffman tree for n bit characters.\\ As the height of all leaf nodes in the resulting tree is n, it too forms a complete binary tree. Thus, they have fixed length encoding as all the characters require n bits to represent them.\\ Hence, By PMI, compression obtained by Huffman encoding of 16 bit characters is same as fixed length encoding of length 16.{\hfill$\qedsymbol$} \end{enumerate} \newpage \section{Graduation Party of Alice} \begin{enumerate}[(a)] \item Alice wants to throw a graduation party and is deciding whom to call. She has n people to choose from, and she has made up a list of which pairs of these people know each other. She wants to pick a largest subset of n people, subject to two constraints: at the party, each person should have at least five other people whom they know and five other people whom they don’t know. Present an efficient algorithm that takes as input the list of n people along with the list of pairs who know each other and outputs the best choice of party invitees. Give the running time in terms of n. \textbf{Algorithm:} \begin{algorithm} \caption{Selection Algorithm}\label{alg:sa} \begin{algorithmic}[1] \Procedure{Select}{list of people, list of pairs}\Comment{Returns the maximal set of people possible} \State $G \to$ Undirected Graph produced using every person as a vertex \State $n \to$ number of vertices/people \State $m \to$ number of edges. An edge is added between two people if they know each other. \State $change \gets true$ \State $current \gets n$ \State $D \gets$ array of buckets where vertices are arranged by their degree \While{$change = true$} \State $change \gets false$ \For{$v \in D$ with $deg(v) <5$ and $current > deg(v) > current-5$} %\If{$v$ is marked} \textbf{continue} %\EndIf %\If{$deg(v) < 5$} \State $change \gets true$ \State $current \gets current-1$ \State \textsc{RemoveVertex}($G,v$) %\EndIf %\If{$current - deg(v) < 5$} %\State $change \gets true$ %\State $current \gets current-1$ %\State \textsc{RemoveVertex}($G,v$) %\EndIf \EndFor \EndWhile\label{saendwhile} \State \textbf{return} the set of unmarked vertices \EndProcedure \algrule \Procedure{RemoveVertex}{$G,v$}\Comment{Reduces degrees of vertices connected to $v$} \State Mark $v$ \Comment{$v$ is removed from invitees list} \State $D[deg(v)].\textsc{remove}(v)$ \For{$j$ from $1$ to $deg(v)$} \State $u \gets adj(v)(j)$ \If{$u$ is marked} \State \textbf{continue} \EndIf %\State Delete $v$ from $adj(u)$ \State $deg(u) \gets deg(u)-1$ \State $D[deg(u)+1].\textsc{remove}(u)$ \State $D[deg(u)].\textsc{insert}(u)$ \EndFor \EndProcedure \end{algorithmic} \end{algorithm} \textbf{Runtime Analysis:}\\ The graph is stored in adjacency list as a HashMap mapping each vertex to an unordered set of connected vertices. The time taken to make the graph is $O(m)$. Degree of all vertices is also stored separately as a HashMap to give $O(1)$ access to degrees of each vertex and updated accordingly. The degree of all the vertices is then sorted using bucket sort with each bucket being an unordered set(Hash) of vertices which allows us to insert and remove in $O(1)$ time. The buckets themselves are stored as an array of buckets of size N, with the index of the bucket representing the degree of the vertices stored in them. The initial sorting takes $O(m)$ as it is done through a single pass. The inner FOR loop inside the WHILE loop in the Select procedure, is also run only $O(n)$ times as each iteration, a vertex is removed, and the maximum no. of vertices that can be removed is n. Since the number of edges are $m$ so the procedure \textsc{RemoveVertex}'s FOR loop runs atmost $2*m$ over all the calls as the procedure is called for each vertex only once, and the for loop is run degree(v) times. Sum of the degrees of all the vertices is $2*m$. So the overall running time of the algorithm is $O(m+n)$. Since $m \leq n^2$ so the time complexity is $O(n^2)$. \textbf{Correctness of Algorithm:}\\ In each iteration of the algorithm, vertices or people which can never be part of the optimal invitation list are removed as such people know less than 5 or such people don't know less than 5 people. At each iteration, we must remove at least one person, so this algorithm terminates. When this algorithm terminates, by definition the subset of invitees is valid. Since we only remove people we have deduced could not possibly be invited, we always produce the largest possible number of invitees. {\hfill\qedsymbol} \newpage \item Suppose finally Alice invited $n_0$ out of her n friends to the party. Her next task is to set a minimum number of dinner tables for her friends under the constraint that each table has a capacity of ten people and the age difference between members of each dining group should be at most ten years. Present a greedy algorithm to solve this problem in $O(n_0)$ time assuming the age of each person is an integer in the range [10, 99]. \begin{algorithm} \caption{Minimum tables Algorithm}\label{alg:arr} \begin{algorithmic}[1] \Procedure{Minimum}{$G$} \State $n_0 \gets$ number of selected friends. \State $Tables \gets \{\}$ \State $S \gets \textsc{sort}(G)$\Comment{sorted in ascending order of Age} \While{$\textsc{size}(S) \not= 0$} \State $start \gets \textsc{front}(S)$.Age \State $T \gets NULL$ \State $T$.\textsc{insert}($S$.\textsc{pop\_front}) \State $k \gets 1$ \While{($\textsc{size}(S) \not= 0$ and $k < 10$ and $\textsc{front}(S).age \leq start + 10$)} \State $T$.\textsc{insert}($S$.\textsc{pop\_front}) \State $k \gets k+1$ \EndWhile \State $Tables$.\textsc{insert}($T$) \EndWhile \State \Return $Tables$ \EndProcedure \end{algorithmic} \end{algorithm} \textbf{Runtime Analysis:}\\ Sorting of people by age is possible in $O(n_0)$ time complexity using bucket sort as the range of age is fixed between $10$ and $99$. Every person present in the sorted list is only visited once because as soon as a vertex is visited, it is popped out of the list. Thus, the overall time complexity of the above algorithm is $O(n_0)$. \textbf{Correctness of Algorithm:}\\ \underline{\textbf{Claim:}} Let P be the set of people that have to be assigned tables. Let $a_1$ be the age of the youngest person. Let T be a table that seats the youngest person in P in optimal arrangement. Let T' be a table that seats the youngest person in P by applying the greedy algorithm. Then $size(T') \geq size(T)$ and $youngest((P \setminus T')) \geq youngest((P \setminus T))$\\ \underline{\textbf{Proof:}} \\ \underline{Case 1:} The no. of people with age $a_1$ is more than 10.\\ In this case, our greedy algorithm would form a table with all the 10 people of the same youngest age $a_1$.\\ $Size(T') = 10$ which is the maximum allowed. $youngest((P \setminus T')) = a_1$ as there is more than 10 people with that age. No matter what arrangement T has, $youngest((P \setminus T)) = a_1$ as there is more than 10 people with that age. So, the claim is proved.\\ \underline{Case 2:} The no. of people with age $a_1$ is less than or equal to 10.\\ \phantom{Case 2:} \underline{Case 2.1:} $T = T'$\\ \phantom{Case 2:Case 2.1:} Claim is proved. \phantom{Case 2:} \underline{Case 2.2:} $T \not= T'$\\ \phantom{Case 2:Case 2.1:} Proof by contradiction\\ \phantom{Case 2:Case 2.1:} Assume $size(T') < size(T)$\\ \phantom{Case 2:Case 2.1:} $=> \exists \ p_1$ present in T but not present in T' and size(T') $<$ 10.\\ \phantom{Case 2:Case 2.1:} But because $p_1 \nexists \ T' => p_1 > a_1 + 10$\\ \phantom{Case 2:Case 2.1:} $=> p_1 \nexists \ T$ as a person with age $a_1$ is present in T.\\ \phantom{Case 2:Case 2.1:} Contradiction arises. So, Assumption is false.\\ \phantom{Case 2:Case 2.1:} Assume $youngest((P \setminus T')) < youngest((P \setminus T))$\\ \phantom{Case 2:Case 2.1:} All persons with age less than $youngest((P \setminus T))$ is present in T, but only persons with \\ \phantom{Case 2:Case 2.1:} age less than $youngest((P \setminus T'))$ are present in T'.\\ \phantom{Case 2:Case 2.1:} $=> size(T') < size(T)$, due to the assumption, which is false as proved above.\\ \phantom{Case 2:Case 2.1:} Contradiction arises. So, Assumption is false.\\ \phantom{Case 2:Case 2.1:} Claim is proved. {\hfill\qedsymbol} \bigskip \underline{\textbf{Proof of Correctness:}}\\ At each step of the algorithm, where a table is assigned for the youngest person of the remaining people, it is shown by the above claim that both size of the assigned table by the algorithm is equivalent to optimal arrangement and the youngest person left to be assigned also is equivalent. As, the youngest person's age not assigned is greater than or equal to the optimal arrangement, this implies that the people not assigned a table at each step is minimised optimally in the greedy algorithm, as all the people below that age have been assigned a table. This shows that the greedy algorithm is working equivalent to an optimal algorithm, and in fact stays ahead of the optimal solution always. As this algorithm continues executing until all people are assigned a table, and at each step, at least 1 person is assigned a table, the algorithm terminates.\\ The correctness of the above greedy algorithm is thus proved. {\hfill\qedsymbol} %\phantom{Case 2:Case 2.1:} \underline{Case 2.2.1:} There exists a person $p_1$ is present in T but not in T'.\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} \\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} This implies, there exists a person $p_2$ present in T' but not in T.\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} $=> age(p_1) \geq youngest((P \setminus T'))$\\ %\geq(p_2) $\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} $=> age(p_2) \geq youngest((P \setminus T)) $\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} $=> youngest((P \setminus T')) \geq youngest((P \setminus T))$\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:}Claim is proved.\\ %\phantom{Case 2:Case 2.1:} \underline{Case 2.2.2:} $Size(T) \not= Size(T')$\\ %\phantom{Case 2:Case 2.1:Case 2.2.1:} \end{enumerate} %------------------------------------------------ \end{document} \documentclass[DIV16,twocolumn,10pt]{scrreprt} \usepackage{paralist} \usepackage{graphicx} \usepackage[final]{hcar} \usepackage{caption} %include polycode.fmt %include forall.fmt %include greek.fmt %include colorcode.fmt % \definecolor{codecolor}{rgb}{1,1,1} % \colorhs % \let\Conid\mathsf \begin{document} \include{Yampa-IY} \end{document} \chapter{Conclusioni} In questa tesi abbiamo proposto un protocollo per la Timed-Release Encryption. A differenza delle soluzioni precedenti \cite{time-capsule-signature, 10.1007/11602897_25, 10.1007/3-540-48910-X_6, 10.1007/11889663_17, 10.1007/978-3-642-15317-4_1, 10.1145/1330332.1330336, chalkias2007improved}, il protocollo proposto è profondamente decentralizzato. Il messaggio è distribuito tra $ N $ agenti orchestrati da smart contract eseguiti su blockchain. Nella gran parte delle situazioni il protocollo proposto è in grado di fornire un buon grado di robustezza, ossia è in grado di resistere ad un certo numero di comportamenti scorretti da parte degli agenti senza compromettere la \textit{certezza di pubblicazione} e la \textit{segretezza del messaggio}. I possibili sviluppi futuri sono diversi. È possibile valutare l'introduzione di uno strumento che utilizza la \textit{zero knowledge proof} per permettere ai destinatari del messaggio di verificare che gli agenti non abbiano smarrito gli share senza dover attendere la deadline. Si può studiare come aggiungere ulteriori disincentivi per dissuadere gli agenti da comportamento scorretti, ad esempio introducendo delle sanzioni attraverso contratti giuridici, appoggiandosi a ledger \textit{permissioned} e meccanismi \textit{KYC}\footnote{know your customer} per validare gli agenti. Si può valutare come gestire messaggi dalle elevate dimensioni che non possono essere economicamente memorizzati nella \mbox{blockchain} appoggiandosi a prodotti commerciali per lo storage in cloud come Amazon S3 \cite{amazon-s3} e Google Cloud Storage \cite{gcs} oppure utilizzando storage distribuiti come IPFS \cite{ipfs} e Ethereum Swarm \cite{ethereum-swarm}. Si può pensare di creare un marketplace per favorire l'incontro tra domanda (client) e offerta (agenti), incrementando la competizione tra gli agenti con l'obiettivo di ottenere un servizio migliore. In particolare, si può pensare ad un sistema di ranking degli agenti o estendere la variante \textit{cauzione per gli agenti} [si veda \ref{subsec:cauzione-agenti}] basandosi sull'idea di \textit{proof of stake}. Infine, si può valutare come implementare il protocollo in particolari contesti, magari eliminando la blockchain e utilizzando incentivi/disincentivi non legati alla criptomoneta ma al particolare dominio applicativo.@book{56fafbf2574947cc9cbbfae578a0a36d, title = "Book with one author", author = "", year = "2017", month = "10", publisher = "Van Gennep", } @article{d79d00c790984ab08240e997d077c332, title = "Article with 5 authors with 'and' notation", abstract = "Since the 1950s the amount of plastics in the marine environment has increased dramatically. Worldwide there is a growing concern about the risks and possible adverse effects of (micro)plastics. This paper reflects on the sources and effects of marine litter and the effects of policies and other actions taken worldwide. Current knowledge offers a solid basis for effective action. Yet, so far the effects of policies and other initiatives are still largely insufficient. The search for appropriate responses could be based on possible interventions and profound understanding of the context specific factors for success. Moreover, the scope, timeframe and dynamics of all initiatives are distinctly different and orchestration at all levels, in close cooperation with one another, is currently lacking.", author = " and and and and and {}, Frank", year = "2017", month = "10", doi = "10.1016/j.cosust.2017.08.009", volume = "28", pages = "90--99", journal = "Current Opinion in Environmental Sustainability", issn = "1877-3435", publisher = "Elsevier", } @article{a8781aa0eae047d1826a658f3545ce3f, title = "Article with 3 authors with mixed notation", abstract = "Using country-level data from 2003–2014, we examine the association between auditing level (measured as number of verification actions taken by tax authorities per 100 taxpayers in each country) and tax compliance (measured as business executives’ perception of tax evasion). Our hypothesis is that compliance increases until a certain auditing level is reached, and decreases beyond that level (i.e., an elevated auditing level backfires). In line with our expectation, the results of a series of tests indicate that there is a U-shaped association between auditing and tax evasion. We discuss how a potential backfiring effect may depend on the extent to which compliance is voluntary.", keywords = "Tax compliance, Auditing, Tax enforcement", author = "{}, J.P. and and {Erich Some Middle Name} Kirchler", year = "2017", month = "10", volume = "62", pages = "284--294", journal = "Journal of Economic Psychology", issn = "0167-4870", publisher = "Elsevier", }\hypertarget{_resources_8_designer_8cs}{}\doxysection{D\+:/\+Repositories/\+Y\+T\+Music\+Uploader/\+Y\+T\+Music\+Uploader/\+Properties/\+Resources.Designer.\+cs File Reference} \label{_resources_8_designer_8cs}\index{D:/Repositories/YTMusicUploader/YTMusicUploader/Properties/Resources.Designer.cs@{D:/Repositories/YTMusicUploader/YTMusicUploader/Properties/Resources.Designer.cs}} \doxysubsection*{Classes} \begin{DoxyCompactItemize} \item class {\bfseries Y\+T\+Music\+Uploader.\+Properties.\+Resources} \begin{DoxyCompactList}\small\item\em A strongly-\/typed resource class, for looking up localized strings, etc. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection*{Namespaces} \begin{DoxyCompactItemize} \item namespace \mbox{\hyperlink{namespace_y_t_music_uploader}{Y\+T\+Music\+Uploader}} \item namespace \mbox{\hyperlink{namespace_y_t_music_uploader_1_1_properties}{Y\+T\+Music\+Uploader.\+Properties}} \end{DoxyCompactItemize} CVs/Viviane.Pons.tex \begin{participant}[type=PI,PM=2,gender=female]{} at the Laboratoire de Recherche en Informatique, is a young researcher in Algebraic Combinatorics. She defended her thesis in 2013 and has 4 papers in international journals and 5 communications in international conferences, including a talk at PyCon US 2015. She also is in the editorial board of the Journal of Open Source Software. Before starting her research career, she worked for two years in industry as a Java and web developer. She discovered \Sage during her first \Sage Days in 2010 and has since been an active user and contributor with 10 (co)authored tickets improving the support of combinatorial objects in \Sage. She is heavily involved in the promotion of \Sage, participating in \Sage Days and running \Sage introduction tutorials or \Sage presentations at various conferences. She is also one of the main developers of the project \software{FindStat} dedicated to databases in combinatorics. Viviane is leading the very successful Community Building and Dissemination work package of the European Research Infrastructures project OpenDreamKit (2015-2019), in which 66 events (development workshops, training sessions, ...) were organized or coorganized, with more than a thousand trainees. Viviane herself organized or coorganized several of them, including two week-long workshops dedicated to women. \end{participant} %%% Local Variables: %%% mode: latex %%% TeX-master: "../proposal" %%% End: hjaremko/bd-egzaminsrc/sections/indexes_btrees.tex \section{Indeksy, typy indeksów, statystyki, wykorzystanie przez optymalizatory kwerend} \label{sec:indeksy} \horrule{0.5pt} Proszę \underline{omówić budowę} indeksu typu \textbf{drzewo B+}. Proszę podać wersje tego indeksu (w systemie Microsoft SQL Server: clustered i non-clustered, w Oracle IOT).\\ \horrule{0.5pt}\\ \textbf{WĘZEŁ WEWNĘTRZNY} z $q-1$ wartości\\ \begin{adjustbox}{width=\columnwidth,center} \includegraphics[scale=1]{internal} \end{adjustbox} \begin{itemize} \item $ \langle P_1, K_1, P_2, K_2, \ldots, P_{q-1}, K_{q-1}, P_q \rangle$, gdzie $q \leqslant p$,\\ $P_i (i = 1, \ldots, q)$ jest wskaźnikiem do poddrzewa. \item $K_1 < K_2 < \ldots < K_{q-1}$ \item Każdy wewnętrzny węzeł ma co najwyżej $p$ wskaźników. \item Każdy węzeł \textit{(z wyjątkiem korzenia)} ma przynajmniej $\left\lceil \frac{p}{2} \right\rceil$ wskaźników do poddrzew.\\ Korzeń ma co najmniej dwa wskaźniki, jeśli jest węzłem wewnętrznym. \item Węzęł wewnętrzny z $q$ wskaźnikami ma $q-1$ wartości. \end{itemize} \pagebreak \textbf{WĘZEŁ ZEWNĘTRZNY}\\ \begin{adjustbox}{width=\columnwidth,center} \includegraphics[scale=1]{external} \end{adjustbox} \begin{itemize} \item $\langle\langle K_1, Pr_1 \rangle , \langle K_2, Pr_2 \rangle , \ldots, \langle K_{q-1}, Pr_{q-1}, \rangle P_{next} \rangle$ , gdzie $q \leqslant p$ i $Pr_i$ jest wskaźnikiem \textbf{do danych}, $P_{next}$ wskazuje na następny liść. \item $K_1 < K_2 < \ldots < K_{q-1}$ \item Każdy $Pr_i$ jest wskaźnikiem do danych, który wsakzuje na rekord, którego wartość w polu indeksowanym jest równa $K_i$ lub wskazuje na blok zawierający rekord lub na blok wskaźników do rekordów z takimi samymi wartościami $K_i$ jeśli klucz indeksu nie jest unikalny. \item Każdy liść przechowuje co najmniej $\left\lceil \frac{p}{2} \right\rceil$ wartości. \item Wszystkie liście są na tym samym poziomie. \end{itemize} % TODO: % \textbf{WERSJE}\\ % \textbf{B DRZEWO A B+ DRZEWO}\\ \documentclass{syllabus} \setheader{CS 5300}{Compiler Construction} \begin{document} \section*{Description} This course covers the basics of compiler construction. Including lexical analysis, parsing, optimization and code generation. \begin{text}{Compilers: Principles, Techniques, and Tools} \textchapter{Lexical Analysis}{3} \textchapter{Syntax Analysis}{4} \textchapter{Intermediate-Code Generation}{5} \textchapter{Code Generation}{8} \textchapter{Machine-Independent Optimizations}{9} \end{text} \begin{grading} \gradecategory{Assignments}{70} \gradecategory{Midterm Exams}{15} \gradecategory{Final Exam}{15} \end{grading} \officehours{By Appointment} %\finalexam{Monday December $9^{th}$ 7:30~a.m.~-~9:20~p.m.} %\adddrop{September $16^{th}$} \cheating \ada \fees{25} \end{document} % Präambel \documentclass[11pt, % Schriftgröße a4paper, % Papierformat oneside, % einseitiges (oneside) oder zweiseitiges (twoside) Dokument listof=totoc, % Tabellen- und Abbildungsverzeichnis ins Inhaltsverzeichnis bibliography=totoc, % Literaturverzeichnis ins Inhaltsverzeichnis aufnehmen titlepage, % Titlepage-Umgebung statt \maketitle headsepline, % horizontale Linie unter Kolumnentitel %abstracton, % Überschrift beim Abstract einschalten, Abstract muss dazu in {abstract}-Umgebung stehen DIV18, % auskommentieren, um den Seitenspiegel zu vergrößern BCOR6mm, % Bindekorrektur, die den Seitenspiegel um 6mm nach rechts verschiebt, cleardoublepage=empty, % Stil einer leeren eingefügten Seite bei Kapitelwechsel parskip % Absatzabstand bei Absatzwechsel einfügen ]{scrbook} \usepackage{ucs} % Dokument in utf8-Codierung schreiben und speichern \usepackage[utf8x]{inputenc} % ermöglicht die direkte Eingabe von Umlauten \usepackage[ngerman]{babel} % deutsche Trennungsregeln und Übersetzung der festcodierten Überschriften \usepackage[T1]{fontenc} % Ausgabe aller zeichen in einer T1-Codierung (wichtig für die Ausgabe von Umlauten!) \setlength{\parindent}{0ex} % bei neuem Abschnitt nicht einrücken \linespread{1.2}\selectfont % Zeilenabstand erhöhen - größere Werte als 1.2 nicht verwenden!! \usepackage{scrpage2} % SCR Headings verwenden \setheadsepline{0.4pt} % Kopfzeile Linien oben \setfootsepline{0.4pt} % Kopfzeile Linien unten \pagestyle{scrheadings} % SCR Headings einschalten \usepackage{graphicx} % Einbinden von Grafiken erlauben \usepackage[format=hang, % Formatierungen von Unter- / Überschriften font=normal, labelfont=bf, justification=RaggedRight, singlelinecheck=true, aboveskip=1mm ]{caption} % Please add the following required packages to your document preamble: \usepackage[table,xcdraw]{xcolor} % If you use beamer only pass "xcolor=table" option, i.e. \documentclass[xcolor=table]{beamer} % Note: It may be necessary to compile the document several times to get a multi-page table to line up properly \usepackage{booktabs} \usepackage{longtable} \usepackage{pdfpages} \usepackage{enumitem} % Erlaubt Änderung der Nummerierung in der Umgebung enumerate \usepackage{amsmath} % Ergänzungen für Formeln \usepackage{textcomp} % zum Einsatz von Eurozeichen u. a. Symbolen \newcommand*\diff{\mathop{}\!\mathrm{d}} % Differentialzeichen \newcommand*\Diff[1]{\mathop{}\!\mathrm{d^#1}} % Differentialzeichen höherer Ableitung \newcommand*\jj{\mathop{}\!\mathrm{j}} % Komplexe Zahl j \usepackage[ % Einstellunge Paket hyperref hyperfootnotes=false % im pfd-Output Fußnoten nicht verlinken ]{hyperref} \usepackage{makeidx} % Paket zur Erstellung eines Index \usepackage[intoc]{nomencl} % zur Erstellung des Abkürzungsberzeichnisses \usepackage[ % Einstellungen für Fußnoten bottom, % Ausrichtung unten multiple, % Trennung durch Seperator bei mehreren Fußnoten hang, marginal ]{footmisc} \usepackage{calc} % Paket zum Berechnen von Längen z.B. 0.8\linewidth \usepackage{xcolor} % einfache Verwendung von Farben in nahezu allen Farbmodellen \usepackage{listings} % Darstellung von Quellcode mit den Umgebungen {lstlisting}, \lstinline und \lstinputlisting \lstset{literate= % Damit können Umlaute innerhalb Listings geschrieben werden {Ö}{{\"O}}1 {Ä}{{\"A}}1 {Ü}{{\"U}}1 {ß}{{\ss}}1 {ü}{{\"u}}1 {ä}{{\"a}}1 {ö}{{\"o}}1 } \definecolor{mygreen}{rgb}{0,0.6,0} \definecolor{mygray}{rgb}{0.5,0.5,0.5} \definecolor{mymauve}{rgb}{0.58,0,0.82} \lstset{ % backgroundcolor=\color{white}, % choose the background color; you must add \usepackage{color} or \usepackage{xcolor}; should come as last argument basicstyle=\footnotesize, % the size of the fonts that are used for the code breakatwhitespace=false, % sets if automatic breaks should only happen at whitespace breaklines=true, % sets automatic line breaking captionpos=t, % sets the caption-position to (b) bottom or (t) top commentstyle=\color{mygreen}, % comment style deletekeywords={...}, % if you want to delete keywords from the given language escapeinside={\%*}{*)}, % if you want to add LaTeX within your code escapeinside={(*@}{@*)}, extendedchars=true, % lets you use non-ASCII characters; for 8-bits encodings only, does not work with UTF-8 frame=none, % "single" adds a frame around the code; "none" keepspaces=true, % keeps spaces in text, useful for keeping indentation of code (possibly needs columns=flexible) keywordstyle=\color{blue}, % keyword style language=[LaTeX]TeX, % the language of the code morekeywords={*,nomenclature}, % if you want to add more keywords to the set numbers=left, % where to put the line-numbers; possible values are (none, left, right) numbersep=5pt, % how far the line-numbers are from the code numberstyle=\tiny\color{mygray}, % the style that is used for the line-numbers rulecolor=\color{black}, % if not set, the frame-color may be changed on line-breaks within not-black text (e.g. comments (green here)) showspaces=false, % show spaces everywhere adding particular underscores; it overrides 'showstringspaces' showstringspaces=false, % underline spaces within strings only showtabs=false, % show tabs within strings adding particular underscores stepnumber=1, % the step between two line-numbers. If it's 1, each line will be numbered stringstyle=\color{mymauve}, % string literal style tabsize=2, % sets default tabsize to 2 spaces title=\lstname % show the filename of files included with \lstinputlisting; also try caption instead of title } \makeindex % Indexverzeichnis erstellen \makenomenclature % Abkürzungsverzeichnis erstellen % ----------------------------------------------------------------------------------------------------------------- % Zum Aktualisieren des Abkürzungsverzeichnisses (Nomenklatur) bitte auf der Kommandozeile folgenden Befehl aufrufen : % makeindex .nlo -s nomencl.ist -o .nls % Oder besser: Kann in TexStudio unter Tools-Benutzer als Shortlink angelegt werden % Konfiguration unter: Optionen-Erzeugen-Benutzerbefehle: makeindex -s nomencl.ist -t %.nlg -o %.nls %.nlo % ----------------------------------------------------------------------------------------------------------------- % Hier die persönlichen Daten eingeben: \newcommand{\titel}{BlindControl} \newcommand{\untertitel}{Eine smarte Steuerung für Jalousien} \newcommand{\arbeit}{Projektdokumentation} \newcommand{\studiengang}{Elektrotechnik} \newcommand{\studienrichtung}{Fahrzeugelektronik} \newcommand{\studienschwerpunkt}{} \newcommand{\autor}{, , } \newcommand{\matrikelnr}{123 456} \newcommand{\kurs}{TFE17-2} \newcommand{\firma}{Firma (Angabe entfällt ggf. bei Studienarbeit)} \newcommand{\abgabe}{30.04.2019} \newcommand{\betreuerdhbw}{} \newcommand{\betreuerfirma}{Gutachter der Firma (Angabe entfällt ggf. bei Studienarbeit)} \newcommand{\jahr}{2019} % für Angabe im Copyright-Vermerk der Titelseite % Folgende Zeilen definieren Abkürzungen, um Befehle schneller eingeben zu können \newcommand{\ua}{\mbox{u.\,a.\ }} \newcommand{\zB}{\mbox{z.\,B.\ }} \newcommand{\bs}{$\backslash$} % Folgende Zeilen weden benötigt, um Tikz und PGF-Plot-Grafiken einzubinden \usepackage{pgfplots} \usepackage{pgfplotstable} \pgfplotsset{compat=newest} \usepgfplotslibrary{smithchart} \usepackage{tikz} % Tikz sollte nach Listings Pakete geladen werden. \usetikzlibrary{arrows} \hyphenation{Schrift-ar-ten} \usepackage{float} % ------------------------------------------------------------------------------------------- % Beginn des Dokumenteninhalts % ------------------------------------------------------------------------------------------- \begin{document} \setcounter{secnumdepth}{3} % Nummerierungstiefe fürs Inhaltsverzeichnis \setcounter{tocdepth}{3} \sffamily % für die Titelei serifenlose Schrift verwenden % ------------------------------ Titelei ----------------------------------------------------- \include{pages/titelseite} % erzeugt die Titelseite \pagenumbering{roman} % kleine, römische Seitenzahlen für Titelei %\include{pages/erklaerung} % Einbinden der eidestattlichen Erklärung %\include{chapter/abstract} % Einbinden des Abstracts \tableofcontents % Erzeugen des Inhalsverzeichnisses \cleardoublepage % -------------------------------------------------------------------------------------------- % Inhalt der Bachelorarbeit %--------------------------------------------------------------------------------------------- \pagenumbering{arabic} % arabische Seitenzahlen für den Hauptteil \rmfamily \include{chapter/einfuehrung} \include{chapter/grundlagen} \include{chapter/anforderungen} \include{chapter/projektplanung} \include{chapter/architektur} \include{chapter/implementierung} \include{chapter/test} \include{chapter/installation} \include{chapter/fazit} % ---- Literaturverzeichnis ---------- % \bibliography{literature/literatur1,literature/literatur2} % Einbindung mehrerer Verzeichnisse in einem \bibliography Befehl mit Kommata trennen - keine Leerzeichen nach den Kommata! \bibliographystyle{alphadin} %plain: alphabetisch, unsrt: nach Zitat, alphadin: NameJahr % -----Ausgabe aller Verzeichnisse --- \setlength{\parskip}{0.5\baselineskip} %\renewcommand{\indexname}{Sachwortverzeichnis} %\printindex % Erzeugen des Indexverzeichnises %\addcontentsline{toc}{chapter}{\indexname} \input{pages/abkuerzungen} % Datei mit allgemeinen Abkürzungen laden \renewcommand{\nomname}{Verzeichnis verwendeter Formelzeichen und Abkürzungen} \setlength{\nomlabelwidth}{.20\hsize} \renewcommand{\nomlabel}[1]{#1 \dotfill} \setlength{\nomitemsep}{-\parsep} \printnomenclature % Erzeugen des Abkürzungsverzeichnises, siehe auch Inhalt der Datei pages/abkuerzungen.tex \cleardoublepage %\renewcommand{\glossaryname}{Glossar} %\printglossaries %\cleardoublepage \listoffigures % Erzeugen des Abbildungsverzeichnisses \cleardoublepage \listoftables % Erzeugen des Tabellenverzeichnisses \cleardoublepage % -----Anhang ------------------------ \begin{appendix} \clearpage %\pagenumbering{Roman} % große, römische Seitenzahlen für Anhang, falls gewünscht \include{chapter/anhang} %\include{chapter/vorlagen/anhang_vorlagen} % Zeile auskommentieren bei finalem Dokument! \end{appendix} \end{document}\section{Conclusions and Implications} \label{sec:conclusion} In \citetitle{qe2paper}, \citeauthor{qe2paper} conclude that the proposed implementation of solving the \gls{nem} equations with the \gls{jfnk} method after performing local elimination and using the physics-based preconditioner provides improved convergence rate and reduced computation time compared to their implementation of the \gls{pi} method. It is difficult to argue with these results as presented. However, \sref{sec:critique} has shown that it is the data omitted from the results that merits consideration. Without additional data, it is difficult to extend the results of \citeauthor{qe2paper} to broader applications of the \gls{jfnk} method for solving the \gls{nem} equations. A first step for better interpreting the results presented is to compare the \gls{jfnk} method to a \gls{pi} method using the \gls{ws} as in the work by \citeauthor{jfnk_wielandt}. It would also be useful to investigate preconditioning with a few \glspl{pi} as this has been demonstrated to be the preferable preconditioning by others \cite{gill_azmy,jfnk_wielandt}. Additionally, the \gls{nem} should be implemented in the \gls{cmfd} formulation to investigate improved computational efficiency and allow for compatibility with existing production-quality computer programs \cite{palmtagThesis,smith_nonlinear}. In their concluding remarks, \citeauthor{qe2paper} claim that their method will be extended to ``improve the computational efficiency for large-scale complicated multiphysics coupled problems in nuclear reactor analysis'' \cite{qe2paper}. This may seem like a straightforward extension, but a new residual function and Jacobian approximation will likely be required. The authors' claim of incorporating multiphysics effects is extremely bold given both the omission of important results and other publications in the field. Specifically, \citeauthor{caslJFNK} investigated this exact application of \gls{jfnk} to large-scale multiphysics simulations. Such an application required significant work in the formation of the Jacobian including constructing approximations of temperature derivatives of cross sections. These cross section derivatives were specific to a particular type of reactor (\glspl{pwr}) but \citeauthor{qe2paper} attempt to simulate both \glspl{pwr} and pebble bed reactor designs. Succinctly, in reactor multiphysics simulations, it was determined that without the calculation of cross section derivatives, the \gls{jfnk} method was unacceptable as it required a burdensome amount of cross section processing time during the simulation \cite{caslJFNK}. The results of \citeauthor{caslJFNK} are not the end of the narrative. Applying the \gls{jfnk} method to the \gls{nem} equations may be useful for certain nuclear reactor multiphysics simulations. However, insufficient data is provided by \citeauthor{qe2paper} to make such a conclusion. Additional attention must be paid in the application of the \gls{jfnk} method to the \gls{nem} equations to determine if such an implementation would provide a benefit compared to existing methods. When drawing conclusions about improvements in computational efficiency due to a new method, it is crucial to compare optimized codes to determine if any improvement is true, or merely due to a suboptimal implementation. varqox/img2tex 1 \leq q_1 \neq q_2 < n @article{ISI:000368458800058, abstract = {Forests play a key role in the carbon balance of terrestrial ecosystems. One of the main uncertainties in global change predictions lies in how the spatiotemporal dynamics of forest productivity will be affected by climate warming. Here we show an increasing influence of climate on the spatial variability of tree growth during the last 120 y, ultimately leading to unprecedented temporal coherence in ring-width records over wide geographical scales (spatial synchrony). Synchrony in growth patterns across cold-constrained (central Siberia) and drought-constrained (Spain) Eurasian conifer forests have peaked in the early 21st century at subcontinental scales (similar to 1,000 km). Such enhanced synchrony is similar to that observed in trees co-occurring within a stand. In boreal forests, the combined effects of recent warming and increasing intensity of climate extremes are enhancing synchrony through an earlier start of wood formation and a stronger impact of year-to-year fluctuations of growing-season temperatures on growth. In Mediterranean forests, the impact of warming on synchrony is related mainly to an advanced onset of growth and the strengthening of drought-induced growth limitations. Spatial patterns of enhanced synchrony represent early warning signals of climate change impacts on forest ecosystems at subcontinental scales.}, author = {Shestakova, . and Gutierrez, Emilia and Kirdyanov, . and , Jesus and Genova, Mar and Knorre, . and , Juan and , Victor and Sanchez-Salguero, Raul and Voltas, Jordi}, doi = {10.1073/pnas.1514717113}, issn = {0027-8424}, journal = {PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA}, month = {JAN 19}, number = {3}, orcid-numbers = {Shestakova, Tatiana/0000-0002-5605-0299 de Dios, Victor Resco/0000-0002-5721-1656 Camarero, J. Julio/0000-0003-2436-2922 Linares, J/0000-0001-8375-6353 Gutierrez, Emilia/0000-0002-6085-5700 A, Kirdyanov V/0000-0002-6797-4964 Voltas, Jordi/0000-0003-4051-1158}, pages = {662-667}, researcherid-numbers = {Shestakova, Tatiana/S-4198-2016 de Dios, Victor Resco/AAH-3655-2019 Camarero, J. Julio/A-8602-2013 Linares, Juan Carlos/G-3474-2011 Gutierrez, Emilia/O-7568-2014 A, Kirdyanov V/J-6789-2013 Voltas, Jordi/D-3280-2013 Genova, Mar/R-8395-2018 Voltas, Jordi/N-9587-2019}, title = {Forests synchronize their growth in contrasting Eurasian regions in response to climate warming}, unique-id = {ISI:000368458800058}, volume = {113}, year = {2016} } Megscammell/METOD-Algorithm0 \begin{tabular}{lrrrr} \toprule {} & 0 & 1 & 2 & 3 \\ \midrule 0 & 95.0 & 5.0 & 0.0 & 1.0 \\ 1 & 97.0 & 3.0 & 0.0 & 1.0 \\ 2 & 97.0 & 3.0 & 0.0 & 1.0 \\ 3 & 96.0 & 4.0 & 0.0 & 1.0 \\ 4 & 97.0 & 3.0 & 0.0 & 1.0 \\ 5 & 98.0 & 2.0 & 0.0 & 1.0 \\ \bottomrule \end{tabular} \documentclass{article} \usepackage[margin=1in]{geometry} \usepackage[outdir=./]{epstopdf} % Avoids errors when input figures \usepackage[labelsep=period,labelfont=bf]{caption} %\usepackage{subcaption} \usepackage{graphicx} %\graphicspath{{../Figures/LPs/LagDep-FX/Target/EM/}{../Figures/LPs/LagDep-FX/Path/EM/}{../Figures/LPs/LagDep-FX/LSAP/EM/}} \begin{document} \begin{figure}[tbph] \caption{Response of 2-Year Emerging Market Yield to U.S. Monetary Policy Surprises} \label{fig:LPEM2Y} \begin{center} \begin{minipage}{\linewidth} \begin{center} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/Target/EM/TargetEMnomyptpphi24m.eps} \\ \vspace{-0.35cm} \caption{Target Surprise: 2000-2008} \label{subfig:LPEM2Ytarget} \vspace{0.4cm} \end{subfigure} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/Path/EM/PathEMnomyptpphi24m.eps} \\ \vspace{-0.35cm} \caption{Forward Guidance Surprise: 2000-2019} \label{subfig:LPEM2Ypath} \end{subfigure} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/LSAP/EM/LSAPEMnomyptpphi24m.eps} \\ \vspace{-0.35cm} \caption{Asset Purchase Surprise: 2009-2019} \label{subfig:LPEM2Ylsap} \end{subfigure} \end{center} \fignotes{This figure shows the response following \cite{Jorda:2005} of the 2-year emerging market nominal yield and its components to U.S. monetary policy surprises. The nominal yield is decomposed into an expected future short-term interest rate (ER), a term premium (TP) and a credit risk premium (CRP). The target, forward guidance and asset purchase surprises are identified using high-frequency data around Fed's monetary policy announcements, see section \ref{sec:USMPS} for details.} \end{minipage} \end{center} \end{figure} \pagebreak[4] \begin{figure}[tbph] \caption{Response of 10-Year Emerging Market Yield to U.S. Monetary Policy Surprises} \label{fig:LPEM10Y} \begin{center} \begin{minipage}{\linewidth} \begin{center} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/Target/EM/TargetEMnomyptpphi120m.eps} \\ \vspace{-0.35cm} \caption{Target Surprise: 2000-2008} \label{subfig:LPEM10Ytarget} \vspace{0.4cm} \end{subfigure} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/Path/EM/PathEMnomyptpphi120m.eps} \\ \vspace{-0.35cm} \caption{Forward Guidance Surprise: 2000-2019} \label{subfig:LPEM10Ypath} \end{subfigure} \begin{subfigure}[t]{\linewidth} \includegraphics[trim={0cm 0cm 0cm 0cm},clip,height=0.24\textheight,width=\linewidth]{../Figures/LPs/LagDep-FX/LSAP/EM/LSAPEMnomyptpphi120m.eps} \\ \vspace{-0.35cm} \caption{Asset Purchase Surprise: 2009-2019} \label{subfig:LPEM10Ylsap} \end{subfigure} \end{center} \fignotes{This figure shows the response following \cite{Jorda:2005} of the 10-year emerging market nominal yield and its components to U.S. monetary policy surprises. The nominal yield is decomposed into an expected future short-term interest rate (ER), a term premium (TP) and a credit risk premium (CRP). The target, forward guidance and asset purchase surprises are identified using high-frequency data around Fed's monetary policy announcements, see section \ref{sec:USMPS} for details.} \end{minipage} \end{center} \end{figure} \end{document} % trim = { }% IUJ Hakkaisan Beamer Color Theme, ver. 0.01 (2016-10-04) % Copyright 2016 by <> % % This file may be distributed and/or modified % % 1. under the LaTeX Project Public License and/or % 2. under the GNU Public License. \ProvidesPackage{beamercolorthemehakkaisan}[10/04/2016] \mode \definecolor{iujblue}{RGB}{0,80,153} \definecolor{iujbrick}{RGB}{196,0,0} \definecolor{iujgray}{RGB}{100,100,100} \setbeamercolor*{normal text}{fg=black,bg=white} \setbeamercolor*{alerted text}{fg=iujbrick} \setbeamercolor*{example text}{fg=iujblue} \setbeamercolor*{structure}{fg=iujblue,bg=white} \setbeamerfont{alerted text}{series=\bfseries} \setbeamercolor*{palette primary}{fg=white,bg=iujblue} \setbeamercolor*{palette secondary}{fg=iujblue,bg=white} \setbeamercolor*{palette tertiary}{fg=black,bg=white} \setbeamercolor*{palette quaternary}{fg=black,bg=white} \setbeamercolor{titlelike}{fg=white, bg=iujblue} \setbeamercolor{frametitle}{fg=iujblue, bg=white} \setbeamercolor{frametitle right}{fg=iujgray, bg=white} \setbeamercolor{sidebar}{bg=iujblue} \setbeamercolor*{palette sidebar primary}{fg=black} \setbeamercolor*{palette sidebar secondary}{fg=black} \setbeamercolor*{palette sidebar tertiary}{fg=black} \setbeamercolor*{palette sidebar quaternary}{fg=black} \setbeamercolor*{item projected}{fg=black,bg=black!20} \setbeamercolor{block title}{fg=white,bg=iujblue} \setbeamercolor{block title alerted}{use=alerted text,fg=white,bg=iujbrick} \setbeamercolor{block title example}{use=example text,fg=white,bg=iujgray} \setbeamercolor{block body}{parent=normal text,use=block title,bg=block title.bg!15!bg} \setbeamercolor{block body alerted}{parent=normal text,use=block title alerted,bg=block title alerted.bg!15!bg} \setbeamercolor{block body example}{parent=normal text,use=block title example,bg=block title example.bg!15!bg} \setbeamercolor*{separation line}{} \setbeamercolor*{fine separation line}{} \mode spl0w/security-courses1-10 \documentclass[Screen16to9,17pt]{foils} \usepackage{zencurity-slides} \externaldocument{software-security-exercises} \selectlanguage{english} \begin{document} \mytitlepage {5. Web Application Security: Recon} {KEA Kompetence OB2 Software Security} \slide{Goals for today} \hlkimage{6cm}{thomas-galler-hZ3uF1-z2Qc-unsplash.jpg} Todays goals: \begin{list2} \item Web Application Hacking -- more structured knowledge about common web apps \item Do reconnaissance on web applications \item Talk about common patterns \end{list2} Photo by on Unsplash \slide{Plan for today} \begin{list1} \item Subjects \item Web Application Security: Recon \begin{list2} \item Structured approach to discovery, information gathering, mapping \item The Structure of a Modern Web Application \item REST APIs, JSON \item Authentication and Authorization Systems \item Finding domains, subdomains and vhosts \end{list2} \item Exercises \begin{list2} \item Run small programs -- notice how libraries can take aways complexity \item Scaning real life sites with Nikto and Whatweb, your sites, my sites and JuiceShop \end{list2} \end{list1} \slide{Reading Summary} \emph{Web Application Security}, , 2020, ISBN: 9781492053118 \begin{list1} \item Part I. Recon, chapters 1-8, very short chapters \item 1. The History of Software Security \item 2. Introduction to Web Application Reconnaissance \item 3. The Structure of a Modern Web Application \item 4. Finding Subdomains \item 5. API Analysis \item 6. Identifying Third-Party Dependencies \item 7. Identifying Weak Points in Application Architectur \item 8. Part I Summary \end{list1} \slide{Goals: Discovery and Reconnaissance} \hlkimage{8cm}{homer-end-is-near.jpg} \begin{list1} \item Be able to discover parts of web application environments \item Perform Information Gathering Web Application Mapping \end{list1} \slide{Selecting Technologies for your enterprise} \hlkimage{14cm}{software.pdf} \slide{Why talk about selecting technologies } \hlkimage{7cm}{johnny_automatic_blueprints.png} \begin{list2} \item A big part of systems integration it to make sure applications can work together \item Data interchange \item Running systems require skills, many different technologies, many humans needed \item Managing complexity with many systems become harder \end{list2} Later today we will discuss this subject more with the hand-in assigment \slide{Secure Infrastructures starts with architecture and design} \hlkimage{3cm}{secure_coding.png} %\vskip 2cm %\hlkrightimage{5cm}{secure_coding.png} {\emph{Secure Coding: Principles and Practices} af , 2003} \begin{list2} \item Architecture and design while you are thinking about the application \item Implementation while you are writing the application \item Operations After the application is in production \item Approx. 200 pages, but very dense with information. \end{list2} \slide{Operating Systems} \hlkimage{4cm}{tux.jpg} \begin{list2} \item Applications need to run within some controlled system \item What is an operating system today? \item Is Docker an operating system? What is Docker? \end{list2} \slide{Use the Modern Operating Systems} \begin{list1} \item Newer versions of Microsoft Windows, Mac OS X and Linux \begin{list2} \item Buffer overflow protection \item Stack protection, non-executable stack \item Heap protection, non-executable heap \item \emph{Randomization of parameters} stack gap m.v. \end{list2} \item Note: these still have errors and bugs, but are better than older versions \item Check end-of-life and when updates will stop for each version \item OpenBSD has shown the way in many cases\\ \link{http://www.openbsd.org/papers/} \end{list1} \vskip 1cm \centerline{Always try to make life worse and more costly for attackers} \slide{Technologies used in enterprises} The following tools and environments are examples that are used in enterprises today: \begin{list2} \item Programming languages and frameworks Java, Spring, Python \item Development environments IDE NetBeans / Eclipse / IntelliJ, Atom \item Systems for running Java: TomCat / GlassFish \item Networking and network protocols: TCP/IP, HTTP, DNS \item Formats XML, JSON, WSDL, GRPC, msgpack, protobuf, apache thrift \item Web technologies and services: REST, API, HTML5, CSS %\item BPL, UML \item Tools like cURL, Git and Github \item Integration tools Camel \item Message queueing systems: MQ \item Aggregated example platforms: Elastic stack, logstash, elasticsearch, kibana, grafana \item Cloud and virtualisation Docker, Kubernetes, Azure, AWS, microservices \end{list2} \centerline{This list is not complete, but what was discussed in System Integration course!} \slide{What about dependencies} \hlkimage{4cm}{kyler-trautner-693525-unsplash.jpg} \begin{list2} \item Are you using some special software, or hardware \item Does your application depend on some tools, library that needs help \end{list2} \slide{Data overview XML data, JSON} \hlkimage{15cm}{chris-lawton-5IHz5WhosQE-unsplash.jpg} Photo by on Unsplash \slide{XML data} \begin{quote} Extensible Markup Language (XML) is a markup language that defines a set of rules for encoding documents in a format that is both human-readable and machine-readable. The World Wide Web Consortium's XML 1.0 Specification[2] of 1998[3] and several other related specifications[4]—all of them free open standards—define XML.[5] The design goals of XML emphasize simplicity, generality, and usability across the Internet.[6] It is a textual data format with strong support via Unicode for different human languages. Although the design of XML focuses on documents, the language is widely used for the representation of arbitrary data structures[7] such as those used in web services. \end{quote} Source: \url{https://en.wikipedia.org/wiki/XML} \begin{list2} \item We have seen XML used for configuration in Apache Tomcat and Camel \item Perfect for computers, less for humans \end{list2} \slide{XML data example - Nmap output} \begin{minted}[fontsize=\footnotesize]{xml}
\end{minted} \slide{XML data - documents} \begin{quote} Hundreds of document formats using XML syntax have been developed,[8] including RSS, Atom, SOAP, SVG, and XHTML. XML-based formats have become the default for many office-productivity tools, including Microsoft Office (Office Open XML), OpenOffice.org and LibreOffice (OpenDocument), and Apple's iWork[citation needed]. XML has also provided the base language for communication protocols such as XMPP. Applications for the Microsoft .NET Framework use XML files for configuration, and property lists are an implementation of configuration storage built on XML.[9] \end{quote} Source: \url{https://en.wikipedia.org/wiki/XML} \begin{list2} \item Document formats using XML may still be proprietary! \item Documents using XML can be validated, are they well-formed according to the Document Type Definition (DTD) \end{list2} \slide{Transforming XML using XSLT} \begin{quote} XSLT (Extensible Stylesheet Language Transformations) is a language for transforming XML documents into other XML documents,[1] or other formats such as HTML for web pages, plain text or XSL Formatting Objects, which may subsequently be converted to other formats, such as PDF, PostScript and PNG.[2] XSLT 1.0 is widely supported in modern web browsers.[3] \end{quote} Source: \url{https://en.wikipedia.org/wiki/XSLT} \begin{list2} \item Can be seen as a mapping between formats, different XML schemas \item Also is Turing complete, is a programming language \end{list2} \slide{XSLT example} \begin{minted}[fontsize=\footnotesize]{xml} \end{minted} \begin{list2} \item XSLT uses XPath to identify subsets of the source document tree and perform calculations. XPath also provides a range of functions \item XSLT functionalities overlap with those of XQuery, which was initially conceived as a query language for large collections of XML documents\\ Source: \url{https://en.wikipedia.org/wiki/XSLT} \end{list2} \slide{xsltproc example using Nmap} \begin{alltt}\footnotesize $ su - # apt install nmap xsltproc # nmap -sP -oA /tmp/router 192.168.3.11 # exit $ xsltproc /tmp/router.xml > /tmp/router.html $ firefox /tmp/router.html \end{alltt} \begin{list2} \item We can use the command line tool \verb+xlstproc+ for transforming documents \item \verb+apt install xsltproc+ \item Its part of the package Libxslt \url{https://en.wikipedia.org/wiki/Libxslt} \vskip 2cm \item Try installing the tools Nmap and \verb+xsltproc+ and reproduce the above \item This is an easy tool to try, and very useful too \end{list2} \slide{Data overview JSON} \begin{quote} JavaScript Object Notation (JSON, pronounced /ˈdʒeɪsən/; also /ˈdʒeɪˌsɒn/[note 1]) is an open-standard file format or data interchange format that uses {\bf human-readable text} to transmit data objects consisting of attribute–value pairs and array data types (or any other serializable value). It is a very common data format, with a diverse range of applications, such as serving as replacement for XML in AJAX systems.[6] \end{quote} Source: \url{https://en.wikipedia.org/wiki/JSON} \begin{list2} \item I like JSON much better than XML \item Many web services can supply data in JSON format \end{list2} \slide{JSON example} \begin{minted}[fontsize=\footnotesize]{json} { "first name": "John", "last name": "Smith", "age": 25, "address": { "street address": "21 2nd Street", "city": "New York", "state": "NY", "postal code": "10021" }, "phone numbers": [ { "type": "home", "number": "212 555-1234" }, ], } \end{minted} \begin{list2} \item This is a basic JSON document, new data attribute-value pairs can be added\\ Source: \url{https://en.wikipedia.org/wiki/JSON} \end{list2} \slide{Note about frameworks and libraries} \begin{minted}[fontsize=\footnotesize]{python} import xml.etree.ElementTree as ET tree = ET.parse('testfile.xml') root = tree.getroot() print(root.tag) print('Nmap version: \t\t{:s} '.format(root.attrib['version'])) print('Nmap started: \t\t{:s} '.format(root.attrib['startstr'])) print('Nmap command line: \t{:s} '.format(root.attrib['args'])) hosts = tree.findall('./host') for host in hosts: print(host.tag) print(host.attrib) for hostvalues in host: print(hostvalues.tag) print(hostvalues.attrib) \end{minted} \begin{list2} \item Dont import JSON or XML using home made programs \item Example uses xml.etree.ElementTree from Python\\ \url{https://docs.python.org/3.7/library/xml.etree.elementtree.html} \end{list2} \slide{Convert XML to JSON} \begin{minted}[fontsize=\footnotesize]{python} import xml.etree.ElementTree as ET import json def etree_to_dict(t): d = {t.tag : map(etree_to_dict, t.getchildren())} d.update(('@' + k, v) for k, v in t.attrib.items()) d['text'] = t.text return d tree = ET.parse('testfile.xml') root = tree.getroot() mydict = etree_to_dict(root) print(type(tree)) print(type(root)) print(type(mydict)) print(mydict) with open('testfile.json', 'w') as json_file: json.dump(mydict, json_file) \end{minted} Converting using Python is easy \slide{Web services SOAP, WSDL} \hlkimage{15cm}{josefina-di-battista-hB25vJNTnNQ-unsplash.jpg} Photo by on Unsplash \slide{Web Services} The term Web service (WS) is either: \begin{list2} \item a service offered by an electronic device to another electronic device, communicating with each other via the World Wide Web, or \item a server running on a computer device, listening for requests at a particular port over a network, serving web documents (HTML, JSON, XML, images), and creating web applications services, which serve in solving specific domain problems over the Web (WWW, Internet, HTTP) \end{list2} Source: \url{https://en.wikipedia.org/wiki/Web_service} \begin{list2} \item Today a generic name for services using the internet \item Web servers such as Apache HTTPD, Nginx etc. provide a service to thew internet allowing access using HTTP \item Source for some parts on this slide, \url{https://en.wikipedia.org/wiki/Web_service} \end{list2} \slide{W3C Web Services} \begin{quote} A web service is a software system designed to support interoperable machine-to-machine interaction over a network. It has an interface described in a machine-processable format (specifically WSDL). Other systems interact with the web service in a manner prescribed by its description using SOAP-messages, typically conveyed using HTTP with an XML serialization in conjunction with other web-related standards. \end{quote} Source -- W3C, Web Services Glossary[3] \slide{WSDL - Web Services Description Language} \begin{quote} The Web Services Description Language (WSDL /ˈwɪz dəl/) is an XML-based interface description language that is used for describing the functionality offered by a web service. The acronym is also used for any specific WSDL description of a web service (also referred to as a WSDL file), which provides a machine-readable description of how the service can be called, what parameters it expects, and what data structures it returns. Therefore, its purpose is roughly similar to that of a type signature in a programming language. The current version of WSDL is WSDL 2.0. The meaning of the acronym has changed from version 1.1 where the "D" stood for "Definition". \end{quote} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} %\begin{list2} %\item %\item %\end{list2} \slide{WSDL XML} \begin{minted}[fontsize=\footnotesize]{xml} This is a sample WSDL 2.0 document. \end{minted} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} \slide{WSDL XML types} \begin{minted}[fontsize=\footnotesize]{xml} ... ... \end{minted} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} Types Describes the data. The XML Schema language (also known as XSD) is used (inline or referenced) for this purpose. \slide{WSDL XML interfaces} \begin{minted}[fontsize=\footnotesize]{xml} \end{minted} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} Interface Defines a Web service, the operations that can be performed, and the messages that are used to perform the operation. \slide{WSDL XML the binding} \begin{minted}[fontsize=\footnotesize]{xml} \end{minted} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} Binding Specifies the interface and defines the SOAP binding style (RPC/Document) and transport (SOAP Protocol). The binding section also defines the operations. \slide{WSDL XML the Service} \begin{minted}[fontsize=\footnotesize]{xml} \end{minted} Source: \url{https://en.wikipedia.org/wiki/Web_Services_Description_Language} Service Contains a set of system functions that have been exposed to the Web-based protocols. \slide{SOAP - Simple Object Access Protocol} \begin{quote} SOAP (abbreviation for Simple Object Access Protocol) is a messaging protocol specification for exchanging structured information in the implementation of web services in computer networks. Its purpose is to provide extensibility, neutrality, verbosity and independence. It uses XML Information Set for its message format, and relies on application layer protocols, most often Hypertext Transfer Protocol (HTTP), although some legacy systems communicate over Simple Mail Transfer Protocol (SMTP), for message negotiation and transmission. \end{quote} Source: \url{https://en.wikipedia.org/wiki/SOAP} Utilizes UDDI (Universal Description, Discovery, and Integration) \slide{Web Service Explained } \begin{quote} The term "Web service" describes a standardized way of integrating Web-based applications using the XML, SOAP, WSDL and UDDI open standards over an Internet Protocol backbone. XML is the data format used to contain the data and provide metadata around it, SOAP is used to transfer the data, WSDL is used for describing the services available and UDDI lists what services are available. \end{quote} Source:\url{https://en.wikipedia.org/wiki/Web_service} \slide{Again frameworks and libraries} A simple “Hello World” http SOAP server: \begin{minted}[fontsize=\footnotesize]{python} import SOAPpy def hello(): return "Hello World" server = SOAPpy.SOAPServer(("localhost", 8080)) server.registerFunction(hello) server.serve_forever() \end{minted} And the corresponding client: \begin{minted}[fontsize=\footnotesize]{python} import SOAPpy server = SOAPpy.SOAPProxy("http://localhost:8080/") print server.hello() \end{minted} \begin{list2} \item Dont process SOAP manually using home made programs \item \url{https://pypi.org/project/SOAPpy/} \end{list2} \slide{SOAP Request } \begin{minted}[fontsize=\footnotesize]{xml} // HTTP here POST / HTTP/1.0 Host: localhost:8080 User-agent: SOAPpy xxx (pywebsvcs.sf.net) Content-type: text/xml; charset=UTF-8 Content-length: 340 SOAPAction: "hello" // SOAP start here \end{minted} \slide{SOAP Response} \begin{minted}[fontsize=\footnotesize]{xml} // HTTP here HTTP/1.0 200 OK Server: SOAPpy xxx (Python 2.7.16) Date: Mon, 24 Feb 2020 13:29:44 GMT Content-type: text/xml; charset=UTF-8 Content-length: 510 // SOAP start here Hello World \end{minted} \slide{REST Service} \hlkimage{9cm}{soabook-9-1-REST.png} \begin{list2} \item Very typical REST URL/method \verb+GET /invoice/{invoice-id}+ \item See also \link{https://en.wikipedia.org/wiki/Representational_state_transfer} \end{list2} Source: {\footnotesize\\ \emph{Service‑Oriented Architecture: Analysis and Design for Services and Microservices}, , 2017} \slide{10.1 RESTful services} \hlkimage{13cm}{camelbook-10-1-restful.png} \begin{quote} RESTful services, also known as REST services, has become a popular architectural style used in modern enterprise projects. REST was defined by Roy Fielding in 2000 when he published his paper, and today REST is a foundation that drives the modern APIs on the web. You can also think of it as a modern web service, in which the APIs are RESTful and HTTP based so they’re easily consumable on the web. \end{quote} Source: {\footnotesize\\ \emph{Camel in action}, and , 2018} \slide{Python and REST} \inputminted{python}{programs/rest-1.py} \begin{list2} \item Lets try to use some Python to access a REST service. \item We will use the JSONPlaceholder which is a free online REST API: \link{https://jsonplaceholder.typicode.com/} \item Start at the site: \link{https://jsonplaceholder.typicode.com/guide.html} and try running a few of the examples with your browser \item Then try using the same URLS in the Requests HTTP library from Python,\\ \link{https://requests.readthedocs.io/en/master/} \end{list2} \slide{Shared Database} \hlkimage{7cm}{SharedDatabaseIntegration.png} Shared Database — Have the applications store the data they wish to share in a common database. Common systems and technologies used: \begin{list2} \item database management system (DBMS) using Structured Query Language (SQL), relational database examples:\\ \item PostgresSQL, Oracle DM, Microsoft SQL, MySQL \url{https://en.wikipedia.org/wiki/SQL} \item NoSQL databases has been a new input with examples like: MongoDB, CouchDB, Redis, RIAK\\ \url{https://en.wikipedia.org/wiki/NoSQL} \end{list2} \slide{Elastic stack Kibana} \hlkimage{10cm}{illustrated-screenshot-hero-kibana.png} Screenshot from \url{https://www.elastic.co/kibana} Elasticsearch is a search engine and ocument store used in a lot of different systems, allowing cross application integration. \slide{Getting started with Elastic Stack} Easy to get started using the tutorial \emph{Getting started with the Elastic Stack} :\\ {\footnotesize\url{https://www.elastic.co/guide/en/elastic-stack-get-started/current/get-started-elastic-stack.html}} Today Elastic Stack contains lots of different parts. \begin{list2} \item Elasticsearch - the core engine \item Logstash - a tool for parsing logs and other data.\\ \url{https://www.elastic.co/logstash} \begin{quote} "Logstash dynamically ingests, transforms, and ships your data regardless of format or complexity. Derive structure from unstructured data with grok, decipher geo coordinates from IP addresses, anonymize or exclude sensitive fields, and ease overall processing." \end{quote} \item Kibana - a web application for accessing and working with data in Elasticsearch\\ \url{https://www.elastic.co/kibana} \end{list2} \slide{TCP/IP and External Data Representation 1987} \begin{quote} External Data Representation (XDR) is a standard data serialization format, for uses such as computer network protocols. It allows data to be transferred between different kinds of computer systems. Converting from the local representation to XDR is called encoding. Converting from XDR to the local representation is called decoding. \end{quote} \hlkrightpic{14cm}{-2cm}{compare-osi-ip.png} Source: {\footnotesize\\ \link{https://en.wikipedia.org/wiki/External_Data_Representation}\\ \link{https://tools.ietf.org/html/rfc1014}} \slide{Designing and Standardizing HTTP Response Codes} \begin{list2} \item 100-199 are informational codes used as low-level signaling mechanisms, such as a confirmation of a request to change protocols \item 200-299 are general success codes used to describe various kinds of success conditions \item 300-399 are redirection codes used to request that the consumer retry a request to a different resource identifier, or via a different intermediary \item 400-499 represent consumer-side error codes that indicate that the consumer has produced a request that is invalid for some reason, example 404 file not found \item 500-599 represent service-side error codes that indicate that the consumer’s request may have been valid but that the service has been unable to process it for internal reasElasticsearch exposes REST APIs that are used by the UI components and can be called directly to configure and access Elasticsearch features.ons. \end{list2} Source: {\footnotesize\\ \emph{Service‑Oriented Architecture: Analysis and Design for Services and Microservices}, , 2017} \slide{Exercise: Communicate with HTTP} Try this - use netcat/ncat, available in Nmap package from \url{Nmap.org}: \begin{alltt} \small $ {\bf netcat www.zencurity.com 80 GET / HTTP/1.0} HTTP/1.1 200 OK Server: nginx Date: Sat, 01 Feb 2020 20:30:06 GMT Content-Type: text/html Content-Length: 0 Last-Modified: Thu, 04 Jan 2018 15:03:08 GMT Connection: close ETag: "5a4e422c-0" Referrer-Policy: no-referrer Accept-Ranges: bytes ... \end{alltt} \slide{Basic test tools TCP - Telnet and OpenSSL} \begin{list1} \item Telnet used for terminal connections over TCP, but is clear-text \item Telnet can be used for testing connections to many older protocols which uses text commands \begin{list2} \item \verb+telnet mail.kramse.dk 25+ create connection to port 25/tcp \item \verb+telnet www.kramse.dk 80+ create connection to port 80/tcp \end{list2} \item Encrypted connections often use TLS and can be tested using OpenSSL command line tool \verb+openssl+ \begin{list2} \item \verb+openssl s_client -host www.kramse.dk -port 443+\\ create connection to port 443/tcp with TLS \item \verb+openssl s_client -host mail.kramse.dk -port 993+\\ create connection to port 993/tcp with TLS \end{list2} \item Using OpenSSL in client-mode allows the use of the same commands like Telnet after connection \end{list1} \slide{Book: Practical Packet Analysis (PPA)} \hlkimage{6cm}{PracticalPacketAnalysis3E_cover.png} \emph{Practical Packet Analysis, Using Wireshark to Solve Real-World Network Problems} by , 3rd Edition April 2017, 368 pp. ISBN-13: 978-1-59327-802-1 \link{https://nostarch.com/packetanalysis3} \exercise{ex:nikto-webscanner} \exercise{ex:whatweb-scanner} \slide{Git intro} \begin{quote} Git (/ɡɪt/)[7] is a distributed version-control system for tracking changes in source code during software development.[8] It is designed for coordinating work among programmers, but it can be used to track changes in any set of files. Its goals include speed,[9] data integrity,[10] and support for distributed, non-linear workflows.[11] \end{quote} Source: \url{https://en.wikipedia.org/wiki/Git} \begin{list2} \item We will introduce Git and Github - commercial Git hosting company\\ \url{https://en.wikipedia.org/wiki/Git} \item Try creating a Git repository, remember to add a README \item Checkout the repository \item Edit a file \item add and commit it \end{list2} Use the Github Hello World example: \url{https://guides.github.com/activities/hello-world/} \exercise{ex:postman-api} \slidenext{} \end{document} %%% Watermark for draft \usepackage{draftwatermark} \def\watermarkloaded{1} @inproceedings{Marwan2019ITiCSE, author = { Lytle, , . and }, booktitle = {Proceedings of the Annual Conference on Innovation and Technology in Computer Science Education}, file = {:C$\backslash$:/Users/twprice/AppData/Local/Mendeley Ltd./Mendeley Desktop/Downloaded/Marwan et al. - 2019 - The Impact of Adding Textual Explanations to Next-step Hints in a Novice Programming Environment.pdf:pdf}, keywords = {Peer Reviewed Conference Paper (Full),computer,intelligent tutoring systems,next step hints,textual hints}, mendeley-groups = {CAREER2020/HINTS Lab}, mendeley-tags = {Peer Reviewed Conference Paper (Full)}, pages = {1--7}, title = {The Impact of Adding Textual Explanations to Next-step Hints in a Novice Programming Environment}, year = {2019} } \documentclass{article} \usepackage{amsmath} \usepackage[margin=1.0in]{geometry} \usepackage{xcolor} \begin{document} \noindent Does $\displaystyle \sum_{n=1}^\infty \frac{n^2+1}{n^3+7}$ diverge, converge absolutely, or converge conditionally? \subsection*{Solution 1} $\displaystyle \sum_{n=1}^\infty \frac1n$ is a $p$-series with $p=1$. Since $p \leq 1$, the series $\displaystyle \sum_{n=1}^\infty \frac1n$ diverges by the $p$-series test. If we use $a_n = \frac{n^2+1}{n^3+7}$ and $b_n = \frac1n$, then \begin{align*} \lim_{n \to \infty} \frac{a_n}{b_n} &= \lim_{n \to \infty} \frac{n^2+1}{n^3+7} \cdot \frac{n}{1}\\ &= \lim_{n \to \infty} \frac{n^3+n}{n^3+7} \\ &= \lim_{n \to \infty} \frac{3n^2+1}{3n^2} \text{ using L'hopital}\\ &= \lim_{n \to \infty} \frac{6n}{6n} \text{ using L'hopital}\\ &= \lim_{n \to \infty} 1\\ &= 1 \end{align*} So by the Limit Comparison Test, the series $\displaystyle \sum_{n=1}^\infty \frac{n^2+1}{n^3+7}$ diverges. \subsection*{Solution 2} We are looking for a fixed $K > 0$ such that \[ K \cdot \frac{n^2+1}{n^3+7} \geq \frac1n\] By multiplying both sides by $n(n^3+7)$, we have \[ K(n^3 + n) \geq n^3+7\] so \[ Kn^3 + Kn \geq n^3+7\] so if we pick $K=8$, then we'd have \[ 8n^3 + 8n \geq n^3+7\] is likely true because \[ n^3+ 7n^3 + 8n \geq n^3+7\] with the $n^3$'s compared together, and $7n^3$ is greater than $7$. So we are ready to present starting from true inequalities. With the scratch work above done, since $7n^3 \geq 7$ for $n \geq 1$, we have \[ n^3+ 7n^3 + 8n \geq n^3+7\] \[ 8n^3 + 8n \geq n^3+7\] \[ 8(n^3 + n) \geq n^3+7\] and dividing both sides by $8n(n^3+7)$ we get \[\frac{n^2+1}{n^3+7} \geq \frac1{8n}\] Since $\displaystyle \sum_{n=1}^\infty \frac1{8n} = \frac18\sum_{n=1}^\infty \frac1{n}$ diverges by the $p$-test, the series $\displaystyle \sum_{n=1}^\infty \frac{n^2+1}{n^3+7}$ diverges by the Direct Comparison Test. \end{document}%%%%%%%%%%%%%%%%% \begin{align*} L&=\lim_{n \to \infty} \sqrt[n]{|a_n|}\\ &= \lim_{n \to \infty} \sqrt[n]{\left| \right|}\\ \end{align*} \begin{align*} L&=\lim_{n \to \infty} \left|\frac{a_{n+1}}{a_n}\right|\\ &= \lim_{n \to \infty} \left| \right|\\ \end{align*} \begin{align*} \lim_{n \to \infty} a_n &= \lim_{n \to \infty} \\ \end{align*} Since $\sum |a_n| = \sum a_n$, the series $\displaystyle \sum_{n=1}^\infty AAAAAAAAAAAAAA$ converges absolutely. Since $|r| < 1$, the series ... converges by the Geometric Series Test. Since $|r| \geq 1$, the series ... diverges by the Geometric Series Test. The function $f(x)=\frac{}{}$ is continuous, positive, and decreasing on $[1,\infty)$. \subsection*{Solution} % ===> this file was generated automatically by noweave --- better not edit it \section{Introduction} The {\Tt{}netdraw-mlab\nwendquote} module implements a Matlab drawing context. It is not the Matlab interface to the drawing system -- that is implemented in {\Tt{}netdraw-mex\nwendquote}. \section{Interface} \nwfilename{netdraw-mlab.nw}\nwbegincode{1}\sublabel{NW2c68Jw-4H9n7s-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-4H9n7s-1}}}\moddef{netdraw-mlab.h~{\nwtagstyle{}\subpageref{NW2c68Jw-4H9n7s-1}}}\endmoddef\nwstartdeflinemarkup\nwenddeflinemarkup #ifndef NETDRAW_MLAB_H #define NETDRAW_MLAB_H #include #include void netdraw_mlab(mesh_t mesh, double* x); #endif /* NETDRAW_MLAB_H */ \nwnotused{netdraw-mlab.h}\nwendcode{}\nwbegindocs{2}\nwdocspar \section{Implementation} \nwenddocs{}\nwbegincode{3}\sublabel{NW2c68Jw-2Czttk-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-2Czttk-1}}}\moddef{netdraw-mlab.c~{\nwtagstyle{}\subpageref{NW2c68Jw-2Czttk-1}}}\endmoddef\nwstartdeflinemarkup\nwenddeflinemarkup #include #include #include #include #include #include #include #include \LA{}static functions~{\nwtagstyle{}\subpageref{NW2c68Jw-1duChy-1}}\RA{} \LA{}functions~{\nwtagstyle{}\subpageref{NW2c68Jw-nRuDO-1}}\RA{} \nwnotused{netdraw-mlab.c}\nwendcode{}\nwbegindocs{4}\nwdocspar \subsection{Drawing beams} \nwenddocs{}\nwbegincode{5}\sublabel{NW2c68Jw-1duChy-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-1duChy-1}}}\moddef{static functions~{\nwtagstyle{}\subpageref{NW2c68Jw-1duChy-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-2Czttk-1}}\nwprevnextdefs{\relax}{NW2c68Jw-1duChy-2}\nwenddeflinemarkup static void draw_beam(void* pself, double* transform, double l, double w, double h, int* vindex) \{ double* x = (double*) pself; double* xlocal; int i; mxArray* prhs[5]; prhs[0] = mxCreateDoubleMatrix(3, 4, mxREAL); memcpy(mxGetPr(prhs[0]), transform, 12*sizeof(double)); prhs[1] = mx_from_double(l); prhs[2] = mx_from_double(w); prhs[3] = mx_from_double(h); prhs[4] = mxCreateDoubleMatrix(12, 1, mxREAL); xlocal = mxGetPr(prhs[4]); if (x) \{ for (i = 0; i < 12; ++i) xlocal[i] = x[vindex[i]]; for (i = 0; i < 12; i += 3) affine_apply_AT(transform, xlocal + i); \} else memset(xlocal, 0, 12*sizeof(double)); mexCallMATLAB(0, NULL, 5, prhs, "cho_draw_mlab_beam"); \} \nwalsodefined{\\{NW2c68Jw-1duChy-2}}\nwused{\\{NW2c68Jw-2Czttk-1}}\nwendcode{}\nwbegindocs{6}\nwdocspar \nwenddocs{}\nwbegincode{7}\sublabel{NW2c68Jw-4NFrDv-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-4NFrDv-1}}}\moddef{cho-draw-mlab-beam.m~{\nwtagstyle{}\subpageref{NW2c68Jw-4NFrDv-1}}}\endmoddef\nwstartdeflinemarkup\nwenddeflinemarkup function cho_draw_mlab_beam(T, L, W, H, q) resolution = 20; \LA{}get local coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-3Dkknx-1}}\RA{} \LA{}compute rotated corner displacements~{\nwtagstyle{}\subpageref{NW2c68Jw-3H1gql-1}}\RA{} \LA{}build interpolants in local coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-3kyIT3-1}}\RA{} \LA{}build interpolants in global coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-4f5wS-1}}\RA{} \LA{}plot the surfaces~{\nwtagstyle{}\subpageref{NW2c68Jw-26zJHS-1}}\RA{} \LA{}hermite cubic function~{\nwtagstyle{}\subpageref{NW2c68Jw-1buIps-1}}\RA{} \nwnotused{cho-draw-mlab-beam.m}\nwendcode{}\nwbegindocs{8}\nwdocspar \nwenddocs{}\nwbegincode{9}\sublabel{NW2c68Jw-3Dkknx-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-3Dkknx-1}}}\moddef{get local coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-3Dkknx-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup x1 = q(1); y1 = q(2); z1 = q(3); rx1 = q(4); ry1 = q(5); rz1 = q(6); x2 = q(7); y2 = q(8); z2 = q(9); rx2 = q(10); ry2 = q(11); rz2 = q(12); \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{10}\nwdocspar \nwenddocs{}\nwbegincode{11}\sublabel{NW2c68Jw-3H1gql-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-3H1gql-1}}}\moddef{compute rotated corner displacements~{\nwtagstyle{}\subpageref{NW2c68Jw-3H1gql-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup A = [0 0 0 0; W/2 W/2 -W/2 -W/2; -H/2 H/2 H/2 -H/2]; B = A; A = rot2local(rx1, ry1, rz1)' * A; B = rot2local(rx2, ry2, rz2)' * B; \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{12}\nwdocspar \nwenddocs{}\nwbegincode{13}\sublabel{NW2c68Jw-3kyIT3-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-3kyIT3-1}}}\moddef{build interpolants in local coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-3kyIT3-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup s = linspace(0, L, resolution); hx = x1 + (1 + (x2-x1)/L)*s; hy = hermite_cubic(L, y1,rz1, y2,rz2, s); hz = hermite_cubic(L, z1,-ry1, z2,-ry2, s); \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{14}\nwdocspar \nwenddocs{}\nwbegincode{15}\sublabel{NW2c68Jw-4f5wS-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-4f5wS-1}}}\moddef{build interpolants in global coordinates~{\nwtagstyle{}\subpageref{NW2c68Jw-4f5wS-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup for corner = 1 : 4 px = A(1,corner) + (B(1,corner)-A(1,corner))*s/L + hx; py = A(2,corner) + (B(2,corner)-A(2,corner))*s/L + hy; pz = A(3,corner) + (B(3,corner)-A(3,corner))*s/L + hz; p = T(1:3,1:3) * [px; py; pz]; u(corner,:) = p(1,:) + T(1,4); v(corner,:) = p(2,:) + T(2,4); w(corner,:) = p(3,:) + T(3,4); end \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{16}\nwdocspar \nwenddocs{}\nwbegincode{17}\sublabel{NW2c68Jw-26zJHS-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-26zJHS-1}}}\moddef{plot the surfaces~{\nwtagstyle{}\subpageref{NW2c68Jw-26zJHS-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup X = [u(1:4,:); u(1,:)]; Y = [v(1:4,:); v(1,:)]; Z = [w(1:4,:); w(1,:)]; surfl(X,Y,Z); \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{18}\nwdocspar \nwenddocs{}\nwbegincode{19}\sublabel{NW2c68Jw-1buIps-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-1buIps-1}}}\moddef{hermite cubic function~{\nwtagstyle{}\subpageref{NW2c68Jw-1buIps-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-4NFrDv-1}}\nwenddeflinemarkup function [f] = hermite_cubic(L, f0, f00, fL, fLL, s) f0L = (fL - f0)/L; f00L = (f0L - f00)/L; f0LL = (fLL - f0L)/L; f00LL = (f0LL - f00L)/L; f = f0 + s.*(f00 + s.*(f00L + (s-L).*f00LL)); \nwused{\\{NW2c68Jw-4NFrDv-1}}\nwendcode{}\nwbegindocs{20}\nwdocspar \subsection{Get displacement vector} \nwenddocs{}\nwbegincode{21}\sublabel{NW2c68Jw-1duChy-2}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-1duChy-2}}}\moddef{static functions~{\nwtagstyle{}\subpageref{NW2c68Jw-1duChy-1}}}\plusendmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-2Czttk-1}}\nwprevnextdefs{NW2c68Jw-1duChy-1}{\relax}\nwenddeflinemarkup static double* get_x(void* pself) \{ return (double*) pself; \} \nwused{\\{NW2c68Jw-2Czttk-1}}\nwendcode{}\nwbegindocs{22}\nwdocspar \subsection{Main routine} \nwenddocs{}\nwbegincode{23}\sublabel{NW2c68Jw-nRuDO-1}\nwmargintag{{\nwtagstyle{}\subpageref{NW2c68Jw-nRuDO-1}}}\moddef{functions~{\nwtagstyle{}\subpageref{NW2c68Jw-nRuDO-1}}}\endmoddef\nwstartdeflinemarkup\nwusesondefline{\\{NW2c68Jw-2Czttk-1}}\nwenddeflinemarkup void netdraw_mlab(mesh_t mesh, double* x) \{ netdraw_fun_t methods; netdraw_gc_t gc; double* xall; int i, n; vars_mgr_t vars_mgr = mesh_vars_mgr(mesh); assembler_t assembler = mesh_assembler(mesh); int nvars = vars_count(vars_mgr); int nactive = assemble_get_active(assembler); double* displacements = assemble_get_displacements(assembler); memset(&methods, 0, sizeof(methods)); methods.beam = draw_beam; methods.line = NULL; methods.plane = NULL; methods.get_x = get_x; gc.methods = &methods; xall = mxCalloc(nvars, sizeof(double)); if (x) \{ memcpy(xall, displacements, nvars * sizeof(double)); memcpy(xall, x, nactive * sizeof(double)); \} gc.data = xall; n = mesh_num_elements(mesh); for (i = 1; i <= n; ++i) element_display( mesh_element(mesh, i), &gc ); mxFree(xall); \} \nwused{\\{NW2c68Jw-2Czttk-1}}\nwendcode{} \nwixlogsorted{c}{{build interpolants in global coordinates}{NW2c68Jw-4f5wS-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-4f5wS-1}}}% \nwixlogsorted{c}{{build interpolants in local coordinates}{NW2c68Jw-3kyIT3-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-3kyIT3-1}}}% \nwixlogsorted{c}{{cho-draw-mlab-beam.m}{NW2c68Jw-4NFrDv-1}{\nwixd{NW2c68Jw-4NFrDv-1}}}% \nwixlogsorted{c}{{compute rotated corner displacements}{NW2c68Jw-3H1gql-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-3H1gql-1}}}% \nwixlogsorted{c}{{functions}{NW2c68Jw-nRuDO-1}{\nwixu{NW2c68Jw-2Czttk-1}\nwixd{NW2c68Jw-nRuDO-1}}}% \nwixlogsorted{c}{{get local coordinates}{NW2c68Jw-3Dkknx-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-3Dkknx-1}}}% \nwixlogsorted{c}{{hermite cubic function}{NW2c68Jw-1buIps-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-1buIps-1}}}% \nwixlogsorted{c}{{netdraw-mlab.c}{NW2c68Jw-2Czttk-1}{\nwixd{NW2c68Jw-2Czttk-1}}}% \nwixlogsorted{c}{{netdraw-mlab.h}{NW2c68Jw-4H9n7s-1}{\nwixd{NW2c68Jw-4H9n7s-1}}}% \nwixlogsorted{c}{{plot the surfaces}{NW2c68Jw-26zJHS-1}{\nwixu{NW2c68Jw-4NFrDv-1}\nwixd{NW2c68Jw-26zJHS-1}}}% \nwixlogsorted{c}{{static functions}{NW2c68Jw-1duChy-1}{\nwixu{NW2c68Jw-2Czttk-1}\nwixd{NW2c68Jw-1duChy-1}\nwixd{NW2c68Jw-1duChy-2}}}% \nwbegindocs{24}\nwdocspar \nwenddocs{} \hypertarget{structcoro__stack}{}\doxysection{Référence de la structure coro\+\_\+stack} \label{structcoro__stack}\index{coro\_stack@{coro\_stack}} \doxysubsection*{Attributs publics} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{structcoro__stack_ac529104462d659a848a4101b0cf750b4}\label{structcoro__stack_ac529104462d659a848a4101b0cf750b4}} void $\ast$ {\bfseries sptr} \item \mbox{\Hypertarget{structcoro__stack_a83072c69c125b4046780f040eeb1835e}\label{structcoro__stack_a83072c69c125b4046780f040eeb1835e}} size\+\_\+t {\bfseries ssze} \end{DoxyCompactItemize} La documentation de cette structure a été générée à partir du fichier suivant \+:\begin{DoxyCompactItemize} \item ext/nanogui/ext/coro/coro.\+h\end{DoxyCompactItemize} \hypertarget{irene3000_8cpp}{}\section{/home/ashiroji/\+Arduino/libraries/\+Cool\+Board/src/irene3000.cpp File Reference} \label{irene3000_8cpp}\index{/home/ashiroji/\+Arduino/libraries/\+Cool\+Board/src/irene3000.\+cpp@{/home/ashiroji/\+Arduino/libraries/\+Cool\+Board/src/irene3000.\+cpp}} {\ttfamily \#include \char`\"{}F\+S.\+h\char`\"{}}\newline {\ttfamily \#include $<$Arduino.\+h$>$}\newline {\ttfamily \#include \char`\"{}Arduino\+Json.\+h\char`\"{}}\newline {\ttfamily \#include $<$math.\+h$>$}\newline {\ttfamily \#include $<$Irene3000.\+h$>$}\newline Include dependency graph for irene3000.\+cpp\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d0/da5/irene3000_8cpp__incl} \end{center} \end{figure} \subsection*{Macros} \begin{DoxyCompactItemize} \item \#define \hyperlink{irene3000_8cpp_ad72dbcf6d0153db1b8d8a58001feed83}{D\+E\+B\+UG}~0 \end{DoxyCompactItemize} \subsection{Macro Definition Documentation} \mbox{\Hypertarget{irene3000_8cpp_ad72dbcf6d0153db1b8d8a58001feed83}\label{irene3000_8cpp_ad72dbcf6d0153db1b8d8a58001feed83}} \index{irene3000.\+cpp@{irene3000.\+cpp}!D\+E\+B\+UG@{D\+E\+B\+UG}} \index{D\+E\+B\+UG@{D\+E\+B\+UG}!irene3000.\+cpp@{irene3000.\+cpp}} \subsubsection{\texorpdfstring{D\+E\+B\+UG}{DEBUG}} {\footnotesize\ttfamily \#define D\+E\+B\+UG~0} Definition at line 44 of file irene3000.\+cpp. \hypertarget{structrl_1_1_continuous_state_value_functor}{}\section{rl\+:\+:Continuous\+State\+Value\+Functor$<$ State\+Type $>$ Struct Template Reference} \label{structrl_1_1_continuous_state_value_functor}\index{rl\+::\+Continuous\+State\+Value\+Functor$<$ State\+Type $>$@{rl\+::\+Continuous\+State\+Value\+Functor$<$ State\+Type $>$}} Inheritance diagram for rl\+:\+:Continuous\+State\+Value\+Functor$<$ State\+Type $>$\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=3.000000cm]{structrl_1_1_continuous_state_value_functor} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hypertarget{structrl_1_1_continuous_state_value_functor_a5d40cf6eb4afb8e1e8a2ea7d52e95aef}{}\label{structrl_1_1_continuous_state_value_functor_a5d40cf6eb4afb8e1e8a2ea7d52e95aef} virtual double {\bfseries operator()} (const State\+Type \&state) const =0 \item \hypertarget{structrl_1_1_continuous_state_value_functor_ab2e62455723c5642d352a74ae124e456}{}\label{structrl_1_1_continuous_state_value_functor_ab2e62455723c5642d352a74ae124e456} virtual void {\bfseries Update} (const State\+Type \&state, double target, double step\+\_\+size)=0 \end{DoxyCompactItemize} \subsection{Detailed Description} \subsubsection*{template$<$typename State\+Type$>$\newline struct rl\+::\+Continuous\+State\+Value\+Functor$<$ State\+Type $>$} Definition at line 56 of file continuous\+\_\+state\+\_\+value\+\_\+functor.\+hpp. The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item /\+Users/davidfridovichkeil/\+Documents/\+Developer/rl/include/value/continuous\+\_\+state\+\_\+value\+\_\+functor.\+hpp\end{DoxyCompactItemize} @inproceedings{li2018visualizing, title={Visualizing the loss landscape of neural nets}, author={ }, booktitle={Advances in Neural Information Processing Systems}, pages={6389--6399}, year={2018} }Chapter4/chapter4.tex %!TEX root = ../thesis.tex %******************************************************************************* %****************************** Third Chapter ********************************** %******************************************************************************* \chapter{Requirements} \graphicspath{{Chapter4/Figs/Raster/}{Chapter4/Figs/}} Background literature review in Chapter 2.1 and 2.2 has driven the direction of the project and provided many ideas for functional requirements. This chapter describes the collection of primary data for further requirements, and provides the formal list of user requirements. \section{Primary Data and Analysis} Collecting primary data from shareholders of higher education e-learning systems would be able to: \begin{itemize} \setlength\itemsep{0em} \item reaffirm and further develop requirements obtained from literature in Chapter 2, and \item obtain further requirements from real-world participants, their pain points and goals. \end{itemize} \subsection{Methodology} A qualitative study was conducted with semi-structured, face-to-face interviews (open-ended questions). The participants were gathered through directly emailing existing contacts (convenience sampling). Stakeholders with lots of experience were targeted with these criteria: \begin{itemize} \setlength\itemsep{0em} \item Teaching staff in higher education with 10+ years of experience, and \item Students in higher education who has academic liaison experience and are exposed to a wide range of student problems. \end{itemize} It was important that a qualitative, semi-structured method is used. This allowed for a flexible scope, encouraging the generation of new perspectives and ideas beyond that of the secondary research in Chapter 2. It also increased validity, as the interviewer can probe for a deeper understanding \citep{mcleod2014interviews}. An ethics submission was completed on BREO and approval was granted on 21st November. See Supporting Materials/Ethics/Requirements/ for the approval letter, participant information sheet, consent form, and example questions documents. A total of five interviews were conducted between December 2017 and February 2018: two with teaching staff and three with student representatives. See table \ref{table:participants-req} for a more detailed description of these participants. \begin{table}[!h] \caption{Participants in primary data collection interviews} \centering \label{table:participants-req} \begin{tabularx}{\textwidth}{>{\bfseries}lX} Participant & Characterisation \\ \toprule Educator A & lecturer in higher education for over 20 years, and an experienced higher education administrator \\\midrule Educator B & lecturer in higher education for over 20 years, with research interests in e-learning interactions, effectiveness and acceptance \\\midrule Student C & university course representative for 3 years, which involves collecting and communicating student feedback and attending staff-student liaison meetings \\\midrule Student D & university peer-assisted learning leader for 2 years, helping out lower level students with their academic work by facilitating peer discussions, and escalating common problems to academic staff in debrief sessions \\\midrule Student E & university course representative for 2 years, peer-assisted learning leader for 1 year \\\bottomrule \end{tabularx} \end{table} \subsection{Interview Results and Analysis} The raw data from interviews (transcripts) were contextually analysed and grouped thematically using thematic analysis techniques from \citet{clarke2014thematic}. These themes are presented below as problem statements (PS). These problem statements were sorted into four affinity groups: statements about 1. assessments, 2. curriculum personalisation, 3. higher education, and 4. e-Learning systems. Below you will find the leading questions asked for each group, the problem statements, explanations and discussions. For the relevant transcript snippets for each problem statements, please go to Appendix C. \subsubsection{On Assessments} Leading Question: What are the problems with assessments in higher education today? / Is there tension between students and teachers over assessments, and why? \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline & Problem Statement & Participants \\ \hline PS1 & \textbf{Poor communication of assessment expectations} & B, C, D, E \\ \hline \multicolumn{3}{|X|}{Four participants have independently confirmed that the problem with transparency in expectations for assessments (as described in Chapter 2.1.1) exists.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS2 & \textbf{Lack of practice and formative feedback} & B, D \\ \hline \multicolumn{3}{|X|}{Students transitioning from school to higher education would experience a drop in formative assessments (that do not affect their final grade) such as homework practices.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS3 & \textbf{Lack of standardisation in marking criteria} & B, C, E \\ \hline \multicolumn{3}{|X|}{Some assessments have clearly defined marking criteria forms for markers, but a high number of them do not.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS4 & \textbf{Oral defence may be necessary to validate assessments} & A, E \\ \hline \multicolumn{3}{|X|}{Concern was raised over the ease of cheating and incomplete learning, especially in automatic assessments. Vivas are proposed as a potential solution.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS5 & \textbf{Lack of feedback and procedural transparency in terminal assessments} & C \\ \hline \multicolumn{3}{|X|}{Students are unhappy with the lack of, or scarcity of feedback from terminal assessments and exams.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS6 & \textbf{The need for synoptic (cross-topics) assessments} & B \\ \hline \multicolumn{3}{|X|}{There is a concern that assessments are too compartmentalised into their modules, not relevant to industry, and not encouraging holistic critical thinking.} \\ \hline \end{tabularx} \end{table} The interviews have confirmed the problem of transparency on assessments (see PS1, PS3, PS5), while new user concerns have surfaced, namely the lack of practice (PS2), the need for assessment validations (PS4), and synoptic assessments (PS6). Most of these problems can be tackled by a blockchain-based system. The transparent provenance of student and teacher inputs, and outputs guaranteed by Smart Contracts, can greatly improve transparency and trust in assessment processes and outcomes. The lack of practice (PS2) however, is seen more as a human factor, as it depends heavily on how the module is designed by its teacher. Synoptic assessments (PS6) are considered a low priority feature, as it is a relatively new concept in higher education, and was only mentioned by one participant. \subsubsection{On Curriculum Personalisation} Leading Question: What do you think are the current roadblocks to offering more multi-disciplinary, personalised, or even multi-institutional arrangements for higher education? \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS7 & \textbf{There is a demand for multi-disciplinary degree offerings but only a few universities are capable of offering them} & B, C, D \\ \hline \multicolumn{3}{|X|}{A defined degree that encompasses two particular fields may not be economically viable, having free-to-choose credits is a more practical model; \newline universities struggle with bureaucracy for curriculum personalisation; \newline students like more freedom and sometimes dislike certain compulsory modules.} \\ \hline \end{tabularx} \end{table} \clearpage \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS8 & \textbf{Dedicated support and guidance is necessary for increased customisation} & B, E \\ \hline \multicolumn{3}{|X|}{There are careers that require multi-disciplinary backgrounds, but there is a risk of students not making informed choices.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS9 & \textbf{National regulations requiring programme outcomes and specifications deter movement} & A \\ \hline \multicolumn{3}{|X|}{The UK higher education academic infrastructure requires all degree programme to lay out programme outcomes and programme specifications, which deters movement.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS10 & \textbf{Managing study load for multi-disciplinary courses may be difficult} & E \\ \hline \multicolumn{3}{|X|}{Switching between the mindsets for different disciplines could be hard. The number of credits for multi-disciplinary programs should not be higher.} \\ \hline \end{tabularx} \end{table} It was encouraging to see that three out of four participants described a clear demand for more freedom in curriculum personalisation (PS7). Also confirming the need for informed educational decisions (Chapter 2.1.2, \citet{green2005futurelab}), participants have asked for dedicated curriculum personalisation support (PS8). The present inflexible degree structures in markets such as the UK is an interesting point raised by participant A (PS9). This is actually what Smart Contracts can be very good at, automating administrative, regulatory work. For this proof-of-concept project, example administrative fields can be added to showcase such a potential. \subsubsection{Other Higher Education Issues} Leading Question: What else needs improving in higher education in general? \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS11 & \textbf{More support, interaction and engagement needed} & D, E \\ \hline \multicolumn{3}{|X|}{Students have complained that institutions are not always good at supporting and engaging with all students.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS12 & \textbf{Administrative middlemen causing delays and bottlenecks in institutions} & E \\ \hline \multicolumn{3}{|X|}{Students complained that feedback was not given on time due to administrative delays} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS13 & \textbf{Lack of support on self-regulated learning skills} & B \\ \hline \multicolumn{3}{|X|}{There is not enough help and guidance for students to make the transition to self-motivated learning} \\ \hline \end{tabularx} \end{table} Only the administrative bottleneck issue (PS12) is considered highly relevant to this project. A blockchain backend would not immediately be able to improve student engagement (PS11) or self-regulated learning skills (PS13). \subsubsection{e-Learning System Issues} Sample Questions: What other features would you like to see in a future e-Learning system for higher education? \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS14 & \textbf{Fine-grained access control for learning history needed to preserve privacy} & D, E \\ \hline \multicolumn{3}{|X|}{Students prefer to be as private as possible, and dislike data aggregation, even by their institution} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS15 & \textbf{Poor mobile friendliness in many e-Learning systems} & C \\ \hline \multicolumn{3}{|X|}{Students want any new system to be built responsive and mobile friendly.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS16 & \textbf{Real-world learning activities not captured in course management systems} & A \\ \hline \multicolumn{3}{|X|}{For example, face-to-face contact time is not captured by most conventional systems.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS17 & \textbf{Systems cannot be customised to fit assessment and grading models} & A \\ \hline \multicolumn{3}{|X|}{assessment features and functions on e-learning systems are incompatible with institutional requirements, or even national requirements, giving staff extra work converting them manually.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS18 & \textbf{Multiple systems used in concoction without login integration} & B \\ \hline \multicolumn{3}{|X|}{No single sign-on for intranet, assessment, content management and student records systems.} \\ \hline \end{tabularx} \end{table} \begin{table}[!ht] \begin{tabularx}{\textwidth}{|c|X|c|} \hline PS19 & \textbf{Lower trust and value associated with online institutions and programmes} & C \\ \hline \multicolumn{3}{|X|}{It is hard to bridge the gap in trust and reputation that comes with campus learning.} \\ \hline \end{tabularx} \end{table} Ease of use is a major theme in this group, with requests for mobile friendliness (PS15), flexibility in record types (PS16, 17), and single sign-on (PS18). It will be important to take these into account while designing our blockchain and client applications. A blockchain has the potential to improve access control (PS14) and create trust (PS19) through provenance. These two user concerns discovered has been added to the aim of the project. \subsubsection{Summary} Overall, the study has been useful in extending our knowledge on higher education and e-Learning issues. The convenience sampling has likely been a limitation, since all participants were recruited from the same university, and the social circle of the interviewer. An affinity diagram is also produced to give a high-level overview of all the problem statements. (See Figure \ref{fig:ps-affinity} on the next page). \clearpage \begin{figure}[!ht] \centering \includegraphics[width=1.0\textwidth]{ps-affinity} \caption[Affinity diagram for primary data] {Affinity diagram for problem statements devised from primary data} \label{fig:ps-affinity} \end{figure} \section{Formal Requirements} Table \ref{table:fx-reqs} lists the functional requirements (FR) adopted by this project going forward. They are related to one or more of the problem statements (PS) gathered above from primary data, or to the literature reviewed in Chapter 2. Table \ref{table:nonfx-reqs} lists the non-functional requirements (NR) adopted by this project going forward. They are related to one or more of the problem statements (PS) gathered, or to usability heuristics. These requirements have been ranked with the MoSCoW prioritisation framework, which specified four levels of priority: Must Have, Should Have, Could Have, and Won’t Have this time \citep{agile2018moscow}. The MoSCoW levels are given from mainly a system engineering perspective in planning the minimum viable product for the demonstrator of this project, and do not necessarily reflect the priorities of the stakeholders. \begin{table}[!h] \caption{Prioritised list of functional requirements for this project} \centering \label{table:fx-reqs} \begin{tabularx}{\textwidth}{>{\bfseries}l>{\hsize=1.5\hsize}X>{\hsize=.5\hsize}Xl} & Requirement Statements & Related To & MoSCoW \\ \toprule FR1 & The system would store learner records on a blockchain & Ch 2.2.1 (LTSA) & Must Have \\\midrule FR2 & Teachers would be able to create and edit learning resources & Ch 2.2.1 (LTSA) & Must Have \\\midrule FR3 & Teachers would be able to create and edit assessments & Ch 2.2.1 (LTSA) & Must Have \\\midrule FR4 & The system would enforce the provision of learning outcomes, knowledge required and assessment goals & Ch 2.1.1 (Transparency), PS1 & Must Have \\\midrule FR5 & The system would enforce predefined assessments rules (eg. marking schemes, transparent procedures and feedback) with Smart Contracts & Ch 2.1.1 (Transparency), PS3, PS5 & Must Have \\\midrule FR6 & The system would allow teachers to configure multiple assessments and formative assessments for modules & PS2 & Could Have \\\midrule FR7 & The system would be able to facilitate vivas (oral defence) as a form of assessments & PS4 & Could Have \\\midrule FR8 & The system would provide multiple ways to define grade schema & PS17 & Could Have \\\midrule FR9 & Teachers would be able to negotiate a customised list of courses for a student within a fixed course credits budget, customising degree specifications & Ch 2.1.2 (Personalisation), PS7, PS9, PS10 & Must Have \\\midrule FR10 & The system should feature dedicated support channels between students and teachers or other advisors & PS8, PS11 & Should Have \\\midrule FR11 & Students would be able to add assessment submissions on the blockchain & Figure \ref{fig:moocon_assess} (Concept) & Must Have \\\midrule FR12 & The system would be able to generate certificates on the blockchain when a course has been completed & Figure \ref{fig:moocon_assess} (Concept) & Must Have \\\midrule FR13 & The system would allow students to control access to their education history on the blockchain & PS14 & Should Have \\\midrule FR14 & The system would provide one login for content delivery, assessment and record keeping & PS18 & Should Have \\\midrule & \multicolumn{2}{c}{Requirements targetting PS6, PS13, PS16} & Won't Have \\\bottomrule \end{tabularx} \end{table} \begin{table}[!h] \caption{Prioritised list of non-functional requirements for this project} \centering \label{table:nonfx-reqs} \begin{tabularx}{\textwidth}{>{\bfseries}l>{\hsize=1.6\hsize}X>{\hsize=.4\hsize}Xl} & Requirement Statements & Related To & MoSCoW \\\toprule NR1 & The client applications would have the same functionalities across devices and a responsive interface & PS15 & Should Have \\\midrule NR2 & The client applications would fail safely and display error messages to the user & & Should Have \\\midrule NR3 & The client applications would notify users of relevant events on the blockchain network & & Should Have \\\midrule NR4 & The system should be able to reduce administrative work & PS12 & Should Have \\\midrule NR5 & The system should be able to create more trust in online education providers and programmes & PS19 & Should Have \\\midrule NR6 & The system should be highly usable, visually consistent and appealing & & Should Have \\\midrule NR7 & The client applications should always display its navigation menu and status of the application & & Should Have \\\bottomrule \end{tabularx} \end{table}@misc{rfc1576, series = {Request for Comments}, number = 1576, howpublished = {RFC 1576}, publisher = {RFC Editor}, doi = {10.17487/RFC1576}, url = {https://rfc-editor.org/rfc/rfc1576.txt}, author = {}, title = {{TN3270 Current Practices}}, pagetotal = 12, year = 1994, month = jan, abstract = {This document describes the existing implementation of transferring 3270 display terminal data using currently available telnet capabilities. The name traditionally associated with this implementation is TN3270. This memo provides information for the Internet community. This memo provides information for the Internet community. This memo does not specify an Internet standard of any kind.}, } UQComputingSociety/subject-guide1-10 \courseTemplate[ code = {DECO2500}, title = {Human-Computer Interaction}, score = {4}, prereq = {DECO1400 or CSSE1001}, contact = {4C}, coordinator = {Prof ()}, assessment = { Practical & 40\% & 10 weekly group work pracs worth 4\% each. \\ Report & 25\% & A 2000 word report from a choice of 20 topics \\ Final Exam & 35\% & A final exam on every chapter from the lectures and the textbook \\ }, review = { Human-Computer Interaction is a theory-oriented course. Many students come into the course expecting to do coding or design work and become sorely disappointed. Assessment involves weekly practicals, a research paper, and a final exam. 40\% of the grade are weekly pracs worth 4\% each. These weekly pracs can be easy marks if you prepare beforehand. As with any group work, be sure to find good people to work with in the first week. The second big assessment is the research document from a choice of 20 topics. Some involve real-world testing, and some are just literature reviews. Feel free to pick whatever you'd like, but some topics are easier than others. The final exam will quiz you on the textbook. Make sure to read it cover to cover, as you will be quizzed on specific case studies in the book. Past exam questions will be your friend when studying for the final exam. Penelope Sanderson is a lecturer from the Psychology department. If you approach the course as more of a psychology course than an engineering course, then you will have better expectation and understanding of the activities. Not everyone will enjoy this course, but the course gives a solid foundation to understanding how usabilitiy testing makes for better software. }, preparation = { \item Prepare for the weeky practical classes beforehand. The criteria sheets for each week are released before the class begins, so no excuses. \item You will be required to find high-quality sources for your research paper. Using bibliography software such as Zotero makes this much easier. There are heavy penalties for going over the word count. \item Spread out your reading of the textbook over the semester, rather than at the last minute. For best results, find classmates who will quiz you on your knowledge. It is easier to understand the material than to memorise it. }]{}WormLabCaltech/mprsq \documentclass[9pt,twocolumn,twoside]{pnas-new} \usepackage{gensymb} \usepackage{siunitx} \usepackage[switch]{lineno} % new commands % q value \newcommand{\qval}[1]{$q<10^{-#1}$} % species names \newcommand{\cel}{\emph{C.~elegans}} \newcommand{\dicty}{\emph{D.~discoideum}} \newcommand{\ecol}{\emph{E.~coli}} % gene names \newcommand{\gene}[1]{\mbox{\emph{#1}}} % \newcommand{\gene}[1]{\emph{#1}} # for MS word typesetting \newcommand{\nlp}{\gene{nlp-31}} \newcommand{\ftna}{\gene{ftn-1}} \newcommand{\ftnb}{\gene{ftn-2}} \newcommand{\cysl}{\gene{cysl-1}} \newcommand{\nog}{\gene{nog-1}} \newcommand{\nhr}{\gene{nhr-57}} \newcommand{\lam}{\gene{lam-3}} \newcommand{\fog}{\gene{fog-2(lf)}} \newcommand{\egl}{\gene{egl-9(lf)}} \newcommand{\rhy}{\gene{rhy-1(lf)}} \newcommand{\vhl}{\gene{vhl-1(lf)}} \newcommand{\eglvhl}{\gene{egl-9(lf); vhl-1(lf)}} \newcommand{\eglhif}{\gene{egl-9(lf) hif-1(lf)}} \newcommand{\hif}{\gene{hif-1(lf)}} % protein names \newcommand{\eglp}{EGL-9} \newcommand{\rhyp}{RHY-1} \newcommand{\nogp}{NOG-1} \newcommand{\vhlp}{VHL-1} \newcommand{\hifp}{HIF-1} \newcommand{\fogp}{FOG-2} \newcommand{\nhrp}{NHR-57} \newcommand{\lamp}{LAM-3} \newcommand{\cyslp}{CYSL-1} % DE genes numbers: \newcommand{\egln}{2,549} \newcommand{\rhyn}{3,005} \newcommand{\vhln}{1,275} \newcommand{\eglvhln}{3,654} \newcommand{\hifn}{1,075} \newcommand{\eglhifn}{744} \newcommand{\fogn}{2,840} \newcommand{\total}{7,609} % downstream targets \newcommand{\egltargets}{4} \newcommand{\rhytargets}{0} \newcommand{\vhltargets}{71} % 72 minus vhl-1 (IDed due to deletion) \newcommand{\hiftargets}{264} \newcommand{\hifohtargets}{56} % website commands \newcommand{\website}{ \url{https://wormlabcaltech.github.io/mprsq/} } \newcommand{\webref}{ \href{https://wormlabcaltech.github.io/mprsq/}{website}} \templatetype{pnasresearcharticle} % Choose template \title{Reconstructing a metazoan genetic pathway with transcriptome-wide epistasis measurements} \author[a, b, 1]{} \author[a, b, c, 1]{} \author[a]{} \author[a]{} \author[a, b]{} \affil[a]{Division of Biology and Biological Engineering, Caltech, Pasadena, CA, 91125, USA} \affil[b]{Howard Hughes Medical Institute, Caltech, Pasadena, CA, 91125, USA} \affil[c]{Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles, California, 90033, USA} % Please give the surname of the lead author for the running footer \leadauthor{Angeles-Albores} % Please add here a significance statement to explain the relevance of your work \significancestatement{ Transcriptome profiling quantitatively measures gene expression genome-wide. There is widespread interest in using transcriptomic profiles as phenotypes for epistasis analysis. Though epistasis measurements can be performed using individual transcripts, this results in many scores that must be interpreted independently. We developed a statistic that summarizes these measurements, simplifying analysis. Moreover, epistasis analysis has previously only been performed on cDNA extracted from single cells. We show that whole-organism RNA-seq can be used to characterize interactions between genes. With the advent of genome engineering, mutants can be created easily in many organisms. Thus, phenotyping is now the rate-limiting step towards reconstructing interaction networks. Our work potentially represents a solution to this problem because RNA-seq is sensitive to a variety of genetic perturbations. } % Please include corresponding author, author contribution and author declaration information \authorcontributions{DAA, CPR and PWS designed experiments. CPR and BAW performed all experiments. BJW provided resources. DAA developed computational methods and performed analysis. DAA, CPR and PWS wrote the paper.} \authordeclaration{The authors declare no conflict of interest.} \equalauthors{\textsuperscript{1} DAA and CPR contributed equally to this work.} \correspondingauthor{\textsuperscript{2}To whom correspondence should be addressed. E-mail: } % Keywords are not mandatory, but authors are strongly encouraged to provide % them. If provided, please include two to five keywords, separated by the pipe % symbol, e.g: \keywords{Epistasis $|$ Genetic Interaction $|$ Transcriptome $|$ Hypoxia} \begin{abstract} RNA-seq is commonly used to identify genetic modules that respond to perturbations. In single cells, transcriptomes have been used as phenotypes, but this concept has not been applied to whole-organism RNA-seq. Also, quantifying and interpreting epistatic effects using expression profiles remains a challenge. We developed a single coefficient to quantify transcriptome-wide epistasis that reflects the underlying interactions and which can be interpreted intuitively. To demonstrate our approach, we sequenced four single and two double mutants of \emph{Caenorhabditis~elegans}. From these mutants, we reconstructed the known hypoxia pathway. In addition, we uncovered a class of \hifohtargets{} genes with \gene{hif-1}-dependent expression that have opposite changes in expression in mutants of two genes which cooperate to negatively regulate \hifp{} abundance; however, the double mutant of these genes exhibits suppression epistasis. This class violates the classical model of HIF-1 regulation, but can be explained by postulating a role of hydroxylated HIF-1 in transcriptional control. \end{abstract} \dates{This manuscript was compiled on \today} \doi{\url{www.pnas.org/cgi/doi/10.1073/pnas.XXXXXXXXXX}} \begin{document} % Optional adjustment to line up main text (after abstract) of first page with % line numbers, when using both lineno and twocolumn options. You should only % change this length when you've finalised the article contents. \verticaladjustment{-2pt} \maketitle \thispagestyle{firststyle} \ifthenelse{\boolean{shortarticle}}{\ifthenelse{\boolean{singlecolumn}}{\abscontentformatted}{\abscontent}}{} \linenumbers{} \dropcap{G}enetic analysis of molecular pathways has traditionally been performed through epistatic analysis. If the mutants of two distinct genes have a quantifiable phenotype, and the double mutant has a phenotype that is not the sum of the phenotypes of the single mutants, this non-additivity is referred to as generalized epistasis, and indicates that these genes interact functionally. Such interactions can occur at the biochemical level between their products or as a consequence of their functions~\cite{Huang2006}. Epistasis analysis remains a cornerstone of genetics today~\cite{Phillips2008}. Recently, biological studies have shifted in focus from studying single genes to studying all genes in parallel. In particular, RNA-seq~\cite{Mortazavi2008} enables biologists to identify genes that change expression in response to a perturbation. % Gene expression % profiling using RNA-seq has become much more sensitive thanks to deeper and more % frequent sequencing due to lower sequencing costs~\cite{Metzker2010}, % better and faster abundance quantification~\cite{Patro2014,Bray2016,Patro2015}, % and improved differential expression analysis % methods~\cite{Pimentel2016,Trapnell2013}. RNA-seq has been used to identify genetic modules involved in a variety of processes, such as in the \emph{Caenorhabditis~elegans} linker cell migration~\cite{Schwarz2012}, planarian stem cell maintenance~\cite{VanWolfswinkel2014,Scimone2014}. The role of transcriptional profiling has been restricted to target gene identification, and so far there are only a few examples where transcriptomes have been used to generate quantitative genetic models of any kind. % \color{purple} In quantitative genetics, eQTL studies have established the power of transcriptomes for genetic mapping~\cite{Brem2002,Schadt2003,Li2006,King2014}. Genetic pathway analysis via epistasis has been performed in \emph{Saccharomyces~cerevisiae}~\cite{Hughes2000,Capaldi2008} and in \emph{Dictyostelium~discoideum}~\cite{VanDriessche2005}. Recently, Dixit \emph{et al} described a protocol for epistasis analysis in dendritic and K562 cells using single-cell RNA-seq~\cite{Dixit2016}. \color{black} Epistasis analysis of single cells or single-celled organisms is popular because of the concern that whole-organism sequencing will mix information from multiple cell types, preventing the accurate reconstruction of genetic interactions. Using whole-organism transcriptome profiling, we have recently identified a new developmental state of \cel{} caused by loss of a single cell type (sperm cells)~\cite{Angeles-Albores2017a}, which suggests that whole-organism transcriptome profiling contains sufficient information for epistatic analysis. To investigate the ability of whole-organism transcriptomes to serve as quantitative phenotypes for epistatic analysis in metazoans, we sequenced the transcriptomes of four well-characterized loss-of-function mutants in the \cel{} hypoxia pathway~\cite{Epstein2001,Shen2006,Shao2009,Jiang2001}. % carmie: Metazoans depend on the presence of oxygen in sufficient concentrations to support aerobic metabolism. Hypoxia inducible factors (HIFs) are an important group of oxygen-responsive genes that are highly conserved in metazoans~\cite{Loenarz2011}. A common mechanism for hypoxia-response induction is heterodimerization between a HIF$\alpha$ and a HIF$\beta$ subunit; the heterodimer then initiates transcription of target genes~\cite{Jiang1996}. The number and complexity of HIFs varies throughout metazoans. In the roundworm \cel{} there is a single HIF$\alpha$ gene, \gene{hif-1}~\cite{Jiang2001}, and a single HIF$\beta$ gene, \gene{ahr-1}~\cite{Powell-Coffman1998}. Levels of HIF$\alpha$ proteins are tightly regulated. Under conditions of normoxia, \hifp{}$\alpha$ exists in the cytoplasm and partakes in a futile cycle of protein production and rapid degradation~\cite{Huang1996}. In \cel{}, \hifp{}$\alpha$ is hydroxylated by a proline hydroxylase (\eglp{})~\cite{Kaelin2008}. \hifp{} hydroxylation increases its binding affinity to Von Hippel-Lindau tumor suppressor 1 (\vhlp{}), which in turn allows ubiquitination of \hifp{} leading to its degradation. In \cel{}, \eglp{} activity is inhibited by binding of \cyslp{}, a homolog of sulfhydrylases/cysteine synthases; and \cyslp{} activity is in turn inhibited by the putative transmembrane O-acyltransferase \rhyp{}, possibly by post-translational modifications to \cyslp{}~\cite{Ma2012} (see Fig.~\ref{fig:pathway}). \begin{figure}[tbhp] \centering \includegraphics[width=\linewidth]{../final_figs/HIF1pathway.pdf} \caption{ Genetic and biochemical representation of the hypoxia pathway in \cel{}. Red arrows are arrows that lead to inhibition of \hifp{}, and blue arrows are arrows that increase \hifp{} activity or are the result of \hifp{} activity. \eglp{} is known to exert \vhlp{}-dependent and independent repression on \hifp{} as shown in the genetic diagram. The \vhlp{}-independent repression of \hifp{} by \eglp{} is denoted by a dashed line and is not dependent on the hydroxylating activity of \eglp{}. RHY-1 inhibits CYSL-1, which in turn inhibits EGL-9, but this interaction was abbreviated in the genetic diagram for clarity. } \label{fig:pathway} \end{figure} Our reconstruction of the hypoxia pathway in \cel{} shows that whole-animal transcriptome profiles can be used as phenotypes for genetic analysis and that epistasis, a hallmark of genetic interaction observed in double mutants, holds at the molecular systems level. We demonstrate that transcriptomes can aid in ordering genes in a pathway using only single mutants. We were able to identify genes that appear to be downstream of \gene{vhl-1}, but not downstream of \gene{hif-1}. Using a single set of transcriptome-wide measurements, we observed most of the known transcriptional effects of \gene{hif-1} as well as novel effects not described before in \cel{}. Taken together, this analysis demonstrates that whole-animal RNA-seq is a fast and powerful method for genetic analyses in an area where phenotypic measurements are now the rate-limiting step. \section*{Results} \subsection*{The hypoxia pathway controls thousands of genes in \cel{}} \label{sub:summary} We selected four null single mutants within the hypoxia pathway for expression profiling: \gene{egl-9(sa307)}, \gene{rhy-1(ok1402)}, \gene{vhl-1(ok161)}, \gene{hif-1(ia4)}. We also sequenced the transcriptomes of two double mutants, \gene{egl-9; vhl-1} and \gene{egl-9 hif-1} as well as wild type (N2). % \color{purple} Each genotype was sequenced in triplicate using mRNA extracted from 30 worms at a depth of 15~million reads per sample. Of these 15~million reads, 50\% of the reads mapped to the \cel{} genome on average. All samples were analyzed under normoxic conditions. \color{black} We measured differential expression of 19,676 isoforms across all replicates and genotypes ($\sim$70\% of the protein coding isoforms in \cel{}; see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/1_basic_stats.html} {Basic Statistics Notebook}). We included in our analysis a \gene{fog-2(q71)} mutant we have previously studied~\cite{Angeles-Albores2017a}, because \gene{fog-2} is not reported to interact with the hypoxia pathway. We analyzed our data using a general linear model on logarithm-transformed counts. Changes in gene expression are reflected in the regression coefficient $\beta$, which is specific to each isoform within a genotype (excluding wild type, which is used as baseline). Statistical significance is achieved when the q-value of a $\beta$ coefficient ($p$-values adjusted for multiple testing) are less than 0.1. Transcripts that are differentially expressed between the wild type and a given mutant have $\beta$ values that are statistically significantly different from 0 (i.e.\ greater than 0 or less than 0). $\beta$ coefficients are analogous to the logarithm of the fold-change between the mutant and the wild type. Larger magnitudes of $\beta$ correspond to larger perturbations (see Fig.~\ref{fig:explain}). When we refer to $\beta$ coefficients and $q$-values, it will always be in reference to isoforms. However, we report the sizes of each gene set by the number of differentially expressed genes (DEGs), not isoforms, they contain. For the case of \cel{}, this difference is negligible since the great majority of protein-coding genes have a single isoform. We have opted for this method of referring to gene sets because it simplifies the language considerably. A complete version of the code used for this analysis with ample documentation, is available at \url{https://wormlabcaltech.github.io/mprsq}. \begin{figure}[tbhp] \centering \includegraphics[width=0.35\textwidth]{../final_figs/meaningofbeta.pdf} \caption{ Analysis workflow. After sequencing, reads are quantified using Kallisto. Bars show estimated counts for each isoform. Differential expression is calculated using Sleuth, which outputs one $\beta$ coefficient per isoform per genotype. $\beta$ coefficients are analogous to the natural logarithm of the fold-change relative to a wild type control. Downstream analyses are performed with $\beta$ coefficients that are statistically significantly different from 0. $q$-values less than 0.1 are considered statistically different from 0. } \label{fig:explain} \end{figure} Transcriptome profiling of the hypoxia pathway revealed that this pathway controls thousands of genes in \cel{} (see Table~\ref{tab:genes}, see SI File 1 for a complete list of differentially expressed genes). The \egl{} transcriptome showed differential expression of \egln{} genes. \rhyn{} genes were differentially expressed in \rhy{} mutants. The \vhl{} transcriptome showed considerably fewer DEGs (\vhln{}), possibly because \gene{vhl-1} is a weaker inhibitor of \gene{hif-1} than \gene{egl-9}~\cite{Shao2009}. The \egl{};\vhl{} double mutant transcriptome showed \eglvhln{} DEGs. The \hif{} mutant showed a transcriptomic phenotype involving \hifn{} genes. The \eglhif{} double mutant showed a similar number of genes with altered expression (\eglhifn{} genes). % \color{purple} We do not think that this transcriptional response is the due to transiently induced hypoxia during harvesting. If the wild type strain had become hypoxic, then the \hif{} genotype should show significantly lower levels of \gene{nhr-57}, a marker that increases during hypoxia. We do not observe altered levels of \gene{nhr-57} when comparing the wild type and \hif{} mutant, nor between the wild type and \eglhif{} double mutant. Finally, the \egl{}, \vhl{}, \rhy{} and \eglvhl{} mutants did show altered \gene{nhr-57} transcript levels (see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/5_quality_check.html}{ Quality Control Notebook}, SI Figure 1). Of the differentially expressed genes in \hif{} mutants, 161/\hifn{} were also differentially expressed in \eglhif{} mutants, which suggests these transcripts are \gene{hif-1}-dependent under normoxia. For the remaining genes, we cannot rule out cumulative effects from loss of \gene{hif-1}, strain-specific eQTLs present in the strain background or that loss of \gene{egl-9} suppresses the mutant phenotype. We designed our experiments to probe the constitutive hypoxia response, and not the effects of \gene{hif-1} under normoxia, which we did not foresee. As a result, we have limited resolving power to explain the transcriptome of \hif{} mutants. \color{black} \begin{table}[tbhp] \centering \begin{tabular}{lr} \toprule{} Genotype & Differentially Expressed Genes\\ \midrule{}\egl{} & \egln{}\\ \rhy{} & \rhyn{}\\ \vhl{} & \vhln{}\\ \hif{} & \hifn{}\\ \eglvhl{} & \eglvhln{}\\ \eglhif{} & \eglhifn{}\\ \fog{} & \fogn{}\\ \bottomrule{} \end{tabular} \caption{Number of differentially expressed genes in each mutant strain with respect to the wild type (N2).} \label{tab:genes} \end{table} \subsection*{Principal Component Analysis visualizes epistatic relationships between genotypes} \label{sub:Clustering} Principal component analysis (PCA) is used to identify relationships between high-dimensional data points~\cite{Yeung2001}. We used PCA examine whether each genotype clustered in a biologically relevant manner. PCA identifies the vector that explains most of the variation in the data; this is called the first principal component. PCA can identify the first $n$ components that explain more than 95\% of the data variance. Clustering in these $n$ dimensions can indicate biological relationships, although interpreting principal components can be difficult. % After applying PCA, we expected \hif{} to cluster near \eglhif{}, because % \hif{} exhibits no phenotypic defects under normoxic conditions, in contrast to % \egl{}, which exhibits an egg-laying (Egl) phenotype in the same environment. % In \eglhif{} mutants the Egl phenotype of \egl{} mutants is suppressed and instead % the grossly wild-type phenotype of \hif{} is observed. On the other hand, we % expected \egl{}, \rhy{}, \vhl{} and \eglvhl{} to form a separate cluster since % each of these genotypes is Egl and has a constitutive hypoxic response. Finally, % we included as a negative control a \fog{} mutant we have analyzed % previously~\cite{Angeles-Albores2016a}. This data was obtained at a different % time from the other genotypes, so we included a batch-normalization term in our % equations to account for this. Since \gene{fog-2} has not been described % to interact with the hypoxia pathway, we expected that it should appear far away % from either cluster. In our analysis, the first principal component discriminated mutants that have constitutive high levels of \hifp{} from mutants that have no \hifp{}, whereas the second component was able to discriminate between mutants within the hypoxia pathway and outside the hypoxia pathway (see Fig.~\ref{fig:pca}; \gene{fog-2} is not reported to act in the hypoxia pathway and acts as a negative control; see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/2_predict_interactions.html} {Genetic Interactions Notebook}). % Therefore, expression profiling measures enough signal to % cluster genes in a meaningful manner in complex metazoans. \begin{figure}[tbhp] \centering \includegraphics[width=0.4\textwidth]{../final_figs/pca.pdf} \caption{ Principal component analysis of various \cel{} mutants. Genotypes that have an constitutive hypoxia response (i.e. \egl{}) cluster far from genotypes that do not have a hypoxic response (i.e. \hif{}) along the first principal component. The second principal component separates genotypes that do not participate hypoxic response pathway. } \label{fig:pca} \end{figure} \subsection*{Reconstruction of the hypoxia pathway from first genetic principles} \label{sec:reconstruct} To reconstruct a genetic pathway, we must assess whether two genes act on the same phenotype. If they do not act on the same phenotype (two mutations do not cause the same genes to become differentially expressed relative to wild type), these mutants are independent. Otherwise, we must measure whether these genes act additively or epistatically on the phenotype of interest; if there is epistasis we must measure whether it is positive or negative, in order to assess whether the epistatic relationship is a genetic suppression or a synthetic interaction. To allow coherent comparisons of different mutant transcriptomes (the phenotype we are studying here), we define the shared transcriptomic phenotype (STP) between two mutants as the shared set of genes or isoforms whose expression in both mutants are different from wild-type, regardless of the direction of change. \subsubsection*{Genes in the hypoxia mutant act on the same transcriptional phenotype} \label{sec:phenotypes} All the hypoxia mutants had a significant STP:\@ the fraction of differentially expressed genes that was shared between mutants ranged from a minimum of 10\% between \hif{} and \eglvhl{} to a maximum of 32\% between \egl{} and \eglvhl{} (see SI Table 1). For comparison, we also analyzed a previously published \fog{} transcriptome~\cite{Angeles-Albores2017a}. The \gene{fog-2} gene is involved in masculinization of the \cel{} germline, which enables sperm formation, and is not known to be involved in the hypoxia pathway. The hypoxia pathway mutants and the \fog{} mutant also had STPs (8.8\%--14\%). % genetic correlations \begin{figure}[tbhp] \centering \includegraphics[width=0.5\textwidth]{../final_figs/triangle_plot.pdf} \caption{ Interacting genes have correlated transcriptional signatures. The rank order of transcripts contained in the shared transcriptional phenotype is plotted for each pairwise combination of genotypes.Correlations between in-pathway genotypes are strong whereas comparisons with a \fog{} genotype are dominated by noise. Comparisons between some genotypes show populations of transcripts that are anticorrelated, possibly as a result of feedback loops. Plots are color-coded by row. Comparisons with genotypes with a constitutive hypoxia response are in blue; comparisons with genotypes negative for \hif{} are black; and comparisons involving \fog{} are red. X- and y-axes show the rank of each transcript within each genotype. } \label{fig:genetic_interactions} \end{figure} Next, we analyzed pairwise correlations between all mutant pairs. We rank-transformed the $\beta$ coefficients of each isoform between the STP of two mutants, and plotted the transcript ranks between genotypes (see Fig~\ref{fig:genetic_interactions}). % For mutants % associated with the hypoxia pathway, these correlations have values higher than % 0.9 with a tight distribution around the line of best fit. % The correlations for % mutants from the hypoxia pathway with the \fog{} mutant were considerably % weaker, with magnitudes between 0.6--0.85 and greater variance around the line % of best fit. Although \gene{hif-1} is known to be genetically repressed by \gene{egl-9}, \gene{rhy-1} and \gene{vhl-1}~\cite{Epstein2001,Shen2006}, all the correlations between mutants of these genes and \hif{} were positive (see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/2_predict_interactions.html} {Genetic Interactions Notebook}). We reasoned that this apparent contradiction could be due to either strain-specific effects in our N2 background (an artifactual signal) or that it could reflect a previously unrecognized aspect of \hifp{} biology. This motivated us to look for genes that exhibited verifiable extreme patterns of anomalous behavior and led us to propose a new model of the hypoxia pathway (see Identification of non-classical epistatic interactions). \subsection*{Transcriptome-wide epistasis} Ideally, any measurement of transcriptome-wide epistasis should conform to certain expectations. First, it should make use of the regression coefficients of as many genes as possible. Second, it should be summarizable in a single, well-defined number. Third, it should have an intuitive behavior, such that special values of the statistic have an unambiguous interpretation. We found an approach that satisfies all of the above conditions and which can be graphed in an epistasis plot (see Fig~\ref{fig:egl9epistasis}) In an epistasis plot, the X-axis represents the expected $\beta$ coefficient for given gene in a double mutant $a^-b^-$ if $a$ and $b$ interact log-additively. In other words, each individual isoform's x-coordinate is the sum of the regression coefficients from the single mutants $a^-$ and $b^-$. The Y-axis represents the deviations from the log-additive (null) model, and can be calculated as the difference between the predicted and the observed $\beta$ coefficients. Only isoforms that are differentially expressed in all three genotypes are plotted. This attempts to ensure that the isoforms to be examined are regulated by both genes. These plots will generate specific patterns that can be described through linear regressions. The slope of these lines, to which we assign the mathematical notation $s({a,b})$, is the transcriptome-wide epistasis coefficient. Importantly, the transcriptome-wide epistasis coefficient is fundamentally distinct from Pearson or Spearman correlation coefficients and need not have a simple linear mapping. In other words, negative correlation coefficients do not imply a specific sign of the epistasis coefficient, and \emph{vice versa}. For suppression to occur, for example, the only requirement is that the phenotype of the double mutant should match one, and only one, of the two single mutants. The value of the correlation coefficient is not relevant. Transcriptome-wide epistasis coefficients can be understood intuitively for simple cases of genetic interactions if complete genetic nulls are used. If two genes act additively on the same set of differentially expressed isoforms then all the plotted points will fall along the line $y=0$. If two genes act positively in an unbranched pathway, then all the mutants should have the same phenotype. It follows that data from this pathway will form line with slope equal to $-\frac{1}{2}$. On the other hand, in the limit of complete genetic inhibition of $b$ by $a$ in an unbranched pathway (i.e., $a$ is in great excess over $b$, such that under the conditions measured $b$ has no activity), the plots should show a line of best fit with slope equal to $-1$. Genes that interact synthetically (\emph{i.e.}, through an OR-gate) will fall along lines with slopes $>0$. When there is epistasis of one gene over another, the points will fall along one of two possible slopes that must be determined empirically from the single mutant data. We can use both single mutant data to predict the distribution of slopes that results for the cases stated above. Thus, the transcriptome-wide epistasis coefficient integrates information from many different isoforms into a single number (see Fig.~\ref{fig:egl9epistasis}). In our experiment, we studied two double mutants, \eglhif{} and \eglvhl{}. We wanted to understand how well an epistatic analysis based on transcriptome-wide coefficients agreed with the epistasis results reported in the literature, which were based on qPCR of single genes. Therefore, we determined the epistasis coefficient of the two gene combinations we studied (\gene{egl-9} and \gene{vhl-1}, and \gene{egl-9} and \gene{hif-1}). In addition to computing an epistasis coefficient from these factors, we would like to know which gene is suppressed in the double mutant. Suppression means that the double mutant should have exactly the phenotype of one and only one mutant, we can simulate the double mutant by replacing the double mutant data with either of the two single mutants and matching the simulated result to the observed result. The result that most closely matches the real data will reveal which gene is being suppressed, which in turn allows us to order the genes along a pathway. \color{black} We measured the epistasis coefficient between \gene{egl-9} and \gene{vhl-1}, $s({\text{\gene{egl-9} \gene{vhl-1}}}) = -0.41\pm 0.01$ (see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/6_epistasis.html} {Epistasis Notebook}). Simulations using just the single mutant data showed that the double mutant exhibited the \egl{} phenotype (see Fig.~\ref{fig:egl9epistasis}). We used Bayesian model selection to reject a linear pathway (odds ratio (OR) $>10^{92}$), which leads us to conclude \gene{egl-9} is upstream of \gene{vhl-1} acting on a phenotype in a branched manner. We also measured epistasis between \gene{egl-9} and \gene{hif-1}, $s({\text{\gene{egl-9}, \gene{hif-1}}}) = -0.80\pm0.01$ (see SI Figure 2), and we found that this behavior could be predicted by modeling \gene{hif-1} downstream of \gene{egl-9}. We also rejected the null hypothesis that these two genes act in a positive linear pathway (OR$> 10^{93}$). Taken together, this leads us to conclude that \gene{egl-9} strongly inhibits \gene{hif-1}. % epistasis graph \begin{figure}[tbhp] \centering \includegraphics[width=0.5\textwidth]{../final_figs/egl9vhl1-epistasis.pdf} \caption{ (\textbf{A}) Schematic diagram of an epistasis plot. The X-axis on an epistasis plot is the expected coefficient for a double mutant under an log-additive model (null model). The Y-axis plots deviations from this model. Double mutants that deviate in a systematic manner from the null model exhibit transcriptome-wide epistasis ($s$). To measure $s$, we find the line of best fit and determine its slope. Genes that act log-additively on a phenotype \textbf{(Ph)} will have $s=0$ (null hypothesis, orange line); whereas genes that act along an unbranched pathway will have $s=-1/2$ (blue line). Strong repression is reflected by $s=-1$ (red line), whereas $s>0$ correspond to synthetic interactions (purple line). (\textbf{B}) Epistasis plot showing that the \eglvhl{} transcriptome deviates significantly from a null additive. Points are colored qualitatively according to density (purple---low, yellow---high) and size is inversely proportional to the standard error (S.E.) of the y-axis. The green line is the line of best fit from an orthogonal distance regression. (\textbf{C}) Comparison of simulated epistatic coefficients against the observed coefficient. Green curve shows the bootstrapped observed transcriptome-wide epistasis coefficient for \gene{egl-9} and \gene{vhl-1}. Dashed green line shows the mean value of the data. Simulations use only the single mutant data to idealize what expression of the double mutant should look like. $a > b$ means that the phenotype of $a$ is observed in a double mutant $a^-b^-$. } \label{fig:egl9epistasis} \end{figure} % \color{purple} \subsubsection*{Epistasis between two genes can be predicted using an upstream component} \color{black} Given our success in measuring epistasis coefficients, we wanted to know whether it would be possible to predict the epistasis coefficient between \gene{egl-9} and \gene{vhl-1} in the absence of the \egl{} genotype. Since \rhyp{} indirectly activates \eglp{}, we reasoned that the \rhy{} transcriptome should contain almost equivalent information to the \egl{} transcriptome. Therefore, we generated predictions of the epistasis coefficient between \gene{egl-9} and \gene{vhl-1} by substituting in the \rhy{} data, predicting $s({rhy-1,vhl-1}) = -0.45$. Similarly, we used the \eglvhl{} double mutant to measure the epistasis coefficient while replacing the \egl{} dataset with the \rhy{} dataset. We found that the epistasis coefficient using this substitution was $-0.38\pm 0.01$. This coefficient was different from $-0.50$ (OR $>10^{102}$), reflecting the same qualitative conclusion that \gene{vhl-1} represents a branch in the hypoxia pathway. We were able to obtain a close prediction of the epistasis coefficient for two mutants using the transcriptome of a related, upstream mutant. \subsection*{Transcriptomic decorrelation can be used to infer functional distance} \label{sub:decorrelation} % What are functional interactions? So far, we have shown that RNA-seq can accurately measure genetic interactions. However, genetic interactions do not require two gene products to interact biochemically, nor even to be physically close to each other. RNA-seq cannot measure physical interactions between genes, but we wondered whether expression profiling contains sufficient information to order genes along a pathway. Single genes are often regulated by multiple independent sources. The connection between two nodes can in theory be characterized by the strength of the edges connecting them (the thickness of the edge); the sources that regulate both nodes (the fraction of inputs common to both nodes); and the genes that are regulated by both nodes (the fraction of outputs that are common to both nodes). In other words, we expected that expression profiles associated with a pathway would respond quantitatively to quantitative changes in activity of the pathway. Targeting a pathway at multiple points would lead to expression profile divergence as we compare nodes that are separated by more degrees of freedom, reflecting the flux in information between them. % decorrelation \begin{figure}[tbhp] \centering \includegraphics[width=.4\textwidth]{../final_figs/decorrelation.pdf} \caption{ Transcriptomes can be used to order genes in a pathway under certain assumptions. Arrows in the diagrams above are intended to show the direction of flow, and do not indicate valence. \textbf{A}. A linear pathway in which \gene{rhy-1} is the only gene controlling \gene{egl-9}, which in turn controls \gene{hif-1} does not contain information to infer the order between genes. \textbf{B}. If \gene{rhy-1} and \gene{egl-9} have transcriptomic effects that are separable from \gene{hif-1}, then the \gene{rhy-1} transcriptome should contain contributions from \gene{egl-9}, \gene{hif-1} and \gene{egl-9}- and \gene{hif-1}-independent pathways. This pathway contains enough information to infer order. \textbf{C}. If a pathway is branched both upstream and downstream, transcriptomes will show even faster decorrelation. Nodes that are separated by many edges may begin to behave almost independently of each other with marginal transcriptomic overlap or correlation. \textbf{D}. The hypoxia pathway can be ordered. We hypothesize the rapid decay in correlation is due to a mixture of upstream and downstream branching that happens along this pathway. Bars show the standard error of the weighted coefficient from the Monte Carlo Markov Chain computations. } \label{fig:decorrelation} \end{figure} We investigated this possibility by weighting the robust Bayesian regression between each pair of genotypes by the size of the shared transcriptomic phenotype of each pair divided by the total number of isoforms differentially expressed in either mutant ($N_\mathrm{Intersection}/N_{\mathrm{Union}}$). We plotted the weighted correlation of each gene pair, ordered by increasing functional distance (see Fig.~\ref{fig:decorrelation}). In every case, we see that the weighted correlation decreases monotonically due mainly, but not exclusively, to a smaller STP (see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/10_decorrelation.html} {Decorrelation Notebook}). We believe that this result is not due to random noise or insufficiently deep sequencing. Instead, we propose a framework in which every gene is regulated by multiple different molecular species, which induces progressive decorrelation. This decorrelation in turn has two consequences. First, decorrelation within a pathway implies that two nodes may be almost independent of each other if the functional distance between them is large. Second, it may be possible to use decorrelation dynamics to infer gene order in a branching pathway, as we have done with the hypoxia pathway. \subsection{Classical epistasis identifies a core hypoxic response} We searched for genes whose expression obeyed the two epistatic equality relationships, \hif{}=\eglhif{} and \egl{}=\eglvhl{}, since these equalities define the hypoxia pathway. We excluded genes whose expression deviated from this relationship by more than 2 standard deviations or that had opposite changes in direction. Using these criteria, we identified 1,258 genes in the hypoxia response. Tissue Enrichment Analysis showed that the intestine and epithelial system were enriched in this response (\qval{10} for both terms), consistent with previous reports~\cite{Budde2010}. Gene Enrichment Analysis~\cite{Angeles-Albores106369} showed enrichment in the mitochondrion and in collagen trimers (\qval{10}) (see \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/3_ea_of_hypoxia_data.html} {Enrichment Analysis Notebook} and SI Figures 3 and 4). This response included 15 transcription factors. Even though \hifp{} is an activator, not all of these genes were up-regulated. We reasoned that only genes that are up-regulated in \hifp{}-inhibitor mutants are candidates for direct regulation by \hifp{}. We found \hiftargets{} such genes. \subsection*{Feedback can be inferred} \label{sub:topology} While some of the rank plots contained a clear positive correlation, others showed a discernible cross-pattern (see Fig.~\ref{fig:genetic_interactions}). In particular, this cross-pattern emerged between \vhl{} and \rhy{} or between \vhl{} and \egl{}, even though \gene{vhl-1}, \gene{rhy-1} and \gene{egl-9} are all inhibitors of \hif{}. Such cross-patterns could be indicative of feedback loops or other complex interaction patterns. If the above is correct, then it should be possible to identify genes that are regulated by \gene{rhy-1} in a logically consistent way: Since loss of \gene{egl-9} causes \gene{rhy-1} mRNA levels to increase, if this increase leads to a significant change in RHY-1 activity, then it follows that the \egl{} and \rhy{} should show anti-correlation in a subset of genes. Since we do not observe many genes that are anti-correlated, we conclude that is unlikely that the change in \gene{rhy-1} mRNA expression causes a significant change in RHY-1 activity under normoxic conditions. We also searched for genes with \gene{hif-1}-independent, \gene{vhl-1}-dependent gene expression and found \vhltargets{} genes (SI File 1). \subsection*{Identification of non-classical epistatic interactions} \label{sub:hifoh} \hif{} has traditionally been viewed as existing in a genetic OFF state under normoxic conditions. However, our dataset indicates that \hifn{} genes show altered expression when \gene{hif-1} function is removed in normoxic conditions. Moreover, we observed positive correlations between \hif{} $\beta$ coefficients and \egl{}, \vhl{} and \rhy{} $\beta$ coefficients in spite of the negative regulatory relationships between these genes and \gene{hif-1}. Such positive correlations could indicate a relationship between these genes that has not been reported previously. We identified genes that exhibited violations of the canonical genetic model of the hypoxia pathway (see Fig.~\ref{fig:hif1oh}; also \href{https://wormlabcaltech.github.io/mprsq/analysis_notebooks/7_hifoh.html} {Non-canonical epistasis notebook}). We searched for genes that changed in different directions between \egl{} and \vhl{}, or, equivalently, between \rhy{} and \vhl{} (we assume that all results from the \rhy{} transcriptome reflect a complete loss of \gene{egl-9} activity) without specifying any further conditions. We found \hifohtargets{} that satisfied this condition (see Fig.~\ref{fig:hif1oh}, SI File 1). When we checked expression of these genes in the double mutant, we found that \gene{egl-9} remained epistatic over \gene{vhl-1} for this class of genes. This class of genes may in fact be larger because it overlooks genes that have wild-type expression in an \egl{} background, altered expression in a \vhl{} background, and suppressed (wild-type) expression in an \eglvhl{} background. % \color{purple} As a result, it could help explain why the \hif{} mutant transcriptome is positively correlated with its inhibitors. \color{black} \begin{figure}[tbhp] \centering \includegraphics[width=.4\textwidth]{../final_figs/hif1oh_epistasis.pdf} \caption{ \textbf{A}. \hifohtargets{} genes in \cel{} exhibit non-classical epistasis in the hypoxia pathway, characterized by opposite effects on gene expression, relative to the wild type, of the \vhl{} compared to \egl{} (or \rhy{}) mutants. Shown are a random selection of 15 out of \hifohtargets{} genes for illustrative purposes. \textbf{B}. Genes that behave non-canonically have a consistent pattern. \vhl{} mutants have an opposite effect to \egl{}, but \gene{egl-9} remains epistatic to \gene{vhl-1} and loss-of-function mutations in \gene{hif-1} suppress the \egl{} phenotype. Asterisks show $\beta$ values significantly different from 0 relative to wild type (\qval{1}). } \label{fig:hif1oh} \end{figure} Although this entire class had similar behavior, we focused on two genes, \nlp{} and \ftna{} which have representative expression patterns. \ftna{} is described to be responsive to mutations in the hypoxia pathway and has been reported to have aberrant behaviors; specifically, loss of function of \gene{egl-9} and \gene{vhl-1} have opposing effects on \ftna{} expression~\cite{Ackerman2012,Romney2011}. These studies showed the same \ftna{} expression phenotypes using RNAi and alleles, allaying concerns of strain-specific interference. We observed that \gene{hif-1} was epistatic to \gene{egl-9}, and that \gene{egl-9} and \gene{hif-1} both promoted \ftna{} expression. Analysis of \ftna{} expression reveals that \gene{egl-9} is epistatic to \gene{hif-1}; that \gene{vhl-1} has opposite effects to \gene{egl-9}, and that \gene{vhl-1} is epistatic to \gene{egl-9}. Analysis of \nlp{} reveals similar relationships. \nlp{} expression is decreased in \hif{}, and increased in \egl{}. However, \gene{egl-9} is epistatic to \gene{hif-1}. Like \ftna{}, \gene{vhl-1} has the opposite effect to \gene{egl-9}, yet is epistatic to \gene{egl-9}. We propose in the Discussion a novel model for how \hifp{} might regulate these targets. \section*{Discussion} \label{sec:discussion} \subsection*{The \cel{} hypoxia pathway can be reconstructed \emph{de novo} from RNA-seq data} We have shown that whole-organism transcriptomic phenotypes can be used to reconstruct genetic pathways and to discern previously uncharacterized genetic interactions. We successfully reconstructed the hypoxia pathway including the order of action of the genetic components and its branching pattern. % \color{purple} These results highlight the potential of whole-animal expression profiles for dissecting molecular pathways that are expressed in a large number of cells within an organism. While our results are promising, it remains to be seen whether our approach will also work for pathways that act in a few cells. We selected a previously characterized pathway because \cel{} is less amenable to high-throughput screens compared to cultured cells. That said, the striking nature of our results makes us optimistic that this technique could be successfully used to reconstruct unknown pathways. \color{black} \subsection*{Interpretation of the non-classical epistasis in the hypoxia pathway} The \hifohtargets{} genes that exhibit a striking pattern of non-classical epistasis suggest the existence of previously undescribed aspects of the hypoxia pathway. Some of these non-classical behaviors had been observed previously~\cite{Ackerman2012,Romney2011,Luhachack2012}, but no satisfactory mechanism has been proposed to explain them. Previous studies~\cite{Romney2011,Ackerman2012} suggested that \hifp{} integrates information on iron concentration in the cell to determine its binding affinity to the \ftna{} promoter, but could not definitively establish a mechanism. It is unclear why deletion of \gene{hif-1} and deletion of \gene{egl-9} both cause induction of \ftna{} expression, but deletion of \gene{vhl-1} abolishes this induction. Moreover, Luchachack et al~\cite{Luhachack2012} have previously reported that certain genes important for the \cel{} immune response against pathogens reflect similar non-canonical expression patterns. Their interpretation was that \gene{swan-1}, which encodes a binding partner to \eglp{}~\cite{Shao2010}, is important for modulating \hifp{} activity in some manner. The lack of a conclusive double mutant analysis in this work means the role of SWAN-1 in modulation of \hifp{} activity remains to be demonstrated. Other mechanisms, such as tissue-specific differences in the pathway~\cite{Budde2010} could also modulate expression, though it is worth pointing out that \ftna{} expression appears restricted to a single tissue, the intestine~\cite{Kim2004}. Another possibility is that \gene{egl-9} controls \gene{hif-1} mRNA stability via other \gene{vhl-1}-independent pathways, but we did not see a decreases in \gene{hif-1} level in \egl{}, \rhy{} or \vhl{} mutants. Another possibility, such as control of protein stability via \gene{egl-9} independently of \gene{vhl-1}~\cite{Chintala2012} will not lead to splitting unless it happens in a tissue-specific manner. \begin{figure}[tbhp] \centering \includegraphics[width=.5\textwidth]{../final_figs/hif1oh_model.pdf} \caption{ A hypothetical model showing a mechanism where \hifp{}-hydroxyl antagonizes \hifp{} in normoxia. \textbf{A}. Diagram showing that RHY-1 activates EGL-9. EGL-9 hydroxylates HIF-1 in an oxygen-dependent manner. HIF-1 is rapidly hydroxylated and the product, HIF-1-OH is rapidly degraded in a VHL-1-dependent fashion. EGL-9 can also inhibit HIF-1 in an oxygen-independent fashion. In our model, HIF-1 and HIF-1-OH have opposing effects on transcription. The width of the arrows represents rates in normoxic conditions. \textbf{B}. Table showing the effects of loss-of-function mutations on HIF-1 and HIF-1-OH activity, showing how this can potentially explain the \gene{ftn-1} expression levels in each case. S.S = Steady-state. } \label{fig:hif1oh_table} \end{figure} One parsimonious solution is to consider \hifp{} as a protein with both activating and inhibiting states. In fact, \hifp{} already exists in two states in \cel{}: unmodified \hifp{} and \hifp{}-hydroxyl (\hifp{}-OH). Under this model, the effects of \hifp{} for certain genes like \ftna{} or \nlp{} are antagonized by \hifp{}-hydroxyl, which is present at only a low level in the cell in normoxia because it is degraded in a \gene{vhl-1}-dependent fashion. This means that loss of \gene{vhl-1} stabilizes \hifp{}-hydroxyl. If \gene{vhl-1} is inactivated, genes that are sensitive to \hifp{}-hydroxyl will be inhibited as a result of the increase in \hifp{}-hydroxyl, despite the increased levels of non-hydroxylated \hifp{}. On the other hand, \egl{} abrogates the generation of \hifp{}-hydroxyl, stimulating accumulation of non-hydroxylated \hifp{} and promoting gene expression. Whether deletion of \hif{} is overall activating or inhibiting will depend on the relative activity of each protein state under normoxia (see Fig.~\ref{fig:hif1oh_table}). \hifp{}-hydroxyl is challenging to study genetically, and if it does have the activity suggested by our genetic evidence this may have prevented such a role from being detected. No mimetic mutations are known with which to study the pure hydroxylated \hifp{} species, and mutations in the Von Hippel-Lindau gene that stabilize the hydroxyl species also increase the quantity of non-hydroxylated \hifp{} by mass action. Because \hifp{} is detected at low levels in cells under normoxic conditions~\cite{Wang1993}, total \hifp{} protein levels are assumed to be so low as to be biologically inactive. However, our data show \hifn{} genes change expression in response to loss of \gene{hif-1} under normoxic conditions, which establishes that there is sufficient total \hifp{} protein to be biologically active. Our analyses also revealed that \hif{} shares positive correlations with \egl{}, \rhy{} and \vhl{}, and that each of these genotypes also shows a secondary negative rank-ordered expression correlation with each other. % These % cross-patterns between all loss of function of inhibitors of \hifp{} and \hif{} % can be most easily explained if \hifp{}-hydroxyl is biologically active. A homeostatic argument can be made in favor of the activity of \hifp{}-hydroxyl. The cell must continuously monitor multiple metabolite levels. The \gene{hif-1}-dependent hypoxia response integrates information from O$_2$, $\alpha$-ketoglutarate and iron concentrations in the cell. One way to integrate this information is by encoding it within the effective hydroxylation rate of \hifp{} by \eglp{}. Then the dynamics in this system will evolve exclusively as a result of the total amount of \hifp{} in the cell. Such a system can be sensitive to fluctuations in the absolute concentration of \hifp{}~\cite{Goentoro2009a}. Since the absolute levels of \hifp{} are low in normoxic conditions, small fluctuations in protein copy-number can represent a large fold-change in \hifp{} levels. These fluctuations might not be problematic for genes that must be turned on only under conditions of severe hypoxia---presumably, these genes would be activated only when \hifp{} levels increase far beyond random fluctuations. For yet other sets of genes that must change expression in response to the hypoxia pathway, it may not be sufficient to integrate metabolite information exclusively via \eglp{}-dependent hydroxylation of \hifp{}. In particular, genes that may function to increase survival in mild hypoxia may benefit from regulatory mechanisms that can sense minor changes in environmental conditions and which therefore benefit from robustness to transient changes in protein copy number. Likewise, genes that are involved in iron or $\alpha$-ketoglutarate metabolism (such as \ftna{}) may benefit from being able to sense, accurately, small and consistent deviations from basal concentrations of these metabolites. For these genes, the information may be better encoded by using \hifp{} and \hifp{}-hydroxyl as an activator/repressor pair. Such circuits are known to possess distinct advantages for controlling output robustly to transient fluctuations in the levels of their components~\cite{Hart2012,Hart2013}. Our RNA-seq data suggests that one of these atypical targets of \hifp{} may be \rhyp{}. Although \gene{rhy-1} does not exhibit non-classical epistasis, all genotypes containing a \hif{} mutation had increased expression levels of \gene{rhy-1}. We speculate that if \gene{rhy-1} is controlled by both \hifp{} and \hifp{}-hydroxyl, then this might imply that \hifp{} auto-regulates both positively and negatively. % \color{purple} \subsection*{Strengths and weaknesses of the methodology} We have described a set of methods that can in principle be applied to any multidimensional phenotype. Although we have not applied these methods to \emph{de novo} pathway discovery, we believe that they will be broadly applicable to a wide variety of genetic problems. One aspect of our methodology is the use of whole-organism expression data. Data collection from whole-organisms can be rapid with low technical barriers. On the other hand, a concern is that whole-organism data will average signals across tissues, which would limit the scope of this technology to the study of genetic pathways that are systemic or expressed in large tissues. In reality, our method may be applicable for pathways that are expressed even in a small number of cells in an organism. If a pathway is active in a single cell, this does not mean that it does not have cell-non-autonomous effects that could be detected on an organism-wide level. Thus, pathways that act in single cells could still be characterized via whole-organism transcriptome profiling. If the non-autonomous effects are long-lasting, then the profiling could take place after the time-of-action of this pathway. In fact, this is how the female-like state in \cel{} was recently identified~\cite{Angeles-Albores2017a}: \gene{fog-2} is involved in translation repression of \gene{tra-2} in the somatic gonad, thereby promoting sperm formation in late larvae~\cite{Clifford2000}. Loss of this gene causes non-cell-autonomous effects that can be detected well after the time-of-action of \gene{fog-2} in the somatic gonad has ended. Therefore, we believe that our methodology will be applicable to many genetic cases, with the exception of pathways that acts in complex, antagonistic manners depending on the cell type, or if the pathway minimally affects gene expression. % Here, we have described a set of methods that can be applied to any vectorial % phenotype studied with an appropriate experimental design. Transcriptome % profiling methods offers a lot of information, but transcriptome-wide % interpretation of the results is often extremely challenging. Each method has % its own advantages and disadvantages. % % \color{purple} % PCA is computationally tractable and clusters are often % visually detectable. However, PCA can be misleading, especially when % the viewed components do not explain a large fraction of the variance % in the data. In addition, principal components are the product of a % linear combination of vectors, limiting their interpretability. In our % case, the first principal component could be regarded as \hifp{} % pseudo-abundance~\cite{Lonnberg2017} by hypothesizing that values in this % component somehow reflect \hifp{} abundance. % \color{black} % % Whereas PCA operates on all genotypes simultaneously, correlation analysis is a % pairwise procedure. Like PCA, correlation analysis is computationally fast. % Unlike PCA, the product of a correlation analysis is a single number with a % straightforward interpretation. However, correlation analysis is sensitive to % outliers. Although outliers can be mitigated via rank-transformations, these % transformations cannot remove outliers resulting from systematic variation % caused, for example, by branching. Such interactions can lead to vanishing % correlations if both are equally strong. Adequately weighted correlations could % be informative for ordering genes along pathways. A drawback of correlation % analysis is that the number of pairwise comparisons increases combinatorially. Genetic analysis of transcriptomic data has proved challenging as a result of its complexity. Although dimensionality reduction techniques such as PCA have emerged as powerful methods with which to understand these data, these methods generate reduced coordinates which are difficult or impossible to interpret. As an example, the first principal component in this paper (see Fig.~\ref{fig:pca}) could be interpreted as \hifp{} pseudo-abundance~\cite{Lonnberg2017}. However, another equally reasonable, yet potentially completely different interpretation, is as a pseudo-\hifp{}/\hifp{}-OH ratio. Another way to analyze genetic interactions is via general linear models (GLMs) that include interaction terms between two or more genes. GLMs can quantify the genetic interactions on single transcripts. We and others~\cite{Dixit2016, Angeles-Albores2017a} have used GLMs to perform epistasis analyses of pathways using transcriptomic phenotypes. GLMs are powerful, but they generate a different interaction coefficient for each gene measured. The large number of coefficients makes interpretation of the genetic interaction between two mutants difficult. Previous approaches~\cite{Dixit2016} visualize these coefficients via clustered heatmaps. However, two clusters cannot be assumed to be evidence that two genes interact via entirely distinct pathways. Indeed, the non-classical epistasis examples we described here might cluster separately even though a reasonable model can be invoked that does not require any new molecular players. \color{black} The epistasis plots shown here are a useful way to visualize epistasis in vectorial phenotypes. We have shown how an epistasis plot can be used to identify interactions between two genes by examining the transcriptional phenotypes of single and double mutants. % In reality, epistasis plots % can be generated for any set of measurements involving a set of $N$ mutants (or % conditions) and an $N$-mutant genotype. Epistasis plots can accumulate an arbitrary number of points within them, possess a rich structure that can be visualized and have straightforward interpretations for special slope values. % \color{purple} Epistasis plots and GLMs are not mutually exclusive. A GLM could be used to quantify epistasis interactions at single-transcript resolution, and the results then analyzed using an epistasis plot (for a non-genetic example, see~\cite{Angeles-Albores2017a}). A benefit of epistasis plots is that they enable the computation of a single, aggregate statistic that describes the ensemble behavior of a set of genes. This aggregate statistic is not enough to describe all possible behaviors in a system, but it can be used to establish whether the genes under study are part of a single pathway. In the case of the hypoxia pathway, phenotypes that are downstream of the hypoxia pathway should conform to the genetic equalities, \eglhif{} = \hif{} \textbf{\texttt{AND}} \eglvhl{} = \egl{}. Genes whose expression levels behave strangely, yet satisfy these equalities are downstream of the hypoxia pathway. These anomalous genes cannot be identified via the epistasis coefficient but the epistasis coefficient does provide a unifying framework with which to analyze them by constraining the space of plausible hypotheses. \color{black} Until relatively recently, the rapid generation and molecular characterization of null mutants was a major bottleneck for genetic analyses. Advances in genomic engineering mean that, for a number of organisms, production of mutants is now rapid and efficient. As mutants become easier to produce, biologists are realizing that phenotyping and characterizing the biological functions of individual genes is challenging. This is particularly true for whole organisms, where subtle phenotypes can go undetected for long periods of time. We have shown that whole-animal RNA-sequencing is a sensitive method that can be seamlessly incorporated with genetic analyses of epistasis. \matmethods{ \subsection*{Nematode strains and culture} Strains used were N2 (Bristol), JT307 \gene{egl-9(sa307)}, CB5602 \gene{vhl-1(ok161)}, ZG31 \gene{hif-1(ia4)}, RB1297 \gene{rhy-1(ok1402)}. CB6088 \gene{egl-9(sa307) hif-1(ia4)} CB6116 \gene{egl-9(sa307)};\gene{vhl-1(ok161)}, Lines were grown on standard nematode growth media Petri plates seeded with OP50 \ecol{} at 20\degree{}C~\cite{Brenner1974}. \subsection*{RNA isolation} Lines were synchronized by harvesting eggs via sodium hypochlorite treatment and subsequently plating eggs on food. Worms were staged and based on the time after plating, vulva morphology and the absence of eggs. 30--50 non-gravid young adults were picked and placed in 100 \si{\micro\liter} of TE pH 8.0 (Ambion AM9849) in $0.2$ \si{\milli\liter} PCR tubes on ice. Worms were allowed to settle or spun down by centrifugation and $\sim 80$ \si{\micro\liter} of supernatant removed before flash-freezing in liquid $N_2$. These samples were digested with Recombinant Proteinase K PCR Grade (Roche Lot No. 03115 838001) for 15 \si{\min} at 60\degree{} in the presence of 1\% SDS and 1.25 \si{\micro\liter} RNA Secure (Ambion AM7005). 5 volumes of Trizol (Tri-Reagent Zymo Research) were added to the RNA samples and treated with DNase I using Zymo Research Quick-RNA MicroPrep R1050. Samples were analyzed run on an Agilent 2100 BioAnalyzer (Agilent Technologies). Replicates were selected that had RNA integrity numbers equal to or greater than 9.0 and without bacterial ribosomal bands, except for the ZG31 mutant where one of three replicates had a RIN of 8.3. \subsection*{Library preparation and sequencing} 10 \si{\nano\gram} of total RNA from each sample was reverse-transcribed into cDNA using the Clontech SMARTer Ultra Low Input RNA for Sequencing v3 kit (catalog \#634848) in the SMARTSeq2 protocol~\cite{Picelli2014}. RNA was denatured at 70\degree{}C for 3 \si{\min} in the presence of dNTPs, oligo dT primer and spiked-in quantitation standards (NIST/ERCC from Ambion, catalog \#4456740). After chilling to 4\degree{}C, the first-strand reaction was assembled using a LNA TSO primer~\cite{Picelli2014}, and run at 42\degree{}C for 90 minutes, followed by denaturation at 70\degree{}C for 10 \si{\min}. The first strand reaction was used as template for 13 cycles of PCR using the Clontech v3 kit. Reactions were purified with Ampure XP SPRI beads (catalog \#A63880). After quantification using the Qubit High Sensitivity DNA assay, a 3 \si{\nano\gram} aliquot of the cDNA was run on the Agilent HS DNA chip to confirm the length distribution of the amplified fragments. The median value for the average cDNA lengths from all length distributions was 1,076 bp. Tagmentation of the full length cDNA was performed using the Illumina/Nextera DNA library prep kit (catalog \#FC-121--1030). Following Qubit quantitation and Agilent BioAnalyzer profiling, the tagmented libraries were sequenced on an Illumina HiSeq2500 machine in single read mode with a read length of 50~nt to a depth of 15~million reads per sample. Base calls were performed with RTA 1.13.48.0 followed by conversion to FASTQ with bcl2fastq 1.8.4. % Spearman correlation of % the estimated counts for each genotype showed that every pairwise correlation % within genotype was $0.9$. \subsection*{Read alignment and differential expression analysis} We used Kallisto~\cite{Bray2016} to perform read pseudo-alignment and performed differential analysis using Sleuth~\cite{Pimentel2016}. We fit a general linear model for an isoform $t$ in sample $i$: \begin{equation} y_{t,i} = \beta_{t, 0} + \beta_{t, genotype}\cdot{}X_{t, i} + \beta_{t, batch}\cdot{}Y_{t, i} + \epsilon_{t, i} \end{equation} where $y_{t, i}$ was the logarithm transformed counts of isoform $t$ in sample $i$; $\beta_{t, genotype}$ and $\beta_{t, batch}$ were parameters of the model for the isoform $t$, and which could be interpreted as biased estimators of the log-fold change; $X_{t, i}, Y_{t, i}$ were indicator variables describing the experimental conditions of the isoform $t$ in sample $i$; and $\epsilon_{t, i}$ was the noise associated with a particular measurement. % \color{purple} After fitting the general linear model, we tested isoforms for differential expression using the built-in Wald-test in Sleuth~\cite{Pimentel2016}, which outputs a $q$-value that has been corrected for multiple hypothesis testing. \color{black} \subsection*{Genetic Analysis, Overview} The processed data were analyzed using Python 3.5. We used the Pandas, Matplotlib, Scipy, Seaborn, Sklearn, Networkx, PyMC3, and TEA libraries~\cite{McKinney2011,Oliphant2007, Pedregosa2012,Salvatier2015,VanDerWalt2011,Hunter2007,Angeles-Albores2016,Waskom}. Our analysis is available in Jupyter Notebooks~\cite{Perez2007}. All code and processed data are available at \url{https://github.com/WormLabCaltech/mprsq} along with version-control information. Our Jupyter Notebook and interactive graphs for this project can be found at \url{https://wormlabcaltech.github.io/mprsq/} in html format, or in the Supplementary File 2 in pdf format. Raw reads were deposited in the Short Read Archive under the study accession number SRP100886 and in the GEO under the accession number GSE97355. \subsection*{Weighted correlations} Correlations between mutants were calculated by identifying their STP. Transcripts were rank-ordered according to their regression coefficient, $\beta$. Regressions were performed using a Student-T distribution with the PyMC3 library~\cite{Salvatier2015} (\texttt{pm.glm.families.StudenT} in Python). If the correlations had an average value $>1$, the average correlation coefficient was set to 1. Weights were calculated as the number of genes that were inliers divided by the number of DEGs present in either mutant. \subsection*{Epistatic analysis} The epistasis coefficient between two null mutants \gene{a} and \gene{b} was calculated as: \begin{equation} s(a, b) = \frac{\beta_{a,b} - \beta_a - \beta_b}{\beta_a + \beta_b} \label{eq:epistasis_coef} \end{equation} Null models for various epistatic relationships were generated by sampling the single mutants in an appropriate fashion. For example, to generate the distribution for two mutants that obey the epistatic relationship $a^- = a^-b^-$, we substituted $\beta_{a, b}$ with $\beta_a$ and bootstrapped the result. To select between theoretical models, we implemented an approximate Bayesian Odds Ratio. We defined a free-fit model, $M_1$, that found the line of best fit for the data: \begin{equation} P(\alpha~|M_1, D) \propto \prod_{(x_i, y_i, \sigma_i)\in D} \exp{ [\frac{{(y_i - \alpha\cdot x_i)}^2} % numerator {2\sigma_i^2}] % denominator } \cdot {(1+\alpha^2)}^{-3/2}, \label{eq:free_model} \end{equation} where $\alpha$ was the slope to be determined, $x_i, y_i$ are the of each point, and $\sigma_i$ was the standard error associated with the y-value. We used equation~\ref{eq:free_model} to obtain the most likely slope given the data, $D$, via minimization (\texttt{scipy.optimize.minimize} in Python). Finally, we approximated the odds ratio as: \begin{equation} OR = \frac{ P(D~|\alpha^*, M_1)\cdot {(2\pi)}^{1/2}\sigma_{\alpha^*} % numerator }{P(D~| M_i)}, % denominator \end{equation} where $\alpha^*$ was the slope found after minimization, $\sigma_\alpha^*$ was the standard deviation of the parameter at the point $\alpha^*$ and $P(D~|M_i)$ was the probability of the data given the parameter-free model, $M_i$. \subsection*{Enrichment analysis} Tissue, Phenotype and Gene Ontology Enrichment Analysis were carried out using the WormBase Enrichment Suite for Python~\cite{Angeles-Albores106369, Angeles-Albores2016}. } % end methods \showmatmethods{} % Display the Materials and Methods section \acknow{ This work was supported by HHMI with whom PWS is an investigator and by the Millard and Genetics and Genomics Laboratory at Caltech. Strains were provided by the CGC, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). We thank , , , and for their advice throughout this project.} \showacknow{} % Display the acknowledgments section % \pnasbreak splits and balances the columns before the references. % If you see % unexpected formatting errors, try commenting out this line % as it can run into % problems with floats and footnotes on the final page. % \pnasbreak{} % Bibliography \bibliography{citations} \end{document} application/latex/_constraint_8php.tex \hypertarget{_constraint_8php}{}\section{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/phpunit/phpunit/src/\+Framework/\+Constraint.php File Reference} \label{_constraint_8php}\index{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/phpunit/phpunit/src/\+Framework/\+Constraint.\+php@{/\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/phpunit/phpunit/src/\+Framework/\+Constraint.\+php}} \subsection*{Data Structures} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{class_p_h_p_unit___framework___constraint}{P\+H\+P\+Unit\+\_\+\+Framework\+\_\+\+Constraint}} \end{DoxyCompactItemize} 0 \hypertarget{cylindersizecalculatorbase_8h}{}\section{Файл Projects/labs/course\+\_\+project\+\_\+cg/src/algorithm/cylindersizecalculatorbase.h} \label{cylindersizecalculatorbase_8h}\index{Projects/labs/course\+\_\+project\+\_\+cg/src/algorithm/cylindersizecalculatorbase.\+h@{Projects/labs/course\+\_\+project\+\_\+cg/src/algorithm/cylindersizecalculatorbase.\+h}} {\ttfamily \#include $<$vector$>$}\\* {\ttfamily \#include $<$utility$>$}\\* {\ttfamily \#include \char`\"{}src/math/line.\+h\char`\"{}}\\* {\ttfamily \#include \char`\"{}src/math/cylindersize.\+h\char`\"{}}\\* Граф включаемых заголовочных файлов для cylindersizecalculatorbase.\+h\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d6/d13/cylindersizecalculatorbase_8h__incl} \end{center} \end{figure} Граф файлов, в которые включается этот файл\+:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d9/d53/cylindersizecalculatorbase_8h__dep__incl} \end{center} \end{figure} \subsection*{Классы} \begin{DoxyCompactItemize} \item class \hyperlink{class_cylinder_size_calculator_base}{Cylinder\+Size\+Calculator\+Base} \end{DoxyCompactItemize} agentzero93/Vargo_lab_website @article{RN695, author = {. and Labadie, . and }, doi = {doi:10.1098/rspb.2011.1030}, journal = {Proceedings of the Royal Society B: Biological Sciences}, number = {1729}, pages = {813-819}, title = {Asexual queen succession in the subterranean termite Reticulitermes virginicus}, type = {Journal Article}, url = {https://royalsocietypublishing.org/doi/abs/10.1098/rspb.2011.1030 %X Termite colonies are founded by a pair of primary reproductives. In many species, including subterranean termites (family Rhinotermitidae), the primary king and queen can be succeeded by neotenic reproductives that are produced from workers or nymphs within the colony. It is generally believed that these neotenics inbreed within the colony, sometimes for many generations. Here, we show that primary queens of the North American subterranean termite, Reticulitermes virginicus, are replaced by numerous parthenogenetically produced female neotenics. We collected functional female neotenics from five colonies of R. virginicus in North Carolina and Texas, USA. Genetic analysis at eight microsatellite loci showed that 91–100% of the neotenics present within a colony were homozygous at all loci, indicating that they were produced through automictic parthenogenesis with terminal fusion. In contrast, workers, soldiers and alates were almost exclusively sexually produced by mating between the female neotenics and a single king. This is the second termite species shown to undergo asexual queen succession, a system first described in the Japanese species, Reticulitermes speratus. Thus, the conditional use of sexual and asexual reproduction to produce members of different castes may be widespread within Reticulitermes and possibly other subterranean termites. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3248721/pdf/rspb20111030.pdf}, volume = {279}, year = {2012} } affifboudaoud/U-scolaritedocumentation/latex/_controller_8php.tex \doxysection{Http/\+Controllers/\+Controller.php File Reference} \label{_controller_8php}\index{Http/Controllers/Controller.php@{Http/Controllers/Controller.php}} \doxysubsection*{Data Structures} \begin{DoxyCompactItemize} \item class \textbf{ Controller} \end{DoxyCompactItemize} \doxysubsection*{Namespaces} \begin{DoxyCompactItemize} \item \textbf{ App\textbackslash{}\+Http\textbackslash{}\+Controllers} \end{DoxyCompactItemize} MA530/MA530ComplexAnalysisHwk13.tex \documentclass[a4paper,12pt]{article} \usepackage{fancyhdr} \usepackage{amssymb} %\usepackage{mathpazo} \usepackage{mathtools} \usepackage{amsmath} \usepackage{slashed} \usepackage{cancel} \usepackage[mathscr]{euscript} \usepackage{MaxPackage} %Note: You need MaxPackage installed or in the same folder as your .tex file or something. \newcommand{\colorcomment}[2]{\textcolor{#1}{#2}} %First of these leaves in comments. Second one kills them. %\newcommand{\colorcomment}[2]{} \pagestyle{fancy} \lhead{} \chead{MA530} \rhead{Assignment 13, Page \thepage} %Number of Problems :8 %Clear : %Begun :4,5,6 %Not started :8 %Can complete via book : %Needs Polish :1,2,3,7 %Absolutely broken : %Pomodoros logged :16 \begin{document} Note: in the below, we adopt the notation $\phi_a: D_1(0) \to D_1(0)$ to be given by $\phi_a(z) = \frac{z-a}{1-\overline{a}z}$. This is the same as the $f_a$ given in class, but that notation lends itself to issues in this homework. I should've said this in the other homework as well, but I use $\overline{\C}$ to denote the Riemann Sphere because I can't figure out how to get $\C$ with a hat over it. {\bf Problem 1:} (I googled facts about ellipses in order to get the last equation.) The map described in class is $f\circ g \circ h$, where $f(z) = \frac{z-1}{z+1}$, $g(z) = \sqrt{z}$, and $h(z) = \frac{z-1}{z+1}$. Its inverse is thus $h^{-1} \circ g^{-1} \circ f^{-1}$, which is $F: D_1(0) \to \overline{\C} \setminus [-1,1]$ where $F(z) = \frac{\left(\frac{z+1}{1-z}\right)^2 + 1}{1-\left(\frac{z+1}{1-z}\right)^2} = \frac{-z^2-1}{2z}$. Consider the set $\partial D_r(0)$ where $r < 1$. We see that $F(\partial D_r(0)) = \{\frac{-z^2-1}{2z}: \absval{z} = r\}$. Now, let $p \in F(\partial D_r(0))$, so that for some $z \in D_1(0)$, we have $\frac{-z^2-1}{2z} = p$. Say that $x = \text{Re}(z)$ and $y = \text{Im}(z)$. Then we have \begin{align*} \absval{p+1} + \absval{p-1} &= \absval{\frac{-z^2 -1}{2z} + 1} + \absval{\frac{-z^2 -1}{2z} -1}\\ &= \absval{\frac{-z^2 +2z-1}{2z}} + \absval{\frac{-z^2 -2z-1}{2z}}\\ &= \frac{1}{2r}\left[\absval{-z^2 +2z-1} + \absval{-z^2 -2z-1}\right]\\ &= \frac{1}{2r}\left[\absval{z-1}^2 + \absval{z+1}^2\right]\\ &= \frac{1}{2r}\left[(x-1)^2+y^2 + (x+1)^2+y^2\right]\\ &= \frac{1}{2r}\left[2y^2+2x^2+2\right]\\ &= \frac{r^2+1}{r}\\ \end{align*} So the sum of the distance from $p$ to $1$ and from $p$ to $-1$ does not depend on $p$; this set is an ellipse with foci at $1$ and $-1$; the length of the semimajor axis is given by half the sum of the distances; that is, it is $\frac{r^2+1}{2r}$. To be more explicit, the set is given by the equation $1= \frac{x}{(\frac{r^2+1}{2r})^2} + \frac{y}{1- (\frac{r^2+1}{2r})^2}$, where $x = \text{Re}(z)$ and $y = \text{Im}(z)$. \shunt {\bf Problem 2:} Consider $\phi_{f(0)}(f(z))$. Taking a derivative, we get: \begin{align*} (\phi_{f(0)}(f(z)))' &= \phi_{f(0)}'(f(z))f'(z)\\ f'(z) &= \frac{(\phi_{f(0)}(f(z)))'}{\phi_{f(0)}'(f(z))}\\ \absval{f'(z)} &= \absval{\frac{(\phi_{f(0)}(f(z)))'}{\phi_{f(0)}'(f(z))}}\\ \absval{f'(0)} &= \absval{\frac{(\phi_{f(0)}(f(0)))'}{\phi_{f(0)}'(f(0))}}\\ \absval{f'(0)} &= (1-\absval{f(0)}^2)\absval{(\phi_{f(0)}(f(0)))'}\\ \end{align*} with the last line being because $\absval{\phi_{a}'(a)} = \frac{1}{1-\absval{a}^2}$, which was discussed in class. Moreover, $\absval{(\phi_{f(0)}(f(0)))'} \leq 1$, by Schwarz's lemma. (Note that $\phi_{f(0)}(f(z))$ is a holomorphic map fixing the origin, so its derivative at the origin is at most $1$.) Thus, we have $\absval{f'(z)} \leq (1-\absval{f(0)}^2)$. \shunt {\bf Problem 3:} Fix $z \in D_1(0)$. Consider $f(\phi_{-z}(w))$ as a function of $w$. Taking a derivative, we get: \begin{align*} (f(\phi_{-z}(w)))' &= f'(\phi_{-z}(w))\phi_{-z}'(w)\\ (f(\phi_{-z}(0)))' &= f'(\phi_{-z}(0))\phi_{-z}'(0)\\ \frac{(f(\phi_{-z}(0)))'}{\phi_{-z}'(0)} &= f'(\phi_{-z}(0))\\ \frac{(f(\phi_{-z}(0)))'}{\phi_{-z}'(0)} &= f'(z)\\ \absval{\frac{(f(\phi_{-z}(0)))'}{\phi_{-z}'(0)}} &= \absval{f'(z)}\\ \absval{f'(z)}&= \absval{f(\phi_{-z}(0)))'} \frac{1}{1-\absval{-z}^2}\\ \absval{f'(z)}&\leq \frac{1}{1-\absval{z}^2}\\ \end{align*} With the last line being by Schwarz's lemma, as above. \shunt {\bf Problem 4:} Consider $\{z \in \C: A \absval{z}^2 + 2\text{Re}(Bz^2) + 2\text{Re}(Cz) + D = 0\}$, with $A,D \in \R$, $B,C \in \C$ ($A,B,C,D$ fixed). We are given that this represents a conic section. This describes a line when $A=B=0$; If $A$ or $B$ is nonzero, then consider any two points $z,z'$ in the set where $z \neq z'$. Then $\frac{z+z'}{2}$ is not in the set; \begin{align*} &A \absval{\frac{z+z'}{2}}^2 + 2\text{Re}(B(\frac{z+z'}{2})^2) + 2\text{Re}(C\frac{z+z'}{2}) + D\\ &= A \absval{\frac{z+z'}{2}}^2 + \frac{1}{2}\text{Re}(B(z+z')^2) + \text{Re}(C(z+z')) + D\\ &= \frac{A}{4} \absval{z+z'}^2 + \frac{1}{2}\text{Re}(Bz^2)+ \frac{1}{2}\text{Re}(Bz'^2) + \text{Re}(Bzz') + \text{Re}(Cz) +\text{Re}(Cz')+ D\\ &= \frac{A}{4} \absval{z+z'}^2 -\frac{A}{2}\absval{z}^2 -\frac{A}{2}\absval{z'}^2- \frac{1}{2}\text{Re}(Bz^2)- \frac{1}{2}\text{Re}(Bz'^2) + \text{Re}(Bzz')\\ &= \frac{A}{4} \left[\absval{z+z'}^2 -2\absval{z}^2 -2\absval{z'}^2\right] + \text{Re}(B(zz'-\frac{z^2+z'^2}{2}))\\ \end{align*} (I admit that I can't see why this only vanishes when $z=z'$; I currently believe that I have under-determined the problem.) %Fix it if possible Thus, this could not have described a line; the midpoint of two points on it fails to be in the set. However, if $A=B=0$, then the set becomes $\{z \in \C: 2\text{Re}(Cz) = D\}$, which is rather clearly a line (if this is not clear, repeating the same analysis as above yields that the midpoint of any two points is in the set...and the only conic section satisfying this is the line.). This describes a circle when $B=0$ and $A \neq 0$; if $B=0$, then the set described is $\{z \in \C: A\absval{z}^2 + 2\text{Re}(Cz) + D = 0\}$. The equation reduces to: \begin{align*} 0=A\absval{z}^2 + 2\text{Re}(Cz) + D &= A\absval{z}^2 + \overline{z}C + \overline{C}z + D\\ &= A\absval{z}^2 + \overline{z}C + \overline{C}z + D\\ &= A(z+C)(\overline{z}+\overline{C}) + D-\absval{C}^2\\ 0&= A\absval{z+C}^2 + D-\absval{C}^2\\ \frac{\absval{C}^2-D}{A}&= \absval{z+C}^2\\ \end{align*} which represents a circle. Performing the same analysis in reverse yields that circles have the general form described above, finishing the problem. \shunt {\bf Problem 5:} (Note: I had read this in Complex Made Simple before this was assigned.) Let $\phi \in \text{Aut}(\overline{\C})$. Say $\scrC$ is the set of all circles and lines in the complex plane. Note that $\text{Aut}(\overline{\C})$ is the set of linear-fractional transformations. Further note that the set of linear-fractional transformations is generated, as a group, by the maps of the form $z \mapsto az+b$ (with $a,b \in \C$) and the map $z \mapsto 1/z$. It suffices to show our result for the generating set. The result is clear for linear maps (note that they're a dilation followed by a translation followed by a rotation.) For the map $f(z) = 1/z$, let $\ell$ be a line through the origin: that is, $\ell = \{z \in \C: \exists r \in \overline{\R}: z=\ep r\}$ for some fixed $\ep \in \C$ with $\absval{\ep} = 1$. Then $f(\ell)$ is another line: $f(\ell) = \{z \in \C: \exists r \in \overline{\R}: z=1/(\ep r)\}$; note that $\absval{1/\ep} = 1$ and $1/r$ is an automorphism of $\overline{\R}$. If $\ell$ is a line that misses the origin: that is, $\ell = \{z \in \C: \exists r \in \overline{\R}: z=\ep r+ c\}$ for some fixed $\ep \in \C$ and $c \in \C$ with $\absval{\ep} = 1$ . Then $f(\ell)$ is a circle: $f(\ell) = \{z \in \C: \exists r \in \overline{\R}: z = 1/(\ep r + c)\}$, which is a circle. (I am somewhat certain we discussed this in class.) Let $\Ga$ be a circle centered at the origin: that is, $\Ga = \{z \in \C: \absval{z} = r\}$ for some fixed $r \in \R$. Then $f(\Ga) = \{1/z \in \C: \absval{z} = r\} = \{z \in \C: \absval{1/z} = r\} = z \in \C: \absval{z} = 1/r\}$. That is, $f(\Ga)$ is a circle centered at the origin of radius $1/r$. Let $\Ga$ be a circle not centered at the origin: that is, $\Ga = \{z \in \C: \absval{z-c} = r\}$ for some fixed $r \in \R$ and $c \in \C$. Then $f(\Ga) =\{\}$ %Finish it. So in all cases, $f(z) = 1/z$ maps $\scrC$ to itself. So we have the desired result. \shunt {\bf Problem 6:} (A fragment of this is an exercise in Complex Made Simple, and a hint is given.) Let $\Om \subset \C$ be open and connected, $f_n \in \scrO(\Om)$, $\sup(\absval{f_n(z)}) = L < \infty$, $\xi_j \in \Om$, (with each $\xi_j$ distinct), $\xi_j \to \xi \in \Om$, and $f_n(\xi_j) \to \Xi_j$ for some $\Xi_j$. By Vitali-Montel, there's a subsequence of $f_n$, call it $f_{n_k}$, that converges locally uniformly to some holomorphic function, $f$. Now, there is only one holomorphic function, $f$, such that $f_{n_k}$ converges to $f$ for some subsequence $f_{n_k}$: this is because for any such function, $f(\xi_j) = \lim_{n \to \infty} f_n(\xi_j)$, so $f$ is determined on $\xi_j$...so $f$ is unique, by the uniqueness theorem. Now, if $f_n(x) \slashed{\to} f(x)$ at any given point, then there's a subsequence of $f_n(x)$ converging to some point other than $f(x)$ (because $f_n$ is bounded, $f_n(x)$ lies in a compact space). But then there's a subsequence of $f_n$ converging to a function other than $f$...which contradicts our result above. So $f_n$ converges to $f$, and $f_n$ has a subsequence ,$f_{n_k}$, that converges locally uniformly to $f$. Therefore, $f_n$ converges locally uniformly to $f$: this is probably an epsilon-delta proof that I ran out of time for. \shunt {\bf Problem 7:} Consider $\text{Aut}(\C \setminus \{0\})$. Let $\phi \in \text{Aut}(\C \setminus \{0\})$. Then $\phi$ is an injective holomorphism with singularities at $0$ and $\infty$. By the exam problem, $\phi$ has removable singularities or (first order) poles at $0$ and $\infty$. If $\phi$ has a removable singularity at $0$, then $\phi$ is extended naturally to an automorphism of $\C$. Thus, $\phi$ is given by $z \mapsto az+b$ for some $a,b \in \C$. Note that $b=0$ in this case, otherwise $\phi(-b/a) = 0$, so that $\phi$ is no longer well defined. Moreover, $a \neq 0$, else $\phi$ isn't injective. So if $\phi$ has a removable singularity at $0$, then $\phi$ is given by $z \mapsto az$ for some $a \in \C$, $a \neq 0$. Next, let $\phi$ have a pole at $0$. Then $1/\phi$ has a removable singularity at $0$, and $1/\phi$ is an automorphism of $\C \setminus \{0\}$. So $1/\phi$ is given by $z \mapsto az$ for some $a \in \C$ , so $\phi$ is given by $z \mapsto 1/(az)$. That is, $\phi \in \text{Aut}(\C \setminus \{0\})$ is given by $z \mapsto az$ or $z \mapsto a/z$ for some $a \in \C$ where $a \neq 0$. It is somewhat clear that all such maps are automorphisms of $\C \setminus \{0\}$, so we have $\text{Aut}(\C \setminus \{0\})$ being the set of maps given by $z \mapsto az$ or $z \mapsto a/z$ for some $a \in \C$ where $a \neq 0$. %Apply 1/z and regress to the above situation. \shunt {\bf Problem 8:} I ran out of time, this one is left undone. \shunt \end{document}@inproceedings{5d2f9a514f21460d9453c57a96ca1ea0, abstract = {Ambient noise is generally seen as an unwanted excitation that disturbs the estimation of vibration parameters. Av- eraging techniques are then used to decrease as much as possible the influence of the noise. However, this noise also excites the mechanical structure and thus increases the vibration response level. Moreover, it is possible that (broadband) noise excites vibration modes that are not well excited by the artificially applied forces. Those modes are missed by clas- sical estimation methods. Recently, classical EMA and OMA were combined into the so-called OMAX framework. In this framework both the artificial force and the ambient excitation are considered useful in determining the modal parameters. In this paper it is shown that the classical frequency-domain modal parameter estimators (rational fraction polynomial based and state space based) can be used without changing them, if the correct non-parametric preprocessing is applied to calculate the frequency response function (FRF) and the power spectrum (PSD). Special attention is paid to the case of structure-exciter interaction, where a direct OMAX approach would result in erroneous results. Also the importance of scaling the FRF and PSD is discussed. The approach is demonstrated on a typical OMAX case: flight flutter test of an airplane wing.}, author = {, and }, booktitle = {Proceedings of IMAC XXVIII, A Conference and Exposition on Structural Dynamics}, keywords = {modal analysis, omax}, language = {English}, month = {2}, title = {Frequency-domain modal analysis in the OMAX framework}, year = {2010} } enukane/c91-capturing-80211-2016 \geometry{top=20mm,bottom=10mm,left=15mm,right=20mm} \setlength\floatsep{2pt} \setlength\textfloatsep{2pt} \setlength\intextsep{2pt} \setlength\abovecaptionskip{2pt} \setlength{\textwidth}{\fullwidth} \setlength{\evensidemargin}{\oddsidemargin} \makeatletter \renewcommand{\reviewappendix}[0]{% \if@openright\cleardoublepage\else\clearpage\fi \thispagestyle{empty} \refstepcounter{part} \addcontentsline{toc}{part}{おわりに\hspace{1zw}} \vspace*{2\Cvs} {\parindent \z@ \raggedright \normalfont \interlinepenalty\@M \Huge \headfont おわりに\par\nobreak \vskip 3\Cvs} } \makeatother @misc{rfc5072, series = {Request for Comments}, number = 5072, howpublished = {RFC 5072}, publisher = {RFC Editor}, doi = {10.17487/RFC5072}, url = {https://rfc-editor.org/rfc/rfc5072.txt}, author = {}, title = {{IP Version 6 over PPP}}, pagetotal = 16, year = 2007, month = sep, abstract = {The Point-to-Point Protocol (PPP) provides a standard method of encapsulating network-layer protocol information over point-to-point links. PPP also defines an extensible Link Control Protocol, and proposes a family of Network Control Protocols (NCPs) for establishing and configuring different network-layer protocols. This document defines the method for sending IPv6 packets over PPP links, the NCP for establishing and configuring the IPv6 over PPP, and the method for forming IPv6 link-local addresses on PPP links. It also specifies the conditions for performing Duplicate Address Detection on IPv6 global unicast addresses configured for PPP links either through stateful or stateless address autoconfiguration. This document obsoletes RFC 2472. {[}STANDARDS-TRACK{]}}, } @inproceedings{yoon2016talking, title={Talking with tact: Polite language as a balance between kindness and informativity}, author={ and Tessler, and Goodman, and }, booktitle={Proceedings of the 38th annual conference of the cognitive science society}, pages={2771--2776}, year={2016}, organization={Cognitive Science Society} } @article{grice1975logic, title={Logic and conversation}, author={ and and and others}, journal={1975}, pages={41--58}, year={1975} } @article{frank2012predicting, title={Predicting pragmatic reasoning in language games}, author={ and Goodman, }, journal={Science}, volume={336}, number={6084}, pages={998--998}, year={2012}, publisher={American Association for the Advancement of Science} } @book{brown1987politeness, title={Politeness: Some universals in language usage}, author={ and Levinson, and Levinson, }, volume={4}, year={1987}, publisher={Cambridge university press} } talentositoazul/proyectomio20 \hypertarget{dir_0b0571c19b25aba665ce34b0cf98ed67}{}\section{C\+:/wamp64/www/proyectomio2/todo/views Directory Reference} \label{dir_0b0571c19b25aba665ce34b0cf98ed67}\index{C\+:/wamp64/www/proyectomio2/todo/views Directory Reference@{C\+:/wamp64/www/proyectomio2/todo/views Directory Reference}} \subsection*{Files} \begin{DoxyCompactItemize} \item file \hyperlink{actualizar__aprendiz_8blade_8php}{actualizar\+\_\+aprendiz.\+blade.\+php} \item file \hyperlink{actualizar__productos_8blade_8php}{actualizar\+\_\+productos.\+blade.\+php} \item file \hyperlink{actualizar__programa__de__formacion_8blade_8php}{actualizar\+\_\+programa\+\_\+de\+\_\+formacion.\+blade.\+php} \item file \hyperlink{actualizar__proveedor_8blade_8php}{actualizar\+\_\+proveedor.\+blade.\+php} \item file \hyperlink{actualizar__referencia_8blade_8php}{actualizar\+\_\+referencia.\+blade.\+php} \item file \hyperlink{aprendiz_8blade_8php}{aprendiz.\+blade.\+php} \item file \hyperlink{home_8blade_8php}{home.\+blade.\+php} \item file \hyperlink{listaraprendiz_8blade_8php}{listaraprendiz.\+blade.\+php} \item file \hyperlink{listarproductos_8blade_8php}{listarproductos.\+blade.\+php} \item file \hyperlink{listarprograma__de__formacion_8blade_8php}{listarprograma\+\_\+de\+\_\+formacion.\+blade.\+php} \item file \hyperlink{listarproveedor_8blade_8php}{listarproveedor.\+blade.\+php} \item file \hyperlink{listarreferencia_8blade_8php}{listarreferencia.\+blade.\+php} \item file \hyperlink{productos_8blade_8php}{productos.\+blade.\+php} \item file \hyperlink{programa__de__formacion_8blade_8php}{programa\+\_\+de\+\_\+formacion.\+blade.\+php} \item file \hyperlink{proveedor_8blade_8php}{proveedor.\+blade.\+php} \item file \hyperlink{referencia_8blade_8php}{referencia.\+blade.\+php} \end{DoxyCompactItemize} -- ist studierter Philosoph. Vielleicht wurde aus ihm, aus heutiger Sicht eher ein politischer Philosoph, denn Philosophie hat in der Zeit nach ihm ihre Unschuld verloren bzw. aufgegeben. Werkzeug, Aufgabe und Ausdrucksform in einem. -- Soziologie ========== De Omnibus Dubitandum (We ought to question everything) — ’s favourite motto. Heute zählt sein Werk zu den Grundlagentexten der Soziologie. Er ist Referenz für eine Betrachtungsweise von Gesellschaft, die Gruppen mit einander entgegengesetzten Interessen und die daraus hervorgehenden Kämpfe und Konflikte untersucht. Sein Lebenswerk, die sozio-ökonomische Analyse der kapitalistischen Gesellschaft, die Darstellung der Faktoren, welche zu ihrer Entstehung führten und die Prognosen zur deren weiteren Entwicklung, werden heute als wichtiger Beitrag zur Entwicklung der Soziologie anerkannt. -- Die Politikwissenschaft knüpft ebenfalls zum Beispiel bei Theorien der internationalen Beziehungen an die Kritik der Politischen Ökonomie an. -- Doch wie sehen wir ihn heute als Ökonom? * Die zentralisierte Planwirtschaft die in sozialistischen Staaten im 20. Jahrhundert praktiziert wurde, hatte neben einzelnen Erfolgen, langfristig wenige Vorteile im Vergleich zur kapitalistischen Marktwirtschaft. * Seine Arbeitswertheorie wird als ideologisch begründet angesehen und sie kann nicht die Entstehung von Marktpreisen erklären. -- Marx war Humanist, aber kein Utopist, denn Utopia liegt im Nirgendwo. -- Für Marx war die Überwindung des Kapitalismus eine Selbstverständlichkeit. Wenn die Menschen die Chance bekommen, diesen Verhältnissen zu entkommen so gab es für ihn keinen Zweifel, dann würden sie diese ergreifen. Die realen Arbeits- und Lebensbedingungen des Proletariats ließen ihn in dieser Frage zumindest ziemlich sicher sein. -- Unter Diktatur des Proletariats ist die Selbstbesimmung der Arbeiter über ihre Arbeit zu verstehen. Ein Zustand der zu Lebzeiten von Karl Marx während der Errichtung der Pariser Kommune angestrebt wurde. Wann und wie hat der Begriff der Diktatur seine heutige Bedeutung erhalten? -- Quellen: Marx and Marxism - Der Staat im globalen Kapitalismus - (2013) \documentclass[14pt]{article} \usepackage{fancyhdr} \usepackage{extramarks} \usepackage{amsmath} \usepackage{amsthm} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{tikz} \usepackage[plain]{algorithm} \usepackage{algpseudocode} \usepackage{enumitem} \usepackage{relsize} \usepackage{scrextend} \usepackage{graphicx} \usepackage{listings} \usetikzlibrary{automata,positioning} % % Assembly language code listing % \lstset{language=[x86masm]Assembler} % % Basic Document Settings % \topmargin=-0.45in \evensidemargin=0in \oddsidemargin=0in \textwidth=6.5in \textheight=9.0in \headsep=0.25in \linespread{1.1} \pagestyle{fancy} \lhead{\hmwkAuthorName} \chead{\hmwkClass\ (\hmwkClassInstructor): \hmwkTitle} \rhead{\firstxmark} \lfoot{\lastxmark} \cfoot{\thepage} \renewcommand\headrulewidth{0.4pt} \renewcommand\footrulewidth{0.4pt} \setlength\parindent{0pt} % % Create Problem Sections % \newcommand{\enterProblemHeader}[1]{ \nobreak\extramarks{}{Problem \arabic{#1} continued on next page\ldots}\nobreak{} \nobreak\extramarks{Problem \arabic{#1} (continued)}{Problem \arabic{#1} continued on next page\ldots}\nobreak{} } \newcommand{\exitProblemHeader}[1]{ \nobreak\extramarks{Problem \arabic{#1} (continued)}{Problem \arabic{#1} continued on next page\ldots}\nobreak{} \stepcounter{#1} \nobreak\extramarks{Problem \arabic{#1}}{}\nobreak{} } \setcounter{secnumdepth}{0} \newcounter{partCounter} \newcounter{homeworkProblemCounter} \setcounter{homeworkProblemCounter}{1} \nobreak\extramarks{Problem \arabic{homeworkProblemCounter}}{}\nobreak{} % % Homework Problem Environment % % This environment takes an optional argument. When given, it will adjust the % problem counter. This is useful for when the problems given for your % assignment aren't sequential. See the last 3 problems of this template for an % example. % \newenvironment{homeworkProblem}[1][-1]{ \ifnum#1>0 \setcounter{homeworkProblemCounter}{#1} \fi \section{Problem \arabic{homeworkProblemCounter}} \setcounter{partCounter}{1} \enterProblemHeader{homeworkProblemCounter} }{ \exitProblemHeader{homeworkProblemCounter} } % % Homework Details % - Title % - Due date % - Class % - Section/Time % - Instructor % - Author % \newcommand{\hmwkClass}{CS 219} \newcommand{\hmwkTitle}{Homework\ \#10} \newcommand{\hmwkDueDate}{November 23rd, 2016} \newcommand{\hmwkDueTime}{4:00pm} \newcommand{\hmwkClassInstructor}{Dr. Egbert} \newcommand{\hmwkAuthorName}{} % % Title Page % \title{ \vspace{2in} \textmd{\textbf{\hmwkClass:\ \hmwkTitle}}\\ \normalsize\vspace{0.1in}\small{Due\ on\ \hmwkDueDate\ at \hmwkDueTime}\\ \vspace{0.1in}\large{\textit{\hmwkClassInstructor}} \vspace{3in} } \author{\textbf{\hmwkAuthorName}} \date{} \renewcommand{\part}[1]{\textbf{\large Part \Alph{partCounter}}\stepcounter{partCounter}\\} % % Various Helper Commands % % Useful for algorithms \newcommand{\alg}[1]{\textsc{\bfseries \footnotesize #1}} % Alias for the Solution section header \newcommand{\solution}{\textbf{\large Solution:}} % Alias for answers \newcommand{\answer}{\textbf{\large Answer: }} \begin{document} \maketitle \pagebreak \begin{homeworkProblem} 1.) Write a program called SUB64 to subtract the 64-bit integer in memory locations 0x0150 and 0x154 from the 64-bit integer in 0x0160 and 0x0164. Store the result in memory location 0x0170 and 0x0174.\\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; SUB64 ; Template by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: mov eax,[0x0150] ; lower 32 bits of first num mov edi,[0x0160] ; lower 32 bits of second num mov edx,[0x0154] ; lower 32 bits of first num mov esi,[0x0164] ; lower 32 bits of second num sub %edi, %eax ; Subtract lowest 32-bits, borrow reflected in carry flag sbb %esi, %edx ; Subtract highest 32-bits, and the borrow if there was one mov [0x170],%esi ; Copy high bits to memory mov [0x174],%edi ; Copy low bits to memory ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here ; N/A ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 2.) Write a program called COMBINE that combines the low-order nibbles of the four bytes in memory locations 0x0150 to 0x0153 into a single 16-bit word. The nibbles should be ordered low-to-high in the result beginning with the data from location 0x0150. Store the result as 16-bits in memory location 0x0154. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; COMBINE ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BL,[0x0150l] ; Store value at 0x0150 in lower bits of reg BX MOV BH,[0x0153h] ; Store value at 0x0154 in higher bits of reg B MOV CP,[0x0154] ; Move offset address to base pointer MOV [CP],BX ; Write value of BX register to memory ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 3.) Write a program called FIND to find the larger of two signed bytes. Assume the two bytes are in memory locations 0x0150 and 0x0151. Store the larger of the two in memory location 0x0152.\\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; Find ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[0x150] MOV CL,0 ; Store larger of two the two bytes LP1: MOV AL,[0x150+1] CMP CL,AL ; Test if byte 1 > byte 2 JA LPC ; continue testing if not MOV CL,AL ; otherwise we've found the larger byte LPC: DEC BX JGE LP1 MOV [0x152],CL ; Store the largest byte ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 4.) Write a program called LSHIFT to shift logically the 32-bit contents of memory location 0x0150 left according to the 8-bit shift count stored in memory location 0x0154 and store the results at memory address 0x0158. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; LSHIFT ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX, [0x0152] ; Store value at 0x0152 in register BX MOV CX, 0x0154 ; Move the address for the result into CX SHL BX, [0x0158] ; Left shift by the offset stored at 0x0158 MOV [CX], BX ; Move the value in BX to memory ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 5.) Write a program called FIND8 to find the largest unsigned 8-bit word in a list. The list begins at address 0x0154. The length of the list is stored in an 8-bit variable at address 0x0150. Store the largest entry in memory location 0x0152. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; Find8 ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 6.) Write a program called FIND32 to find the largest unsigned 32-bit word in a list. The list begins at address 0x0160. The length of the list is stored in an 8-bit variable at address 0x0150. Store the largest entry in memory location 0x0154. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; Find32 ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 7.) Write a program called SCAN to scan a list of unsigned bytes and find the smallest and largest entries in the list. The length of the list is stored in a 16-bit variable at addresses 0x0152 and 0x0154. The list begins at address 0x0160. Store the smallest byte at address 0x0150 and the largest byte at address 0x0151. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; SCAN ; ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 8.) Write a program called COUNT to count the number of characters in a null-terminated ASCII string that are equal to a KEY. The KEY is stored in memory location 0x0150. The string is stored in memory beginning at address 0x0160. Store the 8-bit count in memory location 0x0154. (Assume the maximum count is 255.) \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; COUNT ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 9.) Write a program called ONES to determine the number of bits equal to one in a 32-bit variable. The 32-bit variable is in memory location 0x0154. Store the 8-bit counter in memory location 0x0150. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; ONES ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 10.) Write a subroutine called STRLEN that determines the length of a null-terminated ASCII string. Pass the 16-bit start address of the string to the subroutine in register BX. Return the length, excluding the null byte, in register CX. All registers (except CX) should return to the calling program unchanged. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; STRLEN ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 11.) Write a subroutine called REPLACE that processes a null-terminated string of decimal characters and replaces leading zeros with spaces. Pass the 32-bit address of the string to the subroutine in register BX. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; Replace ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE LP1: MOV AL,[BX+DLST] CMP CL,AL ;TEST IF BL > AL JA LPC ;CONTINUE IF NOT MOV CL,AL ;ELSE STORE NEW MAX LPC: DEC BX JGE LP1 MOV [MAXV],CL ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \pagebreak \begin{homeworkProblem} 12.) Write a program called UNPACK to convert the 16-bit BCD variable in memory locations 0x0150 and 0x0151 to four ASCII characters with the high-order digit first, beginning in memory location 0x0154. \\ \solution \begin{table}[h] \begin{tabular}{|c|} \hline \begin{lstlisting} ; UNPACK ; Templated by ver 1.1 11/14/2016 org 100h section .text ; beginning address of code = 0x0100 ;******************************************************* ; put your code here start: MOV BX,[LSTL] MOV CL,0 ;USE CL FOR LARGEST VALUE ; end of your code ;******************************************************* ILP: JMP ILP ;infinite loop ; TIMES 50H -($-$$) DB 0 section .data ; beginning address of data = 0x0150 ;******************************************************* ; put your data items here LSTL: DB 14 MAXV: DB 0 DLST: DB 254,5,25,250,100,150,30,200,253,15,23,46,73,175,0 ; end of your data ;******************************************************* \end{lstlisting} \\ \hline \end{tabular} \end{table} \end{homeworkProblem} \end{document} 0 \chapter{Numeri complessi} \label{cha:numericomplessibase} \section{Il piano complesso} \tcbstartrecording \begin{exercise} Disegnare nel piano complesso\index{Piano!complesso} il numero $z=2+3\uimm$ \tcblower Disegnare nel piano complesso il numero $z=2+3\uimm$ \begin{center} \includestandalone[width=.6\textwidth]{terzo/grafici/Piano_complesso_01} \captionof{figure}{Piano complesso}\label{fig:disegnopianocomplesso01} \end{center} \end{exercise} \begin{exercise} Disegnare nel piano complesso il numero $z=-4+2\uimm$ \tcblower Disegnare nel piano complesso il numero $z=-4+2\uimm$ \begin{center} \includestandalone[width=.6\textwidth]{terzo/grafici/Piano_complesso_02} \captionof{figure}{Piano complesso}\label{fig:disegnopianocomplesso02} \end{center} \end{exercise} \begin{exercise}[no solution] Disegnare nel piano complesso il numero $z=2-7\uimm$ \end{exercise} \begin{exercise}[no solution] Disegnare nel piano complesso il numero $z=1+2\uimm$ \end{exercise} \begin{exercise}[no solution] Disegnare nel piano complesso il numero $z=+2\uimm$ \end{exercise} \begin{exercise}[no solution] Disegnare nel piano complesso il numero $z=1$ \end{exercise} % \tcbstoprecording %% \newpage % \section{Soluzioni il piano complesso} % \tcbinputrecordstex/texmf-context/tex/context/base/mkii/enco-grk.mkii %D \module %D [ file=enco-grk, %D version=2003.03.01, %D title=\CONTEXT\ Encoding Macros, %D subtitle=Greek, %D author=, %D date=\currentdate, %D copyright={PRAGMA ADE \& \CONTEXT\ Development Team}] %C %C This module is part of the \CONTEXT\ macro||package and is %C therefore copyrighted by \PRAGMA. See mreadme.pdf for %C details. \startmapping[iso-8859-7] % Uppercase Greek letters \definecasemap 193 193 225 % greekAlpha \definecasemap 194 194 226 % greekBeta \definecasemap 195 195 227 % greekGamma \definecasemap 196 196 228 % greekDelta \definecasemap 197 197 229 % greekEpsilon \definecasemap 198 198 230 % greekZeta \definecasemap 199 199 213 % greekEta \definecasemap 200 200 232 % greekTheta \definecasemap 201 201 233 % greekIota \definecasemap 202 202 234 % greekKappa \definecasemap 203 203 235 % greekLambda \definecasemap 204 204 236 % greekMu \definecasemap 205 205 237 % greekNu \definecasemap 206 206 238 % greekXi \definecasemap 207 207 239 % greekOmicron \definecasemap 208 208 240 % greekPi \definecasemap 209 209 241 % greekRho \definecasemap 211 211 243 % greekSigma \definecasemap 212 212 244 % greekTau \definecasemap 213 213 245 % greekUpsilon \definecasemap 214 214 246 % greekPhi \definecasemap 215 215 247 % greekChi \definecasemap 216 216 248 % greekPsi \definecasemap 217 217 249 % greekOmega % Lowercase Greek letters \definecasemap 225 193 225 % greekalpha \definecasemap 226 194 226 % greekbeta \definecasemap 227 195 227 % greekgamma \definecasemap 228 196 228 % greekdelta \definecasemap 229 197 229 % greekepsilon \definecasemap 230 198 230 % greekzeta \definecasemap 213 199 213 % greeketa \definecasemap 232 200 232 % greektheta \definecasemap 233 201 233 % greekiota \definecasemap 234 202 234 % greekkappa \definecasemap 235 203 235 % greeklambda \definecasemap 236 204 236 % greekmu \definecasemap 237 205 237 % greeknu \definecasemap 238 206 238 % greekxi \definecasemap 239 207 239 % greekomicron \definecasemap 240 208 240 % greekpi \definecasemap 241 209 241 % greekrho \definecasemap 242 211 242 % greekfinalsigma \definecasemap 243 211 243 % greekSigma \definecasemap 244 212 244 % greekTau \definecasemap 245 213 245 % greekUpsilon \definecasemap 246 214 246 % greekPhi \definecasemap 247 215 247 % greekChi \definecasemap 248 216 248 % greekPsi \definecasemap 249 217 249 % greekOmega % Accented Uppercase Greek letters \definecasemap 182 193 220 % greekAlphatonos \definecasemap 184 197 221 % greekEpsilontonos \definecasemap 185 199 222 % greekEtatonos \definecasemap 186 201 223 % greekIotatonos \definecasemap 188 207 252 % greekOmicrontonos \definecasemap 190 213 253 % greekUpsilontonos \definecasemap 191 217 254 % greekOmegatonos \definecasemap 218 218 250 % greekIotadialytika \definecasemap 219 219 251 % greekUpsilondialytika % Accented Lowercase Greek letters \definecasemap 220 193 220 % greekalphatonos \definecasemap 221 197 221 % greekepsilontonos \definecasemap 222 199 222 % greeketatonos \definecasemap 223 201 223 % greekiotatonos \definecasemap 252 207 252 % greekomicrontonos \definecasemap 253 213 253 % greekupsilontonos \definecasemap 254 217 254 % greekomegatonos \definecasemap 250 218 250 % greekiotadialytika \definecasemap 251 219 251 % greekupsilondialytika \definecasemap 192 218 192 % greekiotadialytikatonos \definecasemap 224 219 224 % greekupsilondialytikatonos \stopmapping \startencoding[iso-8859-7] % Uppercase Greek letters \definecharacter greekAlpha 193 \definecharacter greekBeta 194 \definecharacter greekGamma 195 \definecharacter greekDelta 196 \definecharacter greekEpsilon 197 \definecharacter greekZeta 198 \definecharacter greekEta 199 \definecharacter greekTheta 200 \definecharacter greekIota 201 \definecharacter greekKappa 202 \definecharacter greekLambda 203 \definecharacter greekMu 204 \definecharacter greekNu 205 \definecharacter greekXi 206 \definecharacter greekOmicron 207 \definecharacter greekPi 208 \definecharacter greekRho 209 \definecharacter greekSigma 211 \definecharacter greekTau 212 \definecharacter greekUpsilon 213 \definecharacter greekPhi 214 \definecharacter greekChi 215 \definecharacter greekPsi 216 \definecharacter greekOmega 217 % Lowercase Greek letters \definecharacter greekalpha 225 \definecharacter greekbeta 226 \definecharacter greekgamma 227 \definecharacter greekdelta 228 \definecharacter greekepsilon 229 \definecharacter greekzeta 230 \definecharacter greeketa 231 \definecharacter greektheta 232 \definecharacter greekiota 233 \definecharacter greekkappa 234 \definecharacter greeklambda 235 \definecharacter greekmu 236 \definecharacter greeknu 237 \definecharacter greekxi 238 \definecharacter greekomicron 239 \definecharacter greekpi 240 \definecharacter greekrho 241 \definecharacter greekfinalsigma 242 \definecharacter greeksigma 243 \definecharacter greektau 244 \definecharacter greekupsilon 245 \definecharacter greekphi 246 \definecharacter greekchi 247 \definecharacter greekpsi 248 \definecharacter greekomega 249 % Accented Uppercase Greek letters \definecharacter greekAlphatonos 182 \definecharacter greekEpsilontonos 184 \definecharacter greekEtatonos 185 \definecharacter greekIotatonos 186 \definecharacter greekOmicrontonos 188 \definecharacter greekUpsilontonos 190 \definecharacter greekOmegatonos 191 \definecharacter greekIotadialytika 218 \definecharacter greekUpsilondialytika 219 % Accented Lowercase Greek letters \definecharacter greekalphatonos 220 \definecharacter greekepsilontonos 221 \definecharacter greeketatonos 222 \definecharacter greekiotatonos 223 \definecharacter greekomicrontonos 252 \definecharacter greekupsilontonos 253 \definecharacter greekomegatonos 254 \definecharacter greekiotadialytika 250 \definecharacter greekupsilondialytika 251 \definecharacter greekiotadialytikatonos 192 \definecharacter greekupsilondialytikatonos 224 % Miscellaneous Greek symbols \definecharacter greekleftquot 171 \definecharacter greekrightquot 187 \definecharacter greektonos 180 \definecharacter greekdialytikatonos 181 \definecharacter greekapostrophos 162 \stopencoding 10-100 \documentclass[aspectratio = 169]{beamer} \usepackage[utf8]{inputenc} % Character encoding. \pdfinfo{ /Author () /Title (The Simplex Algorithm II) /Subject (Linear Programs) } \usepackage{./Style/linearProgramsBeamer} % This is extra styling for Beamer environments. \usepackage{./Style/linearProgramsStyle} % This is a set of commands for maths content. %------------------------------------------------------------------------------- % TITLE PAGE %------------------------------------------------------------------------------- \author[BD]{} \institute[]{EPITA} \title{Linear Programs} % \subtitle{The Simplex Algorithm II} %------------------------------------------------------------------------------- % DOCUMENT BODY %------------------------------------------------------------------------------- \begin{document} \begin{frame}[plain] \titlepage % Print the title page as the first slide \end{frame} \begin{frame}{Où on en est, à quoi on fait face.} La procédure adoptée à ce stade s'est toujours terminée. Notre conjecture de base est qu'elle nous renvoie un point optimal du programme linéaire de départ. Pour l'instant on ne peut pas en être certains. Par bien des égards ce qu'on a fait jusqu'à présent est loin d'être satisfaisant. Étant donné un programme linéaire $L$, voici une liste de points qu'ils nous restent à éclaircir: \begin{itemize} \item[\textcolor<6>{lightgray}{\textbullet}]<1-> \textcolor<6>{lightgray}{Comment savoir si $L$ est admissible?} \item[\textcolor<6>{lightgray}{\textbullet}]<2-> \textcolor<6>{lightgray}{Que faire lorsque la solution de base n'est pas admissible?} \item[\textbullet]<3-> Comment vérifier que $L$ n'est pas majoré? \item[\textbullet]<4-> Est-ce que la procédure qu'on a pu tester jusqu'à présent se termine en général? \item[\textcolor<6>{lightgray}{\textbullet}]<5-> \textcolor<6>{lightgray}{Si la procédure se termine, est-ce qu'on en déduit toujours une valeur optimale?} \end{itemize} \end{frame} \section{Pivoter} \begin{frame}{Pivoter | Fixer la notation} Dans le but de répondre aux questions qu'ils nous restent à aborder il va nous falloir écrire au propre les algorithmes qu'on a testés à la main. On considère le programme linéaire $L$ \begin{figure} \alt<4>{ \textcolor{orange}{ \begin{linearProgG}{ ${\displaystyle z - \sum_{j=1}^n c_jx_j} = \nu$ }{ ${\displaystyle \forall i \in \{1, \ldots, m\}, \quad \sum_{j=1}^n a_{ij}x_j + x_{i+m} = b_i}$ }{ $\forall j \in N\cup B, \quad x_j \geq 0$ } \end{linearProgG} } }{ \alt<3>{ \begin{linearProgG}{ ${\displaystyle z = \nu + \sum_{j=1}^n c_jx_j}$ }{ ${\displaystyle \forall i \in \{1, \ldots m \}, \quad x_{i+m} = b_i - \sum_{j=1}^n a_{ij}x_j}$ }{ $\forall j \in N\cup B, \quad x_j \geq 0.$ } \end{linearProgG} }{ \begin{linearProgG}{ ${\displaystyle z = \nu + \sum_{j=1}^n c_jx_j}$ }{ ${\displaystyle \forall i \in \{1, \ldots, m\}, \quad \sum_{j=1}^n a_{ij}x_j \leq b_i}$ }{ $\forall j \in N, \quad x_j \geq 0$ } \end{linearProgG} } } \end{figure} \begin{overlayarea}{\textwidth}{.3\textheight} \begin{onlyenv}<1> On écrit \begin{itemize} \item[\textbullet] $\bs{A}$ pour la matrice de taille $(m, n)$ des coefficients $\big(a_{ij}\big)_{\substack{1 \leq i \leq m \\ 1 \leq j \leq n}}$ \item[\textbullet] $\bs{b}$ pour le $m$-tuple $(b_1, \ldots, b_m)$ \item[\textbullet] $\bs{c}$ pour le $n$-tuple $(c_1, \ldots, c_n)$. \end{itemize} Noter que $\nu$ est la valeur objective de la solution de base. \end{onlyenv} \begin{onlyenv}<2> Le programme linéaire $L$ dans cette forme est décrit par la donnée $(\bs{A}, \bs{b}, \bs{c}, \nu)$. Le programme initial a en général $\nu = 0$. \end{onlyenv} \begin{onlyenv}<3-> Pour encoder la forme slack on inclus également les données $N$, $B$ des indices de variables hors-base et de base. Une forme slack est donc donnée par $(N, B, \bs{A}, \bs{b}, \bs{c}, \nu)$. L'ensemble $N$ est initialisé à $\{1, \ldots, n\}$ et $B$ à $\{n+1, \ldots, n+m\}$. \pause[4] \textcolor{orange}{Pour se rapprocher de l'écriture d'un système linéaire on modifie légèrement l'écriture de la forme slack.} \end{onlyenv} \end{overlayarea} \end{frame} \begin{frame} \frametitle{Pivoter | Le tableau d'un programme linéaire} Les contraintes linéaires de $L$ dans la forme slack représente une matrice $(m, n+m)$ notée $\underline{\bs{A}}$ et obtenue comme concaténation de $\bs{A}$ et $\bs{I}_m$ le long des lignes de $\bs{A}$. Le \emph{tableau} $T$ de $L$ est obtenu en \begin{itemize} \item<2-> concaténant $\underline{\bs{A}}$ et $b^T$ le long des lignes de $A$ \item<3-> concaténant les résultats avec le vecteur $(-c_1, \ldots, -c_n, 0, \ldots 0, \nu)$ de taille $n+m+1$ le long des colonnes de $\bs{A}$. \end{itemize} \pause[4] \begin{figure} \begin{tabular}{c|ccc|cccc|c|} & \alert{$\bs{x_1}$} & \alert{$\cdots$} & \alert{$\bs{x_n}$} & \alert{$\bs{x_{n+1}}$} & \alert{$\bs{x_{n+2}}$} & \alert{$\cdots$} & \alert{$\bs{x_{n+m}}$} & \\ \hline & $-c_1$ & $\cdots$ & $-c_n$ & $0$ & $0$ & $\cdots$ & $0$ & $\nu$ \\ \hline \alert{$\bs{n+1}$} & $a_{11}$ & $\cdots$ & $a_{1n}$ & $1$ & $0$ & $\cdots$ & $0$ & $b_1$ \\ \alert{$\bs{n+2}$}& $a_{21}$ & $\cdots$ & $a_{2n}$ & $0$ & $1$ & $\cdots$ & $0$ & $b_2$ \\ \alert{$\vdots$}& $\vdots$ & $\ddots$ & $\vdots$ & $\vdots$ & $\vdots$ & $\ddots$ & $\vdots$ & $\vdots$ \\ \alert{$\bs{n+m}$} & $a_{m1}$ & $\cdots$ & $a_{mn}$ & $0$ & $\cdots$ & $\cdots$ & $1$ & $b_m$ \end{tabular} \end{figure} Le tableau $T$ est de taille $(m + 1, n + m + 1)$. \end{frame} \begin{frame}[fragile]{Pivoter} We are given input $(N, B, \bs{A}, \bs{b}, \bs{c}, \nu)$ and two indexes $e \in N$, $l\in B$ respectively corresponding to entering and leaving variables to and from the set of \textit{basic} variables $B$. \begin{columns} \begin{column}{.5\textwidth} \begin{itemize} \item[\textbullet]<2-> Express $x_e$ in terms of other variables in equation $l$ \item[\textbullet]<3-> Replace $x_e$ by previously obtained expression in linear constraints \item[\textbullet]<4-> Replace $x_e$ by corresponding expression in the value function \item[\textbullet]<5-> Update basic and none basic sets of variables. \end{itemize} \end{column} \begin{column}{.5\textwidth} \begin{overlayarea}{.96\textwidth}{.45\textheight} \begin{onlyenv}<2-> \begin{tcolorbox}[ enhanced, parbox = false, colback = mLightBrown!10!white, colframe = mLightBrown, arc = 0mm, ] \small{ \mint{python}{T[l, :] = (1/T[l, e])*T[l, :]} } \end{tcolorbox} \end{onlyenv} \begin{onlyenv}<3-> \begin{tcolorbox}[ enhanced, parbox = false, colback = mLightBrown!10!white, colframe = mLightBrown, arc = 0mm, ] \small{ \begin{minted}[autogobble]{python} if i != l: T[i, :] -= T[i, e]*T[l, :] \end{minted} } \end{tcolorbox} \end{onlyenv} \begin{onlyenv}<5> \begin{tcolorbox}[ enhanced, parbox = false, colback = mLightBrown!10!white, colframe = mLightBrown, arc = 0mm, ] \small{ \begin{minted}[autogobble]{python} N.insert(N.index(e), l).remove(e) B.insert(B.index(l), e).remove(l) \end{minted} } \end{tcolorbox} \end{onlyenv} \end{overlayarea} \end{column} \end{columns} \end{frame} \begin{frame}[fragile]{Pivoter | Full Function} \small{ \begin{minted}[linenos]{python} def pivot(N, B, T, e, l): """Pivoting in linear programs. Pivots entering and leaving variables in linear program given as tableau. Done in place. """ T[l, :] = (1/T[l, e])*T[l, :] for i in range(m+1): if i != l: # ugly T[i, :] -= T[i, e]*T[i, :] N.insert(N.index(e), l).remove(e) B.remove(B.index(l), e).remove(l) \end{minted} } \end{frame} \begin{frame}{Pivoter | Full Function} \begin{itemize} \item<1-> Index $e$ is an element of $N$ and $l$ one of $B$. \item<2-> There better be no entries $e$, $l$ such that \mintinline{python}{T[l, e] == 0}. We shall ensure this is never the case when \mintinline{python}{pivot} is used. \item<3-> The objective value and basic solution of obtained linear program can be read on last column to the right. \end{itemize} \end{frame} \section{Détecter le fait de ne pas être borné} \begin{frame}{Tester la non \emph{bornitude}} On note $L$ le programme linéaire en forme standard \begin{figure} \begin{linearProgG}{ ${\displaystyle z = \nu + \sum_{j=1}^n c_jx_j}$ }{ ${\displaystyle \forall i \in \{1, \ldots, m\}, \quad \sum_{j=1}^n a_{ij}x_j \leq b_i}$ }{ $\forall j \in N, \quad x_j \geq 0$ } \end{linearProgG} \end{figure} \begin{halfshyblock}{Fait} \begin{onlyenv}<1 >S'il y a un indice $j$ tel que $c_j > 0$ et tous les coefficients de $x_j$ dans les contraintes linéaires sont négatifs alors $L$ est non majoré. \end{onlyenv} \begin{onlyenv}<2- >\!\alert{De manière équivalente, s'il y a une colonne dans le tableau de $L$ ayant une première entrée négative non nulle et toutes les autres négatives alors $L$ est non majoré.} \end{onlyenv} \end{halfshyblock} \end{frame} \begin{frame}{Tester la non \emph{bornitude}} En utilisant le fait précédent, à chaque appel de \mintinline{python}{pivot}, on teste si on ne satisfait pas la propriété: \begin{quote} Il existe un indice $j$ tel que $c_j > 0$ et tous les autres coefficients de $x_j$ sont négatifs ou nuls. \end{quote} \pause C'est une condition nécessaire, pour l'instant on a aucune garantie qu'on tombe dans une telle situation quand $L$ n'est pas majoré. Le fait que ce soit le cas vient des résultats de dualité. \pause Pour l'instant on va devoir accepter que ce test naïf sera suffisant\footnote{Il l'est mais on a encore aucune certitude.}. \end{frame} \section{The Simplex Algorithm : Second Try} \begin{frame}[fragile]{The Simplex Algorithm \emph{Restricted}} \begin{overlayarea}{1.1\textwidth}{.75\textheight} \setlength{\columnsep}{-10pt} \begin{multicols}{2} \scriptsize{ \begin{minted}[autogobble, linenos, breaklines]{python} # Under construction function! def _simplex(N, B, T): """Restricted simplex algorithm Runs simplex algorithm on basic feasible linear program in slack form. Args: N, B (list[int]): lists of non-basic and basic variables. T (ndarray[float]): numpy array for tableau of linear program. Output: (ndarray[float]) vector tail of which is maximal objective value, rest is optimal point. """ m, margins = len(B), dict() aug_var = [i for i in N if T[0, i] < 0] while aug_var: e = random.choice(augmenting_var) for i in range(m): if T[i, e] > 0: margins[B[i]] = T[i, -1]/T[i, e] if not margins: raise Exception("Unbounded LP") min_margin = min(margins.values()) minima = [i for i in margins\ if margins[i] == min_margin] l = random.choice(minima) pivot(N, B, T, e, l) aug_var = [i for i in N if T[0, i] < 0] return T[:, -1] \end{minted} } \end{multicols} \end{overlayarea} \end{frame} \begin{frame}{Boucler} À chaque fois qu'on entre dans la boucle \mintinline{python}{while} de \mintinline{python}{_simplex} ou bien on \textit{augmente la valeur objective} ou alors on découvre que le PL est non-majoré. A priori, \mintinline{python}{_simplex} pourrait s'exécuter indéfiniment, en itérant entre des formes slack équivalentes ayant une valeur objective constante. On va chercher à étudier ce phénomène. \pause \begin{prop}[\textbf{C}] Soit $L$ un PL $(N, B, \bs{A}, \bs{b}, \bs{c}, \nu)$ où $\bs{A}$ est une matrice $(m, n)$. Si la boucle de \mintinline{python}{_simplex} s'exécute plus de $\binom{n+m}{m}$ fois, alors l'algorithme boucle, i.e. la boucle s'exécute indéfiniment en alternant entre un nombre fini de formes slack ayant la même valeur objective. \end{prop} \begin{rem} Cela signifie que dès que \mintinline{python}{_simplex} entre dans sa boucle principale plus $\binom{n+m}{m}$ fois alors on peut retourner la valeur objective et la solution de base courante. \end{rem} \end{frame} \begin{frame}{Boucler} La preuve de la proposition ci-dessus se base sur deux faits: \begin{itemize} \item si on tombe sur une forme slack déjà vue au cours de la résolution des précédentes itérations, alors on retrouvera par la suite les mêmes itérations qui s'ensuivaient, on a un comportement périodique ; \item il n'y a qu'un nombre fini de formes slacks possibles pour un même programme linéaire. \end{itemize} \pause Seul le second point nécessite d'être clarifié. On se contente de le montrer dans le cas de PL ayant un lieu admissible d'intérieur non vide. \end{frame} \begin{frame}{Boucler} \begin{lem}[$\bs{B}$] Soit $L$ un PL donné par $(\bs{A}, \bs{b}, \bs{c})$. Une forme slack de $L$ qui apparaît dans \mintinline{python}{_simplex} est déterminée par le choix d'un sous-ensemble $B$ de variables de base. \end{lem} \pause \setlength\columnseprule{.1pt} \begin{multicols}{2} \begin{demo} Deux formes slacks $(N, B, \bs{A}, \bs{b}, \bs{c}, \nu)$ et $(N, B, \bs{A'}, \bs{b'}, \bs{c'}, \nu')$ sont équivalentes. \pause \vspace{\baselineskip} En étudiants la différence des contraintes linéaires, on obtient les relations: \[ 0 = (\nu - \nu') + \sum_{j \in N} (c_j - c_j')x_j \] et pour chaque $i \in B$ \[ 0 = (b_i - b_i') - \sum_{j \in N} (a_{ij} - a_{ij}')x_j. \] \pause Ces relations étant vraie pour tout vecteur $(x_1, \ldots, x_n)$ dans le lieu admissible de LP, si le lieu admissible de LP est d'intérieur non vide, il est clair que pour tout $i \in B$ et $j \in N$ on a $\nu = \nu'$, $b_i = b_i'$, $c_j = c_j'$, $a_{ij} = a_{ij}'$. \end{demo} \end{multicols} \end{frame} \begin{frame}{Boucler} Rajouter un compteur de boucle à l'algo du \mintinline{python}{_simplex} garanti la terminaison de \mintinline{python}{_simplex}. \pause \begin{rem} Cette solution n'est pas intelligente. Il y a plusieurs solutions à ce problème en pratique. Dans notre cas on implémente la règle de \emph{Bland}: à chaque choix d'indice aux lignes $21$ et $30$ de \mintinline{python}{_simplex}, on choisit le plus petit possible. \end{rem} \end{frame} \begin{frame}[fragile]{The Simplex Algorithm \emph{Restricted} | No Cycling Version} \begin{overlayarea}{1.1\textwidth}{.75\textheight} \setlength{\columnsep}{-10pt} \begin{multicols}{2} \scriptsize{ \begin{minted}[autogobble, linenos, breaklines]{python} def _simplex(N, B, T): """Restricted simplex algorithm Runs simplex algorithm on basic feasible linear program in slack form. Args: N, B (list[int]): lists of non-basic and basic variables. T (ndarray[float]): numpy array for tableau of linear program. Output: (ndarray[float]) vector tail of which is maximal objective value, rest is optimal point. """ m = len(B) l, margin = None, float('inf') aug_var = [i for i in N if T[0, i] < 0] while aug_var: e = min(augmenting_var) for i in range(m): if T[i, e] > 0: if T[i, -1]/T[i, e] < margin: margin = T[i, -1]/T[i,e] l = i if not l: raise Exception("Unbounded LP") pivot(N, B, T, e, l) aug_var = [i for i in N if T[0, i] < 0] return T[:, -1] \end{minted} } \end{multicols} \end{overlayarea} \end{frame} \begin{frame} \centering {\huge \textbf{C'est tout pour aujourd'hui!}} \end{frame} \end{document} %%% Local Variables: %%% mode: latex %%% TeX-master: t %%% End: rachelslaybaugh/munk-disseration \section{Description of the Characterization Problems} \label{sec:AngleProbDesc} In characterizing the $\Omega$-methods, we aim to determine in which problems they perform well, and then quantify that success. First, we must determine how effective the $\Omega$-methods are in reducing the variance for a tally result in Monte Carlo. This is done by assessing and comparing the FOMs between different VR methods. Also, the method must be investigated using a diverse set of anisotropic problems. By constructing problems that have different mechanisms causing or inducing anisotropy in the flux, potential strengths or weaknesses of the method can be isolated as a function of these mechanisms. In addition to comparing the FOMs or REs between methods, another desirable metric by which to measure the method's success given the degree of anisotropy in the problem. Recall that different means of quantifying the flux anisotropy are described in Section \ref{sec:anisotropy_quant}. With a diverse selection of characterization problems, we obtain variation in the flux anisotropy in each problem as well as the resultant FOMs. This provides us with a path forward with which to use the $\Omega$-methods in a deeper angular-sensitivity study. \subsection{Identification of Anisotropy-Inducing Physics} \label{subsec:AngleProbID} There exists a rich history of using hybrid methods in problems with strong angular dependence, as summarized in Chapter \ref{ch:lit_review}. Angular dependence may appear in a problem through several means--both physical and computational. Mosher et al.\ \cite{mosher_automated_2009} noted in their threat-detection work with ADVANTG that problems with strongly directional sources and problems with ``thin'' materials like air were difficult for ADVANTG to effectively reduce the variance. They attributed this to strongly anisotropic behavior of the importance function that were not reflected well by the scalar flux. Sweezy~\cite{sweezy_automated_2005} also found that weight windows obtained from a hybrid S$_N$ calculation were not good for a dogleg void problem, where ray effects from the S$_N$ calculation generated poorer weight windows than a method without ray effects \footnote{Recall from Sections \ref{sec:ContributonImportance} and \ref{sec:litsummary} that ray effects are a nonphysical effect seen in the flux solution that arise from the angular discretization of the problem. Ray effects are common in situations where there are strong streaming effects or if a strong source is emitting particles with long mean free paths in the material.}. Though they did not observe ray effects in the importance map for the problem, Peplow et al.~\cite{peplow_consistent_2012} also found that CADIS struggled with thin material streaming in a spherical boat test problem. The examples of angle-dependence in problems affecting hybrid methods' success illustrate that the flux can have anisotropy resulting from more than one mechanism. Based on these examples, we have identified several separate processes that affect the flux anisotropy. These processes can be grouped into three categories: \begin{itemize} \itemsep0em \item anisotropy in the flux resulting from strongly directional sources, \item anisotropy resulting from strong differences between material properties (this can be due to differences in materials spatially or due to changes in interaction probabilities as a function of energy), \item anisotropy in the flux from algorithmic limitations (ray effects). \end{itemize} These processes overlap. Consequently, this section continues with a brief discussion about how each mechanism applies to anisotropic problems. A strongly directional source is one that emits particles in a very small solid angle of angle-space. The most extreme example of this would be a monodirectional source, while an extreme opposite would be an isotropic source. This particular anisotropy-creating process is source-specific and does not depend on the rest of the problem configuration. Our characterization problems will have sources of both types to ensure the full parameter space is covered. % This next paragraph is maybe a little confusing and could use rewording. The next subset of anisotropy-inducing processes are those that result form strong differences between material properties. As noted, this can be from the geometric configuration of the problem, or from variations in the cross sections within a geometric location. To illustrate the differences in the way the problem can physically induce anisotropy in the flux, several simple thought experiments will be presented. Consider first the extreme example of material A which has some low absorption probability, and material B which is a pure absorber. Only particles that travel through material A will eventually reach the tally location. This is an example of a type of problem with strong material heterogeneity. In constructing a set of characterization problems, creating channels through which particles will preferentially travel will induce anisotropy in the flux. These types of flow paths are also of interest in shielding application problems, and were discussed at length in Section \ref{sec:ContributonImportance}. In this type of problem, material A can either have a low scattering probability (airlike), or it can be highly scattering. In scattering events, neutral particles can either lose very little energy with a high Z material, like lead, or they can lose a lot of energy with a low Z material. These are considered separately, because the energy spectrum of the particles affects the particle's interaction probability. Consider another example of an isotropic point source immersed in a pure thin material. Because particles have a very low probability of interaction in the material, they will travel almost uniformly outwards away from the point source. At some distance from the point source, the majority of the particles in a cell will be traveling in the same direction. This is an example of a problem with streaming paths. To summarize, we have identified several sub-distinctions of this type of effect: regions with streaming where particles far from the source are primarily monodirectional, regions that are highly scattering where particles have a preferential flowpath through one material and are downscattered in energy, and regions with strong material heterogeneity where particles have preferential flowpaths but are not necessarily downscattered in energy. It should be noted that while streaming and scattering problems will almost always be subsets of problems with material homogeneity, it is possible to have a highly scattering or a streaming problem without material heterogeneity. The last factor that can influence anisotropy in the flux solution is ray effects. While ray effects are a result of anisotropy in the flux solution, this is a nonphysical effect and can actually affect variance reduction performance. In the case of ray effects, we aim to see if the $\Omega$-methods are more robust in avoiding them in generating VR parameters. Because ray effects are primarily seen in large regions with low interaction probabilities, some of the characterization problems must incorporate these types of regions into their geometries. In this subsection, four primary physical mechanisms by which the flux may be anisotropic were identified. These are: streaming paths, problems with high scattering effects, problems with high material heterogeneity (specifically with materials with strong differences in scattering and absorption probabilities), and problems with monodirectional sources. As described in the preceding paragraphs, a few of these mechanisms may overlap with one another. Together, they compose an assortment of anisotropy-inducing physics. Combined with different geometric arrangements a diverse group of anisotropic problems can be formulated. \subsection{Problem Specifications} \label{subsec:ProbSpecs} With the anisotropy-inducing physics described in Section \ref{subsec:AngleProbID}, a set of characterization problems that have different combinations of each of these effects can be conceptualized. These problems provide an overview of how the $\Omega$-methods perform in an assortment of anisotropic problems. As previously described, these fall into two broad categories: anisotropy caused by the problem materials and geometry, and anisotropy caused by the source definition. In the next several paragraphs, the material and geometric configuration of each problem will be described. This will be supplemented with an explanation of which anisotropy-inducing physics are contained in each problem. A summary of which physics are in each problem is provided in Table \ref{tab:probphysics}. \subsubsection*{Labyrinths} The labyrinth problems have isotropic point sources on the left hand side of the problem emitting a Watt spectrum of neutrons approximating the energy spectrum emitted by that of $^{235}$U fission. On the right hand side of the problem there is a NaI detector recording the flux. They are composed of a concrete maze with an air channel through the maze, and then open air channels at either end of the channels. The first variant of the labyrinth has a single turn, as illustrated in Figure \ref{fig:maze2geom}, and the second labyrinth has multiple turns, as illustrated in Figure \ref{fig:maze1geom}. These problems are both likely to have ray effects in the air region near the forward source. However, because far more scattering events will be required for a particle to exit the channel in the multi-turn maze, ray effects will likely be less prominent in the air region near the detector of that variant problem than in the single turn maze. Both problems have strong differences in interaction probabilities between the air and the concrete, thus they will have material heterogeneity. Further, because the concrete is composed of several lighter-mass elements, these will also be highly scattering. \begin{figure}[h!] \centering \includegraphics[width=15cm]{./chapters/characterization_probs/figures/geometries/maze2geom.png} \caption[Single turn labyrinth geometry.]{Single turn labyrinth geometry.} \label{fig:maze2geom} \end{figure} \begin{figure}[h!] \centering \includegraphics[width=15cm]{./chapters/characterization_probs/figures/geometries/maze1geom.png} \caption[Multi-turn labyrinth geometry.]{Multi-turn labyrinth geometry.} \label{fig:maze1geom} \end{figure} \subsubsection*{Steel beam in Concrete} \begin{figure}[h!] \centering \includegraphics[width=15cm]{./chapters/characterization_probs/figures/geometries/prob1geom.png} \caption[Steel plate embedded in concrete.]{Steel plate embedded in concrete.} \label{fig:prob1geom} \end{figure} Figure \ref{fig:prob1geom} is a variant problem with a steel beam embedded in concrete. A NaI detector is located on the right hand side of the problem to record the response in CADIS problems. The source is a 80x80 centimeter sheet pointed in towards the steel structure in the $+x$ direction emitting 10 MeV neutrons. Because the particles have preferential flow through the steel but do do not have long streaming paths, this problem has material heterogeneity and will be highly scattering, but will not have streaming paths in the shielding region. Further, because the source is emitted from a thin plate in $+x$, it is monodirectional. This problem may have some ray effects occurring from backscattering off of the steel and concrete in the left side air region. It may also have ray effects exiting the beam on the right hand side. However, because significantly more scattering will happen in the concrete, the ray effects on the right hand side will be less pronounced than in the air exits of the labyrinths. \subsubsection*{U-shaped corridor} \begin{figure}[h!] \centering \includegraphics[width=12cm]{./chapters/characterization_probs/figures/geometries/prob2geom.png} \caption[U-shaped corridor in concrete]{U-shaped corridor in concrete.} \label{fig:prob2geom} \end{figure} The U-shaped corridor illustrated in Figure \ref{fig:prob2geom} is somewhat similar to the maze variants from Figs. \ref{fig:maze2geom} and \ref{fig:maze1geom}. On the left-hand side of the corridor there is a point source emitting a Watt spectrum of $^{235}$U neutrons. The right leg of the corridor has a NaI detector. Without the large air voids in the labyrinth variants, the U-shaped corridor will have less prominent ray effects. The heterogeneity between the air and concrete will preferentially transport particles through the air, and particles interacting with the concrete will downscatter in energy. \subsubsection*{Concrete shielding with rebar} \begin{figure}[htb!] \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=15cm]{./chapters/characterization_probs/figures/geometries/prob4geomy100.png} \caption[Slice at $y=100$ centimeters]{Slice at $y=100$ centimeters} \label{fig:prob4yslice} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=8cm]{./chapters/characterization_probs/figures/geometries/prob4geomz105.png} \caption[Slice at $z=105$ centimeters]{Slice at $z=105$ centimeters} \label{fig:prob4zslice} \end{subfigure} \caption[Concrete shielding with rebar]{Concrete shielding with rebar.} \label{fig:prob4geom} \end{figure} The shielding material illustrated in Figure \ref{fig:prob4geom} is built off of the steel structural beam problem in Figure \ref{fig:prob1geom}. However, this is a more realistic illustration of rebar in concrete. In this problem, a NaI detector is used to measure the response on the right hand side of the problem in yellow. The source is both space- and energy-dependent, emitting a Watt spectrum of neutrons characteristic of $^{235}$U fission, and is distributed in a 100x160 centimeter plate on the left hand side of the problem. The source is monodirectional in $+x$. The two images provided show different xy-plane cutaways of the shielding, with steel rebar running through the concrete in different directions. This problem will have angular dependence, but preferential flowpaths through the concrete are not directed towards the detector location on the other side of the shielding in some of the rebar. This problem has material heterogeneity both in the concrete and between the concrete and air. This problem is highly scattering from the concrete, and is unlikely to have ray effects without a strong single preferential flowpath through the shield. \subsubsection*{Nuclear medicine therapy room} \begin{figure}[h!] \centering \includegraphics[width=15cm]{./chapters/characterization_probs/figures/geometries/therapygeom0000.png} \caption[Nuclear medicine therapy room.]{Therapy room geometry.} \label{fig:therapygeom} \end{figure} A small application problem relevant to the interests of this project is the therapy room illustrated in Figure \ref{fig:therapygeom}. This room has concrete walls, a water-based phantom that is being irradiated by a monodirectional source in the room, and a hallway where a therapy technician might walk. In a CADIS run of this problem, we seek to calculate the response in the technician in the hallway from particles that are not absorbed by the patient in the room. Because this problem is primarily air with concrete borders, it will have strong streaming effects in the air. Particles that do make it to the technician will be produced by emission from the patient in the room, by scattering off air or by scattering off walls. Because of the high fraction of air in this problem, we also anticipate ray effects to occur. While there will be scattering in this problem, it will not be as strong of an effect as other characterization problems. Now that the broad subset of characterization problems have been described, the physics that each contains is summarized in Table \ref{tab:probphysics}. The table illustrates that it is difficult to separate one cause of flux anisotropy from another in a characterization problem. This is especially true in generating a problem that has ray effects without streaming paths, and in constructing a highly scattering problem that has preferential flow paths but does not have material heterogeneity. This is a deficiency of the characterization problem construction, and is certainly an area that may be improved upon in future work. \begin{table}[h!] \centering \input{./chapters/characterization_probs/figures/problem_physics} \caption[Anisotropy-inducing physics of each of the characterization problems.] {Anisotropy-inducing physics of each of the characterization problems. Each identified anisotropy-inducing physical metric is used in different combinations for the characterization problems. This will help to aid in extrapolating to which real problems the $\Omega$-methods may be applied.} \label{tab:probphysics} \end{table} % maybe it would be worthwhile to add a table of source definitions between each % of the char problems. Some of them are monoenergetic, some are % monodirectional, and some are monospatial(?) (point sources). A table would be % a good way to summarize this information. \subsection{Introduction to Data Visualization and Analysis} \label{subsec:resultsintro} At this point several characterization problems have been identified for their properties in inducing anisotropy in the particle flux. Prior to going through the results for each of the characterization problems, this section shows how the data for each problem is presented and walks through the reasoning behind this approach to the analysis. This starts with example tables and figures of the FOM and tally results. Then, plots explaining the anisotropy metrics follow. This is accompanied by a discussion about how the anisotropy metrics can be related to the FOM and the relative error. \subsubsection*{Figure of Merit and Timing Tables} In Section \ref{sec:FOMvariants} several equation variants of the FOM were presented as quantifications of method success. The FOMs for each characterization problem are presented in tabular form, similar to Table \ref{tab:fom_defaults}. As discussed in that section, the FOM is dependent on the relative error and the time to obtain that relative error. For the hybrid cases, six different FOMs will be presented: three FOMs based on the tally average relative error, the tally maximum relative error, and the tally minimum relative error, and two FOMs based on the Monte Carlo runtime and the hybrid runtime. The unbiased analog Monte Carlo does not have a deterministic runtime, so only the three FOM variants based on the relative error are presented for those runs. When analyzing the results in the FOM table for each characterization problem, consider that the tally average relative error is calculated from all particles contributing to all tally bins in the problem. Thus the FOM reported for the tally average relative error may be outside of the bounds of the tally minimum or the tally maximum relative error. Table \ref{tab:fom_defaults} summarizes which equations were used to calculate each FOM; each equation number is noted in brackets. \begin{table}[h!] \centering \input{./chapters/characterization_probs/figures/sample_data/fom_defaults} \caption[Table of FOM variants used to measure $\Omega$-method performance.]{ Table of FOM variants used to measure $\Omega$ performance. Relevant equations can be found in Section \ref{sec:FOMvariants} and are referenced in the table in parentheses.} \label{tab:fom_defaults} \end{table} Tables calculating the FOMs summarized in Table \ref{tab:fom_defaults} may not have evaluated FOMS in some locations. These will be noted with a dashed line, or ``--''. These values will generally be in the minimum relative error section of the FOM tables, and they represent a zero relative error. This does not mean that infinite particles have been sampled (so the relative error is infinitely small), but rather that no particles have been binned for that energy bin. This technically results in an infinite FOM, but in reality represents a bin that will never converge. Because this value will hold no meaning in our quantification of the $\Omega$-methods' success, the infinite valued FOM is not included. Table \ref{tab:time_defaults} reports the times used to calculate the FOM values in Table \ref{tab:fom_defaults} more detail. This table is split into three vertical regions: the MCNP time spent doing Monte Carlo transport (T$_{MC}$), the deterministic time spent in ADVANTG/Denovo (T$_{det}$), and the walltime (T$_{hybrid}$), which is the summation of the two. The deterministic time section contains further segmentations of timing. This is because processes in ADVANTG are run using different computational resources. ADVANTG itself is a driver script that can launch a paralellized run in Exnihilo/Denovo, but it also postprocesses the Denovo fluxes into source biasing and weight window parameters. The processes exclusive to ADVANTG, like generating the biasing parameters, are performed in serial on a single processor. Conversely, all of the Denovo calculation is run in parallel on any number of cores specified by the user. To ensure that a comparable time is used when calculating the adjusted FOM, we have chosen to calculate the total walltime spent in each calculation. Thus, the parallelized clock time is multiplied by the total number of cores to obtain T$_{denovo}$. This quantity is summed with the runtimes of the other serial tasks to obtain the total deterministic runtime. \begin{table}[h!] \centering \input{./chapters/characterization_probs/figures/sample_data/time_defaults} \caption[Table of differing times used to measure $\Omega$ performance.]{ Table of differing times used to measure $\Omega$ performance. These times are used to calculate the FOMS in Table \ref{tab:fom_defaults}. } \label{tab:time_defaults} \end{table} Two other times are listed under the deterministic time that may or may not be included in T$_{hybrid}$, which are T$_{\Omega}$ and T$_{dispose}$. T$_{dispose}$ is the reported times that are not included in the calculation of T$_{det}$ in either CADIS or CADIS-$\Omega$. It is a sum of time results that either are not important to comparing the methods--like calculating the anisotropy metrics--or times that are accounted for by other tasks in T$_{det}$. This prevents overlap of times and provides a more realistic comparison between the performance of both methods. The reported $\Omega$ time, T$_{\Omega}$, is the total time spent in the tasks unique to the $\Omega$-methods. This includes reading in the angular flux files, performing the computation of Eq. \eqref{eq:omega_basic}, and writing the $\Omega$-results to a file. The $\Omega$ time, though run in Denovo, is still a serial calculation so is separated out from the total Denovo time. The $\Omega$-method tasks at this time are not parallelized, so the clock time is treated in the same way as the reported ADVANTG time. Because the majority of the $\Omega$-flux generation infrastructure is implemented in Exnihilo rather than ADVANTG, future expansions of the method could be parallelized for faster clock times. Because the adjusted FOM (the FOMs labeled FOM$_{hybrid}$ in Table \ref{tab:fom_defaults}) uses T$_{Hybrid}$, which is the total runtime of the Monte Carlo calculation (T$_{MC}$) and the hybrid/deterministic run preceding it (T$_{det}$), it will differ between the $\Omega$-methods, standard CADIS, and standard FW-CADIS. For CADIS, T$_{det}$ is the sum of the ADVANTG runtime and the wall time of the Denovo transport. For CADIS-$\Omega$, this is the sum of the ADVANTG runtime, the wall time of the Denovo transport, and the time spent in the $\Omega$-flux calculation. How each time is calculated is summarized in Table \ref{tab:time_defaults}. Beyond adding the $\Omega$-flux compute time, CADIS-$\Omega$ will generally have much longer Denovo runtimes than CADIS. This is a combination of the $\Omega$-methods' requirement of both a forward and adjoint calculation (recall that CADIS requires only the adjoint calculation), and that the $\Omega$-methods require full angular flux solutions to calculate the $\Omega$-flux. While standard CADIS has the ability to print the full angular flux solutions as CADIS-$\Omega$, it is neither a requirement nor is it standard practice. The I/O demands to both write the angular fluxes and then read them back in is a potential bottleneck in the method based on the current implementation. \subsubsection*{Tally Result and Relative Error Plots} Each of the problems introduced in Section \ref{subsec:ProbSpecs} has a 10x10x10 cm detector in which the tally response is calculated. The tallies are discretized in energy; the tally result and associated relative error are tabulated for each energy bin. Some of this information can be inferred from Table \ref{tab:fom_defaults}, but seeing the distribution of the relative errors for each energy bin for each method is a useful way of seeing how effective each method is at biasing particles all of the tally bins, without time effects. As described in the previous paragraph, CADIS-$\Omega$'s deterministic time will be longer than CADIS', so the FOM$_{hybrid}$ may be lower for the $\Omega$-methods, even if the relative errors are better. Presenting both the relative error distribution and the FOM will provide a clear picture of the performance of the $\Omega$-methods. The tally results and relative errors for CADIS, CADIS-$\Omega$, and the nonbiased analog Monte Carlo will be presented in figures similar to \ref{fig:sampleresult} and \ref{fig:sampleerror}. In the case where the relative error of the nonbiased analog Monte Carlo far exceeds the errors achieved by CADIS and CADIS-$\Omega$, it will be omitted. The example given in Figure \ref{fig:sampleerror} shows a result where this is the case. The hybrid methods will be marked with a dashed line; the nonbiased analog Monte Carlo will be a solid line. \begin{figure}[ht!] \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=0.8\linewidth]{./chapters/characterization_probs/figures/char/maze2/maze_2_tally_result_compare.pdf} \caption{Comparison between methods of the tally result.} \label{fig:sampleresult} \end{subfigure} \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=0.8\linewidth]{./chapters/characterization_probs/figures/char/maze2/maze_2_tally_error_compare.pdf} \caption{Comparison between methods of the tally relative error.} \label{fig:sampleerror} \end{subfigure} \caption[Sample results for a characterization problem tally.]{Sample results for a characterization problem tally.} \end{figure} \subsubsection*{Anisotropy Metrics} Equations \eqref{eq:metric_one} through \eqref{eq:metric_six} in Section \ref{sec:anisotropy_quant} presented several different ways by which the anisotropy of each problem could be quantified. As discussed in that section, Each metric will show slightly differing effects. For example, the ratio of the $\Omega$- to adjoint-flux in metric two will differ significantly from the angular contributon max to average of metric three. The $\Omega$-flux may be larger or smaller than the adjoint scalar flux depending on the directionality of the adjoint and forward particles relative to one another. If the particles are travelling in opposite directions, this will result in a larger omega flux than the adjoint flux. If they stream in the same direction (away from the tally detector, for example), then the resultant $\Omega$ flux will be smaller than the adjoint. In the case of the angular contributon max to average the distribution will have a lower limit where the maximum is very close to the average contributon flux. It can never be lower than the average. In a isotropic problem, the majority of the cells in the problem will be this ratio, whereas in a strongly anisotropic problem this distribution will shift upwards, but will still have the same limiting lower value as the isotropic case. \begin{figure}[htb!] \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=0.95\linewidth]{./chapters/characterization_probs/figures/sample_data/group_001_strip_full.pdf} \caption{Example distribution of anisotropy metrics for fastest energy group.} \label{fig:samplestrip001} \end{subfigure} \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=0.95\linewidth]{./chapters/characterization_probs/figures/sample_data/group_013_strip_full.pdf} \caption{Example distribution of anisotropy metrics for epithermal energy group.} \label{fig:samplestrip013} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=0.95\linewidth]{./chapters/characterization_probs/figures/sample_data/group_026_strip_full.pdf} \caption{Example distribution of anisotropy metrics for thermal energy group.} \label{fig:samplestrip026} \end{subfigure} \caption[Example distribution of all anisotropy metrics for highest, intermediate, and lowest energy groups.]{Example distribution of all anisotropy metrics for highest, intermediate, and lowest energy groups.} \label{fig:samplestrips} \end{figure} To illustrate the effect of how different the anisotropy metrics' distributions are, Figure \ref{fig:samplestrips} shows stripplots for all of the anisotropy metrics for three different energy groups in one of the characterization problems. The effects of thermalization--and consequently more induced isotropy--on each of the metrics can be seen clearly as one scans from Fig \ref{fig:samplestrip001} to \ref{fig:samplestrip026}. The adjoint anisotropy metric, the forward anisotropy metric, and metric three are all shifted by a factor of $4\pi$. Their natural lowest limit should be near unity but all lie lower. This may be corrected in the future, but for the purposes of this analysis we are more interested in the relative distribution and the consistent factor of $4\pi$ is not important to that effect. A stripplot shows distinct data points, but easily can be overwhelmed if the full number of cells is used in a single strip. The figures in \ref{fig:samplestrips} contain a random selection of 1500 data points from the full anisotropy datasets, which is only a small fraction of the number of cells in the characterization problem meshes. There are other ways to visualize the full distribution of the dataset. Figure \ref{fig:sampledistros} shows three modes by which an anisotropy metric can be visualized. These plots, unlike Figure \ref{fig:samplestrips}, show a single metric but all energy groups. The highest/fastest energy group is plotted in deep red, and the lowest or most thermal energy group is shown in blue. \begin{figure}[htb!] \centering \begin{subfigure}[t]{\textwidth} \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_four_strip_full.pdf} \caption{Example distribution of M$_4$, all energy groups, strip plot.} \label{fig:samplestripM4} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_four_box_full.pdf} \caption{Example distribution of M$_4$, all energy groups, box plot.} \label{fig:sampleboxM4} \end{subfigure} \begin{subfigure}[t]{\textwidth} \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_four_violin_full.pdf} \caption{Example distribution of M$_4$, all energy groups, violin plot.} \label{fig:sampleviolinM4} \end{subfigure} \caption[Different ways of visualizing M$_4$ for a characterization problem.] {Different ways of visualizing M$_4$ for a characterization problem.} \label{fig:sampledistros} \end{figure} All three subfigures in \ref{fig:sampledistros} show the effects of thermalization on the chosen metric distribution and density. The stripplot of \ref{fig:samplestripM4} is a clear representation of the density, but not much more can be ascertained about the distribution of the metric. Figure \ref{fig:sampleboxM4} has box and whisker plots that show the data quartiles, the mean, and outliers. However, in the case where the distribution is heavily towards a limiting value, the mean is hard to separate from the distribution. Further, no data on how the metric is distributed beyond the quartile markers is provided. The violin plot of Figure \ref{fig:sampleviolinM4} is a hybrid of the former two plots. The width of the violin is related to the density of values, but inside the violin the limits of the box plots are marked in black. The violin limits extend to the outliers. The analysis for each of the characterization problems look at the result for the tally average relative error, the tally maximum relative error, and the tally minumum relative error. Because we are interested in how the relative error in each energy bin changes with respect to CADIS-$\Omega$ and CADIS, the plots showing the distributions over all energy groups for a single metric is generally more applicable than the plots for a single energy group but with all metrics. As a result, future plots of the metrics will be in the style of those in Figure \ref{fig:sampledistros} rather than \ref{fig:samplestrips}. \subsubsection*{Filtered Anisotropy Metrics} \begin{figure}[htb!] \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_two_violin_full.pdf} \caption{Example distribution of M$_2$, all energy groups, violin plot} \label{fig:samplefullviolinM2} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_two_violin_median.pdf} \caption{Example distribution of M$_2$, all energy groups, violin plot using only datapoints above the median metric value in each energy group.} \label{fig:samplemedianviolinM2} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_two_violin_mean.pdf} \caption{Example distribution of M$_2$, all energy groups, violin plot using only datapoints above the mean metric value in each energy group.} \label{fig:samplemeanviolinM2} \end{subfigure} \caption[M$_2$ violin plots using different selections of the metric data.] {M$_2$ violin plots using different selections of the metric data.} \label{fig:sampleviolinsM2} \end{figure} Beyond plotting the anisotropy metrics as a function of energy group, we are interested in how the relative error or FOM will respond as a function of each metric. However, not all cells in the problem are as important as others to contributing to the tally. A cell on the problem boundary is very unlikely to contribute to the tally result when compared to a cell next to the adjoint source. As discussed in Section \ref{sec:ContributonImportance}, the contributon flux measures the response importance of a cell. By selectively choosing anisotropy metrics from cells that are likely to induce a response, some of the noise of less important cells can be cut out. To consistently cut out the same number of datapoints across all metrics, we have chosen to use a filtering algorithm based on the contributon flux in each cell. The first filter is choosing metric values from cells where the contributon flux is above the problem median contributon flux. This median is evaluated separately for each energy group to ensure that the same number of cells in each group is plotted. The second filter is choosing metric values from cells where the contributon flux is above the problem mean contributon flux. Again, the mean is computed separately for each energy group such that energy groups with higher contributon fluxes do not cut out important flux values from a different energy group. However, unlike the median filter a different number of cells for each energy group will be filtered. This is dependent on the skew between the contributon mean and median value for each energy group. Because the filter is evaluated based on the contributon flux, it can be applied to each metric consistently, meaning that the same number of cells are filtered out between different metrics. Figure \ref{fig:sampleviolinsM2} shows the effects of cutting out data from unimportant cells on the M$_2$ distribution. The first figure in the series, \ref{fig:samplefullviolinM2}, is the M$_2$ full distribution. As discussed previously, M$_2$ will be above unity in cells where the foward and adjoint angular fluxes travel in opposing directions, and will be below unity in cells where they travel in the same direction. Very unimportant cells should be below unity. Applying the first filter--selecting values above the contributon median--to this distribution results in Figure \ref{fig:samplemedianviolinM2}. The bottom tails of all of the distributions have been shortened, but still many unimportant cells remain. This should be expected, as only half of cells have been removed. Applying the second filter results in Figure \ref{fig:samplemeanviolinM2}. The unimportant tails have been almost completely removed from the M$_2$ distributions. Further, features in the metric distribution once obviscated by the tails are now visible. \subsubsection*{Improvement Factor Correlations with Anisotropy} Now that a way of visualizing the metric distributions has been presented, we seek to find how the metric distributions relate to the relative error or FOM for a given problem. First, an improvement ratio for the relative error and FOM will be defined. For the relative error it is \begin{equation} I_{RE} = \frac{RE_{CADIS-\Omega}}{RE_{CADIS}}\bigg\rvert_{E_g}, \label{eq:I-RE} \end{equation} and for the FOM it is \begin{equation} I_{FOM} = \frac{FOM_{CADIS-\Omega}}{FOM_{CADIS}}\bigg\rvert_{E_g}. \label{eq:I-FOM} \end{equation} These will be henceforth be referred to as the relative error and FOM improvement factors. With this definition of the improvement in the FOM or the relative error from CADIS to CADIS-$\Omega$, we now have a comparison between the updated and standard methods. By relating this metric to the anisotropy metrics, we can see how anisotropy of the problem influences the improvement in the relative error or the FOM. \begin{figure}[htb!] \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_three_err_stats_full.pdf} \caption{M$_3$ average, mean, skew, and variance plotted against the relative error improvement I$_{RE}$} \label{fig:samplestatsfullM3} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_three_err_stats_median.pdf} \caption{M$_3$ data selection above the metric median for each energy group, value average, mean, skew, and variance plotted against the relative error improvement I$_{RE}$} \label{fig:samplestatsmedianM3} \end{subfigure} \end{figure} \begin{figure}[htb!]\ContinuedFloat \centering \begin{subfigure}[t]{\textwidth} \centering \includegraphics[width=\linewidth]{./chapters/characterization_probs/figures/sample_data/metric_three_err_stats_mean.pdf} \caption{M$_3$ data selection above the mean for each energy group, value average, mean, skew, and variance plotted against the relative error improvement I$_{RE}$ } \label{fig:samplestatsmeanM3} \end{subfigure} \caption[Sample scatterplots of M$_3$ distribution against the relative error improvement factor, I$_{RE}$.] {Sample scatterplots of the M$_3$ distribution against the relative error improvement factor, I$_{RE}$.} \label{fig:samplestats} \end{figure} There are several ways in which the improvement factor I$_{RE}$ or I$_{FOM}$ may be compared against the anisotropy metrics. The first are against the metric mean and median values. A plot of I versus either of these values should look very simiar, with some shifting depending on the distribution. However, if the mean and median are shifted significantly, this would indicate a skew of the distribution. This skew may also be correlated with either of the I values. Last, it is possible that the spread of metric values may be correlated with the I factor. Figure \ref{fig:samplestatsfullM3} is an illustration of how I can be plotted with each of these measurements of the metric distribution. Similar to using the filtering algorithms in Figure \ref{fig:sampleviolinsM2}, the data in the statistical trend plots can also be filtered. The subfigures in \ref{fig:samplestats} illustrate how filtering out the data by the contributon flux influences the location of I$_{RE}$ for each energy group. Figure \ref{fig:samplestatsmedianM3} calculates the metric mean, median, skew, and variance for each energy group using only metric values in cells above the contributon median. Conversely, Figure \ref{fig:samplestatsmeanM3} calculates the metric mean, median, skew, and variance for each energy group using only metric values in cells above the contributon median value. The dots in each plot correspond to the same energy groups plotted in \ref{fig:sampleviolinsM2}. That is, the lowest energy is plotted in blue and the highest in red. Note that this type of plot is possible because the Monte Carlo tally has been discretized to have the same binning as the the deterministic code. It would be far more difficult if the energy bin widths of the Monte Carlo tally did not match the deterministic code. The data that will be presented for each characterization problem can be subdivided into three distinct categories: data primarily obtained by the Monte Carlo calculation, data primarily obtained by the deterministic calculation, and data that is a combination of both. The FOM values using Monte Carlo runtimes, for example, is in the first category. The anisotropy metrics presented in Section \ref{sec:anisotropy_quant} are an example of a determinstic-exclusive dataset. The results presented in Figure \ref{fig:samplestats} are a combination of both deterministic and Monte Carlo-influenced results. In studying the $\Omega$ methods, we seek to understand how the $\Omega$ methods' performance influence the Monte Carlo results. Beyond observing the FOM and relative error distribution obtained in the Monte Carlo, the anisotropy metrics will provide another avenue by which to investigate $\Omega$-method performance. One may have deduced that the results for the characterization problems and the subsequent angle sensitivity study will be substantive. Only the most pertinent fraction of the available data will be presented with each problem in Sections \ref{sec:CharResults} and \ref{sec:AngleResults}. For example, in most cases only a single figure--and perhaps only a single metric--from the three presented in \ref{fig:samplestats} will be presented for a particular problem, because only one will show a trend relevant to the $\Omega$-methods' performance. A more extensive set of data and figures is accessible in the public repositories listed in Appendix \ref{ch:codes}. dependencies/tex/latex/cjkpunct/CJKpunct.sty %% %% This is file `CJKpunct.sty', %% generated with the docstrip utility. %% %% The original source files were: %% %% CJKpunct.dtx (with options: `CJKpunct') %% %% Version 4.8.2 (06-May-2009) %% %% This is the file CJKpunct.sty for the CJK package %% %% Authors: %% () %% () %% %% \def\fileversion{4.8.2} \def\filedate{2009/05/06} \ProvidesPackage{CJKpunct}[\filedate\space\fileversion] \endlinechar \m@ne \newif\if@CJKpunct \newif\if@CJKpunct@dokerning \newcount\CJKpunct@cnta \newcount\CJKpunct@cntb \newcount\CJKpunct@cntc \newcount\CJKpunct@cntd \newcount\CJKpunct@cnte \let\CJKo@testLastCJK\CJK@testLastCJK \def\CJKpunct@testLastCJK{ \global\CJK@false \global\edef\CJKpunct@lastkern{\number\lastkern}} \let\CJKo@testLastKern\CJK@testLastKern \def\CJKpunct@testLastKern{ \global\CJK@false} \let\CJKo@testPrePunct\CJK@testPrePunct \let\CJKo@testPostPunct\CJK@testPostPunct \def\CJKpunct@testPrePunct#1#2#3{} \def\CJKpunct@testPostPunct#1#2#3{} \let\CJKo@nobreakglue\CJK@nobreakglue \let\CJKosymbol\CJKsymbol \def\CJKpunct@CJKsymbol#1{ {{{ \ifnum\CJKpunct@lastkern>0\relax \ifnum\CJKpunct@lastcharclass=0\relax \CJKglue \else \CJKpunct@ULspecials \fi \fi \CJKosymbol{#1} \gdef\CJKpunct@lastcharclass{0}}}}} \def\CJKpunct@lastcharclass{0} \def\CJKpunct@lastkern{0} \let\CJKopunctsymbol\CJKpunctsymbol \def\CJKpunct@CJKpunctsymbol#1{ \CJKpunct@setfamily \CJKpunct@setmarginkerning \edef\CJKpunct@currentpunct{\CJK@plane/\the#1} \ifcsname CJKpunct@\CJK@enc @\CJKpunct@currentpunct\endcsname \edef\CJKpunct@currentcharclass{ \csname CJKpunct@\CJK@enc @\CJKpunct@currentpunct\endcsname} {{{% We need three braces for CJKulem to work \@CJKpunctfalse \ifnum\CJKpunct@lastkern>0\relax \ifnum\CJKpunct@lastcharclass>0\relax \unkern \unkern \ifnum\CJKpunct@punctstyle>0\relax \@CJKpuncttrue \else \ifcsname CJKpunct@specialpunct\CJK@enc \CJKpunct@currentpunct\endcsname \@CJKpuncttrue \fi \fi \fi \fi \if@CJKpunct \CJKpunct@unskip \CJKpunct@setkern{\CJKpunct@lastpunct}{\CJKpunct@currentpunct} \kern \csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @kern\CJKpunct@lastpunct @\CJKpunct@currentpunct\endcsname \CJKpunct@nobreak \else \CJKpunct@ULspecials \ifnum\CJKpunct@currentcharclass=1\relax \hskip \csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @lglue@\CJKpunct@currentpunct\endcsname plus 0.1em minus 0.1 em \else \ifcsname CJKpunct@specialpunct\CJK@enc \CJKpunct@currentpunct\endcsname \CJKglue % breakable \else \nobreak \fi \fi \fi \global\edef\CJKpunct@lastpunct{\CJKpunct@currentpunct} \vrule width \csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @lrule@\CJKpunct@currentpunct\endcsname depth \z@ height \z@ \CJKopunctsymbol{#1} \vrule width \csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @rrule@\CJKpunct@currentpunct\endcsname depth \z@ height \z@ \ifnum\CJKpunct@currentcharclass=2\relax \hskip \csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @rglue@\CJKpunct@currentpunct\endcsname plus 0.1em minus 0.1 em \fi \global\let\CJKpunct@lastcharclass\CJKpunct@currentcharclass}}} \else \CJKsymbol{#1} \global\def\CJKpunct@lastcharclass{0} \fi} \def\CJKpunct@setfamily{ \ifcsname \CJK@enc @\CJK@family @\f@series @\f@shape\endcsname \global\edef\CJKpunct@family{\csname \CJK@enc @\CJK@family @\f@series @\f@shape\endcsname} \else \edef\CJKpunct@family{\CJK@family} \fi} \def\CJKpunctmapfamily#1#2#3#4#5{ \expandafter\edef\csname #1@#2@#3@#4\endcsname{#5}} \def\CJKpunct@plainpunctsymbol#1#2{ \CJKpunctsymbol{#2}} \def\CJKpunct@setmarginkerning{ \ifcsname CJKpunct @\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family\endcsname \else \expandafter\gdef\csname CJKpunct @\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family\endcsname{} \ifcsname CJKpunct@\CJKpunct@family @spaces\endcsname \PackageInfo{CJKpunct}{use punctuation spaces for family '\CJKpunct@family' \space with punctstyle (\CJKpunct@currentpunctstyle)}\relax \edef\CJKpunct@spaces{\csname CJKpunct@\CJKpunct@family @spaces\endcsname} \else \ifcsname CJKpunct@spaces@\CJKpunct@family\endcsname \else \PackageInfo{CJKpunct}{punctuation spaces for family '\CJKpunct@family' do not exist. \space Use family 'def' instead.}\relax \global\expandafter\def\csname CJKpunct@spaces@\CJKpunct@family\endcsname{} \fi \edef\CJKpunct@spaces{\csname CJKpunct@def@spaces\endcsname} \fi \CJKpunct@cnta=0\relax \expandafter\CJKpunct@@setmarginkerning\CJKpunct@spaces \fi} \def\CJKpunct@@setmarginkerning#1,#2,{ \edef\CJKpunct@temp{#1} \ifx\CJKpunct@temp\@empty \def\CJKpunct@temp{} \else \def\CJKpunct@temp{\CJKpunct@@setmarginkerning} \ifnum\CJKpunct@cnta<12 \def\CJKpunct@lr{l} \else \def\CJKpunct@lr{r} \fi \edef\CJKpunct@encpn{\csname CJKpunct@pn@\CJK@enc @\the\CJKpunct@cnta\endcsname} \if l\CJKpunct@lr \expandafter\gdef\csname CJKpunct@\CJK@enc @\CJKpunct@encpn\endcsname{1} \else \expandafter\gdef\csname CJKpunct@\CJK@enc @\CJKpunct@encpn\endcsname{2} \fi \@CJKpunct@dokerningtrue \ifnum\CJKpunct@punctstyle=\CJKpunct@ps@plain\relax \@CJKpunct@dokerningfalse \else \ifcsname CJKpunct@specialpunct\CJK@enc\CJKpunct@encpn\endcsname \@CJKpunct@dokerningfalse \fi \fi \ifnum\CJKpunct@punctstyle=\CJKpunct@ps@banjiao \def\CJKpunct@sidespaces{12} \else \def\CJKpunct@sidespaces{15} \fi \ifnum\CJKpunct@cnta=12\relax {\CJKpunct@cntb=#1\relax \advance\CJKpunct@cntb #2\relax \advance\CJKpunct@cntb 2\relax \CJKpunct@numtostring{\CJKpunct@cntb} \edef\CJKpunct@temp{\csname CJKpunct@pn@\CJK@enc @12\endcsname} \CJKpunct@cntc=0\relax \loop \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @kern\CJKpunct@temp @\CJKpunct@temp\endcsname{ -0.\CJKpunct@decimal em} \advance \CJKpunct@cntc 1\relax \ifnum\CJKpunct@cntc<6\repeat} \fi \if@CJKpunct@dokerning \CJKpunct@cntb=#1\relax \advance\CJKpunct@cntb -\CJKpunct@sidespaces\relax \ifnum\CJKpunct@cntb<0\relax \CJKpunct@cntb=0\relax \fi \CJKpunct@cntc=#2\relax \advance\CJKpunct@cntc -\CJKpunct@sidespaces\relax \ifnum\CJKpunct@cntc<0\relax \CJKpunct@cntc=0\relax \fi \CJKpunct@cntd=\CJKpunct@cntb \advance\CJKpunct@cntd\CJKpunct@cntc\relax \ifcase\CJKpunct@punctstyle % hangmobanjiao \or % quanjiao \or % banjiao \advance\CJKpunct@cntd -50\relax \or % kaiming \ifcsname CJKpunct@kaiming\CJK@enc\CJKpunct@encpn\endcsname \else \advance\CJKpunct@cntd -50\relax \fi \or %CCT \advance\CJKpunct@cntd -20\relax \fi \CJKpunct@cnte=\CJKpunct@cntd \ifnum\CJKpunct@cntd<0\relax \CJKpunct@cntd=0\relax \fi \else \CJKpunct@cntb=0\relax \CJKpunct@cntc=0\relax \CJKpunct@cntd=0\relax \CJKpunct@cnte=0\relax \fi \CJKpunct@numtostring{\CJKpunct@cntb} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @lrule@\CJKpunct@encpn\endcsname{ -0.\CJKpunct@decimal em} \CJKpunct@numtostring{\CJKpunct@cntc} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @rrule@\CJKpunct@encpn\endcsname{ -0.\CJKpunct@decimal em} \CJKpunct@numtostring{\CJKpunct@cntd} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @\CJKpunct@lr glue@\CJKpunct@encpn\endcsname{ 0.\CJKpunct@decimal em} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @\CJKpunct@lr oglue@\CJKpunct@encpn\endcsname{ \the\CJKpunct@cnte} \fi \advance \CJKpunct@cnta 1\relax \CJKpunct@temp} \def\CJKpunct@numtostring#1{ \edef\CJKpunct@decimal{\the#1} \ifnum\CJKpunct@decimal<10\relax \edef\CJKpunct@decimal{0\CJKpunct@decimal} \fi} \def\CJKpunct@setkern#1#2{ \ifcsname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @kern#1@#2\endcsname \else \CJKpunct@cnta=0\relax \ifcsname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @roglue@#1\endcsname \advance\CJKpunct@cnta\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @roglue@#1\endcsname \fi \ifcsname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @loglue@#2\endcsname \advance\CJKpunct@cnta\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @loglue@#2\endcsname \fi \relax \ifcase\CJKpunct@punctstyle % hangmobanjiao \or % quanjiao \advance\CJKpunct@cnta -50\relax \or % banjiao \or % kaiming \ifcsname CJKpunct@kaiming#1\endcsname \ifcsname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @loglue@#2\endcsname \advance\CJKpunct@cnta -50\relax \fi \fi \fi \ifnum\CJKpunct@cnta<0\relax \CJKpunct@cnta=0\relax \fi \CJKpunct@numtostring{\CJKpunct@cnta} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @kern#1@#2\endcsname{ 0.\CJKpunct@decimal em} \fi} \let\CJKpunct@unskip\unskip \def\CJKpunct@UL@unskip{ \ifcsname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @rglue@\CJKpunct@lastpunct\endcsname \hskip -\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @rglue@\CJKpunct@lastpunct\endcsname \relax \fi} \@ifundefined{UL@hskip}{\let\UL@hskip\relax}{} \def\CJKpunct@punctUL@group{ \ifx\hskip\UL@hskip \egroup \UL@stop \UL@start \bgroup \fi} \def\CJKpunct@ULspecials{} \AtBeginDocument{ \ifcsname UL@hook\endcsname \addto@hook\UL@hook{\let\CJK@ignorespaces\ignorespaces \let\CJKpunct@unskip\CJKpunct@UL@unskip \let\CJKpunct@ULspecials\CJKpunct@punctUL@group} \fi} \def\CJKpunctallowbreakbetweenpuncts{ \def\CJKpunct@nobreak{ \ifnum\CJKpunct@lastcharclass=2 \hskip 0pt \fi}} \def\CJKpunctnobreakbetweenpuncts{ \let\CJKpunct@nobreak\nobreak} \CJKpunctnobreakbetweenpuncts \def\CJKpunctstyle#1{ \ifcsname CJKpunct@ps@#1\endcsname \edef\CJKpunct@currentpunctstyle{#1} \edef\CJKpunct@punctstyle{\csname CJKpunct@ps@#1\endcsname} \ifnum\CJKpunct@punctstyle=\CJKpunct@ps@plain\relax \CJKpunctallowbreakbetweenpuncts \let\CJK@testLastCJK\CJKo@testLastCJK \let\CJK@testLastKern\CJKo@testLastKern \let\CJK@testPrePunct\CJKo@testPrePunct \let\CJK@testPostPunct\CJKo@testPostPunct \let\CJKpunct@punctsymbol\CJKpunct@plainpunctsymbol \let\CJKsymbol\CJKosymbol \let\CJKpunctsymbol\CJKopunctsymbol \let\CJK@nobreakglue\CJKo@nobreakglue \let\CJKpunct@utfsymbol\CJKpunct@utfbsymbol \else \let\CJK@testLastCJK\CJKpunct@testLastCJK \let\CJK@testLastKern\CJKpunct@testLastKern \let\CJK@testPrePunct\CJKpunct@testPrePunct \let\CJK@testPostPunct\CJKpunct@testPostPunct \let\CJKpunct@punctsymbol\CJKpunct@@punctsymbol \let\CJKsymbol\CJKpunct@CJKsymbol \let\CJKpunctsymbol\CJKpunct@CJKpunctsymbol \let\CJK@nobreakglue\relax \let\CJKpunct@utfsymbol\CJKpunct@utfasymbol \fi \else \PackageWarning{CJKpunct}{Punctstyle #1\space is not defined.}\relax \fi} \let\punctstyle\CJKpunctstyle \def\CJKpunct@ps@hangmobanjiao{0} \def\CJKpunct@ps@marginkerning{0} \def\CJKpunct@ps@quanjiao{1} \def\CJKpunct@ps@fullwidth{1} \def\CJKpunct@ps@banjiao{2} \def\CJKpunct@ps@halfwidth{2} \def\CJKpunct@ps@kaiming{3} \def\CJKpunct@ps@mixedwidth{3} \def\CJKpunct@ps@CCT{4} \def\CJKpunct@ps@plain{5} \AtBeginDocument{\punctstyle{quanjiao}} \def\CJKplainout{\punctstyle{plain}} \let\CJKnormalout\relax \def\CJKpunctsetkern#1#2#3{ \CJKpunct@setplanenumber{#1} \edef\CJKpunct@pna{\CJKpunct@char@pn} \CJKpunct@setplanenumber{#2} \edef\CJKpunct@pnb{\CJKpunct@char@pn} \global\expandafter\edef\csname CJKpunct\CJKpunct@punctstyle @\CJK@enc @\CJKpunct@family @kern\CJKpunct@pna @\CJKpunct@pnb\endcsname{ #3}} \def\CJKpunct@setplanenumber#1{{ \def\CJK@testPrePunct##1##2##3{ \global\edef\CJKpunct@charplane{\CJK@plane} \global\edef\CJKpunct@charnumber{\the\@tempcnta}} \savebox\voidb@x{#1} \global\edef\CJKpunct@char@pn{\CJKpunct@charplane/\CJKpunct@charnumber}}} \def\CJKpunct@punctlist#1{ \CJKpunct@cnta=0\relax \def\CJKpunct@enc{#1} \CJKpunct@setpunctfamilynumber} \def\CJKpunct@setpunctfamilynumber#1,{ \edef\CJKpunct@temp{#1} \ifx\CJKpunct@temp\@empty \def\CJKpunct@temp{} \else \expandafter\def\csname CJKpunct@pn@\CJKpunct@enc @\the\CJKpunct@cnta\endcsname{#1} \advance \CJKpunct@cnta 1\relax \def\CJKpunct@temp{\CJKpunct@setpunctfamilynumber} \fi \CJKpunct@temp} \CJKpunct@punctlist{C70}20/24,20/28,30/12,30/14,30/20,ff/8,ff/59,ff/91,% 30/8,30/10,30/22,30/16,% 20/20,20/38,30/1,30/2,ff/12,ff/14,ff/26,ff/27,ff/1,ff/31,ff/5,30/21,ff/9,% ff/61,ff/93,30/9,30/11,30/23,30/17,20/25,20/29,30/13,30/15,, \CJKpunct@punctlist{C10}01/13,01/15,01/23,01/25,01/17,01/195,01/246,02/22,01/19,% 01/21,01/27,01/29,% 01/9,01/12,01/1,01/2,01/199,01/201,01/213,01/214,01/188,01/218,01/192,01/18,% 01/196,01/248,02/24,01/20,01/22,01/28,01/30,01/14,01/16,01/24,01/26,, \CJKpunct@punctlist{C19}25/45,25/47,25/55,25/57,25/49,26/163,26/214,26/246,25/51,% 25/53,25/59,25/61,% 25/41,25/44,25/33,25/34,26/167,26/169,26/181,26/182,26/156,26/186,26/160,% 25/50,26/164,26/216,26/248,25/52,25/54,25/60,25/62,25/46,25/48,25/56,25/58,, \def\CJKpunct@totalpuncts{35} \ifcsname DeclareUnicodeCharacter\endcsname \DeclareUnicodeCharacter{2018}{\CJKpunct@utfsymbol{"80}{"98}} \DeclareUnicodeCharacter{2019}{\CJKpunct@utfsymbol{"80}{"99}} \DeclareUnicodeCharacter{201C}{\CJKpunct@utfsymbol{"80}{"9C}} \DeclareUnicodeCharacter{201D}{\CJKpunct@utfsymbol{"80}{"9D}} \DeclareUnicodeCharacter{2014}{\CJKpunct@utfsymbol{"80}{"94}} \DeclareUnicodeCharacter{2026}{\CJKpunct@utfsymbol{"80}{"A6}} \fi \def\CJKpunct@utfasymbol#1#2{ \CJK@punctchar{\CJK@uniPunct}{0}{#1}{#2}} \def\CJKpunct@utfbsymbol#1#2{ \ifnum #2=148 \textemdash \else \ifnum #2=166 \textellipsis \else \ifnum #2=152 \textquoteleft \else \ifnum #2=153 \textquoteright \else \ifnum #2=156 \textquotedblleft \else \ifnum #2=157 \textquotedblright \fi \fi \fi \fi \fi \fi} \def\CJKpunct@setspecialpunct#1#2{ \expandafter\def\csname CJKpunct@specialpunct#1#2\endcsname{}} \CJKpunct@setspecialpunct{C70}{20/20} \CJKpunct@setspecialpunct{C70}{20/38} \CJKpunct@setspecialpunct{C19}{25/41} \CJKpunct@setspecialpunct{C19}{25/44} \CJKpunct@setspecialpunct{C10}{01/9} \CJKpunct@setspecialpunct{C10}{01/12} \def\CJKpunct@setkaimingpunct#1#2{ \expandafter\def\csname CJKpunct@kaiming#1#2\endcsname{}} \CJKpunct@setkaimingpunct{C70}{30/02} \CJKpunct@setkaimingpunct{C70}{ff/1} \CJKpunct@setkaimingpunct{C70}{ff/31} \CJKpunct@setkaimingpunct{C19}{25/34} \CJKpunct@setkaimingpunct{C19}{26/156} \CJKpunct@setkaimingpunct{C19}{26/186} \CJKpunct@setkaimingpunct{C10}{01/2} \CJKpunct@setkaimingpunct{C10}{01/188} \CJKpunct@setkaimingpunct{C10}{01/218} \def\CJKpunct@def@spaces{69,18,60,6,63,2,63,3,69,8,69,6,69,1,39,% 37,63,4,56,2,63,5,63,6,6,6,12,11,23,50,24,54,16,71,20,69,12,76,13,% 74,26,61,3,50,3,4,8,69,6,69,2,69,38,39,4,62,2,55,5,62,7,62,16,71,9,% 58,3,62,3,62,,,} \IfFileExists{CJKpunct.spa}{\input{CJKpunct.spa}}{} \endlinechar `\^^M \endinput %% %% End of file `CJKpunct.sty'. % 6-color-theorem.tex %%%%%%%%%%%%%%% \begin{frame}{} \begin{theorem} Every \cyan{simple} \red{planar} graph is \blue{6-colorable}. \end{theorem} \pause \vspace{0.30cm} \begin{center} \red{By induction on the number of vertices.} \pause \vspace{0.30cm} \begin{description}[<+->][Induction Hypothesis:] \setlength{\itemsep}{6pt} \item[Basis Step:] $n = 1$. Trivial. \item[Induction Hypothesis:] Suppose that it holds for simple planar graphs with $n \ge 1$ vertices. \item[Induction Step:] Consider a simple planar graph $G$ with $n+1$ vertices. \\[5pt] \uncover<6->{\purple{$G$ contains a vertex $v$ of degree $\le 5$.} \\[5pt]} \uncover<7->{$G' = G - v$ is 6-colorable. \\[5pt]} \uncover<8->{Thus, $G$ is 6-colorable.} \uncover<9->{\quad \cyan{($\deg(v) \le 5$)}} \end{description} \end{center} \end{frame} %%%%%%%%%%%%%%%GoldbergData/newWorldSimulation \hypertarget{class_saveela}{}\doxysection{Saveela Class Reference} \label{class_saveela}\index{Saveela@{Saveela}} Inheritance diagram for Saveela\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{class_saveela} \end{center} \end{figure} \doxysubsection*{Additional Inherited Members} The documentation for this class was generated from the following files\+:\begin{DoxyCompactItemize} \item \mbox{\hyperlink{_saveela_8h}{Saveela.\+h}}\item \mbox{\hyperlink{_saveela_8cpp}{Saveela.\+cpp}}\end{DoxyCompactItemize} 0 @misc{rfc6747, series = {Request for Comments}, number = 6747, howpublished = {RFC 6747}, publisher = {RFC Editor}, doi = {10.17487/RFC6747}, url = {https://rfc-editor.org/rfc/rfc6747.txt}, author = { and }, title = {{Address Resolution Protocol (ARP) for the Identifier-Locator Network Protocol for IPv4 (ILNPv4)}}, pagetotal = 12, year = 2012, month = nov, abstract = {This document defines an Address Resolution Protocol (ARP) extension to support the Identifier-Locator Network Protocol for IPv4 (ILNPv4). ILNP is an experimental, evolutionary enhancement to IP. This document is a product of the IRTF Routing Research Group. This document defines an Experimental Protocol for the Internet community.}, } lorenchorley/StrangeIoC-Updated1-10 \hypertarget{dir_f3822ae47b8b22233d6c26fd0cf5af66}{\section{Assets/scripts/babel/framework Directory Reference} \label{dir_f3822ae47b8b22233d6c26fd0cf5af66}\index{Assets/scripts/babel/framework Directory Reference@{Assets/scripts/babel/framework Directory Reference}} } \subsection*{Directories} \begin{DoxyCompactItemize} \item directory \hyperlink{dir_f560a35ff485aa3960a08dd89c96baee}{api} \item directory \hyperlink{dir_73211d799b8183164bac4022989f2c65}{impl} \end{DoxyCompactItemize} 1-10 \section{Conclusiones} Se requería implementar un medio para simular batallas entre bandos distintos. Para ello, se implementó un simulador de batallas el cual es personalizable a través de un lenguaje de dominio específico Battle Script. La simulación de batallas en un entorno controlado ayudaría a reducir el costo en vidas humanas en las guerras, así como ahorrar recursos económicos y tomar decisiones estratégicas. Se implementaron unidades que funcionan como agentes casi puramente reactivos, así como bandos que representan los ejércitos, batallones, compañías, escuadrones, etc. Las unidades funcionan a través de un sistema experto que les aporta la inteligencia necesaria para actuar con el fin de eliminar a todas las unidades del o de los bandos contrarios. Se definió un módulo para la generación automática de mapas de alturas. Se probó la simulación con distintos casos de prueba, con mapas con poco o mucho relieve, bandos grandes o pequeños, así como se modificó el comportamiento de nuevas unidades definidas a través del lenguaje. Cabe destacar, que la simulación se ejecuta de forma eficiente, de forma que el usuario final obtiene resultados en poco tiempo. Se sugiere desarrollar las sugerencias o recomendaciones para que el entorno y las unidades se asemejen cada vez más a los soldados, maquinarias de guerra, o ejércitos reales.doc/context/third/simpleslides/solutions/speaker_introduction-2min.tex0 % This file is a solution template for: % - Introducing another speaker. % - Talk length is about 2min. % - Style is informal. % This is adapted from the example by <> % included as part of the beamer package in LaTeX % % In principle, this file can be redistributed and/or modified under % the terms of the GNU Public License, version 2. % % However, this file is supposed to be a template to be modified % for your own needs. For this reason, if you use this file as a % template and not specifically distribute it as part of a another % package/program, the author grants the extra permission to freely % copy and modify this file as you see fit and even to delete this % copyright notice. \usemodule [simpleslides] [style=Boxed, font=Bookman] \starttext \SlideTitle {Speaker's Name} \startitemize \item Current affiliation of Speaker's Name % Examples: \startitemize \item Professor of mathematics, University of Wherever. \item Junior partner at company X. \item Speaker for organization/project X. \stopitemize \stopitemize \SlideTitle {Speaker's Name} \startitemize \item Experience and achievements % Optional. Use this if it is appropriate to slightly flatter the % speaker, for example if the speaker has been invited. % Using subitems, list things that make the speaker look % interesting and competent. % Examples: \startitemize \item Academic degree, but only if appropriate \item Current and/or previous positions, possibly with dates \item Publications (possibly just number of publications) \item Awards, prizes \stopitemize \stopitemize \SlideTitle {Speaker's Name} \startitemize \item Concerning today's talk % Optional. Use this to point out specific experiences/knowledge % of the speaker that are important for the talk and that do not % follow from the above. % Examples: \startitemize \item Expert who has worked in the field/project for X month/years. \item Will present his/her/group's/company's research on the subject. \item Will summarize project report or current project status. \stopitemize \stopitemize \stoptext chapters/05_study/sections/results.tex \section{Results} \label{chp:study:sec:results} In this chapter we present the results of the executed study which we base the later discussion upon. Specifically, we state the calculated metrics of the two runs. First we look at the metrics of the first run in which we use the crowdsourced dataset $D_{crowd}$. Here all three models achieve mostly similar metrics when classifying $D_{crowd_{test}}$. All metrics are listed in \cref{tab:study:results:first_run} for each model. \begin{table}[htpb] \centering \begin{tabular}{l | l l l l } \toprule Model & Precision & Recall & $F_1$ Score & Average Precision \\ \midrule \ac{BERT} & $0.36$ & $0.82$ & $0.51$ & $0.50$\\ \ac{DistilBERT} & $0.31$ & $0.79$ & $0.45$ & $0.47$\\ \ac{ERNIE2.0} & $0.48$ & $0.52$ & $0.50$ & $0.47$\\ \bottomrule \end{tabular} \caption[Study Results on Crowdsourced Dataset]{The results for the crowdsourced dataset $D_{crowd_{test}}$.}\label{tab:study:results:first_run} \end{table} As we describe in \cref{chp:study:sec:execution}, we performed a second run similar to the first one. The only difference is that the more extensive dataset $D_{all_{train}}$ was used for the training and $D_{all_{test}}$ to calculate the resulting metrics correspondingly. The results of this second run are shown in \cref{tab:study:results:second_run}. \begin{table}[htpb] \centering \begin{tabular}{l | l l l l } \toprule Model & Precision & Recall & $F_1$ Score & Average Precision \\ \midrule \ac{BERT} & $0.45$ & $0.53$ & $0.48$ & $0.46$\\ \ac{DistilBERT} & $0.33$ & $0.71$ & $0.48$ & $0.46$\\ \ac{ERNIE2.0} & $0.35$ & $0.85$ & $0.50$ & $0.43$\\ \bottomrule \end{tabular} \caption[Study Results on Complete Dataset]{The results for the dataset $D_{all_{test}}$.}\label{tab:study:results:second_run} \end{table} vignettes/iEN.bib @article{hercus_2009, title = {The granulocyte-macrophage colony-stimulating factor receptor: linking its structure to cell signaling and its role in disease.}, author = {Hercus, and and Guthridge, and Ekert, and King-Scott, Jack and Parker, and Lopez, }, pages = {1289-1298}, url = {http://dx.doi.org/10.1182/blood-2008-12-164004}, year = {2009}, month = {aug}, day = {13}, urldate = {2018-10-23}, journal = {Blood}, volume = {114}, number = {7}, doi = {10.1182/blood-2008-12-164004}, pmid = {19436055}, pmcid = {PMC2727416}, abstract = {Already 20 years have passed since the cloning of the granulocyte-macrophage colony-stimulating factor ({GM}-{CSF}) receptor alpha-chain, the first member of the {GM}-{CSF}/interleukin ({IL})-3/{IL}-5 family of hemopoietic cytokine receptors to be molecularly characterized. The intervening 2 decades have uncovered a plethora of biologic functions transduced by the {GM}-{CSF} receptor (pleiotropy) and revealed distinct signaling networks that couple the receptor to biologic outcomes. Unlike other hemopoietin receptors, the {GM}-{CSF} receptor has a significant nonredundant role in myeloid hematologic malignancies, macrophage-mediated acute and chronic inflammation, pulmonary homeostasis, and allergic disease. The molecular mechanisms underlying {GM}-{CSF} receptor activation have recently been revealed by the crystal structure of the {GM}-{CSF} receptor complexed to {GM}-{CSF}, which shows an unexpected higher order assembly. Emerging evidence also suggests the existence of intracellular signosomes that are recruited in a concentration-dependent fashion to selectively control cell survival, proliferation, and differentiation by {GM}-{CSF}. These findings begin to unravel the mystery of cytokine receptor pleiotropy and are likely to also apply to the related {IL}-3 and {IL}-5 receptors as well as other heterodimeric cytokine receptors. The new insights in {GM}-{CSF} receptor activation have clinical significance as the structural and signaling nuances can be harnessed for the development of new treatments for malignant and inflammatory diseases.} } @article{gaudillire_2014, title = {Clinical recovery from surgery correlates with single-cell immune signatures.}, author = {Gaudillière, Fragiadakis, and Bruggner, and Nicolau, , , , , , , , Goodman, and Davis, and Bendall, and Fantl, and Angst, and Nolan, }, pages = {255ra131}, url = {http://dx.doi.org/10.1126/scitranslmed.3009701}, year = {2014}, month = {sep}, day = {24}, urldate = {2018-10-23}, journal = {Sci Transl Med}, volume = {6}, number = {255}, doi = {10.1126/scitranslmed.3009701}, pmid = {25253674}, pmcid = {PMC4334126}, abstract = {Delayed recovery from surgery causes personal suffering and substantial societal and economic costs. Whether immune mechanisms determine recovery after surgical trauma remains ill-defined. Single-cell mass cytometry was applied to serial whole-blood samples from 32 patients undergoing hip replacement to comprehensively characterize the phenotypic and functional immune response to surgical trauma. The simultaneous analysis of 14,000 phosphorylation events in precisely phenotyped immune cell subsets revealed uniform signaling responses among patients, demarcating a surgical immune signature. When regressed against clinical parameters of surgical recovery, including functional impairment and pain, strong correlations were found with {STAT3} (signal transducer and activator of transcription), {CREB} (adenosine 3',5'-monophosphate response element-binding protein), and {NF}-{\kappaB} (nuclear factor {\kappaB}) signaling responses in subsets of {CD14}(+) monocytes (R = 0.7 to 0.8, false discovery rate \textless 0.01). These sentinel results demonstrate the capacity of mass cytometry to survey the human immune system in a relevant clinical context. The mechanistically derived immune correlates point to diagnostic signatures, and potential therapeutic targets, that could postoperatively improve patient recovery. Copyright \copyright 2014, American Association for the Advancement of Science.} } @article{friedman_2010, title = {Regularization Paths for Generalized Linear Models via Coordinate Descent.}, author = { }, pages = {1-22}, url = {http://dx.doi.org/10.18637/jss.v033.i01}, year = {2010}, urldate = {2019-01-25}, journal = {J Stat Softw}, volume = {33}, number = {1}, doi = {10.18637/jss.v033.i01}, pmid = {20808728}, pmcid = {PMC2929880}, abstract = {We develop fast algorithms for estimation of generalized linear models with convex penalties. The models include linear regression, two-class logistic regression, and multinomial regression problems while the penalties include ℓ(1) (the lasso), ℓ(2) (ridge regression) and mixtures of the two (the elastic net). The algorithms use cyclical coordinate descent, computed along a regularization path. The methods can handle large problems and can also deal efficiently with sparse features. In comparative timings we find that the new algorithms are considerably faster than competing methods.} } @article{bendall_2011, title = {Single-cell mass cytometry of differential immune and drug responses across a human hematopoietic continuum.}, author = {, , and Finck, Bruggner, and Melamed, , , , , Sachs, , , and Nolan, }, pages = {687-696}, url = {http://dx.doi.org/10.1126/science.1198704}, year = {2011}, month = {may}, day = {6}, urldate = {2018-11-08}, journal = {Science}, volume = {332}, number = {6030}, issn = {1095-9203}, doi = {10.1126/science.1198704}, pmid = {21551058}, pmcid = {PMC3273988}, abstract = {Flow cytometry is an essential tool for dissecting the functional complexity of hematopoiesis. We used single-cell "mass cytometry" to examine healthy human bone marrow, measuring 34 parameters simultaneously in single cells (binding of 31 antibodies, viability, {DNA} content, and relative cell size). The signaling behavior of cell subsets spanning a defined hematopoietic hierarchy was monitored with 18 simultaneous markers of functional signaling states perturbed by a set of ex vivo stimuli and inhibitors. The data set allowed for an algorithmically driven assembly of related cell types defined by surface antigen expression, providing a superimposable map of cell signaling responses in combination with drug inhibition. Visualized in this manner, the analysis revealed previously unappreciated instances of both precise signaling responses that were bounded within conventionally defined cell subsets and more continuous phosphorylation responses that crossed cell population boundaries in unexpected manners yet tracked closely with cellular phenotype. Collectively, such single-cell analyses provide system-wide views of immune signaling in healthy human hematopoiesis, against which drug action and disease can be compared for mechanistic studies and pharmacologic intervention.} } @article{ivashkiv_2014, title = {Regulation of type I interferon responses.}, author = {Ivashkiv, and Donlin, }, pages = {36-49}, url = {http://dx.doi.org/10.1038/nri3581}, year = {2014}, month = {jan}, urldate = {2018-10-22}, journal = {Nat Rev Immunol}, volume = {14}, number = {1}, doi = {10.1038/nri3581}, pmid = {24362405}, pmcid = {PMC4084561}, abstract = {Type I interferons ({IFNs}) activate intracellular antimicrobial programmes and influence the development of innate and adaptive immune responses. Canonical type I {IFN} signalling activates the Janus kinase ({JAK})-signal transducer and activator of transcription ({STAT}) pathway, leading to transcription of {IFN}-stimulated genes ({ISGs}). Host, pathogen and environmental factors regulate the responses of cells to this signalling pathway and thus calibrate host defences while limiting tissue damage and preventing autoimmunity. Here, we summarize the signalling and epigenetic mechanisms that regulate type I {IFN}-induced {STAT} activation and {ISG} transcription and translation. These regulatory mechanisms determine the biological outcomes of type I {IFN} responses and whether pathogens are cleared effectively or chronic infection or autoimmune disease ensues.} } @article{malek_2010, title = {Interleukin-2 receptor signaling: at the interface between tolerance and immunity.}, author = { and }, pages = {153-165}, url = {http://dx.doi.org/10.1016/j.immuni.2010.08.004}, year = {2010}, month = {aug}, day = {27}, urldate = {2018-10-23}, journal = {Immunity}, volume = {33}, number = {2}, doi = {10.1016/j.immuni.2010.08.004}, pmid = {20732639}, pmcid = {PMC2946796}, abstract = {Interleukin-2 receptor ({IL}-{2R}) signaling regulates tolerance and immunity. Here, we review recent work concerning the structure, signaling, and function of the {IL}-{2R}, emphasizing the contribution of {IL}-2 for T cell-dependent activity in vivo. {IL}-{2R} signaling influences two discrete aspects of immune responses by {CD8}(+) T cells, terminal differentiation of effector cells in primary responses, and aspects of memory recall responses. {IL}-2 also delivers essential signals for thymic development of regulatory T (Treg) cells and later to promote their homeostasis and function. Each of these outcomes on T effector and Treg cells requires distinct amounts of {IL}-{2R} signaling, with low {IL}-{2R} signaling sufficient for many key aspects of Treg cells. Thus, tolerance is readily maintained and favored with limited {IL}-2. Copyright 2010 Elsevier Inc. All rights reserved.} } @book{efron_1994, title = {An Introduction to the Bootstrap}, author = { Tibshirani, .}, series = {Chapman \& Hall/{CRC} Monographs on Statistics \& Applied Probability}, publisher = {Chapman and Hall/{CRC}}, url = {http://link.springer.com/10.1007/978-1-4899-4541-9}, year = {1994}, urldate = {2018-10-16}, isbn = {978-0-412-04231-7}, doi = {10.1007/978-1-4899-4541-9}, address = {Boca Raton}, abstract = {Statistics is a subject of many uses and surprisingly few effective practitioners. The traditional road to statistical knowledge is blocked, for most, by a formidable wall of mathematics. The approach in An Introduction to the Bootstrap avoids that wall. It arms scientists and engineers, as well as statisticians, with the computational techniques they need to analyze and understand complicated data sets.} } @article{bandura_2009, title = {Mass cytometry: technique for real time single cell multitarget immunoassay based on inductively coupled plasma time-of-flight mass spectrometry.}, author = {Bandura, and Baranov, and Ornatsky, and Antonov, Alexei and Kinach, , Xudong and Pavlov, Serguei and Vorobiev, Sergey and Dick, and Tanner, }, pages = {6813-6822}, url = {http://dx.doi.org/10.1021/ac901049w}, year = {2009}, month = {aug}, day = {15}, urldate = {2018-11-07}, journal = {Anal Chem}, volume = {81}, number = {16}, doi = {10.1021/ac901049w}, pmid = {19601617}, abstract = {A novel instrument for real time analysis of individual biological cells or other microparticles is described. The instrument is based on inductively coupled plasma time-of-flight mass spectrometry and comprises a three-aperture plasma-vacuum interface, a dc quadrupole turning optics for decoupling ions from neutral components, an rf quadrupole ion guide discriminating against low-mass dominant plasma ions, a point-to-parallel focusing dc quadrupole doublet, an orthogonal acceleration reflectron analyzer, a discrete dynode fast ion detector, and an 8-bit 1 {GHz} digitizer. A high spectrum generation frequency of 76.8 {kHz} provides capability for collecting multiple spectra from each particle-induced transient ion cloud, typically of 200-300 micros duration. It is shown that the transients can be resolved and characterized individually at a peak frequency of 1100 particles per second. Design considerations and optimization data are presented. The figures of merit of the instrument are measured under standard inductively coupled plasma ({ICP}) operating conditions (\textless3\% cerium oxide ratio). At mass resolution (full width at half-maximum) M/{DeltaM} \textgreater 900 for m/z = 159, the sensitivity with a standard sample introduction system of \textgreater1.4 x 10(8) ion counts per second per mg L(-1) of Tb and an abundance sensitivity of (6 x 10(-4))-(1.4 x 10(-3)) (trailing and leading masses, respectively) are shown. The mass range (m/z = 125-215) and abundance sensitivity are sufficient for elemental immunoassay with up to 60 distinct available elemental tags. When \textless15 elemental tags are used, a higher sensitivity mode at lower resolution (M/{DeltaM} \textgreater 500) can be used, which provides \textgreater2.4 x 10(8) cps per mg L(-1) of Tb, at (1.5 x 10(-3))-(5.0 x 10(-3)) abundance sensitivity. The real-time simultaneous detection of multiple isotopes from individual 1.8 microm polystyrene beads labeled with lanthanides is shown. A real time single cell 20 antigen expression assay of model cell lines and leukemia patient samples immuno-labeled with lanthanide-tagged antibodies is presented.} } @article{finck_2013, title = {Normalization of mass cytometry data with bead standards.}, author = { Simonds, Jager, Krishnaswamy, , , , , }, pages = {483-494}, url = {http://dx.doi.org/10.1002/cyto.a.22271}, year = {2013}, month = {may}, urldate = {2017-07-24}, journal = {Cytometry A}, volume = {83}, number = {5}, doi = {10.1002/cyto.a.22271}, pmid = {23512433}, pmcid = {PMC3688049}, f1000-projects = {immuEN}, abstract = {Mass cytometry uses atomic mass spectrometry combined with isotopically pure reporter elements to currently measure as many as 40 parameters per single cell. As with any quantitative technology, there is a fundamental need for quality assurance and normalization protocols. In the case of mass cytometry, the signal variation over time due to changes in instrument performance combined with intervals between scheduled maintenance must be accounted for and then normalized. Here, samples were mixed with polystyrene beads embedded with metal lanthanides, allowing monitoring of mass cytometry instrument performance over multiple days of data acquisition. The protocol described here includes simultaneous measurements of beads and cells on the mass cytometer, subsequent extraction of the bead-based signature, and the application of an algorithm enabling correction of both short- and long-term signal fluctuations. The variation in the intensity of the beads that remains after normalization may also be used to determine data quality. Application of the algorithm to a one-month longitudinal analysis of a human peripheral blood sample reduced the range of median signal fluctuation from 4.9-fold to 1.3-fold. Copyright \copyright 2013 International Society for Advancement of Cytometry.} } @article{zunder_2015, title = {Palladium-based mass tag cell barcoding with a doublet-filtering scheme and single-cell deconvolution algorithm.}, author = {Zunder, Finck, , , Krishnaswamy, , , Bjornson, , and Bodenmiller, , , , }, pages = {316-333}, url = {http://dx.doi.org/10.1038/nprot.2015.020}, year = {2015}, month = {feb}, urldate = {2017-07-24}, journal = {Nat Protoc}, volume = {10}, number = {2}, doi = {10.1038/nprot.2015.020}, pmid = {25612231}, pmcid = {PMC4347881}, f1000-projects = {immuEN}, abstract = {Mass-tag cell barcoding ({MCB}) labels individual cell samples with unique combinatorial barcodes, after which they are pooled for processing and measurement as a single multiplexed sample. The {MCB} method eliminates variability between samples in antibody staining and instrument sensitivity, reduces antibody consumption and shortens instrument measurement time. Here we present an optimized {MCB} protocol. The use of palladium-based labeling reagents expands the number of measurement channels available for mass cytometry and reduces interference with lanthanide-based antibody measurement. An error-detecting combinatorial barcoding scheme allows cell doublets to be identified and removed from the analysis. A debarcoding algorithm that is single cell-based rather than population-based improves the accuracy and efficiency of sample deconvolution. This debarcoding algorithm has been packaged into software that allows rapid and unbiased sample deconvolution. The {MCB} procedure takes 3-4 h, not including sample acquisition time of ∼1 h per million cells.} } @article{krutzik_2003, title = {Intracellular phospho-protein staining techniques for flow cytometry: monitoring single cell signaling events.}, author = {Krutzik, and Nolan, }, pages = {61-70}, url = {http://dx.doi.org/10.1002/cyto.a.10072}, year = {2003}, month = {oct}, urldate = {2018-08-15}, journal = {Cytometry A}, volume = {55}, number = {2}, doi = {10.1002/cyto.a.10072}, pmid = {14505311}, f1000-projects = {immuEN}, abstract = {{BACKGROUND}: Recent advances in intracellular staining techniques, cytometer technology, fluorescent reagents, and antibody production have expanded the number of intracellular antigens that can be analyzed by flow cytometry. Measurement of protein phosphorylation with phospho-specific antibodies has given insight into kinase signaling cascades. However, available techniques for phospho-epitope staining can differ greatly, making it necessary to understand the differences between the outcomes when such techniques are applied and to develop robust and reproducible methods of application. {METHODS}: Ten different cellular fixation and permeabilization techniques were tested for their ability to provide phospho-specific staining. Combinations of formaldehyde, methanol, ethanol, acetone, Triton X-100, and saponin were used as fixation and permeabilization reagents. Phospho-specific antibodies were labeled with Alexa Fluor dyes to provide multicolor analysis of different signaling events simultaneously within individual cells. {RESULTS}: Fixing cells with 1.5\% formaldehyde followed by permeabilization in methanol gave optimal results for {pERK}, pp38, {pJNK}, {pStat1}, {pStat5}, and {pStat6} staining. Alteration of formaldehyde fixation and methanol permeabilization times affected measurements of phosphorylation induction. Phospho-specific flow cytometric analyses correlated well with Western blotting, providing cross platform validation of the technique. {CONCLUSIONS}: Measuring phosphorylation events by flow cytometry provides a rapid and efficient way to measure kinase cascades in individual cells. Stability of phospho-epitopes in methanol allows long-term storage of samples prior to analysis. Multiple signaling cascades can be monitored simultaneously through the use of different fluorophore labels to determine specificity of ligands or inhibitors. Application of optimized techniques to heterogeneous cell types such as peripheral blood or murine splenocytes may allow signaling to be analyzed simultaneously in immune cell subsets. Copyright 2003 Wiley-Liss, Inc.} } @article{hanley_1982, title = {The meaning and use of the area under a receiver operating characteristic ({ROC}) curve.}, author = { and {McNeil}, }, pages = {29-36}, url = {http://dx.doi.org/10.1148/radiology.143.1.7063747}, year = {1982}, month = {apr}, urldate = {2018-08-16}, journal = {Radiology}, volume = {143}, number = {1}, doi = {10.1148/radiology.143.1.7063747}, pmid = {7063747}, f1000-projects = {immuEN}, abstract = {A representation and interpretation of the area under a receiver operating characteristic ({ROC}) curve obtained by the "rating" method, or by mathematical predictions based on patient characteristics, is presented. It is shown that in such a setting the area represents the probability that a randomly chosen diseased subject is (correctly) rated or ranked with greater suspicion than a randomly chosen non-diseased subject. Moreover, this probability of a correct ranking is the same quantity that is estimated by the already well-studied nonparametric Wilcoxon statistic. These two relationships are exploited to (a) provide rapid closed-form expressions for the approximate magnitude of the sampling variability, i.e., standard error that one uses to accompany the area under a smoothed {ROC} curve, (b) guide in determining the size of the sample required to provide a sufficiently reliable estimate of this area, and (c) determine how large sample sizes should be to ensure that one can statistically detect differences in the accuracy of diagnostic techniques.} } @article{gaudillire_2015, title = {Implementing Mass Cytometry at the Bedside to Study the Immunological Basis of Human Diseases: Distinctive Immune Features in Patients with a History of Term or Preterm Birth.}, author = {, , , Fragiadakis, , Aghaeepour, , , Cele and El-Sayed, , and Lewis, and Stevenson, and Nolan, and Angst, }, pages = {817-829}, url = {http://dx.doi.org/10.1002/cyto.a.22720}, year = {2015}, month = {sep}, urldate = {2018-08-16}, journal = {Cytometry A}, volume = {87}, number = {9}, doi = {10.1002/cyto.a.22720}, pmid = {26190063}, pmcid = {PMC4758855}, f1000-projects = {immuEN}, abstract = {Single-cell technologies have immense potential to shed light on molecular and biological processes that drive human diseases. Mass cytometry (or Cytometry by Time Of Flight mass spectrometry, {CyTOF}) has already been employed in clinical studies to comprehensively survey patients' circulating immune system. As interest in the "bedside" application of mass cytometry is growing, the delineation of relevant methodological issues is called for. This report uses a newly generated dataset to discuss important methodological considerations when mass cytometry is implemented in a clinical study. Specifically, the use of whole blood samples versus peripheral blood mononuclear cells ({PBMCs}), design of mass-tagged antibody panels, technical and analytical implications of sample barcoding, and application of traditional and unsupervised approaches to analyze high-dimensional mass cytometry datasets are discussed. A mass cytometry assay was implemented in a cross-sectional study of 19 women with a history of term or preterm birth to determine whether immune traits in peripheral blood differentiate the two groups in the absence of pregnancy. Twenty-seven phenotypic and 11 intracellular markers were simultaneously analyzed in whole blood samples stimulated with lipopolysaccharide ({LPS} at 0, 0.1, 1, 10, and 100 ng {mL}(-1)) to examine dose-dependent signaling responses within the toll-like receptor 4 ({TLR4}) pathway. Complementary analyses, grounded in traditional or unsupervised gating strategies of immune cell subsets, indicated that the {prpS6} and {pMAPKAPK2} responses in classical monocytes are accentuated in women with a history of preterm birth ({FDR\textless} 1\%). The results suggest that women predisposed to preterm birth may be prone to mount an exacerbated {TLR4} response during the course of pregnancy. This important hypothesis-generating finding points to the power of single-cell mass cytometry to detect biologically important differences in a relatively small patient cohort. \copyright 2015 International Society for Advancement of Cytometry.} } @article{porter_2011, title = {Chimeric antigen receptor-modified T cells in chronic lymphoid leukemia.}, author = {Porter, and Levine, and Kalos, Bagg, , }, pages = {725-733}, url = {http://dx.doi.org/10.1056/{NEJMoa1103849}}, year = {2011}, month = {aug}, day = {25}, urldate = {2019-02-05}, journal = {N Engl J Med}, volume = {365}, number = {8}, issn = {1533-4406}, doi = {10.1056/{NEJMoa1103849}}, pmid = {21830940}, pmcid = {PMC3387277}, abstract = {We designed a lentiviral vector expressing a chimeric antigen receptor with specificity for the B-cell antigen {CD19}, coupled with {CD137} (a costimulatory receptor in T cells [4-{1BB}]) and {CD3}-zeta (a signal-transduction component of the T-cell antigen receptor) signaling domains. A low dose (approximately 1.5×10(5) cells per kilogram of body weight) of autologous chimeric antigen receptor-modified T cells reinfused into a patient with refractory chronic lymphocytic leukemia ({CLL}) expanded to a level that was more than 1000 times as high as the initial engraftment level in vivo, with delayed development of the tumor lysis syndrome and with complete remission. Apart from the tumor lysis syndrome, the only other grade 3/4 toxic effect related to chimeric antigen receptor T cells was lymphopenia. Engineered cells persisted at high levels for 6 months in the blood and bone marrow and continued to express the chimeric antigen receptor. A specific immune response was detected in the bone marrow, accompanied by loss of normal B cells and leukemia cells that express {CD19}. Remission was ongoing 10 months after treatment. Hypogammaglobulinemia was an expected chronic toxic effect.} } @book{vapnik_2000, title = {The Nature of Statistical Learning Theory}, author = {Vapnik, .}, publisher = {Springer New York}, url = {http://link.springer.com/10.1007/978-1-4757-3264-1}, year = {2000}, urldate = {2018-11-26}, isbn = {978-1-4419-3160-3}, doi = {10.1007/978-1-4757-3264-1}, address = {New York, {NY}}, abstract = {Setting of the learning problem consistency of learning processes bounds on the rate of convergence of learning processes controlling the generalization ability of learning processes constructing learning algorithms what is important in learning theory?.} } @article{arck_2013, title = {Fetomaternal immune cross-talk and its consequences for maternal and offspring's health.}, author = { and }, pages = {548-556}, url = {http://dx.doi.org/10.1038/nm.3160}, year = {2013}, month = {may}, day = {7}, urldate = {2018-12-05}, journal = {Nat Med}, volume = {19}, number = {5}, doi = {10.1038/nm.3160}, pmid = {23652115}, abstract = {An improved mechanistic understanding of the adaptational processes mounted during mammalian reproduction is emerging. Intricate pathways occurring at the fetomaternal interface, such as the formation of a functional synapse between invading fetal trophoblast cells, and the involvement of various maternal immune cell subsets and epigenetically modified decidual stromal cells have now been identified. These complex pathways synergistically create a tolerogenic niche in which the semiallogeneic fetus can develop. New insights into fetomaternal immune cross-talk may help us to understand the pathogenesis of pregnancy complications as well as poor postnatal health. Moreover, the effects of maternal immune adaptation to pregnancy on autoimmune disease activity are becoming increasingly evident. Thus, insights into fetomaternal immune cross-talk not only advance our understanding of pregnancy-related complications but also may be informative on how immune tolerance can be modulated in clinical settings outside the context of reproduction.} } @article{romero_2014, title = {Preterm labor: one syndrome, many causes.}, author = { }, pages = {760-765}, url = {http://dx.doi.org/10.1126/science.1251816}, year = {2014}, month = {aug}, day = {15}, urldate = {2018-12-05}, journal = {Science}, volume = {345}, number = {6198}, doi = {10.1126/science.1251816}, pmid = {25124429}, pmcid = {PMC4191866}, abstract = {Preterm birth is associated with 5 to 18\% of pregnancies and is a leading cause of infant morbidity and mortality. Spontaneous preterm labor, a syndrome caused by multiple pathologic processes, leads to 70\% of preterm births. The prevention and the treatment of preterm labor have been long-standing challenges. We summarize the current understanding of the mechanisms of disease implicated in this condition and review advances relevant to intra-amniotic infection, decidual senescence, and breakdown of maternal-fetal tolerance. The success of progestogen treatment to prevent preterm birth in a subset of patients at risk is a cause for optimism. Solving the mystery of preterm labor, which compromises the health of future generations, is a formidable scientific challenge worthy of investment. Copyright \copyright 2014, American Association for the Advancement of Science.} } @article{lu_2008, title = {{LPS}/{TLR4} signal transduction pathway.}, author = { and and }, pages = {145-151}, url = {http://dx.doi.org/10.1016/j.cyto.2008.01.006}, year = {2008}, month = {may}, urldate = {2018-10-22}, journal = {Cytokine}, volume = {42}, number = {2}, doi = {10.1016/j.cyto.2008.01.006}, pmid = {18304834}, abstract = {The stimulation of Toll-like receptor 4 ({TLR4}) by lipopolysaccharide ({LPS}) induces the release of critical proinflammatory cytokines that are necessary to activate potent immune responses. {LPS}/{TLR4} signaling has been intensively studied in the past few years. Here we review molecules involved in {TLR4}-mediated signaling, including players that are involved in the negative regulation of this important pathway.} } @article{beutler_2009, title = {{TLRs} and innate immunity.}, author = {Beutler, }, pages = {1399-1407}, url = {http://dx.doi.org/10.1182/blood-2008-07-019307}, year = {2009}, month = {feb}, day = {12}, urldate = {2019-03-15}, journal = {Blood}, volume = {113}, number = {7}, doi = {10.1182/blood-2008-07-019307}, pmid = {18757776}, pmcid = {PMC2644070}, abstract = {One of the most fundamental questions in immunology pertains to the recognition of non-self, which for the most part means microbes. How do we initially realize that we have been inoculated with microbes, and how is the immune response ignited? Genetic studies have made important inroads into this question during the past decade, and we now know that in mammals, a relatively small number of receptors operate to detect signature molecules that herald infection. One or more of these signature molecules are displayed by almost all microbes. These receptors and the signals they initiate have been studied in depth by random germline mutagenesis and positional cloning (forward genetics). Herein is a concise description of what has been learned about the Toll-like receptors, which play an essential part in the perception of microbes and shape the complex host responses that occur during infection.} } @article{sabio_2014, title = {{TNF} and {MAP} kinase signalling pathways.}, author = { and }, pages = {237-245}, url = {http://dx.doi.org/10.1016/j.smim.2014.02.009}, year = {2014}, month = {jun}, urldate = {2018-10-22}, journal = {Semin Immunol}, volume = {26}, number = {3}, doi = {10.1016/j.smim.2014.02.009}, pmid = {24647229}, pmcid = {PMC4099309}, abstract = {The binding of tumour necrosis factor \alpha ({TNF\alpha}) to cell surface receptors engages multiple signal transduction pathways, including three groups of mitogen-activated protein ({MAP}) kinases: extracellular-signal-regulated kinases ({ERKs}); the {cJun} {NH2}-terminal kinases ({JNKs}); and the p38 {MAP} kinases. These {MAP} kinase signalling pathways induce a secondary response by increasing the expression of several inflammatory cytokines (including {TNF\alpha}) that contribute to the biological activity of {TNF\alpha}. {MAP} kinases therefore function both upstream and down-stream of signalling by {TNF\alpha} receptors. Here we review mechanisms that mediate these actions of {MAP} kinases during the response to {TNF\alpha}. Copyright \copyright 2014 Elsevier Ltd. All rights reserved.} } @article{heinrich_2003, title = {Principles of interleukin ({IL})-6-type cytokine signalling and its regulation.}, author = { and Behrmann, Haan, , , , Fred}, pages = {1-20}, url = {http://dx.doi.org/10.1042/{BJ20030407}}, year = {2003}, month = {aug}, day = {15}, urldate = {2018-10-23}, journal = {Biochem J}, volume = {374}, number = {Pt 1}, doi = {10.1042/{BJ20030407}}, pmid = {12773095}, pmcid = {PMC1223585}, abstract = {The {IL} (interleukin)-6-type cytokines {IL}-6, {IL}-11, {LIF} (leukaemia inhibitory factor), {OSM} (oncostatin M), ciliary neurotrophic factor, cardiotrophin-1 and cardiotrophin-like cytokine are an important family of mediators involved in the regulation of the acute-phase response to injury and infection. Besides their functions in inflammation and the immune response, these cytokines play also a crucial role in haematopoiesis, liver and neuronal regeneration, embryonal development and fertility. Dysregulation of {IL}-6-type cytokine signalling contributes to the onset and maintenance of several diseases, such as rheumatoid arthritis, inflammatory bowel disease, osteoporosis, multiple sclerosis and various types of cancer (e.g. multiple myeloma and prostate cancer). {IL}-6-type cytokines exert their action via the signal transducers gp (glycoprotein) 130, {LIF} receptor and {OSM} receptor leading to the activation of the {JAK}/{STAT} (Janus kinase/signal transducer and activator of transcription) and {MAPK} (mitogen-activated protein kinase) cascades. This review focuses on recent progress in the understanding of the molecular mechanisms of {IL}-6-type cytokine signal transduction. Emphasis is put on the termination and modulation of the {JAK}/{STAT} signalling pathway mediated by tyrosine phosphatases, the {SOCS} (suppressor of cytokine signalling) feedback inhibitors and {PIAS} (protein inhibitor of activated {STAT}) proteins. Also the cross-talk between the {JAK}/{STAT} pathway with other signalling cascades is discussed.} } @article{finak_2016, title = {Standardizing Flow Cytometry Immunophenotyping Analysis from the Human {ImmunoPhenotyping} Consortium.}, author = {Finak, , , , , , Devine, , Gerlinde and Pekalski, and Pontikos, , , , , , , , and Brandes, Aaron and Ramey, John and Aghaeepour, Nima and Mosmann, Tim and Scheuermann, and Reed, Elaine and Palucka, Karolina and Pascual, Virginia and Blomberg, and Nestle, , and Brinkman, and Gottardo, Raphael and Maecker, Holden and {McCoy}, J Philip}, pages = {20686}, url = {http://dx.doi.org/10.1038/srep20686}, year = {2016}, month = {feb}, day = {10}, urldate = {2018-11-07}, journal = {Sci Rep}, volume = {6}, doi = {10.1038/srep20686}, pmid = {26861911}, pmcid = {PMC4748244}, abstract = {Standardization of immunophenotyping requires careful attention to reagents, sample handling, instrument setup, and data analysis, and is essential for successful cross-study and cross-center comparison of data. Experts developed five standardized, eight-color panels for identification of major immune cell subsets in peripheral blood. These were produced as pre-configured, lyophilized, reagents in 96-well plates. We present the results of a coordinated analysis of samples across nine laboratories using these panels with standardized operating procedures ({SOPs}). Manual gating was performed by each site and by a central site. Automated gating algorithms were developed and tested by the {FlowCAP} consortium. Centralized manual gating can reduce cross-center variability, and we sought to determine whether automated methods could streamline and standardize the analysis. Within-site variability was low in all experiments, but cross-site variability was lower when central analysis was performed in comparison with site-specific analysis. It was also lower for clearly defined cell subsets than those based on dim markers and for rare populations. Automated gating was able to match the performance of central manual analysis for all tested panels, exhibiting little to no bias and comparable variability. Standardized staining, data collection, and automated gating can increase power, reduce variability, and streamline analysis for immunophenotyping.} } @article{tai_2007, title = {Incorporating prior knowledge of gene functional groups into regularized discriminant analysis of microarray data.}, author = { and }, pages = {3170-3177}, url = {http://dx.doi.org/10.1093/bioinformatics/btm488}, year = {2007}, month = {dec}, day = {1}, urldate = {2019-01-16}, journal = {Bioinformatics}, volume = {23}, number = {23}, doi = {10.1093/bioinformatics/btm488}, pmid = {17933851}, abstract = {{MOTIVATION}: Discriminant analysis for high-dimensional and low-sample-sized data has become a hot research topic in bioinformatics, mainly motivated by its importance and challenge in applications to tumor classifications for high-dimensional microarray data. Two of the popular methods are the nearest shrunken centroids, also called predictive analysis of microarray ({PAM}), and shrunken centroids regularized discriminant analysis ({SCRDA}). Both methods are modifications to the classic linear discriminant analysis ({LDA}) in two aspects tailored to high-dimensional and low-sample-sized data: one is the regularization of the covariance matrix, and the other is variable selection through shrinkage. In spite of their usefulness, there are potential limitations with each method. The main concern is that both {PAM} and {SCRDA} are possibly too extreme: the covariance matrix in the former is restricted to be diagonal while in the latter there is barely any restriction. Based on the biology of gene functions and given the feature of the data, it may be beneficial to estimate the covariance matrix as an intermediate between the two; furthermore, more effective shrinkage schemes may be possible. {RESULTS}: We propose modified {LDA} methods to integrate biological knowledge of gene functions (or variable groups) into classification of microarray data. Instead of simply treating all the genes independently or imposing no restriction on the correlations among the genes, we group the genes according to their biological functions extracted from existing biological knowledge or data, and propose regularized covariance estimators that encourages between-group gene independence and within-group gene correlations while maintaining the flexibility of any general covariance structure. Furthermore, we propose a shrinkage scheme on groups of genes that tends to retain or remove a whole group of the genes altogether, in contrast to the standard shrinkage on individual genes. We show that one of the proposed methods performed better than {PAM} and {SCRDA} in a simulation study and several real data examples.} } @article{pihlstrom_2005, title = {Periodontal diseases.}, author = {Pihlstrom, and Michalowicz, and Johnson, }, pages = {1809-1820}, url = {http://dx.doi.org/10.1016/S0140-6736(05)67728-8}, year = {2005}, month = {nov}, day = {19}, urldate = {2005-11-25}, journal = {Lancet}, volume = {366}, number = {9499}, doi = {10.1016/S0140-6736(05)67728-8}, pmid = {16298220}, abstract = {The periodontal diseases are highly prevalent and can affect up to 90\% of the worldwide population. Gingivitis, the mildest form of periodontal disease, is caused by the bacterial biofilm (dental plaque) that accumulates on teeth adjacent to the gingiva (gums). However, gingivitis does not affect the underlying supporting structures of the teeth and is reversible. Periodontitis results in loss of connective tissue and bone support and is a major cause of tooth loss in adults. In addition to pathogenic microorganisms in the biofilm, genetic and environmental factors, especially tobacco use, contribute to the cause of these diseases. Genetic, dermatological, haematological, granulomatous, immunosuppressive, and neoplastic disorders can also have periodontal manifestations. Common forms of periodontal disease have been associated with adverse pregnancy outcomes, cardiovascular disease, stroke, pulmonary disease, and diabetes, but the causal relations have not been established. Prevention and treatment are aimed at controlling the bacterial biofilm and other risk factors, arresting progressive disease, and restoring lost tooth support.} } @article{newell_2016, title = {Mass cytometry: blessed with the curse of dimensionality.}, author = {Newell, and }, pages = {890-895}, url = {http://www.nature.com/doifinder/10.1038/ni.3485}, year = {2016}, month = {jul}, day = {19}, urldate = {2018-08-16}, journal = {Nat Immunol}, volume = {17}, number = {8}, issn = {1529-2908}, doi = {10.1038/ni.3485}, pmid = {27434000}, f1000-projects = {immuEN}, abstract = {Immunologists are being compelled to develop new high-dimensional perspectives of cellular heterogeneity and to determine which applications best exploit the power of mass cytometry and associated multiplex approaches.} } @article{crawford_2014, title = {Concomitant evaluation of {PMA}+ionomycin-induced kinase phosphorylation and cytokine production in T cell subsets by flow cytometry.}, author = {Crawford, and Jalbert, Emilie and Ndhlovu, and Barbour, }, pages = {268-276}, url = {http://dx.doi.org/10.1002/cyto.a.22444}, year = {2014}, month = {mar}, urldate = {2018-10-23}, journal = {Cytometry A}, volume = {85}, number = {3}, doi = {10.1002/cyto.a.22444}, pmid = {24464647}, abstract = {Methods to detect intracellular kinase signaling intermediates by flow cytometry have been recently developed. Termed "phospho-flow," these methods employ fluorescence-conjugated monoclonal antibodies that recognize phosphorylated epitopes of intracellular kinases, and may be combined with surface phenotypic markers to observe changes in kinase pathways by cellular subset. Effector functions, like cytokine production, are processes intrinsically linked to intracellular signaling and kinase activity within each cell. Methodologies that would simultaneously detect changes to signaling pathways as well as effector responses at the single-cell level would allow for mapping of the functional consequences induced by signaling pathway modifications. However, there are challenges to developing such a combined protocol, relating to the different kinetics of rapid signaling events and the more prolonged time required to induce and observe cytokine responses. In this report, we describe the development of an assay that accommodates differences in protocol conditions and response kinetics, merging phospho-flow cytometry, and intracellular cytokine staining methods into a single experimental protocol. We examined intracellular {ERK1}/2 phosphorylation and {IFN}-\gamma production by {CD4}+ and {CD8}+ T cells upon polyclonal stimulation with {PMA} and ionomycin, while monitoring expression of the cytolytic molecule perforin and the T cell activation marker {CD38}. We present a method that allows observation of kinase phosphorylation and cytokine production within the same cell after stimuli, while maintaining a stable cellular phenotype. Monitoring of signaling and effector functions in distinct immune subsets provides a platform to investigate and relate intracellular kinase signaling activity to immune cell effector function and phenotype in disease states.\copyright 2014 International Society for Advancement of Cytometry.} } @article{wolpert_1997, title = {No free lunch theorems for optimization}, author = { .}, pages = {67-82}, url = {http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=585893}, year = {1997}, month = {apr}, urldate = {2018-08-15}, journal = {{IEEE} Trans. Evol. Computat.}, volume = {1}, number = {1}, issn = {1089778X}, doi = {10.1109/4235.585893}, f1000-projects = {immuEN} } @article{eke_2015, title = {Update on Prevalence of Periodontitis in Adults in the United States: {NHANES} 2009 to 2012.}, author = { and Dye, and and Slade, and Thornton-Evans, and Borgnakke, and Taylor, , and Genco, }, pages = {611-622}, url = {http://dx.doi.org/10.1902/jop.2015.140520}, year = {2015}, month = {may}, urldate = {2018-09-14}, journal = {J Periodontol}, volume = {86}, number = {5}, doi = {10.1902/jop.2015.140520}, pmid = {25688694}, pmcid = {PMC4460825}, f1000-projects = {immuEN}, abstract = {{BACKGROUND}: This report describes prevalence, severity, and extent of periodontitis in the {US} adult population using combined data from the 2009 to 2010 and 2011 to 2012 cycles of the National Health and Nutrition Examination Survey ({NHANES}). {METHODS}: Estimates were derived for dentate adults, aged ≥30 years, from the {US} civilian non-institutionalized population. Periodontitis was defined by combinations of clinical attachment loss ({AL}) and periodontal probing depth ({PD}) from six sites per tooth on all teeth, except third molars, using standard surveillance case definitions. For the first time in {NHANES} history, sufficient numbers of non-Hispanic Asians were sampled in 2011 to 2012 to provide reliable estimates of their periodontitis prevalence. {RESULTS}: In 2009 to 2012, 46\% of {US} adults, representing 64.7 million people, had periodontitis, with 8.9\% having severe periodontitis. Overall, 3.8\% of all periodontal sites (10.6\% of all teeth) had {PD} ≥4 mm, and 19.3\% of sites (37.4\% teeth) had {AL} ≥3 mm. Periodontitis prevalence was positively associated with increasing age and was higher among males. Periodontitis prevalence was highest in Hispanics (63.5\%) and non-Hispanic blacks (59.1\%), followed by non-Hispanic Asian Americans (50.0\%), and lowest in non-Hispanic whites (40.8\%). Prevalence varied two-fold between the lowest and highest levels of socioeconomic status, whether defined by poverty or education. {CONCLUSIONS}: This study confirms a high prevalence of periodontitis in {US} adults aged ≥30 years, with almost fifty-percent affected. The prevalence was greater in non-Hispanic Asians than non-Hispanic whites, although lower than other minorities. The distribution provides valuable information for population-based action to prevent or manage periodontitis in {US} adults.} } @article{blencowe_2012, title = {National, regional, and worldwide estimates of preterm birth rates in the year 2010 with time trends since 1990 for selected countries: a systematic analysis and implications.}, author = { and Cousens, , , Doris and Moller, Ann-Beth and , Alma and , Claudia and and and Lawn, }, pages = {2162-2172}, url = {http://dx.doi.org/10.1016/S0140-6736(12)60820-4}, year = {2012}, month = {jun}, day = {9}, urldate = {2018-10-17}, journal = {Lancet}, volume = {379}, number = {9832}, doi = {10.1016/S0140-6736(12)60820-4}, pmid = {22682464}, abstract = {{BACKGROUND}: Preterm birth is the second largest direct cause of child deaths in children younger than 5 years. Yet, data regarding preterm birth (\textless 37 completed weeks of gestation) are not routinely collected by {UN} agencies, and no systematic country estimates nor time trend analyses have been done. We report worldwide, regional, and national estimates of preterm birth rates for 184 countries in 2010 with time trends for selected countries, and provide a quantitative assessment of the uncertainty surrounding these estimates. {METHODS}: We assessed various data sources according to prespecified inclusion criteria. National Registries (563 datapoints, 51 countries), Reproductive Health Surveys (13 datapoints, eight countries), and studies identified through systematic searches and unpublished data (162 datapoints, 40 countries) were included. 55 countries submitted additional data during {WHO}'s country consultation process. For 13 countries with adequate quality and quantity of data, we estimated preterm birth rates using country-level loess regression for 2010. For 171 countries, two regional multilevel statistical models were developed to estimate preterm birth rates for 2010. We estimated time trends from 1990 to 2010 for 65 countries with reliable time trend data and more than 10,000 livebirths per year. We calculated uncertainty ranges for all countries. {FINDINGS}: In 2010, an estimated 14·9 million babies (uncertainty range 12·3-18·1 million) were born preterm, 11·1\% of all livebirths worldwide, ranging from about 5\% in several European countries to 18\% in some African countries. More than 60\% of preterm babies were born in south Asia and sub-Saharan Africa, where 52\% of the global livebirths occur. Preterm birth also affects rich countries, for example, {USA} has high rates and is one of the ten countries with the highest numbers of preterm births. Of the 65 countries with estimated time trends, only three (Croatia, Ecuador, and Estonia), had reduced preterm birth rates 1990-2010. {INTERPRETATION}: The burden of preterm birth is substantial and is increasing in those regions with reliable data. Improved recording of all pregnancy outcomes and standard application of preterm definitions is important. We recommend the addition of a data-quality indicator of the per cent of all live preterm births that are under 28 weeks' gestation. Distinguishing preterm births that are spontaneous from those that are provider-initiated is important to monitor trends associated with increased caesarean sections. Rapid scale up of basic interventions could accelerate progress towards Millennium Development Goal 4 for child survival and beyond. {FUNDING}: Bill \& Melinda Gates Foundation through grants to Child Health Epidemiology Reference Group ({CHERG}) and Save the Children's Saving Newborn Lives programme; March of Dimes; the Partnership for Maternal Newborn and Childe Health; and {WHO}, Department of Reproductive Health and Research. Copyright \copyright 2012 Elsevier Ltd. All rights reserved.} } @article{liu_2016, title = {Global, regional, and national causes of under-5 mortality in 2000-15: an updated systematic analysis with implications for the Sustainable Development Goals.}, author = { Oza, Shefali and and Chu, Yue and Perin, Jamie and and Lawn, and Cousens, Simon and Mathers, Colin and }, pages = {3027-3035}, url = {http://dx.doi.org/10.1016/S0140-6736(16)31593-8}, year = {2016}, month = {dec}, day = {17}, urldate = {2018-10-17}, journal = {Lancet}, volume = {388}, number = {10063}, doi = {10.1016/S0140-6736(16)31593-8}, pmid = {27839855}, pmcid = {PMC5161777}, abstract = {{BACKGROUND}: Despite remarkable progress in the improvement of child survival between 1990 and 2015, the Millennium Development Goal ({MDG}) 4 target of a two-thirds reduction of under-5 mortality rate ({U5MR}) was not achieved globally. In this paper, we updated our annual estimates of child mortality by cause to 2000-15 to reflect on progress toward the {MDG} 4 and consider implications for the Sustainable Development Goals ({SDG}) target for child survival. {METHODS}: We increased the estimation input data for causes of deaths by 43\% among neonates and 23\% among 1-59-month-olds, respectively. We used adequate vital registration ({VR}) data where available, and modelled cause-specific mortality fractions applying multinomial logistic regressions using adequate {VR} for low {U5MR} countries and verbal autopsy data for high {U5MR} countries. We updated the estimation to use Plasmodium falciparum parasite rate in place of malaria index in the modelling of malaria deaths; to use adjusted empirical estimates instead of modelled estimates for China; and to consider the effects of pneumococcal conjugate vaccine and rotavirus vaccine in the estimation. {FINDINGS}: In 2015, among the 5·9 million under-5 deaths, 2·7 million occurred in the neonatal period. The leading under-5 causes were preterm birth complications (1·055 million [95\% uncertainty range ({UR}) 0·935-1·179]), pneumonia (0·921 million [0·812 -1·117]), and intrapartum-related events (0·691 million [0·598 -0·778]). In the two {MDG} regions with the most under-5 deaths, the leading cause was pneumonia in sub-Saharan Africa and preterm birth complications in southern Asia. Reductions in mortality rates for pneumonia, diarrhoea, neonatal intrapartum-related events, malaria, and measles were responsible for 61\% of the total reduction of 35 per 1000 livebirths in {U5MR} in 2000-15. Stratified by {U5MR}, pneumonia was the leading cause in countries with very high {U5MR}. Preterm birth complications and pneumonia were both important in high, medium high, and medium child mortality countries; whereas congenital abnormalities was the most important cause in countries with low and very low {U5MR}. {INTERPRETATION}: In the {SDG} era, countries are advised to prioritise child survival policy and programmes based on their child cause-of-death composition. Continued and enhanced efforts to scale up proven life-saving interventions are needed to achieve the {SDG} child survival target. {FUNDING}: Bill \& Melinda Gates Foundation, {WHO}. Copyright \copyright 2016 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the {CC} {BY} license. Published by Elsevier Ltd.. All rights reserved.} } @article{zieglerheitbrock_1993, title = {{CD14}: cell surface receptor and differentiation marker.}, author = {Ziegler-Heitbrock, and Ulevitch, }, pages = {121-125}, url = {http://dx.doi.org/10.1016/0167-5699(93)90212-4}, year = {1993}, month = {mar}, urldate = {2018-11-08}, journal = {Immunol Today}, volume = {14}, number = {3}, doi = {10.1016/0167-5699(93)90212-4}, pmid = {7682078}, abstract = {In the past, {CD14} has been viewed simply as a useful marker molecule for monocytes and macrophages. Now, new findings on its role in binding of {LPS}-{LBP} complexes and in signal transduction have engendered renewed interest in the properties of {CD14}. Here, {CD14} function, its expression in different cell types and the regulation of expression, including the generation of soluble {CD14}, are described, and the diagnostic value of {CD14} in various diseases is discussed.} } @article{rieckmann_2017, title = {Social network architecture of human immune cells unveiled by quantitative proteomics.}, author = {Rieckmann, and , , , , , , , , , Felix}, pages = {583-593}, url = {http://dx.doi.org/10.1038/ni.3693}, year = {2017}, month = {mar}, day = {6}, urldate = {2018-12-05}, journal = {Nat Immunol}, volume = {18}, number = {5}, doi = {10.1038/ni.3693}, pmid = {28263321}, abstract = {The immune system is unique in its dynamic interplay between numerous cell types. However, a system-wide view of how immune cells communicate to protect against disease has not yet been established. We applied high-resolution mass-spectrometry-based proteomics to characterize 28 primary human hematopoietic cell populations in steady and activated states at a depth of \textgreater10,000 proteins in total. Protein copy numbers revealed a specialization of immune cells for ligand and receptor expression, thereby connecting distinct immune functions. By integrating total and secreted proteomes, we discovered fundamental intercellular communication structures and previously unknown connections between cell types. Our publicly accessible (http://www.immprot.org/) proteomic resource provides a framework for the orchestration of cellular interplay and a reference for altered communication associated with pathology.} } @article{davis_2017, title = {Systems immunology: just getting started.}, author = { and Tato, and }, pages = {725-732}, url = {http://dx.doi.org/10.1038/ni.3768}, year = {2017}, month = {jun}, day = {20}, urldate = {2018-08-17}, journal = {Nat Immunol}, volume = {18}, number = {7}, doi = {10.1038/ni.3768}, pmid = {28632713}, pmcid = {PMC5790187}, f1000-projects = {immuEN}, abstract = {Systems-biology approaches in immunology take various forms, but here we review strategies for measuring a broad swath of immunological functions as a means of discovering previously unknown relationships and phenomena and as a powerful way of understanding the immune system as a whole. This approach has rejuvenated the field of vaccine development and has fostered hope that new ways will be found to combat infectious diseases that have proven refractory to classical approaches. Systems immunology also presents an important new strategy for understanding human immunity directly, taking advantage of the many ways the immune system of humans can be manipulated.} } @article{aghaeepour_2017, title = {An immune clock of human pregnancy.}, author = {, Mcilwain, , , , , , Quentin and {McNeil}, Leslie and Okada, , and Furman, , , , and El-Sayed, and Quaintance, Cecele and Gibbs, Ronald and Darmstadt, and Shaw, and Stevenson, and Tibshirani, Robert and Nolan, and Lewis, and Angst, and Gaudilliere, Brice}, url = {http://dx.doi.org/10.1126/sciimmunol.aan2946}, year = {2017}, month = {sep}, day = {1}, urldate = {2018-08-10}, journal = {Sci Immunol}, volume = {2}, number = {15}, doi = {10.1126/sciimmunol.aan2946}, pmid = {28864494}, pmcid = {PMC5701281}, f1000-projects = {immuEN}, abstract = {The maintenance of pregnancy relies on finely tuned immune adaptations. We demonstrate that these adaptations are precisely timed, reflecting an immune clock of pregnancy in women delivering at term. Using mass cytometry, the abundance and functional responses of all major immune cell subsets were quantified in serial blood samples collected throughout pregnancy. Cell signaling-based Elastic Net, a regularized regression method adapted from the elastic net algorithm, was developed to infer and prospectively validate a predictive model of interrelated immune events that accurately captures the chronology of pregnancy. Model components highlighted existing knowledge and revealed previously unreported biology, including a critical role for the interleukin-2-dependent {STAT5ab} signaling pathway in modulating T cell function during pregnancy. These findings unravel the precise timing of immunological events occurring during a term pregnancy and provide the analytical framework to identify immunological deviations implicated in pregnancy-related pathologies. Copyright \copyright 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License 4.0 ({CC} {BY}).} } @article{jain_2000, title = {Statistical pattern recognition: a review}, author = {. and . and ,}, pages = {4-37}, url = {http://ieeexplore.ieee.org/document/824819/}, year = {2000}, urldate = {2018-12-04}, journal = {{IEEE} Trans Pattern Anal Mach Intell}, volume = {22}, number = {1}, issn = {01628828}, doi = {10.1109/34.824819} } @article{bakker_2018, title = {Integration of multi-omics data and deep phenotyping enables prediction of cytokine responses.}, author = {Bakker, and Aguirre-Gamboa, Raul and Sanna, Serena and Oosting, Marije and Smeekens, and Jaeger, Martin and Zorro, Maria and Võsa, Urmo and Withoff, Sebo and Netea-Maier, and Koenen, and Joosten, Irma and and Franke, Lude and Joosten, and Kumar, Vinod and Wijmenga, Cisca and Netea, and }, pages = {776-786}, url = {http://www.nature.com/articles/s41590-018-0121-3}, year = {2018}, month = {jul}, urldate = {2018-08-17}, journal = {Nat Immunol}, volume = {19}, number = {7}, issn = {1529-2908}, doi = {10.1038/s41590-018-0121-3}, pmid = {29784908}, pmcid = {PMC6022810}, f1000-projects = {immuEN}, abstract = {The immune response to pathogens varies substantially among people. Whereas both genetic and nongenetic factors contribute to interperson variation, their relative contributions and potential predictive power have remained largely unknown. By systematically correlating host factors in 534 healthy volunteers, including baseline immunological parameters and molecular profiles (genome, metabolome and gut microbiome), with cytokine production after stimulation with 20 pathogens, we identified distinct patterns of co-regulation. Among the 91 different cytokine-stimulus pairs, 11 categories of host factors together explained up to 67\% of interindividual variation in cytokine production induced by stimulation. A computational model based on genetic data predicted the genetic component of stimulus-induced cytokine production (correlation 0.28-0.89), and nongenetic factors influenced cytokine production as well.} } @article{ghaemi_2019, title = {Multiomics modeling of the immunome, transcriptome, microbiome, proteome and metabolome adaptations during human pregnancy.}, author = {Ghaemi, and {DiGiulio}, and Contrepois, Kévin and Callahan, Benjamin and Ngo, and Lee-{McMullen}, Brittany and Lehallier, Benoit and Robaczewska, Anna and Mcilwain, David and Rosenberg-Hasson, , , , Anthony and Stanley, Natalie and Tanada, Athena and Tsai, Amy and Gaudilliere, Dyani and Ganio, and and {McNeil}, Leslie and Tingle, Martha and Wise, Paul and Maric, Ivana and Sirota, Marina and Wyss-Coray, Tony and Winn, and Druzin, and Gibbs, Ronald and Darmstadt, and Lewis, and , Vahid and Agard, Bruno and Tibshirani, Robert and Nolan, Garry and Snyder, and Relman, and Quake, and Shaw, and Stevenson, and Angst, and Gaudilliere, Brice and Aghaeepour, Nima}, pages = {95-103}, url = {https://academic.oup.com/bioinformatics/advance-article/doi/10.1093/bioinformatics/bty537/5047759}, year = {2019}, month = {jan}, day = {1}, urldate = {2019-02-04}, journal = {Bioinformatics}, volume = {35}, number = {1}, issn = {1367-4803}, doi = {10.1093/bioinformatics/bty537}, pmid = {30561547}, pmcid = {PMC6298056}, abstract = {Motivation: Multiple biological clocks govern a healthy pregnancy. These biological mechanisms produce immunologic, metabolomic, proteomic, genomic and microbiomic adaptations during the course of pregnancy. Modeling the chronology of these adaptations during full-term pregnancy provides the frameworks for future studies examining deviations implicated in pregnancy-related pathologies including preterm birth and preeclampsia. Results: We performed a multiomics analysis of 51 samples from 17 pregnant women, delivering at term. The datasets included measurements from the immunome, transcriptome, microbiome, proteome and metabolome of samples obtained simultaneously from the same patients. Multivariate predictive modeling using the Elastic Net ({EN}) algorithm was used to measure the ability of each dataset to predict gestational age. Using stacked generalization, these datasets were combined into a single model. This model not only significantly increased predictive power by combining all datasets, but also revealed novel interactions between different biological modalities. Future work includes expansion of the cohort to preterm-enriched populations and in vivo analysis of immune-modulating interventions based on the mechanisms identified. Availability and implementation: Datasets and scripts for reproduction of results are available through: https://nalab.stanford.edu/multiomics-pregnancy/. Supplementary information: Supplementary data are available at Bioinformatics online.} } @article{kveler_2018, title = {Immune-centric network of cytokines and cells in disease context identified by computational mining of {PubMed}.}, author = { Ziv-Kenet, , , Yuri and Shalev-Malul, Gali and Aizenbud-Reshef, Netta and Dubovik, Tania and Briller, Mayan and Campbell, John and Rieckmann, and Asbeh, Nuaman and Rimar, Doron and Meissner, Felix and Wiser, Jeff and Shen-Orr, }, pages = {651-659}, url = {http://www.nature.com/doifinder/10.1038/nbt.4152}, year = {2018}, month = {jun}, day = {18}, urldate = {2019-01-10}, journal = {Nat Biotechnol}, volume = {36}, number = {7}, issn = {1087-0156}, doi = {10.1038/nbt.4152}, pmid = {29912209}, pmcid = {PMC6035104}, abstract = {Cytokines are signaling molecules secreted and sensed by immune and other cell types, enabling dynamic intercellular communication. Although a vast amount of data on these interactions exists, this information is not compiled, integrated or easily searchable. Here we report {immuneXpresso}, a text-mining engine that structures and standardizes knowledge of immune intercellular communication. We applied {immuneXpresso} to {PubMed} to identify relationships between 340 cell types and 140 cytokines across thousands of diseases. The method is able to distinguish between incoming and outgoing interactions, and it includes the effect of the interaction and the cellular function involved. These factors are assigned a confidence score and linked to the disease. By leveraging the breadth of this network, we predicted and experimentally verified previously unappreciated cell-cytokine interactions. We also built a global immune-centric view of diseases and used it to predict cytokine-disease associations. This standardized knowledgebase (http://www.immunexpresso.org) opens up new directions for interpretation of immune data and model-driven systems immunology.} } @article{tibshirani_1996, title = {Regression Shrinkage and Selection via the Lasso}, author = {}, pages = {267-288}, publisher = {Wiley for the Royal Statistical Society}, url = {https://www.jstor.org/stable/2346178}, year = {1996}, urldate = {2018-08-10}, journal = {Journal of the Royal Statistical Society. Series B (Methodological)}, volume = {58}, number = {1}, f1000-projects = {immuEN} } @article{zou_2005, title = {Regularization and variable selection via the elastic net}, author = { and }, publisher = {Wiley/Blackwell (10.1111)}, url = {https://rss.onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-9868.2005.00503.x\%4010.1111/\%{28I\SSN\}%291467-9868.{TOP\_SERIES\_B\_RESEARCH}}, year = {2005}, month = {apr}, day = {1}, urldate = {2018-08-10}, journal = {Journal of the Royal Statistical Society: Series B (Statistical Methodology)}, f1000-projects = {immuEN} } @article{maaten_2008, title = {Visualizing Data using t-{SNE}}, author = {Maaten, and Hinton, Geoffrey}, url = {http://www.jmlr.org/papers/v9/vandermaaten08a.html}, year = {2008}, urldate = {2018-08-16}, journal = {Journal of Machine Learning Research} } @article{hoerl_1970, title = {Ridge regression: biased estimation for nonorthogonal problems}, author = {Hoerl, . and .}, pages = {55}, url = {https://www.jstor.org/stable/1267351?origin=crossref}, year = {1970}, month = {feb}, urldate = {2018-09-27}, journal = {Technometrics}, volume = {12}, number = {1}, issn = {00401706}, doi = {10.2307/1267351} } @article{kellywelch_2005, title = {Interleukin-4 ({IL}-4) pathway.}, author = { }, pages = {cm9}, url = {http://dx.doi.org/10.1126/stke.2932005cm9}, year = {2005}, month = {jul}, day = {19}, urldate = {2018-10-23}, journal = {Sci {STKE}}, volume = {2005}, number = {293}, doi = {10.1126/stke.2932005cm9}, pmid = {16030287}, abstract = {Interleukin-4 ({IL}-4) is a cytokine produced by T(H)2 type helper T cells and by mast cells, basophils, and eosinophils. This cytokine can elicit many responses, some of which are associated with allergy and asthma. Studies with long-term cell lines and primary cells have revealed differences in the signaling between these two experimental systems. Understanding these differences is important because therapeutic strategies targeting {IL}-4 and its signaling pathways are currently being tested to treat allergy and asthma.} } @article{cui_2014, title = {{TLR4} ligands lipopolysaccharide and monophosphoryl lipid a differentially regulate effector and memory {CD8}+ T Cell differentiation.}, author = { and and and and and }, pages = {4221-4232}, url = {http://dx.doi.org/10.4049/jimmunol.1302569}, year = {2014}, month = {may}, day = {1}, urldate = {2018-10-23}, journal = {J Immunol}, volume = {192}, number = {9}, doi = {10.4049/jimmunol.1302569}, pmid = {24659688}, pmcid = {PMC4071140}, abstract = {Vaccines formulated with nonreplicating pathogens require adjuvants to help bolster immunogenicity. The role of adjuvants in Ab production has been well studied, but how they influence memory {CD8}(+) T cell differentiation remains poorly defined. In this study we implemented dendritic cell-mediated immunization to study the effects of commonly used adjuvants, {TLR} ligands, on effector and memory {CD8}(+) T cell differentiation in mice. Intriguingly, we found that the {TLR4} ligand {LPS} was far more superior to other {TLR} ligands in generating memory {CD8}(+) T cells upon immunization. {LPS} boosted clonal expansion similar to the other adjuvants, but fewer of the activated {CD8}(+) T cells died during contraction, generating a larger pool of memory cells. Surprisingly, monophosphoryl lipid A ({MPLA}), another {TLR4} ligand, enhanced clonal expansion of effector {CD8}(+) T cells, but it also promoted their terminal differentiation and contraction; thus, fewer memory {CD8}(+) T cells formed, and {MPLA}-primed animals were less protected against secondary infection compared with those primed with {LPS}. Furthermore, gene expression profiling revealed that {LPS}-primed effector cells displayed a stronger pro-memory gene expression signature, whereas the gene expression profile of {MPLA}-primed effector cells aligned closer with terminal effector {CD8}(+) T cells. Lastly, we demonstrated that the {LPS}-{TLR4}-derived "pro-memory" signals were {MyD88}, but not Toll/{IL}-{1R} domain-containing adapter inducing {IFN}-\beta, dependent. This study reveals the influential power of adjuvants on the quantity and quality of {CD8}(+) T cell memory, and that attention to adjuvant selection is crucial because boosting effector cell expansion may not always equate with more memory T cells or greater protection.} } @article{ryu_2018, title = {Fibrin-targeting immunotherapy protects against neuroinflammation and neurodegeneration.}, author = { Rafalski, and Meyer-Franke, , and Poda, and , and Pedersen, and and Baeten, and Sikorski, and Bedard, Catherine and Hanspers, Kristina and Bardehle, Sophia and Mendiola, and Davalos, Dimitrios and Machado, and Chan, and Plastira, Ioanna and Petersen, and Pfaff, and Ang, and Hallenbeck, and Syme, Catriona and Hakozaki, Hiroyuki and Ellisman, and Swanson, and Zamvil, and Arkin, and Zorn, and Pico, and Mucke, , and Stavenhagen, and Nelson, and Akassoglou, Katerina}, pages = {1212-1223}, url = {http://www.nature.com/articles/s41590-018-0232-x}, year = {2018}, month = {nov}, urldate = {2019-02-05}, journal = {Nat Immunol}, volume = {19}, number = {11}, issn = {1529-2908}, doi = {10.1038/s41590-018-0232-x}, pmid = {30323343}, pmcid = {PMC6317891}, abstract = {Activation of innate immunity and deposition of blood-derived fibrin in the central nervous system ({CNS}) occur in autoimmune and neurodegenerative diseases, including multiple sclerosis ({MS}) and Alzheimer's disease ({AD}). However, the mechanisms that link disruption of the blood-brain barrier ({BBB}) to neurodegeneration are poorly understood, and exploration of fibrin as a therapeutic target has been limited by its beneficial clotting functions. Here we report the generation of monoclonal antibody {5B8}, targeted against the cryptic fibrin epitope \gamma377-395, to selectively inhibit fibrin-induced inflammation and oxidative stress without interfering with clotting. {5B8} suppressed fibrin-induced nicotinamide adenine dinucleotide phosphate ({NADPH}) oxidase activation and the expression of proinflammatory genes. In animal models of {MS} and {AD}, {5B8} entered the {CNS} and bound to parenchymal fibrin, and its therapeutic administration reduced the activation of innate immunity and neurodegeneration. Thus, fibrin-targeting immunotherapy inhibited autoimmunity- and amyloid-driven neurotoxicity and might have clinical benefit without globally suppressing innate immunity or interfering with coagulation in diverse neurological diseases.} } @article{nettey_2018, title = {{OMIP}-050: A 28-color/30-parameter Fluorescence Flow Cytometry Panel to Enumerate and Characterize Cells Expressing a Wide Array of Immune Checkpoint Molecules.}, author = {Nettey, Giles, and Chattopadhyay, }, pages = {1094-1096}, url = {http://dx.doi.org/10.1002/cyto.a.23608}, year = {2018}, month = {nov}, urldate = {2018-11-07}, journal = {Cytometry A}, volume = {93}, number = {11}, doi = {10.1002/cyto.a.23608}, pmid = {30347136} } @article{saphire_2018, title = {Antibody-mediated protection against Ebola virus.}, author = {Saphire, Schendel, and Gunn, and Milligan, and Alter, Galit}, pages = {1169-1178}, url = {http://dx.doi.org/10.1038/s41590-018-0233-9}, year = {2018}, month = {nov}, urldate = {2019-02-05}, journal = {Nat Immunol}, volume = {19}, number = {11}, doi = {10.1038/s41590-018-0233-9}, pmid = {30333617}, abstract = {Recent Ebola virus disease epidemics have highlighted the need for effective vaccines and therapeutics to prevent future outbreaks. Antibodies are clearly critical for control of this deadly disease; however, the specific mechanisms of action of protective antibodies have yet to be defined. In this Perspective we discuss the antibody features that correlate with in vivo protection during infection with Ebola virus, based on the results of a systematic and comprehensive study of antibodies directed against this virus. Although neutralization activity mediated by the Fab domains of the antibody is strongly correlated with protection, recruitment of immune effector functions by the Fc domain has also emerged as a complementary, and sometimes alternative, route to protection. For a subset of antibodies, Fc-mediated clearance and killing of infected cells seems to be the main driver of protection after exposure and mirrors observations in vaccination studies. Continued analysis of antibodies that achieve protection partially or wholly through Fc-mediated functions, the precise functions required, the intersection with specificity and the importance of these functions in different animal models is needed to identify and begin to capitalize on Fc-mediated protection in vaccines and therapeutics alike.} } @book{hastie_2016, title = {The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition (Springer Series in Statistics)}, author = { and Tibshirani, Robert and }, publisher = {Springer}, url = {https://www.amazon.com/Elements-Statistical-Learning-Prediction-Statistics/dp/0387848576}, year = {2016}, urldate = {2018-12-04}, edition = {2nd}, isbn = {9780387848570} } @article{zhou_2008, title = {Angiotensin receptor agonistic autoantibodies induce pre-eclampsia in pregnant mice.}, author = { and , , and Kellems, and }, pages = {855-862}, url = {http://dx.doi.org/10.1038/nm.1856}, year = {2008}, month = {aug}, urldate = {2018-12-05}, journal = {Nat Med}, volume = {14}, number = {8}, issn = {1546-{170X}}, doi = {10.1038/nm.1856}, pmid = {18660815}, pmcid = {PMC3267158}, abstract = {Pre-eclampsia affects approximately 5\% of pregnancies and remains a leading cause of maternal and neonatal mortality and morbidity in the United States and the world. The clinical hallmarks of this maternal disorder include hypertension, proteinuria, endothelial dysfunction and placental defects. Advanced-stage clinical symptoms include cerebral hemorrhage, renal failure and the {HELLP} (hemolysis, elevated liver enzymes and low platelets) syndrome. An effective treatment of pre-eclampsia is unavailable owing to the poor understanding of the pathogenesis of the disease. Numerous recent studies have shown that women with pre-eclampsia possess autoantibodies, termed {AT}(1)-{AAs}, that bind and activate the angiotensin {II} receptor type 1a ({AT}(1) receptor). We show here that key features of pre-eclampsia, including hypertension, proteinuria, glomerular endotheliosis (a classical renal lesion of pre-eclampsia), placental abnormalities and small fetus size appeared in pregnant mice after injection with either total {IgG} or affinity-purified {AT}(1)-{AAs} from women with pre-eclampsia. These features were prevented by co-injection with losartan, an {AT}(1) receptor antagonist, or by an antibody neutralizing seven-amino-acid epitope peptide. Thus, our studies indicate that pre-eclampsia may be a pregnancy-induced autoimmune disease in which key features of the disease result from autoantibody-induced angiotensin receptor activation. This hypothesis has obvious implications regarding pre-eclampsia screening, diagnosis and therapy.} } @article{jensen_2012, title = {{CD19}+{CD5}+ cells as indicators of preeclampsia.}, author = { }, pages = {861-868}, url = {http://dx.doi.org/10.1161/{HYPERTENSIONAHA}.111.188276}, year = {2012}, month = {apr}, urldate = {2018-12-05}, journal = {Hypertension}, volume = {59}, number = {4}, doi = {10.1161/{HYPERTENSIONAHA}.111.188276}, pmid = {22353610}, abstract = {Preeclampsia is a devastating pregnancy-associated disorder affecting 5\% to 8\% of pregnant women worldwide. It emerges as an autoimmune-driven disease, and, among others, the autoantibodies against angiotensin type 1 receptor {II} have been proposed to account for preeclampsia symptoms. Despite much attention focused on describing autoantibodies associated with preeclampsia, there is no clue concerning the cell population producing them. {CD19}(+){CD5}(+) B-1a B cells constitute the main source of natural and polyreactive antibodies, which can be directed against own structures. Here, we aimed to identify the B-cell subpopulation responsible for autoantibody production during preeclampsia and to study their regulation, as well as their possible use as markers for the disease. The frequency of {CD19}(+){CD5}(+) cells in peripheral blood of preeclamptic patients is dramatically increased compared with normal pregnant women as analyzed by flow cytometry. This seems to be driven by the high human chorionic gonadotropin levels present in the serum and placenta supernatant of preeclamptic patients versus normal pregnant women. Not only ≈95\% of {CD19}(+){CD5}(+) cells express the human chorionic gonadotropin receptor, but these cells also expand on human chorionic gonadotropin stimulation in a lymphocyte culture. Most importantly, isolated {CD19}(+){CD5}(+) cells produce autoantibodies against angiotensin type 1 receptor {II}, and {CD19}(+){CD5}(+) cells were further detected in the placenta of preeclamptic but not of normal pregnancies where barely B cells are present. Our results identify a B-cell population able to produce pregnancy-pathological autoantibodies as possible markers for preeclampsia, which opens vast diagnostic and therapeutic applications.} } @article{matthiesen_2005, title = {Immunology of preeclampsia.}, author = {, Ernerudh, , Jonsson, }, pages = {49-61}, url = {http://dx.doi.org/10.1159/000087912}, year = {2005}, urldate = {2018-12-05}, journal = {Chem Immunol Allergy}, volume = {89}, doi = {10.1159/000087912}, pmid = {16129952}, abstract = {Preeclampsia is a placenta-dependent disorder with both local and systemic anomalies with neonatal and maternal morbidity. It is manifested late in pregnancy, but the onset is during early stages of gestation. The current hypothesis regarding the aetiology of preeclampsia is focused on maladaptation of immune responses and defective trophoblast invasion. Thus, an excessive maternal inflammatory response, perhaps directed against foreign fetal antigens, results in a chain of events including shallow trophoblast invasion, defective spiral artery remodelling, placental infarction and release of pro-inflammatory cytokines and placental fragments in the systemic circulation. During normal pregnancy, trophoblasts interact in the decidua with the unique uterine {NK} cells, modifying their cytokine repertoire, regulating adhesion molecules and matrix metalloproteinases. The inability of trophoblasts to accomplish these changes might be a critical factor for the onset of preeclampsia. Several cytokines, produced at the maternal-fetal interface, have an impact on trophoblast invasion. It is suggested that deficiency of interleukin-10 may contribute to enhanced inflammatory responses towards the trophoblasts elicited by e.g. tumour necrosis factor-alpha and interferon-gamma. Consequently, trophoblasts subjected to a high rate of apoptosis are hampered in their invasive capacity resulting in defective transformation of spiral arteries, hypoxia, thrombosis and infarction of the placenta. The ensuing infarction of placenta leads to leakage of increasing amounts of placental fragments and cytokines in the maternal circulation and an exaggerated systemic endothelial activation as identified in preeclampsia. So far, treatment of preeclampsia is focused on signs like hypertension, whereas attempts of modifying immune responses may be a possibility in the future.} } @article{deshmukh_2018, title = {Immunological basis for recurrent fetal loss and pregnancy complications.}, author = { }, url = {http://dx.doi.org/10.1146/annurev-pathmechdis-012418-012743}, year = {2018}, month = {sep}, day = {5}, urldate = {2018-12-10}, journal = {Annu Rev Pathol}, volume = {14}, number = {1}, doi = {10.1146/annurev-pathmechdis-012418-012743}, pmid = {30183507}, abstract = {Pregnancy stimulates an elaborate assortment of dynamic changes, allowing intimate approximation of genetically discordant maternal and fetal tissues. Although the cellular and molecular details about how this works remain largely undefined, important clues arise from evaluating how a prior pregnancy influences the outcome of a future pregnancy. The risk of complications is consistently increased when complications occurred in a prior pregnancy. Reciprocally, a prior successful pregnancy protects against complications in a future pregnancy. Here, we summarize immunological perturbations associated with fetal loss, with particular focus on how both harmful and protective adaptations may persist in mothers. Immunological aberrancy as a root cause of pregnancy complications is also considered, given their shared overlapping risk factors and the sustained requirement for averting maternal-fetal conflict throughout pregnancy. Understanding pregnancy-induced immunological changes may expose not only new therapeutic strategies for improving pregnancy outcomes but also new facets of how immune tolerance works, and these may be applicable to other physiological and pathological contexts. Expected final online publication date for the Annual Review of Pathology: Mechanisms of Disease Volume 14 is January 24, 2019. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.} } @article{lux_2018, title = {{flowLearn}: fast and precise identification and quality checking of cell populations in flow cytometry.}, author = {Lux, , and Chauve, , , , , Barbara}, pages = {2245-2253}, url = {http://dx.doi.org/10.1093/bioinformatics/bty082}, year = {2018}, month = {jul}, day = {1}, urldate = {2019-01-11}, journal = {Bioinformatics}, volume = {34}, number = {13}, doi = {10.1093/bioinformatics/bty082}, pmid = {29462241}, pmcid = {PMC6022609}, abstract = {Motivation: Identification of cell populations in flow cytometry is a critical part of the analysis and lays the groundwork for many applications and research discovery. The current paradigm of manual analysis is time consuming and subjective. A common goal of users is to replace manual analysis with automated methods that replicate their results. Supervised tools provide the best performance in such a use case, however they require fine parameterization to obtain the best results. Hence, there is a strong need for methods that are fast to setup, accurate and interpretable. Results: {flowLearn} is a semi-supervised approach for the quality-checked identification of cell populations. Using a very small number of manually gated samples, through density alignments it is able to predict gates on other samples with high accuracy and speed. On two state-of-the-art datasets, our tool achieves median(F1)-measures exceeding 0.99 for 31\%, and 0.90 for 80\% of all analyzed populations. Furthermore, users can directly interpret and adjust automated gates on new sample files to iteratively improve the initial training. Availability and implementation: {FlowLearn} is available as an R package on https://github.com/mlux86/{flowLearn}. Evaluation data is publicly available online. Details can be found in the Supplementary Material. Supplementary information: Supplementary data are available at Bioinformatics online.} } @article{bergersen_2011, title = {Weighted lasso with data integration.}, author = {Bergersen, }, url = {http://dx.doi.org/10.2202/1544-6115.1703}, year = {2011}, month = {aug}, day = {29}, urldate = {2019-01-16}, journal = {Stat Appl Genet Mol Biol}, volume = {10}, number = {1}, doi = {10.2202/1544-6115.1703}, pmid = {23089821}, abstract = {The lasso is one of the most commonly used methods for high-dimensional regression, but can be unstable and lacks satisfactory asymptotic properties for variable selection. We propose to use weighted lasso with integrated relevant external information on the covariates to guide the selection towards more stable results. Weighting the penalties with external information gives each regression coefficient a covariate specific amount of penalization and can improve upon standard methods that do not use such information by borrowing knowledge from the external material. The method is applied to two cancer data sets, with gene expressions as covariates. We find interesting gene signatures, which we are able to validate. We discuss various ideas on how the weights should be defined and illustrate how different types of investigations can utilize our method exploiting different sources of external data. Through simulations, we show that our method outperforms the lasso and the adaptive lasso when the external information is from relevant to partly relevant, in terms of both variable selection and prediction.} } @article{mollaysa_2017, title = {Regularising Non-linear Models Using Feature Side-information}, author = { Kalousis, Alexandros}, url = {https://arxiv.org/abs/1703.02570}, year = {2017}, month = {mar}, day = {7}, urldate = {2019-01-16}, journal = {arXiv}, abstract = {Very often features come with their own vectorial descriptions which provide detailed information about their properties. We refer to these vectorial descriptions as feature side-information. In the standard learning scenario, input is represented as a vector of features and the feature side-information is most often ignored or used only for feature selection prior to model fitting. We believe that feature side-information which carries information about features intrinsic property will help improve model prediction if used in a proper way during learning process. In this paper, we propose a framework that allows for the incorporation of the feature side-information during the learning of very general model families to improve the prediction performance. We control the structures of the learned models so that they reflect features similarities as these are defined on the basis of the side-information. We perform experiments on a number of benchmark datasets which show significant predictive performance gains, over a number of baselines, as a result of the exploitation of the side-information.} } @article{hans_2011, title = {Elastic net regression modeling with the orthant normal prior}, author = {}, pages = {1383-1393}, url = {http://www.tandfonline.com/doi/abs/10.1198/jasa.2011.tm09241}, year = {2011}, month = {dec}, urldate = {2019-01-17}, journal = {J Am Stat Assoc}, volume = {106}, number = {496}, issn = {0162-1459}, doi = {10.1198/jasa.2011.tm09241}, abstract = {The elastic net procedure is a form of regularized optimization for linear regression that provides a bridge between ridge regression and the lasso. The estimate that it produces can be viewed as a Bayesian posterior mode under a prior distribution implied by the form of the elastic net penalty. This article broadens the scope of the Bayesian connection by providing a complete characterization of a class of prior distributions that generate the elastic net estimate as the posterior mode. The resulting model-based framework allows for methodology that moves beyond exclusive use of the posterior mode by considering inference based on the full posterior distribution. Two characterizations of the class of prior distributions are introduced: a properly normalized, direct characterization, which is shown to be conjugate for linear regression models, and an alternate representation as a scale mixture of normal distributions. Prior distributions are proposed for the regularization parameters, resulting in an infinite mixture of elastic net regression models that allows for adaptive, data-based shrinkage of the regression coefficients. Posterior inference is easily achieved using Markov chain Monte Carlo ({MCMC}) methods. Uncertainty about model specification is addressed from a Bayesian perspective by assigning prior probabilities to all possible models. Corresponding computational approaches are described. Software for implementing the {MCMC} methods described in this article, written in C++ with an R package interface, is available at http://www.stat.osu.edu/\~hans/software/.} } 0 \begin{table} \centering \begin{tabular}{|l|c|c|c|c|} \hline &\textbf{Stand}&\textbf{Walk}&\textbf{Sit}&\textbf{Lie}\\\hline \textbf{Only ADLs (# of instances)}&1094&1095&90&40\\\hline \textbf{ADLs + Drills (# of instances)}&1711&1733&169&40\\\hline \textbf{Subject 1 only ADLs (# of instances)}&252&271&11&10\\\hline \textbf{Subject 1 ADLs + Drill (# of instances)}&424&463&31&10\\\hline \textbf{Subject 2 only ADLs (# of instances)}&286&285&23&10\\\hline \textbf{Subject 2 ADLs + Drill (# of instances)}&437&434&43&10\\\hline \textbf{Subject 3 only ADLs (# of instances)}&259&245&32&10\\\hline \textbf{Subject 3 ADLs + Drill (# of instances)}&406&394&50&10\\\hline \textbf{Subject 4 only ADLs (# of instances)}&297&294&24&10\\\hline \textbf{Subject 4 ADLs + Drill (# of instances)}&444&442&45&10\\\hline \end{tabular} \caption{} \end{table}\section*{Eidesstattliche Erklärung} Hiermit erkläre ich an Eides statt, dass ich die vorgelegte Diplomarbeit selbstständig und ohne Benutzung anderer als der angegebenen Hilfsmittel angefertigt habe. Gedanken, die aus fremden Quellen direkt oder indirekt übernommen wurden, sind als solche gekennzeichnet. Die Arbeit wurde bisher in gleicher oder ähnlicher Weise keiner anderen Prüfungsbehörde vorgelegt und auch noch nicht veröffentlicht. \\[1em] Leonding, am \duedatede \\[5em] \ifthenelse{\isundefined{\firstauthor}}{}{\firstauthor} \ifthenelse{\isundefined{\secondauthor}}{}{\kern-1ex, \secondauthor} \\[5em] \begin{otherlanguage}{english} \section*{Declaration of Academic Honesty} Hereby, I declare that I have composed the presented paper independently on my own and without any other resources than the ones indicated. All thoughts taken directly or indirectly from external sources are properly denoted as such. This paper has neither been previously submitted to another authority nor has it been published yet. \\[1em] Leonding, \duedateen \\[5em] \ifthenelse{\isundefined{\firstauthor}}{}{\firstauthor} \ifthenelse{\isundefined{\secondauthor}}{}{\kern-1ex, \secondauthor} \end{otherlanguage} \begin{abstract} \begin{itemize} \item {\em Aufgabenstellung} \\ Um den Kunden der Firma {\projectpartner} die Visualisierung eines für sie designten Bades zu erleichtern, ist es notwendig, ihnen eine möglichst genaue und detaillierte Darstellung anzubieten. Dies bezieht sich sowohl auf das Design, die Komponenten als auch für die Bauanleitung. Da nicht nur die Hotelketten das Modell zu Gesicht bekommen, sondern auch Monteure, die sich daran beim Zusammenbauen orientieren können. Dadurch dass der Fortschritt und das fertige Bad aus verschiedenen Perspektiven angesehen werden kann, wird die Vision für alle klarer und einfacher. Da die ganze Applikation webbasiert ist, ist sie plattformunabhängig und kann dank Electron lokal und ohne Internet ausgeführt werden. Bisher wurde dies immer mit Präsentationsvideos umgesetzt, die jedoch sehr unflexibel sind und zeitintensiv waren. Die Probleme waren dabei, dass das Bad immer nur aus einer Perspektive zu sehen war, das Modell nicht interaktiv war und Kundenwünsche nicht sofort umgesetzt werden konnten. Der Umgang mit dem Tool wird möglichst intuitiv erfolgen und besondere Ressourcenanforderungen sind nicht präsent. \item {\em Umsetzung} \\ Die Web-Anwendung basiert auf den gängigen Technologien HTML, JavaScript, Three.js und WEB.GL. Dies ermöglicht es über einen beliebigen Browser und beliebiges Betriebssystem darauf zuzugreifen. Die einzige Anforderung ist eine Internetverbindung. Die Wahl für die oben genannten Technologien ist darauf zurückzuführen, dass alle robust, zukunftssicher, gut dokumentiert und weitverbreitet sind. Damit eignen sie sich perfekt für die Applikation und sind ein wichtiger Beitrag dafür das die Anwendung wartbar ist und bleibt. \item {\em Ergebnisse} \\ Die Software wurde nach Abschluss der Arbeit an das Unternehmen {\projectpartner} übergeben. Demonstrationen und Tutorials an die zukünftigen Anwender wurden durch das Entwicklerteam durchgeführt. Unter folgender Adresse \url{http://vm85.htl-leonding.ac.at/} kann die Diplomarbeit begutachtet und verwendet werden. \end{itemize} \clearpage \newpage \begin{figure}[h] \centering \includegraphics[width=0.65\linewidth]{images/Screenshot_front.png} \caption{Bad Designer; Frontalansicht} \end{figure} \begin{figure}[h] \centering \includegraphics[width=0.65\linewidth]{images/Screenshot_schraeg.png} \caption{Bad Designer; Schrägansicht} \end{figure} \begin{figure}[h] \centering \includegraphics[width=0.65\linewidth]{images/Screenshot_seite.png} \caption{Bad Designer; Seitenansicht} \end{figure} \end{abstract} \newpage \clearpage \begin{otherlanguage}{english} \begin{abstract} % \begin{enumerate} \begin{itemize} \item {\em Definition of the project} \\ In order to present the customers of {\projectpartner} a visualization of a custom-made modular bath it is necessary to have a precise and detailed illustration. This is especially important for the custom design, the components and for the construction manual. As the model will be showed to the assembler as well it is important that it is possible to build it with the manual the application provides. The progress and the complete bath can be viewed from multiple angles with the intention of making the concept clearer and simpler for all. As it is a web application it can be run on every operating system and thanks to Electron locally runable without an internet connection. Up to now this was done with presentation videos, but they were too inflexible and time-consuming. Moreover, the bath was only visible from one angle, the model was not interactive and customer requirements could not be implemented immediately. The tool is as intuitive as possible and does not require special resources. \item {\em Implementation} \\ The web-app is based on the commonly used technologies HTML, JavaScript, Three.js and WebGL. This allows it to be run on every operating system and any modern browser. It only requires a internet connection. The mentioned technologies were uses because they are all robust, future-proof, well documented and extendable. Thanks to these attributes they are perfect fit for the application and make it easier to service. \item {\em Results} \\ The software was delivered after the thesis to the company {\projectpartner}. Demonstrations and tutorials were provided to the future users by the developing team. The work is publicly accessible on this website \url{http://vm85.htl-leonding.ac.at/}. \end{itemize} \clearpage \newpage \begin{figure}[h] \centering \includegraphics[width=0.65\textwidth]{images/Screenshot_front.png} \caption{Bad Designer; Frontal view} \end{figure} \begin{figure}[h] \centering \includegraphics[width=0.65\textwidth]{images/Screenshot_schraeg.png} \caption{Bad Designer; Oblique view } \end{figure} \begin{figure}[h] \centering \includegraphics[width=0.65\textwidth]{images/Screenshot_seite.png} \caption{Bad Designer; Side view} \end{figure} \end{abstract} \end{otherlanguage} \newpage \clearpage \section*{Danksagungen} Wir möchten uns sehr herzlich bei unserem Diplomarbeitsbetreuer {\supervisor} bedanken der uns stets fachlich und persönlich unterstützt hat und immer an unserer Seite stand. Einen großen Dank möchten wir auch an unseren Auftraggeber die {\projectpartner} richten die uns das Vertrauen geschenkt hat. Für das Vermitteln der Diplomarbeit möchten wir uns bei Prof. Dipl.-Ing. bedanken. \\ \\ Ein großer Dank gilt auch unseren Familien und Freunden die in der Zeit der Erstellung dieser Arbeit für uns da waren und uns unterstützt haben.0 \newpage \subsection*{Weak Authentication } 10-100 \chapter*{Erklärung zur Selbstständigkeit} Ich versichere, dass ich diese Arbeit selbstständig verfasst habe und keine % anderen als die angegebenen Quellen und Hilfsmittel benutzt habe, die % wörtlich oder inhaltlich übernommenen Stellen als solche kenntlich gemacht und % die Satzung des KIT zur Sicherung guter wissenschaftlicher Praxis in der % gültigen Fassung vom 24.05.2018 beachtet habe.\\ \vspace{1cm} \renewcommand{\arraystretch}{0} % for spacing in the tabular environment \begin{flushright} \begin{tabular}{rr} Karlsruhe, den \thesistimehandin, & \hspace*{5cm}\\[0mm] \cline{2-2}\\[2mm] % the last line has height 2mm due & \thesisauthor % to \arraystretch=0 \end{tabular} \end{flushright} \vfill \begin{flushright} Als Prüfungsexemplar genehmigt von\\ \vspace{1cm} \begin{tabular}{rr} Karlsruhe, den \thesistimehandin, & \hspace*{5cm}\\[0mm] \cline{2-2}\\[2mm] % the last line has height 2mm due & \thesisreviewerone % to \arraystretch=0 \end{tabular} \end{flushright} \renewcommand{\arraystretch}{1} \cleardoublepage pubs/SPIE2018/latex/main.tex \documentclass[]{spie} % US letter paper %\documentclass[a4paper]{spie} % A4 paper %\documentclass[nocompress]{spie} % no compression of citations \renewcommand{\baselinestretch}{1.0} % 1.65 for double spacing \usepackage{amsmath,amsfonts,amssymb} \usepackage{graphicx} \usepackage{wrapfig} \usepackage{csquotes} \usepackage{xcolor} \usepackage{framed} \usepackage{soul} \usepackage{float} \usepackage{stackengine} \usepackage{gensymb} \usepackage{booktabs} % \usepackage{lipsum} \usepackage{colortbl} \usepackage[colorlinks=true, allcolors=blue]{hyperref} \definecolor{tablegrey}{HTML}{808080} % style \pagestyle{empty} %\pagestyle{plain} % for page numbers %\setcounter{page}{301} % start numbering at 301 %\sethlcolor{green} % default highlighter %\definecolor{example}{RGB}{28, 69, 135} % custom color %\colorlet{shadecolor}{orange} % macros \DeclareQuoteStyle[american]{english}{\itshape\textquotedblleft}[\textquotedblleft]{\textquotedblright}[0.05em]{\textquoteleft}{\textquoteright} \DeclareRobustCommand{\hlnote}[1]{{\sethlcolor{green}\hl{\textsc{#1}}}} \DeclareRobustCommand{\framenote}[1]{{\colorlet{shadegreen}{yellow}\begin{shaded}#1\end{shaded}}} % title \title{Learning and estimating whole sky visible, VNIR, SWIR radiance distributions from a commercial camera} % authors \author[a]{} \author[b]{} \author[c]{} \author[a]{ Jr.} \affil[a]{Institute for Simulation and Training, University of Central Florida, Orlando, FL, USA} \affil[b]{Full Sail University, Winter Park, FL, USA} \affil[c]{University of Miami, Coral Gables, FL, USA} \authorinfo{Send correspondence to: , } % main document \begin{document} \maketitle % teaser \vfill \begin{figure}[H] \begin{center} \begin{tabular}{!{\color{tablegrey}\vrule}cc!{\color{tablegrey}\vrule}cc!{\color{tablegrey}\vrule}} \arrayrulecolor{tablegrey}\hline \includegraphics[width=0.448\textwidth]{img/story_train.png} & & & \includegraphics[width=0.448\textwidth]{img/story_predict.png} \\\arrayrulecolor{tablegrey}\hline \cellcolor{blue!11} \footnotesize {Offline Learning (Precomputation)} &\cellcolor{blue!11} &\cellcolor{red!11} & \cellcolor{red!11}\footnotesize{Whole Sky Spectral Radiance Estimation (Real-time)} \\\arrayrulecolor{tablegrey}\hline \end{tabular} \end{center}\vspace{-2mm} \caption{We estimate sky radiance distribution curves between 350-2500nm from images captured with a digital camera. (Left) We use measurements from a commercial digital camera and a sky scanning spectroradiometer to train machine learning algorithms. (Right) We then utilize this ML model to take new sky images and produce the whole sky spectral illumination in real-time. We have validated this approach to measured data. } \label{fig:teaser} \end{figure} \vfill % text \input{tex/0_abstract} \keywords{Sky Radiance, Radiance Distribution, Multispectral, Machine Learning, Regression, Sky Viewer} \input{tex/1_introduction} \input{tex/2_data} \input{tex/3_methods} \input{tex/4_results} \input{tex/5_conclusions} \input{tex/6_acknowledgements} % references \bibliography{citations} % bibliography data in report.bib \bibliographystyle{spiebib} % makes bibtex use spiebib.bst \end{document} \section{What-If Scenarios} In our report~%% \cite{climate_report} , we mentioned that another approach to addressing climate issues is to rethink the conference culture that has emerged, organically over many decades, in SIGPLAN (and computer science generally) and consider more radical changes to it. We then list a few alternatives. This section presents a forecast of carbon footprint for three of those alternatives, based on the attendance data we have. \emph{Again, we need to make a lot of assumptions and explain them, because this is not a linear system...} \subsection{What If: Mega Conferences} \subsection{What If: Regional Conferences} \subsection{What If: Multi-Site Conferences} \ifLaomianLinkColors@ \definecolor{LaomianCiteColor}{rgb}{0,0,0.6} \definecolor{LaomianLinkColor}{rgb}{0.8,0,0} \definecolor{LaomianURLColor}{rgb}{0,0.3,0.8} \usepackage[colorlinks=true,citecolor=LaomianCiteColor,linkcolor=LaomianLinkColor,filecolor=LaomianURLColor,urlcolor=LaomianURLColor]{hyperref} \else \usepackage[colorlinks=true,allcolors=blue]{hyperref} \fi % Implement ORCID icon. \ifLaomianORCIDIcon@ \usepackage{orcidlink} \def\orcidID#1{\unskip\,\orcidlink{#1}} \fi \chapter*{编著者的话} 无线电资源是全人类共同的财产。提到无线电，我们再熟悉不过的是日常生活中的手机和Wi-Fi，在军事上，人们利用无线电控制导弹、飞机，%用来杀人 在救险活动上，人们利用无线电辅助实施灾害时的救援，%用来救人 在业余无线电领域，爱好者们互相通信，以提高技能，同时学习新知识。 笔者原本对业余无线电一无所知，因为精通业余无线电的朋友的介绍，才逐渐开始对其有所了解。在日本留学期间，笔者考取了日本的操作证书和电台执照，建立了第一个自己的业余无线电台，开始了业余无线电爱好者的旅途。 在归国后，笔者通过业余无线电操作证考试拿到了A类的操作证。在操作证考试应试学习过程中，笔者深深感到，国内现有的操作证考试应试书籍对于很多小学生读者来说，缺乏细致的解释，题目里的术语艰涩难懂，计算题不知道如何计算，用这些书籍学习的读者，想必难以通过操作证考试。在这样的背景下，笔者萌生了撰写一本老少皆能读懂的操作证考试的应试书籍的想法。 本书在写作过程中，为了让业余无线电知识几乎完全不了解的初学者也能读懂，笔者经过了反复的推敲，尽可能的把复杂的业余无线电知识简单易懂地展现给读者们。本书在解题的过程中，适当地介绍相关的术语，并把重点难点用加粗的字体标出，方便应试者快速记忆概念、理解计算方法。 %希望本书能帮助您顺利通过考试。 %关我啥事 %咋的，编不出来了？ 17-3_latex/Times_New_Roman/Times_New_Roman.tex % !TeX program = xelatex \documentclass[12pt,a4paper]{article} \usepackage{siunitx} \sisetup{math-rm = \ensuremath,math-micro = \symup{μ}} % für Textschriftart \usepackage{fontspec} % für Matheschriftarten \usepackage{unicode-math} % legt Schriftart für Text fest, kann eine beliebige Systemschriftart sein \setmainfont{Times New Roman} % legt Schriftarten für Mathematik fest: % Schriftart für Sonderzeichen, zum Beispiel: % latinmodern-math.otf, xits-math.otf, stixmath-regular.otf, Asana-Math.otf, texgyrepagella-math.otf, euler.otf, texgyredejavu-math.otf \setmathfont[Scale=MatchUppercase]{xits-math.otf} % Schriftarten für grosse und kleine, griechische und lateinische Buchstaben sowie Zahlen, beliebige Systemschriftart \setmathfont[range=up/{greek,Greek,latin,Latin,num}]{Times New Roman} \setmathfont[range=it/{greek,Greek,latin,Latin,num}]{Times New Roman} \setmathfont[range=bb/{greek,Greek,latin,Latin,num}]{Times New Roman} \setlength{\parindent}{0pt} \begin{document} \begin{center}\Large XeLaTeX mit Times New Roman-Font \end{center} \begin{center} \textbf{Wichtig:} Mit XeLaTeX kompilieren. \end{center} \section{Text}\sloppy Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut labore et dolore magna aliquyam erat, sed diam voluptua. At vero eos et accusam et justo duo dolores et ea rebum. Stet clita kasd gubergren, no sea takimata sanctus est Lorem ipsum dolor sit amet. Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut labore et dolore magna aliquyam erat, sed diam voluptua. At vero eos et accusam et justo duo dolores et ea rebum. Stet clita kasd gubergren, no sea takimata sanctus est Lorem ipsum dolor sit amet.\\ \SI{5}{\mu mol} bei einer Ausbeute von \SI{75}{\%} bei $\Delta T=\SI{50}{K}$. \\ By employing the Eyring equation of the transition state theory, the activation enthalphy $\Delta H=\SI{43(3)}{kJ\:mol^{-1}}$ and activation entropy $\Delta S=\SI{-91(10)}{J\:K^{-1}\:mol^{-1}}$ were acquired. \section{Gleichungen} \begin{equation} k(T)=A\cdot\exp\left(-\frac{E_A}{RT}\right)\qquad\Leftrightarrow\qquad \ln k=-\frac{E_A}{RT}+\ln A \end{equation} \begin{equation} q_v=\prod_{i=1}^s\left(1-e^{-\frac{h\nu_i}{k_\textrm{B}T}}\right)^{-1} \end{equation} \begin{equation} \textrm{Gr}=\frac{L_c^3\,g\,\beta\,\Delta T\,\rho^2}{\mu^2} \end{equation} \end{document}jb80/jbaggio.info @misc{Cumming2020, abstract = {Institutions are vital to the sustainability of social-ecological systems, balancing individual and group interests and coordinating responses to change. Ecological decline and social conflict in many places, however, indicate that our understanding and fostering of effective institutions for natural resource management is still lacking. We assess theoretical and methodological challenges facing positivist institutional analysis, focusing on natural resource governance according to Ostrom's social-ecological systems (SES) framework. Rather than adding more variables, progress requires a clearer, more consistent approach to selecting, defining and measuring institutional elements; stronger links between theory and empirical research; a greater focus on mechanisms and causality; and the development and application of new methods, including quantitative approaches. Strengthening the connections between theory, models, and data suggests several promising avenues for advancing institutional analysis through the study of relationships between institutional structure, process, function, context, and outcomes.}, author = { . and . and . and and . and . and . and .}, booktitle = {Current Opinion in Environmental Sustainability}, doi = {10.1016/j.cosust.2020.02.005}, file = {:Users/jacapobaggio/Documents/A_STUDI/AAA_ALL_Papers/Cumming et al. - 2020 - Advancing understanding of natural resource governance a post-Ostrom research agenda.pdf:pdf}, issn = {18773435}, month = {jun}, pages = {26--34}, title = {Advancing understanding of natural resource governance: a post-Ostrom research agenda}, url = {https://linkinghub.elsevier.com/retrieve/pii/S1877343520300129}, volume = {44}, year = {2020} } \hypertarget{_intake_8cpp}{}\doxysection{C\+:/robotcm/training2020/src/main/cpp/subsys/\+Intake.cpp File Reference} \label{_intake_8cpp}\index{C:/robotcm/training2020/src/main/cpp/subsys/Intake.cpp@{C:/robotcm/training2020/src/main/cpp/subsys/Intake.cpp}} {\ttfamily \#include $<$vector$>$}\newline {\ttfamily \#include $<$memory$>$}\newline {\ttfamily \#include $<$string$>$}\newline {\ttfamily \#include $<$controllers/\+Control\+Modes.\+h$>$}\newline {\ttfamily \#include $<$controllers/\+Control\+Data.\+h$>$}\newline {\ttfamily \#include $<$subsys/\+Intake.\+h$>$}\newline {\ttfamily \#include $<$hw/\+Dragon\+Solenoid.\+h$>$}\newline {\ttfamily \#include $<$hw/interfaces/\+I\+Dragon\+Motor\+Controller.\+h$>$}\newline {\ttfamily \#include $<$utils/\+Logger.\+h$>$}\newline @article{zarybnicka2009tengmalm, title={Do Tengmalm’s Owls alter parental feeding effort under varying conditions of main prey availability?}, author={Z{\'a}rybnick{\'a}, Mark{\'e}ta and Sedl{\'a}{\v{c}}ek, Ond{\v{r}}ej and Korpim{\"a}}, journal={Journal of Ornithology}, volume={150}, number={1}, pages={231--237}, year={2009}, publisher={Springer} } @book{graham2003hayman, title={Hayman fire case study}, author={}, year={2003}, publisher={US Department of Agriculture, Forest Service, Rocky Mountain Research Station} } @article{griffis2001understory, title={Understory response to management treatments in northern Arizona ponderosa pine forests}, author={Griffis, and Crawford, and Wagner, and }, journal={Forest Ecology and Management}, volume={146}, number={1-3}, pages={239--245}, year={2001}, publisher={Elsevier} } @article{nappi2010effect, title={Effect of fire severity on long-term occupancy of burned boreal conifer forests by saproxylic insects and wood-foraging birds}, author={ and Saint-Germain, Michel and Angers, }, journal={International Journal of Wildland Fire}, volume={19}, number={4}, pages={500--511}, year={2010}, publisher={CSIRO} } @article{veblen2000climatic, title={Climatic and human influences on fire regimes in ponderosa pine forests in the Colorado Front Range}, author={ }, journal={Ecological applications}, volume={10}, number={4}, pages={1178--1195}, year={2000}, publisher={Wiley Online Library} } @article{linkhart2013flammulated, title={Flammulated owl (Psiloscops flammeolus)}, author={Linkhart, Brian and McCallum, DA}, journal={The Birds of North America (PG Rodewald, Editor). Cornell Lab of Ornithology, Ithaca, NY, USA. https://birdsna. org/Species-Account/bna/species/flaowl}, year={2013} } @inproceedings{reynolds1992flammulated, title={Flammulated owls in ponderosa pine: evidence of preference for old growth}, author={Reynolds, and Linkhart, }, booktitle={MR Kaufman, WH Moir, and RL Bassett, technical coordinators. Proceedings of the workshop on old-growth in the Southwest and Rocky Mountain Region. Portal, Arizona, USA}, pages={166--169}, year={1992} } @article{fornwalt2016did, title={Did the 2002 Hayman Fire, Colorado, USA, burn with uncharacteristic severity?}, author={Fornwalt, and Huckaby, and Alton, and Kaufmann, and Brown, and Cheng, }, journal={Fire Ecology}, volume={12}, number={3}, pages={117--132}, year={2016}, publisher={Springer} } \documentclass[../document.tex]{subfiles} \begin{document} \section*{Preliminary Study} \addcontentsline{toc}{section}{Preliminary Study} The project, internet of things, is quite abstract. We did not have a clear goal at first, but we were asked by the customer to come up with a solution that fits the customer needs. This chapter outlines these ideas and also why we chose to utilize the scrum model as well as which tools we will use, both hardware and software, for this project. \subsection*{Initial Ideas} \addcontentsline{toc}{subsection}{Initial Ideas} Early in the project, we explored a wide range of ideas and application areas for the cheap sensor technology supplied by Altran. These ideas were then further developed into more concrete designs, with pros and cons listed. The outcome of this process is summarized below. These ideas are also presented in a table at the bottom of this section. \subsection*{Elder health system} \addcontentsline{toc}{subsection}{Elder health system} Elderly people are often subject to poor health. For this reason it is important to detect and avoid any harm that may come to them in their homes. In many cases, the application of sensors may achieve this goal. Sound or pressure sensors may be installed into walls, floor and stairs to register a fall accident. Even better, sensor devices placed on the body of the person could register abnormal heart rate, blood pressure or body temperature. Moreover, sensors can also be used to create a panic button. Sensor data would be collected at a central hub, and then transmitted to a standby team that would react to the event. Microphones placed in the room or on the person might be used to create a communication channel directly to the standby team. Panic buttons are already widely used. There are also systems for both video monitoring and sound monitoring of your home. More advanced systems actually include motion sensors, contact sensors for doors and windows, and pressure sensors with a user-friendly web service for presenting data. \subsection*{High risk work environment} \addcontentsline{toc}{subsection}{High risk work environment} Oil platforms, construction sites, or mines are high risk work environments. These rely on strict security routines to keep workers safe. There are various applications for sensors in such an environment. A small device carried by the workers or attached to their helmets could emit a sound whenever the worker enters a “hazard area”. When an area needs to be cleared of workers, such sensors could signal if a person enters the area, preventing incidents. The sensors would need to have a large range, and be noise resistant. There are currently no electronic systems in place in high risk work environments. However, the systems that are in place use statistics, training and safer work practices. While our system would enable the workers to feel safer, it would only add another layer of safety. It is debatable whether or not the customer would pay for the extra layer when training still is mandatory, and many security measures are already in effect. The other constraint of this system is its accuracy. We may not be able to guarantee the consistent up-time and accuracy the high risk work environment needs. \subsection*{Exploring} \addcontentsline{toc}{subsection}{Exploring} Cheap sensors can be an aid when exploring unknown or hazardous environments. The sensors can be spread over an area to provide data on temperature, humidity, and similar before human explorers go in. The area of exploration could be deep sea, caverns, or even space. Robots are already widely used in exploration. In deep sea ROVs are used. NASA has also conducted research on robots for space exploration. \subsection*{Smart Home/Office, or the smart room} \addcontentsline{toc}{subsection}{Smart Home/Office, or the smart room} Our original idea for the project, and one of the ways that this concept can be applied in real world situations, is through a smart room. A smart room would have a panel where one would be able to adjust settings, such as temperature and humidity in the room. The system would then adjust this by interacting with other devices in the room, such as air conditioning and humidifiers. Sunlight could be blocked by automatic curtains, or the curtains could open in the morning letting the sunlight in. This is a fairly young field of study and application, but it is growing fast. Already there are solutions for smart home systems on the market, but they cost quite a lot in comparison to what our system would cost. Most of these systems also handle only one part of the house automation, such as the security system or entertainment system. Our system would include all of the systems of the house. However, it would be difficult to implement and it would require universal plugins from all the other devices, such as air conditioning, which severely limits its modifiability. \subsection*{Initial Ideas Table} \addcontentsline{toc}{subsection}{Initial Ideas Table} Here we will present the ideas discussed above in a more concrete way, in the form of a table: \begin{table}[H] \caption{Initial Ideas Table} \centering \begin{tabularx}{\textwidth}{ |X|X|X|} \hline \textbf{Initial Idea} & \textbf{Pros} & \textbf{Cons} \\ \hline Elder health system & Health tracking on the fly \newline \ \newline Fast response in case of an emergency.newline \newline \ \newline Cheap. & More advanced systems already exist, including motion sensors and more detailed and advanced statistics tracking. \newline \ \newline Panic buttons are already widely used and it is a well established market. \\ \hline High risk work environment & Prevent work-related accidents by alerting the workers if they are walking into a dangerous zone, or if there is danger in the vicinity and they need to evacuate. \newline \ \newline No need for complex visual and audio systems. \newline \ \newline Cheap. & There are other effective non-electronic methods in place and it would provide just another layer of safety, which may not be that effective in comparison. \newline \ \newline We may not be able to guarantee full accuracy with the sensors, which would diminish their purpose. \\ \hline Exploration & Would act as a cheap way of detecting hazards in unexplored areas. & There are far more sophisticated, if more expensive, robots and methods created for exploration. \\ \hline Smart Room & Very cheap in comparison to the current systems. \newline \ \newline Would be able to control multiple aspects of the home, not just one. \newline \ \newline Very easy to use solution for the customer with a simple user interface. & Most other devices that would connect to the system would have to have universal plugins for the system. This system would then become either very expensive or much less modifiable. \\ \hline \end{tabularx} \end{table} \subsection*{Visualization Ideas} \addcontentsline{toc}{subsection}{Visualization Ideas} While the above ideas deal with what we can make of the project, our customer has clearly stated that what they want is visualization of data and a basic product that can be portable and modifiable for the future. Therefore our second main goal was to think of ideas on how to visualize the data we receive from the sensors. On this topic we have thought of a few ideas that are listed below. This is the actual overarching goal of the project. Moreover, the ideas above are representations of the possible future of the project. A table at the end of this section is also included as a summary of these ideas, including their pros and cons. \subsection*{Simple Image Manipulation} \addcontentsline{toc}{subsection}{Simple Image Manipulation} Our first and simplest idea for visualization would include a simple screen representation of all the data combined and averaged out. The background would be the temperature, which would change colour depending on how hot or cold it is in the room. The intensity of the colour would change with the amount of light in the room. And transparent layers would be added in the form of simple geometrical shapes to represent other factors, such as sound, humidity, and pressure. This idea is very simple to implement, however, it does not show data per sensor. It shows data as an average, which while it is good for some environments is unacceptable for others. Therefore this idea was quickly abandoned in favour of the next idea. \subsection*{Simple Image Manipulation in a grid} \addcontentsline{toc}{subsection}{Simple Image Manipulation in a grid} Instead of having one image that would take all the data and average out the calculations, we would have a grid where a sensor or a group of sensors would provide the same image manipulation but on a grid scale. While the grid solves the problems of whether individual sensors or a group of sensors are showing, it introduces another major problem. How to make the grid portable and modifiable. This was solved by using grid in a fairly “loose” meaning. Instead of it meaning a grid in its strict sense, the grid would be a sphere around the sensor. In this fashion the system can visualize a sensors immediate surrounding as well as average out values that overlap with the spheres. While this adds a layer of complexity to our implementation, we believe that it still remains fairly simple to implement. \subsection*{Manipulation of sprites} \addcontentsline{toc}{subsection}{Manipulation of sprites} Instead of using geometrical shapes or simple colours, we can use small pre-made images commonly known as sprites. The system works in quite the same fashion. Every sensor or groups of sensors would be presented by a sprite, which would change based on the changes in the environment. Sprites may be a bit more intuitive than colours and geometrical shapes, as they can speak more than the shapes can. For example a water sprite rising or lowering could represent the change in humidity. On the other hand it becomes more complex and time consuming to draw the sprites and even potentially animate them. Sprites also may have difficulty representing data that changes slowly, like temperature, and the change might not even become apparent that easily. \subsection*{Combination of sprites and shapes} \addcontentsline{toc}{subsection}{Combination of sprites and shapes} Since sprites may not be able to represent the entire spectrum of data well enough, we could use a combination of sprites and shapes to represent data. Sprites could represent changes that happen more quickly in the environment, such as sound recognition or pressure, while temperature could be represented by colours. The problem with this approach is that it becomes less intuitive and more cluttered the more we add. While we could make the approach more complex, complexity often hurts intuitiveness. As a result, we would have to carefully balance the two in order to produce a good visual representation. \subsection*{Visualization Ideas Table} \addcontentsline{toc}{subsection}{Visualization Ideas Table} Here we will present the ideas discussed above in the form of a table: \begin{table}[H] \caption{Visualization Ideas Table} \centering \begin{tabularx}{\textwidth}{ |X|X|X|} \hline \textbf{Visualization idea} & \textbf{Pros} & \textbf{Cons} \\ \hline Simple Image Manipulation & Very simple to implement. & Not enough data per sensor is shown. \newline \ \newline No overarching image of what sensors senses what attribute and the value of the attribute. \\ \hline Simple Image Manipulation in a grid & Simple to implement. \newline \ \newline Shows each sensor (or small groups of sensors) and shows their specific attributes. & Requires accurate motion tracking to represent the sensors in a 2D environment. \\ \hline Manipulation of sprites & Adds intuitiveness. \newline \ \newline Shows each sensor as a sprite and animates the sprite according to the data. & The need to animate and draw sprites may be costly in the long run. \newline \ \newline Some attributes, such as temperature, might not be that intuitive, as they will change very slowly. \\ \hline Combination of sprites and shapes & Shows each sensors data clearly and well. & Complex implementation. \newline \ \newline Need to animate and draw sprites may be costly in the long run. \newline \ \newline Too many sprites and shapes might cause clutter on the screen and actually make it harder to understand. \\ \hline \end{tabularx} \end{table} \subsection*{Similar Products and Projects} \addcontentsline{toc}{subsection}{Similar Products and Projects} \subsubsection*{Internet of Things} The Internet of Things is about connecting devices in our everyday environments to the internet. Many novel uses of internet-connected devices have already been demonstrated, and the future looks promising. It is expected that we will have 100 billion internet-connected objects by 2020. The possibilities might seem limitless. Moreover, the cheap sensor technology supplied by Altran is highly applicable in this area, and such sensors opens up many interesting opportunities. \subsubsection*{Products and Projects} While there are similar projects that deal with the concept of “Internet of Things”, typically these projects are not limited to the visualization stage. They become a concrete project that uses data to achieve some effect besides visualization. Therefore, most of the products are internal projects that become full fledged projects or products later. It is difficult to find any products that match what we are trying to achieve and most visualizations of data are well out of our field of study, such as weather systems that create cloud or wind patterns and predict their movement. \subsubsection*{Different possible architectural models for the prototype} While working on the initial stages of the project, we had to prepare countermeasures to counteract risk 1, which was the possibility of us not getting the hardware necessary to complete our prototype. In order to make the prototype work without any hardware from the customer, he suggested that we could substitute the sensors with android phones. With this in mind we had to prepare for both situations and we found out that the architecture in both cases actually differs. In both cases we would be using a client-server architecture, however, the roles of the client and the server would be reversed. When using the sensors, we would have a raspberry pi acting as a server. In this case our client requests information from the server, and stores it in a database of some sort. Then our visualizer uses this data to show it in a visual way. The image of the architecture can be seen below. \begin{figure}[H] \centering \includegraphics[width=\textwidth]{Architecture_1.png} \caption{Architecture 1} \end{figure} The architecture that deals with the android phones is the opposite. The phones are sensors, but also servers at the same time, and we have only one client, that requests information from the phones, or servers. This information is stored in some sort of a database, and the same process is repeated as above with the visualizer. The image of the architecture can be seen below. \begin{figure}[H] \centering \includegraphics[width=\textwidth]{Architecture_2.png} \caption{Architecture 2} \end{figure} The two architectures are both client server, but they are quite different as can be seen from the images. They are both worth mentioning, as they will influence our implementation based on the hardware provided. \subsection*{Tools} \addcontentsline{toc}{subsection}{Tools} Below we will list the tools we will use to create our project. This encompasses both the hardware that we will be using as well as major programming languages or protocols that we will use to complete this project. The full list of tools used can be found in appendix A. \subsection*{Hardware} \addcontentsline{toc}{subsection}{Hardware} As stated before, we need to prepare for both architectures. The hardware then encompasses both the sensors and the android phones. We will also note down the DASH7 standard for wireless networking, which is used by the sensors. \subsubsection*{Sensors} We were given a short demonstration of the sensor capabilities by the customer on our first meeting. We know that the sensors are able to detect temperature, pressure, humidity and light. It also has a built in accelerometer. The sensors also have two led lights, one green and one red, that can be remotely turned on or off, although they are not very interesting in our project, the lights can be used in different projects. There is also a large (comparative to the sensor) antenna mounted at the end of the sensor. The sensors use DASH7 standard (explained below in more detail) to connect to a central hub of some sort. From a client we are then able to view and extract this information. The sensors are quite small in size, even with an antenna they are not more than 15 centimetres long and 5 centimetres wide, with a thickness (including the antenna) of about 1-2 centimetres. The sensors are quite cheap to manufacture as well. The sensors should be provided to us by the customer when we start working on the implementation. \subsubsection*{Android phone} In the case that we would not receive sensors, we could use android phones. An android phone works very similar to our sensors. The sensors that exist on the android phone can measure light, ambient temperature, pressure, humidity, proximity, gravity and acceleration to the device. The data gathering process would be done through an application, commonly also referred to as an app, that would run in the background, listen for requests and reply with data. This app would be written Java, as it is the main programming language for android phones. Except for humidity, all of the other sensors were implemented as of android 2.3. \subsubsection*{DASH7} DASH7 is a RFID-standard for wireless networking, that operates on 433 Mhz unlicensed SRD band. It provides long battery life, up to 2 km connection distance, low latency with moving objects and 1 metre accuracy. It is frequently used in the military. No one in our group is familiar with DASH7, therefore we needed to spend at least some time studying it’s capabilities. It is not an overly important part of the system, after the only concern that we have with it is in between the sensors and the central hub, the part of the system we will not modify in any fashion. Still it is worth mentioning as a separate part of hardware, since it demonstrates the network capability of the sensors in comparison to standard WiFi networking and also demonstrates further the capabilities of the sensors. \subsection*{Software} \addcontentsline{toc}{subsection}{Software} In this project we will mainly be using Java as a programming language. We will also use LaTeX for formatting our presentation into a more presentable and clean document. \subsubsection*{Java} Java is a high level, class-based, object-oriented programming language, introduced in 1995. There are two major advantages for using Java in our system. The first one is that the entire development team is familiar with the Java programming language and we do not have to spend time learning to program in Java. The second advantage is in it’s class-based, object-oriented design. It allows us to work in pairs on classes, and allows classes to synchronize better with each other, so it makes programming easier. Not only that, but Java also has some very good ways of implementing a user interface, such as JavaFX, which will help us visualize the project easier. Since visualization is the main goal, Java is, we feel, an excellent choice of programming language for this project. \subsubsection*{LaTeX} LaTeX is a document markup language. LaTeX is widely used in document preparation, as it has many functions that improve the readability of the document, as well as many automatic functions such as generation of a table of contents and chapter numbering. It also has a clean and consistent way of presenting tables as well as symbols for any kind of mathematical formula. At least half of our team is familiar with LaTeX and it is a markup language that is fairly easy to learn, in our opinion, so it will not constrain our time further. In fact, with the automatic table and diagram marking it will make our document much more readable and presentable and that is why we choose to use it. \end{document} @inproceedings{pmlr-v48-mniha16, title = {Asynchronous Methods for Deep Reinforcement Learning}, author = { and and and and and and and }, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {1928--1937}, year = {2016}, editor = { and }, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/mniha16.pdf}, url = {http://proceedings.mlr.press/v48/mniha16.html}, }The {\module{mathdata}} module contains the code for representing types, terms and proofs and the code for type checking and proof checking. This is arguably the most important module in Qeditas. A bug in this module could lead to non-theorems being accepted as theorems undermining the primary purpose of the Qeditas system. Fortunately the {\file{mathdata.ml}} file is not long (currently less than 1500 lines of code) and depends very little on the rest of Qeditas (using only code for serialization and cryptographic hashing). It is intended to satisfy the {\defin{de Bruijn criterion}} in that the code can be manually audited to determine its correctness. We attempt to give enough information in this chapter for someone who wishes to undertake such an audit. The original version of the code for this module was taken from the code for the Egal system~\cite{Brown2014}, but has since undergone extensive changes. One major difference in the syntax is the explicit support for type variables in Qeditas. Support for theories and signatures have also been added, and the type of documents has been modified (adding support for importing signatures and declaring conjectures but removing all presentation level items). Additionally, the checking functions are parameterized by functions to verify a term identified only by its hash root has a type in a theory and to verify a proposition identified only by its hash root is known to be a theorem in a theory. Such information will be looked up in the ledger tree (see Chapter~\ref{chap:ctre}) by checking what is held at corresponding term addresses. Finally, a significant portion of the Egal proof checking code was apparently intended to avoid expanding definitions unnecessarily. This code has been deleted and replaced by simpler code to expand all definitions during proof checking. One might argue that it would be safer to use an older, established proof checker. However, experience has shown that even established systems can be vulnerable to ``tricks'' which can be used to prove what should be a non-theorem. For example, on {\tt{proofmarket.org}}~\cite{ProofMarket} a bitcoin bounty was placed on the proposition {\sf{False}} in Coq~\cite{Coq:manual}. In spite of the fact that Coq is an advanced tool used by many people for many projects, such a ``proof'' of {\sf{False}} was given.\footnote{In fact, two different proofs were given.} The ``proofs'' were related to implementation issues rather than an inconsistency in the underlying logic, but only the implementation will matter in a system like Qeditas. By using a simple underlying logic (simple type theory) and isolating the implementation in the reasonably small module {\module{mathdata}} it is hoped that such apparent inconsistencies can be avoided. The underlying logic is a form of simple type theory~\cite{Church40} with support for prefix polymorphism. The basic proof calculus is natural deduction~\cite{gent36,praw65} with Curry-Howard style $\lambda$-terms proof terms~\cite{howa80}. This leads the type checker and proof checker in {\module{mathdata}} to be very similar to the oldest proof checker, AUTOMATH~\cite{DeBruijn80}. The logic is designed to allow for multiple theories to be declared and for signatures to be used to import previous typed terms and proven propositions. Of the popular proof assistants at the moment, the closest would probably be Isabelle~\cite{Nipkow-Paulson-Wenzel:2002}, although Isabelle follows the LCF style~\cite{GORDON79} instead of Curry-Howard. {\bf{Note:}} Unit tests for the {\module{mathdata}} module are in {\file{mathunittests.ml}} in the {\dir{src/unittests}} directory in the {\branch{testing}} branch. These unit tests give a number of examples demonstrating how the functions described below should behave. The {\branch{testing}} branch is, however, out of date with the code in the {\branch{dev}} and {\branch{master}} branches. A few examples of types, terms and proof terms used in these unit tests are in {\file{unittestsaux.ml}} in the same directory. Likewise, examples of publications (encoded versions of documents released with Egal~\cite{Brown2014}) are in {\file{testpubs1.ml}} and {\file{testpubs2.ml}} in the same directory. {\bf{Note:}} The Coq module {\coqmod{MathData}} is intended to correspond to {\module{mathdata}}, except that the checking code is omitted and left abstract. \section{Simple Types} Simple types ($\alpha$, $\beta$) are described by the following grammar: $$ \alpha,\beta ::= \delta_n |o|\iota_n|(\alpha\to\beta)|(\Pi \alpha) $$ We treat $\to$ as right associative to omit parentheses. For example, $\iota_0\to\iota_0\to o$ means $(\iota_0\to (\iota_0\to o))$. Also, we will omit parentheses in $\Pi \alpha$ since $\Pi$ will always be used above $\to$ and so no ambiguity can result. Simple types are implemented as the inductive type {\type{tp}}. We describe each constructor: \begin{itemize} \item ${\mbox{\constr{TpVar}}}(n)$ means the type variable $\delta_n$, where the $n$ should be interpreted as a de Bruijn index~\cite{deBruijn72}. For example, $\Pi \Pi \delta_1 \to \delta_0 \to \delta_1$ means the type of a function which expects two types $\alpha$ and $\beta$, a term of type $\alpha$, a term of type $\beta$ and returns a term of type $\alpha$. \item {\constr{Prop}} means the type $o$ of propositions. \item ${\mbox{\constr{Base}}}(n)$ means the $n^{th}$ base type $\iota_n$. Only finitely many base types will be explicitly used in a theory. In fact, so far only theories using one base type $\iota_0$ have been considered, but the support for multiple base types is included in case it is needed later. \item ${\mbox{\constr{TpAll}}}(\alpha)$ means $\Pi \alpha$, binding a type variable. Only types of the form $\Pi\cdots\Pi\alpha$ where $\alpha$ has no occurrence of a $\Pi$ will be used in practice. \end{itemize} The functions {\serfunc{seo\_tp}} and {\serfunc{sei\_tp}} serialize and deserialize types. % {\func{tp\_to\_str}} returns a string representation of the type and {\func{str\_to\_tp}} returns a type given a string representation of the type.\footnote{These are included to help with testing, and are not currently used outside {\module{mathdata}} otherwise.} {\func{hashtp}} takes a type and returns a hash value obtained by serializing the type to a string, hashing the string, and then hashing the result tagged with $64$. (The intention of hashing tagged results is to ensure that, for example, the hash value associated with a type will not accidentally be the same as the the hash value associated with a term, proof or anything else.) \section{Terms and Propositions} Terms $s,t,u$ are described by the following grammar: $$ s,t ::= x_n |\tmh{h}|c_n|(st)|(\lambda_\alpha s)|(s\to t)|(\forall_\alpha s)|(s \alpha)|(\Lambda s)|(\tforall s) $$ Here $n$ ranges over non-negative integers and $h$ ranges over hash values. Terms $x_n$ are variables, where $n$ should be interpreted as a de Bruijn index~\cite{deBruijn72}. For example, $\lambda_o x_0 \to \forall_o x_1\to x_0$ would be written as $\lambda y:o . y\to\forall z:o.y\to z$ in a named representation. A term $\tmh{h}$ is an abbreviation for a term which has $h$ as its hash root (see {\func{tm\_hashroot}} below). Note that there are two kinds of application: (1) $(st)$ of a term $s$ to a term $t$ and (2) $(s\alpha)$ of a term $s$ to a type $\alpha$. Likewise there are two abstractions and two universal quantifiers: one for the term level and one for the type level. First, $(\lambda_\alpha s)$ is a term level abstraction representing a function expecting an input of type $\alpha$ with return value determined by this input and $s$. Likewise, $(\forall_\alpha s)$ corresponds to universally quantifying over the elements of type $\alpha$. On the other hand, $(\Lambda s)$ is a type level abstraction and represents a function which expects a type $\alpha$ and then returns a value determined by $\alpha$ and $s$. Likewise, $(\tforall s)$ corresponds to universally quantifying over all types. We refer to $\lambda_\alpha$, $\forall_\alpha$, $\Lambda$ or $\tforall$ collectively as {\defin{binders}} and say the term $s$ in $(\lambda_\alpha s)$, $(\forall_\alpha s)$, $(\Lambda s)$ or $(\tforall s)$ is in the {\defin{scope}} of the binder. We often omit parentheses. Application is assumed to be left associative and so $s\alpha\beta t u$ means $((((s\alpha)\beta)t)u)$ If parenthesis around the body of a binder are omitted, then they are assumed to be such that the scope of the binder is as large as possible. For example, $\forall_o x_0\to x_0$ means $(\forall_o (x_0\to x_0))$. The corresponding type in the OCaml code is {\type{tm}}. We describe each constructor: \begin{itemize} \item ${\mbox{\constr{DB}}}(n)$ corresponds to the variable $x_n$ (i.e., the de Bruijn index). \item ${\mbox{\constr{TmH}}}(h)$ corresponds to the term $\tmh{h}$ and should be considered an abbreviation (which is sometimes opaque and sometimes transparent, depending on the current signature). \item ${\mbox{\constr{Prim}}}(n)$ corresponds to the primitive $c_n$. \item ${\mbox{\constr{Ap}}}(s,t)$ corresponds to term level application $st$. \item ${\mbox{\constr{Lam}}}(\alpha,s)$ corresponds to term level abstraction $\lambda_\alpha s$. \item ${\mbox{\constr{Imp}}}(s,t)$ corresponds to implication $s\to t$. \item ${\mbox{\constr{All}}}(\alpha,s)$ corresponds to term level universal quantification $\forall_\alpha s$. \item ${\mbox{\constr{TTpAp}}}(s,\alpha)$ corresponds to type level application $s\alpha$. \item ${\mbox{\constr{TTpLam}}}(s)$ corresponds to type level abstraction $\Lambda s$ \item ${\mbox{\constr{TTpAll}}}(s)$ corresponds to type level universal quantification $\tforall s$. \end{itemize} The functions {\serfunc{seo\_tm}} and {\serfunc{sei\_tm}} serialize and deserialize terms. % {\func{tm\_to\_str}} returns a string representation of the term and {\func{str\_to\_tm}} returns a term given a string representation of the type.\footnote{These are included to help with testing, and are not currently used outside {\module{mathdata}} otherwise.} There are two functions {\func{hashtm}} and {\func{tm\_hashroot}} which take terms and return a corresponding hash value. In the case of {\func{hashtm}}, a hash value is obtained by serializing the term to a string, hashing the string, and then hashing the result tagged with $66$. This (effectively) guarantees that different terms will always be given different hash values. On the other hand, {\func{tm\_hashroot}} takes a term and computes its {\defin{hash root}}. The hash root of a term does not distinguish between a term $\tmh{h}$ and a term $t$ which has $h$ as its hash root. In effect, {\func{tm\_hashroot}} views all such abbreviations as transparent. The {\defin{hash root}} $\tmhr{t}$ of a term $t$ can be defined as follows: \begin{itemize} \item $\tmhr{\tmh{h}}$ is $h$ \item $\tmhr{c_n}$ is the hash of $n$ tagged with $96$. \item $\tmhr{x_n}$ is the hash of $n$ tagged with $97$. \item $\tmhr{st}$ is the hash of the hashed pair of $\tmhr{s}$ and $\tmhr{t}$ tagged with $98$. \item $\tmhr{\lambda_\alpha s}$ is the hash of the hashed pair of the hash of $\alpha$ and $\tmhr{s}$ tagged with $99$. \item $\tmhr{s\to t}$ is the hash of the hashed pair of $\tmhr{s}$ and $\tmhr{t}$ tagged with $100$. \item $\tmhr{\forall_\alpha s}$ is the hash of the hashed pair of the hash of $\alpha$ and $\tmhr{s}$ tagged with $101$. \item $\tmhr{s\alpha}$ is the hash of the hashed pair of $\tmhr{s}$ and the hash of $\alpha$ tagged with $102$. \item $\tmhr{\Lambda s}$ is the hash of $\tmhr{s}$ tagged with $103$. \item $\tmhr{\tforall s}$ is the hash of $\tmhr{s}$ tagged with $104$. \end{itemize} The reader can verify that this corresponds to the definition of {\func{tm\_hashroot}} in the code. The tags are used to record which term constructor was traversed and is also used to ensure that hash roots of terms are not the hash values computed in other contexts. A proposition is a certain kind of term (in a given context). In short, propositions are always of the form $\tforall \cdots \tforall t$ where $t$ has type $o$. Usually a proposition is simply of the form $t$ where $t$ has type $o$. \section{Proof Terms} Proof terms $\cD,\cE$ are described by the following grammar: $$ \cD,\cE ::= \gpa{h} | \hyp{n} | \known{h} | (\cD s) | (\cD \cE) | (\lambda_s \cD) | (\lambda_\alpha \cD) | (\cD \alpha) | (\Lambda \cD) $$ Here $n$ ranges over non-negative integers and $h$ ranges over hash values. We sometimes simply say ``proofs'' instead of ``proof terms.'' The proof term $\gpa{h}$ is an abbreviation for a proof term which has hash root $h$ (see {\func{pf\_hashroot}} below). The proof term $\hyp{n}$ is the proof of a hypothesis (in a hypothesis context). The proof term $\known{h}$ simply asserts that the proposition with hash root $h$ is known. (The current signature maintains a list of known propositions and their hash root. Inclusion of such a proposition in the signature may require checking that the term address corresponding to $h$ is owned as a proposition in the ledger. The only way this could have happened is if the term is the axiom of the current theory or was previously proven.) There are three kinds of application and three kinds of abstractions. At the proof level there are applications $(\cD\cE)$ and abstractions $(\lambda_s \cD)$. These correspond to the elimination and introduction rules for implication. At the term level there are applications $(\cD t)$ and abstractions $(\lambda_\alpha\cD)$. These correspond to the elimination and introduction rules for universal quantification. Finally at the type level there are applications $(\cD\alpha)$ and abstractions $(\Lambda\cD)$. Type level application is the way polymorphic known propositions are applied at specific types. Type level abstraction is the way polymorphic propositions are proven. As with terms, we omit parentheses assuming application associates to the left and assuming abstraction (binders) have as large a scope as possible. The corresponding type in the OCaml code is {\type{pf}}. We describe each constructor: The functions {\serfunc{seo\_pf}} and {\serfunc{sei\_pf}} serialize and deserialize proof terms. % {\func{pf\_to\_str}} returns a string representation of the proof term and {\func{str\_to\_pf}} returns a proof term given a string representation of the type.\footnote{These are included to help with testing, and are not currently used outside {\module{mathdata}} otherwise.} Again, there are two functions taking a proof term and returning a hash value: {\func{hashpf}} and {\func{pf\_hashroot}}. The function {\func{hashpf}} takes a term and returns a hash value obtained by serializing the term to a string, hashing the string, and then hashing the result tagged with $67$. This implies {\func{hashpf}} returns an effectively unique hash value for each proof term. The function {\func{pf\_hashroot}} computes a {\defin{hash root}} similar to the way hash roots for terms are computed. In this case, the hash root for a proof term abbreviation $\gpa{h}$ is $h$. \section{Publications} There are three kinds of publications: theories, signatures and documents. A theory declares the types of some primitives $c_n$ and gives some axioms. A signature is to be interpreted within a given theory and is intended to make some terms and propositions accessible for use within another publication (a document or another signature). A signature declares some parameters (opaque terms of the form $\tmh{h}$) giving the hash root and the simple type, declares some definitions and declares some propositions to be known (either axioms of the theory or previously proven propositions). A document is similar to a signature except proofs of theorems are also allowed. In addition, a document may declare a proposition to be a conjecture. For all three kinds of publications there is a representation as a list of ``items.'' This list is perhaps best thought of as being in reverse order. The idea is that after one has processed the ``rest'' of the list, then one has sufficient information to process the ``head'' of the list. In practice there is a distinction between the specification of a theory and the theory itself. The same is true of signatures. In essence a theory specification or signature specification corresponds to a list of declarations, where a theory or signature itself is a ``compiled'' format which other publications may used. This ``compiled'' format must be stored by every node in order to check later publications. For this reasons, Qeditas currency units must be burned in order to publish a theory or publication. In particular, $21$ zerms must be burned for each byte in the serialized representation of the theory or signature. The idea behind a fee of $21$ zerms is that since there is an upper bound of $21$ million fraenks ($21$ billion zerms) we can be sure that no more than $1$ GB worth of theories and signatures will ever be published. \subsection{Theories} A {\defin{theory item}} is one of the following: \begin{itemize} \item a declaration of a primitive to have type $\alpha$, \item a declaration of a definition of type $\alpha$ defined by a term $s$, or \item a declaration of proposition $s$ as an axiom. \end{itemize} A {\defin{theory specification}} is a list of theory items. A {\defin{theory}} $\cT$ is a pair $(\cP,\cA)$ of a list $\cP$ of simple types $\alpha_0,\ldots,\alpha_{n-1}$ and a list $\cA$ of hash values $\overline{h}$. The idea is that the primitive $c_i$ has the type $\alpha_i$ for $ij$. This corresponds to the ``removal'' of the variable $\delta_j$ during the substitution. \item $\dbsub{j}{\beta}{(\alpha_1\to\alpha_2)} = (\dbsub{j}{\beta}{\alpha_1})\to (\dbsub{j}{\beta}{\alpha_2})$ \item $\dbsub{j}{\beta}{(\Pi \alpha)} = (\Pi (\dbsub{j+1}{\beta}{\alpha}))$ \item $\dbsub{j}{\beta}{\alpha} = \alpha$ otherwise. \end{itemize} \item {\func{tmtpsubst}} defines $\dbsub{j}{\beta}{s}$ for terms $s$ and types $\beta$. The defining cases are: \begin{itemize} \item $\dbsub{j}{\beta}{(st)} = (\dbsub{j}{\beta}{s}) (\dbsub{j}{\beta}{t})$. \item $\dbsub{j}{\beta}{(\lambda_\alpha t)} = \lambda_{\dbsub{j}{\beta}{\alpha}} (\dbsub{j}{\beta}{t})$. The term level $\lambda_\alpha$ binder does not affect which type variables are locally bound. \item $\dbsub{j}{\beta}{(s\to t)} = (\dbsub{j}{\beta}{s}) \to (\dbsub{j}{\beta}{t})$. \item $\dbsub{j}{\beta}{(\forall_\alpha t)} = \forall_{\dbsub{j}{\beta}\alpha} (\dbsub{j}{\beta}{t})$. The term level $\forall_\alpha$ binder does not affect which type variables are locally bound. \item $\dbsub{j}{\beta}{(s\alpha)} = (\dbsub{j}{\beta}{s}) \alpha$. \item $\dbsub{j}{\beta}{(\Lambda t)} = \Lambda (\dbsub{j+1}{\beta}{t})$. The type level $\Lambda$ binder means one more type variable is locally bound. \item $\dbsub{j}{\beta}{(\tforall t)} = \tforall (\dbsub{j+1}{\beta}{t})$. The type level $\tforall$ binder means one more type variable is locally bound. \item $\dbsub{j}{\beta}{t} = t$ otherwise. In particular, $\dbsub{j}{\beta}{\tmh{h}} = \tmh{h}$. Assuming $h$ is the hash root of a term $s$ where every type variable is locally bound, then $\dbsub{j}{\beta}{s} = s$ and $h$ is still the term root of $\dbsub{j}{\beta}{s}$. \end{itemize} \item {\func{tmsubst}} defines $\dbsub{j}{u}{s}$ for terms $s$ and $u$. The defining cases are: \begin{itemize} \item $\dbsub{j}{u}{x_j} = \dbsh{0}{j}{u}$. In the special case where $j=0$ we know $\dbsh{0}{j}{u} = u$ and so we can simply take $\dbsub{0}{u}{x_0} = u$. \item $\dbsub{j}{u}{x_i} = x_{i-1}$ if $i>j$. This corresponds to the ``removal'' of the variable $x_j$ during the substitution. \item $\dbsub{j}{u}{(st)} = (\dbsub{j}{u}{s}) (\dbsub{j}{u}{t})$. \item $\dbsub{j}{u}{(\lambda_\alpha t)} = \lambda_\alpha (\dbsub{j+1}{u}{t})$. The term level $\lambda_\alpha$ binder makes one more term variable locally bound. \item $\dbsub{j}{u}{(s\to t)} = (\dbsub{j}{u}{s}) \to (\dbsub{j}{u}{t})$. \item $\dbsub{j}{u}{(\forall_\alpha t)} = \forall_\alpha (\dbsub{j+1}{u}{t})$. The term level $\forall_\alpha$ binder makes one more term variable locally bound. \item $\dbsub{j}{u}{(s\alpha)} = (\dbsub{j}{u}{s}) \alpha$. \item $\dbsub{j}{u}{(\Lambda t)} = \Lambda (\dbsub{j}{u}{t})$. The type level $\Lambda$ binder does not change which term level variables are locally bound. \item $\dbsub{j}{u}{(\tforall t)} = \tforall (\dbsub{j}{u}{t})$. The type level $\tforall$ binder does not change which term level variables are locally bound. \item $\dbsub{j}{u}{t} = t$ otherwise. In particular, $\dbsub{j}{u}{\tmh{h}} = \tmh{h}$. Assuming $h$ is the hash root of a term $s$ where every term variable is locally bound, then $\dbsub{j}{u}{s} = s$ and $h$ is still the term root of $\dbsub{j}{u}{s}$. \end{itemize} \end{itemize} Next we need to say what it means for a type variable $\delta_j$ to be {\defin{free}} in a type or term, and what it means for a term variable $x_j$ to be {\defin{free}} in a term. There are three relevant definitions: \begin{itemize} \item The function {\func{free\_tpvar\_in\_tp\_p}} determines if a type variable $\delta_j$ is {\defin{free}} in a type $\alpha$. The definition is by recursion on $\alpha$ and the $j$ must be increased by 1 in the $\Pi$ binder case to account for the new locally bound variable. \item The function {\func{free\_tpvar\_in\_tm\_p}} determines if a type variable $\delta_j$ is {\defin{free}} in a term $t$. The definition is by recursion on $t$ and the $j$ must be increased by 1 in the $\Lambda$ and $\tforall$ binder cases. \item The function {\func{free\_in\_tm\_p}} determines if a term variable $x_j$ is {\defin{free}} in a term $t$. The definition is by recursion on $t$ and the $j$ must be increased by 1 in the $\lambda_\alpha$ and $\forall_\alpha$ binder cases. \end{itemize} We can now turn to $\beta\eta$-normalization. We begin by considering four kinds of {\defin{redexes}} and their corresponding {\defin{reducts}}. Normalization is performed by reducing redexes to their reducts until no more redexes remain. A theorem of various type theories is that normalization terminates in a unique normal form for well-typed terms, and that is true for the type theory under consideration here. \begin{itemize} \item A term of the form $(\lambda_\alpha s)t$ is a {\defin{term level $\beta$-redex}} with reduct $\dbsub{0}{t}{s}$. \item A term of the form $(\Lambda s)\alpha$ is a {\defin{type level $\beta$-redex}} with reduct $\dbsub{0}{\alpha}{s}$. \item A term of the form $\lambda_\alpha (s x_0)$ where $x_0$ is not free in $s$ is a {\defin{term level $\eta$-redex}} with reduct $\dbsh{0}{-1}{s}$. (The shift of term variables by $-1$ is required since $s$ was in the scope of one term level binder $\lambda_\alpha$ which is removed from the reduct.) \item A term of the form $\Lambda (s \delta_0)$ where $\delta_0$ is not free in $s$ is a {\defin{type level $\eta$-redex}} with reduct $\dbtpsh{0}{-1}{s}$. (The shift of type variables by $-1$ is required since $s$ was in the scope of one type level binder $\Lambda$ which is removed from the reduct.) \end{itemize} A term is {\defin{normal}} if it has no redexes. The function {\func{tm\_norm\_p}} checks if a term is normal. In theory specifications, signature specifications and documents all definitions, knowns, conjectures and theorems are required to be normal. The exception {\exc{NonNormalTerm}} is raised if this requirement is violated. The normalization procedure is {\func{tm\_beta\_eta\_norm}}. It proceeds by repeatedly calling {\func{tm\_beta\_eta\_norm\_1}}. In simple terms, the function {\func{tm\_beta\_eta\_norm\_1}} recursively traverses a term reducing each redex it finds and returning the reduced term along with a boolean indicating if at least one reduction was performed. If no reductions were performed, then the term is normal and the procedure ends. In certain examples, $\beta$-normalization (and hence $\beta\eta$-normalization) leads to large terms and can require an unrealistic number of $\beta$-reductions. This problem is dealt with by having resource bounds represented internally by {\var{beta\_count}} and {\var{term\_count}}. Before a signature specification or document is checked, these resource bounds are reset (by {\func{reset\_resource\_limits}}). (Checking a theory specification requires no $\beta$-reductions.) There are $200,000$ beta reductions and $10$ million term traversal steps allowed per signature specification or document. Each $\beta$-reduction step decrements {\var{beta\_count}} by one. If {\var{beta\_count}} reaches $0$, then the exception {\exc{BetaLimit}} is raised. For each recursive call of a shift or substitution function decrements {\var{term\_count}}. If {\var{term\_count}} reaches $0$, then the exception {\exc{TermLimit}} is raised. If either of these exceptions are thrown, it essentially means that checking the signature specification or document is too resource intensive and it cannot be published in its current form. If this occurs, a possible solution is to factor the publication into multiple publications. It is worth noting that some well-known ill-typed terms do not have a normal form. For example, $(\lambda_o x_0 x_0) (\lambda_o x_0 x_0)$ is a term level $\beta$-redex with itself as a reduct. Without resource bounds, calling {\func{tm\_beta\_eta\_norm}} with this term would result in an infinite loop. With the resource bound, {\exc{BetaLimit}} would be raised. In practice, {\func{tm\_beta\_eta\_norm}} should never be called with such a term since during the checking of publications {\func{tm\_beta\_eta\_norm}} is only called with terms which are already known to be well-typed. We write $\benorm{s}$ for the $\beta\eta$-normal form of $s$, assuming it exists. \section{Type Checking and Proof Checking} We now turn to the most important functions: those which check that a type is valid, check that a term has a type, check that a term is a proposition, and check that a proof term is a proof of a proposition. Checking attempts typically either succeed (possibly returning some information) or raise an exception. The exception {\exc{CheckingFailure}} is raised if checking fails. One of the exceptions {\exc{BetaLimit}} or {\exc{TermLimit}} is raised if one of the corresponding resource bounds is reached. All the properties defined will be relative to a {\defin{type context}}. Since type variables are represented as de Bruijn indices, the {\defin{type context}} can be taken to simply be a non-negative integer $v$. We say a type $\alpha$ is {\defin{valid as a simple type}} in type context $v$ (and write $v\vdash \alpha \stp$) if it contains no occurrence of $\Lambda$ and every type variable $\delta_i$ satisfies $isubfigure.sty %% %% This is file `subfigure.sty', %% generated with the docstrip utility. %% %% The original source files were: %% %% subfigure.dtx (with options: `package') %% %% Copyright (C) 1988-1995 . %% %% This file is NOT the source for subfigure, because almost all comments %% have been stripped from it. It is NOT the preferred form of subfigure %% for making modifications to it. %% %% Therefore you can NOT redistribute and/or modify THIS file. You can %% however redistribute the complete source (subfigure.dtx and %% subfigure.ins) and/or modify it under the terms of the GNU General %% Public License as published by the Free Software Foundation; either %% version 2, or (at your option) any later version. %% %% The subfigure package is distributed in the hope that it will be %% useful, but WITHOUT ANY WARRANTY; without even the implied warranty %% of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the %% GNU General Public License for more details. %% %% You should have received a copy of the GNU General Public License %% along with this program; if not, write to the Free Software %% Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% @LaTeX-style-file{ %% Author = "", %% Version = "2.0", %% Date = "1995/03/06", %% Time = "14:43:14", %% Filename = "subfigure.sty", %% Address = "Digital Mapping Laboratory, School of Computer Science %% Carnegie-Mellon University, 5000 Forbes Avenue %% Pittsburgh, PA 15213-3891, USA", %% Telephone = "(412) 268-5654", %% FAX = "(412) 268-5576", %% Email = " (Internet)", %% CodeTable = "ISO/ASCII", %% Keywords = "LaTeX2e, float, figure, table", %% Supported = "yes", %% Abstract = "LaTeX package for providing support for the %% inclusion of small, `sub,' figures and tables. It %% simplifies the positioning, captioning and %% labeling of them within a single figure or table %% environment. In addition, this package allows %% such sub-captions to be written to the List of %% Figures or List of Tables if desired." %% } %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \ifx\if@compatibility\undefined\else \NeedsTeXFormat{LaTeX2e} \ProvidesPackage{subfigure}[1995/03/06 v2.0 subfigure package] \typeout{Package: subfigure 1995/03/06 v2.0} \fi \newif\ifsubcaphang \newif\ifsubcapcenter \newif\ifsubcapcenterlast \newif\ifsubcapnooneline \newcommand{\subfigtopskip}{10pt} \newcommand{\subfigbottomskip}{10pt} \newcommand{\subfigcapskip}{10pt} \newcommand{\subfigcapmargin}{10pt} \newcommand{\subcapsize}{} \newcommand{\subcaplabelfont}{} \newcounter{subfigure}[figure] \def\thesubfigure{(\alph{subfigure})} \newcommand{\@thesubfigure}{{\subcaplabelfont\thesubfigure}\space} \let\p@subfigure\thefigure \let\ext@subfigure\ext@figure \newcommand{\l@subfigure}{% \@dottedxxxline{\ext@subfigure}{2}{3.9em}{2.3em}} \newcounter{lofdepth} \setcounter{lofdepth}{1} \newcounter{subtable}[table] \def\thesubtable{(\alph{subtable})} \newcommand{\@thesubtable}{{\subcaplabelfont\thesubtable}\space} \let\p@subtable\thetable \let\ext@subtable\ext@table \newcommand{\l@subtable}{% \@dottedxxxline{\ext@subtable}{2}{3.9em}{2.3em}} \newcounter{lotdepth} \setcounter{lotdepth}{1} \ifx\if@compatibility\undefined \subcaphangfalse \subcapcenterfalse \subcapcenterlastfalse \def\subcapsize{\footnotesize} \else \DeclareOption{normal}{% \subcaphangfalse \subcapcenterfalse \subcapcenterlastfalse \subcapnoonelinefalse} \DeclareOption{hang}{\subcaphangtrue} \DeclareOption{center}{\subcapcentertrue} \DeclareOption{centerlast}{\subcapcenterlasttrue} \DeclareOption{nooneline}{\subcapnoonelinetrue} \DeclareOption{isu}{\ExecuteOption{hang}} \DeclareOption{anne}{\ExecuteOption{centerlast}} \DeclareOption{scriptsize}{\renewcommand{\subcapsize}{\scriptsize}} \DeclareOption{footnotesize}{\renewcommand{\subcapsize}{\footnotesize}} \DeclareOption{small}{\renewcommand{\subcapsize}{\small}} \DeclareOption{normalsize}{\renewcommand{\subcapsize}{\normalsize}} \DeclareOption{large}{\renewcommand{\subcapsize}{\large}} \DeclareOption{Large}{\renewcommand{\subcapsize}{\Large}} \DeclareOption{up}{\renewcommand{\subcaplabelfont}{\upshape}} \DeclareOption{it}{\renewcommand{\subcaplabelfont}{\itshape}} \DeclareOption{sl}{\renewcommand{\subcaplabelfont}{\slshape}} \DeclareOption{sc}{\renewcommand{\subcaplabelfont}{\scshape}} \DeclareOption{md}{\renewcommand{\subcaplabelfont}{\mdseries}} \DeclareOption{bf}{\renewcommand{\subcaplabelfont}{\bfseries}} \DeclareOption{rm}{\renewcommand{\subcaplabelfont}{\rmfamily}} \DeclareOption{sf}{\renewcommand{\subcaplabelfont}{\sffamily}} \DeclareOption{tt}{\renewcommand{\subcaplabelfont}{\ttfamily}} \ExecuteOptions{normal,footnotesize} \ProcessOptions \fi \newcommand{\subfigure}{% \bgroup \advance\csname c@\@captype\endcsname\@ne \refstepcounter{sub\@captype}% \leavevmode \@ifnextchar [% {\@subfloat{sub\@captype}}% {\@subfloat{sub\@captype}[\@empty]}} \let\subtable\subfigure \def\@subfloat#1[#2]#3{% \setbox\@tempboxa \hbox{#3}% \@tempdima=\wd\@tempboxa \vtop{% \vbox{ \vskip\subfigtopskip \box\@tempboxa}% \ifx \@empty#2\relax \else \vskip\subfigcapskip \@subcaption{#1}{#2}% \fi \vskip\subfigbottomskip}% \egroup} \newcommand{\@subfigcaptionlist}{} \newcommand{\@subcaption}[2]{% \begingroup \let\label\@gobble \def\protect{\string\string\string}% \xdef\@subfigcaptionlist{% \@subfigcaptionlist,% {\protect\numberline {\@currentlabel}% \noexpand{\ignorespaces #2}}}% \endgroup \@nameuse{@make#1caption}{\@nameuse{@the#1}}{#2}} \newcommand{\@makesubfigurecaption}[2]{% \setbox\@tempboxa \hbox{% \subcapsize \ignorespaces #1% \ignorespaces #2}% \@tempdimb=-\subfigcapmargin \multiply\@tempdimb\tw@ \advance\@tempdimb\@tempdima \hbox to\@tempdima{% \hfil \ifdim \wd\@tempboxa >\@tempdimb \subfig@caption{#1}{#2}% \else\ifsubcapnooneline \subfig@caption{#1}{#2}% \else \box\@tempboxa \fi\fi \hfil}} \let\@makesubtablecaption\@makesubfigurecaption \newcommand{\subfig@caption}[2]{% \ifsubcaphang \sbox{\@tempboxa}{% \subcapsize \ignorespaces #1}% \addtolength{\@tempdimb}{-\wd\@tempboxa}% \usebox{\@tempboxa}% \subfig@captionpar{\@tempdimb}{#2}% \else \subfig@captionpar{\@tempdimb}{#1#2}% \fi} \newcommand{\subfig@captionpar}[2]{% \parbox[t]{#1}{% \strut \ifsubcapcenter \setlength{\leftskip}{\@flushglue}% \setlength{\rightskip}{\@flushglue}% \setlength{\parfillskip}{\z@skip}% \else\ifsubcapcenterlast \addtolength{\leftskip}{0pt plus 1fil}% \addtolength{\rightskip}{0pt plus -1fil}% \setlength{\parfillskip}{0pt plus 2fil}% \fi\fi \subcapsize \ignorespaces #2% \par}} \newcommand{\@dottedxxxline}[6]{% \ifnum #2>\@nameuse{c@#1depth}\else \@dottedtocline{0}{#3}{#4}{#5}{#6} \fi} \let\subfig@oldcaption\@caption \long\def\@caption#1[#2]#3{% \subfig@oldcaption{#1}[{#2}]{#3}% \@for \@tempa:=\@subfigcaptionlist \do {% \ifx\@empty\@tempa\relax \else \addcontentsline {\@nameuse{ext@sub#1}}% {sub#1}% {\@tempa}% \fi}% \gdef\@subfigcaptionlist{}} \endinput %% %% End of file `subfigure.sty'. Common Lagrangian in introductory courses of classical mechanics depends of $q$ and $\dot{q}$. We refer to them as \emph{first-order Lagrangian} because the highest derivative term is $\dot{q}$, i.e.\ the first derivative of $q(t)$ with respect to time. If in the Lagrangian appears also $\ddot{q}$ it will be a \emph{second-order Lagrangian} and so on. \\ In the following, the mathematical machinery develop in Section~\ref{section: lagrangin and hamiltonian formalism}, will be used to study two apparently similar Lagrangian and introduce the core problem of HD systems. \subsection{First-order vs Second-order Lagrangian} The simplest non-trivial Lagrangians are those in polynomial form like \begin{equation*} L_1(q, \dot{q}) = \frac{\dot{q}^2}{2} - \alpha(q) \qquad L_2(q, \ddot{q}) = \frac{\ddot{q}^2}{2} - \alpha(q) \end{equation*} where $\alpha$ is a smooth function of $q$ only~\cite{Chen13}. No Greek indices appear on $q^{(i)}$ because we limit ourself to one-dimensional systems. $L_1$ is a first-order Lagrangian while $L_2$ it's a second-order one. \paragraph{First-order Lagrangian} In order to get the equation of motion we used the Euler-Lagrange equation~\eqref{eq:euler-lagrange} \begin{equation} \label{eq: first-order_motion_eq_lagr} \frac{\partial L_1}{\partial q} - \frac{d}{dt}\frac{\partial L_1}{\partial \dot{q}} + \underbrace{ \frac{d^2}{dt^2}\frac{\partial L_1}{\partial \ddot{q}} - \ldots }_0 = 0 \qquad \Rightarrow \qquad \ddot{q} = - \frac{d\alpha(q)}{dq} \end{equation} If $q$ is the position coordinate of a unit mass point, the function $\alpha(q)$ can be interpreted as the potential energy and~\eqref{eq: first-order_motion_eq_lagr} is nothing less than Newton's second law. To get the Hamiltonian description apply the recipe in Section~\ref{subsection: hamiltonian_formalism} \begin{equation*} \begin{cases} Q_1 := q \\ P_1 := \frac{\delta L_1}{\delta \dot{q}} = \dot{q} \end{cases} \qquad \tilde{H}_1(q, \dot{q}) := \frac{\delta L_1}{\delta \dot{q}} \dot{q} - L_1(q, \dot{q}) \end{equation*} \begin{equation} \label{eq: first-order_motion_eq_ham} H_1(Q_1, P_1) = \frac{P_1^2}{2} + \alpha(Q_1) \qquad \begin{cases} \dot{Q_1} = \frac{\partial H_1}{\partial P_1} = P_1 \\ \dot{P_1} = - \frac{\partial H_1}{\partial Q_1} = - \frac{d\alpha(Q_1)}{dQ_1} \end{cases} \end{equation} Since $L_1$ does not explicitly depend on time (only through $q$ and $\dot{q}$), the Hamiltonian $H_1$ can be interpreted as the total energy of the system (sum of kinetic and potential energy). Combining the two equation in~\eqref{eq: first-order_motion_eq_ham} we restore~\eqref{eq: first-order_motion_eq_lagr} proving the equivalence between Lagrangian and Hamiltonian mechanics for simple systems describe by $L_1$. \paragraph{Second-order Lagrangian} Proceeding as before, equation of motion follow from Euler-Lagrange equation~\eqref{eq:euler-lagrange} \begin{equation} \label{eq: second-order_motion_eq_lagr} \frac{\partial L_2}{\partial q} - \frac{d}{dt}\frac{\partial L_2}{\partial \dot{q}} + \frac{d^2}{dt^2}\frac{\partial L_2}{\partial \ddot{q}} - \underbrace{ \frac{d^3}{dt^3}\frac{\partial L_2}{\partial q^{(3)}} + \ldots }_0 = 0 \quad \Rightarrow \quad q^{(4)} = - \frac{d\alpha(q)}{dq} \end{equation} Equation~\eqref{eq: second-order_motion_eq_lagr} never appear in classical mechanics, but can it describe some kind of system? The study of the Hamiltonian can give us some clues about the total energy of the system. \begin{equation*} \begin{cases} Q_1 := q \\ P_1 := \frac{\delta L_2}{\delta \dot{q}} = \frac{\partial L_2}{\partial \dot{q}} - \frac{d}{dt} \left( \frac{\partial L_2}{\partial \ddot{q}} \right) = 0 - \frac{d}{dt} \left( \ddot{q} \right) = -q^{(3)} \end{cases} \begin{cases} Q_2 := \dot{q} \\ P_2 := \frac{\delta L_2}{\delta \ddot{q}} = \frac{\partial L_2}{\partial \ddot{q}} = \ddot{q} \end{cases} \end{equation*} Therefore \begin{equation}\label{eq:canonical_coordinates_second_order_lagrangian} \begin{cases} q = Q_1 \\ \dot{q} = Q_2 \\ \ddot{q} = P_2 \end{cases} \qquad \begin{cases} \frac{\delta L_2}{\delta \dot{q}} = P_1 \\ \frac{\delta L_2}{\delta \ddot{q}} = P_2 \end{cases} \end{equation} \begin{equation*} \tilde{H}_2(q, \dot{q}) := \frac{\delta L_2}{\delta \dot{q}} \dot{q} + \frac{\delta L_2}{\delta \ddot{q}} \ddot{q} - L_2(q, \ddot{q}) = \frac{\delta L_2}{\delta \dot{q}} \dot{q} + \frac{\delta L_2}{\delta \ddot{q}} \ddot{q} - \frac{\ddot{q}^2}{2} + \alpha(q) \end{equation*} Substituting relations~\eqref{eq:canonical_coordinates_second_order_lagrangian} in the expression for $\tilde{H}_2$ we get the Hamiltonian \begin{align} \label{eq: second-order_motion_eq_ham} H_2(Q_1, Q_2, P_1, P_2) &= P_1 Q_2 + P_2 P_2 - \frac{P_2^2}{2} + \alpha(Q_1) \notag \\ &= P_1 Q_2 + \frac{P_2^2}{2} + \alpha(Q_1) \end{align} There is a significant difference between the two spectra of $H_1$ and $H_2$: the first is bounded from below while the latter is not (and both are not bounded from above). \subsection{Linear Ostrogradskian instability}~\label{subsection: linear_ostrogradskian_instability} Consider an \emph{isolated} system describe by the Hamiltonian $H_2$; if the energy is conserve even though the spectra is not bounded the energy stay constant. Things start to going wrong when ones consider \emph{interacting} system (e.g. $H_1$-system interacting with $H_2$-system). The phase space of $H_2$-system infinitely extend where the Hamiltonian is negative. In the example this is due to the \emph{linear} term $P_1$. When two systems interact the $H_1$-system tends to occupy higher and higher energies states while the $H_2$-system fall lower and lower in negative energies; this behavior entails the conservation of energy~\cite{Kallosh08, Eliezer89}. This is the so called \emph{linear Ostrogradskian instability}. \\ One can encounter the same kind of instability in quantum mechanics when try to canonically quantized higher order Hamiltonian (e.g. $H_2$). This process lead to \emph{negative norm states} (or \emph{negative energy states}) which are often called \emph{``ghosts''}. Systems (even classical ones) involving this type of ghosts are called \emph{ghost-like} or \emph{ghost-ridden} systems. \\ One may wonder when these ghosts are summoned; the answer is contains in the Ostrogradsky theorem \begin{theorem}[Ostrogradsky]\label{th:ostrogradsky_classical} If in the Lagrangian~\eqref{eq:general_lagrangian} $n \geq 2$ and the canonical momentum $\bm{P}_n$ does not vanish, the Hamiltonian~\eqref{eq:general_hamiltonian} may acquire an arbitrary real value. \end{theorem} \begin{proof} The proof is given for a one-dimensional system (i.e. $\mu=1$) but it is generalize straight to higher dimensions. Consider the Hamiltonian~\eqref{eq:general_hamiltonian} in terms of conjugate coordinates \begin{equation} \label{eq:general_hamiltonian_1dim} H = P_n h + P_{n-1} Q_n + \cdots + P_1 Q_2 - L(Q_1, Q_2, \ldots, h) \end{equation} The function $h$ express $q^{(n)}$ in terms of conjugate coordinates under the assumption of regular Lagrangian, i.e. \begin{equation*} q^{(n)} = h(Q_1, \ldots, Q_n, P_n) \end{equation*} Observe that $h$ does not depend on $P_1, \ldots, P_{n-1}$ so the only contribution of these momenta to the Hamiltonian is in the linear factor \begin{equation*} P_{n-1} Q_n + \cdots + P_1 Q_2 \end{equation*} which is linear in momenta and hence can acquire any real value and so does the Hamiltonian~\eqref{eq:general_hamiltonian_1dim}. \end{proof} For many years, due to consequences of Theorem~\ref{th:ostrogradsky_classical}, HD theories are consider intrinsically sick and not worthy of further study. However a deeper investigation reveals that Ostrogradsky instability can be cure and HD theories should not be discard in the first instance. One way to exorcising Ostrogradsky's ghost is through imposition of constraints~\cite{Chen13}. \subsection{Removing Ostrogradskian instability with constraints} In Hamiltonian mechanics, \emph{constraints} are relation between coordinates and conjugate momenta. Constraints are classify in two group: \emph{first class constraints} and \emph{second class constraints}. Second class constraints can be thought as ``physical'' (e.g.\ train on a railway, particle on a plane) so that the solutions of equations of motion are different with or without them. First-class are instead related to a gauge freedom: the solutions of equations of motion differ by a function of time and hence represent the same system. Moreover the Poisson Bracket of a first class constraint with all the other constraints vanishes on the constraints surface in the phase space (i.e. $\{\phi, \tilde{\phi}\} \approx 0$, see later for the notation used). Another independent classification can be made for constraints (and not confused with the previous one): \emph{primary constraints} and \emph{secondary constrains}. Primary constraints are relation between $Q$s and $P$s which are given at the beginning. \begin{equation} \label{eq:constraint} \phi_1(Q, P) = 0 \end{equation} As the name suggests, secondary constraints are derived from the primary ones imposing the conservations of the constraints during the evolution of the system. This generates a series of constraints where the constraints relations are called \emph{consistency relations}. \begin{equation} \label{eq:consistency_relations} \phi_1 \approx 0 \quad \Rightarrow \quad \left\{ \phi_1 , H \right\} =: \phi_2 \approx 0 \quad \Rightarrow \quad \left\{ \phi_2 , H \right\} =: \phi_3 \approx 0 \quad \Rightarrow \quad \ldots \end{equation} The weak equality symbol ``$\approx$'' has been used to highlight the fact that these relations vanish only on the hypersurface where all constraints are satisfied. When we will be said that a constrain $\phi_i$ \emph{vanishes} it is to be intended in this weak sense. \subsubsection{Lagrangian and Hamiltonian with $m$ auxiliary variables} Constraints in Hamiltonian mechanics can be imposed with the help of auxiliary variables $\lambda_i$ in the Lagrangian: \begin{equation*} L = L( q, \dot{q}, \ddot{q}, \ldots, q^{(n)}, \lambda_1, \lambda_2, \ldots, \lambda_m ) \end{equation*} So the Lagrangian, besides the equations of motion~\eqref{eq:euler-lagrange}, need to satisfy the constraints equation \begin{equation} \frac{\partial L}{\partial \lambda_i} = 0 \qquad i=1, 2, \ldots, m \end{equation} Canonical coordinates have to be given also for $\lambda_i$ and the following choice can be done \begin{equation} \label{eq:def_canonical_coordinates_lambda} \Lambda_{i}:= \lambda_{i} \quad \leftrightarrow \quad \Pi_{i} := \frac{\delta L}{\delta \dot{\lambda_{i}}} = 0 \qquad i = 1, 2, \ldots, m \end{equation} so the primary constraints in~\eqref{eq:constraint} simply become $\phi_{1, i}: \Pi_i = 0$. To the Hamiltonian~\eqref{eq:Ham_in_q} are now added the auxiliary variables terms and it becomes \begin{equation} \label{eq:Ham_constraints_in_q} \tilde{H} := \sum_{j=1}^{n} \frac{\delta L}{\delta q^{(j)}} q^{(j)} + \sum_{i=1}^{m} \frac{\delta L}{\delta \dot{\lambda_i}} \dot{\lambda_i} - L(q, \ldots, q^{(n)}, \lambda_1, \ldots, \lambda_m) \end{equation} The $\dot{\lambda_i}$ can be obtain by the consistency relations~\eqref{eq:consistency_relations} but there is no need to explicitly calculate them. $\dot{\lambda_i}$ can be written as functions of canonical coordinates: $\dot{\lambda_i} = u_i(Q_1, \ldots, Q_n, P_1, \ldots, P_n)$. Expressing~\eqref{eq:Ham_constraints_in_q} using canonical coordinates we obtain the Hamiltonian \begin{align} \label{eq:Ham_constraints} H =\ & P_n h + P_{n-1} Q_n + \cdots + P_1 Q_2 \nonumber \\ & + \phi_m u_m + \phi_{m-1} u_{m-1} + \cdots + \phi_1 u_1 \nonumber \\ & - L ( Q_1, Q_2, \ldots, h, \Lambda_1, \ldots, \Lambda_m) \end{align} After calculating the secondary constraints with~\eqref{eq:consistency_relations} \begin{equation*} \phi_{2, i} := \left\{ \phi_{1,i}, H \right\} = \frac{\partial \phi_{1,i}}{\partial \Pi_i} \frac{\partial H}{\partial \Lambda_i} = \frac{\partial L}{\partial \lambda_i} \Big|_{\lambda_i = \Lambda_i} \approx 0 \end{equation*} two cases are possible: \begin{enumerate} \item $\{\phi_{1,i}, \phi_{2,j}\} \not\approx 0$: $\phi_{1,i}$, $\phi_{2,j}$ are both second-class constraints thus there are no further constraints derivable from consistency relations. \item $\{\phi_{1,i}, \phi_{2,j}\} \approx 0$: $\phi_{1,i}$, $\phi_{2,j}$ are not second-class constraints then further constraints can be found employing the chain~\eqref{eq:consistency_relations} until either \emph{case 1} is reached or the new constraint can be expressed using those found previously. \end{enumerate} After all constraints have been found and the gauge symmetries are fixed, one can use these equations to write the auxiliary canonical variables in terms of the ``usual'' canonical variables \begin{equation} \label{eq:constraints_in_terms_of_canonical_vars} \begin{cases} \Lambda_i =\ f_i(Q_1, \ldots, Q_n, P_n) \\ \Pi_i =\ 0 \end{cases} \qquad i = 1, 2, \ldots, m \\ \end{equation} and then substitute in the Hamiltonian~\eqref{eq:Ham_constraints} \begin{equation*} H =\ P_n h + P_{n-1} Q_n + \cdots + P_1 Q_2 - L ( Q_1, Q_2, \ldots, h, f_1, \ldots, f_m) \end{equation*} To recap, we started form a phase space of dimension $2n$ (n$Q$ + n$P$), then we enlarged it by adding $2m$ auxiliary variables (m$\Lambda$ + m$\Pi$) so the total phase space dimension is $2(n+m)$. Then we associated with every pair of canonical auxiliary variables two constraints (\emph{case 1}) or more (\emph{case 2}) have been found. During the last substitution the dimension of phase space is bring down to it's original dimensions (\emph{case 1}: $2(n+m) - 2m = 2n$) or further reduced (\emph{case 2}: e.g. $2(n+m) - 3m < 2n$). \subsubsection{Example of non-stable system with constraints} We introduce here a widely studied HD Lagrangian: the \emph{Pais-Uhlenbeck oscillator} (PU). \begin{equation} \label{eq:lagrangian_PU} L_{PU} = \frac{1}{2} \left[ \ddot{q}^2 - (\omega_1^2 + \omega_2^2) \dot{q}^2 + \omega_1^2 \omega_2^2 q^2 \right] \end{equation} According to Ostrogradsky Theorem~\ref{th:ostrogradsky_classical} this system shows a problematic instability. Then we can try to cure that by introducing a constraint such as $\ddot{q}^2 - \dot{q}^2 = 0$ and see if the Hamiltonian is still two-side unbounded. Using the auxiliary variable $\lambda$ one can write the constraint version of~\eqref{eq:lagrangian_PU}: \begin{equation} \label{eq:lagrangian_PUC} L_{PUC} = L_{PU} + \frac{\lambda}{2} \left(\ddot{q}^2 - \dot{q}^2\right) \end{equation} and from it we derive the Hamiltonian following the given prescription. \begin{align*} & \begin{cases} Q_1 := q \\ P_1 := \frac{\delta L_{PUC}}{\delta \dot{q}} = - (\lambda + 1) q^{(3)} - \dot{\lambda}\ddot{q} - (\lambda + \omega_1^2 + \omega_2^2)\dot{q} \end{cases} \\ & \begin{cases} Q_2 := \dot{q} \\ P_2 := \frac{\delta L_{PUC}}{\delta \ddot{q}} = (\lambda + 1) \ddot{q} \quad \Rightarrow \quad \ddot{q} = \frac{P_2}{(\Lambda + 1)} \end{cases} \\ & \begin{cases} \Lambda := \lambda \\ \Pi := \frac{\delta L_{PUC}}{\delta \lambda} = 0 \quad \Rightarrow \quad \phi_1: \Pi \approx 0 \end{cases} \end{align*} \begin{equation} \label{eq:ham_PUC_with_aux_vars} H_{PUC} =\ P_1 Q_2 + \frac{P_2^2}{2(\Lambda + 1)} + \phi_1 u_1 + \frac{Q_2^2}{2} \left(\Lambda + \omega_1^2 + \omega_2^2 \right) - \frac{Q_1^2}{2} \omega_1^2 \omega_2^2 \end{equation} \begin{equation} \label{eq:secondary_constraint_PU} \phi_2 := \left\{ \phi_1, H_{PUC} \right\} = \frac{1}{2} \left[ \frac{P_2}{\Lambda + 1} - Q_2 \right] \left[ \frac{P_2}{\Lambda + 1} + Q_2 \right] \approx 0 \end{equation} The secondary constraint equation~\eqref{eq:secondary_constraint_PU} vanish if one of the factors in square brackets vanish. The two square bracket represent two hypersurfaces and one solution is chosen instead of the other based on initial conditions. In the following, we choose the bracket with the minus sign. Moreover note that $\phi_1$ and $\phi_2$ are already second-class constraints, indeed \begin{equation*} \begin{cases} \phi_1: \Pi = 0 \\ \phi_2: \frac{P_2}{\Lambda + 1} - Q_2 \approx 0 \end{cases} \quad \Rightarrow \quad \left\{ \phi_1, \phi_2 \right\} = - \frac{\partial\phi_1}{\partial\Pi} \frac{\partial\phi_2}{\partial\Lambda} = \frac{P_2}{{(\Lambda + 1)}^2} \not\approx 0 \end{equation*} so no further constraints are needed (\emph{case 1}). The equation for $\phi_1$ (trivial) and for $\phi_2$ can be locally inverted~\eqref{eq:constraints_in_terms_of_canonical_vars} obtaining that $\Pi = 0$ and $\Lambda = P_2/Q_2 - 1$. Substituting them in~\eqref{eq:ham_PUC_with_aux_vars} we get the final Hamiltonian of the system but unfortunately the Ostrogradsky instability persists. \begin{equation} \label{eq:ham_PUC_instable} H_{PUC} =\ P_1 Q_2 + P_2 Q_2 + \frac{Q_2^2}{2} \left(\omega_1^2 + \omega_2^2 - 1 \right) - \frac{Q_1^2}{2} \omega_1^2 \omega_2^2 \end{equation} \subsubsection{Example of stable system with constraints} Now consider the following constrained Pais-Uhlenbeck oscillator~\cite{Chen13} \begin{equation} L_{PUC} = L_{PU} + 4 \omega_1^2\omega_2^2 q^2 \lambda (1+ \lambda) + 2 \sqrt{2} \omega_1\omega_2 \lambda q \ddot{q} \end{equation} Analogously to the previous example, canonical variable and hamiltonian $H_{PUC}$ can be derived in similar fashion \begin{align*} & \begin{cases} Q_1 := q \\ P_1 := - q^{(3)} - \left( 2\sqrt{2} \omega_1\omega_2\lambda + \omega_1^2 + \omega_1^2 \right) \dot{q} - 2\sqrt{2}\omega_1\omega_2\dot{\lambda} q \end{cases} \\ & \begin{cases} Q_2 := \dot{q} \\ P_2 := \ddot{q} + 2\sqrt{2} \omega_1\omega_2\lambda q \quad \Rightarrow \quad \ddot{q} = P_2 - 2\sqrt{2} \omega_1\omega_2\Lambda Q_1 \end{cases} \\ & \begin{cases} \Lambda := \lambda \\ \Pi := 0 \quad \Rightarrow \quad \phi_1: \Pi \approx 0 \end{cases} \end{align*} \begin{align} \label{eq:ham_PUC_with_aux_vars_stable} H_{PUC} =\ & P_1 Q_2 + \frac{P_2^2}{2} - \omega_1^2\omega_2^2\frac{Q_1^2}{2} + \left( \omega_1^2 + \omega_2^2 \right) \frac{Q_2^2}{2} \\ & - 4 \omega_1^2\omega_2^2\Lambda Q_1^2 - 2\sqrt{2}\omega_1\omega_2\Lambda P_2 Q_1 + \phi_1 u_1 \end{align} Secondary constraints relations carry useful information up to the forth order \footnote{ To obtain $\phi_4$, information from $\phi_1$ and $\phi_2$ have been used in order to write $\phi_4$ in terms of $\Lambda$. } \begin{align*} & \phi_1 : \Pi = 0 \\ & \phi_2 := \{\phi_1, H_{PUC}\} = P_2 + \sqrt{2} \omega_1\omega_2 Q_1 \approx 0 \\ & \phi_3 := \{\phi_2, H_{PUC}\} = P_1 + \left(\omega_1^2 + \omega_2^2 - \sqrt{2}\omega_1\omega_2 \right) Q_2 \approx 0 \\ & \phi_4 := \{\phi_3, H_{PUC}\} = \omega_1 \omega_2 (3+8\Lambda) - \sqrt{2} \left( \omega_1^2 + \omega_2^2 \right) \left(1 + 2\Lambda \right) \approx 0 \end{align*} These consistency relations can be inverted so that $P_2$, $Q_2$ and $\Lambda$ are express in terms of $P_1$ and $Q_1$. Substituting them into~\eqref{eq:ham_PUC_with_aux_vars_stable} obtaining \begin{equation} \label{eq:ham_PUC_stable} H_{PUC} =\ \frac{\omega_1^2\omega_2^2}{2} Q_1^2 + \frac{\omega_1\omega_2}{\sqrt{2} {\left(\sqrt{2} \omega_1\omega_2 - \omega_1^2 - \omega_2^2 \right)}^2} P_1^2 \end{equation} The Hamiltonian~\eqref{eq:ham_PUC_stable} is bounded from below (and not from above, just like an harmonic oscillator) and hence this system does not suffer of Ostrogradsky instability unlike the previous example. \\ The main difference between the two is the role played by constraints. In the first example the dimensions of phase space of the unconstrained and constrained Hamiltonian are both four (i.e.\ four variable appear in the reduced Hamiltonian~\eqref{eq:ham_PUC_instable}: $Q_1$, $P_1$, $Q_2$ and $P_2$). In the second example the constraints reduce the dimensionality of the phase space from the original four to two (indeed just $Q_1$ and $P_1$ appear in~\eqref{eq:ham_PUC_stable}). As shown in~\cite{Chen13} a reduction of the phase space, due to imposed constraints, can cure the typical Ostrogradsky instability of HD theories. \documentclass{article} \usepackage[utf8]{inputenc} \usepackage{pgfplots,multicol} \usepackage{tikz-qtree} \usepackage{url} \usepackage{hyperref} \usepackage{xcolor} \usepackage{avm} \usepackage{rtrees} \usepackage{forest} \usepackage{rotating, graphicx} \useforestlibrary{linguistics} \newcommand{\comment}[1]{} \forestapplylibrarydefaults{linguistics} \hypersetup{ colorlinks = true, %Colours links instead of ugly boxes urlcolor = red, %Colour for external hyperlinks linkcolor = blue, %Colour of internal links citecolor = blue %Colour of citations } \pgfplotsset{compat=newest} \mathchardef\period=\mathcode`. \newcommand{\textarray}[1]{\ensuremath{\left[ \mbox{\ttfamily\begin{tabular}{l} #1 \end{tabular}}\right]}} \begin{document} \title{566 Midterm} \author{ \tt {}} \date{11/08/2019} \maketitle \section{Chapter 10, Problem 1} For the examples (i) - (viii) the treatment of passives sketched in the text does not always correctly predict the grammar. Our grammar will improperly predict example (v) since it will incorrectly fail the Anaphoric Agreement Principle (AAP). For the examples that include the active subject of the passive verb(e.g. by the doctor) I have included it in the ARG-ST usually with the index j. \subsection{She was introduced to herself (by the doctor).} Based on our text, this example is properly predicted as a grammatical sentence. As we can see below in the ARG-ST, in this sentence, the anaphora(herself) NP is outranked by its co-indexed element(She) which satisfies Principle A of the AAP. Moreover, in using the Passive Lexical Rule, the PP is resolved, and the Case Constraint and the Binding Theory comes into play. \\ \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[PP \\ \[INDEX & i \\CASE & {\it acc} \\MODE & {\it ana} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & j \\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \subsection{*She was introduced to her (by the doctor).} Based on our text, this example is properly predicted as an ungrammatical sentence. As we can see below in the ARG-ST, in this sentence, the co-indexed element(her) it is a [MODE ref] element that is co indexed with another element that outranks it(the first NP on the list, she). Consequently, the co-indexing indicated is not permitted because its a violation of Principle B of the AAP.\\ \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[PP \\ \[INDEX & i \\CASE & {\it acc} \\MODE & {\it ref} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & j \\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \subsection{The barber was shaved (only) by himself} Based on our text, this example is properly predicted as a grammatical sentence. As we can see below in the ARG-ST, in this sentence, the anaphora(himself) NP is outranked by its co-indexed element(barber) which satisfies Principle A of the Anaphoric Agreement Principle (AAP).\\ \begin{avm} \< shaved , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[ PP \\ \[FORM & {\it by} \\ INDEX & i \\ MODE & {\it ana} \] \] \avmr \> \] \> \end{avm} \subsection{*The barber was shaved (only) by him} Based on our text, this example is properly predicted as an ungrammatical sentence. As we can see below in the ARG-ST, in this sentence, the co-indexed element(him) it is a [MODE ref] element that is co indexed with another element that outranks it(the first NP on the list, barber). Consequently, the co-indexing indicated is not permitted based on Principle B of the AAP. \\ \begin{avm} \< shaved , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[ PP \\ \[FORM & {\it by} \\ INDEX & i \\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \subsection{The students were introduced to other (by Leslie)} Based on our text this, example is improperly predicted as an ungrammatical sentence. As we can see below in the ARG-ST, in this sentence, the co-indexed element(other) it is a [MODE ref] element that is co indexed with another element that outranks it(the first NP on the list, students). Consequently, the co-indexing indicated is not permitted based on Principle B of the AAP which will rule this sentence as ungrammatical despite it being grammatical in real English. \\ \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[PP \[INDEX & i \\ CASE & {\it acc} \\MODE & {\it ref} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & j \\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \subsection{*The students were introduced to them (by Leslie)} Based on our text this example is properly predicted as an ungrammatical sentence. As we can see below in the ARG-ST, in this sentence, the co-indexed element(them) it is a [MODE ref] element that is co indexed with another element that outranks it(the first NP on the list, students). Consequently, the co-indexing indicated is a violation of Principle B of the AAP. \\ \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[PP \[INDEX & i \\CASE & {\it acc} \\MODE & {\it ref} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & j \\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \subsection{Kim was introduced to Larry by himself} Based on our text, this example is properly predicted as a grammatical sentence. As we can see below in the ARG-ST, in this sentence, the anaphora(himself) NP is outranked by its co-indexed element(Larry) which satisfies Principle A of the Anaphoric Agreement Principle (AAP). \\ \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[NP_j \[CASE & {\it acc} \\MODE & {\it ref} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & j \\ MODE & {\it ana} \] \] \avmr \> \] \> \end{avm} \subsection{*Kim was introduced to himself by Larry.} Based on our text this example is properly predicted as an ungrammatical sentence. As we can see below in the ARG-ST, in this sentence, the co-indexed element(himself) it is a [MODE ana] element that is co indexed with another element that it outranks(the last NP on the list, Larry). Consequently, the co-indexing indicated is not permitted based on Principle B and Principle Aof the AAP. \begin{avm} \< introduced , \[ ARG-ST & \< \avml \[NP_i \\ \[MODE & {\it ref} \] \] , \[NP_j \[CASE & {\it acc} \\MODE & {\it ana} \]\] , \[ PP \\ \[FORM & {\it by} \\ INDEX & ij\\ MODE & {\it ref} \] \] \avmr \> \] \> \end{avm} \section{Chapter 10, Problem 3} These underline once again the need for a theory of lexical irregularity and exceptions to lexical rules \subsection{Is this alternation productive?} I do not believe this alternation is productive because it opens the flood gates for the grammar to license sentences as grammatical that are not. As mentioned in the text, there are only certain verbs particularly in the idiomatic form that break this alteration this alteration still introduces a non zero amount of exceptions that need to be dealt with elsewhere in the grammar. If instead of having these alterations we maintain two separate types of datives(and two forms of most verbs in our lexicon) we do see a likely large growth in our lexicon size(since most verbs would work under this alteration) but we do not have to add secondary and terciary measure to deal with the exceptions and ensure they do not license wrong sentences. Some examples of the incorrectly licensed alterations would be: \\ 1. The IRS audited a return to Trump. \\ 2. * The IRS audited Trump a return. \\ Example below refers to cat the linux command. \\ 3. Susan catted the file to bash \\ 4. * Susan catted bash the file. \subsection{Formulate a lexical rule for the dative alternation.} For this rule I will use a D-Rule because our rule is going to specify a different ARG-ST value, change the order of our COMPS values, and changing one of our comps' NP to a PP. Each one of these changes would be inconsistent with the constraints on i-rule which means we need a d-rule. I have a helper function $F_{DAR}$ which modifies our verb to become a causative verb.\\ Dative Alteration Lexical Rule \\ \begin{avm} \[{\it d-rule} \\ INPUT & \< \avml{\@1}, \[{\it verb-lxm} \\ SYN & \[VAL & \[SPR & \< \avml {\@2} \avmr \> \\ COMPS & \< \avml {\@3}NP , {\@4}NP\avmr \> \] \\ ARG-ST & \< \avml {\@2},{\@3},{\@4} \avmr \> \] \] \avmr \> \\ OUTPUT & \< \avml F_{DAR}({\@1}) , \[{\it verb-lxm} \\ SYN & \[VAL & \[SPR & \< \avml {\@2} \avmr \> \\ COMPS & \< \avml {\@4}NP ,{\@3}PP \avmr \> \] \\ ARG-ST & \< \avml {\@2} {\@4},{\@3} \avmr \> \] \] \avmr \> \] \end{avm} \subsection{Dative Alteration Lexical Rule interacts with the Passive Lexical Rule} Given the way these rules are structured, we first want to apply the Dative Alteration Lexical Rule(DALR) followed by the Passive Lexical Rule(PLR). We choose this order because the DALR changes the COMPS values and the PLR can function independent of them. If we were to apply the PLR before the DALR we would need a second DALR that can modify passive verbs. When these rules are chained together we can think of the effects as below. \\ \begin{avm} \[{\it d-rule} \\ INPUT & \< \avml{\@1}, \[{\it verb-lxm} \\ SYN & \[VAL & \[SPR & \< \avml {\@2} \avmr \> \\ COMPS & \< \avml {\@3}NP , {\@4}NP \avmr \> \] \\ ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2},{\@3},{\@4} \avmr \> \] \] \avmr \> \\ DALR-OUTPUT & \< \avml F_{DAR}({\@1}) , \[{\it verb-lxm} \\ SYN & \[VAL & \[SPR & \< \avml {\@2} \avmr \> \\ COMPS & \< \avml {\@4}NP ,{\@3}PP \avmr \> \] \\ ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2}, {\@4},{\@3} \avmr \> \] \] \avmr \> \\ PLR-OUTPUT & \< \avml F_{PSP}({\@1}) , \[{\it verb-lxm} \\ SYN & \[HEAD &\[FORM & {\it pass} \] \\ VAL & \[SPR & \< \avml {\@2} \avmr \> \\ COMPS & \< \avml {\@4}NP ,{\@3}PP \avmr \> \] \\ ARG-ST & \< \avml {\@4},{\@3} \oplus {\@2}(\[PP \\FORM & by \\ INDEX & i\) \avmr \> \] \] \avmr \> \] \end{avm} \\ In context of sentence iii:(Merle was handed a book by Dale) we apply the DALR and shift the AGR-ST values from (1) to (2) and then we apply the PLR to shift the ARG-ST values from (2) to (3) where \begin{avm}{\@2}\end{avm} refers to Merle, \begin{avm}{\@3}\end{avm} refers to book and \begin{avm}{\@3}\end{avm} refers to Dale \\ \begin{equation} \begin{avm} \[ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2},{\@3},{\@4} \avmr \> \] \end{avm} \end{equation} \begin{equation} \begin{avm} \[ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2},{\@4},{\@3} \avmr \> \] \end{avm} \end{equation} \begin{equation} \begin{avm} \[ARG-ST & \< \avml {\@4} {\@3} \oplus {\@2}(\[PP \\FORM & by \\ INDEX & i\) \avmr \> \] \end{avm} \end{equation} In context of sentence iv:(A book was handed to Merle by Dale) we apply the DALR (taking (4) to (5) followed by the PLR which takes the AGR-ST to (6) where \begin{avm}{\@2}\end{avm} refers to Merle, \begin{avm}{\@3}\end{avm} refers to book and \begin{avm}{\@3}\end{avm} refers to Dale \\ \begin{equation} \begin{avm} \[ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2},{\@3},{\@4} \avmr \> \] \end{avm} \end{equation} \begin{equation} \begin{avm} \[ARG-ST & \< \avml \[INDEX & i\] \oplus {\@2},{\@4},{\@3} \avmr \> \] \end{avm} \end{equation} \begin{equation} \begin{avm} \[ARG-ST & \< \avml {\@4},{\@3} \oplus {\@2}(\[PP \\FORM & by \\ INDEX & i\) \avmr \> \] \end{avm} \end{equation} \\ \subsection{Fail to license A book was handed Merle by Dale.} The grammar fails to license v(correctly for a simple reason, Merle is a NP instead of a PP which a dative altered passive verb needs. If a to/by was added before Merle then the sentence would be licensed by our grammar. \section{Send the postcard or flyer to Sandy's address!} \subsection{Lexical Types of lexical entries} 1. send is a dtv-lm. \\ 2. the is a det-lxm. \\ 3. postcard is a cntn-lxm. \\ 4.or is a conj-lxm. \\ 5. flyer is a cntn-lxm. \\ 6. to is a argmkp-lxm. \\ 7. Sandy is a cntn-lxm. \\ 8. 's is a det-lxm \\ 9. Address is a cntn-lxm. \\ \comment{\begin{avm} \< \avml send , \[{\it dtv-lm} \\ SYN & \[HEAD & \[{\it verb} \\CASE & NOM \\ ARG & {\@1} \] \\ VAL & \[SPR & \< \avml \[AGR {\@1} \] \avmr \> \] \] \\ ARG-ST & \< \avml NP_i, NP_j , NP_k \avmr \> \\ SEM & \[MODE & {\it prop} \\ INDEX & s \\ RESTR & \<\[RELN & send \\ SIT & {\it s} \\ SENDER & {\it i} \\ RECIPIENT & {\it j} \\ SENT & {\it k} \] \avmr \> \] \] \avmr \> \end{avm} \\ \begin{avm} \< \avml the, \[{\it det-lxm} \\ SYN & \[HEAD & \[{\it det} \\ ARG & {\it 3sing} \\ COUNT & + \] \\ VAL & \[SPR & \< \avml \avmr \> \] \] \\ ARG-ST & \< \avml \avmr \> \\ SEM & \[MODE & {\it none} \\ INDEX & i \\ RESTR & \<\[RELN & exists \\ BV & {\it i} \] \avmr \> \] \] \avmr \> \end{avm} \\ \begin{avm} \< \avml postcard , \[{\it cntn-lxm} \\ SYN & \[ HEAD & \[{\it noun} \\ AGR & {\@1}\[PER & {\it 3rd} \\NUM & {\it sg} \] \] \\ VAL & \[SPR & \< \avml {\@2}\[AGR & {\@1} \] \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[MODE & {\it ref} \\ INDEX & {\it i} \\ RESTR & \< \avml \[RELN & postcard \\ INST & {\it i} \] \avmr \> \] \\ AGR-ST & \< \avml {\@2}\[DP \\ COUNT & + \] \avmr \> \] \avmr \> \end{avm} \\ \begin{avm} \< \avml or , \[{\it conj-lxm} \\ SYN & \[ HEAD & \[{\it conj} \] \\ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[MODE & {\it none} \\ INDEX & {\it s} \\ RESTR & \< \avml \[RELN & or \\ SIT & {\it s} \] \avmr \> \] \\ AGR-ST & \< \avml \avmr \> \] \avmr \> \end{avm} \\ \begin{avm} \< \avml flyer , \[{\it cntn-lxm} \\ SYN & \[ HEAD & \[{\it noun} \\ AGR & {\@1}\[PER & {\it 3rd} \\NUM & {\it sg} \] \] \\ VAL & \[SPR & \< \avml {\@2}\[AGR & {\@1} \] \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[MODE & {\it ref} \\ INDEX & {\it i} \\ RESTR & \< \avml \[RELN & flyer \\ INST & {\it i} \] \avmr \> \] \\ AGR-ST & \< \avml {\@2}\[DP \\ COUNT & + \] \avmr \> \] \avmr \> \end{avm} \\ \begin{avm} \< \avml to , \[{\it predp-lxm} \\ SYN & \[ HEAD & \[{\it prep} \] \\ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[ INDEX & {\it s} \\ RESTR & \< \avml \[RELN & to \\ ITEM & i \\ RECIPIENT & j \\ SIT & {\it s} \] \avmr \> \] \\ AGR-ST & \< \avml NP_i , NP_j \avmr \> \] \avmr \> \end{avm} \\ \begin{avm} \< \avml Sandy, \[{\it cntn-lxm} \\ SYN & \[ HEAD & \[{\it noun} \\ AGR & {\@1}\[PER & {\it 3rd} \\NUM & {\it sg} \] \] \\ VAL & \[SPR & \< \avml {\@2}\[AGR & {\@1} \] \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[MODE & {\it ref} \\ INDEX & {\it i} \\ RESTR & \< \avml \[RELN & name \\ NAMED & {\it i} \\ NAME & Sandy\] \avmr \> \] \\ AGR-ST \< \avml {\@2}\[DP \\ COUNT & + \] \avmr \> \] \avmr \> \end{avm} \begin{avm} \< \avml 's, \[{\it det-lxm} \\ SYN & \[HEAD & \[{\it det} \\ COUNT & + \] \\ VAL & \[SPR & \< \avml NP \avmr \> \] \] \\ ARG-ST & \< \avml \avmr \> \\ SEM & \[MODE & {\it none} \\ INDEX & i \\ RESTR & \<\[RELN & poss \\ POSSESOR & i \\ POSSESED & J \] \[RELN & the \\ BV & {\it i} \] \] \avmr \> \] \] \avmr \> \end{avm} \\ \begin{avm} \< \avml Address , \[{\it cntn-lxm} \\ SYN & \[ HEAD & \[{\it noun} \\ AGR & {\@1}\[PER & {\it 3rd} \\NUM & {\it sg} \] \] \\ VAL & \[SPR & \< \avml {\@2}\[AGR & {\@1} \] \avmr \> \\ COMPS \< \avml \avmr \> \] \] \\ SEM & \[MODE & {\it ref} \\ INDEX & {\it i} \\ RESTR & \< \avml \[RELN & address \\ INST & {\it i} \] \avmr \> \] \\ AGR-ST & \< \avml {\@2}\[DP \\ COUNT & + \] \avmr \> \] \avmr \>} \end{avm} For each word in the sentence, identify the lexical type of the lexical entry that licenses it. \subsection{Lexical Rules} 1. Send is licensed by Non-3rd-Singular Verb Lexical Rule and Passive Lexical Rule. \\ 2. the is licensed by a Constant Lexeme Lexical Rule.\\ 3. Postcard is licensed by Singular Noun Lexical Rule.\\ 4. or is licensed by a Constant Lexeme Lexical Rule.\\ 5. flyer is licensed by Singular Noun Lexical Rule.\\ 6. to is licensed by a Constant Lexeme Lexical Rule.\\ 7. Sandy is licences by Singular Noun Lexical Rule.\\ 8. 's is licensed by a Constant Lexeme Lexical Rule.\\ 9. address is licensed by Singular Noun Lexical Rule. \subsection{Simple Tree} Send the postcard or flyer to Sandy's address!\\ \begin{forest} [S [VP [V [Send]] [DP [D [the ] ] [NP [NP [NOM [postcard] ] ] [CONJ [or] ] [NP [NOM [flyer] ] ] ] ] [PP [P [ to] ] [NP [DP [NOM [Sandy]] [D ['s] ] ] [NP [NOM [address] ] ] ] ] ] ] \end{forest} \subsection{Tree with feature structure} Tree rotated for ease of reading. \\ \\ \scalebox{0.47}{ \begin{turn}{90} \begin{forest} [S [VP \\ [ \begin{avm}\[V \\\[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml {\@1} {\@2} \avmr \> \] \] \\ AGR-ST & \< \avml {\@A} \oplus{\@B} \oplus{\@C} \avmr \> \] \] \end{avm} [Send]] [ \begin{avm}\[ DP \\ {\@1} \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [\begin{avm} \[D \\ {\@3} \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [the ] ] [\begin{avm}\[ NP \\ {\@A} \[SYN & \[ VAL & \[SPR & \< \avml {\@3} \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [\begin{avm} \[NP \\ \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm}[postcard] ] [\begin{avm} \[CONJ \\ \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [or] ] [\begin{avm}\[NP \\ \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \]\] \] \end{avm} [flyer] ] ] ] [\begin{avm}\[PP \\ {\@2} \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [\begin{avm}\[prep \\ \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [ to] ] [\begin{avm}\[NP \\\[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [\begin{avm}\[DP \\ {\@4} \[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [\begin{avm}\[NP \\ {\@B}\[SYN & \[ VAL & \[SPR & \< \avml \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \]\end{avm} [Sandy]] [\begin{avm}\[DET \\ \[SYN & \[ VAL & \[SPR & \< \avml {\@B} \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} ['s] ] ] [\begin{avm}\[NP \\ {\@C} \[SYN & \[ VAL & \[SPR & \< \avml {\@4} \avmr \> \\ COMPS \< \avml \avmr \> \] \] \] \] \end{avm} [address] ] ] ] ] ] \end{forest} \end{turn} } \subsection{Chain of Identities that link DESTINATION role to INST role} 1. The Noun Address identifies the INST role of Address from the lexical entry via Head-Specifier Rule. \\ 2. The Verb Send identifies the Destination role of the send predication from the lexical entry via Head-Specifier Rule. \\ 3. The NP 'Sandy's address' identifies the INST role of address from the lexical entry via Head-Specifier Rule. \\ 4. The SPR values for Sandy's cause its ARG-ST include the index of INST from address due to the Argument Realization Principle. \\ 5. The NP 'Sandy's address' assumes the INST role of Sandy via semantic inheritance principle. \\ 6. The PP 'to Sandy's address' assumes the joint INSTS of address and sandy via valence principle. \\ 7. The Argument Realization Principle joins the ARG-ST values in the VP and the NP. \\ 8. Global representation of ARG-ST created via Semantic Compositionality Principle. \\ 9. Information about the INST of destination is updated in the verb Send via Semantic Inheritance Principle. \subsection{PER value of node above SEND} In the fully resolved tree, the PER value of the node is Verb because of the Head-Specifier Rule, The Head Feature Principle and The Valence Principle. \end{document}jacopok/notesphd_courses/experimental_gravitation_cosmology/dec21.tex \documentclass[main.tex]{subfiles} \begin{document} \marginpar{Tuesday\\ 2021-12-21} The radiation pressure term is written as % \begin{align} \text{RP} = \frac{1}{c} \left( P + 2 \sqrt{P} \sqrt{\hbar \omega_0 } \hat{a}_1 (\Omega ) \right) \,, \end{align} % with quadratic, \(\order{\hat{a}^2}\) corrections. This corresponds to a displacement \(\text{RP} / m \Omega^2 = \delta x (\Omega) \). This displacement is connected to a phase fluctuation % \begin{align} \delta \phi (\Omega ) = 2 \frac{ \delta x (\Omega ) \omega_0 }{c} \,. \end{align} The thing is then that the modulation can be written as % \begin{align} E_1 &= \sqrt{P} \cos(\omega_0 t + \delta \phi (\Omega )) \\ &= \sqrt{P} \cos(\omega_0 t) - \delta \phi (\Omega ) \sqrt{O} \sin(\omega_0 t) \,, \end{align} % while the phase quadrature reads % \begin{align} E_2 = \sqrt{P} \delta \phi (\Omega ) \propto - \frac{\sqrt{P}}{\Omega^2} \,. \end{align} This is the reason why the \SI{}{MHz} sidebands are not affected by radiation pressure noise. \subsection{Wiener filtering} If we have an estimated signal \(\hat{y} = F(x)\) and an observed signal \(y\), we want to minimize \(\expval{(y - \hat{y})^2} = e^2\). We choose this by determining a maximum, with \(\pdv*{e^2}{F} = 0\) and \(\pdv*{e^2}{F} < 0\). The optimal choice is then \(F = \expval{xy} / \expval{x^2}\). In the frequency domain, a filter looks like \(\hat{y}(\Omega ) = F(\Omega ) x(\Omega )\). There are ways to implement this and keep track of continuous variations of the environment (Kanman filtering?). \subsection{Environmental noise} Part of the site selection process is looking at seismic noise. We need to have noise globally below a certain RMS value, even microseismic peaks at a few Hz create issues since they can throw the interferometer out of alignment, even though we do not detect signals at those frequencies. A big issue is also Newtonian gravitational noise, gravitational coupling of the environment to the test mass, which scales like \(f^{-2}\). Temperature gradients of sufficiently small scales in the atmosphere can also create issues. The acoustic noise in the cave also creates issues. One can subtract the measurement of noise actively, but this is very expensive for ET since we would need many new boreholes! Measuring acoustic fields in the atmosphere is hard: eddies can form around the microphone if the wind is too high. Lasers are quite good for this purpose. There are magnetic disturbances correlating all the way around the world. These are Schumann resonances. \subsection{Thermal noise} Material science is very poorly understood. We are not able to compute things like Young's modulus from first principles. The fundamental theorem describing this is the fluctuation-dissipation theorem: % \begin{align} S_x(\Omega ) = \frac{ 8 \pi k_B T}{\Omega^2} \frac{W _{\text{diss}}}{F_p^2} \,. \end{align} This describes the thermal noise spectrum due to mechanically dissipated power. The thermal noise we measure is average over the beam, so low-wavelength thermal noise are not really a problem. Some high-order Laguerre-Gauss modes have been proposed as a way to moderate this issue. Heat links can be a shortcut for vibrations! The way coating of the mirror is done is relevant. Tunneling is a big source of dissipation in materials. The study of internal friction of materials is often studied by perturbing them and looking at the ringdown. Exam: one can say which part they want to focus on. There is a short, 20-minute presentation to give. \end{document} \hypertarget{a00776}{}\doxysection{Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text Class Reference} \label{a00776}\index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} Collaboration diagram for Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=250pt]{a00774} \end{center} \end{figure} \doxysubsection*{Static Public Attributes} \begin{DoxyCompactItemize} \item const int \mbox{\hyperlink{a00776_a410c7e4a86b4c97abd2c11093f64b0e2}{Annotation\+Text\+Offset}} = 5 \item const string \mbox{\hyperlink{a00776_ac21e92415393f152a581f0a9e6fd3f8e}{Profound\+HL}} = \char`\"{}Profound Hearing Loss\char`\"{} \item const string \mbox{\hyperlink{a00776_a3961324337eb487d4e7ed9cf4d1686a7}{Severe\+HL}} = \char`\"{}Severe Hearing Loss\char`\"{} \item const string \mbox{\hyperlink{a00776_a600042dc1fa62cf0aba25080ae0241d1}{Moderate\+HL}} = \char`\"{}Moderate Hearing Loss\char`\"{} \item const string \mbox{\hyperlink{a00776_a45c456fe52c651e204bbf8b19627df6c}{Mild\+HL}} = \char`\"{}Mild Hearing Loss\char`\"{} \item const string \mbox{\hyperlink{a00776_a44b0a5a4408e77fee130f91296e4d4bd}{Normal\+HL}} = \char`\"{}Normal Hearing\char`\"{} \end{DoxyCompactItemize} \doxysubsection{Member Data Documentation} \mbox{\Hypertarget{a00776_a410c7e4a86b4c97abd2c11093f64b0e2}\label{a00776_a410c7e4a86b4c97abd2c11093f64b0e2}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!AnnotationTextOffset@{AnnotationTextOffset}} \index{AnnotationTextOffset@{AnnotationTextOffset}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{AnnotationTextOffset}{AnnotationTextOffset}} {\footnotesize\ttfamily const int Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Annotation\+Text\+Offset = 5\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{a00776_a45c456fe52c651e204bbf8b19627df6c}\label{a00776_a45c456fe52c651e204bbf8b19627df6c}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!MildHL@{MildHL}} \index{MildHL@{MildHL}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{MildHL}{MildHL}} {\footnotesize\ttfamily const string Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Mild\+HL = \char`\"{}Mild Hearing Loss\char`\"{}\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{a00776_a600042dc1fa62cf0aba25080ae0241d1}\label{a00776_a600042dc1fa62cf0aba25080ae0241d1}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!ModerateHL@{ModerateHL}} \index{ModerateHL@{ModerateHL}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{ModerateHL}{ModerateHL}} {\footnotesize\ttfamily const string Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Moderate\+HL = \char`\"{}Moderate Hearing Loss\char`\"{}\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{a00776_a44b0a5a4408e77fee130f91296e4d4bd}\label{a00776_a44b0a5a4408e77fee130f91296e4d4bd}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!NormalHL@{NormalHL}} \index{NormalHL@{NormalHL}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{NormalHL}{NormalHL}} {\footnotesize\ttfamily const string Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Normal\+HL = \char`\"{}Normal Hearing\char`\"{}\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{a00776_ac21e92415393f152a581f0a9e6fd3f8e}\label{a00776_ac21e92415393f152a581f0a9e6fd3f8e}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!ProfoundHL@{ProfoundHL}} \index{ProfoundHL@{ProfoundHL}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{ProfoundHL}{ProfoundHL}} {\footnotesize\ttfamily const string Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Profound\+HL = \char`\"{}Profound Hearing Loss\char`\"{}\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{a00776_a3961324337eb487d4e7ed9cf4d1686a7}\label{a00776_a3961324337eb487d4e7ed9cf4d1686a7}} \index{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}!SevereHL@{SevereHL}} \index{SevereHL@{SevereHL}!Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText@{Audiometry.ViewModel.PureToneVM.Audiogram.AnnotationText}} \doxysubsubsection{\texorpdfstring{SevereHL}{SevereHL}} {\footnotesize\ttfamily const string Audiometry.\+View\+Model.\+Pure\+Tone\+V\+M.\+Audiogram.\+Annotation\+Text.\+Severe\+HL = \char`\"{}Severe Hearing Loss\char`\"{}\hspace{0.3cm}{\ttfamily [static]}} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item Audiometry/\+View\+Model/\+Pure\+Tone\+V\+M/\+Audiogram/\mbox{\hyperlink{a00257}{Audiogram\+Types.\+cs}}\end{DoxyCompactItemize} @book{johnson2007ghost, address = {London}, edition = {Reprint edition}, title = {The {{Ghost Map}}: {{The Story}} of {{London}}'s {{Most Terrifying Epidemic}}--and {{How It Changed Science}}, {{Cities}}, and the {{Modern World}}}, isbn = {978-1-59448-269-4}, shorttitle = {The {{Ghost Map}}}, abstract = {A National Bestseller, a New York Times Notable Book, and an Entertainment Weekly Best Book of the Year It's the summer of 1854, and London is just emerging as one of the first modern cities in the world. But lacking the infrastructure-garbage removal, clean water, sewers-necessary to support its rapidly expanding population, the city has become the perfect breeding ground for a terrifying disease no one knows how to cure. As the cholera outbreak takes hold, a physician and a local curate are spurred to action-and ultimately solve the most pressing medical riddle of their time. In a triumph of multidisciplinary thinking, Johnson illuminates the intertwined histories of the spread of disease, the rise of cities, and the nature of scientific inquiry, offering both a riveting history and a powerful explanation of how it has shaped the world we live in.}, language = {English}, publisher = {{Riverhead Books}}, author = {}, month = oct, year = {2007}, doi = {10.1080/01944360802146329} } @incollection{arribas-bel2017looking, address = {Cham}, series = {Advances in {{Spatial Science}}}, title = {Looking at {{}}'s {{Cholera Map}} from the {{Twenty First Century}}: {{A Practical Primer}} on {{Reproducibility}} and {{Open Science}}}, isbn = {978-3-319-50590-9}, shorttitle = {Looking at {{}}'s {{Cholera Map}} from the {{Twenty First Century}}}, abstract = {This chapter (This manuscript is a chapter version of the original document, which is a reproducible online notebook. The entire, version-controlled project can be found online at: https://bitbucket.org/darribas/reproducible\_john\_snow.) presents an entirely reproducible spatial analysis of the classic John Snow's map of the 1854 cholera epidemic in London. The analysis draws on many of the techniques most commonly used by regional scientists, such as choropleth mapping, spatial autocorrelation, and point pattern analysis. In doing so, the chapter presents a practical roadmap for performing a completely open and reproducible analysis in regional science. In particular, we deal with the automation of (1) synchronizing code and text, (2) presenting results in figures and tables, and (3) generating reference lists. In addition, we discuss the significant added value of version control systems and their role in enhancing transparency through public, open repositories. With this chapter, we aim to practically illustrate a set of principles and techniques that facilitate transparency and reproducibility in empirical research, both keys to the health and credibility of regional science in the next 50 years to come.}, language = {en}, booktitle = {Regional {{Research Frontiers}} - {{Vol}}. 2: {{Methodological Advances}}, {{Regional Systems Modeling}} and {{Open Sciences}}}, publisher = {{Springer International Publishing}}, author = {{Arribas-Bel}, Daniel and {}, Rey, .}, editor = { and }, year = {2017}, keywords = {Regional Science,Spatial Autocorrelation,Spatial Outlier,Spatial Weight Matrix,Street Segment}, pages = {283-306}, doi = {10.1007/978-3-319-50590-9_17} } @book{tufte2001visual, title = {The Visual Display of Quantitative Information}, publisher = {{Graphics Press Cheshire, CT, USA}}, author = {.}, year = {2001} } @article{goodchild2007citizen, title = {Citizen as Sensors: The World of Volunteered Geography}, volume = {69}, journal = {GeoJournal}, author = {Goodchild, }, year = {2007}, keywords = {volunteer geographic information}, pages = {211--221} } @article{arribas-bel2014accidental, title = {Accidental, Open and Everywhere: {{Emerging}} Data Sources for the Understanding of Cities}, volume = {49}, issn = {01436228}, shorttitle = {Accidental, Open and Everywhere}, abstract = {In this paper, I review the recent emergence of three groups of data sources and assess some of the opportunities and challenges they pose for the understanding of cities, particularly in the context of the Regional Science and urban research agenda. These are data collected from mobile sensors carried by individuals, data derived from businesses moving their activity online and government data released in an open format. Although very different from each other, they are all becoming available as a side-effect since they were created with different purposes but their degree of popularity, pervasiveness and ease of access is turning them into interesting alternatives for researchers. Existing projects and initiatives that conform to each class are featured as illustrative examples of these new potential sources of knowledge. \'O 2013 Elsevier Ltd. All rights reserved.}, language = {en}, journal = {Applied Geography}, doi = {10.1016/j.apgeog.2013.09.012}, author = {{Arribas-Bel}, Daniel}, month = may, year = {2014}, pages = {45-53}, file = {/home/lw17329/Zotero/storage/I4QFYAPM/Arribas-Bel - 2014 - Accidental, open and everywhere Emerging data sou.pdf} } @article{anselin1988spatial, title = {Do Spatial Effects Really Matter in Regression Analysis?}, volume = {65}, journal = {Papers Regional Science}, author = { }, year = {1988}, keywords = {Spatial dependence}, pages = {11--34}, doi = {10.1111/j.1435-5597.1988.tb01155.x} } @book{anselin2014modern, address = {Chicago, IL}, title = {Modern {{Spatial Econometrics}} in {{Practice}}, a {{Guide}} to {{GeoDa}}, {{GeoDaSpace}}, and {{PySAL}}}, publisher = {{GeoDa Press}}, author = { Rey, .}, year = {2014} } @article{rey2009show, title = {Show Me the Code: Spatial Analysis and Open Source}, volume = {11}, issn = {1435-5930, 1435-5949}, shorttitle = {Show Me the Code}, language = {en}, number = {2}, journal = {Journal of Geographical Systems}, doi = {10.1007/s10109-009-0086-8}, author = {Rey, .}, month = jun, year = {2009}, pages = {191--207}, file = {/home/lw17329/Dropbox/literature/Rey - 2009 - Show me the code spatial analysis and open source.pdf;/home/lw17329/Zotero/storage/E6UEYX4F/art\%3A10.1007\%2Fs10109-009-0086-8.pdf;/home/lw17329/Zotero/storage/K9JRHXCG/art\%3A10.1007\%2Fs10109-009-0086-8.pdf} } @incollection{rey2018code, address = {Cham}, series = {Advances in {{Geographic Information Science}}}, title = {Code as {{Text}}: {{Open Source Lessons}} for {{Geospatial Research}} and {{Education}}}, isbn = {978-3-319-59511-5}, shorttitle = {Code as {{Text}}}, abstract = {This chapter examines the potential opportunities that open source offers for research and education in spatial analysis. Drawing on lessons learned in the development of PySAL: Python Library for Spatial Analysis, it touches on the opportunities and challenges related to the adoption of open source practices and culture. While open source has had major impacts on pedagogy and research in spatial analysis, these are somewhat under-appreciated and at times seen as separate spheres. A central argument is that a mind shift is required that comes to see code not just as a tool for doing research, but rather to view code as text in the sense it becomes an object of research. The chapter reconsiders open source spatial analysis teaching and research from this lens of code as text.}, language = {en}, booktitle = {{{GeoComputational Analysis}} and {{Modeling}} of {{Regional Systems}}}, publisher = {{Springer International Publishing}}, author = {.}, editor = { and }, year = {2018}, pages = {7-21}, doi = {10.1007/978-3-319-59511-5_2} } @techreport{ucgis2019geographic, title = {Geographic {{Information Science}} and {{Technology Body}} of {{Knowledge}}}, author = {{University Consortium of Geographic Information Science}}, year = {2019}, file = {/home/lw17329/Zotero/storage/9INJ4PZ7/gistbok.ucgis.org.html} } @misc{kelsey_jordahl_2019_3333010, author = { and and and and and and and and and and and and and and maxalbert and and and and and and and and and and \" and }, title = {geopandas/geopandas: v0.5.1}, month = jul, year = 2019, doi = {10.5281/zenodo.3333010}, url = {https://doi.org/10.5281/zenodo.3333010} } @misc{aleksey_bilogur_2019_3475569, author = { and and and }, title = {ResidentMario/geoplot 0.3.3}, month = oct, year = 2019, doi = {10.5281/zenodo.3475569}, url = {https://doi.org/10.5281/zenodo.3475569} } @article{roth2010vba, author = { and and }, title = {Value-by-alpha maps: {An} alternative to the cartogram}, journal = {The Cartographic Journal}, year = {2010}, volume = 47, issue = 2, pages = {130--140}, doi = {10.1179/000870409X12488753453372} } volt_distr/doc/volt_distr/html/latex/files.tex \section{\-File \-List} \-Here is a list of all files with brief descriptions\-:\begin{DoxyCompactList} \item\contentsline{section}{{\bf volt\-\_\-distr.\-cpp} }{\pageref{volt__distr_8cpp}}{} \item\contentsline{section}{{\bf volt\-\_\-distr.\-h} }{\pageref{volt__distr_8h}}{} \item\contentsline{section}{{\bf volt\-\_\-distr\-\_\-creator.\-cpp} }{\pageref{volt__distr__creator_8cpp}}{} \item\contentsline{section}{{\bf volt\-\_\-distr\-\_\-creator.\-h} }{\pageref{volt__distr__creator_8h}}{} \item\contentsline{section}{{\bf volt\-\_\-distr\-\_\-viz.\-cpp} }{\pageref{volt__distr__viz_8cpp}}{} \item\contentsline{section}{{\bf volt\-\_\-distr\-\_\-viz.\-h} }{\pageref{volt__distr__viz_8h}}{} \end{DoxyCompactList} akhandelwal8/akhandelwal.github.io0 @article{GSTCHGWSCKCB2018rse, title = "Using Landsat and nighttime lights for supervised pixel-based image classification of urban land cover", journal = "Remote Sensing of Environment", volume = "205", pages = "253 - 275", year = "2018", issn = "0034-4257", doi = "https://doi.org/10.1016/j.rse.2017.11.026", url = "http://www.sciencedirect.com/science/article/pii/S0034425717305758", author = " and and and and and and and and and and ", keywords = "Urbanization, Built-up land cover, Nighttime light, Image classification, Google Earth Engine", abstract = "Reliable representations of global urban extent remain limited, hindering scientific progress across a range of disciplines that study functionality of sustainable cities. We present an efficient and low-cost machine-learning approach for pixel-based image classification of built-up areas at a large geographic scale using Landsat data. Our methodology combines nighttime-lights data and Landsat 8 and overcomes the lack of extensive ground-reference data. We demonstrate the effectiveness of our methodology, which is implemented in Google Earth Engine, through the development of accurate 30m resolution maps that characterize built-up land cover in three geographically diverse countries: India, Mexico, and the US. Our approach highlights the usefulness of data fusion techniques for studying the built environment and is a first step towards the creation of an accurate global-scale map of urban land cover over time." }disputatio/propositum/software.tex \section{Computational Methods and Software} \label{sec:software} The previous section described a series of models for the system of interest. %These models are too %complicated to solve exactly, and must instead be %instantiated with software to produce a numerical %result. This section details the numerical formulation and simulation of these models. It begins with a discussion of the numerical discretization of the equations of interest. The mesh discretization is then described. Next, the scientific software in which these numerical models are used is discussed. Finally, the tool chain and supercomputer systems are briefly introduced. \subsection{Discretization Scheme} To numerically solve the Navier-Stokes equations on a computer, a Galerkin finite element method (FEM) is used, which requires that the equations in section~\ref{sub_sec:ns_en} be cast into a weak form. Manipulating these partial differential equations into a variational formulation is accomplished by multiplying the equations by appropriate test functions and integrating over the domain, $\Omega$. The resulting weak problem is to find, $(u,p,T) \in H^1(\Omega)^3 \times L_2(\Omega) \times H^1(\Omega)$ such that % % http://www.numerik.uni-hd.de/Oberwolfach-Seminar/CFD-Course.pdf % \begin{align} (\frac{\partial u}{\partial t}, v) + (u \cdot \nabla u,v) + \nu (\nabla u, \nabla v) -(p,\nabla \cdot u) &= (gT'/T_0,v) \label{eqn:ns_weak} \\ (\nabla \cdot u,q) &= 0 \label{eqn:cont_weak} \\ (\frac{\partial T}{\partial t}, w) + (u \cdot \nabla T, w) + (k \, \nabla T, \nabla w) &= 0.\label{eqn:en_weak} \end{align} $\forall (v,q,w) \in H^1(\Omega)^3 \times L_2(\Omega) \times H^1(\Omega)$, where $(u,v) = \int_\Omega u \cdot v \, dx$. Some of the simulations presented here were conducted under steady conditions, for which the $\frac{\partial}{\partial t}$ terms vanish. A Galerkin FEM scheme is obtained by posing the weak form in terms of discrete subspaces of the function spaces specified above defined using piecewise-polynomial basis functions. All of the simulations discussed in this work were accomplished using linear basis functions for both the velocity and pressure. %Typically, the use of equal order elements for velocity and pressure is %ruled out in the standard Galerkin FEM formulation by the Babuska-Brezzi %condition\cite{bb-cond}. The scheme is stable with equal-order elements for velocity and pressure due to the introduction of streamline upwind/Petrov-Galerkin (SUPG) stabilization terms as first described by Hughes\cite{Hughes198685,supg} and extended to natural convection as in Becker and Braack\cite{Becker2002428}. The stabilization terms add artificial dissipation that approaches zero as the residual converges. This scheme is ``consistent'' because the underlying order of convergence of the numerical method is not affected\cite{hughes2000finite}. This stabilization is accomplished by introducing an additional term, $\langle Lc,S\phi \rangle_\tau$, to the weak form defined in Equations \ref{eqn:ns_weak}-\ref{eqn:en_weak}. Here $L()$ is the operator for the PDEs in \ref{sub_sec:ns_en}, S is a stabilization operator which is chosen to be the negative adjoint of the differential terms of $L()$, and $c$ and $\phi$ are state and test functions, i.e. $ c= (u,p,T)$, and $\phi = ( v,w,q )$. The angle brackets $\langle \cdot,\cdot \rangle$ signify integration of the element interiors for each of the K elements, that is: \begin{equation} \langle u,v \rangle_\tau = \sum_K \tau_K(u,v)_K. \end{equation} This results in three stabilization parameters, $\tau_P, \tau_v, \tau_T$ which are selected as proposed by Becker and Braack. A full derivation of the weak form and stabilization terms will be provided in an appendix of the full thesis. The system of ODEs are discretized in time using the backward Euler method\cite{moin2010fundamentals}. The time interval $(0,T)$ is sliced into $N_t$ steps of uniform temporal length, $\Delta t$, where $n = 0,\dots,N_t$. This has the form, \begin{equation} y_{n+1} = y_n + \Delta t \, f(y_{n+1},t_{n+1}). \end{equation} As $f$ is non-linear, a Newton-Raphson method is used to solve the resulting implicit nonlinear problem. While an iterative method is significantly more computationally expensive per timestep than a similar explicit method, the method was selected due to its unconditional stability and ease of statistical sampling for a uniform timestep. % % gave not completely described numerical methods % for instance, have not indicated the stabilization schemes % do not need complete equations, but should permit someone to access % the literature and construct precisely the numerical formulations used % \subsection{Mesh Discretization} % % what about mesh... % The domain's described in subsection~\ref{sec:bc} are consistently discretized. The domain extents are scaled by system diameter but the same number of grid points are used for every simulation. Thus, while the ratio of the domain length to system diameter remains fixed, the grid spacing increases proportionally with domain length. The mesh has a uniform spacing in the lateral directions, except for a single refinement in the region of the vanes. Typically, the grid is roughly one hundred points in the streamwise and spanwise directions before the refinement. The refinement halves the spacing (doubles the number of points) in all three coordinate directions, \{x,y,z\} in this region. The refinement is made from the ground to 1.5 times the height of the vanes and cone. The mesh is non-uniform in height to resolve the boundary layer. This is accomplished by redistributing a mesh uniform in height, z, according to, \begin{equation} z = \begin{cases} C_1(z-L_z)+L_z,& \text{if } z \geq z_\delta\\ C_2 \text{ exp}(C_3 z - 1), & \text{otherwise} \end{cases} \end{equation} where $z_\delta$ is the chosen height of the boundary layer mesh, and $C_1-C_3$ are scaling coefficients. This gives the mesh an exponentially varying character, with the coefficients chosen to ensure ten or more points in the boundary layer, isotropic spacing in cells outside of it, and smooth blending between these two regimes. Each boundary layer spacing was tested against a finer spacing to ensure that the results were not sensitive to the choice of spacing. A horizontal slice though a representative domain is shown in Figure \ref{fig:meshing}. The single refinement in the region of the vanes is visible, as well as the finer meshed boundary layer region near the ground. \begin{figure}[!htb] \begin{center} \includegraphics[width = 10 cm]{figs/meshing} \caption{Horizontal slide through the domain, to show a representative meshing. The single refinement region around the vanes is visible, as well as the finer boundary layer mesh near the ground.} \label{fig:meshing} \end{center} \end{figure} Finally, the diffusivities are proportionally scaled with grid size to ensure that the cell Reynolds number, \begin{equation} \text{Re}_\text{cell} = \frac{\text{max}(\Delta x,\Delta y) u}{\nu_T} \end{equation} is maintained for every simulation, in order to ensure stability. %% hmin = 0.001 %% hmax = 0.4 %% zb = 2.0 %% hrat = ${/ ${mesh-options/hmax} ${mesh-options/hmin}} %% zmax = ${mesh-options/domain_x3_max} %% loghrat = ${= log ${mesh-options/hrat}} %% c2 = ${/ ${mesh-options/zb} ${- ${mesh-options/hrat} 1}} %% zetab = ${/ ${mesh-options/zmax} ${+ 1 ${/ ${- ${mesh-options/zmax} ${mesh-options/zb}} ${* ${mesh-options/c2} ${mesh-options/hrat} ${mesh-options/loghrat}}}}} %% c3 = ${/ ${mesh-options/loghrat} ${mesh-options/zetab}} %% mesh_nx3 = ${= ceil ${/ ${* ${mesh-options/zmax} ${mesh-options/c2} ${mesh-options/c3}} ${mesh-options/hmin}}} %% c1 = ${* ${mesh-options/c2} ${mesh-options/c3} ${= exp ${* ${mesh-options/c3} ${mesh-options/zetab}}}} %% redistribute = '{x}{y}{if(z>${mesh-options/zetab},${mesh-options/c1}*(z-${mesh-options/zmax})+${mesh-options/zmax},${mesh-options/c2}*(exp(${mesh-options/c3}*z)-1))}' %After operation, solutions are evaluated to ensure that %the qualitative character of the solution does not change. \subsection{Software Stack} The numerical approximations described above had been implemented using the GRINS library\cite{GRINSpaper} by Stogner using the Libmesh\cite{libMeshPaper} FEM infrastructure. It was designed to support multiphysics FEM applications, the reusability and extensibility of mathematical modeling kernels, supporting interfaces to existing solver and discretization libraries to enable modern solution strategies, while, at the same time, retaining flexibility to effectively address a wide range of science or engineering problems. GRINS provides a platform that enables powerful numerical algorithms such as adjoint-based AMR, adaptive modeling, sensitivity analysis, and, eventually, enabling uncertainty quantification. While few of these capabilities are in use for the present work, they could be useful in future investigations. GRINS stands for, ``General Reacting Incompressible Navier-Stokes'', which roughly encapsulates the physical regimes it was originally designed to simulate. GRINS is open-source, and available on \hyperref[www.github.com/grinsfem/grins]{github}. It is released under LGPL2.1. GRINS is heavily unit tested, with over 60 tests available to ensure the reliability of results regardless of install platform. %The remainder of this subsection is devoted to %discussing the underlying libraries used and the description of the %GRINS framework. % PETSC\cite{petsc} trilinos\cite{trilinos} % GRINS also uses the fparser\cite{fparser} % library to support both parsing and compilation of mathematical % functions into high % performance kernels. This capability allows for easy specification of % boundary conditions, initial conditions, or constitutive equations from an input file. % Currently, libMesh has been scaled tens of thousands of cores and has % been run on over 100,000 cores on the BG/Q machine Mira at Argonne National % Lab\cite{libmesh-scaling} %In principle, alternative software libraries/frameworks such as %FEniCS\cite{fenics}, OpenFOAM\cite{openfoam}, etc. would likely be %capable of simulating this regime. % % INCLUDE IN THESIS % % \subsection{Solver Options} % GRINS uses PETSC\cite{petsc} and trilinos\cite{trilinos} for numerical % linear algebra, such as constructing and using sparse matrices, finding % the iterative solution of linear systems, and for preconditioning. % While a variety of solver options have been tested in PETSC, all the % results shown in this document use GMRES with block Jacobi for % preconditioning\cite{Saad:2003} for the linear solve. % This uses the inverse of the diagonal block for that processor for % preconditioning. % the preconditioner it's going to use to precondition the linear system % for the solution of the diagonal block. To approximate this, incomplete % LU factorization is used. % ILU(0) factorization %% (11:41:54 AM) nick: ``-ksp_view -ksp_type gmres -pc_type bjacobi -sub_pc_type ilu -sub_pc_factor_levels 0'' %% (11:42:00 AM) : OK %% (11:42:17 AM) : -pc_type is the preconditioner for the entire linear system %% (11:42:25 AM) : You're doing bjacobi = block Jacobi %% (11:42:28 AM) nick: right %% (11:42:36 AM) nick: and does anyone have a good reference I can learn this better? i feel as if I cant look this up, for some reason %% (11:42:39 AM) : That is just using the inverse of the diagonal block for that processor %% (11:42:46 AM) : Now %% (11:42:55 AM) : that is a linear solve %% (11:43:17 AM) : So you can use all the linear solver technology to solve or approximately that block %% (11:43:24 AM) : Hence, -sub_pc_type %% (11:43:32 AM) hil left the room. %% (11:43:46 AM) : That's the preconditioner it's going to use to precondition the linear system for the solution of the diagonal block %% (11:43:53 AM) : You're telling it to use incomplete lu %% (11:44:26 AM) : Now the -sub_pc_factor_levels options applies to ilu %% (11:44:28 AM) hil entered the room. %% (11:44:44 AM) : The incomplete part of imcomplete LU is about the level of fill you use %% (11:45:05 AM) : The more levels of fill you have, the more ``complete'' the incomplete LU will be %% (11:45:08 AM) : Does that make sense? %% (11:45:23 AM) nick: no, that is where you lost me %% (11:45:39 AM) nick: i dont think i know this level of fill %% (11:46:17 AM) : Check out Yousse's book if more curious about the subject %% (11:46:37 AM) nick: cool thanks %% (11:46:38 AM) : Suffice it to say, you heopfully shouldn't ever need to go past 3 or 4 levels of fill %% (11:47:03 AM) : Also, if you've got superlu installed with the PETSc, consider using -sub_pc_factor_mat_solver_package superlu %% (11:47:15 AM) : That's a *much* faster/better implementation than PETSc's \subsection{Tool Chain and Simulation Custodianship} Simulations are performed on the Texas Advanced Computing Center (TACC) supercomputers Lonestar Four and Stampede. Run durations for transient cases are typically twelve hours to perform several hundred timesteps. These runs are submitted to the production queue and are 264-528 processing cores, or 22-44 nodes on Lonestar (with 12 cores per node), and a similar number for Stampede. The runs typically have several million degrees of freedom (DoF), and the local number of DoF per core is maintained at $O(10^4)$. This was selected due to memory constraints, after a strong scaling analysis of the performance of the code on these resources, and after consulting with the software developers. After a run terminates, several scripts are automatically invoked. These scripts archive the run (outside of the volatile /scratch production directories) and simultaneously, label the concluded run with unique metadata that defines the system environment, the jobs input files and run definitions, as well as information detailing the hypothesis or physics the job was intended to investigate. Finally, once a week a script performs \textbf{rsync} on the entire archived database to ensure more than single redundancy for the runs. In other words, the workflow is designed to permit rapid queuing of a series of runs (in parallel) to investigate a variety of conditions or scenario parameters. This capability is necessary for the optimization campaign detailed in \ref{sec:proposed_work}, where running many concurrent investigations will be required to adequately sample the configuration space. \hypertarget{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command}{}\section{Lint\+Command Class Reference} \label{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command}\index{Lint\+Command@{Lint\+Command}} Inheritance diagram for Lint\+Command\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a167134f30ff3446d19da8c08695d07c7}{\+\_\+\+\_\+construct}} (string \$name=null, callable \$directory\+Iterator\+Provider=null, callable \$is\+Readable\+Provider=null) \end{DoxyCompactItemize} \subsection*{Protected Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a9be5e0bdb5720efed6ddb6426c5c16ee}{configure}} () \item \mbox{\hyperlink{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_ab31c72b72ddaf7116db5d84c055d3c0b}{execute}} (Input\+Interface \$input, Output\+Interface \$output) \end{DoxyCompactItemize} \subsection*{Static Protected Attributes} \begin{DoxyCompactItemize} \item static \mbox{\hyperlink{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a8f95e09946a819ad70ca9e71e8c0d153}{\$default\+Name}} = \textquotesingle{}lint\+:yaml\textquotesingle{} \end{DoxyCompactItemize} \subsection{Detailed Description} Validates Y\+A\+ML files syntax and outputs encountered errors. \begin{DoxyAuthor}{Author} \href{mailto:}{\tt } \href{mailto:}{\tt } \end{DoxyAuthor} \subsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a167134f30ff3446d19da8c08695d07c7}\label{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a167134f30ff3446d19da8c08695d07c7}} \index{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}!\+\_\+\+\_\+construct@{\+\_\+\+\_\+construct}} \index{\+\_\+\+\_\+construct@{\+\_\+\+\_\+construct}!Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}} \subsubsection{\texorpdfstring{\+\_\+\+\_\+construct()}{\_\_construct()}} {\footnotesize\ttfamily \+\_\+\+\_\+construct (\begin{DoxyParamCaption}\item[{string}]{\$name = {\ttfamily null}, }\item[{callable}]{\$directory\+Iterator\+Provider = {\ttfamily null}, }\item[{callable}]{\$is\+Readable\+Provider = {\ttfamily null} }\end{DoxyParamCaption})} \subsection{Member Function Documentation} \mbox{\Hypertarget{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a9be5e0bdb5720efed6ddb6426c5c16ee}\label{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a9be5e0bdb5720efed6ddb6426c5c16ee}} \index{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}!configure@{configure}} \index{configure@{configure}!Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}} \subsubsection{\texorpdfstring{configure()}{configure()}} {\footnotesize\ttfamily configure (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \{\} \mbox{\Hypertarget{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_ab31c72b72ddaf7116db5d84c055d3c0b}\label{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_ab31c72b72ddaf7116db5d84c055d3c0b}} \index{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}!execute@{execute}} \index{execute@{execute}!Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}} \subsubsection{\texorpdfstring{execute()}{execute()}} {\footnotesize\ttfamily execute (\begin{DoxyParamCaption}\item[{Input\+Interface}]{\$input, }\item[{Output\+Interface}]{\$output }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \subsection{Field Documentation} \mbox{\Hypertarget{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a8f95e09946a819ad70ca9e71e8c0d153}\label{class_symfony_1_1_component_1_1_yaml_1_1_command_1_1_lint_command_a8f95e09946a819ad70ca9e71e8c0d153}} \index{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}!\$default\+Name@{\$default\+Name}} \index{\$default\+Name@{\$default\+Name}!Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command@{Symfony\+::\+Component\+::\+Yaml\+::\+Command\+::\+Lint\+Command}} \subsubsection{\texorpdfstring{\$default\+Name}{$defaultName}} {\footnotesize\ttfamily \$default\+Name = \textquotesingle{}lint\+:yaml\textquotesingle{}\hspace{0.3cm}{\ttfamily [static]}, {\ttfamily [protected]}} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item /\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/symfony/yaml/\+Command/\mbox{\hyperlink{_lint_command_8php}{Lint\+Command.\+php}}\end{DoxyCompactItemize} \documentclass{standalone} \RequirePackage{tikz} \RequirePackage{ifthen} \usetikzlibrary{arrows,shapes} \tikzstyle{btreeptr} = [draw, ultra thin, fill=blue!50, minimum height=1cm, inner sep=0cm, minimum width=0cm] \tikzstyle{btreeval} = [draw, semithick, fill=yellow!30, minimum size=1cm] \tikzstyle{btlink} = [draw, thick, ->, >=triangle 45] \newcommand{\xyshift}[3]{ \begin{scope}[xshift=#1, yshift=#2] #3 \end{scope} } \newcommand{\btreelink}[2]{ \draw[btlink] (#1.south) -- ([yshift=-0.1cm] #2.north); \& \fill (#1.south) circle[radius=0.15cm]; \\ } \newcommand{\btreenodea}[2]{ \matrix [ampersand replacement=\&] (#1) { \node[btreeptr] (#1-a) {\vphantom{1}}; \& \node[btreeval] {#2}; \& \node[btreeptr] (#1-b) {\vphantom{1}}; \\ }; } \newcommand{\btreenodeb}[3]{ \matrix [ampersand replacement=\&] (#1) { \node[btreeptr] (#1-a) {\vphantom{1}}; \& \node[btreeval] {#2}; \& \node[btreeptr] (#1-b) {\vphantom{1}}; \& \node[btreeval] {#3}; \& \node[btreeptr] (#1-c) {\vphantom{1}}; \\ }; } \newcommand{\btreenodec}[4]{ \matrix [ampersand replacement=\&] (#1) { \node[btreeptr] (#1-a) {\vphantom{1}}; \& \node[btreeval] {#2}; \& \node[btreeptr] (#1-b) {\vphantom{1}}; \& \node[btreeval] {#3}; \& \node[btreeptr] (#1-c) {\vphantom{1}}; \& \node[btreeval] {#4}; \& \node[btreeptr] (#1-d) {\vphantom{1}}; \\ }; } \newcommand{\btreenoded}[5]{ \matrix [ampersand replacement=\&] (#1) { \node[btreeptr] (#1-a) {\vphantom{1}}; \& \node[btreeval] {#2}; \& \node[btreeptr] (#1-b) {\vphantom{1}}; \& \node[btreeval] {#3}; \& \node[btreeptr] (#1-c) {\vphantom{1}}; \& \node[btreeval] {#4}; \& \node[btreeptr] (#1-d) {\vphantom{1}}; \& \node[btreeval] {#5}; \& \node[btreeptr] (#1-e) {\vphantom{1}}; \\ }; } \newcommand{\btreenodee}[6]{ \matrix [ampersand replacement=\&] (#1) { \node[btreeptr] (#1-a) {\vphantom{1}}; \& \node[btreeval] {#2}; \& \node[btreeptr] (#1-b) {\vphantom{1}}; \& \node[btreeval] {#3}; \& \node[btreeptr] (#1-c) {\vphantom{1}}; \& \node[btreeval] {#4}; \& \node[btreeptr] (#1-d) {\vphantom{1}}; \& \node[btreeval] {#5}; \& \node[btreeptr] (#1-e) {\vphantom{1}}; \& \node[btreeval] {#6}; \& \node[btreeptr] (#1-f) {\vphantom{1}}; \\ }; } \begin{document} \begin{tikzpicture} \tikzstyle{every node}=[font=\fontsize{15}{0}\selectfont] \xyshift{27.5mm}{-30mm}{\btreenoded{b}{1}{2}{3}{4}} \xyshift{82.5mm}{-30mm}{\btreenoded{c}{5}{6}{7}{8}} \xyshift{137.5mm}{-30mm}{\btreenoded{d}{11}{12}{13}{14}} \xyshift{192.5mm}{-30mm}{\btreenoded{e}{16}{17}{18}{19}} \xyshift{247.5mm}{-30mm}{\btreenodee{f}{21}{22}{23}{24}{25}} \xyshift{137.5mm}{0mm}{\btreenoded{a}{5}{10}{15}{20}} \btreelink{a-a}{b} \btreelink{a-b}{c} \btreelink{a-c}{d} \btreelink{a-d}{e} \btreelink{a-e}{f} \end{tikzpicture} \end{document} Arrangement_on_surface_2/doc/Arrangement_on_surface_2/fig_src/spherical_concept_hierarchy.tex1000+ \documentclass[12pt]{standalone} \input{header} \pagestyle{empty} \begin{document} \psset{treevshift=0,unit=1em,xunit=2em,yunit=1em,everytree={},etcratio=.75,triratio=.5} \jtree[everylabel=\sl,xunit=70pt,arrows=->] \! = {} [scaleby=2.81 0,arrows=-,linestyle=none]{\psframebox{\concept{ArrangementIdentifiedVerticalTraits\_2}}}@uvt !uvt ^[scaleby=1 0,arrows=-,linestyle=none]{\psframebox{\concept{ArrangementContractedBottomTraits\_2}}}@ubt !ubt ^[scaleby=2.73 0,arrows=-,linestyle=none]{\psframebox{\concept{ArrangementContractedTopTraits\_2}}}@utt !utt . \!ubt = [scaleby=1 1]{\psframebox{\concept{ArrangementSphericalBoundaryTraits\_2}}}@ut !ut . \ncline{uvt:b}{ut:t} \ncline{utt:b}{ut:t} \endjtree \psset{treevshift=0,unit=1cm,xunit=1cm,yunit=1cm,everytree={}, etcratio=.75,triratio=.5} \end{document} \subsection{Kernel Entry Layer Model} A thread performing kernel operations is ``controlled'' by the kernel entry point and the return code. All kernel entry points are ``clean.'' A thread that gets canceled will never reveal itself to be in the middle of a kernel operation; it will always appear to the canceling thread that it is about to begin a kernel operation, or that it is in user mode. To prevent long or complicated kernel functions from becoming a bottleneck, these operations are broken into sub-sequences. For example the long {\tt fluke_ipc_client_connect_send_over_receive()} operation that connects to a server, sends a request, waits for a reply and receives it into a buffer, can take quite a while. As portions of the operation are completed the kernel entrypoint is advanced. So, in this example, after the ``connect'' phase is completed, the kernel sets the thread's entrypoint to be {\tt fluke_ipc_send_over_receive().} \begin{figure} {\small \begin{enumerate} \item {\tt KR_USER_EXCEPTION} A processor exception occurred which should be blamed on the user (e.g. because the exception was generated while accessing user space). The current thread's exception_state contains the details. \item {\tt KR_PAGE_FAULT} Page fault occurred. This gets turned into a real KR_USER_EXCEPTION by the kentry layer if the page fault cannot be resolved in the kernel and no appropriate region keeper can be found to handle the fault \item {\tt KR_CANCEL} Another thread is trying to manipulate us and has asynchronously canceled us, e.g. due to thread_interrupt(), thread_get_state(), thread_set_state(). \item {\tt KR_NO_MEMORY} Ran out of kernel memory. \item {\tt KR_RESTART} This return code indicates that we have context switched due to a wait, and we need to restart execution in user mode before doing anything else in case dependent things have changed. \end{enumerate} } \caption{The set of return codes used within the kernel} \label{ReturnCodes-fig} \end{figure} All of the major functions within the kernel return one of a small, well defined set of error codes. See Figure~\ref{ReturnCodes-fig} for a complete list. The return code is used to signal special conditions. By convention, any function returning a non-zero return code signals that the kernel operation in progress has been canceled or must be restarted. \hypertarget{namespaceDNAStrandTest}{}\doxysection{D\+N\+A\+Strand\+Test Namespace Reference} \label{namespaceDNAStrandTest}\index{DNAStrandTest@{DNAStrandTest}} \doxysubsection*{Classes} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{classDNAStrandTest_1_1DNAStrandTest}{D\+N\+A\+Strand\+Test}} \begin{DoxyCompactList}\small\item\em Class for testing certain aspects of the behavior of \mbox{\hyperlink{namespaceDNAStrand}{D\+N\+A\+Strand}}. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Detailed Description} Class for testing the \mbox{\hyperlink{namespaceDNAStrand}{D\+N\+A\+Strand}} matching. \begin{DoxyAuthor}{Author} \end{DoxyAuthor} \begin{DoxySince}{Since} 05/02/2020 \end{DoxySince} doc/manual/insns/ADC.tex \begin{instruction}{ADC}{Add Register with Carry} \begin{encoding} \mnemonic & \op{5}{00010} & \reg{d} & \reg{a} & \op{2}{01} & \reg{b} \\ \end{encoding} \assembly{\mnemonic{} Rd, Ra, Rb} \purpose{To add 16-bit integers in registers, with carry input.} \restrictions{None.} \begin{operation}\aluRR{opA + opB + C}\wb\flagZSCV\end{operation} \begin{remarks} A 32-bit addition with both operands in registers can be performed as follows: \begin{alltt} ; Perform \string{R1, R0\string} ← \string{R3, R2\string} + \string{R5, R4\string} ADD R0, R2, R4 ADC R1, R3, R5 \end{alltt} \end{remarks} \end{instruction} \hypertarget{classserver_1_1Model}{}\doxysection{server.\+Model Class Reference} \label{classserver_1_1Model}\index{server.Model@{server.Model}} Inheritance diagram for server.\+Model\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{classserver_1_1Model} \end{center} \end{figure} \doxysubsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{classserver_1_1Model_a9a37ce5d48d65d1f73e3310dd540b5cd}\label{classserver_1_1Model_a9a37ce5d48d65d1f73e3310dd540b5cd}} def {\bfseries \+\_\+\+\_\+init\+\_\+\+\_\+} (self, seed) \item \mbox{\Hypertarget{classserver_1_1Model_a24d947d23fffd4f1fa2adffdffa7ad46}\label{classserver_1_1Model_a24d947d23fffd4f1fa2adffdffa7ad46}} def {\bfseries start} (self) \item \mbox{\Hypertarget{classserver_1_1Model_a6ea5ba19abb493d10867d27edba9bb82}\label{classserver_1_1Model_a6ea5ba19abb493d10867d27edba9bb82}} def {\bfseries run} (self) \item \mbox{\Hypertarget{classserver_1_1Model_a3c6c94870bd808a1a8868f6c68610828}\label{classserver_1_1Model_a3c6c94870bd808a1a8868f6c68610828}} def {\bfseries enqueue} (self, func, $\ast$args, $\ast$$\ast$kwargs) \item \mbox{\Hypertarget{classserver_1_1Model_ad0103204c6bad7d0409368d426e971fb}\label{classserver_1_1Model_ad0103204c6bad7d0409368d426e971fb}} def {\bfseries dequeue} (self) \item \mbox{\Hypertarget{classserver_1_1Model_a3666c45a716bd0b1ee7d58b16e695be4}\label{classserver_1_1Model_a3666c45a716bd0b1ee7d58b16e695be4}} def {\bfseries execute} (self, $\ast$args, $\ast$$\ast$kwargs) \item \mbox{\Hypertarget{classserver_1_1Model_ab427a4018c7e7144f0d7dedfe7e87957}\label{classserver_1_1Model_ab427a4018c7e7144f0d7dedfe7e87957}} def {\bfseries commit} (self) \item \mbox{\Hypertarget{classserver_1_1Model_a010bcdc35e6520fc20eaad27781960d6}\label{classserver_1_1Model_a010bcdc35e6520fc20eaad27781960d6}} def {\bfseries create\+\_\+tables} (self) \item \mbox{\Hypertarget{classserver_1_1Model_abeaec0fd82dbf66b7ca21bb464517c20}\label{classserver_1_1Model_abeaec0fd82dbf66b7ca21bb464517c20}} def {\bfseries get\+\_\+default\+\_\+block} (self, x, y, z) \item \mbox{\Hypertarget{classserver_1_1Model_a3acb5e5f1b3c11a61caf47887414d0e1}\label{classserver_1_1Model_a3acb5e5f1b3c11a61caf47887414d0e1}} def {\bfseries get\+\_\+block} (self, x, y, z) \item \mbox{\Hypertarget{classserver_1_1Model_a3e13a7313fc94f05ecf271dd4a9664d1}\label{classserver_1_1Model_a3e13a7313fc94f05ecf271dd4a9664d1}} def {\bfseries next\+\_\+client\+\_\+id} (self) \item \mbox{\Hypertarget{classserver_1_1Model_a5ab44bf7f5122e3b5572079e02ff0b92}\label{classserver_1_1Model_a5ab44bf7f5122e3b5572079e02ff0b92}} def {\bfseries on\+\_\+connect} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_a2307c8a450ed2e43325c7d182fcf9b69}\label{classserver_1_1Model_a2307c8a450ed2e43325c7d182fcf9b69}} def {\bfseries on\+\_\+data} (self, client, data) \item \mbox{\Hypertarget{classserver_1_1Model_ab461be206915068246242c80eb7d75bb}\label{classserver_1_1Model_ab461be206915068246242c80eb7d75bb}} def {\bfseries on\+\_\+disconnect} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_a7a130d4cef3951790bf3f27b9a2114c6}\label{classserver_1_1Model_a7a130d4cef3951790bf3f27b9a2114c6}} def {\bfseries on\+\_\+version} (self, client, version) \item \mbox{\Hypertarget{classserver_1_1Model_a2e228fba3228beb3da42cd5506cd5606}\label{classserver_1_1Model_a2e228fba3228beb3da42cd5506cd5606}} def {\bfseries on\+\_\+authenticate} (self, client, username, access\+\_\+token) \item \mbox{\Hypertarget{classserver_1_1Model_a210e38620f762c3ae492b8f6e4d78a3a}\label{classserver_1_1Model_a210e38620f762c3ae492b8f6e4d78a3a}} def {\bfseries on\+\_\+chunk} (self, client, p, q, key=0) \item \mbox{\Hypertarget{classserver_1_1Model_a8d5d9e3cda0e8aa384b5f5f2f45cd4bd}\label{classserver_1_1Model_a8d5d9e3cda0e8aa384b5f5f2f45cd4bd}} def {\bfseries on\+\_\+block} (self, client, x, y, z, w) \item \mbox{\Hypertarget{classserver_1_1Model_ae356a9dde121cdda833d37b96012fa98}\label{classserver_1_1Model_ae356a9dde121cdda833d37b96012fa98}} def {\bfseries on\+\_\+light} (self, client, x, y, z, w) \item \mbox{\Hypertarget{classserver_1_1Model_a5b0e4c36467deb17ff65f6559d15e139}\label{classserver_1_1Model_a5b0e4c36467deb17ff65f6559d15e139}} def {\bfseries on\+\_\+sign} (self, client, x, y, z, face, $\ast$args) \item \mbox{\Hypertarget{classserver_1_1Model_ad1f30bfa4395afc29844822700c296db}\label{classserver_1_1Model_ad1f30bfa4395afc29844822700c296db}} def {\bfseries on\+\_\+position} (self, client, x, y, z, rx, ry) \item \mbox{\Hypertarget{classserver_1_1Model_a56115cc5a89c16019a31deb7b4165c8d}\label{classserver_1_1Model_a56115cc5a89c16019a31deb7b4165c8d}} def {\bfseries on\+\_\+talk} (self, client, $\ast$args) \item \mbox{\Hypertarget{classserver_1_1Model_a2563a1a444d9ae83b1513d1e2ef35d54}\label{classserver_1_1Model_a2563a1a444d9ae83b1513d1e2ef35d54}} def {\bfseries on\+\_\+nick} (self, client, nick=None) \item \mbox{\Hypertarget{classserver_1_1Model_a73a69870e10b530cf11b257b14b7d24a}\label{classserver_1_1Model_a73a69870e10b530cf11b257b14b7d24a}} def {\bfseries on\+\_\+spawn} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_aca0fbfd43f14f14ce061e0c01b81d00d}\label{classserver_1_1Model_aca0fbfd43f14f14ce061e0c01b81d00d}} def {\bfseries on\+\_\+goto} (self, client, nick=None) \item \mbox{\Hypertarget{classserver_1_1Model_a2eca244ca01a155ce894d286f22cb1a0}\label{classserver_1_1Model_a2eca244ca01a155ce894d286f22cb1a0}} def {\bfseries on\+\_\+pq} (self, client, p, q) \item \mbox{\Hypertarget{classserver_1_1Model_af8707fd1875a0d52f50c829e143eaa9f}\label{classserver_1_1Model_af8707fd1875a0d52f50c829e143eaa9f}} def {\bfseries on\+\_\+help} (self, client, topic=None) \item \mbox{\Hypertarget{classserver_1_1Model_ae6b5f5c24fa673873cf459d12996e635}\label{classserver_1_1Model_ae6b5f5c24fa673873cf459d12996e635}} def {\bfseries on\+\_\+list} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_aa37bc39f34c8cb3a0f72c47c6cb44f6e}\label{classserver_1_1Model_aa37bc39f34c8cb3a0f72c47c6cb44f6e}} def {\bfseries send\+\_\+positions} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_af0f14f4dd4c97c00023ba014ae6974b0}\label{classserver_1_1Model_af0f14f4dd4c97c00023ba014ae6974b0}} def {\bfseries send\+\_\+position} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_ab39036fff36add7d7dced2892d4f7749}\label{classserver_1_1Model_ab39036fff36add7d7dced2892d4f7749}} def {\bfseries send\+\_\+nicks} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_a9acbf2a15a0a96e0232b9ac22173496d}\label{classserver_1_1Model_a9acbf2a15a0a96e0232b9ac22173496d}} def {\bfseries send\+\_\+nick} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_a1fb58f37e3b98007a6e220a18d0abb4d}\label{classserver_1_1Model_a1fb58f37e3b98007a6e220a18d0abb4d}} def {\bfseries send\+\_\+disconnect} (self, client) \item \mbox{\Hypertarget{classserver_1_1Model_a24dcce0e4633051852b7bff4901ef13d}\label{classserver_1_1Model_a24dcce0e4633051852b7bff4901ef13d}} def {\bfseries send\+\_\+block} (self, client, p, q, x, y, z, w) \item \mbox{\Hypertarget{classserver_1_1Model_a06aaa76ac9e9f36bbc9fd35d131b150d}\label{classserver_1_1Model_a06aaa76ac9e9f36bbc9fd35d131b150d}} def {\bfseries send\+\_\+light} (self, client, p, q, x, y, z, w) \item \mbox{\Hypertarget{classserver_1_1Model_ad9bb38de307f92f2e34059e8c21217b8}\label{classserver_1_1Model_ad9bb38de307f92f2e34059e8c21217b8}} def {\bfseries send\+\_\+sign} (self, client, p, q, x, y, z, face, text) \item \mbox{\Hypertarget{classserver_1_1Model_ac72b296e1388a69afa59ccb6a1412770}\label{classserver_1_1Model_ac72b296e1388a69afa59ccb6a1412770}} def {\bfseries send\+\_\+talk} (self, text) \end{DoxyCompactItemize} \doxysubsection*{Public Attributes} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{classserver_1_1Model_a6107594ebcdc6ff20794651a2f12e5b7}\label{classserver_1_1Model_a6107594ebcdc6ff20794651a2f12e5b7}} {\bfseries world} \item \mbox{\Hypertarget{classserver_1_1Model_ae8aefbf418082334aba83df648f7597e}\label{classserver_1_1Model_ae8aefbf418082334aba83df648f7597e}} {\bfseries clients} \item \mbox{\Hypertarget{classserver_1_1Model_a65c1208dbbe27fc4c2ebf9dbb465e8b2}\label{classserver_1_1Model_a65c1208dbbe27fc4c2ebf9dbb465e8b2}} {\bfseries queue} \item \mbox{\Hypertarget{classserver_1_1Model_a9cc9e8300304ed383f576d97b357b074}\label{classserver_1_1Model_a9cc9e8300304ed383f576d97b357b074}} {\bfseries commands} \item \mbox{\Hypertarget{classserver_1_1Model_a44f6ce8a304b6e72d5bbe4bcbd6795db}\label{classserver_1_1Model_a44f6ce8a304b6e72d5bbe4bcbd6795db}} {\bfseries patterns} \item \mbox{\Hypertarget{classserver_1_1Model_afb3cee7bcfdd6157b3607d823ab5145e}\label{classserver_1_1Model_afb3cee7bcfdd6157b3607d823ab5145e}} {\bfseries connection} \item \mbox{\Hypertarget{classserver_1_1Model_af17091b5639b2ea262bf50746c532dd6}\label{classserver_1_1Model_af17091b5639b2ea262bf50746c532dd6}} {\bfseries last\+\_\+commit} \end{DoxyCompactItemize} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item server.\+py\end{DoxyCompactItemize} % Digits of pi using the Rabinowitz and Wagon spigot algorithm [1] % % This TeX script calculates the first n digits of pi. It uses \count registers % for working memory and is runnable on both Texcraft and pdfTeX. The variable n % is specified at the end of the script, and is restricted as follows: % % - pdfTeX: n cannot be much larger than 1000 as pdfTeX runs out of stack space. % % - Texcraft: n cannot be larger than 7559 as the algorithm requires n + (10n)//3 + 12 % integers in memory, and we assume 2^15 registers, as in pdfTeX. % The playground has a version of this script that uses Texcraft allocation commands % instead of \count registers and is thus not subject to this restriction. % But also, for some currently unknown value of n, 32-bit integer overflow will occur % and the result will be incorrect. From the original paper [1] we know that this n % is at least 5000. % % \count register layout: % % - [0, 25196): used for the length 10n/3 working array % - [25196, 32756): used for storing the n results % - [32756, 32768=2^15): used for 12 named variables via \countdef % % [1] http://www.cs.williams.edu/~heeringa/classes/cs135/s15/readings/spigot.pdf % While loop: repeatedly executes #2 while `\ifnum #1` is true \def\while#1#2{ \ifnum #1 #2% \while{#1}{#2}\fi } % Modulus: calculates `#1 % #2` and puts the result in #1 \countdef \tempTwo 32756 \def\modulus#1#2{ \tempTwo = #1 \divide \tempTwo by #2 \multiply \tempTwo by #2 \multiply \tempTwo by -1 \advance \tempTwo by #1 #1= \tempTwo } \countdef \n 32767 % Any changes to the following line will break the benchmark Rust code as it assumes % the string '\n = 100' appears exactly. \n = 100 \def\result{\count} % \newarray\result\n \countdef \resultIndex 32766 \resultIndex = 25196 % allocate an array of length (10n)/3 \countdef \len 32765 \len = \n \multiply \len by 10 \divide \len by 3 \def\r{\count} % initialize each element of the array to 2 \countdef \i 32764 \i = 0 \while{\i < \len}{ \r \i = 2 \advance \i by 1 } \countdef \j 32763 \countdef \k 32762 % todo: can remove \countdef \carry 32761 \countdef \preDigit 32760 \countdef \firstPreDigit 32759 \firstPreDigit = -1 \countdef \numTrailingPreDigits 32758 \numTrailingPreDigits = 0 \countdef \outerLoopIndex 32757 \outerLoopIndex = 0 \while{\outerLoopIndex < \n}{ \advance \outerLoopIndex by 1 % \i = \len \while{\i > 0}{ \advance \i by -1 % % r[i] = r[i] * 10 + carry \multiply \r \i by 10 \advance \r \i by \carry % % Calculate j = 2i+1 \j = 2 \multiply \j by \i \advance \j by 1 % % carry = (r[i]/j)(i) \carry = \r \i \divide \carry by \j \multiply \carry by \i % % r[i] = r[i] % j \ifnum \i > 0 \modulus{\r\i}{\j} \fi } % \preDigit = \r 0 \divide \preDigit by 10 % \ifnum \preDigit < 9 \ifnum \firstPreDigit > -1 \result\resultIndex = \firstPreDigit \advance\resultIndex by 1 \fi \while{\numTrailingPreDigits > 0}{ \result\resultIndex = 9 \advance\resultIndex by 1 \advance\numTrailingPreDigits by -1 } \firstPreDigit = \preDigit \fi \ifnum \preDigit = 9 \advance\numTrailingPreDigits by 1 \fi \ifnum \preDigit = 10 \ifnum \firstPreDigit < 9 \advance \firstPreDigit by 1 \else \firstPreDigit = 0 \fi \result\resultIndex = \firstPreDigit \advance\resultIndex by 1 \while{\numTrailingPreDigits > 0}{ \result\resultIndex = 0 \advance\resultIndex by 1 \advance\numTrailingPreDigits by -1 } \firstPreDigit = 0 \fi \modulus{\r 0}{10} } \while{\resultIndex < \n}{\result\resultIndex-1\advance\resultIndex1} \advance\n by 25196 \i=25196 \while{\i<\n}{% \ifnum \result\i > -1 \the\result\i \fi \advance\i by 1% } % To run on pdfTeX, uncomment the following line. \end doc/report/rev.tex0 \gdef\therev{c6957fc} \gdef\thedate{2021-04-22 19:55:02 +0800} \documentclass[10pt, a4paper]{article} \usepackage[margin=2cm]{geometry} % \usepackage{mdwlist} \usepackage[T1]{fontenc} \usepackage[utf8]{inputenc} \usepackage{textcomp} \pagestyle{empty} \setlength{\tabcolsep}{0em} \usepackage{hyperref} \usepackage[T1]{fontenc} \usepackage{fouriernc} % \usepackage{multibib} % \newcites{theses}{Theses} % \newcites{invited}{Invited Talk} % \newcites{articles}{Peer-reviewed Journal Articles} % \newcites{conferences}{Conference and Seminar Presentations} % from http://texblog.org/2012/04/25/writing-a-cv-in-latex/ \usepackage{array, xcolor} \definecolor{lightgray}{gray}{0.8} \newcolumntype{L}{>{\raggedleft}p{0.14\textwidth}} \newcolumntype{R}{p{0.8\textwidth}} \newcommand\VRule{\color{lightgray}\vrule width 2.0pt} \setlength{\tabcolsep}{5pt} \renewcommand{\arraystretch}{1.1} \usepackage{titlesec} \titleformat*{\section}{\large\bfseries} \renewcommand{\refname}{Relevant Communications} \begin{document} \fontencoding{T1}\selectfont \begin{center} {\LARGE \textbf{}} \\ Centre for Ecological and Evolutionary Synthesis (CEES)\\ Department of Biosciences, University of Oslo\\ P.O. Box 1066 Blindern\ \ \textbullet \ \ 0316 Oslo\ \ \textbullet \ \ Norway \\ \end{center} \hrule \vspace{-0.4em} \section*{Summary} I am an environmental scientist with a specialisation in scientific high-performance computing. My application fields to date have been \textbf{climate change}, \textbf{anthropogenic pollution}, \textbf{biogeochemistry}, and \textbf{freshwater systems}. My interests in computing and numerical methods include \textbf{machine learning}, \textbf{parameter estimation}, \textbf{quantification of modelling uncertainty} and \textbf{data visualisation}. Highlights in HPC experiences: \begin{itemize} \itemsep-0.1em \item Environmental model implementation on \textbf{3 HPC platforms} (Notur/Norstore, Sharcnet, Amazon EC2), \item Management of terabyte data arrays, and \item Expert skills in \textbf{Unix,} \textbf{Python} and \textbf{R} languages. \end{itemize} \input{education} \input{exchanges} \input{work} \nocite{couture_modelling_2014} \nocite{tominaga_lake_2013} \nocite{tominaga_future_2012} \nocite{tominaga_predicting_2010} \nocite{tominaga_voyage_2009-1} \bibliographystyle{plos2009} \bibliography{koji} \flushleft In total, I have 6 peer-reviewed publications showcasing the uses of HPC methods, approximately 30 conference abstracts, 4 times peer-reviewer contributions, and attended numerous international scientific meetings in 13 countries. \end{document} @incollection{yu2016cold, author = " and and and and ", title = "International Young Physicists' Tournament", editor = " and ", booktitle = "International Young Physicists' Tournament", publisher = "World Scientific Publishing Company", year = 2016, pages = "113-131", chapter = 9, }0 \documentclass[letterpaper]{article} \usepackage{fullpage} \usepackage{nopageno} \usepackage{amsmath} \usepackage{amssymb} \usepackage[utf8]{inputenc} \allowdisplaybreaks \newcommand{\abs}[1]{\left\lvert #1 \right\rvert} \begin{document} \title{Homework} \date{September 22, 2014} \author{} \maketitle \begin{enumerate} \item Habitez-vous en ville ou à la campagne? J'habite en ville. \item Connaissez-vous bien vos voisins? Je connais bien certains de mes voisins. \item Rencontrez-vous souvent dans la rue quelqu'un que vous connaissez? Quelquefois je rencontre dans la rue quelqu'un que je connais. Je vous ai rencontrée hier. \item Faites-vous facilement des rencontres (amicales ou romantiques) ici à l'université (ou ici à Fargo)? Non, je ne suis pas une personne extroverte. Il est difficile faire des rencontres. \item Est-ce qu'habiter en ville rapproche ou éloigne les gens? Habiter en ville rapproche les gens. Il les aide à trouver leur tribu. Mais il faut plus de temps faire des rencontres. \end{enumerate} \end{document} @inproceedings{li_lin_multicasting_2001, abstract = {This paper addresses the problem of multicasting and broadcasting in undirected wavelength-division multiplexing (WDM) networks. Given an undirected network G=(VE) with a source nodes and a set of destination nodes D, /spl Lambda/ is the set of wavelength that can be used in the network. Associated with every edge e, there is a set of available wavelengths on it. The multicast problem is to find a tree rooted at s including all nodes in D such that the cost of the tree is minimum in terms of the cost of wavelength conversion at nodes and the cost of using wavelength on edges. This paper proves that the multicast problem is NP-complete and can not be approximated within a constant factor, unless P=NP. Then we construct an auxiliary graph for the original WDM networks and reduce the multicast problem to a group Steiner tree problem on the auxiliary graph. Employing the known algorithm for the group Steiner tree problem, we derive an algorithm for the problem, which delivers a solution within O(log/sup 2/ (nk)loglog(nk)logp) times the optimum.}, author = { and and and }, booktitle = {Proceedings 2001 International Conference on Computer Networks and Mobile Computing}, doi = {10.1109/ICCNMC.2001.962634}, keywords = {Bandwidth, computational complexity, broadcasting, Costs, Approximation algorithms, Multicast algorithms, multicast communication, NP-complete problem, auxiliary graph, Broadcasting, Intelligent networks, multicasting, optical communication, optical communication network, Optical fiber communication, Optical wavelength conversion, source node, Steiner tree, trees (mathematics), wavelength division multiplexing, Wavelength division multiplexing, wavelength-division multiplexing, WDM networks}, month = {October}, pages = {467--472}, title = {Multicasting and broadcasting in undirected WDM networks}, year = {2001} } \documentclass[__main__.tex]{subfiles} \begin{document} \qtitle{33} Описание пространства сплайнов третьей степени единичного дефекта, его стандартный базис. Формулировка теоремы о корректности интерполирования дефектными сплайнами нулевой, первой, второй и третьей степеней непрерывной на отрезке функции.\\ Пусть $A = \langle a = \tau_0, \tau_1, ..., \tau_k = b\rangle$ - сетка $[a, b]$.\\ $^>y = [y_0, y_1, ..., y_n\rangle \in\;^>\mathbb{R}^{|A|}(A)$ и $(\delta_1, ..., \delta_k)$ - подразделения $[a, b]$ на подотрезки, где $k = [\tau_{i-1}, \tau_i]$ для $i = \overline{1, k}$.\\ Сплайн $\varphi = spl_3(A, \;^>y)$ интерполяция степени 3(дефект 1) на $[a, b]$ определяется сплайном полиномов $(P_1, ..., P_k)$, где $P_i(\tau) = \varphi(\tau) = a_i + b_i(\tau - \tau_{i-1}) + c_i(\tau - \tau_i)^2 + d_i(\tau - \tau_{i-1})^3$ для $\tau \in \delta_i(i = \overline{1, k})$. При этом признаётся, что \begin{gather} \begin{cases} c_1 = c_k = 0\\ P_i(\tau_{i-1}) = y_{i-1} \;\; для\; i = \overline{1, k}\;\; P_k(\tau_k) = y_k\\ P_i(\tau_i) = P_{i+1}(\tau_i) \;\; для\; i = \overline{1, (k-1)}\; (склейка\; значений)\\ P'_i(\tau_i) = P'_{i+1}(\tau_i) \;\; для\; i = \overline{1, (k-1)}\; (склейка\; производных)\\ P''_i(\tau_i) = P''_{i+1}(\tau_i) \;\; для\; i = \overline{1, (k-1)}\; (склейка\; вторых\; производных) \end{cases} \label{33-systema} \end{gather} Условия \ref{33-systema} обеспечивают определение полиномов $P_1, ..., P_k$. Для пространства $Spl_3(A)$ интерполируемых сплайнов третьей степени (дефект 1) на сетке $A$, индуцируемого условием \ref{33-systema}, вводится базис $H = (h_0, ..., h_k)$, в котором $h_i = spl_3(A, \;^>c_{k+1})$ для $i = \overline{0, k}$ где $(^>c_1, ..., \;^>c_{|A|})$ - стандартный базис $^>\mathbb{R}^{|A|}$.\\ Тогда $spl_3(A,\; ^>y) = \sum_{i=0}^{+\infty}{y_ispl_3(A, \;^>c_{i+1})}$ \begin{theorem} а) Корректность интерполирования непрерывной на $[a, b]$ функции для систем сеток с помощью интерполируемых сплайнов третьей степени дефекта 1.\\ б) Если $f \in \underline{C}'([a, b], \mathbb{R})$ и $A_{(.)} = (A_k)_{\mathbb {N}}$ - схема сеток.\\ $f' = \underset{k \rightarrow \infty}{lim}{spl_3}(A_k, \hat{A_k}(spl'_3(A, \;^>y)))$ \end{theorem} \end{document}ayushmathur/ostack-hpc \noindent {\bf Base Linux Edition}: This edition of the guide highlights installation without the use of a companion configuration management system and directly uses distro-provided package management tools for component selection. The steps that follow also highlight specific changes to system configuration files that are required as part of the cluster install process. %% %% $Id$ %% %% Copyright 1989-2014 MINES ParisTech %% %% This file is part of PIPS. %% %% PIPS is free software: you can redistribute it and/or modify it %% under the terms of the GNU General Public License as published by %% the Free Software Foundation, either version 3 of the License, or %% any later version. %% %% PIPS is distributed in the hope that it will be useful, but WITHOUT ANY %% WARRANTY; without even the implied warranty of MERCHANTABILITY or %% FITNESS FOR A PARTICULAR PURPOSE. %% %% See the GNU General Public License for more details. %% %% You should have received a copy of the GNU General Public License %% along with PIPS. If not, see . %% %\documentstyle[psfig,11pt]{article} %\title{Code Generation for distributed memory machines} %\author{} %\begin{document} %\maketitle To provide accurate results in reasonable time, real applications such as seismic, fluid mechanics and structure computation applications (that use large set of data and costly algorithms) need to be widely parallelized. Distributed memory machines are good candidates for these applications because they support large number of processors without the shared memory bottle neck problem. However, distributed memory machines are much more difficult to program efficiently than shared memory machines. Shared memories are often managed by the hardware. Conscientious programmer must only restructure its program in order to take care of cache use. In distributed memory machines, the distribution of data onto the local memories must be designed carefully in order to keep good performance and to avoid too much communications. Many approaches to generate distributed code have been suggested. Some are language-based. The programmer has to specify the data distribution pattern and the parallelism but he does not have to generate processes nor send/receive pairs between processors. Processes are automatically derived from the data distribution using the so-called {\em owner computes rule} and a SPMD model: each processor executes instruction producing a piece of data located in its memory. Send/recei\-ve pairs are derived from the data distribution and from the instruction reference patterns. Other approaches are based on the operating system or/and hardware support. A virtual shared memory provides the uniform address space. Some software mechanisms or complex cache systems maintains the memory consistency. Three different approaches have been implemented in Pips. The first one is language-based. High Performance Fortran is designed to express data parallel applications for distributed memory machines. It provides the user with a set of directives to specify data distribution and parallelism. A prototype HPF compiler is implemented in Pips. The second approach is also language-based but the data distribution and the scheduling are automatically computed. Designed for static control programs, this technique generates SIMD programs expressed in CRAFT (Fortran CRAY-T3D) or CM-Fortran. Finally, the third approach suggests the emulation of a shared memory onto a distributed memory machine. Classical parallelization techniques are used to generate SPMD code. The compiler manages the emulated shared memory and the maintenance of memory coherency. The following section introduces some characteristics of these three approaches. It focuses on the scheduling differences. More details are given in the associated presentation papers. Example presented in Figure~\ref{ex1} illustrates the approaches. \begin{figure}[htp] \begin{verbatim} DO I=1,10 S1 B(I,I)=1 ENDDO DO I=1,10 DO J= I+1,10 S2 B(I,J)=1 S3 A(I,J)=B(I,J-1) ENDDO ENDDO \end{verbatim} \caption{Program 1} \label{ex1} \end{figure} \subsubsection{HPF program} In HPF, the data distribution and the parallelism are specified via directives. The parallel execution of the distributed application is guided by the {\it owner computes} rule: a processor can update only its local variables. Let's take our example. \begin{figure}[htp] \psfig{file=figures/hpfc1.idraw} \caption{Data distribution} \label{hpfc} \end{figure} Due to the data dependences existing on the second dimension of Array B, the data distribution that minimizes the communications groups the array elements by rows. Here, blocks of 2 rows are distributed onto the 5 processor local memories as depicted in Figure~\ref{direct}. \begin{figure}[htp] \begin{verbatim} CHPF$ ALIGN A WITH B CHPF$ PROCESSORS p1(5) CHPF$ DISTRIBUTE B(block,*) ONTO p1 \end{verbatim} \caption{ HPF Directives} \label{direct} \end{figure} According to the {\it owner computes} rule, 2 blocks of iterations of \verb+J+ are executed on each processor. The corresponding generated code is presented in Figure \ref{hpfcc}. \begin{figure}[htp] \begin{verbatim} C DO I=2*PROC_ID, 2*PROC_ID+1 S1 B(I,I)=1 ENDDO DO I=2*PROC_ID, 2*PROC_ID+1 DO J= I+1,10 S2 B(I,J)=1 S3 A(I,J)=B(I,J-1) ENDDO ENDDO \end{verbatim} \caption{Code generated from HPF} \label{hpfcc} \end{figure} The "HPFC" presentation details all the characteristics of our HPF compiler. \subsubsection{Automatic placement} The problem of solving automatically data and control distributions is NP-complet. Thus, they are sequentially solved. In the suggested approach, scheduling is first computed. Then, the data distribution computes the mapping of each array onto a virtual processor grid so as to minimize the communication cost. First, the array data flow graph is built. It characterizes the precedence relations between instruction instances and contains only true dependences because the program is put into a single assignment form. Figure \ref{plac} presents these precedence relations for program 1. Instructions \verb+S1+ and \verb+S2+ that assign array elements of B, should be executed before instruction \verb+S3+. Some array elements are assigned by \verb+S1+ and others by \verb+S2+. The resulting code generated from the DFG is presented in Figure \ref{plac2} \begin{figure}[htp] \psfig{file=figures/plac.idraw} \caption{Precedence Relations} \label{plac} \end{figure} \begin{figure}[htp] \begin{verbatim} CDIR$ SHARED B1(:BLOCK, :BLOCK) CDIR$ SHARED B2(:BLOCK, :BLOCK) CDIR$ SHARED A(:BLOCK, :BLOCK) CDIR$ DOSHARED(I) ON B1(I,I) DO I=1,9 B1(I,I)=1 ENDDO CDIR$ DOSHARED(I,J) ON B2(I,J) DO I=1,10 DO J= I+1,10 B2(I,J)=1 ENDDO ENDDO CDIR$ DOSHARED(I) ON A(I,I+1) DO I=1,9 J= I+1 A(I,J)=B1(I,J-1) ENDDO ENDDO CDIR$ DOSHARED(I,J) ON A(I,J) DO I=1,8 DO J= I+2,10 A(I,J)=B2(I,J-1) ENDDO ENDDO \end{verbatim} \caption{CRAFT Code} \label{plac2} \end{figure} This technique is detailed in the "Polyhedric method" presentation. \subsubsection{Emulated Shared Memory} Since the management of data distribution is complex, this approach suggests the emulation of a shared memory onto a distributed memory machine. Control distribution is applied through {\it tiling}. The iteration domain is partitioned into tiles that can be executed concurrently on different processors. Figure \ref{wp65d} illustrates this tiling and the assignment to the processors. Tiles are distributed in a cyclic way. \begin{figure}[htp] \psfig{file=figures/wp651.idraw} \caption{Control distribution} \label{wp65d} \end{figure} Data distribution is implicit. One half of the processors perform computations and the other half emulate memory banks. Figure \ref{wp65c} presents the corresponding generated code for the computations. The necessary communications are inserted in the generated code automatically by the compiler. The "Distributed code generation" presentation details all the compilation phases of this approach. \begin{figure}[htp] \begin{verbatim} DO I=PROC_ID,10,5 S1 B(I,I)=1 ENDDO DO I=PROC_ID,10,5 DO J= I+1,10 S2 B(I,J)=1 S3 A(I,J)=B(I,J-1) ENDDO ENDDO \end{verbatim} \caption{Wp65 code} \label{wp65c} \end{figure} %\end{document} @preamble{ " \newcommand{\noop}[1]{} " } % a do-nothing command that serves a purpose @ARTICLE{astropy, author = {{Astropy Collaboration} and {Robitaille}, T.~P. and {Tollerud}, E.~J. and {Greenfield}, P. and {Droettboom}, M. and {Bray}, E. and {Aldcroft}, T. and {Davis}, M. and {Ginsburg}, A. and {Price-Whelan}, A.~M. and {Kerzendorf}, W.~E. and {Conley}, A. and {Crighton}, N. and {Barbary}, K. and {Muna}, D. and {Ferguson}, H. and {Grollier}, F. and {Parikh}, M.~M. and {Nair}, P.~H. and {Unther}, H.~M. and {Deil}, C. and {Woillez}, J. and {Conseil}, S. and {Kramer}, R. and {Turner}, J.~E.~H. and {Singer}, L. and {Fox}, R. and {Weaver}, B.~A. and {Zabalza}, V. and {Edwards}, Z.~I. and {}, K. and {Burke}, D.~J. and {Casey}, A.~R. and {Crawford}, S.~M. and {Dencheva}, N. and {Ely}, J. and {Jenness}, T. and {Labrie}, K. and {Lim}, P.~L. and {Pierfederici}, F. and {Pontzen}, A. and {Ptak}, A. and {Refsdal}, B. and {Servillat}, M. and {Streicher}, O.}, title = "{Astropy: A community Python package for astronomy}", journal = {\aap}, archivePrefix = "arXiv", eprint = {1307.6212}, primaryClass = "astro-ph.IM", keywords = {methods: data analysis, methods: miscellaneous, virtual observatory tools}, year = 2013, month = oct, volume = 558, eid = {A33}, pages = {A33}, doi = {10.1051/0004-6361/201322068}, adsurl = {http://adsabs.harvard.edu/abs/2013A%26A...558A..33A}, adsnote = {Provided by the SAO/NASA Astrophysics Data System} } @ARTICLE{emcee, author = {{Foreman-Mackey}, D. and {. and {. and {. }, title = "{emcee: The MCMC Hammer}", journal = {\pasp}, archivePrefix = "arXiv", eprint = {1202.3665}, primaryClass = "astro-ph.IM", year = 2013, month = mar, volume = 125, pages = {306}, doi = {10.1086/670067}, adsurl = {http://adsabs.harvard.edu/abs/2013PASP..125..306F}, adsnote = {Provided by the SAO/NASA Astrophysics Data System}} @article{gala, doi = {10.21105/joss.00388}, url = {https://doi.org/10.21105%2Fjoss.00388}, year = 2017, month = {oct}, publisher = {The Open Journal}, volume = {2}, number = {18}, author = {}, title = {Gala: A Python package for galactic dynamics}, journal = {The Journal of Open Source Software} } @Article{ipython, Author = {, .}, Title = {{IP}ython: a System for Interactive Scientific Computing}, Journal = {Computing in Science and Engineering}, Volume = {9}, Number = {3}, Pages = {21--29}, month = may, year = 2007, url = "https://ipython.org", ISSN = "1521-9615", doi = {10.1109/MCSE.2007.53}, publisher = {IEEE Computer Society} } @INPROCEEDINGS{iraf, author = {{.}, title = "{The IRAF Data Reduction and Analysis System}", booktitle = {Instrumentation in astronomy VI}, year = 1986, series = {\procspie}, volume = 627, editor = {{.}, month = jan, pages = {733}, doi = {10.1117/12.968154}, adsurl = {http://adsabs.harvard.edu/abs/1986SPIE..627..733T}, adsnote = {Provided by the SAO/NASA Astrophysics Data System} } @ARTICLE{matplotlib, author = {{Hunter}, J.~D.}, title = "{Matplotlib: A 2D Graphics Environment}", journal = {Computing in Science and Engineering}, keywords = {Python, Scripting languages, Application development, Scientific programming }, year = 2007, month = may, volume = 9, pages = {90-95}, doi = {10.1109/MCSE.2007.55}, adsurl = {http://adsabs.harvard.edu/abs/2007CSE.....9...90H}, adsnote = {Provided by the SAO/NASA Astrophysics Data System} } @ARTICLE{numpy, author={ and and }, journal={Computing in Science Engineering}, title={The NumPy Array: A Structure for Efficient Numerical Computation}, year={2011}, volume={13}, number={2}, pages={22-30}, keywords={data structures;high level languages;mathematics computing;numerical analysis;Python programming language;high level language;numerical computation;numerical data;numpy array;Arrays;Computational efficiency;Finite element methods;Numerical analysis;Performance evaluation;Resource management;Vector quantization;NumPy;Python;numerical computations;programming libraries;scientific programming}, doi={10.1109/MCSE.2011.37}, ISSN={1521-9615}, month={March} } @InProceedings{pandas, author = { }, title = { Data Structures for Statistical Computing in Python }, booktitle = { Proceedings of the 9th Python in Science Conference }, pages = { 51 - 56 }, year = { 2010 }, editor = { and } } @article{schwimmbad, doi = {10.21105/joss.00357}, url = {https://doi.org/10.21105/joss.00357}, year = {2017}, month = {sep}, publisher = {The Open Journal}, volume = {2}, number = {17}, author = { and }, title = {schwimmbad: A uniform interface to parallel processing pools in Python}, journal = {The Journal of Open Source Software} } @Misc{scipy, author = { and and and others}, title = {{SciPy}: Open source scientific tools for {Python}}, year = {2001--}, url = "http://www.scipy.org/", note = {[Online; accessed ]} }\hypertarget{group__BAUDR}{}\doxysection{Baud Rate Select (B\+A\+U\+DR) Register} \label{group__BAUDR}\index{Baud Rate Select (BAUDR) Register@{Baud Rate Select (BAUDR) Register}} Baud Rate Select (B\+A\+U\+DR) Register. \doxysubsection*{Modules} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{group__TXFTLR}{Transmit F\+I\+F\+O Threshold Level (\+T\+X\+F\+T\+L\+R) Register}} \begin{DoxyCompactList}\small\item\em Transmit F\+I\+FO Threshold Level (T\+X\+F\+T\+LR) Register. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection*{Classes} \begin{DoxyCompactItemize} \item struct \mbox{\hyperlink{struct__ADI__SPIM__REGS__BAUDR__t}{\+\_\+\+A\+D\+I\+\_\+\+S\+P\+I\+M\+\_\+\+R\+E\+G\+S\+\_\+\+B\+A\+U\+D\+R\+\_\+t}} \item struct \mbox{\hyperlink{structADI__SPIM__REGS__BAUDR__t}{A\+D\+I\+\_\+\+S\+P\+I\+M\+\_\+\+R\+E\+G\+S\+\_\+\+B\+A\+U\+D\+R\+\_\+t}} \end{DoxyCompactItemize} \doxysubsection*{Typedefs} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{group__BAUDR_gaa38ecacfa017438d6c3ff0561b60467b}\label{group__BAUDR_gaa38ecacfa017438d6c3ff0561b60467b}} typedef struct \mbox{\hyperlink{struct__ADI__SPIM__REGS__BAUDR__t}{\+\_\+\+A\+D\+I\+\_\+\+S\+P\+I\+M\+\_\+\+R\+E\+G\+S\+\_\+\+B\+A\+U\+D\+R\+\_\+t}} {\bfseries A\+D\+I\+\_\+\+S\+P\+I\+M\+\_\+\+R\+E\+G\+S\+\_\+\+B\+A\+U\+D\+R\+\_\+t} \end{DoxyCompactItemize} \doxysubsection{Detailed Description} Baud Rate Select (B\+A\+U\+DR) Register. bertamiro/Selecting_GRMA heuristic algorithm to select GRM/A-heuristic-algorithm-to-select-GRM.tex % Template for PLoS % Version 3.5 March 2018 % % % % % % % % % % % % % % % % % % % % % % % % % -- IMPORTANT NOTE % % This template contains comments intended % to minimize problems and delays during our production % process. Please follow the template instructions % whenever possible. % % % % % % % % % % % % % % % % % % % % % % % % % % Once your paper is accepted for publication, % PLEASE REMOVE ALL TRACKED CHANGES in this file % and leave only the final text of your manuscript. % PLOS recommends the use of latexdiff to track changes during review, as this will help to maintain a clean tex file. % Visit https://www.ctan.org/pkg/latexdiff?lang=en for info or contact us at . % % % There are no restrictions on package use within the LaTeX files except that % no packages listed in the template may be deleted. % % Please do not include colors or graphics in the text. % % The manuscript LaTeX source should be contained within a single file (do not use \input, \externaldocument, or similar commands). % % % % % % % % % % % % % % % % % % % % % % % % % % -- FIGURES AND TABLES % % Please include tables/figure captions directly after the paragraph where they are first cited in the text. % % DO NOT INCLUDE GRAPHICS IN YOUR MANUSCRIPT % - Figures should be uploaded separately from your manuscript file. % - Figures generated using LaTeX should be extracted and removed from the PDF before submission. % - Figures containing multiple panels/subfigures must be combined into one image file before submission. % For figure citations, please use "Fig" instead of "Figure". % See http://journals.plos.org/plosone/s/figures for PLOS figure guidelines. % % Tables should be cell-based and may not contain: % - spacing/line breaks within cells to alter layout or alignment % - do not nest tabular environments (no tabular environments within tabular environments) % - no graphics or colored text (cell background color/shading OK) % See http://journals.plos.org/plosone/s/tables for table guidelines. % % For tables that exceed the width of the text column, use the adjustwidth environment as illustrated in the example table in text below. % % % % % % % % % % % % % % % % % % % % % % % % % % % -- EQUATIONS, MATH SYMBOLS, SUBSCRIPTS, AND SUPERSCRIPTS % % IMPORTANT % Below are a few tips to help format your equations and other special characters according to our specifications. For more tips to help reduce the possibility of formatting errors during conversion, please see our LaTeX guidelines at http://journals.plos.org/plosone/s/latex % % For inline equations, please be sure to include all portions of an equation in the math environment. % % Do not include text that is not math in the math environment. % % Please add line breaks to long display equations when possible in order to fit size of the column. % % For inline equations, please do not include punctuation (commas, etc) within the math environment unless this is part of the equation. % % When adding superscript or subscripts outside of brackets/braces, please group using {}. % % Do not use \cal for caligraphic font. Instead, use \mathcal{} % % % % % % % % % % % % % % % % % % % % % % % % % % % Please contact with any questions. % % % % % % % % % % % % % % % % % % % % % % % % % \documentclass[10pt,letterpaper]{article} \usepackage[top=0.85in,left=2.75in,footskip=0.75in]{geometry} % amsmath and amssymb packages, useful for mathematical formulas and symbols \usepackage{amsmath,amssymb} % Use adjustwidth environment to exceed column width (see example table in text) \usepackage{changepage} % Use Unicode characters when possible \usepackage[utf8x]{inputenc} % textcomp package and marvosym package for additional characters \usepackage{textcomp,marvosym} % cite package, to clean up citations in the main text. Do not remove. % \usepackage{cite} % Use nameref to cite supporting information files (see Supporting Information section for more info) \usepackage{nameref,hyperref} % line numbers \usepackage[right]{lineno} % ligatures disabled \usepackage{microtype} \DisableLigatures[f]{encoding = *, family = * } % color can be used to apply background shading to table cells only \usepackage[table]{xcolor} % array package and thick rules for tables \usepackage{array} % create "+" rule type for thick vertical lines \newcolumntype{+}{!{\vrule width 2pt}} % create \thickcline for thick horizontal lines of variable length \newlength\savedwidth \newcommand\thickcline[1]{% \noalign{\global\savedwidth\arrayrulewidth\global\arrayrulewidth 2pt}% \cline{#1}% \noalign{\vskip\arrayrulewidth}% \noalign{\global\arrayrulewidth\savedwidth}% } % \thickhline command for thick horizontal lines that span the table \newcommand\thickhline{\noalign{\global\savedwidth\arrayrulewidth\global\arrayrulewidth 2pt}% \hline \noalign{\global\arrayrulewidth\savedwidth}} % Remove comment for double spacing %\usepackage{setspace} %\doublespacing % Text layout \raggedright \setlength{\parindent}{0.5cm} \textwidth 5.25in \textheight 8.75in % Bold the 'Figure #' in the caption and separate it from the title/caption with a period % Captions will be left justified \usepackage[aboveskip=1pt,labelfont=bf,labelsep=period,justification=raggedright,singlelinecheck=off]{caption} \renewcommand{\figurename}{Fig} % Use the PLoS provided BiBTeX style % \bibliographystyle{plos2015} % Remove brackets from numbering in List of References \makeatletter \renewcommand{\@biblabel}[1]{\quad#1.} \makeatother % Header and Footer with logo \usepackage{lastpage,fancyhdr,graphicx} \usepackage{epstopdf} %\pagestyle{myheadings} \pagestyle{fancy} \fancyhf{} %\setlength{\headheight}{27.023pt} %\lhead{\includegraphics[width=2.0in]{PLOS-submission.eps}} \rfoot{\thepage/\pageref{LastPage}} \renewcommand{\headrulewidth}{0pt} \renewcommand{\footrule}{\hrule height 2pt \vspace{2mm}} \fancyheadoffset[L]{2.25in} \fancyfootoffset[L]{2.25in} \lfoot{\today} %% Include all macros below \newcommand{\lorem}{{\bf LOREM}} \newcommand{\ipsum}{{\bf IPSUM}} \usepackage{forarray} \usepackage{xstring} \newcommand{\getIndex}[2]{ \ForEach{,}{\IfEq{#1}{\thislevelitem}{\number\thislevelcount\ExitForEach}{}}{#2} } \setcounter{secnumdepth}{0} \newcommand{\getAff}[1]{ \getIndex{#1}{Universitat de Barcelona,International Rice Research Institute} } \providecommand{\tightlist}{% \setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}} \begin{document} \vspace*{0.2in} % Title must be 250 characters or less. \begin{flushleft} {\Large \textbf\newline{\emph{A heuristic algorithm to select genes potentially regulated by methylation}} % Please use "sentence case" for title and headings (capitalize only the first word in a title (or heading), the first word in a subtitle (or subheading), and any proper nouns). } \newline % Insert author names, affiliations and corresponding author email (do not include titles, positions, or degrees). \\ \textsuperscript{\getAff{Universitat de Barcelona, (UB), Spain}}\textsuperscript{*}, \textsuperscript{\getAff{International Rice Research Institute, (IRRI), Philippines}}\\ \bigskip \textbf{\getAff{Universitat de Barcelona}}Departament of Genetics Microbiology and Statistics, Avda Diagonal 645, Barcelona, 08028\\ \textbf{\getAff{International Rice Research Institute}}Department, Street, City, State, Zip\\ \bigskip * Corresponding author: \\ \end{flushleft} % Please keep the abstract below 300 words \section*{Abstract} Methylation is a key process in cancer. Usually it acts by inhibiting the expression of the gene but if methylation is low then any values of expression, high or low, can be found. This suggests that to select genes regulated by methylation one may look for patterns in the relation between gene expression and methylation showing either an L-shape or negative correlation between expression and methylation. We have developed a heuristic algorithm that mimics the process of visually selecting an ``L-shape'', that is genes that can show a wide range of expression values (low to high) when methylation is low, but only low expressions for intermediate or high methylation. We have compared the method with naïve correlation and, despite not being able to quantify its accuracy -because no dataset with ``TRUE'' L-shaped genes is available- its performance seems to be very good especially due to its flexibility. The method has been implemented in an R package, ``Lheuristic'' and a Shiny application, both available from GitHub (http://github.com/alexsanchezpla). Given two matrices -expression and methylation values - with the same row and column names the program offers the possibility to select genes based on either negative correlation, the heuristic algorithm or both methods at once. Once genes have been selected, results can be interactively reviewed, plotted or downloaded. % Please keep the Author Summary between 150 and 200 words % Use first person. PLOS ONE authors please skip this step. % Author Summary not valid for PLOS ONE submissions. \linenumbers % Use "Eq" instead of "Equation" for equation citations. \hypertarget{introduction-and-background}{% \section{Introduction and Background}\label{introduction-and-background}} \hypertarget{introduction-to-methylation}{% \subsection{Introduction to methylation}\label{introduction-to-methylation}} Epigenetic marks modulate gene expression without affecting the DNA nucleotide sequence. These potentially heritable changes are, for example, DNA methylation or histone acetylation ({[}1{]}). DNA methylation is the most studied epigenetic process in humans. The process is based on the addition of a methyl group, mostly in CpG dinucleotides. The CpG dinucleotides tend to group in areas of less than 500kb and with higher than 55\% C and G content , these regions are named islands; further from the island the region is called shore and further from the shore it is called shelf. More than 60\% of promoter regions are associated with CpG islands ({[}2{]}) and the methylation of these is linked to gene silencing and gene expression inhibition. DNA methylation has been linked to the regulation of numerous cellular processes, including embryonic development, or X-chromosome inactivation and preservation of chromosome stability among others. DNA methylation has also been observed in autoimmune diseases, metabolic disorders, neurological disorders, and other processes that despite being natural they are debilitating, like ageing for example; and it can also be correlated with drug or treatment response ({[}3{]}; {[}4{]}; {[}5{]}; {[}6{]}). Most research on this area has been, however, focused on tumor repressor genes, which are often silenced in cancer cells due to hypermethylation. This is an important mechanism of gene silencing during tumor progression ({[}7{]}). On the contrary, a general level of hypomethylation has been observed in human tumors ({[}8{]}); therefore, hypomethylation is a useful mechanism to distinguish genes of some human cancers from their normal counterparts.\\ In the human genome, about 80\% of cytosines in the 56 million CpG sites are methylated to 5-methylcytosines. The methylation pattern of DNA is highly variable among cells types and developmental stages and influenced by disease processes and genetic factors. The relationship between gene expression and methylation has been associated with cancer and extensively studied, therefore it has produced fruitful results ({[}9{]}). \hypertarget{analysis-of-genes-regulatated-by-methylation}{% \subsection{Analysis of genes regulatated by methylation}\label{analysis-of-genes-regulatated-by-methylation}} With the abundance of emerging evidence indicating the important role of DNA methylation in common diseases, researchers have attempted to use DNA methylation as a biomarker to identify epigenetic changes that are associated with disease status.~While the genetic events that drive the tumorigenic process are relatively well characterized for colorectal cancer, the epigenetic events and their impact on the transcriptional reprogramming observed in colorectal tumors have not been extensively characterized. Although recent genome-wide studies have analyzed the genomic distribution of hypermethylated CpGs in a small number of colorectal tumors (ref), a detailed analysis of the subset of these events that are important for gene expression regulation is currently lacking. Just as gene expression microarrays accelerated and revolutionized the study of transcriptional regulation, rapidly improving technologies are increasingly enabling researchers to assess locus-specific DNA methylation on a genome-wide scale. Recently various high-throughput approaches based on bisulfite conversion combined with next generation sequencing have been developed and applied for the genome wide analysis of DNA methylation. These methods provide single base pair resolution, quantitative DNA methylation data with genome wide coverage. There are various experimental types of methylation assays, but overall, methylation levels can be represented in one of three types: discrete, continuous or categorical. Therefore, methylation can be quantified by directly using read count information , ratio data (which may lose biological variability) or both. Once the DNA samples are processed, an important issue to be considered is the influence of the statistical analysis on the accuracy of the genomic methylation level estimation from bisulfite sequencing data. The accuracy of the statistical approach to methylation quantification increases with the sequencing depth of the particular cytosine residue ({[}10{]}). However, there are regression and neighboring analysis techniques that can counteract the lack of sequence depth in a particular CpG ({[}11{]}). \hypertarget{existing-methods-and-analyses}{% \subsubsection{Existing methods and analyses}\label{existing-methods-and-analyses}} The association between gene expression and DNA methylation in the CpG islands in particular has been long studied; and as a result, mostly negative correlations have been found to relate to cancer driven mechanisms (\textbf{???}), but this inverse relationship between DNA methylation of the first intron in particular and gene expression is a broad mechanism to down-regulate gene expression and it is found in numerous processes, organisms and tissues ({[}12{]}). There have been various studies analysing this correlation using various approaches. For example, Massie et al., (2017) looked at the relationship between gene expression and DNA methylation at the probe level rather than at the gene level. They narrowed a list of genes regulated by methylation that were identified in more than 3 out of 17 studies. Another study analysed the TCGA database to identify patterns in DNA CpG methylation and gene expression and tumor status. They found that the association involved a reduced number of genes linked to cancer than originally anticipated (around the hundreds) and that not all correlations were negative ({[}13{]}). Another recent paper reported two different models for analysis of DNA methylation and regulation of gene expression, one for negatively correlated genes and one for positively correlated genes (Klett et al., 2018). They used expression (GSE106582) and methylation datasets (GSE101764) containing 194 samples, 77 tumors and 117 of the mucose. By random forest analysis they were able to classify genes into cancer related and not related. Still methodologies to find tune classification into cancer/disease related and not cancer/disease related are still needed. A previously developed method was the selection of genes with an L-shape association between the expression and the methylation datasets (Sanchez-Pla et al., 2015). In this research, they focused on the CMI and on a method based on spline regression. They observed that the first method would detect L-shaped genes more accurately in big datasets. On the other hand, the splines clustering was not size dependent, but it would yield a smaller number of samples. Other research exists that aimed to identify genes regulated by methylation according to the expression methylation patterns; however, they only use a particular methodology like the CMI (\textbf{???}) with positive results. A paper focused on the identification of genes regulated by methylation through unsupervised clustering techniques to identify CRC subtypes was able to confirm existing subtypes ({[}14{]}).There has been other work that focused on the development of platforms for the identification of genes regulated by methylation. One of these packages is MEXPRESS (Koch et al., 2017). This package has a web interface that allows the user to visualize expression and methylation data from genes in the TCGA data. The visualization collocates for each selected gene, CpG islands, with transcripts expression together with other clinical values such as gender and age. The tool also generates p-values in relation to the variables specified. Another one of these packages is Methylmix (ref). The algorithm is based on a beta mixture model that identifies methylation states and compares them with what they call normal conditions to find hypo- and hyper-methylated genes. They developed a new statistic coeficient, the Differential Methylation value or DM-value which is defined as the difference of a methylation state with the normal methylation state. Then, they correlate that coefficient with gene expression data to characterize the association between methylation level and gene expression. For expression and methylation correlation analyses of both RNA and DNA molecules there is also a web based tool that analyses methylated genes based on TCGA data, called MethHC (http://methhc.mbc.nctu.edu.tw/php/search.php?opt=gene, Huang et al., 2015). This database has an analysis tool that provides gene-specific analysis for various diseases, and the information is displayed as a comparison between diseased and normal (non-diseased) conditions; list of highest and lowest methylated (hyper and hypo) genes; as well as correlations between expression and methylation. In this, methylation is a binary value (0,1) Other methodologies to identify methylated genes associated with cancer is through text mining analysis, as in the PubMeth database (www.pubmeth.org, {[}15{]}). In this, they identified 5000 genes of 1000 publications. However, high-throughput methodologies that offer an impartial approach to the identification of genes regulated by methylation still need further development and fine-tuning. Here we present such a methodology that will select, out of a gene expression and DNA methylation subsets, those genes that present a negative correlation, and are therefore regulated by methylation. The L-shaped heuristic method to identify genes regulated by methylation was tested and tuned for experimental expression and methylation paired datasets after normalization using other standard methods. \hypertarget{material-and-methods}{% \section{Material and Methods}\label{material-and-methods}} As we have described in the previous section, although there are various approaches to selecting genes based on the relationship between methylation levels and gene expression, none of them are completely satisfactory. In this section we present the method we have developed to select genes in which the pattern of the relationship is ``L-shaped.'' In fact, taking biological processes into account, this is a very common and very reasonably expected pattern when genes are regulated by methylation. Furthermore, as we will see later, it is not only important but can be partially missed by ``naïve'' methods such as significant negative correlation, which increases the interest of our proposal. \hypertarget{rationale-of-the-approach}{% \subsection{Rationale of the approach}\label{rationale-of-the-approach}} After trying different approaches to detect L-shapes, one often comes back to an intuïtive idea: If we are looking for genes whose expression can take any value when methylation is low, and tends to decrease as methylation increases one should observe that points in the methylation-expression scatterplot tend to be scattered near the vertical and horizontal positive axes showing an L-shape. If this does not happen genes can be found anywhere in the scatterplot and we can call it a non-L-shape. That is: \begin{itemize} \tightlist \item The more the points cluster near the vertical and horizontal axes, the more L-shaped can be considered the scatterplot. \item The more the points move away from the axes, the least L-shaped the scatterplot is. \end{itemize} Figure 1 illustrates these two possibilities in two real but non-identified genes. \hypertarget{an-algorithm-to-select-l-shape-scatterplots}{% \subsection{An algorithm to select L-shape scatterplots}\label{an-algorithm-to-select-l-shape-scatterplots}} The intuitive idea presented above can be made more explicit by introducing a ``three-band rule'' as follows: \begin{enumerate} \item Overimpose a $3\times 3$ grid on the scatterplot. \item Classify the scatterplot as \textbf{``L'' or ``non-L''} based on a small set of conditions: \begin{enumerate} \item There must be a \emph{minimum} number of points in the upper-left (cell (1,1)) and lower right (cell (3,3)) corners of the grid. \item There must be a \emph{maximum} number of points in the upper right (cell (1,3)) because points there mean hypermethylation and hyperexpression which is the opposite of what we are looking for. \item We will usually \emph{not require to have a minimum of points in cell (3,1)} unless we are really willing to have an L-shape (in our setting we will also be happy tho recover diagonals, which also reflect a negative correlation!). \end{enumerate} \item Score points on each subgrid in such a way that \begin{enumerate} \item Points in permitted regions (left-outer margin, i.e. cells: (1,1), (2,2), (3,1), (3,2), (3,3)) score positively if the scatterplot has been classified as L or zero if it has been classified as non-L. \item Points in non-desired regions (outer band. i.e. cells (1,2), (1,3), (2,3)) score negatively in all cases. \item Some regions may be declared neutral and not-score, such as cell (2,2). \end{enumerate} \item Use cross-validation to tune scoring parameters (\textit{if a set of positive and negative L-shaped genes is available}). \end{enumerate} The previous scheme can be summarized using the following equation. \begin{equation} S(X) = W_L \circ X \times \mathbf{1}_L(X) + W_{L^C} \circ X \times \mathbf{1}_{L^c}(X), \end{equation} where \begin{itemize} \item ${X}$ is the matrix of \emph{counts}, i.e. the number of counts in each cell of the grid, \item ${W_L}$ is the matrix of scores per cell and point \emph{if the scatterplot has been classified as $L$}, \item ${W_{L^c}}$ is the matrix of scores per cell and point \emph{if the scatterplot has been classified as non-$L$ ($L^c$)}, \end{itemize} and \(\circ\) represents the hadamard product of the two matrices \(W_{L/L^c}\) (i.e.~elementwise multiplication of the two matrices) and \(\mathbf{1}_{L/L^c}()\) is the indicator function for \(L\) or \(L^c\). The fact that the scatterplot is assigned to \(L\) or \(L^c\) can also be described as the hadamard product of three matrices: \begin{equation} \mathbf{1}_L(X) = \bigwedge_{i,j} X \circ C \circ \left( mMP \times \sum_{i,j}x_{ij}\right), \end{equation} where \begin{itemize} \item ${X}$ is the matrix of \emph{counts}, i.e. the number of counts in each cell of the grid, \item $C$ is the matrix of conditions to be verified \emph{if the scatterplot has to be classified as $L$}, \item $mMP$ is the matrix of minimum and Maximum Percentages of points to have in each cell \emph{if the scatterplot has to be classified as $L$}, \item $\circ$ represents the pointwise logical operation which allows that the product of the three cells becomes a logical operation and \item $\bigwedge_{i,j}$ represents an logical ``AND'' operation of all cells, that is if all cells are TRUE the result is assigned to $L$ and if one fails it is assigned to $L^c$. \end{itemize} \hypertarget{on-the-sensitivity-and-specificity-of-the-method}{% \subsection{On the sensitivity and specificity of the method}\label{on-the-sensitivity-and-specificity-of-the-method}} \hypertarget{tuning-the-algorithms-parameters}{% \subsubsection{Tuning the algorithm's parameters}\label{tuning-the-algorithms-parameters}} \hypertarget{synthetic-dataset-generation-for-the-simulation-studies}{% \subsection{Synthetic dataset generation for the simulation studies}\label{synthetic-dataset-generation-for-the-simulation-studies}} The R package \texttt{simstudy} was used to create 4 artificial datasets by using the splines method (https://cran.r-project.org/web/packages/simstudy/simstudy.pdf). The package allows for designing data points on a pre-defined spline, in which knots, limits and dispersion can be tuned. The splines are generated based on a fixed X variable representing the methylation values. The artificial datasets contained a total of 1000 genes, and the data points were developed based on 2 parameters with 2 levels each. The first parameter was the number of samples and the second the \% of true regulated by methylation genes that a sample would contain (with an expression by methylation scatterplot or spline following an L-shape). The number of samples considered was of 50 and 1000, and the \% of true methylated genes in each dataset was 1\% and 10\%. Additionally, the shape of the negative genes (not regulated by methylation) was also pre-defined and classified into 3 different scatterplot shapes (Figure XXX) and the percentage of genes in each category equaled to 1/3 prior subtraction of the true GRM genes. \hypertarget{simulation-of-heuristic-classification-with-the-synthetic-datasets}{% \subsection{Simulation of heuristic classification with the synthetic datasets}\label{simulation-of-heuristic-classification-with-the-synthetic-datasets}} The heuristic method was tested with the 4 artificial datasets: 50 samples, 1\% of GRM genes; 50 samples, 10\% GRM genes; 1000 samples, 1\% GRM genes; and 1000 samples, 10\% GRM genes. After running the model, sensitivity, specificity, and accuracy were measured and compared between datsets. \hypertarget{results}{% \section{Results}\label{results}} \hypertarget{measures-of-performance-for-the-heuristic-method-with-synthetic-datasets}{% \subsection{Measures of performance for the heuristic method with synthetic datasets}\label{measures-of-performance-for-the-heuristic-method-with-synthetic-datasets}} Sensitivity, specificity and accuracy for the heuristic model were measured for the 4 synthetic datsets (Figure {[}ssa{]}) with predefined parameters described in the above section. Specificity was the parameter that scored highest in all datsets, with values between 0.99 (for the datasets with 50 samples) and 1 (for the datasets with 1000 samples). The second highest parameter was accuracy. In this, both datasets containing 1\% of GRM genes scored 0.99, whereas the datasets with 10\% of GRM genes scored 0.93 (for the one with 50 samples) and 0.92 (for the one with 1000 samples). Finally, the sensitivity values were the lowest in the combination of 1\% GRM and 1000 samples (0.1) and highest for 1\% GRM and 50 samples (0.5). These results indicate that the classification scored better non-L shaped scatterplots (true negatives) than L-shaped scatterplots (true positives). \hypertarget{discussion}{% \section{Discussion}\label{discussion}} \hypertarget{supporting-information}{% \section{Supporting information}\label{supporting-information}} \hypertarget{references}{% \section*{References}\label{references}} \addcontentsline{toc}{section}{References} \clearpage \hypertarget{figures-and-captions}{% \section{Figures and captions}\label{figures-and-captions}} \begin{figure} \hypertarget{id}{% \centering \includegraphics[width=0.5\textwidth,height=0.5\textheight]{figures/nonLshapeVSLshape1.png} \caption{Example of non-Lshape vs L-shape for the methylation--expression scatterplots associated with two fictitial genes}\label{id} } \end{figure} \hypertarget{refs}{} \leavevmode\hypertarget{ref-Berger2009}{}% 1. , , , . An operational definition of epigenetics. Genes and Development. 2009;23: 781--783. doi:\href{https://doi.org/10.1101/gad.1787609}{10.1101/gad.1787609} \leavevmode\hypertarget{ref-Saxonov2006}{}% 2. , , . A genome-wide analysis of CpG dinucleotides in the human genome distinguishes two distinct classes of promoters. Proceedings of the National Academy of Sciences of the United States of America. 2006;103: 1412--1417. doi:\href{https://doi.org/10.1073/pnas.0510310103}{10.1073/pnas.0510310103} \leavevmode\hypertarget{ref-Reik2007}{}% 3. . Stability and flexibility of epigenetic gene regulation in mammalian development. Nature Publishing Group; 2007. pp. 425--432. doi:\href{https://doi.org/10.1038/nature05918}{10.1038/nature05918} \leavevmode\hypertarget{ref-Portela2010}{}% 4. , . Epigenetic modifications and human disease. 2010. pp. 1057--1068. doi:\href{https://doi.org/10.1038/nbt.1685}{10.1038/nbt.1685} \leavevmode\hypertarget{ref-Feil2012}{}% 5. , . Epigenetics and the environment: Emerging patterns and implications. 2012. pp. 97--109. doi:\href{https://doi.org/10.1038/nrg3142}{10.1038/nrg3142} \leavevmode\hypertarget{ref-Benayoun2015}{}% 6. , Pollina EA, . Epigenetic regulation of ageing: Linking environmental inputs to genomic stability. Nature Publishing Group; 2015. pp. 593--610. doi:\href{https://doi.org/10.1038/nrm4048}{10.1038/nrm4048} \leavevmode\hypertarget{ref-Jones2002}{}% 7. , . The fundamental role of epigenetic events in cancer. 2002. pp. 415--428. doi:\href{https://doi.org/10.1038/nrg816}{10.1038/nrg816} \leavevmode\hypertarget{ref-Feinberg1983}{}% 8. , . Hypomethylation distinguishes genes of some human cancers from their normal counterparts. Nature. 1983;301: 89--92. doi:\href{https://doi.org/10.1038/301089a0}{10.1038/301089a0} \leavevmode\hypertarget{ref-Yang2014}{}% 9. , , , , , . Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer Cell. Cell Press; 2014;26: 577--590. doi:\href{https://doi.org/10.1016/j.ccr.2014.07.028}{10.1016/j.ccr.2014.07.028} \leavevmode\hypertarget{ref-Zhang2010}{}% 10. , . The application of next generation sequencing in DNA methylation analysis. 2010. pp. 85--101. doi:\href{https://doi.org/10.3390/genes1010085}{10.3390/genes1010085} \leavevmode\hypertarget{ref-Wreczycka2017}{}% 11. , , , , , . Strategies for analyzing bisulfite sequencing data. Journal of Biotechnology. 2017;261: 105--115. doi:\href{https://doi.org/10.1016/j.jbiotec.2017.08.007}{10.1016/j.jbiotec.2017.08.007} \leavevmode\hypertarget{ref-Anastasiadi2018}{}% 12. , , . Consistent inverse correlation between DNA methylation of the first intron and gene expression across tissues and species. Epigenetics and Chromatin. BioMed Central Ltd. 2018;11. doi:\href{https://doi.org/10.1186/s13072-018-0205-1}{10.1186/s13072-018-0205-1} \leavevmode\hypertarget{ref-Long2017}{}% 13. Long MD, J, . The Genomic Impact of DNA CpG Methylation on Gene Expression; Relationships in Prostate Cancer. Biomolecules. Multidisciplinary Digital Publishing Institute (MDPI); 2017;7. doi:\href{https://doi.org/10.3390/biom7010015}{10.3390/biom7010015} \leavevmode\hypertarget{ref-Barat2015}{}% 14. , , , . Integrating Colon Cancer Microarray Data: Associating Locus-Specific Methylation Groups to Gene Expression-Based Classifications. Microarrays. MDPI AG; 2015;4: 630--646. doi:\href{https://doi.org/10.3390/microarrays4040630}{10.3390/microarrays4040630} \leavevmode\hypertarget{ref-Ongenaert2007}{}% 15. , , , , , . PubMeth: a cancer methylation database combining text-mining and expert annotation. Nucleic Acids Research. 2007;36: D842--D846. doi:\href{https://doi.org/10.1093/nar/gkm788}{10.1093/nar/gkm788} \nolinenumbers \end{document} Experiments/Normal Behavior/N2/Behavior N2.tex LINUX Test1 Log in as logon on RHEL: ip a ping 8.8.8.8 nano document.txt ls cat document.txt mkdir documents mv documents.txt documents/ cd documents/ touch text.txt Test2 Login as root on RHEL: df -h top ps -aux yum check-update yum update glibc yum search git yum search htop yum install git cd / ls -l \documentclass[a4paper,10pt]{article} \pdfoptionpdfminorversion=5 \usepackage[utf8]{inputenc} \usepackage{amsmath} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{multicol} \usepackage{xcolor,colortbl} \usepackage[yyyymmdd,hhmmss]{datetime} \usepackage{xfrac} \newcommand{\mc}[2]{\multicolumn{#1}{c}{#2}} \definecolor{Gray}{gray}{0.85} \newcolumntype{a}{>{\columncolor{Gray}}c} \newcolumntype{b}{>{\columncolor{white}}c} \newcolumntype{R}{>{\columncolor{lime}}c} \newcolumntype{B}{>{\columncolor{pink}}c} \usepackage{diagbox} \usepackage[numbers]{natbib} \usepackage[utf8]{inputenc} \usepackage{array,multirow,graphicx} \usepackage{graphics} \usepackage{color} \usepackage{ifpdf} \ifpdf \DeclareGraphicsRule{*}{mps}{*}{} \fi \usepackage[utf8]{inputenc} \hoffset = 0pt \voffset = 0pt \oddsidemargin = 0pt \headheight = 15pt \footskip = 30pt \topmargin = 0pt \marginparsep = 5pt \headsep =25pt \marginparwidth = 54pt \marginparpush = 7pt \textheight = 621pt %621/ \textwidth = 500pt % Default fixed font does not support bold face \DeclareFixedFont{\ttb}{T1}{txtt}{bx}{n}{12} % for bold \DeclareFixedFont{\ttm}{T1}{txtt}{m}{n}{12} % for normal \newcommand{\I}{\ensuremath{\mathbb{I}}} \newcommand{\R}{\ensuremath{\mathbb{R}}} \newcommand{\norm}[1]{\ensuremath{\left\|#1\right\|}} \newcommand{\seminorm}[1]{\ensuremath{\left|#1\right|}} \newcommand{\dual}[2]{\ensuremath{\langle#1, #2 \rangle}} \newcommand{\inner}[2]{\ensuremath{\left(#1, #2\right)}} \newcommand{\Inner}[2]{\ensuremath{\left(\left(#1, #2\right)\right)}} \newcommand{\deriv}[2]{\ensuremath{\frac{\mathrm{d}#1}{\mathrm{d}#2}}} \newcommand{\meas}[1]{\ensuremath{\,\mathrm{d}#1}} \newcommand{\Div}[1] {\ensuremath{\text{div}#1}} \newcommand{\Grad}[1]{\ensuremath{\text{grad}#1}} \newcommand{\Curl}[1]{\ensuremath{\text{curl}#1}} \newcommand{\jump}[1]{\ensuremath{[\![#1]\!]} } \newcommand{\citeneed}{\ensuremath{^\text{\textcolor{blue}{cite?}}}} \usepackage{mathtools} \newcommand{\eqdef}{\mathrel{\mathop=}:} \newcommand{\mm}{\ensuremath{\mathbf{m}}} \newcommand{\Vp}{\ensuremath{V_{\mathcal{P}}}} \newcommand{\Vb}{\ensuremath{V_{\mathcal{B}}}} \newcommand{\Ep}{\ensuremath{E_{\mathcal{P}}}} \newcommand{\Eb}{\ensuremath{E_{\mathcal{B}}}} %matrices \newcommand{\Ap}{\ensuremath{\mathbb{A}_{\mathcal{P}}}} \newcommand{\Ab}{\ensuremath{\mathbb{A}_{\mathcal{B}}}} \newcommand{\Bp}{\ensuremath{\mathbb{B}_{\mathcal{P}}}} \newcommand{\Bb}{\ensuremath{\mathbb{B}_{\mathcal{B}}}} \newcommand{\Amat}{\ensuremath{\mathbb{A}}} \newcommand{\Bmat}{\ensuremath{\mathbb{B}}} \newcommand{\Bmatt}{\ensuremath{\mathbb{B}^{\text{T}}}} \newcommand{\Mmat}{\ensuremath{\mathbb{M}}} \newcommand{\Nmat}{\ensuremath{\mathbb{N}}} %components \newcommand{\Apij}[2]{\ensuremath{\left(\mathbb{A}_{\mathcal{P}}\right)_{#1, #2}}} \newcommand{\Abij}[2]{\ensuremath{\left(\mathbb{A}_{\mathcal{B}}\right)_{#1, #2}}} \newcommand{\Bpij}[2]{\ensuremath{\left(\mathbb{B}_{\mathcal{P}}\right)_{#1, #2}}} \newcommand{\Bbij}[2]{\ensuremath{\left(\mathbb{B}_{\mathcal{B}}\right)_{#1, #2}}} %vectors \newcommand{\Uvec}{\ensuremath{\mathbf{U}}} \newcommand{\bvec}{\ensuremath{\mathbf{b}}} \newcommand{\Pvec}{\ensuremath{\mathbf{P}}} \newcommand{\up}{\ensuremath{u_{\mathcal{P}}}} \newcommand{\ub}{\ensuremath{u_{\mathcal{B}}}} \newcommand{\vp}{\ensuremath{v_{\mathcal{P}}}} \newcommand{\vb}{\ensuremath{v_{\mathcal{B}}}} \newcommand{\TODO}[1]{\textcolor{red}{#1}} \newcommand{\ASK}[1]{\textcolor{blue}{#1}} \newcommand{\CODE}[1]{\textcolor{orange}{Produced by {#1}}} \newcommand{\W}[1]{\ensuremath{w\!\left[#1\right]\!}} \newcommand{\E}[1]{\ensuremath{\epsilon \!\left[#1\right]\!}} \newcommand{\T}[1]{\ensuremath{\sigma \! \left[#1\right]\!}} \newcommand{\Tr}[1]{\ensuremath{\text{tr}#1}} \newcommand*{\defeq}{\mathrel{\vcenter{\baselineskip0.5ex \lineskiplimit0pt \hbox{\scriptsize.}\hbox{\scriptsize.}}}% =} \usepackage{lipsum} \usepackage{authblk} \usepackage{fancyhdr} \usepackage[colorlinks=true, linkcolor=cyan, citecolor=gray]{hyperref} \usepackage{chngcntr} \counterwithin{table}{subsection} \DeclareMathOperator{\spn}{span} % DEBUGGING \usepackage{lineno} \linenumbers %\usepackage{setspace} %\doublespacing % %\pagestyle{fancy} % \renewenvironment{abstract}{% \hfill\begin{minipage}{0.95\textwidth} \rule{\textwidth}{1pt}} {\par\noindent\rule{\textwidth}{1pt}\end{minipage}} % \makeatletter \renewcommand\@maketitle{% \hfill \begin{minipage}{0.95\textwidth} \vskip 2em \let\footnote\thanks {\LARGE \@title \par } \vskip 1.5em {\large \@author \par} \end{minipage} \vskip 1em \par } \makeatother % \begin{document} % %title and author details \title{\begin{center} Some theory behind LEGA\footnote{Last modified on \today@\currenttime} \end{center}} \author[1]{ } \maketitle \begin{abstract} Foo \end{abstract} bar \cite{shen_leg} \newpage \bibliographystyle{plain} \bibliography{lega_theory} \end{document} DataManagementLab/VLDBJ_1000_cores_measurements0 % This file was created by tikzplotlib v0.9.8. \definecolor{color0}{rgb}{0.945098039215686,0.345098039215686,0.329411764705882} \definecolor{color1}{rgb}{0.870588235294118,0.811764705882353,0.247058823529412} \definecolor{color2}{rgb}{0.980392156862745,0.643137254901961,0.227450980392157} \definecolor{color3}{rgb}{0.364705882352941,0.647058823529412,0.854901960784314} \definecolor{color4}{rgb}{0.698039215686274,0.462745098039216,0.698039215686274} \definecolor{color5}{rgb}{1,0,0.501960784313725} \begin{axis}[ axis line style={white!80!black}, grid=both, height=\figH, legend entries={DL DETECT,WAIT DIE,NO WAIT,MVCC,OCC,HSTORE,SILO,TICTOC}, legend style={legend columns=3,fill=none,draw=none,align=center}, legend to name={CommonLegend}, major tick length=0, scaled y ticks={base 10:-3}, tick align=outside, title style={yshift=-1.5ex}, width=\figW, x grid style={white!80!black}, xlabel shift=-1ex, xlabel={Number of Cores}, xmajorgrids, xmajorticks=false, xmajorticks=true, xmin=0, xmax=1578, xtick style={color=white!15!black}, xtick={0,200,400,600,800,1000,1200,1400,1600}, xticklabel style={align=left}, xticklabels={−200,0,200,400,600,800,1000,1200,1400}, y grid style={white!80!black}, ylabel shift=-1.4ex, ylabel={Million txn/s}, ymajorgrids, ymajorticks=false, ymajorticks=true, ymin=0, ymax=30732.8534548381, yminorticks=false, ytick scale label code/.code={}, ytick style={color=white!15!black}, ytick={0,5000,10000,15000,20000,25000,30000,35000}, yticklabels={−5000,0,5000,10000,15000,20000,25000,30000} ] \addplot [line width=0.48pt, color0, mark=*, mark size=2.4, mark options={solid,draw=white}] table {% 24 1781.04978450288 48 2681.9462661045 96 3613.39020497591 192 6647.43346723232 384 10502.3210011685 464 9755.52941467211 752 11375.1549866683 928 19661.0932689419 1120 22823.4714625405 1312 15710.6607479178 1504 15795.0936224739 }; \addplot [line width=0.48pt, color1, mark=square*, mark size=2.4, mark options={solid,draw=white}] table {% 24 1781.98209851295 48 2663.03307059217 96 3591.15866243736 192 6680.42761769607 384 10730.1472339415 464 9860.01646786728 752 11398.1745322224 928 22322.1777121807 1120 25493.8146947844 1312 16335.0687659481 1504 16726.9478461132 }; \addplot [line width=0.48pt, color2, mark=diamond*, mark size=2.4, mark options={solid,draw=white}] table {% 24 1797.16888455049 48 2707.65792045797 96 3666.81793481604 192 6793.95573299115 384 10732.3431223697 464 9941.11985147709 752 11449.6626953694 928 22455.4025802365 1120 25911.5123502317 1312 16376.6890498494 1504 16846.2303779584 }; \addplot [line width=0.48pt, color3, mark=pentagon*, mark size=2.4, mark options={solid,draw=white}] table {% 24 1397.12236344995 48 1984.86751417098 96 2881.98261478472 192 5111.04334071862 752 9972.61128412226 928 14516.8999336468 1312 13136.8028346289 1504 14168.3423218242 }; \addplot [line width=0.48pt, color4, mark=*, mark size=2.4, mark options={solid,draw=white}] table {% 24 1301.74414378678 48 1827.62849534708 96 2408.57508308424 192 4665.6655408815 384 8303.36268794654 464 8208.41429532378 752 10367.5178614057 928 16876.7685688631 1120 19481.0186865792 1312 14938.40104978 1504 15445.4197359547 }; \addplot [line width=0.48pt, white!30.1960784313725!black, mark=square*, mark size=2.4, mark options={solid,draw=white}] table {% 24 4194.70548938545 48 6489.4113404977 96 9707.42793275052 192 13447.1196696532 384 17034.5770782302 464 12561.6825216724 752 13541.4981996518 928 27787.6958873127 1120 29331.3720590738 1312 18155.5385297634 1504 17521.9089010426 }; \addplot [line width=0.48pt, color5, mark=pentagon*, mark size=2.4, mark options={solid,draw=white}] table {% 24 2273.04932062952 48 3116.85104282834 96 4280.71777669968 192 6370.00166995045 384 9925.0012693464 464 9488.50198955631 752 11375.5611619163 928 19475.8365270376 1120 21359.8020701045 1312 14992.0143430306 1504 16308.222541922 }; \addplot [line width=0.48pt, white!63.1372549019608!black, mark=triangle*, mark size=2.4, mark options={solid,draw=white}] table {% 24 2360.47658030149 48 3333.64502213294 96 4733.93254511026 192 7809.29217311004 384 11890.4548459794 464 10614.9730759968 752 12275.477277197 928 23965.9125862383 1120 26390.3848948082 1312 17028.0977938072 1504 17789.501856234 }; \end{axis} version https://git-lfs.github.com/spec/v1 oid sha256:080d6f0dafd1f0aff28576f14c5c48e0f63d93462a0a8711b4d27cb7e6be692c size 1018 1-10 @article{DBLP:journals/tog/ChenLDSM13, author = { and and and and }, title = {Spec2Fab: a reducer-tuner model for translating specifications to 3D prints}, journal = {ACM Trans. Graph.}, volume = {32}, number = {4}, year = {2013}, pages = {135}, ee = {http://doi.acm.org/10.1145/2461912.2461994}, bibsource = {DBLP, http://dblp.uni-trier.de} } 1-10 d_1, d_2, \ldots, d_s @article{Biroli2021, abstract = {In this paper we study the out-of-equilibrium phase diagram of the quantum version of Derrida's random energy model, which is the simplest model of mean-field spin glasses. We interpret its corresponding quantum dynamics in Fock space as a one-particle problem in very high dimension to which we apply different theoretical methods tailored for high-dimensional lattices: the forward-scattering approximation, a mapping to the Rosenzweig-Porter model, and the cavity method. Our results indicate the existence of two transition lines and three distinct dynamical phases: a completely many-body localized phase at low energy, a fully ergodic phase at high energy, and a multifractal "bad metal"phase at intermediate energy. In the latter, eigenfunctions occupy a diverging volume yet an exponentially vanishing fraction of the total Hilbert space. We discuss the limitations of our approximations and the relationship with previous studies.}, author = { and and and and }, day = {29}, doi = {10.1103/PhysRevB.103.014204}, issn = {1098-0121}, journal = {Physical Review B (Condensed Matter and Materials Physics)}, language = {English}, month = {January}, note = {Funding Information: We warmly thank for helpful discussions. This work was partially supported by the Simons Foundation through Grant No. 454935 (G.B.). D.F. was partially supported by the EPSRC (CANES, Grant No. EP/L015854/1) and the European Research Council under the European Union's Horizon 2020 research and innovation program (Grant Agreement No. 723955, GlassUniversality). Publisher Copyright: © 2021 American Physical Society. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.}, number = {1}, publisher = {American Physical Society}, title = {Out-of-equilibrium phase diagram of the quantum random energy model}, volume = {103}, year = {2021} } \documentclass{aastex61} %Packages \usepackage{graphicx} \usepackage{soul, listings, xcolor} \usepackage{amsmath} \usepackage{url} %Code environment details \definecolor{dkgreen}{rgb}{0,0.6,0} \definecolor{gray}{rgb}{0.5,0.5,0.5} \definecolor{mauve}{rgb}{0.58,0,0.82} \definecolor{grey}{gray}{0.95} \lstset{frame=tb, escapeinside={(*@}{@*)}, frame = single, aboveskip=3mm, belowskip=3mm, showstringspaces=false, columns=flexible, basicstyle={\small\ttfamily}, numbers=none, numberstyle=\tiny\color{gray}, keywordstyle=\color{blue}, commentstyle=\color{dkgreen}, stringstyle=\color{mauve}, breaklines=true, breakatwhitespace=true, } %% The default is a single spaced, 10 point font, single spaced article. %% There are 5 other style options available via an optional argument. They %% can be envoked like this: %% %% \documentclass[argument]{aastex61} %% %% where the arguement options are: %% %% twocolumn : two text columns, 10 point font, single spaced article. %% This is the most compact and represent the final published %% derived PDF copy of the accepted manuscript from the publisher %% manuscript : one text column, 12 point font, double spaced article. %% preprint : one text column, 12 point font, single spaced article. %% preprint2 : two text columns, 12 point font, single spaced article. %% modern : a stylish, single text column, 12 point font, article with %% wider left and right margins. This uses the Daniel %% Foreman-Mackey and design. %% %% Note that you can submit to the AAS Journals in any of these 6 styles. %% %% There are other optional arguments one can envoke to allow other stylistic %% actions. The available options are: %% %% astrosymb : Loads Astrosymb font and define \astrocommands. %% tighten : Makes baselineskip slightly smaller, only works with %% the twocolumn substyle. %% times : uses times font instead of the default %% linenumbers : turn on lineno package. %% trackchanges : required to see the revision mark up and print its output %% longauthor : Do not use the more compressed footnote style (default) for %% the author/collaboration/affiliations. Instead print all %% affiliation information after each name. Creates a much %% long author list but may be desirable for short author papers %% %% these can be used in any combination, e.g. %% %% \documentclass[twocolumn,linenumbers,trackchanges]{aastex61} %% AASTeX v6.* now includes \hyperref support. While we have built in specific %% defaults into the classfile you can manually override them with the %% \hypersetup command. For example, %% %%\hypersetup{linkcolor=red,citecolor=green,filecolor=cyan,urlcolor=magenta} %% %% will change the color of the internal links to red, the links to the %% bibliography to green, the file links to cyan, and the external links to %% magenta. Additional information on \hyperref options can be found here: %% https://www.tug.org/applications/hyperref/manual.html#x1-40003 %% This is the end of the preamble. Indicate the beginning of the %% manuscript itself with \begin{document}. \begin{document} \title{Lab 1: Exoplanet Transit \\ AST 443: Observational Techniques in Astronomy} \date{Performed October 7-8, 2016 and Submitted \today} \author{} \author{} \begin{abstract} Our goal for this experiment was to successfully detect a hot Jupiter-type exoplanet, and from this to see whether it has the known mid-transit time and transit duration. Also, from the depth of the transit we wanted to get the planet to stellar radius ratio, which would then give us the planet radius. In our experiment we choose to view the star HD 209458 while its planet HD 209458b, also known as Osiris, was transiting. We used a 14-inch Meade LX200-ACF telescope at Mt. Stony Brook Observatory to locate HD 209458, and a SBIG STL-1001E CCD camera to measure the flux received before, during and after the transit. Using various Python scripts shown in the Appendix and other astronomy programs, we were able to reduce the data and get a light curve of HD 209458. From this light curve, we deduced that ratio of the radii of the HD 209458 and Osiris is 0.138 $\pm$ 0.005, with the radius of Osiris being 1.541 $\pm$ 0.087 $R_{\text{Jup}}$. We also measured the transit duration of Osiris to be 3.0804 $\pm$ 0.0041 hours and the mid-transit time to be on October 8$^{\text{th}}$, 2016 at 03:09:27 $\pm$ 00:00:12 UTC. All of our experimental results agreed with theory within their respective error, except for the transit duration, and the mid-transit time. Although, this was mainly due to not taking into account systematic error. \end{abstract} \section{Introduction (JM)} \label{sec:intro} It has only been within the past decade and a half that we have been able to detect planets around other stars. This has been a major breakthrough in all fields of astronomy because it allows us to see what kind of planets form around different types of stars. Also, it allows us to speculate more on whether we are alone in the universe or not. The most recently discovered method used to detect exoplanets is through transit photometry. This method allows us to point our telescope to a star and see if theres a change in observed flux. Depending on the duration of the observed flux from the star we can determine if there is a transiting body. This is the method we used in our experiment and it has proven very successful over the years for hundreds of other astronomers. Although it has already been confirmed that there is an exoplanet around HD 209458, our experiment was significant because it allowed us to see first hand how powerful the method of transit photometry is. By just knowing the dip in relative flux, we can determine the ratio between the radii of the planet and star. Then from knowing the radius of the host star, we can determine the radius of the exoplanet. This is the reason why this method is so powerful: it allows us to learn the properties of planets hundreds of light years away without even viewing or seeing the planet. \section{Observations (JM)} We decided to observe HD 209458 because of the unique characteristics of the system. The star is of the G type, making it very similar to our own star, having a mass of 1.148 $\pm$ 0.022 $M_\sun$ \citep{2010MNRAS.408.1689S} and a radius of 1.146 $\pm$ 0.050 $R_\sun$ \citep{2001ApJ...552..699B}. The planet has a radius 1.451 times that of Jupiter with an uncertainty of 0.074, while orbiting at nearly one eighth the orbital radius of Mercury to the Sun \citep{2015MNRAS.447..846B}. This makes Osiris a hot Jupiter type planet because its the size of Jupiter but extremely close to it's star. This type of system is a great candidate for exoplanet detection through transit photometry. This is because the star is of average size, while having a huge Jupiter size planet orbiting it. This makes detecting a dip in the relative flux when the planet transits a lot easier. The only negative about this system is the orbital radius of Osiris, because the closer a planet is to its star the harder it will be to detect it. Although, it makes up for it by the size of Osiris and the relatively bright star of 7.65 apparent magnitude. We were able to observe HD 209458 and its transiting planet by using a 14-inch Meade LX200-ACF telescope with a SBIG STL-1001E CCD camera attached. This yields a field of view of 26 arcminutes for the exposures. All observations were held at the Mt. Stony Brook Observatory. The weather that night was mostly clear with some thin, high-altitude clouds in the beginning of observations. By the end of the experiment the sky was largely clear of clouds. The seeing was mostly good throughout the night as well, with there being very little wind. We calculated the total transit duration to be 3.024288 hours, with a mid-transit time of October $8^{\text{th}}$, 2016 at 03:11:37 UTC. The transit duration was calculated using the following equation: \begin{equation} T_{\text{duration}} = \frac{P}{\pi}\arcsin\left(\frac{\sqrt{(R_*+R_P)^{2}-(bR_*)^{2}}}{a}\right) \end{equation} Where $P$, $a$ and $b$ are all the period, semi-major axis and impact parameter of the planet respectively, and $R_*$ and $R_P$ are the radii of the star and planet respectively. The impact parameter is given by: \begin{equation} b = \frac{a \cos(i)}{R_*} \end{equation} Where $i$ is the inclination angle of the planets orbit. In this case, the semi-major axis, inclination angle and period of HD 209458 b are 0.04747 AU, 86.59 degrees and 3.52472 days respectively. All these values are from the The Extrasolar Planets Encyclopedia \footnote{\url{http://exoplanet.eu/catalog/hd_209458_b/}}. The mid-transit time was calculated from knowing the period of the planet, and knowing a specific time when the transit began in the past. From this information we simply just have to keep adding the period time to the time of transit until we reach current day. We figured out where our star should be in the night sky by using the program StarAlt. It helps show whether stars are viewable depending on your location on Earth and your elevation. We first tried to find the star by going to its right accession and declination, and made sure that the star we found was HD 209458 by looking at our finder charts. Once we confirmed the star we made sure it was centered on the cross hairs of the eyepiece. We then made sure the CCD camera was connected correctly to the telescope and the program CCDSoft, which is used for adjusting the exposure times and other settings of the CCD. We first tried to take four exposures of our star with an exposure time (exp. time) of three seconds each. This resulted in counts that were too high because the star was focused too well on the CCDSoft. High counts are bad because the pixels on the CCD can be oversaturated at around 60,000 counts and not yield good data. So for the next exposure we defocused the star, while increasing the exposure time to five seconds, which resulted in lower counts around 20,000. We kept this exposure time for the rest of the science exposures. By the end of the experiment we took 1,461 science exposures, all with an exposure time of five seconds. After this we started to take our calibration data, starting with the darks. In order to take the darks we put the telescope cap back on so we have complete darkness. We then took 19 dark exposures each of five seconds, then we took out flats. We took the cap off and pointed the telescope towards a equally lit part of the dome. At first we took 19 flats each of five seconds, but it turned out the counts were too high. So we lowered the exposure time to three seconds and lowered the lights in the observatory. This yielded in lower counts and better flat fields. Finally we took another set of flats with a shorter exposure time of one second. This is so later we can create a bad pixel mask so we know where the bad pixels are on the CCD. Below is a table of the logsheet we used during observations. The time written down is in EST, and was taken from the computer we were using. \begin{deluxetable*}{ccCrlc}[h!] \tablecaption{Logsheet of Observations \label{tab:logsheet}} \tablecolumns{13} \tablenum{1} \tablewidth{0pt} \tablehead{ \colhead{File Number} & \colhead{Object} & \colhead{Exp. Time} & \colhead{Time} & \colhead{Comments} \\ \colhead{} & \colhead{} & \colhead{(s)} & \colhead{(EST)} & \colhead{} } \startdata science.0-4 & transit & 3 & 8:41 pm & Focused too well \\ science.5 & $\downarrow$ & 5 & 8:42 pm & Max. count of 23,000 \\ science.6 & $\downarrow$ & $\downarrow$ & 8:43 pm & First of 800 \\ science.7-75 & $\downarrow$ & $\downarrow$ & 8:55 pm & \nodata \\ science.76-797 & $\downarrow$ & $\downarrow$ & 10:55 pm & Last of 800 set \\ science.798 & $\downarrow$ & $\downarrow$ & 10:55 pm & First of next 800 \\ science.1087-1088 & $\downarrow$ & $\downarrow$ & 11:42 pm & Long streak of light appeared \\ science.1461 & transit & $\downarrow$ & 12:43 am & Last transit exposure \\ darks.0-19 & darks & $\downarrow$ & 12:58 am & 19 darks taken \\ flats.0-19 & flats & 5 & 1:05 am & Exposures was too long \\ flats.20-29 & $\downarrow$ & 3 & 1:07 am & Lowered lights and exposure, counts at 18,000 \\ flats.30-39 & $\downarrow$ & 1 & 1:08 am & Shorter exposure, counts at 6,000 \\ \enddata \end{deluxetable*} \section{Data Reduction (JM and YM)} The data we obtained from all the exposures were in the form of raw .FIT files. So in order to make the science .FIT files usable we needed to create a Master Flat field, a Master Dark field and a bad pixel mask from all the calibration data. The master dark image was used to correct for the dark current and the master flat field was used to scale each pixel by its sensitivity relative to the other pixels on the CCD. The bad pixel mask found any pixels with a response that was nonlinear in time. If the pixels are linear, the ratio of the counts from the long exposure flats to those of the short exposures should equal the ratio of the time of the long exposure to that of the short. Any pixel that deviated from this was deemed "bad." The ratio of the long exposure to the short was found to be 3. This was all done using the Python code from appendix \ref{code: reduction}, which created clean science images. We found two bad pixels. Ordinarily, we would have taken an average of the surrounding pixels to extrapolate what the each bad pixel should have given us. However, since these pixels were in areas of the field that were never used, we simply ignored any information from them. All of the data reduction was done through the uhura astronomy computer using a secure shell client. When we performed the experiment, the auto-guider on the telescope did not work, so we had to use the astronometry.net software to get the position and coordinates of all the stars in our .FIT files. We wrote a script to let the software run through all of our images. Next, we used the program Source Extractor to return the properties of all the stars in our images, most importantly the relative flux. However, we needed to know the aperture size Source Extractor should use to get the data from the stars. This could be found out by opening up one of our clean images in ds9 and seeing what the radius in pixels is of our target star. Although this aperture is larger than what we would like to maximize the signal to noise ratio, the larger size reduces the effects of seeing. Once we found this radius, we inputed it into the Source Extractor configuration file and obtained the properties of every object in all of our images. We once again wrote a script that runs all our images through Source Extractor. All of the code for this part can be found in appendix \ref{code: astro sex}. We finally now had the three properties we need to get our analyzed data: the time of each image, flux received from all the stars in one image, and the error in that flux. Next, we chose ten reference stars in our clean images and took their respective times, flux, and the error in the flux from the table. We did this because it allows us to calibrate the flux received from the target star, correcting for changes in the atmosphere, seeing, and airmass. All of these stars have similar brightness when we viewed them in ds9. We used the code from appendix \ref{code: flux} to find our stars in each image and get their fluxes and errors in each exposure. Unfortunately, our code would sometimes match to the wrong star in an exposure, resulting in an outlier in our lightcurves. This would not happen for any particular exposure or star. Although it was applied randomly, it was consistent. It would mismatch a particular star for a particular image the same way every time it ran. This might indicate a problem with the code's logic, but the fact that it would not affect every star in the same place or with the same frequency indicates otherwise. There were no syntactic errors found. We even made sure all of the indents were spaces instead of a mix of spaces and tabs. Another possibility was that astronometry.net was not solving our fields properly. To address this, we opened up an image with an outlier for a particular star and 3 images that were matched correctly for the same star. We locked their WCS coordinates and had them disappear and reappear on top of each other. This would let us see if the whole image would shift because astronometry gave it a wrong WCS coordinate. There was no observed shift. Unable to find the cause, we set the flux for the stars to zero for every exposure in which the star was mismatched. From the fluxes of each reference star, we rescaled them by the star's average flux over all exposures taken. This was done also to the error in the flux, shown in \eqref{eq: norm}, where $f_{j}(t)$ is the flux for some reference $j$ at time $t$ and $\sigma_{f_{j}(t)}$ is the error in that particular flux. \begin{equation} \label{eq: norm} f_{j}(t) \rightarrow \frac{f_{j}(t)}{\langle f_{j} \rangle} \;\;\;\;\;\;\;\sigma_{f_{j}(t)} \rightarrow \frac{\sigma_{f_{j}(t)}}{\langle f_{j} \rangle} \end{equation} We normalized the flux and its error because it would allow us to visualize any environmental changes that occurred throughout the night, as well as the overall shape of the lightcurves for each star. Once we had this, we plotted each of the stars' flux as a function of time. The time is the number of seconds after midnight on October $8^{\text{th}}$ in UTC time. Although the averages did take into account the non-mismatched outliers of each star, they were so few in number that the average remained unperturbed. These are shown in Fig. \ref{fig: refcurve1} - \ref{fig: scicurve1}. This was done with the code from appendix \ref{code: curves}. %Light Curves \begin{figure}[hbt!] \centering \includegraphics[scale = .45]{exo_curves1.pdf} \caption{Reference star 1. This variable star was removed for the duration of the experiment} \label{fig: refcurve1} \end{figure} \begin{figure}[hbt!] \centering \includegraphics[scale = .45]{exo_curves2.pdf} \caption{Reference star 2} \label{fig: refcurve2} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves3.pdf} \caption{Reference star 3} \label{fig: refcurve3} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves4.pdf} \caption{Reference star 4} \label{fig: refcurve4} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves5.pdf} \caption{Reference star 5} \label{fig: refcurve5} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves6.pdf} \caption{Reference star 6} \label{fig: refcurve6} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves7.pdf} \caption{Reference star 7} \label{fig: refcurve7} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves8.pdf} \caption{Reference star 8} \label{fig: refcurve8} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves9.pdf} \caption{Reference star 9. The outliers rendered this star useless.} \label{fig: refcurve9} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves10.pdf} \caption{Reference star 10. This star was scrapped because of its outliers and large spread} \label{fig: refcurve10} \end{figure} \begin{figure}[h] \centering \includegraphics[scale = .45]{exo_curves11.pdf} \caption{Normalized light curve of science target} \label{fig: scicurve1} \end{figure} From these curves we can see vast changes in flux caused by the atmospheric conditions early that night. Due to this much variability, these images were not factored into the normalization of the plots. The normalization also ignored the fluxes at zero, which correspond to the mismatched outliers. The remaining outliers were removed via sigma clipping. Any value more than 4 sigma off from the average was removed for the duration of the experiment. Based on the lightcurves, we had to remove some reference stars entirely. The first of these, shown in Fig. \ref{fig: refcurve1}, is a variable star. Its intrinsic flux variations would throw off our calibration, making it seem as though there were atmospheric changes where there were none. Another reject is the ninth reference star, seen in Fig. \ref{fig: refcurve9}. This star appears to have been mismatched more often than not. Combining that with the large spread on the correctly matched fluxes and it becomes unusable. The final star we had to scrap was the tenth reference star, found in Fig. \ref{fig: refcurve10}. This was because of the unusually large spread and high number of correctly matched outliers. Now that we have our reference stars, we can account for environmental influences on the flux from the target star by calculating it relative to the flux of the reference stars. In order to get all the flux from all the reference stars we need to calculate the weighted mean in each clean image. This can be found by using \eqref{eq: average}, where $\mu_{i}^{ref}$ is the weighted mean for any image $i$. This average also has an error, given by \eqref{eq: error}. \begin{equation} \label{eq: average} \mu_{i}^{ref} = \frac{\sum_{j}\left( \frac{f_{j}^{ref}}{\left( \sigma_{j}^{ref}\right) ^{2}}\right) }{\sum_{j}\left( \frac{1}{\left( \sigma_{j}^{ref}\right) ^{2}}\right) } \end{equation} \begin{equation} \label{eq: error} \sigma_{i}^{ref} = \sqrt{\frac{1}{\sum_{j}\left( \frac{1}{\left( \sigma_{j}^{ref}\right) ^{2}}\right)}} \end{equation} Now that we have the weighted mean of the reference stars, we can get rid of the atmospheric changes in the science images by getting the ratio: $r_{i} = \frac{f_{i}^{sci}}{\mu_{i}^{ref}}$. The error in this ratio is given by Eq. \eqref{error: ratio}. \begin{equation}\label{error: ratio} \sigma_{r_{i}} = r_{i} \sqrt{\left( \frac{\sigma_{f_{i}^{sci}}}{f_{i}^{sci}}\right) ^{2} + \left( \frac{\sigma_{i}^{ref}}{\mu_{i}^{ref}}\right) ^{2}} \end{equation} This as all done by the code in appendix \ref{code: bigtable}. We now have all the information needed: the time, science flux, weighted mean and the ratio of the two latter. The last step involves normalizing the ratio by the baseline flux of our target star (the flux when the planet is not transiting). The baseline flux was calculated by looking at the clean images taken before transiting occurred, and taking the average flux of those images. Since it was so noisy, we only used the exposures from 4000-6000 seconds. The new ratio and its error will then be given by Eq. \eqref{eq: baseline}, where $f_{\text{blf}}$ is the baseline flux for a given star. \begin{equation}\label{eq: baseline} r_{i} \rightarrow \frac{r_{i}}{f_{\text{blf}}} \;\;\;\;\;\;\;\;\;\;\; \sigma_{ f_{\text{blf}}} \rightarrow f_{\text{blf}} \sqrt{\left( \frac{\sigma_{ f_{\text{blf}}}}{f_{\text{blf}}}\right) ^{2} + \left( \frac{\sigma_{ r_{i}}}{r_{i}}\right) ^{2}} \end{equation} We then binned this normalized data into intervals of 12 images. Since each exposure was 5 seconds, and there was about 4-5 seconds of processing time between exposures, this corresponds to about 2 minutes. This lightcurve is shown in Fig. \ref{fig: lightcurve}. %Final Lightcurve \begin{figure}[hbt!] \centering \includegraphics[scale = .75]{exo_lightcurve.pdf} \caption{Exoplanet transit. The lightcurve was binned in intervals of about 2 minutes} \label{fig: lightcurve} \end{figure} \section{Data Analysis and Results (JM and YM)} We finally reduced all our data to produce a finished light curve of HD 209458 b transiting its host star. Now our goal is to determine the radius of the transiting exoplanet from the planet to stellar radius ratio. This can be done by first noticing the change in flux based on the transit depth. When the planet is not transiting we should expect to see that the normalized ratio $r_{i}$ should be around 1.0 since we normalized it by the baseline flux. When the transit is occurring we should expect the ratio to be 1.0 - $\epsilon$ , where $\epsilon$ is lowest flux received during the transit. We see that the ratio between the flux during the transit to the flux before is equal to $1-\epsilon$, since the flux before is normalized to 1. We call this quantity $\Delta f$, or the total transit depth. From this, we can get the ratio of the radii between the planet and star. Since we are assuming we are looking at the transit head on, we can assume that both objects are flat disks of area $\pi r^{2}$, where $r$ is the radius of the object. Setting it all up, we have the equation: \begin{equation} \Delta f = \frac{\pi R_{P}^{2}}{\pi R_{*}^{2}} \end{equation} Solving for the planet radius we get: \begin{equation} R_{P} = \sqrt{\Delta f \; R_{*}^{2}} \end{equation} The error in the radius is given by: \begin{equation} \sigma_{R_{P}} = \frac{R_{P}}{2}\sqrt{\left( \frac{\sigma_{ \Delta f}}{\Delta f}\right)^{2} + 4\left( \frac{\sigma_{ R_{*}}}{R_{*}}\right)^{2} } \end{equation} Now that we have all the equations set up, we can plug in our values. Looking at the light curve, it can be seen that near the middle of the transit the fluxes get very messy. In order to get a value for the lowest flux in the transit, we took the median of the values from 8,000 to 14,000 seconds. We chose the median over the average as the former is more insensitive to outliers. The error in this is the median absolute deviation, as shown in Eq. \eqref{error: mad}. Here, $f_i$ is the flux at a specific point. \begin{equation} \label{error: mad} \sigma_{\Delta f} = \text{median}(\lvert \Delta f - f_i \rvert) \end{equation} This resulted in a lower flux of 0.9809 $\pm$ 0.00136. This yields a transit depth of 0.0190875 $\pm$ 0.0013581. As stated in the Observations section, we have a stellar radius of 1.146 $\pm$ 0.050 $R_{\sun}$. Plugging the values into equations (8) and (9) and converting to $R_{\text{Jup}}$, we get a planet radius of 1.541 $\pm$ 0.087 $R_{\text{Jup}}$, and a planet to stellar radius ratio of 0.138 $\pm$ 0.005. Our next goal is to get the transit duration and the mid transit time from the light curve. We first looked at the values of each point on the graph and noticed when the flux went below the baseline in the beginning of the transit, and when it went back to the baseline near the end of the transit. Doing this we get a time of 5,822.814 s $\pm$ 10.000 s for the beginning of the transit, and 16,912.108 $\pm$ 11.000 s at the end of the transit. Now simply by subtracting the end time by the beginning, we will get the transit duration: \begin{equation} T_{\text{duration}} = T_{\text{end}} - T_{\text{beginning}} \end{equation} Where the error in the duration is: \begin{equation} \sigma_{T_{\text{duration}}} = \sqrt{\sigma_{T_{\text{end}}}^{2}+\sigma_{T_{\text{beginning}}}^{2}} \end{equation} Plugging in the values, we get a transit duration of 11,089.294 $\pm$ 14.866 s. Converting to hours, we get a duration of 3.0804 $\pm$ 0.0041 hours. Finally, the mid transit time can be found by simply adding half the transit duration to when the transit begins: \begin{equation} T_{\text{mid}} = T_{\text{beginning}} +\frac{T_{\text{duration}}}{2} \end{equation} The error in this value would be: \begin{equation} \sigma_{T_{\text{mid}}} = \sqrt{\sigma_{T_{\text{beginning}}}^{2}+0.25 \; \sigma_{T_{\text{duration}}}^{2}} \end{equation} Once again, doing the calculations we get a mid transit time of 11,367 $\pm$ 12 s. This time is in UTC time on October $8^{\text{th}}$, 2016. Converting to hours, the mid transit time would have occurred at 03:09:27 $\pm$ 00:00:12 UTC. \section{Discussion (JM and YM)} Comparing the experimental planet radius to the theoretical value of 1.451 $\pm$ 0.074 $R_{\text{Jup}}$, we can see that they agree within one sigma. The theoretical planet to stellar radius ratio is given by 0.1301 $\pm$ 0.0087, which also agrees within error to the experimental value. Since the planet radius agrees within error, we should expect to see the same in the ratio between the radii of the planet and star. As for the transit duration and mid transit time, we have experimental values of 3.158 $\pm$ 0.003 hours and 03:09:27 $\pm$ 00:00:12 UTC time, respectively. Compared to the theoretical values of 3.024288 hours and 03:11:37 UTC, both values are several sigma off. However, they are both quite close to what we expect. The reason why they seemingly disagree is because the error for the experimental values are quite small. This is because we only took into account the statistical error when binning but did not take into account the systematic error when deciding when the transit began and ended. Due to the large fluctuations before the transit, as well as the fact that we looked at the binned data instead of all of the unbinned exposures, there was some extra error when choosing the beginning of the transit. Also, since we just got the end of the transit but did not get data afterwards, the last bin may not have been the end of the transit like we assumed. This uncertainty in the ends of the transit means that there is inherent error in the duration and, in extension, the mid-transit time that we did not adequately take into account. Since the discrepancies between these theoretical and experimental values are small, it seems safe to assume that these values would agree had we properly taken into account the systematic errors. \section{Conclusion (JM and YM)} For our experiment, we set out with the goal of accurately detecting a transit of HD 209458b in front of its host star.From this, we deduced the transit duration, mid-transit time, and the planet to stellar radius ratio of the system. Finally, based on this ratio and knowing the radius of HD 209458, we could determine the radius of the transiting body. After reducing and analyzing all of our data, we can say that most of our results agree with the literature values within error. The experimental planet to stellar radius ratio, and hence the experimental planet radius, agree within error with the theoretical values of both values. However, the experimental transit duration and mid-transit time do not agree with their respective theoretical values within error, despite being extremely close. If we had taken into account systematic error properly, they would have most likely agreed with their theoretical values. Throughout this entire experiment, we have learned just how powerful the method of transit photometry is. By just pointing our modest telescope at a star 47 pc away \citep{2010MNRAS.408.1689S} and taking repeated exposures over a long period, we could detect an exoplanet by simply noticing a change in flux received from the star. After you get the change in relative flux, you only need to know the radius of the host star to determine the planet radius. This is why the detection method of transit photometry is so powerful. \begin{thebibliography}{9} \bibitem[Boyajian et al.(2015)]{2015MNRAS.447..846B} ., ., ., et al.\ 2015, \mnras, 447, 846 \bibitem[Brown et al.(2001)]{2001ApJ...552..699B} .~M., ., .~L., .~W., \& Burrows, A.\ 2001, \apj, 552, 699 \bibitem[Southworth(2010)]{2010MNRAS.408.1689S} Southworth, J.\ 2010, \mnras, 408, 1689 \end{thebibliography} \appendix \section{Clean images code} \label{code: reduction} This was the first step in the data reduction and includes the starter code. This python program takes 5 arguments. The first four are text files containing the list of fits files to be used. These lists include the darks, long exposure flats, short exposure flats, and the science images. The last argument is a name with which all of the output files will begin. It begins by calculating the master dark and flat images, as well as the pixel mask. This is done via function calls. In these functions, the images are loaded up from the current directory by the names provided in the input files for the program. The master images are calculated as the median instead of the mean since the former is less sensitive to outliers, and is therefore better representative of the data. The bad pixel mask was made to record the bad pixels, which was determined to be those with a value less than 2.8. This decision was made after inspecting the ds9 images and noting pixels that were more than 6 sigma away from the average. We found 2 bad pixels. The next part of the code applies these master images and the pixel mask to the science images. After looking at where the bad pixels were, we decided to ignore the data from the bad pixels entirely rather than try to extrapolate the values that they should have been based on the surrounding pixels. This was because the pixels were in a region such that they were never used for the rest of the lab. With this, we created new fits files containing the clean data and copied the headers from the old science images. We also created fits files containing the master flat, master dark, and bad pixels images. \begin{lstlisting}[language=Python, caption= Cleans science images (YM)] ########################################################################################## # This program takes 5 arguments: # # 1) A final name for a file containing a list of dark exposure file names # # 2) A final name for a file containing a list of the long exposure flats file names # # 3) A final name for a file containing a list of the short exposure flats file names # # 4) A final name for a file containing a list of science exposure file names # # 5) A basename that all files written by this program will start with # # # # There are four functions defined below: # # 1) PixelMask(longfiles, shortfiles), whice creates and returns an image of the ratio # # of the non-normalized master flat images for the long and short exposures # # 2) AverageDark(darkfiles), which creates and returns the master dark image # # 3) AverageFlat(flatfiles,masterdark) which creats and returns the master dark # # subtracted flat image # # 4) ScienceExposure(rawscidata,masterdark,masterflat), which applied the master # # dark and flat images to a raw science image. Also gets rid of bad pixels # # # # Below these functions is the main body of the program, which applies the master # # dark and master flat to all of the science images and writes that clean science # # images to files # # # # This code was adapted from code written by the TAs of the Physics 100 course at # # Stanford University ( et al.). # # Modified for PHY 517 / AST 443 at Stony Brook University by and # # # # Modified for specific lab group in AST 443 at Stony Brook University by # # and # # # ########################################################################################## # This is the incomplete version of the calibration script that you will be # using to process all of your observatory data. You will need to finish # this script to complete the data reduction. Remember at any point # during development you can try running the script on a real data and see # if the output products make sense. # Here are the libraries we need. Note that wherever you see np, that # stands for Numpy. 'import' loads an external library. import pyfits import numpy as np import sys,os import pdb # Python is an interpreted programming language, so we have to put all of our functions BEFORE # the main body of the code! #This function finds the bad pixels def PixelMask(longfiles, shortfiles): #Open the long exposure files and store in 2D numpy array containing doubles longflats = np.array([pyfits.open(i.rstrip('\n'))[0].data for i in open(longfiles)]) longflats = longflats.astype(np.float64) #Open the short exposure files and store in 2D numpy array containing doubles shortflats = np.array([pyfits.open(i.rstrip('\n'))[0].data for i in open(shortfiles)]) shortflats = shortflats.astype(np.float64) masterlong=np.median(longflats,axis=0) # Median combines long flat images mastershort=np.median(shortflats,axis=0) # Median combines short flat images image = masterlong/mastershort #These pixels should all be 3 (ratio of our exposure times) badpixels = np.array(np.where(image < 2.8))#Upon inspecting the image with ds9, we determined #that any pixel with a value less than 2.8 was a bad pixel #Making our bad pixel mask mask = np.ones(image.shape, dtype = np.float64) for i in range(0, badpixels.shape[1]): y = badpixels[0,i] x = badpixels[1,i] mask[y,x] = 0.0 return mask # This function does the combining of dark currents def AverageDark(darkfiles): # opens each dark image file and stores the 2d images in a numpy array darkdata=np.array([pyfits.open(i.rstrip('\n'))[0].data for i in open(darkfiles)]) # make the master dark file (uses median) masterdark = np.median(darkdata, axis = 0) return masterdark # This function creates a combined flat field image def AverageFlat(flatfiles): # opens each flat image file and stores the 2d images in a numpy array flatdata=np.array([pyfits.open(i.rstrip('\n'))[0].data for i in open(flatfiles)]) flatdata = flatdata.astype(np.float64) # normalizes each image by its median (useful especially if the flats have very different count level): for i in range(0,flatdata.shape[0]): flatdata[i] = flatdata[i]/np.median(flatdata[i]) masterflat=np.median(flatdata,axis=0) # Median combines flat images masterflat = masterflat/np.mean(masterflat) # Normalizes to the mean of the flats return masterflat # This function creates the processed science image after combined dark, and flat images have been created. def ScienceExposure(rawscidata,masterdark,masterflat,badpixel): rawimage = np.array([rawscidata.data]) #Gets the data from the header of the science image file rawimage = rawimage.astype(np.float64) scienceimage= badpixel*((rawimage - masterdark)/masterflat) #creates final science image return scienceimage # This is the end of the functions. The main body of the code begins below. # Each of these is an argument that needs to be on the calling of the script. # Make sure you run with all arguments provided or you will run into errors! darkfilelist=sys.argv[1] # First argument is a text file that lists the names of all dark current image file names longflatfilelist=sys.argv[2] # Second argument is a text file that lists the names of all of the long exposure flat field images shortflatfilelist = sys.argv[3] # Third argument is a text file that lists the names of all of the short exposure flat field images sciencefilelist=sys.argv[4] # Fourth argument is a text file that lists the names of all the science images basename=sys.argv[5] # All of the output files will start with the string value of basename. finaldark=AverageDark(darkfilelist) # Find function aboved finalflat=AverageFlat(longflatfilelist) # Find function aboved pixelmask = PixelMask(longflatfilelist, shortflatfilelist) for sciencefile in open(sciencefilelist): # Loops though all science files to apply finaldark and finalflat corrections sciencefile = sciencefile.rstrip(' \n') rawdata=pyfits.open(sciencefile+'.FIT')[0] # This gets the 1st extension (starts with 0!), this is an example of # using pyfits.open, this is a FITS file object finalimage=ScienceExposure(rawdata,finaldark,finalflat,pixelmask) # Find function above sciheader=rawdata.header # This grabs the header object from the FITS object rawdata newscience=basename+'_'+sciencefile+'_clean.fits' # Appending filenames onto the base sciencehdu=pyfits.PrimaryHDU(finalimage,header=sciheader) # This converts a numpy array into a FITS object with a # data block (finalimage) and a header (sciheader) sciencehdu.writeto(newscience, clobber=True) # This writes the fits object to the file name newscience, which is # defined above The clobber means to overwrite the file if it already exists. newdark=basename+'_Master_Dark.fits' newflat=basename+'_Master_Flat.fits' newpixel = basename + '_Pixel_Map.fits' darkhdu=pyfits.PrimaryHDU(finaldark) darkhdu.writeto(newdark, clobber=True) flathdu = pyfits.PrimaryHDU(finalflat) flathdu.writeto(newflat, clobber = True) pixelhdu = pyfits.PrimaryHDU(pixelmask) pixelhdu.writeto(newpixel, clobber = True) ###################################### End of Program ########################################## \end{lstlisting} \section{Astrometry and Source Extractor scripts} \label{code: astro sex} These are simple bash scripts that run the clean images through astronometry.net and Source Extractor. The former is given a text file list containing the names of all of the clean images. It then runs them through astronometry, solving the images for the WCS of each image. The new FITS files returned have headers with the correct WCS entry. \begin{lstlisting}[language = bash, caption= Runs clean images through astronometry.net (YM)] #! /bin/bash -u #Argument is list of file filename=$1 while read file do solve-field --ra 330.896 --dec 18.914 --radius 0.35 ${file} done < ${filename} \end{lstlisting} This second script runs the solved images through Source Extractor to find images. It takes a text file list of all the solved images names. The default parameters and configuration files are also provided. Important to note is the \verb|FLUX_APER| and \verb|FLUXERR_APER| parameters only have one value. The \verb|PHOT_APERTURES| keyword was decided using ds9; it corresponds to roughly the diameter of our target star in our image. Source Extractor then created catalog files for each exposure with the parameters specified in the default parameter file. \begin{lstlisting} [language = bash, caption= Runs solved images through Source Extractor (YM)] #! /bin/bash -xv filelist=$1 while read -r file do basename=$(echo ${file} | sed 's/\.new/\.cat/') #rename files sex ${file} -c default.se -CATALOG_NAME ${basename} done < ${filelist} \end{lstlisting} \begin{lstlisting}[caption = Default parameters for Source Extractor (YM)] NUMBER # Running object number X_IMAGE # Object position along x [pixel] Y_IMAGE # Object position along y [pixel] ALPHA_J2000 # Right ascension of barycenter (J2000) [deg] DELTA_J2000 # Declination of barycenter (J2000) [deg] FLUX_APER(1) # Flux vector within fixed circular aperture(s) [count] FLUXERR_APER(1) # RMS error vector for aperture flux(es) [count] FLUX_RADIUS # Fraction-of-light radii [pixel] FWHM_IMAGE # FWHM assuming a gaussian core [pixel] BACKGROUND # Background at centroid position [count] THRESHOLD # Detection threshold above background [count] FLUX_MAX # Peak flux above background [count] ISOAREA_IMAGE # Isophotal area above Analysis threshold [pixel**2] A_IMAGE # Profile RMS along major axis [pixel] B_IMAGE # Profile RMS along minor axis [pixel] THETA_IMAGE # Position angle (CCW/x) [deg] FLAGS # Extraction flags \end{lstlisting} \begin{lstlisting}[caption = Configuration file for Source Extractor (YM)] # Default configuration file for SExtractor 2.19.5 # EB 2014-03-19 # #-------------------------------- Catalog ------------------------------------ CATALOG_NAME test.cat # name of the output catalog CATALOG_TYPE ASCII_HEAD # NONE,ASCII,ASCII_HEAD, ASCII_SKYCAT, # ASCII_VOTABLE, FITS_1.0 or FITS_LDAC PARAMETERS_NAME default.param # name of the file containing catalog contents #------------------------------- Extraction ---------------------------------- DETECT_TYPE CCD # CCD (linear) or PHOTO (with gamma correction) DETECT_MINAREA 5 # min. # of pixels above threshold DETECT_THRESH 2.5 # or , in mag.arcsec-2 ANALYSIS_THRESH 2.5 # or , in mag.arcsec-2 FILTER N # apply filter for detection (Y or N)? FILTER_NAME default.conv # name of the file containing the filter DEBLEND_NTHRESH 32 # Number of deblending sub-thresholds DEBLEND_MINCONT 0.005 # Minimum contrast parameter for deblending CLEAN Y # Clean spurious detections? (Y or N)? CLEAN_PARAM 1.0 # Cleaning efficiency MASK_TYPE CORRECT # type of detection MASKing: can be one of # NONE, BLANK or CORRECT #------------------------------ Photometry ----------------------------------- PHOT_APERTURES 21.0 # MAG_APER aperture diameter(s) in pixels PHOT_AUTOPARAMS 2.5, 3.5 # MAG_AUTO parameters: , PHOT_PETROPARAMS 2.0, 3.5 # MAG_PETRO parameters: , # SATUR_LEVEL 50000.0 # level (in ADUs) at which arises saturation SATUR_KEY SATURATE # keyword for saturation level (in ADUs) MAG_ZEROPOINT 0.0 # magnitude zero-point MAG_GAMMA 4.0 # gamma of emulsion (for photographic scans) GAIN 0.0 # detector gain in e-/ADU GAIN_KEY GAIN # keyword for detector gain in e-/ADU PIXEL_SCALE 1.0 # size of pixel in arcsec (0=use FITS WCS info) #------------------------- Star/Galaxy Separation ---------------------------- SEEING_FWHM 1.2 # stellar FWHM in arcsec STARNNW_NAME default.nnw # Neural-Network_Weight table filename #------------------------------ Background ----------------------------------- BACK_SIZE 64 # Background mesh: or , BACK_FILTERSIZE 3 # Background filter: or , BACKPHOTO_TYPE GLOBAL # can be GLOBAL or LOCAL #------------------------------ Check Image ---------------------------------- CHECKIMAGE_TYPE NONE # can be NONE, BACKGROUND, BACKGROUND_RMS, # MINIBACKGROUND, MINIBACK_RMS, -BACKGROUND, # FILTERED, OBJECTS, -OBJECTS, SEGMENTATION, # or APERTURES CHECKIMAGE_NAME check.fits # Filename for the check-image #--------------------- Memory (change with caution!) ------------------------- MEMORY_OBJSTACK 3000 # number of objects in stack MEMORY_PIXSTACK 300000 # number of pixels in stack MEMORY_BUFSIZE 1024 # number of lines in buffer #----------------------------- Miscellaneous --------------------------------- VERBOSE_TYPE NORMAL # can be QUIET, NORMAL or FULL HEADER_SUFFIX .head # Filename extension for additional headers WRITE_XML N # Write XML file (Y/N)? XML_NAME sex.xml # Filename for XML output \end{lstlisting} \section{Stellar fluxes and time} \label{code: flux} This python program takes text file lists as arguments. The first two are the names of the catalog files from Source Extractor and solved clean science images from astronometry. The last takes the coordinates of the reference and science stars. The script finds all of these stars in each image from the catalog files. This is done by finding the star that is closest to the coordinates of each star in each image. If the matching radius is more than one arcsecond, it ignores that exposure for that star. It then outputs in a text file the time of the image and the flux, flux error, matching distance, and catalog number for each star in each image. The last two are only used for reference and troubleshooting; they are not used in the rest of the lab. The increase in the matching radius for certain stars in specific images indicated to us that they were being incorrectly matched to another star. The catalog number allowed us to use ds9 and confirm that the stars were mismatched. To determine which stars were mismatched, we set the maximum matching radius to one arcsecond. If a star was ignored in an image, the flux and its error is set to zero. The last part of the code tells the user how many stars were ignored. \begin{lstlisting}[language = Python, caption = Outputs the flux and its error for each star in every exposure as well as the time for each image (YM)] import numpy as np import pyfits import sys import pdb import math #********************************Functions************************************** #Returns index of the star in catalog for a given image def findStar(starcoordinates, rawdata): #Epsilon is the distance from our ideal star position and the actual position epsilon = 1000 #Other variables starRA = starcoordinates[0] starDEC = starcoordinates[1] for i in range(0, rawdata.shape[0]): #Find distance testRA = rawdata[i,1] testDEC = rawdata[i,2] distance = math.sqrt((starRA-testRA)**2 + (starDEC-testDEC)**2) if (distance < epsilon): epsilon = distance index = i #Check if our star is still in the image if (epsilon >= 0.00031): index = -1 return (index, epsilon) #Create file with table of fluxes and errors w.r.t. time def makeTable(table, list, starindex, epsilon, catNum): #Open text file to hold tables starnumber = starindex + 1 filename = 'exo_fluxtable' + str(starnumber) + '.txt' outputfile = open(filename, 'w+') #Set Counter imagenumberindex = 0 #Write table to file for imagenames in open(list): #Open FITS files hdulist = pyfits.open(imagenames.rstrip('\n')) #Get time of image time = str(hdulist[0].header['date-obs']) outputfile.write(time + ' ' + str(table[starindex, :, imagenumberindex]) + ' ' + str(epsilon[starindex, imagenumberindex]) + ' ' + str(catNum[starindex, imagenumberindex]) + '\n') imagenumberindex += 1 outputfile.close() return #**********************************Main***************************************** #Get list of file names catfilelist = sys.argv[1] #file with list of .cat files used for fluxes fitsfilelist = sys.argv[2] #file with list of .new files used for times refstarlist = sys.argv[3] #file with list of RA and DEC of all stars #Define some variables coordinates = np.loadtxt(refstarlist) #array of RA and DEC of stars abort = False #if one of our stars isn't in every picture, we abort code to replace it passedstars = 0 catlist = open(catfilelist) #Create some more variables currentimage = 0 #index of current image for mastertable numberofimages = len(catlist.readlines()) numberofstars = coordinates.shape[0] e = np.zeros((numberofstars, numberofimages)) #array of epsilons for each star in each exposure catalognumber = np.zeros((numberofstars, numberofimages)) #number in the catalog #Create 3D array to hold time, flux, and error for all stars in every image #axis = 0: stars #axis = 1: data (flux and flux error) #axis = 2: images (each index in this axis refers to a different picture) mastertable = np.zeros((numberofstars, 2, numberofimages)) catlist.close() for catfilename in open(catfilelist): #go through images catfilename = catfilename.rstrip('\n') #Extract relevent data (catalog number, RA, DEC, flux, and flux error) alldata = np.loadtxt(catfilename) data = np.column_stack([alldata[:,0], alldata[:,3], alldata[:,4], alldata[:,5], alldata[:,6]]) #Go through stars for currentstar in range(0, numberofstars): #Get index of star in catalog starindex, e[currentstar, currentimage] = findStar(coordinates[currentstar, :], data) catalognumber[currentstar, currentimage] = data[starindex, 0] #If the star isn't in the picture, ignore the star in this exposure if (starindex == -1): passedstars +=1 continue #If it is in the picture else: #Add the fluxes and errors to the table mastertable[currentstar, 0, currentimage] = data[starindex,3] mastertable[currentstar, 1, currentimage] = data[starindex,4] #Increment image number counter currentimage += 1 #Print out the images to the files for i in range(0, numberofstars): makeTable(mastertable, fitsfilelist, i, e, catalognumber) #Lemme know how many times stars were ignored print('\n\n' + str(passedstars) + ' stars were ignored') \end{lstlisting} \section{Clean tables and convert times} \label{code: multicommand} This bash script takes as argument a text file list containing the names of the files from appendix \ref{code: flux}. It also gets rid of the square brackets and extra spaces in the table that results from writing a numpy array directly to a file. Finally, it removes the date from the time column and converts the time to seconds. Like all of the bash scripts in this lab, it replaces the files it changes rather than makes copies. It does this by making the changes on clones of the files, removes the old files, and then renames the new files to match the names of the old ones. \begin{lstlisting}[language = bash, caption = Cleans up file and readies for the rest of the lab (YM)] #! /bin/bash -u #Does multiple commands on multiple files and saves them to the file #Argument is text file of list of files to modify filelist=$1 while read -r oldfilename do #Make temporary file to hold changes newfilename=${oldfilename}.temp timesfile=${oldfilename}.time.temp newnewfilename=${oldfilename}.temp2 #Get rid of date, brackets, and excess space sed -e 's|2016-10-08T||' -e 's|\[||g' -e 's|\]||g' -e 's| \+| |g' <$oldfilename >$newfilename #Convert time to seconds and save to a file cut -d ' ' -f 1 <$newfilename | awk 'BEGIN{ FS = ":"}; {print ($1*3600 + $2*60 + $3)}' >$timesfile awk 'FNR==NR{a[NR] = $1; next}{$1 = a[FNR]}1' $timesfile $newfilename >$newnewfilename #Make changes permenent rm -f $oldfilename $timesfile $newfilename mv $newnewfilename $oldfilename done < $filelist \end{lstlisting} \section{Plotting the curves} \label{code: curves} This program takes a list of all of the files produced by the code in appendix \ref{code: multicommand} as its argument. In the first part, it normalizes each star to its average flux. This does not take into account the noisy beginning exposures, outliers, or the mismatched stars. It then plots each star's lightcurve using this normalized flux. These plots are saved as files. The second half of this program outputs text files containing the time of each exposure and the normalized flux and its error for each star. These tables set any remaining outliers to zero. It defines an outlier as one that is more than 4 sigma away from the average flux for that star. The exception is the science target, which does not get normalized to its average. However, it does receive the same sigma clipping treatment. \begin{lstlisting}[language = Python, caption = Plots the lightcurves for each star and outputs fluxes to files (YM and JM)] import pyfits import numpy as np import math as m import sys import matplotlib.pyplot as plt import pdb #Get list of table files fluxlist = sys.argv[1] #Make the figure fig = plt.figure() plt.grid(True) plt.xlabel('Time') plt.ylabel('Relative Flux') plt.title('Lightcurve of Reference Stars') ctr = 0 for tablefile in open(fluxlist): ctr += 1 fig.clf() #Open file tablefile = tablefile.rstrip('\n') data = np.loadtxt(tablefile) #Get the information fluxdata = data[:,1] #fluxdata holds all raw data fluxerrordata = data[:,2] time = data[:,0] #Normalize flux and ignore outliers flux = fluxdata[np.where(fluxdata!=0)] #flux used for average flux = flux[287:] #flux ignores beginning exposures (no outliers) fluxerror = fluxerrordata[287:] #fluxerror ignores beginning images average = np.mean(flux) #average ignores beginning exposures n = flux.shape[0] sigma = m.sqrt( np.sum((flux-average)**2) / (n-1)) normfluxdata = fluxdata/average #normfluxdata normalizes ALL exposures normfluxerrordata = fluxerrordata/average #normfluxerrordata normalizes ALL exposures #Plot data plt.errorbar(time, normfluxdata, yerr = normfluxerrordata, fmt = 'r.') fig.savefig('exo_curves' + str(ctr) + '.pdf', bbox_inches='tight', dpi=fig.dpi) #Make files containing time, flux, and error #For the science target, we won't normalize to average if (ctr == 11): #Scrap non-zero outliers more that 2 sigma away for non-beginning exposures tempflux = fluxdata[287:] #no beginning exposures, HAS outliers tempflux[np.where(tempflux < (average - 4*(sigma)))] = 0.0 tempflux[np.where(tempflux > (average + 4*(sigma)))] = 0.0 fluxerror[np.where(tempflux < (average - 4*(sigma)))] = 0.0 fluxerror[np.where(tempflux > (average + 4*(sigma)))] = 0.0 #Write average fluxes to file filename = ('exo_fluxdata' + str(ctr) + '.txt') finalflux = np.concatenate((fluxdata[:287], tempflux), axis = 0) #flux with beginning finalfluxerror = np.concatenate((fluxerrordata[:287], fluxerror), axis=0) np.savetxt(filename, np.column_stack((time, finalflux, finalfluxerror)), fmt='%f') #For the others, we will normalize to average else: #Scrap non-zero outliers more that 2 sigma away for non-beginning exposures tempflux = normfluxdata[287:] #ignores first exposures tempfluxerror = normfluxerrordata[287:] #ignores first exposures tempflux[np.where( tempflux < ((average - 4*sigma)/average) )] = 0.0 tempflux[np.where( tempflux > ((average + 4*sigma)/average) )] = 0.0 tempfluxerror[np.where( tempflux < ((average - 4*sigma)/average) )] = 0.0 tempfluxerror[np.where( tempflux > ((average + 4*sigma)/average) )] = 0.0 #Write average fluxes to file filename = ('exo_fluxdata' + str(ctr) + '.txt') normflux = np.concatenate((normfluxdata[:287], tempflux), axis = 0) normfluxerror = np.concatenate((normfluxerrordata[:287], tempfluxerror), axis = 0) np.savetxt(filename, np.column_stack((time, normflux, normfluxerror)), fmt='%f') \end{lstlisting} \section{Master table} \label{code: bigtable} This Python program takes as argument some of the files created by the code from appendix \ref{code: curves}. It ignores the first, ninth, and tenth reference stars. It begins by calculating the weighted average and its error of the fluxes from the reference stars in each exposure. It then normalizes the science flux by this average. The error in this ratio is also calculated. The output is a text file containing the time, unnormalized science flux and its error, weighted average of the reference stars, and the normalized science flux and its error. \begin{lstlisting}[language = Python, caption = Creates table of values to be used for the final light curve (YM and JM)] import pdb import numpy as np import math as m import sys #Read in the tables for each star referencestarfile2 = sys.argv[1] referencestarfile3 = sys.argv[2] referencestarfile4 = sys.argv[3] referencestarfile5 = sys.argv[4] referencestarfile6 = sys.argv[5] referencestarfile7 = sys.argv[6] referencestarfile8 = sys.argv[7] referencestarfile11 = sys.argv[8] #Import the data into arrays. #We will only work with the first 3 columns #These are the times, fluxes, and flux errors data2 = np.loadtxt(referencestarfile2) data3 = np.loadtxt(referencestarfile3) data4 = np.loadtxt(referencestarfile4) data5 = np.loadtxt(referencestarfile5) data6 = np.loadtxt(referencestarfile6) data7 = np.loadtxt(referencestarfile7) data8 = np.loadtxt(referencestarfile8) data11 = np.loadtxt(referencestarfile11) #The following is storing the relevent data into arrays #-------------------------------------- fluxdata2 = data2[:,1] fluxerrordata2 = data2[:,2] time2 = data2[:,0] #------------------------------------- fluxdata3 = data3[:,1] fluxerrordata3 = data3[:,2] time3 = data3[:,0] #------------------------------------- fluxdata4 = data4[:,1] fluxerrordata4 = data4[:,2] time4 = data4[:,0] #------------------------------------- fluxdata5 = data5[:,1] fluxerrordata5 = data5[:,2] time5 = data5[:,0] #-------------------------------------- fluxdata6 = data6[:,1] fluxerrordata6 = data6[:,2] time6 = data6[:,0] #-------------------------------------- fluxdata7 = data7[:,1] fluxerrordata7 = data7[:,2] time7 = data7[:,0] #--------------------------------------- fluxdata8 = data8[:,1] fluxerrordata8 = data8[:,2] time8 = data8[:,0] #------------------------------------- fluxdata11 = data11[:,1] fluxerrordata11 = data11[:,2] time11 = data11[:,0] #------------------------------------ #Concatanate all of the fluxes and their errors fluxdata = np.column_stack([fluxdata2, fluxdata3, fluxdata4, fluxdata5, fluxdata6, fluxdata7, fluxdata8]) fluxerrordata = np.column_stack([fluxerrordata2, fluxerrordata3, fluxerrordata4, fluxerrordata5, fluxerrordata6, fluxerrordata7, fluxerrordata8]) #Initialize some variables mu_ref = np.zeros(fluxdata.shape[0]) sigma_ref = np.zeros(fluxdata.shape[0]) ratio = np.zeros(fluxdata.shape[0]) sigma_ratio = np.zeros(fluxdata.shape[0]) #Calculate average reference star flux for each exposure and its error for i in range(0, fluxerrordata.shape[0]): #i = image index denominator = 0 numerator = 0 for n in range(0, fluxerrordata.shape[1]): #n = star index #Ignore certain exposures from each reference star if fluxerrordata[i, n] == 0: continue #Calculate parts of average and sigma numerator += fluxdata[i,n]/(fluxerrordata[i,n]**2) denominator += (1.0/(fluxerrordata[i, n]**2)) #Actual average and error for each image if (numerator == 0) and (denominator == 0): #ignore the beginning exposures continue mu_ref[i] = numerator/denominator sigma_ref[i] = m.sqrt(1.0/denominator) #Get ratio between science flux and average ref star flux #and its error for k in range(0, mu_ref.shape[0]): #Ignore certain exposures from each reference star if fluxdata11[k] == 0: continue ratio[k] = (fluxdata11[k])/(mu_ref[k]) sigma_ratio[k] = ratio[k]*m.sqrt((fluxerrordata11[k]/fluxdata11[k])**2+(sigma_ref[k]/mu_ref[k])**2) #Write everything to a text file with open('exo_masterrecord.txt', 'w+') as outfile: for i in range(0,ratio.shape[0]): outfile.write(str(time11[i]) + ' ' + str(fluxdata11[i]) + ' ' + str(fluxerrordata11[0]) + ' ' + str(mu_ref[i]) + ' ' + str(ratio[i]) + ' ' + str(sigma_ratio[i]) + '\n') \end{lstlisting} \section{Lightcurve} \label{code: lightcurve} The following code takes as input the table produced by the code in appendix \ref{code: bigtable}. It calculates the baseline flux, which is used to normalize the ratio defined in the same earlier code. The baseline flux is the average flux when the planet is not transiting the star. The exposures to be used for calculating the baseline can be set within the main body of the program. The function \verb|getLim| matches the range of exposures to the range of indices for the array holding the images. The program then gets rid of the ignored exposures. Finally, it bins the lightcurve, plots it, and saves the image to a file. \begin{lstlisting}[language = Python, caption= Plots the lightcurve for our transit] import pdb import numpy as np import math as m import sys import matplotlib.pyplot as plt #Calculate average and its error for a given data set def Average(values, error= None): #No error in the values if (error is None): #Calculate the average mu = np.sum(values)/values.shape[0] #Calculate the error s = np.sum((values - mu)**2)/(values.shape[0] - 1) sigma = m.sqrt(s/values.shape[0]) #Error in the values else: #Initialize stuff num = 0 denom = 0 #Caluclations for n in range(0, values.shape[0]): if values[n] == 0: #ignore outliers continue num += values[n]/(error[n]**2) denom += 1.0/(error[n]**2) #Mu and Sigma if (num == 0) and (denom == 0): #in case series of outliers mu = 0 sigma = 0 elif denom == 0: #in case no error mu = num/denom sigma == 0 else: #if normal mu = num/denom sigma = m.sqrt(1.0/denom) return (mu, sigma) #Gives limits (indices) of exposures used for baseline flux def getLim(images, lowexp, upexp): l = np.array(np.where(images > lowexp)) u = np.array(np.where(images < upexp)) lowlim = l[0, 1] uplim = u[0, -1] return (lowlim, uplim) #Extract data from input tablefile = sys.argv[1] mastertable = np.loadtxt(tablefile) exposures = mastertable[:,0] sci_flux = mastertable[:, 1] sci_error = mastertable[:, 2] ratio = mastertable[:, 4] ratio_error = mastertable[:, 5] #Get times for exposures ignore = np.where(sci_flux == 0) #locations of outliers exposures = np.delete(exposures, ignore) #ignore outliers #Get exposure indicies for the range of fluxes from which to calculate baseline flux lowlimit, uplimit = getLim(exposures, 4000, 6000) #picture number for lower limit of fluxes used for baseline #Ignore outliers and caluclate the baseline flux sci_flux = np.delete(sci_flux, ignore) #ignore outliers sci_error = np.delete(sci_error, ignore) #ignore outliers baseline, baseline_error = Average(sci_flux[(lowlimit-1) : (uplimit+1)], sci_error[(lowlimit-1) : (uplimit+1)]) #Get ratios ratio = np.delete(ratio, ignore) #ignore outliers ratio_error = np.delete(ratio_error, ignore) #ignore outliers #Normalize the ratio by baseline flux flux = ratio/baseline flux_error = np.zeros(flux.shape[0]) for i in range(0,flux.shape[0]): if ratio[i] == 0: continue flux_error[i] = flux[i] * np.sqrt((baseline_error/baseline)**2 + (ratio_error[i]/ratio[i])**2) #Set up for binning binsize = 12 times = np.zeros(((exposures.shape[0]/binsize) + 1)) times_error = np.zeros(times.shape[0]) sciencecurve = np.zeros(((flux.shape[0]/binsize) + 1)) sciencecurve_error = np.zeros(sciencecurve.shape[0]) ctr = 0 #counter for index of times array rmLastBin = False #Bin times and fluxes for n in range(0, exposures.shape[0], binsize): #Times if (n + binsize)>= exposures.shape[0]: #last exopsure has smaller bin size #binflux, binflux_error = Average(exposures[n:exposures.shape[0]]) #Uncomment above line if you want the last bin #The line below is if you want to ignore it rmLastBin = True continue else: #rest of exposures have normal bin sizes bintime, bintime_error = Average(exposures[n:n+binsize]) times[ctr] = bintime times_error[ctr] = bintime_error #Fluxes binflux, binflux_error = Average(flux[n:n+binsize], flux_error[n:n+binsize]) sciencecurve[ctr] = binflux sciencecurve_error[ctr] = binflux_error ctr += 1 #Remove last bin if needed if rmLastBin: times = times[:-1] times_error = times_error[:-1] sciencecurve = sciencecurve[:-1] sciencecurve_error = sciencecurve_error[:-1] #Plot our light curve fig = plt.figure() plt.grid(True) plt.xlabel('Time (s)') plt.ylabel('Relative flux') plt.title('Lightcurve for HD 209458') plt.errorbar(x = times, y = sciencecurve, xerr = times_error, yerr = sciencecurve_error, fmt = 'k.') plt.show() fig.savefig('exo_lightcurve.pdf', bbox_inches='tight', dpi=fig.dpi) \end{lstlisting} \end{document}\documentclass{article} \usepackage{amssymb} \usepackage{comment} \usepackage{courier} \usepackage{fancyhdr} \usepackage{fancyvrb} \usepackage[T1]{fontenc} \usepackage[top=.75in, bottom=.75in, left=.75in,right=.75in]{geometry} \usepackage{graphicx} \usepackage{lastpage} \usepackage{listings} \lstset{basicstyle=\small\ttfamily} \usepackage{mdframed} \usepackage{parskip} \usepackage{ragged2e} \usepackage{soul} \usepackage{upquote} \usepackage{xcolor} \usepackage[ampersand]{easylist} % http://www.monperrus.net/martin/copy-pastable-ascii-characters-with-pdftex-pdflatex \lstset{ upquote=true, columns=fullflexible, literate={*}{{\char42}}1 {-}{{\char45}}1 {^}{{\char94}}1 } \lstset{ moredelim=**[is][\color{blue}\bf\small\ttfamily]{@}{@}, } % http://tex.stackexchange.com/questions/40863/parskip-inserts-extra-space-after-floats-and-listings \lstset{aboveskip=6pt plus 2pt minus 2pt, belowskip=-4pt plus 2pt minus 2pt} \usepackage[colorlinks,urlcolor={blue}]{hyperref} \begin{document} \fancyfoot[L]{\color{gray} C4CS -- F'18} \fancyfoot[R]{\color{gray} Revision 1.0} \fancyfoot[C]{\color{gray} \thepage~/~\pageref*{LastPage}} \pagestyle{fancyplain} \title{\textbf{Office Hours ++ (Unit Testing and Python)\\}} \author{\textbf{\color{red}{Due: Sunday, November 11th, 11:59PM (Hard Deadline)}}} \date{} \maketitle \section*{Submission Instructions} Submit this assignment on \href{https://gradescope.com/courses/24368}{Gradescope}. You may find the free online tool \href{https://www.pdfescape.com}{PDFescape} helpful to edit and fill out this PDF. You may also print, handwrite, and scan this assignment. \section*{Adding Features to Your Python RPN Calculator} Building off of what was covered in lecture, work on using the test-driven development methodology and implement \textbf{three (3)} of the features from the list below. Note that this list is not comprehensive, so if there is a different feature that you would like to build that is non-trivial (e.g. implementing a function that would just change the sign on a number would be trivial), you may choose to build that instead. In order to receive credit, you will need to prove the test cases for your additional functions initially fail, and then later pass when the function is fully implemented. \subsection*{Feature List:} Choose \textbf{three (3)} of the following features to build. You may elect to build any number of custom features and each will count as a separate feature - you are not restricted to just one custom feature. \begin{itemize} \item[$\square$] \textbf{Implement additional calculator keys} For this feature, implement the following subfeatures: \begin{itemize} \item[$\square$] Calculate percentages using the \texttt{\%}. You may find \href{https://blogs.msdn.microsoft.com/oldnewthing/20080110-00/?p=23853}{this article} helpful. \end{itemize} \begin{itemize} \item[$\square$] Calculate exponent using \texttt{\^}. \end{itemize} \begin{itemize} \item[$\square$] Perform integer division using \texttt{//}. \end{itemize} \end{itemize} \begin{itemize} \item[$\square$] \textbf{Implement bitwise operators (and, or and not)} \end{itemize} \begin{itemize} \item[$\square$] \textbf{Implement a basic math library} For this feature, implement the following subfeatures: \begin{itemize} \item[$\square$] Allow for usage of constants (\texttt{pi}, \texttt{e}, ...) \end{itemize} \begin{itemize} \item[$\square$] Binary functions \end{itemize} \begin{itemize} \item[$\square$] Unary functions (\texttt{sin}, \texttt{cos}, etc.) \end{itemize} \end{itemize} \begin{itemize} \item[$\square$] \textbf{Degrees and Radians Mode} This would follow from the above feature. Add a command/method which can set whether trigonometric operations use degrees or radians. For example, by entering the keyword \texttt{rad}, if the operation \texttt{3.1415 sin} is entered, the output would be \texttt{0}. Equivalently, if the mode is set using \texttt{deg}, if the operation \texttt{360 cos} is entered, the output would be \texttt{1}.\newline \end{itemize} \newpage \begin{itemize} \item[$\square$] \textbf{Use the Results of a Previous Calculation} Add the ability to use the results of your previous calculation in the next calculation. For example, if we have the first input as \texttt{2 3 +} followed by the next command \texttt{:ans 7 +}, the output should be \texttt{12}. In order to implement this, you will need to define a language - in the example we used the colon(:) to denote this special variable \newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Summation} Implement a command that find the sum of all of the elements on the stack and adds this result to the top of the stack.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Different Base Number Systems} Add a command or method which will allow the user to use your calculator for calculations in different base number systems. For example, using an option \texttt{hex} and then entering the input \texttt{A A +} should result in the output \texttt{14}.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Factorial Operator} Implement the factorial operator. For example, the input \texttt{4 !} should return the output \texttt{24}. \newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Error Handling for Division by Zero} Implement an error handling method that perevents division by 0 errors and saves the user's existing state. For example, if the stack initually contained \{\texttt{1}\} and a user enters \texttt{4 0 /}, output a helpful error message and preserve the initial stack before these values were entered (i.e. \{\texttt{1}\}). \newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Session History} Implement a command such that when it is called, it outputs the standard format of the previous operation. For example, if the previous input was \texttt{3 2 +}, calling this command will print \texttt{3 + 2 = 5}.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Convert Between Decimal and Fraction} Implement a command to convert the item on the stack from decimal to fraction and vice-versa. For example, calling this command when the item at the top of the stack is \texttt{0.75}, the output should return \texttt{3/4}.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Repeat Operator} Implement a repeat operator that will repeatedly carry out a provided binary operation on all items provided in the input line. For example, the input \texttt{4 2 6 * !} (where ! is the repeat operator) would result in the output \texttt{48}, regardless of whatever was previously on the stack.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Rotate} Implement a command that will rotate the order of all items currently on the stack. For example, if the stack currently contains \{\texttt{2 4 6}\}, after calling this operator the stack should be \{\texttt{6 4 2}\}.\newline \end{itemize} \begin{itemize} \item[$\square$] \textbf{Copy} Implement a command that will add a copy of the current top element to the stack. For example, if the stack currently contains \{\texttt{2 4 6}\}, after calling this operator the stack should be \{\texttt{2 2 4 6}\}.\newline \end{itemize} \newpage \begin{itemize} \item[$\square$] \textbf{Allow a Persistent Stack} Make the stack for your RPN calculator persistent. For example, the input \texttt{1 2 3 +} should not produce any errors. From this, the prompt should also be customized to display information about the stack. Using the previous example, the prompt could now be customized to be \texttt{rpn calculator [1, 5]}, or something similar. \end{itemize} \begin{itemize} \item[$\square$] \textbf{Add a Memory System} Add a basic memory system to your calculator. This would be equivalent to the \texttt{M+}, \texttt{M-}, \texttt{MR} and \texttt{MC} on a regular calculator. This could be extended to add a near infinite amount of memory registers by defining another special character (as in the "Use the Results of a Previous Calculation). For example, these registers could be called using the \texttt{\&} key: \texttt{\&myval+}, \texttt{\&c4cs-}, etc. \end{itemize} \begin{itemize} \item[$\square$] \textbf{Read data in from an external file} Add the ability to read in data from an external file for your calculator. This external file could be formatted in a style of your choice (csv, tsv, etc.). \end{itemize} \begin{itemize} \item[$\square$] \textbf{Custom Feature!} If there is a meaningful command not listed above and currently not already implemented by the calculator, you may choose to implement this instead. For example, you could choose to use other Python libraries (such as numbers, cmath, decimal, fractions, statistics, random, NumPy, SciPy, etc) to add new features. \newline \end{itemize} \newpage \section*{Feature 1:} \subsection*{Test Code:} \vspace*{5cm} \subsection*{Screenshot of Failing Test:} \vspace*{5cm} \subsection*{Implementation Code:} \vspace*{5cm} \subsection*{Screenshot of Passing Test:} \vspace*{5cm} \newpage \section*{Feature 2:} \subsection*{Test Code:} \vspace*{5cm} \subsection*{Screenshot of Failing Test:} \vspace*{5cm} \subsection*{Implementation Code:} \vspace*{5cm} \subsection*{Screenshot of Passing Test:} \vspace*{5cm} \newpage \section*{Feature 3:} \subsection*{Test Code:} \vspace*{5cm} \subsection*{Screenshot of Failing Test:} \vspace*{5cm} \subsection*{Implementation Code:} \vspace*{5cm} \subsection*{Screenshot of Passing Test:} \vspace*{5cm} \end{document} xdocs/user-guide/plugins/memory/src/biblio.tex \addcontentsline{toc}{section}{\numberline{}Bibliography} \begin{thebibliography}{99} \bibitem{freericks} M.~Freericks. \emph{The nML Machine Description Formalism}. Technical Report TR SM-IMP/DIST/08, TU Berlin CS Department, 1993. \end{thebibliography} @phdthesis{dissertation, title = {Anomaly Detection Through Explanations}, author = {Gilpin, .}, school = {Massachusetts Institute of Technology}, year = {2020} } flower-go/diplomka \chapter{Title of the first chapter} An~example citation: \cite{Andel07} \cite{ZonkyParametryST} \cite{BERT_ORIG} \section{Title of the first subchapter of the first chapter} \section{Title of the second subchapter of the first chapter} \hypertarget{dir_f4895f05f69d3971d84bff35f7083e67}{}\section{code/hwlib/demo/db103/db103-\/\#0060-\/blink-\/blink Directory Reference} \label{dir_f4895f05f69d3971d84bff35f7083e67}\index{code/hwlib/demo/db103/db103-\/\#0060-\/blink-\/blink Directory Reference@{code/hwlib/demo/db103/db103-\/\#0060-\/blink-\/blink Directory Reference}} pglutz/OpenLogiccontent/set-theory/spine/idea.tex \documentclass[../../../include/open-logic-section]{subfiles} \begin{document} \olfileid{sth}{spine}{valpha} \olsection{Defining the Stages as the $V_\alpha$s} In \olref[sth][ordinals][]{chap}, we defined well-orderings and the (von Neumann) ordinals. In this chapter, we will use these to characterise the hierarchy of sets \emph{itself}. To do this, recall that in \olref[sth][ordinals][opps]{sec}, we defined the idea of successor and limit ordinals. We use these ideas in following definition: \begin{defn}\ollabel{defValphas} \begin{align*} V_\emptyset &:= \emptyset\\ V_{\ordsucc{\alpha}} &:= \Pow{V_\alpha} & & \text{for any ordinal }\alpha\\ V_{\alpha} &:= \bigcup_{\gamma < \alpha} V_\gamma & & \text{when }\alpha\text{ is a limit ordinal} \end{align*} \end{defn} This will be a definition by \emph{transfinite recursion} on the ordinals. In this regard, we should compare this with recursive definitions of functions on the natural numbers.\footnote{Cf.\ the definitions of addition, multiplication, and exponentiation in \olref[sfr][infinite][dedekind]{sec}.} As when dealing with natural numbers, one defines a base case and successor cases; but when dealing with ordinals, we also need to describe the behaviour of \emph{limit} cases. This definition of the $V_\alpha$s will be an important milestone. We have informally motivated our hierarchy of sets as forming sets by \emph{stages}. The $V_\alpha$s are, in effect, just those stages. Importantly, though, this is an \emph{internal} characterisation of the stages. Rather than suggesting a possible \emph{model} of the theory, we will have defined the stages \emph{within} our set theory. \end{document} TS7-development/jsonrpcdocs/latex/structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json.tex \hypertarget{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}{}\doxysection{ts7\+::jsonrpc\+\_\+playground\+::json\+\_\+call\+::As\+Json$<$ T $>$ Struct Template Reference} \label{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}\index{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}} Convert to json. {\ttfamily \#include $<$t5.\+h$>$} \doxysubsection*{Public Member Functions} \begin{DoxyCompactItemize} \item constexpr \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a0b4692ef749dbd0621ba94b4a14dbf21}{As\+Json}} (const T \&\mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a77484329d622abbd87dcc7f6a35557ef}{ref}}) \begin{DoxyCompactList}\small\item\em constructor \end{DoxyCompactList}\item \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_ac4afe3de70ad247247d3063689232811}{operator boost\+::json\+::value}} () const \begin{DoxyCompactList}\small\item\em Cast operator. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection*{Public Attributes} \begin{DoxyCompactItemize} \item const T \& \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a77484329d622abbd87dcc7f6a35557ef}{ref}} \begin{DoxyCompactList}\small\item\em Stored const reference. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Detailed Description} \subsubsection*{template$<$typename T$>$\newline struct ts7\+::jsonrpc\+\_\+playground\+::json\+\_\+call\+::\+As\+Json$<$ T $>$} Convert to json. Converts any type to a json. \begin{DoxyNote}{Note} This version does not contain the static\+\_\+assert. Check \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_from_json}{From\+Json}}, to see how this can be achieved. \end{DoxyNote} \begin{DoxyTemplParams}{Template Parameters} {\em T} & The type that shall be converted to a json value.\\ \hline \end{DoxyTemplParams} \begin{DoxySince}{Since} 1.\+0 \end{DoxySince} \begin{DoxyAuthor}{Author} \href{mailto:}{\texttt{ }} \end{DoxyAuthor} \doxysubsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a0b4692ef749dbd0621ba94b4a14dbf21}\label{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a0b4692ef749dbd0621ba94b4a14dbf21}} \index{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}!AsJson@{AsJson}} \index{AsJson@{AsJson}!ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}} \doxysubsubsection{\texorpdfstring{AsJson()}{AsJson()}} {\footnotesize\ttfamily template$<$typename T $>$ \\ constexpr \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}{ts7\+::jsonrpc\+\_\+playground\+::json\+\_\+call\+::\+As\+Json}}$<$ T $>$\+::\mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}{As\+Json}} (\begin{DoxyParamCaption}\item[{const T \&}]{ref }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [inline]}, {\ttfamily [explicit]}, {\ttfamily [constexpr]}} constructor Creates an instance that stores the const reference to the provided variable. \begin{DoxyParams}{Parameters} {\em ref} & Const reference that shall be used.\\ \hline \end{DoxyParams} \begin{DoxySince}{Since} 1.\+0 \end{DoxySince} \begin{DoxyAuthor}{Author} \href{mailto:}{\texttt{ }} \end{DoxyAuthor} \doxysubsection{Member Function Documentation} \mbox{\Hypertarget{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_ac4afe3de70ad247247d3063689232811}\label{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_ac4afe3de70ad247247d3063689232811}} \index{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}!operator boost::json::value@{operator boost::json::value}} \index{operator boost::json::value@{operator boost::json::value}!ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}} \doxysubsubsection{\texorpdfstring{operator boost::json::value()}{operator boost::json::value()}} {\footnotesize\ttfamily template$<$typename T $>$ \\ \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}{ts7\+::jsonrpc\+\_\+playground\+::json\+\_\+call\+::\+As\+Json}}$<$ T $>$\+::operator boost\+::json\+::value (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption}) const} Cast operator. Casts the provided const reference to a json value. \begin{DoxyAttention}{Attention} This operator is not implemented and will cause a linker error. For any type that shall be supported a corresponding template specialization needs to be created. \end{DoxyAttention} \begin{DoxySince}{Since} 1.\+0 \end{DoxySince} \begin{DoxyAuthor}{Author} \href{mailto:}{\texttt{ }} \end{DoxyAuthor} \doxysubsection{Member Data Documentation} \mbox{\Hypertarget{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a77484329d622abbd87dcc7f6a35557ef}\label{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json_a77484329d622abbd87dcc7f6a35557ef}} \index{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}!ref@{ref}} \index{ref@{ref}!ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$@{ts7::jsonrpc\_playground::json\_call::AsJson$<$ T $>$}} \doxysubsubsection{\texorpdfstring{ref}{ref}} {\footnotesize\ttfamily template$<$typename T $>$ \\ const T\& \mbox{\hyperlink{structts7_1_1jsonrpc__playground_1_1json__call_1_1_as_json}{ts7\+::jsonrpc\+\_\+playground\+::json\+\_\+call\+::\+As\+Json}}$<$ T $>$\+::ref} Stored const reference. The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item playground/003-\/json-\/call/\mbox{\hyperlink{t5_8h}{t5.\+h}}\end{DoxyCompactItemize} \hypertarget{group___u_a_r_t___exported___functions___group2}{}\section{U\+A\+R\+T\+\_\+\+Exported\+\_\+\+Functions\+\_\+\+Group2} \label{group___u_a_r_t___exported___functions___group2}\index{U\+A\+R\+T\+\_\+\+Exported\+\_\+\+Functions\+\_\+\+Group2@{U\+A\+R\+T\+\_\+\+Exported\+\_\+\+Functions\+\_\+\+Group2}} \subsection*{Functions} \begin{DoxyCompactItemize} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Transmit} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size, uint32\+\_\+t Timeout)\hypertarget{group___u_a_r_t___exported___functions___group2_ga210329848c1873957034e129ccf8944e}{}\label{group___u_a_r_t___exported___functions___group2_ga210329848c1873957034e129ccf8944e} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Receive} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size, uint32\+\_\+t Timeout)\hypertarget{group___u_a_r_t___exported___functions___group2_gab868edc590e3b827a14528a25c999e2f}{}\label{group___u_a_r_t___exported___functions___group2_gab868edc590e3b827a14528a25c999e2f} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Transmit\+\_\+\+IT} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size)\hypertarget{group___u_a_r_t___exported___functions___group2_gaf223f2bcc2f5734f147cc5c626d757b0}{}\label{group___u_a_r_t___exported___functions___group2_gaf223f2bcc2f5734f147cc5c626d757b0} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Receive\+\_\+\+IT} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size)\hypertarget{group___u_a_r_t___exported___functions___group2_gadc0c3ef2109881d011601f0d41e70e40}{}\label{group___u_a_r_t___exported___functions___group2_gadc0c3ef2109881d011601f0d41e70e40} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Transmit\+\_\+\+D\+MA} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size)\hypertarget{group___u_a_r_t___exported___functions___group2_ga039ce4af3997f11f55c3c92d043cce77}{}\label{group___u_a_r_t___exported___functions___group2_ga039ce4af3997f11f55c3c92d043cce77} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Receive\+\_\+\+D\+MA} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart, uint8\+\_\+t $\ast$p\+Data, uint16\+\_\+t Size)\hypertarget{group___u_a_r_t___exported___functions___group2_gad674cce054e58927720cd689620ffa08}{}\label{group___u_a_r_t___exported___functions___group2_gad674cce054e58927720cd689620ffa08} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+D\+M\+A\+Pause} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_ga8a713fd976d8ef02b818ea6ff0d4e41a}{}\label{group___u_a_r_t___exported___functions___group2_ga8a713fd976d8ef02b818ea6ff0d4e41a} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+D\+M\+A\+Resume} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_gaf2b3e6004d0200857781809baa16072d}{}\label{group___u_a_r_t___exported___functions___group2_gaf2b3e6004d0200857781809baa16072d} \item \hyperlink{stm32f4xx__hal__def_8h_a63c0679d1cb8b8c684fbb0632743478f}{H\+A\+L\+\_\+\+Status\+Type\+Def} {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+D\+M\+A\+Stop} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_gab21aa06cfbaa1665b1062a803fcb4217}{}\label{group___u_a_r_t___exported___functions___group2_gab21aa06cfbaa1665b1062a803fcb4217} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+I\+R\+Q\+Handler} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_gaad01472c507ceee3c5f2274c775ff3bf}{}\label{group___u_a_r_t___exported___functions___group2_gaad01472c507ceee3c5f2274c775ff3bf} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Tx\+Cplt\+Callback} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_gabcdf9b59049eccbc87d54042f9235b1a}{}\label{group___u_a_r_t___exported___functions___group2_gabcdf9b59049eccbc87d54042f9235b1a} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Tx\+Half\+Cplt\+Callback} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_ga49b287e7de94cd0a38d333629298f7c4}{}\label{group___u_a_r_t___exported___functions___group2_ga49b287e7de94cd0a38d333629298f7c4} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Rx\+Cplt\+Callback} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_gae494a9643f29b87d6d81e5264e60e57b}{}\label{group___u_a_r_t___exported___functions___group2_gae494a9643f29b87d6d81e5264e60e57b} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Rx\+Half\+Cplt\+Callback} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_ga1884970cc493d8efba5aec28c0d526e7}{}\label{group___u_a_r_t___exported___functions___group2_ga1884970cc493d8efba5aec28c0d526e7} \item void {\bfseries H\+A\+L\+\_\+\+U\+A\+R\+T\+\_\+\+Error\+Callback} (\hyperlink{struct_u_a_r_t___handle_type_def}{U\+A\+R\+T\+\_\+\+Handle\+Type\+Def} $\ast$huart)\hypertarget{group___u_a_r_t___exported___functions___group2_ga0e0456ea96d55db31de947fb3e954f18}{}\label{group___u_a_r_t___exported___functions___group2_ga0e0456ea96d55db31de947fb3e954f18} \end{DoxyCompactItemize} \subsection{Detailed Description} 0 \hypertarget{class_main_window}{}\doxysection{Main\+Window Class Reference} \label{class_main_window}\index{MainWindow@{MainWindow}} Inheritance diagram for Main\+Window\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{class_main_window} \end{center} \end{figure} \doxysubsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_main_window_a996c5a2b6f77944776856f08ec30858d}{Main\+Window}} (QWidget $\ast$parent=nullptr) \begin{DoxyCompactList}\small\item\em Main Window\textquotesingle{}s Constuctor. \end{DoxyCompactList}\item \mbox{\hyperlink{class_main_window_ae98d00a93bc118200eeef9f9bba1dba7}{$\sim$\+Main\+Window}} () \begin{DoxyCompactList}\small\item\em A deconstructor. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{class_main_window_a996c5a2b6f77944776856f08ec30858d}\label{class_main_window_a996c5a2b6f77944776856f08ec30858d}} \index{MainWindow@{MainWindow}!MainWindow@{MainWindow}} \index{MainWindow@{MainWindow}!MainWindow@{MainWindow}} \doxysubsubsection{\texorpdfstring{MainWindow()}{MainWindow()}} {\footnotesize\ttfamily Main\+Window\+::\+Main\+Window (\begin{DoxyParamCaption}\item[{QWidget $\ast$}]{parent = {\ttfamily nullptr} }\end{DoxyParamCaption})} Main Window\textquotesingle{}s Constuctor. Sets up the UI and connects the pushbuttons to their designated functions. Also connects the Qtimer to the hourglass function and initializes the pointer to Timermode and the mode. \mbox{\Hypertarget{class_main_window_ae98d00a93bc118200eeef9f9bba1dba7}\label{class_main_window_ae98d00a93bc118200eeef9f9bba1dba7}} \index{MainWindow@{MainWindow}!````~MainWindow@{$\sim$MainWindow}} \index{````~MainWindow@{$\sim$MainWindow}!MainWindow@{MainWindow}} \doxysubsubsection{\texorpdfstring{$\sim$MainWindow()}{~MainWindow()}} {\footnotesize\ttfamily Main\+Window\+::$\sim$\+Main\+Window (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} A deconstructor. deletes dynamically allocated resources The documentation for this class was generated from the following files\+:\begin{DoxyCompactItemize} \item C\+:/\+Users/\+User/\+Documents/\+Qt/\+COP3003-\/\+Integration-\/\+Project/\mbox{\hyperlink{mainwindow_8h}{mainwindow.\+h}}\item C\+:/\+Users/\+User/\+Documents/\+Qt/\+COP3003-\/\+Integration-\/\+Project/\mbox{\hyperlink{mainwindow_8cpp}{mainwindow.\+cpp}}\end{DoxyCompactItemize} report/tex/glossary/glossary.tex % !TEX root = glossary_subfile.tex \chapter{Glossary, index \& nomenclature} \section{Glossary} The \Gls{computer} is an entry defined in the glossary. Use \texttt{glossary.tex} to define new entries. You can also define acronyms like \acrfull{fpsLabel} and refer to it with \verb|\acrlong| (\acrlong{fpsLabel}) or \verb|\acrshort| (\acrshort{fpsLabel}). \section{Index} Use the \verb|\index| macro to register a word in the index. For example, here we index the word index\index{index}. We can also create subentries\index{index!first subentry}\index{index!second subentry}\index{other entry}\index{last entry}. \section{Nomenclature} Use the \verb|\nomenclature| macro to add symbols to the nomenclature. Use the optional argument to specify a group. For instance, we can add the symbol $\Omega$\nomenclature[P]{$\Omega$}{Sample space} to the nomenclature. We can also add $\setN$\nomenclature[S]{$\setN$}{Set of the natural numbers} and $\setR$\nomenclature[S]{$\setR$}{Set of the real numbers}. You can customize the headers of the nomenclature by editing \texttt{nomencl\_header.tex}. Bibliographies/binding.bib @Incollection{abeille:depend98, author = " and and and ", title = "{French bounded Dependencies}", editor = " and ", booktitle = "{Romance in HPSG}", publisher = "CSLI Publications", address = "Stanford", year = "1998", pages = "1--54" } @Incollection{abeille:composition98, author = " and and ", title = "{Two Kinds of Composition in French complex Predicates}", editor = " and and ", booktitle = "{Complex Predicates in non transformational Syntax}", publisher = "Academic Press", address = "Cambridge MA", year = "1998", pages = "1--41" } @Incollection{jaeger:2001b, author = " ", title = "{Anaphora and Quantification in Categorial Gramar}", editor = "", booktitle = "{ Logical Aspects of Computational Linguistics}", publisher = "Springer", address = "Berlin", year = "2001", pages = "70--90" } @Booklet{jaeger:2001, author = "", title = "{Anaphora and Ellipsis in Type Logical Grammar}", howpublished = "{Course notes}", note = "Utrecht Institute of Linguistics-OTS PhD courses", address = "Utrecht", year = "2001", } @Booklet{morrill:2000, author = "", title = "{Type-Logical Anaphora}", howpublished = "{Report de Recerca LSI-00-77-R}", note = "Departament de Llenguatges i Sistemes Inform\'{a}tics, Universitat Polit\'{e}cnica de Catalunya", address = "Barcelona", year = "2000" } @Booklet{dowty:99, author = "", title = "{Anaphora and Type Logical Grammar}", howpublished = "{ms.}", note = "Stanford University Linguistics Colloquium", address = "Stanford", year = "1995" } @Incollection{moortgat:96, author = "", title = "{In Situ Binding: A Modal Analysis}", editor = " and ", booktitle = "{Proceedings of the 10th Amsterdam Colloquium}", publisher = "ILLC", address = "Amsterdam", year = "1996", pages = "539--549" } @PhdThesis{hepple:90, author = "", title = "{The Grammar and Processing of Order and Dependency: A Categorial Approach}", school = "University of Edinburgh", address = " Edinburgh", year = "1990" } @Incollection{szabol:89, author = "", title = "{Bound Variables in Syntax (are there any?)}", editor = " and and ", booktitle = "{Semantics and Contextual Expressions}", publisher = "Foris", address = "Dordrecht", year = "1989", pages = "295--318" } @Incollection{szabol:2003, author = "", title = "{Binding on the Fly: Cross-Sentential Anaphora in Variable-free Semantics}", editor = " and ", booktitle = "{Resource Sensitivity in Binding and Anaphora}", publisher = "Kluwer", address = "Dordrecht", year = "2003" } @Article{stal:context98, author = "", title = "{On the Representation of Context}", journal = "Journal of Logic, Language and Information", year = "1998", volume = "7", pages = "3--19" } @Article{jacobson:paycheck2000, author = "", title = "{Paycheck Pronouns, Bach-Peters Sentences and Variable-free Semantics}", journal = "Natural Language Semantics", year = "2000", volume = "8", pages = "77--155" } @Article{evans:pron80, author = "", title = "{Pronouns}", journal = "Linguistic Inquiry", year = "1980", volume = "2", pages = "337--362" } @Article{reinhart:bound83, author = "", title = "{Coreference and Bound Anaphora: A Restatement of the Anaphora Questions}", journal = "Linguistics and Philosophy", year = "1983", volume = "6", pages = "47--88" } @Incollection{richter:quant99, author = " and and ", title = "{A Formal Interpretation of Relations and Quantification in HPSG}", editor = ", , and ", booktitle = "{Constraints and Resources in Natural Language Syntax and Semantics}", publisher = "CSLI Publications", address = "Stanford", year = "1999", pages = "281--298" } @Article{reyle:udrt93, author = "", title = "{Dealing with Ambiguities by Underspecification: Construction, Representation and Deduction}", journal = "Journal of Semantics", year = "1993", volume = "10", pages = "123--179" } @Inproceedings{mitkov:resol97, author = "", title = "{Factors in Anaphora Resolution: They are not the only Things that Matter}", booktitle = "{Proceedings of the ACL/EACL'97 Workshop on Operational Factors in Practical, Robust Anaphora Resolution}", year = "1997", pages = "36--51" } @Article{lappin:pron94, author = " and ", title = "{An Algorithm for Pronominal Anaphora Resolution}", journal = "Computational Linguistics", year = "1994", volume = "20", pages = "535--561" } @Article{asher:resol89, author = " and ", title = "{A Computational Account of Syntactic, Semantic and Discourse Principles for Anaphora Resolution}", journal = "Journal of Semantics", year = "1989", volume = "6", pages = "309--344" } @Inproceedings{richluper:resol88, author = " and ", title = "{An Architecture for Anaphora Resolution}", booktitle = "{Proceedings of the 2nd Conference on Applied Natural Language Processing (ANLP'88)}", year = "1988", pages = "18--24" } @Inproceedings{carb:resol88, author = " and ", title = "{Anaphora Resolution: A Multi-strategy Approach}", booktitle = "{Proceedings of the 12th International Conference on Computational Linguistics (COLING'88)}", year = "1988", pages = "96--101" } @Article{zribi:pview89, author = "", title = "{Anaphor Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse}", journal = "Language", year = "1989", volume = "65", pages = "695--727" } @Book{kuno:func87, author = "", title = "{Functional Syntax}", publisher = "University of Chicago Press", year = "1987", address = "Chicago" } @PhdThesis{golde:diss99, author = "", title = "{Binding Theory and Beyond}", school = "Ohio State University", address = "Columbus", year = "1999" } @Inproceedings{erbach:profit95, author = "", title = "{Prolog with Features, Inheritance and Templates}", booktitle = "{Proceedings of the 7th European Chapter of the Association of Computational Linguistics (EACL'95)}", year = "1995", pages = "180--187" } @Book{gawron:anaph90, author = " and ", title = "{Anaphora and Quantification in Situation Semantics}", publisher = "CSLI Publications", year = "1990", address = "Stanford" } @Article{rein:refl93, author = " and ", title = "{Reflexivity}", journal = "Linguistic Inquiry", year = "1993", volume = "24", pages = "657--720" } @Article{poesio:ana98, author = " and ", title = "{A Corpus-based Investigation of Definite Description Use}", journal = "Computational Linguistics", year = "1998", volume = "24", pages = " 183--216" } @Inproceedings{xue:ziji94, author = " and and ", title = "{A New Perspective on Chinese {\it Ziji}}", booktitle = "{Proceedings of the the West Coast Conference on Formal Linguistics (WCCFL'94)}", publisher = "CSLI Publications", address = "Stanford", year = "1994" } @Book{polsag:hpsg87, author = " and ", title = "{Information-based Syntax and Semantics}", publisher = "CSLI Publications", year = "1987", address = "Stanford" } @Article{polsag:binding92, author = " and ", title = "{Anaphors in English and the Scope of Binding Theory}", journal = "Linguistic Inquiry", year = "1992", volume = "23", pages = "261--303" } @Book{polsag:hpsg94, author = " and ", title = "{Head-Driven Phrase Structure Grammar}", publisher = "The University of Chicago Press", year = "1994", address = "Chicago" } @Incollection{manningSag:1999, author = " and ", title = "{Dissociations between Argument Structure and Grammatical Relations}", editor = " and and ", booktitle = "{Lexical and {Constructional} Aspects of Linguistic Explanation}", publisher = "CSLI", address = "Stanford", year = "1999", pages = "63--78" } @Incollection{wechsler:1999, author = "", title = "{HPSG, GB, and the Balinese Bind}", editor = " and and ", booktitle = "{Lexical and {Constructional} Aspects of Linguistic Explanation}", publisher = "CSLI", address = "Stanford", year = "1999", pages = "179--196" } @Book{WKK99a-ed-duplicate-removed, address = stanford, booktitle = {Lexical and {Constructional} Aspects of Linguistic Explanation}, editor = { and and ol}, number = {1}, publisher = cslip, series = {Studies in Constraint Based Lexicalism}, title = {Lexical and {Constructional} Aspects of Linguistic Explanation}, year = {1999} } @Inproceedings{pianesi:index91, author = "", title = "{Indexing and Referential Dependencies within Binding Theory: a Computational Framework}", booktitle = "{Proceedings of the 5th Conference of the European Chapter of the Association of Computational Linguistics (EACL'91)}", year = "1991", pages = "39--44" } @Inproceedings{merlo:incrm93, author = "", title = "{For an Incremental Computation of Intra-sentential Coreference}", booktitle = "{Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI'93)}", year = "1993", pages = "1216--1221" } @Inproceedings{kaplan:uncert88, author = " and ", title = "{An Algorithm for Functional Uncertainty}", booktitle = "{Proceedings of the 12th International Conference on Computational Linguistics (COLING'88)}", year = "1988", pages = "297--302" } @Book{kamp:drt93, author = " and ", title = "{From Discourse to Logic: Introduction to Model-theoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory}", publisher = "Kluwer", year = "1993", address = "Dordrecht" } @Inproceedings{johnson:disc90, author = " and ", title = "{Discourse, Anaphora and Parsing}", booktitle = "{Proceedings of the 28th Annual Meeting of the Association of Computational Linguistics (ACL'90)}", year = "1990", pages = "197--302" } @Booklet{johnson:bind95, author = "", title = "{Constraint-based Natural Language Parsing}", howpublished = "{Course notes}", note = "7th European Summer School in Logic, Language and Information", address = "Barcelona", year = "1995", } @Inproceedings{ingria:pron89, author = " and ", title = "{A Computational Mechanism for Pronominal Reference}", booktitle = "{Proceedings of the 27th Annual Meeting of the Association of Computational Linguistics (ACL'89)}", year = "1989", pages = "262--271" } @Inproceedings{giorgi:bind90, author = " and and ", title = "{A Computational Approach to Binding Theory}", booktitle = "{Proceedings of the 13rd International Conference on Computational Linguistics (COLING'90)}", year = "1990", pages = "1--6" } @Inproceedings{frank:sem95, author = " and ", title = "{Principle Based Semantics for HPSG}", booktitle = "{Proceedings of the 7th Conference of the European Chapter of the Association for Computational Linguistics (EACL'95)}", year = "1995", pages = "9--16" } @Inproceedings{fong:index90, author = "", title = "{Free Indexation: Combinatorial Analysis and a Compositional Algorithm}", booktitle = "{Proceedings of the 28th Annual Meeting of the Association for Computational Linguistics (ACL'90)}", year = "1990", pages = "105--110" } @Book{dal:bind93, author = "", title = "{The Syntax of Anaphoric Binding}", publisher = "CSLI Publications", year = "1993", address = "Stanford" } @Inproceedings{correa:bind88, author = "", title = "{A Binding Rule for Government-binding Parsing}", booktitle = "{Proceedings of the 12th International Conference on Computational Linguistics (COLING'88)}", year = "1988", pages = "123--129" } @Article{higg:split83, author = "", title = "{Logical Form, Binding and Nominals}", journal = "Linguistic Inquiry", year = "1983", volume = "14", pages = "395--420" } @Book{chom:lect81, author = "", title = "{Lectures on Government and Binding}", publisher = "Foris", year = "1981", address = "Dordrecht" } @Book{chom:knowledge86, author = "", title = "{Knowledge of Language}", publisher = "Praeger", year = "1986", address = "New York" } @Article{barker:com90, author = " and ", title = "{A Theory of Command Relations}", journal = "Linguistics and Philosophy", year = "1990", volume = "13", pages = "1--34" } @Book{chi:dyn95, author = "", title = "{Dynamics of Meaning: Anaphora, Presupposition and the Theory of Grammar}", publisher = "The University of Chicago Press", year = "1995", address = "Chicago" } @TechReport{back:eagles96, author = " and and and and and and and and and and and and ", title = "{Final Report of the EAGLES Formalisms Working Group}", institution = "EAGLES-Expert Advisory Group on Language Engineering Standards", year = "1996", address = "Luxemburg" } @Incollection{link:isums83, author = "", title = "{The Logical Analysis of Plural and Mass Terms: A Lattice-theoretical Approach}", editor = " and and ", booktitle = "{Meaning, Use and Interpretation of Language}", publisher = "Walter de Gruyter", address = "Berlin", year = "1983", pages = "302--323" } @Inproceedings{eschen:plural89, author = " and and and ", title = "{Remarks on Plural Anaphora}", booktitle = "{Proceedings of the 27th Annual Meeting of the Association for Computational Linguistics (ACL'89)}", year = "1989", pages = "161--167" } @Inproceedings{branco:branch96, author = "Ant\'{", title = "{Branching Split Obliqueness at the Syntax-Semantics Interface}", booktitle = "{Proceedings of the 16th International Conference on Computational Linguistics (COLING'96)}", year = "1996", pages = "149--156" } @Inproceedings{ branco:duall01, author = "Ant\'{", title = "{Duality and Anaphora}", booktitle = "{Proceedings of the 13th Amsterdam Colloquium (AC'2001)}", year = "2001" } @Incollection{branco:ldrefl99-removed, author = "Ant\'{ and ", title = "{Long-distance Reflexives and the Binding Square of Opposition}", editor = " and and ", booktitle = "{Lexical and Constructional Aspects of Linguistics Explanation}", publisher = "CSLI Publications", address = "Stanford", year = "1999", optchapter = "11", pages = "163--177" } @Inproceedings{brancoMarrafa:subject97, author = "Ant\'{ and ", title = "{Subject-oriented and non Subject-oriented Long-distance Anaphora: an Integrated Approach}", editor = " and ", booktitle = "{Proceedings of the 11th Pacific Asia Conference on Language, Information and Computation (PACLIC1997)}", publisher = "Kyung Hee University", address = "Seoul", year = "1997", pages = "21--30" } @Inproceedings{branco:1998-removed-wrong-page-numbers, author = "", title = "{The Logical Structure of Binding}", booktitle = "{Proceedings of the 17th International Conference on Computational Linguistics and 36th Annual Meeting of the Association for Computational Linguistics (COLING-ACL'98)}", year = "1998", pages = "181--187" } @Inproceedings{brancoDaarc:1998, author = "", title = "{Towards a Lean Constraint-based Implementation of Binding Theory}", editor = " and ", booktitle = "{New Approaches to Discourse Anaphora: Proceedings of the Second Colloquium on Discourse, Anaphora and Anaphor Resolution}", publisher = "Lancaster University", address = "Lancaster", volume = "11", year = "1998", pages = "54--67" } @PhdThesis{branco:diss99, author = "Ant\'{", title = "{Reference Processing and its Universal Constraints}", school = "University of Lisbon", address = "Lisbon", year = "1999" } @ Book{branco:livro00, author = "Ant\'{o}", title = "{Reference Processing and its Universal Constraints}", publisher = "Publica\c{c}\~{o}es Colibri", year = "2000", address = "Lisbon" } @Inproceedings{branco:2000, author = "", title = "{Binding Classes and Search Optimisation in Anaphor Resolution}", editor = " and and and ", booktitle = "{Proceedings of the 3rd Discourse Anaphora and Anaphor Resolution Colloquium (DAARC'2000)}", publisher = "Lancaster University", address = "Lancaster", year = "2000", pages = "125--132" } @Inproceedings{brancoColing:2000, author = "", title = "{Binding Constraints as Instructions of Binding Machines}", booktitle = "{Proceedings of the 18th International Conference on Computational Linguistics (COLING'2000)}", year = "2000", pages = "104--111" } @Inproceedings{branco:2001, author = "", title = "{Duality and Anaphora}", editor = " and ", booktitle = "{Proceedings of the 13th Amsterdam Colloquium (AC'2001)}", publisher = "ILLC", address = "Amsterdam", year = "2001", pages = "49--54" } @Article{branco:2002a, author = "", title = "{Binding Machines}", journal = "Computational Linguistics", year = "2002", volume = "28", pages = "1--18" } @Inproceedings{branco:2002b-removed, author = "", title = "{Without an Index: a Lexicalist Account of Binding Theory}", editor = " and and ", booktitle = "{Proceedings of the 8th International on Head-Driven Phrase Structure Grammar Conference (HPSG'2001)}", publisher = "CSLI Publications", address = "Stanford", year = "2002", pages = "71--86" } @Inproceedings{branco:2002c, author = "", title = "{Integrating Binding Principles into Formal Grammar Processing: An HPSG Account}", editor = ", and ", booktitle = "{Proceedings of the 4th Discourse Anaphora and Anaphor Resolution Colloquium (DAARC'2002)}", publisher = "Edi\c{c}\~oes Colibri", address = "Lisbon", year = "2002", pages = "19--24" } @Inproceedings{branco:2004, author = "", title = "{Cognitive Anaphor Resolution and the Binding Principles}", editor = " and and \c{c}alves", booktitle = "{Linguagem, Cultura e Cogni\c{c}\~ao: Estudos de Lingu\'istica Cognitiva}", publisher = "Almedina", address = "Coimbra", year = "2004", pages = "629--640" } @Article{branco:2003, author = "", title = "{Anaphor Resolution: Is the Search Optimization Flawed?}", journal = "Psychologica", year = "2003", volume = "32", pages = "77--89" } @Article{branco:2005, author = "", title = "{Anaphoric Constraints and Dualities in the Semantics of Nominals}", journal = "Journal of Language Logic and Information", year = "2005", volume = "2", pages = "149--171" } @Inproceedings{brancoHpsg:2005, author = "", title = "{Reflexives: Escaping Exemption via Domain Reshuffling}", editor = "", booktitle = "{Proceedings of the 12th International Conference on Head-Driven Phrase Structure Grammar (HPSG2005)}", publisher = "CSLI Publications", address = "Stanford", year = "2005", pages = "467--481" } @Inproceedings{branco:2006, author = "", title = "{Binding Symmetries and Nominal Dualities}", editor = " and ", booktitle = "{Where Semantics Meets Pragmatics}", publisher = "Elsevier", address = "Amsterdam", series = "Current Research in the Semantics/Pragmatics Interface Series", volume = "16", year = "2006", pages = "225--236" } @PhdThesis{breden:diss96, author = "", title = "{Towards a Binding Theory for Head-driven Phrase Structure Grammar}", school = "University of Essex", address = "Essex", year = "1996" } @Incollection{koenig:equa99, author = "", title = "{Inside-Out Constraints and Description Languages for HPSG}", editor = " and and ", booktitle = "{Lexical and Constructional Aspects of Linguistics Explanation}", publisher = "CSLI Publications", address = "Stanford", year = "1999", chapter = "17", pages = "265--280" } @Inproceedings{kiss:2001, author = "", title = "{Anaphora and Exemptness: {A} Comparative Treatment of Anaphoric Binding in {German} and {English}}", editor = " and ", booktitle = "{Proceedings of the 7th International on Head-Driven Phrase Structure Grammar Conference (HPSG'2000)}", publisher = "CSLI Publications", address = "Stanford", year = "2001", pages = "182--197" } @Article{chom:bind80, author = "", title = "{On Binding}", journal = "Linguistic Inquiry", year = "1980", volume = "11", pages = "1--46" } @Book{ Jackendoff72a-u-duplicate-removed, address = {Cambridge}, author = {}, publisher = "MIT Press", title = "{Semantic Interpretation in Generative Grammar}", year = {1972} } @Incollection{huangTang:longdistance91, author = " and ", title = "{The Local Nature of the Long-distance Reflexive in Chinese}", editor = " and ", booktitle = "{Long-Distance Anaphora}", publisher = "CUP", address = "Cambridge", year = "1991", pages = "263--282" } @Book{ hellan:book88, address = {Dordrecht}, author = {}, publisher = "Foris", title = "{Anaphora in Norwegian and the Theory of Grammar}", year = {1988} } @Article{manziniWexler:parameters87, author = " and ", title = "{Parameters, Binding Theory and Learnability}", journal = "Linguistic Inquiry", year = "1987", volume = "18", pages = "413--444" } @Book{ kosterReuland:longdistance91, address = {CUP}, editor = { and }, publisher = "Cambridge", title = {Long-Distance Anaphora}, year = {1991} } @Article{iatridou:86, author = "", title = "{An Anaphor not Bound in Its Governing Category}", journal = "Linguistic Inquiry", year = "1986", volume = "17", pages = "766--772" } @Article{varlokostaHornstein:93, author = " and ", title = "{A Bound Pronoun in Modern Greek}", journal = "Natural Language and Linguistic Theory", year = "1993", volume = "11", pages = "175--195" } @Incollection{kiss:2003, author = "", title = "{The Local Nature of the Long-distance Reflexive in Chinese}", editor = " and {\"u}ller and ", booktitle = "{Arbeiten zur Reflexivierung}", publisher = "Narr", address = "T{\"u}bingen", year = "1991", pages = "157--188" } @Inproceedings{branco:esslli2000, author = "", title = "{Specification and Verification of Binding Constraints: An Integrated Account}", editor = " and and ", booktitle = "{Proceedings of Workshop on Linguistic Theory and Grammar Implementation}", publisher = "12th European Summer School in Logic, Language and Information (ESSLLI2000)", address = "Birmingham", year = "2000", pages = "201--218" } @Article{ Sag97a-duplicate-removed, author = {}, journal = {Journal of Linguistics}, number = {2}, pages = {431--484}, title = "{English Relative Clause Constructions}", volume = {33}, year = {1997} } @Book{ dimitriadisDatabase:2005, address = {University of Utrectht}, author = { and and and }, publisher = "http://languagelink.let.uu.nl/anatyp/", title = {Anaphora Typology Database}, year = {2005} } @Article{brancoNullSubject:2007, author = "", title = "{{Null Subjects are Reflexives, not Pronouns}}", journal = "Lecture Notes in Artificial Intelligence", year = "2007", volume = "4410", pages = "59--76", publisher = "Springer", address = "Berlin" } @Incollection{Lobner1987, author = "", title = "{Quantification as a Major Module of Natural Language Semantics}", editor = " and and ", booktitle = "{Studies in DRT and the Theory of Generalized Quantifiers}", publisher = "Foris", address = "Dordrecht", year = "1987", pages = "53--85" } @Article{Lobner1989, author = "", title = "{{German {\em schon-erst-noch}: An Integrated Analysis}}", journal = "Linguistics and Philosophy", year = "1989", volume = "12", pages = "167--212", } @Article{Lobner1999, author = "", title = "{{Why German {\em schon} and {\em noch} are still Duals: a Reply to van der Auwera}}", journal = "Linguistics and Philosophy", year = "1999", volume = "22", pages = "45--107" } @Incollection{terMeulen1988, author = "Alice {}", title = "{The Semantic Properties of English Aspectual Verbs}", booktitle = "{NELS}", year = "1988", volume = "21", publisher = "University of Massachusetts", address = "Amherst" } @Incollection{Konig1991, author = "", title = "{Concessive Relations as the Dual of Causal Relations}", editor = "", booktitle = "{Semantic Universals and Universal Semantics}", publisher = "Foris", address = "Dordrecht", year = "1991", pages = "190--209" } @Incollection{Smessaert1997, author = "", title = "{Aspectual Duality Regained}", editor = " and and ", booktitle = "{Proceedings of the 11th Amsterdam Colloquium}", publisher = "ILLC", address = "Amsterdam", year = "1997", pages = "271--276" } @Incollection{vanBenthem1991, author = "", title = "{Linguistic Universals in Logical Semantics}", editor = "", booktitle = "{Semantic Universals and Universal Semantics}", publisher = "Foris", address = "Dordrecht", year = "1991", pages = "17--36" } @Incollection{Karttunen1976, author = "", title = "{Discourse Referents}", editor = "", booktitle = "{Syntax and Semantics 7: Notes from the Linguistic Underground}", publisher = "Academic Press", address = "New York", year = "1976", pages = "363--385" } @Incollection{Kamp1981, author = "", title = "{A Theory of Truth and Discourse Representation}", editor = " and and ", booktitle = "{Formal Methods in the Study of Language}", publisher = "Mathematical Centre", address = "Amsterdam", year = "1981", pages = "277--322" } @PhdThesis{ Heim1982, author = {}, title = {The Semantics of Definite and Indefinite Noun Phrases}, year = {1982}, school = {University of Massachusetts}, address = {Amherst} } @Book{ Seuren1985, address = {Oxford}, author = {}, title = "{Discourse Semantics}", year = {1985}, publisher = "Blackwell" } @Incollection{seeley93, author = "", title = "{Binding Plural Pronominals}", editor = " and and and and and ", booktitle = "{CLS29: Papers from the 29th Regional Meeting of the Chicago Linguistic Society}", publisher = "University of Chicago", address = "Chicago", year = "1993", volume = "2", pages = "305--317" } sjgknight/starter-hugo-research-group @article{schuckEducationalScenariosDigital2010, author = { and Aubusson, Peter}, doi = {10.1080/17439884.2010.509351}, fulltext = {https://opus.lib.uts.edu.au/bitstream/10453/17220/1/2009006976.pdf}, journal = {Learning, media and Technology}, number = {3}, pages = {293--305}, publisher = {Taylor & Francis}, related = {https://scholar.google.com/scholar?q=related:HzWeSoyflYYJ:scholar.google.com/&scioq=author:%22aubusson+peter%22+sydney&hl=en&as\textsubscriptsdt=2007}, title = {Educational Scenarios for Digital Futures}, volume = {35}, year = {2010} } 1-10 \section{Finite Sets} \begin{definition} A set is said to be \textbf{finite} if there is a bijective correspondence of $A$ with some section of the positive integers. That is, $A$ is finite if it is empty or if there is a bijection \begin{equation} f:A \to \pbrac{1, \dots, n} \end{equation} for some positive integer $n$. In the former case, we say that $A$ as \textbf{cardinality 0}, and in the latter case, we say that $A$ has \textbf{cardinality $n$}. \end{definition} \section*{Exercises} \bx{ \ea{ \item There are $4 \times 3 \times 2 = 24$ injective mappings, \begin{align*} (1, 1), (2, 2), (3, 3)\\ (1, 1), (2, 2), (3, 4)\\ (1, 1), (2, 3), (3, 2)\\ (1, 1), (2, 3), (3, 4)\\ (1, 1), (2, 4), (3, 2)\\ (1, 1), (2, 4), (3, 3)\\ (1, 2), (2, 1), (3, 3)\\ (1, 2), (2, 1), (3, 4)\\ (1, 2), (2, 3), (3, 1)\\ (1, 2), (2, 3), (3, 4)\\ (1, 2), (2, 4), (3, 1)\\ (1, 2), (2, 4), (3, 3)\\ (1, 3), (2, 1), (3, 2)\\ (1, 3), (2, 1), (3, 4)\\ (1, 3), (2, 2), (3, 1)\\ (1, 3), (2, 2), (3, 4)\\ (1, 3), (2, 4), (3, 1)\\ (1, 3), (2, 4), (3, 2)\\ (1, 4), (2, 1), (3, 2)\\ (1, 4), (2, 1), (3, 3)\\ (1, 4), (2, 2), (3, 1)\\ (1, 4), (2, 2), (3, 3)\\ (1, 4), (2, 3), (3, 1)\\ (1, 4), (2, 3), (3, 2) \end{align*} \item $10 \cdot 9 \cdot \dots \cdot 3 = 1814400$ \dots, you can tell this is not a fun time. } } \bx{ AFSOC $A$ is finite. Then $B$ must be finite since it is a subset of $A$, but this is a contradiction since $b$ is not finite by assumption. } \bx{ Let the propr subset be $X^\omega - {0, 0, \dots}$. Let the bijection be \begin{align*} f(000000\dots) &= 1000000\dots\\ f(100000\dots) &= 0100000\dots\\ f(110000\dots) &= 0010000\dots\\ &\cdots \end{align*} It’s just binary written in R to L significant digits, + 1 to shift for the missing $000\dots$. } \bx{ \ea{ \item Induction from empty set as base case. Adding an element gives you two cases, either the new element is the largest, or the existing largest element remains the largest. \item \TODO: What is an order type? I think the idea here is map $A$ into $\mathbb{Z}$ in the order that the elements are in $A$, and then they will have the order type of the $\Zp$. } } \bx{ No. $A = \emptyset$ and $B$ could be infinite. } \bx{ \ea{ \item $X^n$ is basically an $n$-tuple showing whether or not an element of A is included in the set, so you can make that bijection to $\mathcal{P}(A)$. \item $\mathcal{P}(A)$ has a bijection with $X^n$, which is finite, so $\mathcal{P}(A)$ is finite as well. } \label{chap1:sec6:prob6} } \bx{ Consider $C = A \times B$. Any function $f$ is some subset of $C$. Then the set of all functions is a subset of $\mathcal{P}(C)$, which from \ref{chap1:sec6:prob6} we know is finite since $C$ is finite. } mike715/assignment \documentclass[11pt]{article} \usepackage{amsfonts} \usepackage{fancyvrb} \usepackage{url} \usepackage{graphicx} \usepackage{caption} \usepackage{subcaption} \setlength{\oddsidemargin}{0in} \setlength{\evensidemargin}{0in} \setlength{\textwidth}{6.5in} \setlength{\topmargin}{0in} \setlength{\headsep}{0.5in} \setlength{\textheight}{8.5in} \setcounter{page}{1} %\pagestyle{empty} %\hbadness=10000 \begin{document} \huge \noindent {Discrete Optimization Assignment:} \vspace{0.25cm} \noindent {\bf Graph Coloring} \normalsize \section{Problem Statement} In this assignment you will design an algorithm to find the smallest {\em coloring} of a graph.\footnote{See \url{http://mathworld.wolfram.com/ChromaticNumber.html}} You are provided with a graph and your task is to label the graph's nodes with as few colors as possible such that all pairs of nodes joined by an edge do not have the same color. Figure \ref{fig:graph} illustrates a graph and a three coloring of that graph. The nodes of the graph are labeled with black numbers while the coloring of the graph is labeled with white numbers. You may notice that a three coloring is not a minimal coloring of this graph. In fact, a two coloring is possible. \begin{figure}[h] \centering \begin{subfigure}[b]{8.0cm}%{0.3\textwidth} \centering \includegraphics[width=8cm]{figures/coloring_1.pdf} \caption{A graph.} \label{fig:graph:input} \end{subfigure}% ~ %add desired spacing between images, e. g. ~, \quad, \qquad etc. %(or a blank line to force the subfigure onto a new line) \begin{subfigure}[b]{8.0cm} \centering \includegraphics[width=8cm]{figures/coloring_2.pdf} \caption{A three-coloring of the graph.} \label{fig:graph:coloring} \end{subfigure} \caption{A Graph Coloring Example}\label{fig:graph} \end{figure} \section{Assignment} Write an algorithm to minimize the coloring of a graph. The problem is mathematically formulated in the following way. Given a graph $G = \langle N, E \rangle$ with nodes $N = 0 \ldots n-1$ and edges $E$, let $c_i \in \mathbb{N}$ be a variable denoting the color of node $i$. Then the graph coloring problem is formalized as the following optimization problem, $$ \begin{array}{ll} \mbox{minimize:} & \displaystyle \max_{i \in 0 \ldots n-1} c_i \\ \mbox{subject to:} & \\ & c_i \neq c_j \;\;\; (\langle i,j \rangle \in E) \end{array} $$ \section{Data Format Specification} The input consists of $|E| + 1$ lines. The first line contains two numbers $|N|$ and $|E|$. It is followed by $|E|$ lines, each line represents an edge $\langle u_i, v_j \rangle$ where $u_i, v_j \in 0 \ldots |N|-1$. \vspace{0.2cm} \noindent Input Format \vspace{-0.2cm} \begin{Verbatim}[frame=single] |N| |E| u_0 v_0 u_1 v_1 ... u_|E|-1 v_|E|-1 \end{Verbatim} % The output has two lines. The first line contains two values $obj$ and $opt$. $obj$ is the numbers of colors used in the coloring (i.e. the objective value). $opt$ should be $1$ if your algorithm proved optimality and $0$ otherwise. The next line is a list of $n$ values in $\mathbb{N}$, one for each of the $c_i$ variables. This line encodes the solution. \vspace{0.2cm} \noindent Output Format \vspace{-0.2cm} \begin{Verbatim}[frame=single] obj opt c_0 c_1 c_2 ... c_n-1 \end{Verbatim} % %It is essential that the value order in the solution output matches the value order of the input. Otherwise the grader will misinterpret the output. \paragraph{Examples} \mbox{} %\vspace{0.1cm} \noindent (based on Figure \ref{fig:graph}) \vspace{0.2cm} \noindent Input Example \vspace{-0.2cm} \begin{Verbatim}[frame=single] 4 3 0 1 1 2 1 3 \end{Verbatim} \vspace{0.2cm} \noindent Output Example \vspace{-0.2cm} \begin{Verbatim}[frame=single] 3 0 0 1 2 2 \end{Verbatim} \section{Instructions} \input{instructions.tex} %We use \texttt{stdout} for output. %Output to other stream will be ignored (you may want to send runtime information to \texttt{stderr}). Your submission will be tested on a department linux machine. If your algorithm is a standalone program, please name it \texttt{nr}, %otherwise, please specify the compilation procedure, %it is appreciated if you also provide a script that follows the above format to run the program. \paragraph{Resources} You will find several graph coloring instances in the \texttt{data} directory provided with the handout. %An example output file, \texttt{blabla.out}, is also provided. %\section{Remarks} \input{handin.tex} \input{grading.tex} \input{collaboration.tex} %\paragraph{Questions} Please contact the class GTA Carleton (). \input{warnings.tex} \paragraph{Hint} The optimal value for \texttt{data/gc\_1000\_5} is near $85$. \input{techReqs.tex} \end{document} w-yuchen/nusskylab \chapter{Background} \label{background} The implementation of Skylab faced many design choices that shaped its fundamental development. We describe and justify important major design choices here. Skylab is built using Ruby on Rails, a well-known web development framework with a great support from a large community, used widely in the industries by companies like Twitter, Groupon, Bloomberg, Airbnb and many more\cite{citation1}. There are many reasons why we have chosen Ruby on Rails. Firstly, Ruby is clean, elegant and easy to read and readability and elegance of Ruby enables programmers to be more productive. Secondly, Ruby on Rails adopts many advanced industrial conventions and this enables contributors to have good exposure to programming in the industry. What is more, scaffolding and the contribution of many libraries (gems) by the Ruby community can significantly simplify programmers' work by saving time and effort of the development team from ``reinventing the wheel''. The popularity and activeness in Ruby on Rails' community ensures that getting continuous support would be easy and many resources can also be available online. Last but not least, Ruby on Rails community has a favor for open source contribution which aligns well with the open source nature of Skylab. For the selection of database, we used PostgreSQL. Part of the reason is that it is open source and quite mature, with good drivers available in many languages\cite{citation2}. Besides, we need full ACID (atomicity, consistency, isolation and durability) compliance for consistency of data and we do not need scalability to multiple servers in the foreseeable future\cite{citationACID}. And PostgreSQL has recently added implementation for rich data structures such as JSON which eases development\cite{citation3}. Puma is the web server we have chosen for Skylab. Among common Ruby web servers such as Passenger, Unicorn, Rainbows!, Puma is considered to be fast and memory friendly according to online benchmarking\cite{citation4}. Puma is built for high-concurrency and speed and we observe that more developers are switching to Puma within the Rails community\cite{citation5}. We have selected Nginx as our web (HTTP) server. Nginx has grown in popularity since its release due to its light-weight resource utilization and its ability to scale easily with low memory usage. It excels at serving static content quickly and is designed to pass dynamic requests off to other software that is better suited for those purposes\cite{citation6}. There are also some benchmarking results that indicate the superiority in Nginx handling concurrent access and of its small in-memory footprint\cite{citation7}. A high-level overview of the architecture of Skylab is illustrated in Figure~\ref{fig:Skylabarch}. Incoming requests to server will first be forwarded to Puma worker processes by Nginx. After that the corresponding Skylab code in the Ruby on Rails framework will be executed. Where database access is required, PostgreSQL serves the data. \begin{figure}[h] \centering \includegraphics[width=0.9\textwidth]{Images/Skylab_Arch.png} \caption{Architecture of Skylab} \label{fig:Skylabarch} \end{figure} \section{System design} \subsection{MVC pattern in Ruby on Rails} The basic MVC structure of Rails is shown in Figure~\ref{fig:RailsMVC}: \begin{figure}[h] \centering \includegraphics[width=0.85\textwidth]{Images/Rails_MVC.png} \caption{Illustration of how MCV works in Rails. Excerpted from \cite{citationMVC}} \label{fig:RailsMVC} \end{figure} After request is received by the Rails framework, the \textit{router} will look at the pattern of the requested URL path and send it to the corresponding method of the target \textit{controller} class. The \textit{controller} is supposed to query \textit{models} and gather necessary information for rendering the \textit{view}. Last but not least, the response will be sent back to user for viewing. So the most fundamental component in this whole process is the \textit{model}, which stores all the business data. A proper model design can save a lot of trouble when it comes to writing controllers. \subsection{Skylab Model Overview} Figure~\ref{fig:SkylabER} (source available at \cite{citationERSource}) gives an overview of the current model design. \begin{figure}[h] \centering \includegraphics[width=\textwidth]{Images/Skylab_ER.png} \caption{Current ER diagram for Skylab} \label{fig:SkylabER} \end{figure} As is seen from the entity relation diagram shown in Figure~\ref{fig:SkylabER}, the users' basic information is captured in the \textit{User} model and each user may have one or more different roles such as \textit{Admin}, \textit{Student}, \textit{Adviser}, \textit{Mentor}, \textit{Tutor} and \textit{Facilitator}. Each user can have only one of any type of role in a cohort —-- meaning that if \textit{User} A is an \textit{Adviser} B for cohort 2015, then his/her adviser role for cohort 2015 will only be B (and of course he/she can also be a mentor, a tutor or another role type in cohort 2015). But for cohort 2016 if \textit{User} A is an adviser again, then it is another \textit{Adviser} role C. Roles like \textit{Facilitator}, \textit{Tutor} do not have practical meanings in the system but only for display in public staff page. \textit{Admin} is supposed to overlook manage everything in Skylab. The remaining roles, \textit{Adviser}, \textit{Mentors} and \textit{Student} are connected via different relationships. Each student has a \textit{Team} and a team has an \textit{Adviser} and a \textit{Mentor} (optional). A group of teams under the same adviser is called an \textit{Evaluation Group} and evaluating relationships will be assigned from team to team in the same \textit{Evaluation Group}. Each team will create \textit{Submissions} to \textit{Milestones} to report their progress and their adviser and assigned evaluator teams will submit \textit{PeerEvaluations} to the team's \textit{Submissions}. After receiving \textit{PeerEvaluation} from evaluator teams and adviser, the team need to provide \textit{Feedback} to rate the helpness of those \textit{PeerEvaluations}. Before Orbital officially starts, interested students need to register in Skylab first. A \textit{Registration} will be created to record information about the student and his/her interested topics of project in the summer for recommending potential teammates if they do not have a partner in mind at the time of registration. \textit{PeerEvaluation}, \textit{Feedback} and \textit{Registration} can be considered as \textit{Surveys}. Basically a \textit{Survey} contains some instructions and consists of questions to be answered. So in Skylab, a \textit{SurveyTemplate} would record information like instructions to students and have many \textit{Questions} to form a \textit{Survey}. Responses of \textit{Survey} would be stored in \textit{PeerEvaluation}, \textit{Feedback} and \textit{Registration}. \subsection{Overview of Controllers in Skylab} Controllers will query models to get information needed to present views or make modifications to models if requested. Therefore they are crucial in powering a Rails application and in Skylab we have well-defined controllers to carry out different tasks. A class diagram of all controllers in Skylab is shown in Figure~\ref{fig:SkylabControllers}. \begin{figure}[h] \centering \includegraphics[width=\textwidth]{Images/Skylab_Controllers.png} \caption{Current Class diagram of controllers in Skylab} \label{fig:SkylabControllers} \end{figure} \textit{ApplicationController} is the base of all other controllers. Utility methods such as access control, page title and handling of exceptions are included in \textit{ApplicationController} and it also includes auxiliary modules such as \textit{CohortHelper} to enrich its functionality. \textit{RolesController} is the base class of all roles' controller classes: \textit{AdminsController}, \textit{AdvisersController}, \textit{MentorsController}, \textit{StudentsController}, \textit{FacilitatorsController} and \textit{TutorsController}. These roles share quite a lot in common so they will share request processing methods and they will override data related methods or provide some handlers for special cases. Other controllers are as follows: \begin{itemize} \item \textbf{HomeController}: serves home page to user only. \item \textbf{UsersController}: manages listing, creation, editing and display of users and also includes functionality for admin to masquerade as any user, for new users to register as students in Orbital. \item \textbf{TagsController}: lists tags based on specified filters. \item \textbf{MilestonesController}: handles listing, creation, editing and display of milestones. \item \textbf{SurveyTemplatesController}: handles listing, creation, editing and display of survey templates and also serves as interface for editing of questions belonging to current survey templates. \item \textbf{QuestionsController}: manages Ajax requests to create, edit or delete a question. \item \textbf{SubmissionsController}: manages creation, editing and display of submissions by students. \item \textbf{PeerEvaluationsController}: handles creation, editing and display of peer evaluations from evaluator teams or advisers to evaluated teams. \item \textbf{FeedbacksController}: manages creation, editing and display of feedback from evaluated team to evaluator teams or advisers. \item \textbf{ReceivedEvalsController}: compiles a set of received peer evaluations for a team in one milestone. \item \textbf{ReceivedFeedbacksConroller}: compiles a set of received feedback for a team or an adviser in one milestone. \item \textbf{PublicProjectsController}: serves completed teams' projects to the public. \item \textbf{PublicStaffController}: serves staff of Orbital program to the public. \item \textbf{TeamsController}: manages listing, creation, editing and display of teams. \item \textbf{EvaluatingsController}: handles listing, creation, editing and display of evaluating relation among teams. \end{itemize} Controllers manage a collection of \textit{resources} in Skylab, meaning that each controller is managing one model class, to facilitate higher cohesion. If there is a necessary dependency on other models --- for example, SurveyTemplate will manage questions belonging to current SurveyTemplate --- the requests will be handled by Ajax requests to separate concerns. Skylab follows the Ruby on Rails convention for controller classes. A basic controller that manages a collection has the following structure illustrated in Figure~\ref{fig:SkylabSampleController}. \begin{figure}[h] \centering \includegraphics[width=0.5\textwidth]{Images/Skylab_Sample_Controller.png} \caption{Sample controller class in Skylab} \label{fig:SkylabSampleController} \end{figure} These methods —-- \textit{index}, \textit{new}, \textit{create}, \textit{edit}, \textit{update}, \textit{show} and \textit{destroy} —-- have included all the basic actions to be done with a collection of resources. Following this convention makes controller class design in Skylab consistent and easy to read, and also aligns well with the goal of creating a more human-readable RESTful URL design. \section{Development / source code management process} GitHub flow is used in Skylab's development process. It is simple but agile enough for project management. Skylab is a web application for which releases of new versions do not require any user action. Each release can be deployed to server without users noticing it. This flow is quite different from traditional Git flow, which is far more complex and puts a lot of focus on each release\cite{citation8}. In fact, in GitHub flow, each commit on master branch should be releasable and continuous deployment is recommended\cite{citation8}. Each feature will be developed on a feature branch and once the feature is ready, a pull request is created against master. After code review and running test suite, pull requests are merged into the master branch and is then ready for deployment (Figure~\ref{fig:GithubFlow}). \begin{figure}[h] \centering \includegraphics[width=0.85\textwidth]{Images/Github_Flow_Branching_Model.png} \caption{Branching diagram for GitHub Flow. Excerpted from \cite{citation8}} \label{fig:GithubFlow} \end{figure} By using GitHub flow and services such as Travis as the continuous integration service\cite{citationtravis}, CodeClimate as the code quality monitoring service\cite{citationcodeclimate}, we keep the development of Skylab agile and fast. The use of GitHub flow enables Skylab to deployed regularly, which is essential for timely improvement of user experience\cite{citation9}. 1-10 \hypertarget{structbt_typed_constraint_data}{ \section{btTypedConstraintData Struct Reference} \label{structbt_typed_constraint_data}\index{btTypedConstraintData@{btTypedConstraintData}} } this structure is not used, except for loading pre-2.82 .bullet files {\tt \#include $<$btTypedConstraint.h$>$} \subsection{Detailed Description} this structure is not used, except for loading pre-2.82 .bullet files Definition at line 393 of file btTypedConstraint.h. The documentation for this struct was generated from the following file:\begin{CompactItemize} \item C:/Users/New/Documents/Games\_\-Technology/Year4\_\-Semester1/ICT397/$\sim$My Work/Assignment2/ICT397Carre/CarreGameEngine/Dependencies/BulletPhysicsEngine/include/BulletDynamics/ConstraintSolver/btTypedConstraint.h\end{CompactItemize} 0 \newglossaryentry{Echtzeit}{name=Echtzeit, description={Bereitstellen/Anzeigen von Daten mit einer durch die Verarbeitung bedingten Verzögerung von bis zu ca. 2 Sekunden zwischen dem Anfallen der (Roh-)Daten und der Ausgabe bzw. Visualisierung.}} \newglossaryentry{Vorgang}{name=Vorgang, description={Als Vorgang wird in diesem Pflichtenheft bezeichnet, wenn ein Modus ausgewählt ist und Start gedrückt wurde. Der Vorgang endet mit dem Drücken von Stopp bzw. wird mit dem Wechseln des Modus.}} \newglossaryentry{BLE}{name=BLE, description={Bluetooth Low Energy ist eine Technologie, die Teil des Industriestandards Bluetooth ist und eine energiesparende, kabellose Kommunikation zwischen Geräten in einer Entfernung von bis zu ca. 10 Metern ermöglicht.}} \newglossaryentry{Wearable Computer}{name=Wearable Computer, description={Unter dem Begriff Wearable Computer versteht man Computersysteme, die am Körper, unter der Kleidung oder als Implantat unter der Haut getragen werden können.}} \newglossaryentry{IMU}{name=6-Achsen IMU, description={Ein 6-Achsen IMU ist ein Beschleunigungssensor mit Gyroskop.}} \newglossaryentry{GUI}{name=GUI, description={GUI ist die Abkürzung für den englischen Begriff \glqq graphical user interface\grqq . Sie ist die Schnittstelle zwischen Mensch und Maschine und ermöglicht dem Nutzer die Eingabe/Steuerung, der Maschine.}} \newglossaryentry{CPB}{name=Cross-Platform Bibliothek, description={Eine Cross-Platform Bibliothek ist nichts weiter als eine Bibliothek die auf Rechnersystemen mit verschiedener Architektur laufen kann.}} \newglossaryentry{Steuerungsparameter}{name=Steuerungsparameter, description={Unter Steuerungsparametern fassen wir die Länge der Verbindungsintervalle zwischen Earables und Smartphone sowie die Abtastrate und den Wertebereich von integriertem Gyroskop und Beschleunigungssensor zusammen.}} \newglossaryentry{Schrittfrequenz}{name=Schrittfrequenz, description={Die Schrittfrequenz gibt an wie viele Schritte pro Zeiteinheit gemacht werden.}} \newglossaryentry{Rohdaten}{name=Rohdaten, description={Als Rohdaten werden unverarbeitete Daten bezeichnet.}} \newglossaryentry{TTS}{name=Text-To-Speech, description={Ein Text-to-Speech-System (TTS) (oder Vorleseautomat) wandelt Fließtext in eine akustische Sprachausgabe um. Dabei erfolgt diese auf Deutsch oder auf Englisch, abhängig davon, was als Sprache eingestellt ist.}} \newglossaryentry{Earables}{name=Earables, description={Eine Zusammenschließung des Wortes Wearable und Earphone. Dabei handelt es sich um Kopfhörer, die mit Sensoren ausgestattet sind.}} \newglossaryentry{Vorgangsdaten}{name=Vorgangsdaten, description={Daten, die bei der Ausführung eines Vorgangs gespeichert werden (z.B. Schritte, Sit-ups,\dots).}} \newglossaryentry{Datenbank}{name={Datenbank}, description={Die Datenbank speichert die Trainingsdaten in Form von DBEntries. Dabei wird die Datenbank von dem Plugin SQLite benutzt.}} \newglossaryentry{SQLite}{name={SQLite-net-pcl}, description={SQLite-net-pcl basiert auf dem PlugIn SQLite und ist eine portable Version für die Arbeit mit einer SQLite Datenbank. Dabei ist SQLite eine Erweiterung, mit der man lokale Datenbanken ansprechen und verwalten kann. Dabei bietet die Erweiterung eine lokale Datenbank plattformunabhängig gleich anzusprechen. Diese Erweiterung wird für die Speicherung der Trainingsdaten benutz}} \newglossaryentry{CSV}{name={CSV}, description={CSV steht für "Comma-Seperated-Value" ein Dateityp, bei dem die Attribute durch ein Komma getrennt werden. Eine Zeile bildet dabei immer einen Eintrag ab. Das Format ist human-readable und veränderbar.}} \newglossaryentry{Rg.Plugins.Popup}{name={Rg.Plugins.Popup}, description={Rg.Plugins.Popup ist eine Erweiterung (Plugin), welche es einem ermöglicht Pop-up Fenster zu erstellen und anzuzeigen. Dabei ist das Plugin über den Pluginmanager 'NuGet' einbindbar.}} \newglossaryentry{DependencyInjection}{name={Microsoft.Extensions.DependencyInjection}, description={Microsoft.Extensions.DependencyInjection ist eine Erweiterung, welches ermöglicht Komponenten als Services zu registrieren und sie somit abgekapselt und modularer zu betrachten. Mit Klassen, wie der ServiceCollection, oder dem ServiceProvider, bietet die Erweiterung Microsoft.Extensions.DependencyInjection das Framework.}} \newglossaryentry{NuGet}{name={NuGet},description={NuGet ist ein Plugin-manager für Visual Studio. Dieser ermöglicht es einem einfach neue Erweiterungen dem Projekt hinzuzufügen. Er kümmert sich um die Kompatibilität und bietet ein Verzeichnis der unterschiedlichen Plug-ins.}} \newglossaryentry{Bluetooth LE Plugin}{name={Bluetooth LE Plugin for Xamarin},description={Diese Erweiterung ermöglicht es die Verbindung mit den \Gls{Earables} herzustellen. Zudem wird die Kommunikation zwischen Device und \Gls{Earables} geregelt. Von dieser Erweiterung erhalten wir die \Gls{IMU}-Daten.}} \newglossaryentry{Xamarin.Essentials}{name={Xamarin.Essentials},description={Die Erweiterung Xamarin.Essentials ermöglicht einem, Devicespezifische Services anzusprechen. Benutzt wird hiervon die Funktion TextToSpeech. }} \newglossaryentry{Xamarin.Plugin.Filepicker}{name={Xamarin.Plugin.Filepicker},description={Die Erweiterung Xamarin.Plugin.Filepicker ergänzt die Funktionalität eines Dateiauswähler. Mit diesem kann man in den Dateimanager des Devices und Dateien auswählen. Gebraucht wird dies beim Importieren und Exportieren der Trainingsdaten.}} jorgeluis8ar/Revised-reproduction-package-for-Abebe-et-al-20210 \begin{tabular}{lccccc} \hline \multicolumn{1}{l}{\emph{Outcome}} & \multicolumn{1}{c}{Transport} &\multicolumn{1}{c}{Job App. Workshop} & \multicolumn{1}{c}{Control Mean} & \multicolumn{1}{c}{F} & \multicolumn{1}{c}{N} \\ \hline \\ Life satisfcation (0 - 10) & 0.164 & 0.147 & 4.676 & 0.901 & 2503 \\ & (.132) & (.134) & & & \\ & [1] & [1] & & & \\ & & & & & \\ How much freedom \& control do you feel you have over your life (0-10)? & 0.0150 & -0.0400 & 6.114 & 0.853 & 2505 \\ & (.299) & (.285) & & & \\ & [1] & [1] & & & \\ & & & & & \\ Onenness with society (1-7)$^{a}$ & -0.0260 & 0.0530 & 4.694 & 0.554 & 2505 \\ & (.14) & (.14) & & & \\ & [1] & [1] & & & \\ & & & & & \\ How much do you trust others in this country? (1-4) & 0.0790 & 0.0400 & 2.048 & 0.655 & 2504 \\ & (.081) & (.092) & & & \\ & [1] & [1] & & & \\ & & & & & \\ \hline \end{tabular} @article{ReWa13WC, author = { and }, journal = {IEEE Wireless Communications}, number = {3}, pages = {128-135}, title = {Opportunistic Spectrum Access: From Stochastic Channels to Non-stochastic Channels}, volume = {20}, year = {2013} } estado_del_arte.tex \chapter{Estado del arte} \label{chap:Estado del arte} \drop{E}{n} en capítulo que se presenta a continuación, se definirán primero los conceptos teóricos en los que se basa el desarrollo de projecto, así como se introducen los principios de algunas de las herramientas y lenguajes de programación que han sido utilizados. Los primeros tres puntos desarrollan los conceptos de ingeniería inversa, transformación de modelos y la modernización del software, son básicos para entender el planteamiento que se ha llevado a cabo para cumplir los objetivos propuestos en el capítulo siguiente. Además en la sección Powerbuilder y Powerscript se introduce el lenguaje usado en la aplicación a la que aplicaremos las técnicas explicadas. En las siguientes secciones se numeraran las diferencias y mejoras mas significativas de las versiones actuales de las herramientas utilizadas durante el desarrollo llevado a cabo. \section{Ingenieria inversa} \label{sec:Ingeniería inversa} De acuerdo con la definición de ingeniería inversa propuesta por Chikofsky y Cross en el artículo \cite{Chikofsky1990}, establecemos que en el contexto de un proceso de ingeniería basado en modelos, el resultado de este análisis es un modelo. Para llegar a este modelo, se proponen como posibles soluciones, el análisis estático del código fuente o el seguimiento de la ejecución de la aplicación mediante herramientas diseñadas a tal efecto. En el caso de no disponer del código fuente por no estar disponible, el proceso de ingeniería debe también llevar a cabo la descompilación o desensamblado de los archivos binarios del sistema, para obtener un códido válido para la tarea que se intenta realizar. \section{Transformación de modelos} \label{sec:Transformación de modelos} Una transformación basada en modelos toma como entrada un sistema, que está basado en determinado meta-modelo y generar como resultado otro modelo que se determina por un metamodelo distinto. Existen diferentes aproximaciones a la trasformación basada en modelos como la manipulación directa, relacional, operacional, guiada por estructuras, etc [czarnecki and helsen 2003]. Todas estas aproximaciones implementan actividades ampliamente utilizadas dentro del desarrollo del software tales como son la refactorización, ingeniería inversa y el uso de patrones. \section{Modernización del software} \label{sec:Modernización del software} La modernización del software es un proceso de apliación de técnicas de reingeniería a una aplicación obsoleta, para conseguir que cumpla una serie de nuevos requerimientos e incrementar la calidad del sistema. El proceso de modernización de un sistema puede ser visualizado como una herradura, donde el lado izquierdo se compone de la extracción de información e ingeniería inversa, el lado derecho el proceso de desarrollo e ingeniería y la conexión entre ambas partes apliación de la transformación al sistema antiguo para llegar al sistema objetivo.\cite{Kazman et al. 1998] \begin{figure}[!h] \begin{center} \includegraphics[width=0.8\textwidth]{/horseshoe.jpg} \caption{Diagrama del proceso de modernización de software} \label{fig:informatica} \end{center} \end{figure} \section{PowerBuilder y Powerscript}\label{sec:PowerBuilder y Powerscript} PowerBuilder es un entorno completo para el desarrollo de aplicaciones de negocio, propiedad de la empresa Sysbase. Este sistema está compuesto por frameworks de desarrollo, herramientas de conexión, gestión y tratamiento de bases de datos, así como de un lenguaje propio de programación PowerScript®. Actualmente la versión de trabajo para el entorno de desarrollo y la aplicación a estudio es la 11.5., mientras que la versión más actual del software de PowerBuilder® es la 12.5. Para la creación de aplicaciones con PowerBuilder se dispone de un IDE de desarrollo propio. Este IDE permite realizar la implementación de programas de un modo visual, abstrayendo al desarrollador de la codificación de los comandos. Powerscript es el usado para especificar el comportamiento de la aplicación en respuesta a eventos del sistema o del usuario, tal como cerrar una ventana o presionar un botón. Las aplicaciones desarrolladas con PowerBuilder® se ejecutan exclusivamente en el sistema operativo Microsoft Windows®. \subsubsection{Objectos PowerBuilder} Un programa Powerbuilder® está construido como una colección de objetos proveidos por el propio entorno de desarrollo y los objetos hijos que el desarrollador crea mediante su extensión. \begin{table}[hp] \centering \resizebox{\textwidth}{!} {\input{tables/PowerscriptObjects.tex} } \caption[Semánticas de \acs{RPC} en presencia de distintos fallos] {Semánticas de \acs{RPC} en presencia de distintos fallos (\textsc{Puder}~\cite{puder05:_distr_system_archit})} \label{tab:rpc-semantics} \end{table} \paragraph{Objetos de tipo Aplicación} Estos objetos representan el punto de entrada a una aplicación. Se trata de un objeto que lista el conjunto de librerias que conforman la aplicación en su totalidad además de identificarla. Al igual que el resto de objetos (DataWindow, Estructura, etc) se guarda en una librería (archivo PBL) que generará el compilador. \paragraph{Objetos de tipo Ventanas} Las ventanas son la interfaz principal de comunicación entre el usuario y la aplicación PowerBuilder. En ellas se muestra información, y se recupera de la entrada que el usuario utilice, ya sea respondiendo a eventos lanzados por clicks de ratón o texto introducido por teclado. Una ventana consiste en: \begin{itemize} \item Propiedades de definición de la apariencia y comportamiento, tales como nombre de la barra de titulo, botón de minimizado/maximizado, tamaño, etc. \item Eventos lanzados por las acciones del usuario. \item Controles establecidos en la ventana. \end{itemize} \paragraph{Objectos DataWindow} Un DataWindow es un objeto que el desarrollador utiliza para recuperar y manipular datos desde una fuente externa. Así pues este tipo de objetos se comunicarán utilizando sentencias del lenguaje SQL con bases de datos o hojas de cálculo Microsoft Excel. Normamente se utilizaran integrados dentro de objetos ventana para mostrar los datos recuperados al realizar una acción de esta. \paragraph{Objetos de tipo menú} Los objetos de tipo menú contienen listas de items que el usuario puede seleccionar desde la barra de menú de la ventana activa. Normalmente se trata de agrupaciones de elementos relacionados, y cada uno de ellos permite al usuario lanzar una orden al sistema como puede ser la apertura de una ventana, la ejecución de un proceso o la edición del estilo de un campo de texto. \paragraph{Objetos de funciones} PowerBuilder permite al desarrollador definir dos tipos de clases de funciones: \begin{itemize} \item Funciones a nivel de objetos definidas para un menu o ventana particular. A su vez se pueden subdividir en funciones de sistema (disponibles siempre para objetos de una cierta clase) y funciones definidas por el usuario. \item Funciones globales que no están asociadas a un objeto en particular y que se encuentran ubicadas en un objeto independiente. Al contrario que las funciones a nivel de objetos realizan procesos de propósito general y puden ser utilizadas en cualquier tipo de objeto. Un ejemplo de este tipo de objetos serían funciones de cálculos matemáticos o manejo de cadenas. \end{itemize} \paragraph{Objetos de estructuras} Una estructura es una colección de uno o mas variables relacionadas del mismo tipo o de diferentes tipos, que se encuentran definidas bajo un único nombre que las identifica. Por ejemplo, una estructura llamada s-user-struct que contiene las variables que identifican al usuario: Identificador, dirección, nombre, una imágen, etc. Al igual que en los objetos de funciones disponemos de dos tipos de estructuras: \begin{itemize} \item Estructuras a nivel de objeto que se asocian a un determinado objeto tal como una ventana o menu. Estas estructuras se utilizaran en los scripts definidos para el propio objeto que lo contiene. \item Estructuras globales no asociadas a un objeto determinado y que pueden ser declaradas para su usuo en cualquier script de la aplicación. \end{itemize} \paragraph{Objectos definidos por el usuario} Normalmente las aplicaciones tienen características en común. Un ejemplo claro de ello son los botones existentes en la mayoría de ventanas que permiten al usuario cerrar, minimizar o maximizar un objeto. Al identificar este tipo de agrupación de se puede crear un objeto propio definido por el usuario una única vez y utilizarlo en los puntos de la apliación que lo necesiten. Los objetos definidos por el usuario pueden agruparse en objetos de usuario estándarizados y objetos de usuario específicos. La división establece aquellos que pueden ser exportados a otras aplicaciones PowerBuilder y los que son específicos para una en concreto. Además dentro de ellos podemos diferenciarlos entre los objetos que implican elementos visuales como pueden ser agrupaciones de botones, y los que contienen componentes no visuales como por ejemplo agrupaciones de funciones de cálculo que representan reglas de negocio y que pueden heredar eventos y propiedades de objetos definidos por el sistema. \paragraph{Proyectos y librerías} Los ficheros de extensión PBL y PBT definen estructuralmente la forma en la que el compilador generará la aplicación ejecutable resultante. Así pues, el fichero PBT contiene las relaciones establecidas entre las distintas librerias a generar y el nombre del producto final. Por otro lado los ficheros PBL definen las librerias resultantes de la compilación de objetos relacionados, que pueden encontrarse en el mismo directorio o en directorios separados. \cite{PowerBuilder} \section{KDM} \label{sec:KDM} En Junio de 2003, OMG\cite{OMG} creó un equipo de trabajo para modelar artefactos software en el contexto de sistemas obsoletos. Incialmente el grupo fue llamado \textit{Equipo de trabajo para la transformación de sistemas obsoletos}, \footnote{traducción de «Legacy Transformation Task Force»} aunque pronto se les renombró a \textit{Equipo de trabajo para la modernización enfocada en la arquitectura} \footnote{traducción de «Architecture-Driven Modernization Task Force»} En Noviembre de 2003 la \textit{ADMTF}\footnote{Siglas en inglés del nombre del equipo de trabajo} incorporó la solicitud de propuesta para la especificación \textit{Knowledge Discovery Metamodel}(KDM). La solicitud de propuesta establecía que el estándar del metamodelo KDM debía: \begin{itemize} \item Representar los artefactos de los sistemas obsoletos como entidades, relaciones y atributos. \item Incluir los artefactos externos con los que el interactuen los artefactos del software. \item Soportar diversos lenguajes y plataformas. \item Consistir en un núcleo independiente del lenguaje y la plataforma que pueda extenderse en caso necesario. \item Definir una terminología unificada para los artefactos de software obsoleto. \item Describir las estructuras lógicas y físicas de los sistemas obsoletos. \item La posibilidad de realizar agregaciones o modificaciones de la estructura física del sistema. \item Facilitar la identificación y trazabilidad de los artefactos desde la estructura lógica hacia la física. \item Representar el comportamiento de los artefactos hacia abajo, pero no por debajo, del nivel procedural. \end{itemize} \cite{KDM} \subsection{El metamodelo KDM}\label{sec:KDM-metamodel} El metamodelo KDM se divide en varias capas que representan tanto los artefactos físicos como los lógicos de un sistema obsoleto. Mas allá cada capa de abstracción diferente ssepara el conocimiento sobre el sistema obsoleto en diversos estructuras de sofware conocidas como \textit{vistas de la arquitectura}\footnote{Traducción de «Architecture views»}. Las cuatro capas definidas en el estándar se describen a continuación: \subsubsection{Capa de infraestructura} La capa de infraestructura define el nivel mas bajo de las capas de abstracción y contiene una pequeña lista de conceptos utilizados a traves de toda la especificación. \paragraph{Core} Define las abstracciones básicas de KDM, que son \textit{KDMEntity} y \textit{KDMRelationship} \paragraph{KDM} Proporciona el contexto compartido por todos los modelos KDM. Este paquete define los elementos que constituyen el «framework» de cada representación KDM. Por ejemplo, cada representación KDM consiste en uno o mas elementos de tipo \textit{Segment} que contienen diversos modelos de KDM. \paragraph{Source} Define el conjunto de artefactos físicos del sistema de información heredado y permite referenciar partes del código fuente. Para ello se genera el \textit{Inventory Model}, que enumera todos los artefactos físicos del sistema obsoleto (como ficheros de código, imágenes, ficheros de configuración, etc). Además este artefacto es la base que se utilizará para referenciar a los artefactos físicos desde los modelos KDM. Mas aún, los elementos del \textit{Inventory Model} permiten identificar mediante el uso de \textit{AbstractInventoryRelationships} relaciones de dependencia definidas en el sistema heredado del tipo \textit{DependsOn} que declaran la interrelación de elementos durante pasos del proceso de reingeniería. \end{itemize} \subsubsection{Capa de elementos del programa} Proporciona una representación intermedia, independiente del lenguaje de programación, para representar los constructores comunes a varios lenguajes de programación. Los dos paquetes que lo forman son: \paragraph{Code} Se trata de un paquete que define un conjunto de \textit{CodeItems} que representan elementos comunes presentes en diversos lenguajes como pueden ser; métodos, clases, tipos de datos, funciones o intefaces. Los elementos \textit{CodeItem} se especializan en 3 tipos base: \begin{itemize} \item \textbf{Module}: una unidad de programa discreta e identificable que contiene otros elementos y puede ser utilizada como componente lógico del software. Y a su vez se divide en \textit{Package}, \textit{CompilationUnit} ,\textit{CodeAssembly}, etc. \item \textbf{ComputationalObject}: representación de métodos, funciones, etc. \item \textbf{DataType}: que definen items nombrables del sistema obsoleto heredado como variables, parámetros de funciones, etc. \end{itemize} \paragraph{Action} Define las acciones llevadas a cabo por los elementos del paquete code. Los elementos de ambos paquetes se representan dentro de un modelo de código \textit{CodeModel}. \subsubsection{Capa de recursos} Permite representar conocimiento sobre el entorno y los recursos de ejecución utilizados por los sistemas de información heredados. Dispone de cuatro paquetes: \begin{itemize} \item \textbf{Data} Define los aspectos de los datos. \item \textbf{Event} Define el modelo de eventos, condiciones y acciones del sistema de información heredado. \item \textbf{UI} Define los aspectos de la interfaz de usuario del sistema de información heredado. \item \textbf{Platform} Define las características de la plataforma de ejecución. \end{itemize} \subsubsection{Capa de Abstracción} Permite representar el conocimiento específico de dominio a la vez que da una visión de negocio de los sistemas de información heredados. Dispone de tres paquetes. \begin{itemize} \item \textbf{Conceptual} Define los elementos específicos de dominio del sistema de información heredado. \item \textbf{Structure} Define los componentes estructurales de los sistema de información heredados, es decir, los subsistemas, capas, paquetes, etc. \item \textbf{Build} Define los artefactos finales relativos al sistema de información heredado. \end{itemize} \section{ANTLRv4} \label{sec:ANTLRv4} Un parser es un programa, normalmente parte de un compilador, que recibe entradas de forma secuencial como instrucciones de un fichero de código, etiquetas de marcado, o cualquier tipo de texto y lo divide en trozos que pueden ser utilizados por otro programa. ANTLR es un generador de parsers que permite crear automaticamente árboles de representación de las estructuras que cumplen las gramáticas que el usuario diseña. También es capaz de generar objetos para recorrer el árbol generado nodo a nodo y a traves de ellos realizar tareas programadas. En su desarrollo se ha utilizado el lenguaje de programación Java, y su utilización está ampliamente extendida, destacando por su popularidad entre el resto de competidores. \begin{figure}[!h] \begin{center} \includegraphics[width=0.8\textwidth]{/antlr-tree.png} \caption{Vista de un árbol generado mediante ANTLR4} \label{fig:arbol-antlr} \end{center} \end{figure} \paragraph{Mejoras de ANTLRv4 sobre sus versiones anteriores} La versión 4 de ANTLR tiene una serie de nuevas capacidades que permiten reducir la curva de aprendizaje y hace el desarrollo de gramaticas y aplicaciones de reconocimiento de lenguajes mucho mas sencillas: \begin{itemize} \item La principal mejora de esta versión es la aceptación por defecto de toda gramática, con la excepción de la recursión indirecta a izquierda. ANTLRv4 no genera conflictos con la gramática o lanza avisos relacionados con la ambiguedad de las gramáticas. \item Esta última versión utiliza una nueva tecnología llamada \textit{Adaptative LL(*) o ALL(*)}. Por ella se realiza un análisis dinámico en tiempo de ejecución en vez de el estático realizado en versiones anteriores, antes de la ejecución del parser generado. Como los parsers \textit{ALL(*)} tienen acceso a las secuencias que se le introducen, pueden directamente encontrar la forma de reconocer las secuencias hilando la gramática. Hasta ahora, por el contrario, el análisis estático, tenía que considerar todas las posibles sentencias de entrada que pudieran existir. \item ANTLR v4 automáticamente reescribe las reglas con recursión a izquierda en sus equivalentes sin ella, donde las reglas se referencian inmediatamente así mismas. La única restricción que se pone es que existe es que las reglas no pueden referencias a otra regla en la parte izquierda de una alternativa que pueda volver a referenciar a la regla inicial sin emparejarse a un token. \begin{listing}[ float=ht, language = ANTLRv4, caption = {ALL(*)}, label = code:all*] expr : expr '*' expr | expr '+' expr | INT ; \end{listing} \item El mayor cambio de la nueva versión v4 es que se desenfatiza la inclusión de código embebido dentro de la gramática, en favor de objetos \textit{listener y visitor}. Este nuevo mecanismo de desarrollo permite desacoplar la gramática del código. Sin acciones embebidas, la reutilización de gramáticas es mucho mas sencilla sin ni siquiera tener que recompilar los parsers generados. \end{itemize} \cite{Parr2012} \section{Java 1.8 y Lambdas} \section{MoDisco} \label{sec:MoDisco} MoDisco es un projecto «open source» que forma parte de manera oficial de la \textit{Eclipse Foundation}(EF)\cite{EclipseFoundation} y está integrado en el projecto base de modelado de dicha fundación, promocionando las técnicas de \textit{Ingeniería dirigida a modelos}(MDE) \footnote{Traducción de \textit{Model Driven Engineering}} dentro de la comunidad de la comunidad de Eclipse. Además está reconocida por la OMG como proveedor de referencia para la implementación de diversos estándares como: \begin{itemize} \item \textit{Knowledge Discovery Metamodel} (KDM) \item \textit{Software Measurement Metamodel} (SMM) \item \textit{Generic Abstract Syntax Tree Metamodel} (GASTM) \end{itemize} MoDisco provee de una serie de componentes que permiten eleborar soluciones de ingeniería inversa para la transformación, con idependencia del lenguaje en el que esté desarrollado, de sistemas obsoletos utilizando metamodelos. De manera nativa se ofrecen soluciones para Java, pero gracias a la API que proporciona se puede representar cualquier otro lenguaje. El soporte al proceso de reingeniería comienza con la especificación de los metamodelos, cuyo detalle puede variar en funcion de la tecnología con la que se trabaje. Con objeto de obtener el modelo, se han de usar los llamados \textindent{Discoverers}. Todos los \textindent{Discoverers} pueden ser acoplados al «framework», y usados a traves de él, únicamente con su registro en el \textit{Discoverer Manager}. \begin{figure}[!h] \begin{center} \includegraphics[width=0.8\textwidth]{/MoDisco.png} \caption{Estructura MoDisco} \label{fig:3phase-steps} \end{center} \end{figure} De manera similar a los \textit{Discoverers}, MoDisco proporciona también \textit{Generators} que se corresponden con los metamodelos soportados. Estos \textit{Generators} pueden igualmente ser acoplados al «framework» y habilitan la transformación de los modelos generados en otros tipos de artefactos. Por ejemplo, para Java se disponen de \textit{Generators} que permiten transformar los sistemas en estudio en estándares como KDM y su representación estándar basada en XMI (XML Metadata Interchange) o UML (Unified Modeling Language). \section{Aproximación a la transformación eficiente de lenguajes de programación a KDM en 3 pasos} \label{sec:3stepsapproach} La propuesta de aproximación a la transformación de sistemas heredados obsoletos en 3 pasos o fases hecha por \cite{Wulf2012}, define una estructuración del proceso que permite la generación de un modelo en KDM desde el sistema inicial mediante los siguientes pasos: \begin{itemize} \item \textbf{Tranformación de tipos internos}: En esta primera fase el sistema genera un fichero XMI con los artefactos KDM \textit{Segment} que a su vez contienen el \textit{InventoryModel} y los \textit{CodeModel} de cada uno de los elementos contenedores del código fuente del sistema. \item \textbf{Transformación de miembros internos y métodos}: La segunda fase utiliza el artefacto base \textit{Segment} generado en el paso anterior, y lo amplía añadiendo los \textit{CodeElments} que representan tanto a los miembros instanciados dentro de cada contenedor de código, como los métodos de los mismos. Además recupera las relaciones entre \textit{CodeElements} y añade el \textit{LanguageUnit} de cada uno. \item \textbf{Transformación de sentencias}: La fase final del proceso es la encargada de mapear las sentencias del código y de generar los \textit{ActionElements} definidos en él. \end{itemize} \begin{figure}[!h] \begin{center} \includegraphics[width=1\textwidth]{/3phase.png} \caption{Esquema de los pasos llevados a cabo en la transformación} \label{fig:3phase-steps} \end{center} \end{figure} @article{MATHIMALAR201533, title = "Characterization of Neutron Transmutation Doped (NTD) Ge for low temperature sensor development", journal = "Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms", volume = "345", pages = "33 - 36", year = "2015", issn = "0168-583X", doi = "https://doi.org/10.1016/j.nimb.2014.12.020", url = "http://www.sciencedirect.com/science/article/pii/S0168583X14010362", author = " and and and and and and and and and and and and and ", keywords = "NTD Ge, PALS, Channeling", abstract = "Development of NTD Ge sensors has been initiated for low temperature (mK) thermometry in The India-based TIN detector (TIN.TIN). NTD Ge sensors are prepared by thermal neutron irradiation of device grade Ge samples at Dhruva reactor, BARC, Mumbai. Detailed measurements have been carried out in irradiated samples for estimating the carrier concentration and fast neutron induced defects. The Positron Annihilation Lifetime Spectroscopy (PALS) measurements indicated monovacancy type defects for all irradiated samples, while Channeling studies employing RBS with 2MeV alpha particles, revealed no significant defects in the samples exposed to fast neutron fluence of ∼4×1016/cm2. Both PALS and Channeling studies have shown that vacuum annealing at 600°C for ∼2h is sufficient to recover the damage in the irradiated samples, thereby making them suitable for the sensor development." }The-Makers-of-things/thesis-computational-artificial-market % PDF Latex version: % pdfTeX 3.14159265-2.6-1.40.20 (TeX Live 2019/W32TeX) % kpathsea version 6.3.1 % Compiled with libpng 1.6.36; using libpng 1.6.36 % Compiled with zlib 1.2.11; using zlib 1.2.11 % Compiled with xpdf version 4.01 % Setting document properties \documentclass[a4paper,twoside,12pt,notitlepage,openright,12pt]{article} % Load relevant packages % ---------------------- \usepackage[ansinew]{inputenc} \usepackage{graphicx} %[dvips] to used with .eps figures \usepackage{natbib} % https://www.overleaf.com/learn/latex/natbib_citation_styles \usepackage[nottoc,notlof,notlot]{tocbibind} % Include references to TOC but exclude list of figures & tables \usepackage{lastpage} % page count \usepackage{amsmath} \usepackage{amssymb} \usepackage[english]{babel} \usepackage[intoc]{nomencl} % Includes nomencl (scientific listing of figures, list of tables etc.) to table of contents \usepackage{ifthen} \usepackage[hmargin=3.0cm,vmargin=3.6cm]{geometry} % setting marginals \usepackage{fancyhdr,extramarks} % header ja footer manipulation \usepackage{times} % to change font to times \usepackage{setspace} % for linespacing \usepackage{caption} \usepackage{booktabs} \usepackage{subcaption} \usepackage{url} \usepackage{tabularx} \usepackage{adjustbox} \usepackage{makecell} \usepackage{geometry} \usepackage[ruled,vlined]{algorithm2e} \usepackage{algpseudocode} \usepackage[toc,page,title]{appendix} % Putting empty line after each paragraph \usepackage[parfill]{parskip} % To put figures pericsely where they are in the code \usepackage{float} % Setup % ----- \newcommand{\nomunit}[1]{\renewcommand{\nomentryend}{\hspace*{\fill}#1}} % Inserts units on the right at symbol list \renewcommand{\nomgroup}[1]{% \ifthenelse{\equal{#1}{C}}{\item[\textbf{Latin alphabet}]\item}{% \ifthenelse{\equal{#1}{G}}{\item\item[\textbf{Greek alphabet}]\item}{}}{% \ifthenelse{\equal{#1}{L}}{\item\item[\textbf{Subscripts}]\item}{}}{% \ifthenelse{\equal{#1}{H}}{\item\item[\textbf{Superscripts}]\item}{}}{% \ifthenelse{\equal{#1}{W}}{\item\item[\textbf{Abbreviations}]\item}{}}{}} \singlespacing % HEADING % ------- \pagestyle{empty} % Alternatives: fancy, empty \fancyhf{} % clearing the header and footer \fancyhead[RE,RO]{\thepage} \renewcommand{\headrulewidth}{0pt} % ruler thickness between head and body \renewcommand{\footrulewidth}{0pt} % no ruler between body and footer \setlength{\nomitemsep}{-\parsep} % removing default extra skip between entries at nomenclature \linespread{1.25} % Set line spacing to 1.5 https://tex.stackexchange.com/a/65985 \numberwithin{equation}{section} % equation numbers with section numbers \numberwithin{table}{section} % table numbers with section numbers \numberwithin{figure}{section} % figure numbers with section numbers % To rename "bibliography" to "References". % For some reason having references in TOC (with tocbibind) renames the references title \addto{\captionsenglish}{\renewcommand{\bibname}{References}} % makeindex command needs to run at command prompt to create nomenclature list file \makenomenclature % makeindex main_v2.nlo -s nomencl.ist -o main_v2.nls % Instead of trying to fit the text to the page evenly, the content of a page has variable height. % This prevent the effect of huge amount of white spaces between paragraphs or especially after images \raggedbottom \usepackage[T1]{fontenc} \usepackage{beramono} \usepackage{listings} \usepackage[usenames,dvipsnames]{xcolor} % Bibliography & citation style % ----------------------------- \bibliographystyle{LUTapa_modif} % Programming language definitions % -------------------------------- %% %% Python definition (c) 2014 Jubobs %% % Default fixed font does not support bold face \DeclareFixedFont{\ttb}{T1}{txtt}{bx}{n}{12} % for bold \DeclareFixedFont{\ttm}{T1}{txtt}{m}{n}{12} % for normal % Custom colors \usepackage{color} \definecolor{deepblue}{rgb}{0,0,0.5} \definecolor{deepred}{rgb}{0.6,0,0} \definecolor{deepgreen}{rgb}{0,0.5,0} \newcommand\pythonstyle{\lstset{ language=Python, basicstyle=\ttm, otherkeywords={self}, % Add keywords here keywordstyle=\ttb\color{deepblue}, emph={MyClass,__init__}, % Custom highlighting emphstyle=\ttb\color{deepred}, % Custom highlighting style stringstyle=\color{deepgreen}, frame=tb, % Any extra options here showstringspaces=false % }} % Python environment \lstnewenvironment{python}[1][] { \pythonstyle \lstset{#1} } {} %% %% Julia definition (c) 2014 Jubobs %% \lstdefinelanguage{Julia}% {morekeywords={abstract,break,case,catch,const,continue,do,else,elseif,% end,export,false,for,function,immutable,import,importall,if,in,% macro,module,otherwise,quote,return,switch,true,try,type,typealias,% using,while},% sensitive=true,% alsoother={$},% morecomment=[l]\#,% morecomment=[n]{\#=}{=\#},% morestring=[s]{"}{"},% morestring=[m]{'}{'},% }[keywords,comments,strings]% \lstset{% language = Julia, basicstyle = \ttfamily, keywordstyle = \bfseries\color{blue}, stringstyle = \color{magenta}, commentstyle = \color{ForestGreen}, showstringspaces = false, } % Actual document % --------------- \begin{document} \pagenumbering{gobble} \include{cover} \include{abstract} \inputencoding{utf8} \include{abstract_finnish} \inputencoding{ansinew} \include{acknowledgments} % Contents, figures and tables \tableofcontents \listoffigures \listoftables \thispagestyle{empty} \pagestyle{fancy} %\cleardoublepage% \printnomenclature[2.0cm]% % Page numbering % -------------- \newpage % Informing Latex to start the numbering from new page (start of introduction page) \pagenumbering{arabic} % Setting page numbering to arabic numbers (1, 2, 3, etc.) \setcounter{page}{1} % Force page numbering starting from 1 from here on % Content of the Thesis % --------------------- \include{content/introduction} \include{content/microstructure} \include{content/artificial_stock_market} \include{content/literature_review} \include{content/empirics} \include{content/results} \include{content/conclusion} \bibliography{ref} \newpage \pagenumbering{gobble} \fancyhead[LO]{\nouppercase{\bfseries\firstleftxmark}} \fancyhead[RE]{\nouppercase{\bfseries\lastrightxmark}} \end{document} \chapter{Installation} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{System Requirements} MicroTESK is a set of Java-based utilities that are run from the command line. It can be used on \textbf{\textit{Windows}}, \textbf{\textit{Linux}} and \textbf{\textit{OS X}} machines that have \textbf{\textit{JDK 1.7 or later}} installed. To build MicroTESK from source code or to build the generated Java models, \textbf{\textit{Apache Ant version 1.8 or later}} is required. To generate test data based on constraints, MicroTESK needs the \textbf{\textit{Microsoft Research Z3}} or \textbf{\textit{CVC4}} solver that can work on the corresponding operating system. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Installation Steps} To install MicroTESK, the following steps should be performed: \begin{enumerate} \item Download from \url{http://forge.ispras.ru/projects/microtesk/files} and unpack the MicroTESK installation package (the \texttt{.tar.gz} file, latest release) to your computer. The folder to which it was unpacked will be further referred to as the installation directory (\texttt{}). \item Declare the \texttt{MICROTESK{\_}HOME} environment variable and set its value to the path to the installation directory (see the \hyperref[Setting_Environment_Variables] {Setting Environment Variables} section). \item Set the \texttt{/bin} folder as the working directory (add the path to the \texttt{PATH} environment variable) to be able to run MicroTESK utilities from any path. \item Note: Required for constraint-based generation. Download and install constraint solver tools to the \texttt{/tools} directory (see the \hyperref[Installing_Constraint_Solvers]{Installing Constraint Solvers} section). \end{enumerate} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsection{Setting Environment Variables} \label{Setting_Environment_Variables} \paragraph{Windows} \begin{enumerate} \item Open the \texttt{System Properties} window. \item Switch to the \texttt{Advanced} tab. \item Click on \texttt{Environment Variables}. \item Click \texttt{New..} under \texttt{System Variables}. \item In the \texttt{New System Variable} dialog, specify variable name as \texttt{MICROTESK{\_}HOME} and variable value as \texttt{}. \item Click \texttt{OK} on all open windows. \item Reopen the command prompt window. \end{enumerate} \paragraph{Linux and OS X} ~\\ Add the command below to the \texttt{\textasciitilde{}.bash{\_}profile} file (\textbf{\textit{Linux}}) or the \texttt{\textasciitilde{}/.profile} file (\textbf{\textit{OS X}}): \begin{lstlisting}[language=bash] export MICROTESK_HOME= \end{lstlisting} To start editing the file, type \texttt{vi \textasciitilde{}/.bash{\_}profile} (or \texttt{vi \textasciitilde{}/.profile}). Changes will be applied after restarting the command-line terminal or reboot. You can also run the command in your command-line terminal to make temporary changes. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsection{Installing Constraint Solvers} \label{Installing_Constraint_Solvers} To generate test data based on constraints, MicroTESK requires external constraint solvers. The current version supports the \href{https://github.com/z3prover}{Z3} and \href{http://cvc4.cs.nyu.edu}{CVC4} constraint solvers. Constraint executables should be downloaded and placed to the \texttt{/tools} directory. \paragraph{Using Environment Variables} ~\\ If solvers are already installed in another directory, to let MicroTESK find them, the following environment variables can be used: \texttt{Z3{\_}PATH} and \texttt{CVC4{\_}PATH}. They specify the paths to the Z3 and CVC4 excutables correspondingly. \paragraph{Installing Z3} \begin{itemize} \item \textbf{\textit{Windows}} users should download Z3 (32 or 64-bit version) from the following page:\url{http://z3.codeplex.com/releases} and unpack the archive to the \texttt{/tools/z3/windows} directory. Note: the executable file path is \texttt{/z3/bin/z3.exe}. \item \textbf{\textit{UNIX}} and \textbf{\textit{Linux}} users should use one of the links below and and unpack the archive to the \texttt{/tools/z3/unix} directory. Note: the executable file path is \texttt{/z3/bin/z3}. \begin{tabular} {| l | r |} \hline Debian x64 & \url{http://z3.codeplex.com/releases/view/101916} \\ \hline Ubuntu x86 & \url{http://z3.codeplex.com/releases/view/101913} \\ \hline Ubuntu x64 & \url{http://z3.codeplex.com/releases/view/101911} \\ \hline FreeBSD x64 & \url{http://z3.codeplex.com/releases/view/101907} \\ \hline \end{tabular} \item \textbf{\textit{OS X}} users should download Z3 from \url{http://z3.codeplex.com/releases/view/101918} and unpack the archive to the \texttt{/z3/osx} directory. Note: the executable file path is \texttt{/z3/bin/z3}. \end{itemize} \paragraph{Installing CVC4} \begin{itemize} \item \textbf{\textit{Windows}} users should download the latest version of CVC4 binary from \url{http://cvc4.cs.nyu.edu/builds/win32-opt/} and save it to the \texttt{/tools/cvc4/windows} directory as \texttt{cvc4.exe}. \item \textbf{\textit{Linux}} users download the latest version of CVC4 binary from \url{http://cvc4.cs.nyu.edu/builds/i386-linux-opt/unstable/} (32-bit version) or \url{http://cvc4.cs.nyu.edu/builds/x86_64-linux-opt/unstable/} (64-bit version) and save it to the \texttt{/tools/cvc4/unix} directory as \texttt{cvc4}. \item \textbf{\textit{OS X}} users should download the latest version of CVC4 distribution package from \url{http://cvc4.cs.nyu.edu/builds/macos/} and install it. The CVC4 binary should be copied to \texttt{/tools/cvc4/osx} as \texttt{cvc4} or linked to this file name via a symbolic link. \end{itemize} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \section{Installation Directory Structure} The MicroTESK installation directory contains the following subdirectories: \\ \begin{tabular}{ | l | l |} \hline \textbf{arch} & Microprocessor specifications and test templates \\ \hline \textbf{bin} & Scripts to run modeling and test generation tasks \\ \hline \textbf{doc} & Documentation \\ \hline \textbf{etc} & Configuration files \\ \hline \textbf{gen} & Generated code of microprocessor models \\ \hline \textbf{lib} & JAR files and Ruby scripts to perform modeling and \\ ~ & test generation tasks \\ \hline \textbf{src} & Source code of MicroTESK \\ \hline \end{tabular} fatemeh-azadi/MarianDocs/doxygen_output/latex/db/dd1/classmarian_1_1Vocab.tex \hypertarget{classmarian_1_1Vocab}{}\section{marian\+:\+:Vocab Class Reference} \label{classmarian_1_1Vocab}\index{marian\+::\+Vocab@{marian\+::\+Vocab}} {\ttfamily \#include $<$vocab.\+h$>$} Collaboration diagram for marian\+:\+:Vocab\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=334pt]{d8/dc2/classmarian_1_1Vocab__coll__graph} \end{center} \end{figure} \subsection*{Classes} \begin{DoxyCompactItemize} \item class \hyperlink{classmarian_1_1Vocab_1_1VocabFreqOrderer}{Vocab\+Freq\+Orderer} \end{DoxyCompactItemize} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hyperlink{classmarian_1_1Vocab_a0e99d044198e23cd5187d09883e326ca}{Vocab} () \item size\+\_\+t \hyperlink{classmarian_1_1Vocab_aac39bcbded54b4a3d173a8db4c37d5fe}{operator\mbox{[}$\,$\mbox{]}} (const std\+::string \&word) const \item \hyperlink{namespacemarian_a5385eef6e49dd8f789f616ef579dea3f}{Words} \hyperlink{classmarian_1_1Vocab_a2d83592871711da3b20c7a582be4b743}{operator()} (const std\+::vector$<$ std\+::string $>$ \&line\+Tokens, bool add\+E\+OS=true) const \item \hyperlink{namespacemarian_a5385eef6e49dd8f789f616ef579dea3f}{Words} \hyperlink{classmarian_1_1Vocab_adf744607f7b211cf64541d00d0fd9ab6}{operator()} (const std\+::string \&line, bool add\+E\+OS=true) const \item std\+::vector$<$ std\+::string $>$ \hyperlink{classmarian_1_1Vocab_a7e70e0e9345e36e8cf8960595e86d409}{operator()} (const \hyperlink{namespacemarian_a5385eef6e49dd8f789f616ef579dea3f}{Words} \&sentence, bool ignore\+E\+OS=true) const \item const std\+::string \& \hyperlink{classmarian_1_1Vocab_a3e12519d40e2f1193741a86303b842e0}{operator\mbox{[}$\,$\mbox{]}} (size\+\_\+t id) const \item size\+\_\+t \hyperlink{classmarian_1_1Vocab_a459153ecfdd0c3ad6b256752adda0e84}{size} () const \item void \hyperlink{classmarian_1_1Vocab_a68171b2369e02232e4e24b04934256a9}{load\+Or\+Create} (const std\+::string \&vocab\+Path, const std\+::string \&text\+Path, int max=0) \item void \hyperlink{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{load} (const std\+::string \&vocab\+Path, int max=0) \item void \hyperlink{classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1}{create} (const std\+::string \&vocab\+Path, int max, const std\+::string \&train\+Path) \end{DoxyCompactItemize} \subsection*{Private Types} \begin{DoxyCompactItemize} \item typedef std\+::map$<$ std\+::string, size\+\_\+t $>$ \hyperlink{classmarian_1_1Vocab_a3ff740ce19200167d306de3643bd49e3}{Str2\+Id} \item typedef std\+::vector$<$ std\+::string $>$ \hyperlink{classmarian_1_1Vocab_a9208af73b752aeed1ce03ae5c429d897}{Id2\+Str} \end{DoxyCompactItemize} \subsection*{Private Attributes} \begin{DoxyCompactItemize} \item \hyperlink{classmarian_1_1Vocab_a3ff740ce19200167d306de3643bd49e3}{Str2\+Id} \hyperlink{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739}{str2id\+\_\+} \item \hyperlink{classmarian_1_1Vocab_a9208af73b752aeed1ce03ae5c429d897}{Id2\+Str} \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\+\_\+} \end{DoxyCompactItemize} \subsection{Detailed Description} Definition at line 11 of file marian/src/data/vocab.\+h. \subsection{Member Typedef Documentation} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!Id2\+Str@{Id2\+Str}} \index{Id2\+Str@{Id2\+Str}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{Id2\+Str}{Id2Str}}]{\setlength{\rightskip}{0pt plus 5cm}typedef std\+::vector$<$std\+::string$>$ {\bf marian\+::\+Vocab\+::\+Id2\+Str}\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classmarian_1_1Vocab_a9208af73b752aeed1ce03ae5c429d897}{}\label{classmarian_1_1Vocab_a9208af73b752aeed1ce03ae5c429d897} Definition at line 41 of file marian/src/data/vocab.\+h. \index{marian\+::\+Vocab@{marian\+::\+Vocab}!Str2\+Id@{Str2\+Id}} \index{Str2\+Id@{Str2\+Id}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{Str2\+Id}{Str2Id}}]{\setlength{\rightskip}{0pt plus 5cm}typedef std\+::map$<$std\+::string, size\+\_\+t$>$ {\bf marian\+::\+Vocab\+::\+Str2\+Id}\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classmarian_1_1Vocab_a3ff740ce19200167d306de3643bd49e3}{}\label{classmarian_1_1Vocab_a3ff740ce19200167d306de3643bd49e3} Definition at line 38 of file marian/src/data/vocab.\+h. \subsection{Constructor \& Destructor Documentation} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!Vocab@{Vocab}} \index{Vocab@{Vocab}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{Vocab()}{Vocab()}}]{\setlength{\rightskip}{0pt plus 5cm}marian\+::\+Vocab\+::\+Vocab ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )}\hypertarget{classmarian_1_1Vocab_a0e99d044198e23cd5187d09883e326ca}{}\label{classmarian_1_1Vocab_a0e99d044198e23cd5187d09883e326ca} Definition at line 15 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 15 \{\} \end{DoxyCode} \subsection{Member Function Documentation} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!create@{create}} \index{create@{create}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{create(const std\+::string \&vocab\+Path, int max, const std\+::string \&train\+Path)}{create(const std::string &vocabPath, int max, const std::string &trainPath)}}]{\setlength{\rightskip}{0pt plus 5cm}void marian\+::\+Vocab\+::create ( \begin{DoxyParamCaption} \item[{const std\+::string \&}]{vocab\+Path, } \item[{int}]{max, } \item[{const std\+::string \&}]{train\+Path} \end{DoxyParamCaption} )}\hypertarget{classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1}{}\label{classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1} Definition at line 126 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 128 \{ 129 \hyperlink{amun_2common_2logging_8h_a8cad147aca8c526d3c8a03ae14d5c87d}{LOG}(data) 130 ->info(\textcolor{stringliteral}{"Creating vocabulary \{\} from \{\} (max: \{\})"}, 131 vocabPath, 132 trainPath, 133 max); 134 135 UTIL\_THROW\_IF2(boost::filesystem::exists(vocabPath), 136 \textcolor{stringliteral}{"Vocab file "} << vocabPath << \textcolor{stringliteral}{" exists. Not overwriting"}); 137 138 \hyperlink{classInputFileStream}{InputFileStream} trainStrm(trainPath); 139 140 std::string line; 141 std::unordered\_map counter; 142 143 std::unordered\_set seenSpecial; 144 145 \textcolor{keywordflow}{while}(getline((std::istream&)trainStrm, line)) \{ 146 std::vector toks; 147 \hyperlink{namespaceamunmt_a10b7486d36b130609c77e8356218c0a7}{Split}(line, toks); 148 149 \textcolor{keywordflow}{for}(\textcolor{keyword}{const} std::string& tok : toks) \{ 150 \textcolor{keywordflow}{if}(\hyperlink{namespacemarian_a53db7e8e290296787decc0fd95948535}{SPEC2SYM}.count(tok)) \{ 151 seenSpecial.insert(\hyperlink{namespacemarian_a53db7e8e290296787decc0fd95948535}{SPEC2SYM}.at(tok)); 152 \textcolor{keywordflow}{continue}; 153 \} 154 155 \textcolor{keyword}{auto} iter = counter.find(tok); 156 \textcolor{keywordflow}{if}(iter == counter.end()) 157 counter[tok] = 1; 158 \textcolor{keywordflow}{else} 159 iter->second++; 160 \} 161 \} 162 163 std::vector vocabVec; 164 \textcolor{keywordflow}{for}(\textcolor{keyword}{auto}& p : counter) 165 vocabVec.push\_back(p.first); 166 167 std::sort(vocabVec.begin(), vocabVec.end(), VocabFreqOrderer(counter)); 168 169 YAML::Node vocabYaml; 170 vocabYaml.force\_insert(\hyperlink{namespacemarian_ae9a4ff3b7d4df0a6ea4a710810c5b6f6}{EOS\_STR}, \hyperlink{namespacemarian_a58eb44c6e8969831f40ee27feeed33ed}{EOS\_ID}); 171 vocabYaml.force\_insert(\hyperlink{namespacemarian_a85fabda4ef9ec625d0c992f81292dc92}{UNK\_STR}, \hyperlink{namespacemarian_abe23e32294ac45b7fe50b5ee8d7da9ba}{UNK\_ID}); 172 173 \textcolor{keywordflow}{for}(\textcolor{keyword}{auto} word : seenSpecial) 174 vocabYaml.force\_insert(\hyperlink{namespacemarian_a70398af3e3d1cc14ff372edcdc1a2595}{SYM2SPEC}.at(word), word); 175 176 \hyperlink{namespacemarian_a5db8bee455c97a62d6a525dc48efe4c2}{Word} maxSpec = 1; 177 \textcolor{keywordflow}{for}(\textcolor{keyword}{auto} i : seenSpecial) 178 \textcolor{keywordflow}{if}(i > maxSpec) 179 maxSpec = i; 180 181 \textcolor{keywordflow}{for}(\textcolor{keywordtype}{size\_t} i = 0; i < vocabVec.size(); ++i) 182 vocabYaml.force\_insert(vocabVec[i], i + maxSpec + 1); 183 184 \hyperlink{classOutputFileStream}{OutputFileStream} vocabStrm(vocabPath); 185 (std::ostream&)vocabStrm << vocabYaml; 186 \} \end{DoxyCode} Here is the call graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=306pt]{db/dd1/classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1_cgraph} \end{center} \end{figure} Here is the caller graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{db/dd1/classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1_icgraph} \end{center} \end{figure} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!load@{load}} \index{load@{load}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{load(const std\+::string \&vocab\+Path, int max=0)}{load(const std::string &vocabPath, int max=0)}}]{\setlength{\rightskip}{0pt plus 5cm}void marian\+::\+Vocab\+::load ( \begin{DoxyParamCaption} \item[{const std\+::string \&}]{vocab\+Path, } \item[{int}]{max = {\ttfamily 0}} \end{DoxyParamCaption} )}\hypertarget{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{}\label{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76} Definition at line 84 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 84 \{ 85 \hyperlink{amun_2common_2logging_8h_a8cad147aca8c526d3c8a03ae14d5c87d}{LOG}(data)->info(\textcolor{stringliteral}{"Loading vocabulary from \{\}"}, vocabPath); 86 YAML::Node vocab = YAML::Load(\hyperlink{classInputFileStream}{InputFileStream}(vocabPath)); 87 88 std::unordered\_set seenSpecial; 89 90 \textcolor{keywordflow}{for}(\textcolor{keyword}{auto}&& pair : vocab) \{ 91 \textcolor{keyword}{auto} str = pair.first.as(); 92 \textcolor{keyword}{auto} \textcolor{keywordtype}{id} = pair.second.as<\hyperlink{namespacemarian_a5db8bee455c97a62d6a525dc48efe4c2}{Word}>(); 93 94 \textcolor{keywordflow}{if}(\hyperlink{namespacemarian_a53db7e8e290296787decc0fd95948535}{SPEC2SYM}.count(str)) \{ 95 seenSpecial.insert(\textcolor{keywordtype}{id}); 96 \} 97 98 \textcolor{keywordflow}{if}(!max || \textcolor{keywordtype}{id} < (\hyperlink{namespacemarian_a5db8bee455c97a62d6a525dc48efe4c2}{Word})max) \{ 99 \hyperlink{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739}{str2id\_}[str] = id; 100 \textcolor{keywordflow}{if}(\textcolor{keywordtype}{id} >= \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}.size()) 101 \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}.resize(\textcolor{keywordtype}{id} + 1); 102 \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}[id] = str; 103 \} 104 \} 105 UTIL\_THROW\_IF2(\hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}.empty(), \textcolor{stringliteral}{"Empty vocabulary "} << vocabPath); 106 107 \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}[\hyperlink{namespacemarian_a58eb44c6e8969831f40ee27feeed33ed}{EOS\_ID}] = \hyperlink{namespacemarian_ae9a4ff3b7d4df0a6ea4a710810c5b6f6}{EOS\_STR}; 108 id2str\_[\hyperlink{namespacemarian_abe23e32294ac45b7fe50b5ee8d7da9ba}{UNK\_ID}] = \hyperlink{namespacemarian_a85fabda4ef9ec625d0c992f81292dc92}{UNK\_STR}; 109 \textcolor{keywordflow}{for}(\textcolor{keyword}{auto} \textcolor{keywordtype}{id} : seenSpecial) 110 id2str\_[id] = \hyperlink{namespacemarian_a70398af3e3d1cc14ff372edcdc1a2595}{SYM2SPEC}.at(\textcolor{keywordtype}{id}); 111 \} \end{DoxyCode} Here is the caller graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{db/dd1/classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76_icgraph} \end{center} \end{figure} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!load\+Or\+Create@{load\+Or\+Create}} \index{load\+Or\+Create@{load\+Or\+Create}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{load\+Or\+Create(const std\+::string \&vocab\+Path, const std\+::string \&text\+Path, int max=0)}{loadOrCreate(const std::string &vocabPath, const std::string &textPath, int max=0)}}]{\setlength{\rightskip}{0pt plus 5cm}void marian\+::\+Vocab\+::load\+Or\+Create ( \begin{DoxyParamCaption} \item[{const std\+::string \&}]{vocab\+Path, } \item[{const std\+::string \&}]{text\+Path, } \item[{int}]{max = {\ttfamily 0}} \end{DoxyParamCaption} )}\hypertarget{classmarian_1_1Vocab_a68171b2369e02232e4e24b04934256a9}{}\label{classmarian_1_1Vocab_a68171b2369e02232e4e24b04934256a9} Definition at line 63 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 65 \{ 66 \textcolor{keywordflow}{if}(vocabPath.empty()) \{ 67 \textcolor{keywordflow}{if}(boost::filesystem::exists(trainPath + \textcolor{stringliteral}{".json"})) \{ 68 \hyperlink{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{load}(trainPath + \textcolor{stringliteral}{".json"}, max); 69 \textcolor{keywordflow}{return}; 70 \} 71 \textcolor{keywordflow}{if}(boost::filesystem::exists(trainPath + \textcolor{stringliteral}{".yml"})) \{ 72 \hyperlink{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{load}(trainPath + \textcolor{stringliteral}{".yml"}, max); 73 \textcolor{keywordflow}{return}; 74 \} 75 \hyperlink{classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1}{create}(trainPath + \textcolor{stringliteral}{".yml"}, max, trainPath); 76 \hyperlink{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{load}(trainPath + \textcolor{stringliteral}{".yml"}, max); 77 \} \textcolor{keywordflow}{else} \{ 78 \textcolor{keywordflow}{if}(!boost::filesystem::exists(vocabPath)) 79 \hyperlink{classmarian_1_1Vocab_aeb5be1eaae536e408eac5ab06f50d9c1}{create}(vocabPath, max, trainPath); 80 \hyperlink{classmarian_1_1Vocab_a8387e1e9b7923b9d84c0b4defd25bd76}{load}(vocabPath, max); 81 \} 82 \} \end{DoxyCode} Here is the call graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{db/dd1/classmarian_1_1Vocab_a68171b2369e02232e4e24b04934256a9_cgraph} \end{center} \end{figure} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!operator()@{operator()}} \index{operator()@{operator()}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{operator()(const std\+::vector$<$ std\+::string $>$ \&line\+Tokens, bool add\+E\+O\+S=true) const }{operator()(const std::vector< std::string > &lineTokens, bool addEOS=true) const }}]{\setlength{\rightskip}{0pt plus 5cm}{\bf Words} marian\+::\+Vocab\+::operator() ( \begin{DoxyParamCaption} \item[{const std\+::vector$<$ std\+::string $>$ \&}]{line\+Tokens, } \item[{bool}]{add\+E\+OS = {\ttfamily true}} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_a2d83592871711da3b20c7a582be4b743}{}\label{classmarian_1_1Vocab_a2d83592871711da3b20c7a582be4b743} Definition at line 25 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 26 \{ 27 \hyperlink{namespacemarian_a5385eef6e49dd8f789f616ef579dea3f}{Words} words(lineTokens.size()); 28 std::transform(lineTokens.begin(), 29 lineTokens.end(), 30 words.begin(), 31 [&](\textcolor{keyword}{const} std::string& w) \{ \textcolor{keywordflow}{return} (*\textcolor{keyword}{this})[w]; \}); 32 \textcolor{keywordflow}{if}(addEOS) 33 words.push\_back(\hyperlink{namespacemarian_a58eb44c6e8969831f40ee27feeed33ed}{EOS\_ID}); 34 \textcolor{keywordflow}{return} words; 35 \} \end{DoxyCode} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!operator()@{operator()}} \index{operator()@{operator()}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{operator()(const std\+::string \&line, bool add\+E\+O\+S=true) const }{operator()(const std::string &line, bool addEOS=true) const }}]{\setlength{\rightskip}{0pt plus 5cm}{\bf Words} marian\+::\+Vocab\+::operator() ( \begin{DoxyParamCaption} \item[{const std\+::string \&}]{line, } \item[{bool}]{add\+E\+OS = {\ttfamily true}} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_adf744607f7b211cf64541d00d0fd9ab6}{}\label{classmarian_1_1Vocab_adf744607f7b211cf64541d00d0fd9ab6} Definition at line 37 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 37 \{ 38 std::vector lineTokens; 39 \hyperlink{namespaceamunmt_a10b7486d36b130609c77e8356218c0a7}{Split}(line, lineTokens, \textcolor{stringliteral}{" "}); 40 \textcolor{keywordflow}{return} (*\textcolor{keyword}{this})(lineTokens, addEOS); 41 \} \end{DoxyCode} Here is the call graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=320pt]{db/dd1/classmarian_1_1Vocab_adf744607f7b211cf64541d00d0fd9ab6_cgraph} \end{center} \end{figure} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!operator()@{operator()}} \index{operator()@{operator()}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{operator()(const Words \&sentence, bool ignore\+E\+O\+S=true) const }{operator()(const Words &sentence, bool ignoreEOS=true) const }}]{\setlength{\rightskip}{0pt plus 5cm}std\+::vector$<$ std\+::string $>$ marian\+::\+Vocab\+::operator() ( \begin{DoxyParamCaption} \item[{const {\bf Words} \&}]{sentence, } \item[{bool}]{ignore\+E\+OS = {\ttfamily true}} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_a7e70e0e9345e36e8cf8960595e86d409}{}\label{classmarian_1_1Vocab_a7e70e0e9345e36e8cf8960595e86d409} Definition at line 43 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 44 \{ 45 std::vector decoded; 46 \textcolor{keywordflow}{for}(\textcolor{keywordtype}{size\_t} i = 0; i < sentence.size(); ++i) \{ 47 \textcolor{keywordflow}{if}((sentence[i] != \hyperlink{namespacemarian_a58eb44c6e8969831f40ee27feeed33ed}{EOS\_ID} || !ignoreEOS)) \{ 48 decoded.push\_back((*\textcolor{keyword}{this})[sentence[i]]); 49 \} 50 \} 51 \textcolor{keywordflow}{return} decoded; 52 \} \end{DoxyCode} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!operator\mbox{[}$\,$\mbox{]}@{operator[]}} \index{operator\mbox{[}$\,$\mbox{]}@{operator[]}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{operator[](const std\+::string \&word) const }{operator[](const std::string &word) const }}]{\setlength{\rightskip}{0pt plus 5cm}size\+\_\+t marian\+::\+Vocab\+::operator\mbox{[}$\,$\mbox{]} ( \begin{DoxyParamCaption} \item[{const std\+::string \&}]{word} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_aac39bcbded54b4a3d173a8db4c37d5fe}{}\label{classmarian_1_1Vocab_aac39bcbded54b4a3d173a8db4c37d5fe} Definition at line 17 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 17 \{ 18 \textcolor{keyword}{auto} it = \hyperlink{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739}{str2id\_}.find(word); 19 \textcolor{keywordflow}{if}(it != \hyperlink{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739}{str2id\_}.end()) 20 \textcolor{keywordflow}{return} it->second; 21 \textcolor{keywordflow}{else} 22 \textcolor{keywordflow}{return} \hyperlink{namespacemarian_abe23e32294ac45b7fe50b5ee8d7da9ba}{UNK\_ID}; 23 \} \end{DoxyCode} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!operator\mbox{[}$\,$\mbox{]}@{operator[]}} \index{operator\mbox{[}$\,$\mbox{]}@{operator[]}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{operator[](size\+\_\+t id) const }{operator[](size_t id) const }}]{\setlength{\rightskip}{0pt plus 5cm}const std\+::string \& marian\+::\+Vocab\+::operator\mbox{[}$\,$\mbox{]} ( \begin{DoxyParamCaption} \item[{size\+\_\+t}]{id} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_a3e12519d40e2f1193741a86303b842e0}{}\label{classmarian_1_1Vocab_a3e12519d40e2f1193741a86303b842e0} Definition at line 54 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 54 \{ 55 UTIL\_THROW\_IF2(\textcolor{keywordtype}{id} >= \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}.size(), \textcolor{stringliteral}{"Unknown word id: "} << id); 56 \textcolor{keywordflow}{return} \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}[id]; 57 \} \end{DoxyCode} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!size@{size}} \index{size@{size}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{size() const }{size() const }}]{\setlength{\rightskip}{0pt plus 5cm}size\+\_\+t marian\+::\+Vocab\+::size ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} ) const}\hypertarget{classmarian_1_1Vocab_a459153ecfdd0c3ad6b256752adda0e84}{}\label{classmarian_1_1Vocab_a459153ecfdd0c3ad6b256752adda0e84} Definition at line 59 of file marian/src/data/vocab.\+cpp. \begin{DoxyCode} 59 \{ 60 \textcolor{keywordflow}{return} \hyperlink{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{id2str\_}.size(); 61 \} \end{DoxyCode} \subsection{Member Data Documentation} \index{marian\+::\+Vocab@{marian\+::\+Vocab}!id2str\+\_\+@{id2str\+\_\+}} \index{id2str\+\_\+@{id2str\+\_\+}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{id2str\+\_\+}{id2str_}}]{\setlength{\rightskip}{0pt plus 5cm}{\bf Id2\+Str} marian\+::\+Vocab\+::id2str\+\_\+\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac}{}\label{classmarian_1_1Vocab_a9ec3e11b136d7d75d8647471c28a5fac} Definition at line 42 of file marian/src/data/vocab.\+h. \index{marian\+::\+Vocab@{marian\+::\+Vocab}!str2id\+\_\+@{str2id\+\_\+}} \index{str2id\+\_\+@{str2id\+\_\+}!marian\+::\+Vocab@{marian\+::\+Vocab}} \subsubsection[{\texorpdfstring{str2id\+\_\+}{str2id_}}]{\setlength{\rightskip}{0pt plus 5cm}{\bf Str2\+Id} marian\+::\+Vocab\+::str2id\+\_\+\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739}{}\label{classmarian_1_1Vocab_ab79ee07a9432b1273fc4c92ea7390739} Definition at line 39 of file marian/src/data/vocab.\+h. The documentation for this class was generated from the following files\+:\begin{DoxyCompactItemize} \item \hyperlink{marian_2src_2data_2vocab_8h}{marian/src/data/vocab.\+h}\item \hyperlink{marian_2src_2data_2vocab_8cpp}{marian/src/data/vocab.\+cpp}\end{DoxyCompactItemize} lsst-dm/dmtn-159 \addtocounter{table}{-1} \begin{longtable}{p{0.145\textwidth}p{0.8\textwidth}}\hline \textbf{Acronym} & \textbf{Description} \\\hline & \\\hline CCD & Charge-Coupled Device \\\hline COVID & COrona VIrus Disease \\\hline ComCam & The commissioning camera is a single-raft, 9-CCD camera that will be installed in LSST during commissioning, before the final camera is ready. \\\hline DAQ & Data Acquisition System \\\hline DB & DataBase \\\hline DM & Data Management \\\hline DMTN & DM Technical Note \\\hline Gb & Gigabit \\\hline IT & Information Technology \\\hline L3 & Lens 3 \\\hline LDM & LSST Data Management (Document Handle) \\\hline LHN & long haul network \\\hline Mb & Megabit (1000000 bit) \\\hline N2 & Nitrogen \\\hline NCSA & National Center for Supercomputing Applications \\\hline OCS & Observatory Control System \\\hline OPS & Operations \\\hline PST & Project Science Team \\\hline QA & Quality Assurance \\\hline RSP & Rubin Science Platform \\\hline USDF & United States Data Facility \\\hline \end{longtable} cbjuan/juancb.es0 @article{8240912, abstract = {This paper presents an original study with the aim of improving users' performance in completing large questionnaires through adaptability in web forms. Such adaptability is based on the application of machine-learning procedures and an A/B testing approach. To detect the user preferences, behavior, and the optimal version of the forms for all kinds of users, researchers built predictive models using machine-learning algorithms (trained with data from more than 3000 users who participated previously in the questionnaires), extracting the most relevant factors that describe the models, and clustering the users based on their similar characteristics and these factors. Based on these groups and their performance in the system, the researchers generated heuristic rules between the different versions of the web forms to guide users to the most adequate version (modifying the user interface and user experience) for them. To validate the approach and confirm the improvements, the authors tested these redirection rules on a group of more than 1000 users. The results with this cohort of users were better than those achieved without redirection rules at the initial stage. Besides these promising results, the paper proposes a future study that would enhance the process (or automate it) as well as push its application to other fields.}, author = { and and and and and }, doi = {10.1109/ACCESS.2017.2782678}, issn = {2169-3536}, journal = {IEEE Access}, keywords = {human factors;learning (artificial intelligence);user interfaces;enabling adaptability;user characteristics detection;machine learning;original study;questionnaires;machine-learning procedures;A/B testing approach;user preferences;optimal version;predictive models;machine-learning algorithms;relevant factors;similar characteristics;different versions;adequate version;user interface;user experience;redirection rules;Web forms;user behavior;heuristic rules;Predictive models;Testing;Employment;Tools;Machine learning algorithms;Software;Electronic mail;Adaptability;machine learning;user profiles;web forms;clusters;hierarchical clustering;random forest;A/B testing;human-computer interaction;HCI}, month = {}, number = {}, pages = {2251-2265}, title = {Enabling Adaptability in Web Forms Based on User Characteristics Detection Through A/B Testing and Machine Learning}, volume = {6}, year = {2018} } \subsection*{I - Cas discret} \begin{enumerate} \item On considère deux familles $a_1\leq \cdots \leq a_n$ et $b_1\leq \cdots \leq b_n$. Introduisons une nouvelle famille en posant \begin{displaymath} A = \sum_{k=1}^n a_k \text{ et } \forall k\in \llbracket 1, n \rrbracket, \; a'_k = a_k - \frac{A}{n} \end{displaymath} de sorte que la famille des $a'_k$ est croissante et de somme nulle. Si l'inégalité est prouvée pour une telle somme, on peut écrire: \begin{multline*} \frac{1}{n}\sum_{k=1}^na'_kb_k \geq 0 \Rightarrow \frac{1}{n}\sum_{k=1}^n(a_k-\frac{A}{n})b_k \geq 0 \Rightarrow \frac{1}{n}\sum_{k=1}^na_kb_k -\frac{A}{n}\left(\frac{1}{n}\sum_{k=1}^nb_k\right)\geq 0 \\ \Rightarrow \frac{1}{n}\sum_{k=1}^na_kb_k \geq \left(\frac{1}{n}\sum_{k=1}^na_k\right)\left(\frac{1}{n}\sum_{k=1}^nb_k\right) \end{multline*} \item Commençons par une remarque sur les familles croissantes de somme nulle.\newline Dans une telle famille, le dernier terme (le plus grand) est positif ou nul. S'il ne l'était pas tous les termes seraient strictement négatifs et la somme serait strictement négative au lieu d'être nulle. De plus, si le dernier terme (le plus grand) est nul, alors \emph{tous} les termes sont nuls sinon encore la somme serait strictement négative. On peut donc en conclure que si la famille n'est pas la famille triviale (dont tous les termes sont nuls) alors le dernier terme est strictement positif. Si tous les $a_k$ sont nuls, on prend $a'_n=0$ et $b'_n = b_n$ et les contraintes sont vérifiées (somme nulle). Si la famille des $a_k$ n'est pas la famille triviale nulle, la remarque préliminaire indique que $a_{n+1}>0$. La condition sur la somme des $a_k$ entraîne alors que $a'_n = a_{n}+a_{n+1} > a_n$. La remarque préliminaire s'applique encore à la famille qui se termine par $a'_n$. On en déduit $a'_n\geq 0$.\newline Si $a'_n > 0$, de la condition sur la somme de produits, on tire \begin{displaymath} a'_nb'_n = a_{n}b_{n}+ a_{n+1}b_{n+1} \Rightarrow b'_n = \frac{a_{n}b_{n}+ a_{n+1}b_{n+1}}{a_{n}+a_{n+1}} \end{displaymath} De plus, \begin{displaymath} b'_n - b_n = \frac{a_{n+1}(b_{n+1} - b_n)}{a_n + a_{n+1}} = \frac{a_{n+1}}{a'_n}(b_{n+1}-b_n)\geq 0 \end{displaymath} Il est impossible que $a'_n=0$. Sinon, toujours d'après la remarque du début, on aurait $a_1 = \cdots = a_{n-1} = 0$ et $a_n + a_{n+1}=0$ ce qui entraîne $a_n = - a_{n+1} <0$ et $a_1 \leq \cdots \leq a_{n-1} \leq a_n <0$ ce qui est contradictoire. \item Démontrons l'inégalité par récurrence sur $n$. Elle est évidente pour $n=1$. Montrons que l'inégalité pour une valeur de $n$ entraîne celle pour $n+1$.\newline On considère des familles comme dans la question précédente et on applique l'inégalité aux familles qui se terminent en $a'_n$ et $b'_n$. \begin{displaymath} \frac{1}{n}\left(a_1b_1+\cdots+a_{n+1}b_{n+1} \right) = \frac{1}{n}\left(a_1b_1+\cdots+a_{n-1}b_{n-1}+a'_nb'_n \right)\geq 0 \end{displaymath} \item La famille des $b_k$ est croissante, celle des $b_{n-k+1}$ est décroissante et celle des $-b_{n-k+1}$ est croissante. On peut lui appliquer l'inégalité de Chebychev (avec les $a_k$): \begin{multline*} \frac{1}{n}\sum_{k=0}^na_k(-b_{n-k+1}) \geq \left(\frac{1}{n}\sum_{k=0}^na_k\right) \left(\frac{-1}{n}\sum_{k=0}^nb_{n-k+1}\right) \\ \Rightarrow -\frac{1}{n}\sum_{k=0}^na_kb_{n-k+1} \geq -\left(\frac{1}{n}\sum_{k=0}^na_k\right) \left(\frac{1}{n}\sum_{k=0}^nb_{n-k+1}\right) \\ \Rightarrow \frac{1}{n}\sum_{k=0}^na_kb_{n-k+1} \leq \left(\frac{1}{n}\sum_{k=0}^na_k\right) \left(\frac{1}{n}\sum_{k=0}^nb_{k}\right) \end{multline*} après changement d'indice dans la dernière somme de $b$. \item Inégalité de Nesbitt. \begin{enumerate} \item En écrivant $1=\frac{b+c}{b+c}=\frac{c+a}{c+a}=\frac{a+b}{a+b}$ et en affectant un "1" du "3" à chaque terme, on fait apparaitre le même numérateur $a+b+c$. La mise en facteur conduit à la factorisation demandée. \item Si on permute deux lettres, par exemple $a$ et $b$, les deux premiers termes de la somme sont intervertis et le troisième est inchangé. La somme est donc globalement conservée. Plus généralement, la somme est conservée par toute permutation des lettres. On peut donc supposer que $a\leq b \leq c$. \item L'ordre sur les lettres induit un ordre sur les sommes: \begin{displaymath} \left. \begin{aligned} a\leq b &\Rightarrow c+a \leq b+c \\ b\leq c &\Rightarrow a+b \leq c+a \end{aligned} \right\rbrace \Rightarrow a+b \leq c+a \leq b+c \Rightarrow \frac{1}{b+c} \leq \frac{1}{c+a} \leq \frac{1}{a+b} \end{displaymath} L'inégalité de Chebychev s'écrit alors \begin{displaymath} \frac{1}{3}\left(\frac{a}{b+c}+\frac{b}{c+a}+\frac{c}{a+b} \right) \geq \frac{1}{9}\left( a+b+c\right)\left( \frac{1}{b+c} + \frac{1}{c+a} + \frac{1}{a+b}\right) \end{displaymath} Notons $S$ la somme à minorer et utilisons la question 1., l'inégalité devient \begin{displaymath} \frac{1}{3}S\geq\frac{1}{9}(S+3)\Rightarrow \frac{2}{9}S \geq \frac{1}{3} \Rightarrow S\geq \frac{3}{2} \end{displaymath} \end{enumerate} \end{enumerate} \subsection*{II - Cas continu.} \begin{enumerate} \item Lorsque $f$ est croissante et que son intégrale $I$ sur $[0,1]$ n'est pas nulle, on peut considérer la fonction $\overline{f}=f-I$. Cette fonction est encore croissante mais d'intégrale nulle. Si on peut lui appliquer l'inégalité, on tire \begin{displaymath} \int_0^1(f - I)g \geq 0 \Rightarrow \int_0^1fg \geq I\int_0^1g = \left( \int_0^1f\right) \left( \int_0^1g\right) \end{displaymath} \item \begin{enumerate} \item La fonction doit changer de signe dans l'intervalle ouvert car sinon son intégrale ne pourrait être nulle. Comme elle est continue, d'après le théorème des valeurs intermédiaires, elle doit prendre la valeur nulle. Elle est de plus strictement croissante elle ne prend donc qu'une seule fois cette valeur nulle. On note $a$ l'unique élément de l'intervalle en lequel elle s'annule. \item Comme $f$ est strictement croissante, elle est strictement négative avant $a$ et strictement positive après. La primitive $F$ est strictement décroissante entre $0$ et $a$ puis strictement croissante entre $a$ et $1$. On remarque que $A=F(a)<0$. Elle définit une bijection décroissante $F_1$ entre $[0,a]$ et $[A,0]$ et une bijection croissante $F_2$ entre $[a,1]$ et $[A,0]$. \item Changement de variable $u=F_1(t)$ dans $\int_0^af(t)g(t)\,dt$. \begin{itemize} \item[Bornes.] Quand $t$ est en $0$, $u$ est en $F_1(0)=F(0)=0$. Quand $t$ est en $a$, $u$ est en $F_1(a)=F(a)=A$. \item[\'Elément différentiel.] $u=F_1(t)$, $du = f(t)\,dt$. \item[Fonction.] $u=F_1(t)\Leftrightarrow t = \varphi_1(u)$, $g(t)=g(\varphi_1(u))$. \end{itemize} Le changement de variable s'écrit donc: \begin{displaymath} \int_0^af(t)g(t)\,dt = \int_{0}^{A}g(\varphi_1(u))\,du \end{displaymath} Noter que les bornes sont \og dans le mauvais sens\fg car $A<0$, cela traduit le fait que la bijection $F_1$ est décroissante. \end{enumerate} L'autre changement de variable est tout à fait analogue et conduit à \begin{displaymath} \int_a^1f(t)g(t)\,dt = \int_{A}^{0}g(\varphi_2(u))\,du \end{displaymath} \item On se place dans le cas particulier où l'intégrale de $f$ est nulle puisque cela suffit à montrer le ca général (d'après la question 1). On peut alors décomposer par la relation de Chasles et utiliser les changements de variable \begin{multline*} \int_0^1f(t)g(t)\,dt = \int_0^af(t)g(t)\,dt + \int_a^1f(t)g(t)\,dt = \int_{0}^{A}g(\varphi_1(u))\,du + \int_{A}^{0}g(\varphi_2(u))\,du\\ = \int_{A}^{0}\left( g(\varphi_2(u))- g(\varphi_1(u))\right) \,du \end{multline*} or $\varphi_1(u) \leq a \leq \varphi_2(u)$ et $g$ croissante entraîne $g(\varphi_2(u))- g(\varphi_1(u))\geq 0$ puis la positivité de l'intégrale. \end{enumerate} \subsection*{III - Relation de Lagrange.} \begin{enumerate} \item Deux couples symétriques $(i,j)$ et $(j,i)$ du carré $\mathcal{C}_n$ ont la même contribution à la somme. Les couples $(i,i)$ de la diagonale ont une contribution nulle. On en déduit que la somme étendue au carré est égale à deux fois la somme étendue au triangle $\mathcal{T}_n$ strictement au dessus de la diagonale. \item Développons la somme étendue au carré \begin{multline*} \sum_{(i,j)\in \mathcal{C}_n}(a_j-a_i)(b_j-b_i) = \sum_{(i,j)\in \mathcal{C}_n}(a_jb_j -a_ib_j -a_jb_i + a_ib_i)\\ = \sum_{i=1}^n\left( \sum_{j=1}^na_jb_j\right) - \sum_{i=1}^n \left(a_i \sum_{j=1}^nb_j\right) - \sum_{j=1}^n \left(b_j \sum_{i=1}^na_i\right) +\sum_{j=1}^n \left(\sum_{i=1}^na_ib_i\right) \\ = 2n \sum_{k=1}^n a_kb_k - 2\left(\sum_{k=1}^n a_k \right)\left(\sum_{k=1}^n b_k \right) \end{multline*} On en déduit la relation demandée en simplifiant par $2$ et en utilisant la question 1. \item Lorsque les familles sont monotones, on connait le signe des $(a_j-a_i)(b_j-b_i)$ pour $i1-10 @article{mendez2018investmentTFP, title={Investment constraints and productivity cycles in Bolivia}, author={}, journal={Coyuntural Economics}, volume={3}, number={4}, pages={31--54}, year={2018}, publisher={IIES} } docs/latex/classMPLWidget.tex \hypertarget{classMPLWidget}{}\section{M\+P\+L\+Widget Class Reference} \label{classMPLWidget}\index{M\+P\+L\+Widget@{M\+P\+L\+Widget}} provides G\+UI to all the functionalities of the library \subsection{Detailed Description} provides G\+UI to all the functionalities of the library The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item biomedical\+\_\+signal\+\_\+processing/tools/\hyperlink{func__plot__app_8py}{func\+\_\+plot\+\_\+app.\+py}\end{DoxyCompactItemize} OfficiumDivinum/SundayVespers \no{2}\evenverse Avertísti nos retrórsum post i\textit{ni}\textit{mí}\textit{cos} \textbf{no}stros:~\*\\ \evenverse et qui odérunt nos, diripié\textit{bant} \textbf{si}bi.\\ \no{3}\oddverse Dedísti nos tamquam \textit{o}\textit{ves} \textit{e}\textbf{scá}rum:~\*\\ \oddverse et in Géntibus di\textit{sper}\textbf{sí}sti nos.\\ \no{4}\evenverse Vendidísti pópulum tu\textit{um} \textit{si}\textit{ne} \textbf{pré}tio:~\*\\ \evenverse et non fuit multitúdo in commutatiónibus \textit{e}\textbf{ó}rum.\\ \no{5}\oddverse Posuísti nos oppróbrium \textit{vi}\textit{cí}\textit{nis} \textbf{no}stris,~\*\\ \oddverse subsannatiónem et derísum his, qui sunt in circúi\textit{tu} \textbf{no}stro.\\ \no{6}\evenverse Posuísti nos in simili\textit{tú}\textit{di}\textit{nem} \textbf{Gén}tibus:~\*\\ \evenverse commotiónem cápitis \textit{in} \textbf{pó}pulis.\\ \no{7}\oddverse Tota die verecúndia me\textit{a} \textit{con}\textit{tra} \textbf{me} est,~\*\\ \oddverse et confúsio faciéi meæ coopé\textit{ru}\textbf{it} me.\\ \no{8}\evenverse A voce exprobrántis, \textit{et} \textit{o}\textit{blo}\textbf{quén}tis:~\*\\ \evenverse a fácie inimíci, et per\textit{se}\textbf{quén}tis.\\ \no{9}\oddverse Hæc ómnia venérunt super nos, nec \textit{o}\textit{blí}\textit{ti} \textbf{su}mus te:~\*\\ \oddverse et iníque non égimus in testamén\textit{to} \textbf{tu}o.\\ \no{10}\evenverse Et non recéssit \textit{re}\textit{tro} \textit{cor} \textbf{no}strum:~\*\\ \evenverse et declinásti sémitas nostras a vi\textit{a} \textbf{tu}a:\\ \no{11}\oddverse Quóniam humiliásti nos in loco \textit{af}\textit{fli}\textit{cti}\textbf{ó}nis,~\*\\ \oddverse et coopéruit nos um\textit{bra} \textbf{mor}tis.\\ \no{12}\evenverse Glória \textit{Pa}\textit{tri}, \textit{et} \textbf{Fí}lio,~\*\\ \evenverse et Spirítu\textit{i} \textbf{San}cto.\\ \no{13}\oddverse Sicut erat in princípio, \textit{et} \textit{nunc}, \textit{et} \textbf{sem}per,~\*\\ \oddverse et in sǽcula sæculó\textit{rum}. \textbf{A}men.\\ \documentclass{article} \begin{document} \begin{figure}[tb] % \includegraphics{paper-count-w-2015-new} works, but the optional comment doesn't? \includegraphics[width=\columnwidth]{paper-count-w-2015-new} \end{figure} \end{document} %\textcolor{red}{Adaptive Mesh Refinement} \chapter{AST Processing} \label{AstProcessing:astProcessing} \fixme{ This chapter should cover both the classic Object-Oriented Visitor Pattern Traversal (which has yet to be build, but which I understand Markus is prepared to implement) and the two new traversals based on iteration over the memory pools which Dan built in the latest internal release of ROSE in December 2005. The three of these that are implemented are presenting in the ROSE Tutorial where there is a section for the unimplemented classic Object-Oriented Visitor Pattern Traversal not based on memory pools as a place holder (only the code need be updated).} \section{Introduction} \label{AstProcessing:introduction} ROSE aids the library writer by providing a traversal mechanism that visits all the nodes of the AST in a predefined order and to compute attributes. Based on a fixed traversal order, we provide inherited attributes for passing information down the AST (top-down processing) and synthesized attributes for passing information up the AST (bottom up processing). Inherited attributes can be used to propagate context information along the edges of the AST whereas synthesized attributes can be used to compute values based on the information of the subtree. One function for computing inherited attributes and one function for computing synthesized attributes must be implemented when attributes are used. We provide different interfaces that allow that both, only one, or no attribute is used; in the latter case it is a simple traveral with a visit method called at each node. The AST processing mechanism can be used to gather information about the AST, or ``query'' the AST. Only the functions that are invoked by the AST processing mechanism need to be implemented by the user of AstProcessing classes; no traversal code must be implemented. All 4 Ast*Processing classes provide three different functions for invoking a traversal on the AST: \begin{description} \item[T traverse(SgNode* node, ...)] : traverse full AST (including nodes which represent code from include files) \item[T traverseInputFiles(SgProject* projectNode, ...)] : traverse the subtree of the AST which represents the file(s) specified on the command line to a translator; files which are the 'input' to the translator. \item[T traverseWithinFile(SgNode* node, ...)] : traverse only those nodes which represent code of the same file where the traversal started. The traversal stays 'within' the file. \end{description} The return type {\tt T} and the other parameters are discussed for each {\tt Ast*Processing} class in the following sections. \section{AstSimpleProcessing} \label{AstProcessing:AstSimpleProcessing} This class is called 'Simple' because, in contrast to the other three processing classes, it does not provide the computation of attributes. It implements a traversal of the AST and calls a visit function at each node of the AST. This can be done as a preorder or postorder traversal. \begin{verbatim} typedef {preorder,postorder} t_traversalOrder; class AstSimpleProcessing { public: void traverse(SgNode* node, t_traversalOrder treeTraversalOrder); void traverseWithinFile(SgNode* node, t_traversalOrder treeTraversalOrder); void traverseInputFiles(SgProject* projectNode, t_traversalOrder treeTraversalOrder); protected: void virtual visit(SgNode* astNode)=0; }; \end{verbatim} To use the class AstSimpleProcessing the user needs to implement the function {\tt visit} for a user-defined class which inherits from class AstSimpleProcessing. To invoke a traversal one of the three traverse functions needs to be called. \subsection{Example (in preparation)} In the example we traverse the AST in preorder and print the name of each node in the order in which they are visited. The following steps are necessary: \begin{description} \item[Interface:] Create a class, 'MyVisitor', which inherits from AstSimpleProcessing. \item[Implementation:] Implement the function 'visit(SgNode* astNode)' for class 'MyVisitor' \item[Usage:] Create an object of type 'MyVisitor' and invoke the function traverse(SgNode* node, t\_traverseOrder treeTraversalOrder); \end{description} Interface: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyVisitor.h} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyVisitor.h} \end{htmlonly} \caption{Headerfile 'MyVisitor.h'.} \label{AstProcessing:myvisitor1} \end{figure} Implementation: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyVisitor.C} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyVisitor.C} \end{htmlonly} \caption{Implementation file 'MyVisitor.C'.} \label{AstProcessing:myvisitor1} \end{figure} Usage: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyVisitorMain.C} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyVisitorMain.C} \end{htmlonly} \caption{Example main program 'MyVisitorMain.C'.} \label{AstProcessing:myvisitor1} \end{figure} \section{AstTopDownProcessing} \label{AstProcessing:AstTopDownProcessing} This class allows to use a restricted form of inherited attributes to be computed for the AST. The user needs to implement the function evaluateInheritedAttribute. This function is called for each node when the AST is traversed. The inherited attributes are restricted such that a single attribute of a parent node is inherited by all its child nodes; i.e., the return value computed by the function {\tt evaluateInheritedValue} at the parent node is the input value to the function {\tt evaluateInheritedValue} at all child nodes. \begin{verbatim} template class AstTopDownProcessing { public: void traverse(SgNode* node, t_traversalOrder treeTraversalOrder); void traverseWithinFile(SgNode* node, t_traversalOrder treeTraversalOrder); void traverseInputFiles(SgProject* projectNode, t_traversalOrder treeTraversalOrder); protected: InheritedAttributeType virtual evaluateInheritedAttribute(SgNode* astNode, InheritedAttributeType inheritedValue)=0; }; \end{verbatim} The function {\tt evaluateInheritedAttribute} is called at each node. The traversal is a preorder traversal. \subsection{Example (in preparation)} In the example we traverse the AST and print the node names properly indented, according to the nesting level of C++ basic blocks. The function {\tt evaluateInheritedAttribute} is implemented and an inherited attribute is used to compute the nesting level. The following steps are necessary: \begin{description} \item[Interface:] Create a class, 'MyIndenting', which inherits from AstTopDownProcessing, and a class {\tt MyIndentLevel}. The latter will be used for attributes. Note that the constructor of the class {\tt MyIndentLevel} initializes the attribute value. \item[Implementation:] Implement the function {\tt evaluateInheritedAttribute(SgNode* astNode)} for class {\tt MyIndenting}. \item[Usage:] Create an object of type 'MyIndenting' and invoke the function traverse(SgNode* node, t\_traverseOrder treeTraversalOrder); \end{description} Interface: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyIndenting.h} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyIndenting.h} \end{htmlonly} \caption{Headerfile 'MyIndenting.h'.} \label{AstProcessing:myvisitor1} \end{figure} Implementation: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyIndenting.C} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyIndenting.C} \end{htmlonly} \caption{Implementation file 'MyIndenting.C'.} \label{AstProcessing:myvisitor1} \end{figure} Usage: \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/MyVisitorMain.C} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/MyIndentingMain.C} \end{htmlonly} \caption{Example main program 'MyIndentingMain.C'.} \label{AstProcessing:myvisitor1} \end{figure} Note that we could also use {\tt unsigned int} as attribute type in this simple example. But in general, the use of objects as attributes is more flexible and necessary, if you need to compute more than one attribute value (in the same traversal). \section{AstBottomUpProcessing} \label{AstProcessing:AstBottomUpProcessing} This class allows to use synthesized attributes. The user needs to implement the function evaluateSynthesizedAttribute to compute from a list of synthesized attributes a single return value. Each element in the list is the result computed at one of the child nodes in the AST. The return value is the synthesized attribute value computed at this node and passed upwards the AST. \begin{verbatim} template class AstBottomUpProcessing { public: SynthesizedAttributeType traverse(SgNode* node); SynthesizedAttributeType traverseWithinFile(SgNode* node); SynthesizedAttributeType traverseInputFiles(SgProject* projectNode); typedef vector SynthesizedAttributesList; protected: SynthesizedAttributeType virtual evaluateInheritedAttribute(SgNode* astNode, SynthesizedAttributesList synList)=0; SynthesizedAttributeType defaultSynthesizedAttribute(); }; \end{verbatim} The type {\tt SynthesizedAttributesList} is an STL container that holds the synthesized attribute values of the child nodes. Therefore an iterator can be used to operate on this list. This is necessary when the number of child nodes is arbitrary, e.g. in a SgBasicBlock the number of {\tt SgStatement} nodes which are child nodes, ranges from 0 to {\tt n}, where {\tt n = synList.size()}. For AST nodes with a fixed number of child nodes these values can be accessed by name, using enums defined for each AST node class. The naming scheme for attribute access is {\tt \_}. The method {\tt defaultSynthesizedAttribute} must be used to initialize attributes of primitive type (such as int, bool, etc.). This method is called when a synthesized attribute needs to be created for a non-existing subtree, i.e. when a node-pointer is null. A null pointer is never passed to an evaluate function. If a class is used to represent a synthesized attribute this method does not need to be implemented because the default constructor is called. In order to define an default value for attributes of primitive type, this method must be used. Note that there exist two cases when a default value is used for a synthesized attribute and the defaultSynthesizedAttribute method is called: \begin{itemize} \item When the traversal encounters a null-pointer it will not call an evaluate method but instead calls defaultSynthesizedAttribute. \item When the traversal skips nodes. For example traverseInputFiles only calls the evaluate method on nodes which represent the input-file(s) but skips all other nodes (of header files for example). \end{itemize} % \subsection{Example: Node pointers as synthesized attribute values} %In this example we show how to make all node pointers accessible as synthesized attributes. %TODO \subsection{Example: access of synthesized attribute by name} The enum definition used to access the synthesized attributes by name at a {\tt SgForStatement} node is \begin{verbatim} enum E_SgForStatement {SgForStatement_init_stmt, SgForStatement_test_expr_root, SgForStatement_increment_expr_root, SgForStatement_loop_body}; \end{verabatim} The definitions of the enums for all AST nodes can be found in the generated file \verb+/SAGE/Cxx_GrammarTreeTraversalAccessEnums.h+. For example, to access the synthesized attribute value of the SgForStatement's test-expression the synthesized attributes list is accessed using the enum definition for the test-expr. In the example we assign the pointer to a child node to a variable {\tt myTestExprSynValue}: \begin{verbatim} SgNode* myTestExprSynValue=synList[SgForStatement_test_expr_root].node; \end{verbatim} Note that for each node with a fixed number of child nodes, the size of the synthesized attributes value list is always of the same size, independent of whether the children exist or not. For example, for the SgForStatement it is always of size 4. If a child does not exist, the synthesized attribute value is the default value of the respective type used for the synthesized attribute (as template parameter). \section{AstTopDownBottomUpProcessing} \label{AstProcessing:AstTopDownBottomUpProcessing} This class combines all features from the above two classes. It allows to use inherited and synthesized attributes and therefore, the user needs to provide an implementation for two virtual functions, for evaluateInheritedAttribute and evaluateSynthesizedAttribute. The signature for evaluateSynthesizedAttribute has an inherited attribute as additional parameter. This allows to combine the results of inherited and synthesized attributes. You can use the inherited attribute that is computed at a node A by the evaluateInheritedAttribute method in the evaluateSynthesizedAttribute method at node A. But you cannot use synthesized attributes for computing inherited attributes (which is obvious from the method signatures). If such a data dependence needs to be represented member variables of the traversal object can be used to ``simulate'' such a behaviour to some degree. Essentially this allows to implement a pattern also called ``accumulation''. For example, building a list of all nodes of the AST can be implemented using this technique. %\section{Traversal Order And Attribute Access} %\label{AstProcessing:TraversalOrderAndAttributeAccess} %TODO: generate a latex version using tables, a text version, and a dot %version (without inheritance), and the inheritance graph %(=doxygenoutput) \begin{verbatim} InheritedAttributeType virtual evaluateInheritedAttribute(SgNode* astNode, InheritedAttributeType inheritedValue)=0; SynthesizedAttributeType traverse(SgNode* node, InheritedAttributeType inheritedValue); SynthesizedAttributeType defaultSynthesizedAttribute(InheritedAttributeType); \end{verbatim} \section{AST Node Attributes} To each node in the AST user-defined attributes can be attached ``by name''. The user needs to implement a class which inherits from AstAttribute. Instances of this class can be attached to an AST node by using member functions of SgNode::attribute. example: let node be a pointer to an object of type SgNode \begin{verbatim} class MyAstAttribute : public AstAttribute { public: MyAstAttribute(int v):value(v) {} ... private: int value; ... } node->attribute.setAttribute("mynewattribute",new MyAstAttribute(5)); \end{verbatim} Using above expression an attribute with name ``mynewattribute'' can be attached to the AST node pointed to by node. Similarily the same attribute can be accessed ``by name'' using the member function getAttribute: \begin{verbatim} MyAstAttribute* myattribute=node->attribute.getAttribute("mynewattribute"); \end{verbatim} AST attributes can be used to combine the results of different processing phases. Different traversals which are performed in sequence can store and read results to and from each node of the AST. For example, the first traversal may attache its results for each node as attributes to the AST and the second traversal can read and use these results. \section{Conclusions} All AST*Processing classes provide similar interfaces that differ only by the attributes used. AST node attributes can be used to attach data to each AST node and to share information between different traversals. Additional examples for traversal, attributes, pdf, and dot output can be found in \begin{itemize} \item \verb+ROSE/exampleTranslators/documentedExamples/astProcessingExamples+. \end{itemize} \section{Visualization} \subsection{Example Graphs} % Do this when processing latex to generate non-html (not using latex2html) \begin{figure} \begin{latexonly} \lstinputlisting{AstProcessing/astprocessingdoc_example1.C} \end{latexonly} % Do this when processing latex to build html (using latex2html) \begin{htmlonly} \verbatiminput{AstProcessing/astprocessingdoc_example1.C} \end{htmlonly} \caption{Example program used as running example} \label{AstProcessing:example1} \end{figure} \begin{figure} \centerline{\psfig{file=AstProcessing/astprocessingdoc_example1.Preorder.ps,height=1.5\linewidth,width=1.2\linewidth,angle=0}} \caption{Numbers at nodes show the order in which the visit function is called in a preorder traversal.} \label{AstProcessing:PreorderAst} \end{figure} \begin{figure} \centerline{\psfig{file=AstProcessing/astprocessingdoc_example1.Postorder.ps,height=1.0\linewidth,width=1.0\linewidth,angle=0}} \caption{Numbers at nodes show the order in which the visit function is called in a postorder traversal.} \label{introduction:PostorderAst} \end{figure} The graph shown in fig. \ref{AstProcessing:PreorderAst} is the AST of the program in fig. \ref{AstProcessing:example1}. Such an output can be generated for an AST with: \begin{verbatim} AstDOTGeneration dotgen; dotgen.generateInputFiles(projectNode, AstDOTGeneration::PREORDER); \end{verbatim} where {\tt projectNode} is a node of type {\tt SgProjectNode} and the order in which the AST is traversed is specified to be {\tt AstDOTGeneration::PREORDER} (or {\tt AstDOTGeneration::POSTORDER}). \begin{figure} \centerline{\psfig{file=AstProcessing/astprocessingdoc_example1.TopDown.ps,height=1.0\linewidth,width=1.0\linewidth,angle=0}} \caption{Numbers at nodes show the order in which the function evaluateInheritedAttribute is called in a top-down processing.} \label{AstProcessing:TopDownAst} \end{figure} \begin{figure} \centerline{\psfig{file=AstProcessing/astprocessingdoc_example1.BottomUp.ps,height=1.0\linewidth,width=1.0\linewidth,angle=0}} \caption{Numbers at nodes show the order in which the function evaluateSynthesizedAttribute is called in a bottom up processing.} \label{AstProcessing:BottomUpAst} \end{figure} \begin{figure} \centerline{\psfig{file=AstProcessing/astprocessingdoc_example1.TopDownBottomUp.ps,height=1.0\linewidth,width=1.0\linewidth,angle=0}} \caption{The pair of numbers at nodes shows the order in which the function evaluateInheritedAttribute (1st number) and evaluateSynthesizedAttribute (2nd number) is called in a top-down-bottom-up processing.} \label{AstProcessing:TopDownBottomUpAst} \end{figure} matcdac/IETF_RFCs @misc{rfc5126, series = {Request for Comments}, number = 5126, howpublished = {RFC 5126}, publisher = {RFC Editor}, doi = {10.17487/RFC5126}, url = {https://rfc-editor.org/rfc/rfc5126.txt}, author = { and and }, title = {{CMS Advanced Electronic Signatures (CAdES)}}, pagetotal = 141, year = 2008, month = mar, abstract = {This document defines the format of an electronic signature that can remain valid over long periods. This includes evidence as to its validity even if the signer or verifying party later attempts to deny (i.e., repudiates) the validity of the signature. The format can be considered as an extension to RFC 3852 and RFC 2634, where, when appropriate, additional signed and unsigned attributes have been defined. The contents of this Informational RFC amount to a transposition of the ETSI Technical Specification (TS) 101 733 V.1.7.4 (CMS Advanced Electronic Signatures -- CAdES) and is technically equivalent to it. The technical contents of this specification are maintained by ETSI. The ETSI TS and further updates are available free of charge at: http://www.etsi.org/WebSite/Standards/StandardsDownload.aspx This memo provides information for the Internet community.}, } % Datatracker information for RFCs on the Legacy Stream is unfortunately often % incorrect. Please correct the bibtex below based on the information in the % actual RFC at https://rfc-editor.org/rfc/rfc1375.txt @misc{rfc1375, series = {Request for Comments}, number = 1375, howpublished = {RFC 1375}, publisher = {RFC Editor}, doi = {10.17487/RFC1375}, url = {https://rfc-editor.org/rfc/rfc1375.txt}, author = {}, title = {{Suggestion for New Classes of IP Addresses}}, pagetotal = 7, year = 1992, month = oct, abstract = {This RFC suggests a change in the method of specifying the IP address to add new classes of networks to be called F, G, H, and K, to reduce the amount of wasted address space, and to increase the available IP address number space, especially for smaller organizations or classes of connectors that do not need or do not want a full Class C IP address. This memo provides information for the Internet community. It does not specify an Internet standard.}, } 0 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % GLE - Graphics Layout Engine % % % % Modified BSD License % % % % Copyright (C) 2009 GLE. % % % % Redistribution and use in source and binary forms, with or without % % modification, are permitted provided that the following conditions % % are met: % % % % 1. Redistributions of source code must retain the above copyright % % notice, this list of conditions and the following disclaimer. % % % % 2. Redistributions in binary form must reproduce the above % % copyright notice, this list of conditions and the following % % disclaimer in the documentation and/or other materials provided with % % the distribution. % % % % 3. The name of the author may not be used to endorse or promote % % products derived from this software without specific prior written % % permission. % % % % THIS SOFTWARE IS PROVIDED BY THE AUTHOR "AS IS" AND ANY EXPRESS OR % % IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED % % WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE % % ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY % % DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL % % DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE % % GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS % % INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER % % IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR % % OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN % % IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{Primitives} \label{prim:chap} A GLE command is a sequence of keywords and values separated by white space (one or more spaces or tabs). Each command must begin on a new line. Keywords may not be abbreviated, the case is not significant. All coordinates are expressed in centimeters from the bottom left corner of the page. GLE uses the concept of a {\bf current point} which most commands use. For example, the command {\sf aline 2 3} will draw a line from the {\bf current point} to the coordinates (2,3). The current graphics state also includes other settings like line width, colour, font, 2d transformation matrix. All of these can be set with various GLE commands. \section{Graphics Primitives (a summary)} {\sf ! {\it comment}} \\ {\sf @{\it xxx}} \\ {\sf abound {\it x y}} \\ {\sf aline {\it x y} [arrow start] [arrow end] [arrow both] [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]} \\ {\sf amove {\it x y}} \\ {\sf arc {\it radius a1 a2 [arrow end] [arrow start] [arrow both]}} \\ {\sf arcto {\it x1 y1 x2 y2 rad}} \\ {\sf begin box [fill {\it pattern}] [add {\it gap}] [nobox] [name {\it xyz}] [round {\it val}]} \\ {\sf begin clip } \\ {\sf begin length {\it var}} \\ {\sf begin name {\it name}} \\ {\sf begin object {\it name}} \\ {\sf begin origin} \\ {\sf begin path [stroke] [fill {\it pattern}] [clip]} \\ {\sf begin rotate {\it angle}} \\ {\sf begin scale {\it x y}} \\ {\sf begin table } \\ {\sf begin tex } \\ {\sf begin text [width {\it exp}] } \\ {\sf begin translate {\it x y}} \\ {\sf bezier {\it x1 y1 x2 y2 x3 y3}} \\ {\sf bitmap {\it filename width height} [type {\it type}]} \\ {\sf bitmap\_info {\it filename width height} [type {\it type}]} \\ {\sf box {\it x y} [justify {\it jtype}] [fill {\it color}] [name {\it xxx}] [nobox] [round {\it val}]} \\ {\sf circle {\it radius} [fill {\it pattern}]} \\ {\sf closepath } \\ {\sf colormap {\it fct} {\it xmin} {\it xmax} {\it ymin} {\it ymax} {\it pixels-x} {\it pixels-y} {\it width} {\it height} [color] [palette {\it pal}]} \\ {\sf curve {\it ix iy }[{\it x1 y1 x y x y ... xn yn}]{\it ex ey }} \\ {\sf define marker {\it markername subroutine-name}} \\ {\sf draw {\it name.point [{\it arg1} ... {\it argn}] [name {\it name}]}} \\ {\sf ellipse dx dy [options]} \\ {\sf elliptical\_arc dx dy theta1 theta2 [options]} \\ {\sf for {\it var} = {\it exp1} to {\it exp2} [step {\it exp3}] {\it command [...]} next {\it var}} \\ {\sf grestore} \\ {\sf gsave} \\ {\sf if {\it exp} then {\it command [...]} else {\it command [...]} end if} \\ {\sf include {\it filename}} \\ {\sf join {\it object1.just sep object2.just} [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]} \\ {\sf local {\it var}$_1$, $\ldots$, {\it var}$_n$} \\ {\sf margins {\it top} {\it bottom} {\it left} {\it right}} \\ {\sf marker {\it marker-name} [{\it scale-factor}]} \\ {\sf orientation {\it o}} \\ {\sf papersize {\it size}} \\ {\sf postscript {\it filename.eps width-exp height-exp}} \\ {\sf print {\it string\$} $\ldots$} \\ {\sf psbbtweak} \\ {\sf pscomment} {\it exp} \\ {\sf rbezier {\it x1 y1 x2 y2 x3 y3}} \\ {\sf return} {\it exp} \\ {\sf reverse } \\ {\sf rline {\it x y} [arrow end] [arrow start] [arrow both] [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]} \\ {\sf rmove {\it x y}} \\ {\sf save {\it objectname}} \\ {\sf set alabeldist {\it d}} \\ {\sf set alabelscale {\it s}} \\ {\sf set arrowangle {\sf angle}} \\ {\sf set arrowsize {\sf size}} \\ {\sf set arrowstyle {\sf simple | filled | empty}} \\ {\sf set atitledist {\it s}} \\ {\sf set atitlescale {\it s}} \\ {\sf set background {it c}} \\ {\sf set cap {\sf butt | round | square}} \\ {\sf set color {\it col}} \\ {\sf set dashlen {\it dashlen-exp}} \\ {\sf set fill {\it fill color/pattern}} \\ {\sf set font {\it font-name}} \\ {\sf set fontlwidth {\it line-width}} \\ {\sf set hei {\it character-size}} \\ {\sf set join {\sf mitre | round | bevel }} \\ {\sf set just left $|$ center $|$ right $|$ tl $|$ etc...} \\ {\sf set lstyle {\it line-style}} \\ {\sf set lwidth {\it line-width}} \\ {\sf set pattern {\it fill pattern}} \\ {\sf set texscale {\it scale} $|$ {\it fixed} $|$ {\it none}} \\ {\sf set titlescale {\it s}} \\ {\sf set ticksscale {\it s}} \\ {\sf size {\it w} {\it h}} \\ {\sf sub {\it sub-name parameter1 parameter2 etc}} \\ {\sf tex {\it string} [name {\it xxx}] [add {\it val}]} \\ {\sf text {\it unquoted-text-string}} \\ {\sf write {\it string\$} $\ldots$} \section{Graphics Primitives (in detail)} \begin{commanddescription} \item[{\sf ! {\it comment}}]\index{comment}\index{"!} Indicates the start of a comment. GLE ignores everything from the exclamation point to the end of the line. This works both in GLE scripts and in data files used in, e.g., graph blocks. \item[{\sf @{\it xxx}}] Executes subroutine {\it xxx}. \index{"@} \item[{\sf abound {\it x y}}] \index{abound} Update the current bounding box to include the point $(x,y)$ without drawing anything. This command is useful in combination with `begin box', `begin name', etc., e.g., to add empty space to the box. \item[{\sf aline {\it x y} [arrow start] [arrow end] [arrow both] [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]}] \index{aline} \index{arrow} Draws a line from the current point to the absolute coordinates {\it (x,y)}, which then becomes the new current point. The arrow qualifiers are optional, they draw arrows at the start or end of the line, the size of the arrow is proportional to the current font height. If the curve option is given, then a Bezier curve\index{Bezier curve}\index{curve} is drawn instead of a line. The first control point is located at a distance $d1$ and angle $\alpha1$ from the current point and the second control point is located at distance $d2$ and angle $\alpha2$ from {\it (x,y)}. \item[{\sf amove {\it x y}}] \index{amove} Changes the current point to the absolute coordinates {\it (x,y)}. \item[{\sf arc {\it radius a1 a2 [arrow end] [arrow start] [arrow both]}}] \index{arc}\index{narc} Draws an arc of a circle in the anti-clockwise direction, centered at the current point, of radius {\it radius}, starting at angle {\it a1} and finishing at angle {\it a2}. Angles are specified in degrees. Zero degrees is at three o'clock and Ninety degrees is at twelve o'clock. \preglecode{} \begin{Verbatim} arc 1.2 20 45 \end{Verbatim} \postglecode{} The command {\sf narc} is identical but draws the arc in the clockwise direction. This is important when constructing a path. \begin{minipage}[c]{8cm} \begin{Verbatim} amove 0.5 0.5 rline 1 0.5 arrow end set lwidth 0.1 arc 1 10 160 arc 0.5 -90 0 \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_arc}} \end{minipage} \item[{\sf arcto {\it x1 y1 x2 y2 rad}}] \index{arcto} Draws a line from the current point to {\it (x1,y1)} then to {\it (x2,y2)} but fits an arc of radius {\it rad} joining the two vectors instead of a vertex at the point {\it (x1,y1)}. \begin{minipage}[c]{8cm} \begin{Verbatim} amove 1.5 .5 rline 1 0 set lwidth .1 arcto 2 0 -1 1 .5 set lwidth 0 rline -1 1 \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_arcto}} \end{minipage} \item[{\sf begin {\it block\_name} ... {\it end block\_name}}] There are several block structured commands in GLE. Each {\sf begin} must have a matching {\sf end}. Blocks which change the current graphics state (e.g. scale, rotate, clip etc) will restore whatever they change at the end of the block. Indentation is optional but should be used to make the GLE program easier to read. \item[{\sf begin box [fill {\it pattern}] [add {\it gap}] [nobox] [name {\it xyz}] [round {\it val}]} ] \index{add} \index{begin!box} \index{nobox} \index{name (box)} \label{cmd:beginbox} Draws a box around everything between {\sf begin box} and {\sf end box}. The option {\sf add} adds a margin of {\sf margin} cm to each side of the box to make the box slightly larger than the area defined by the graphics primitives in the {\sf begin box} \ldots {\sf end box} group (to leave a gap around text for example). The option {\sf nobox} stops the box outline from being drawn. The {\sf name} option saves the coordinates of the box for later use with among others the {\sf join} command. If the {\sf round} option is used, a box with rounded corners will be drawn. \begin{minipage}[c]{8cm} \begin{Verbatim} begin box add 0.2 begin box fill gray10 add 0.2 round .3 text John end box end box \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_beginbox}} \end{minipage} \item[{\sf begin clip }] \index{clip} \index{begin!clip} This saves the current clipping region. A clipping region is an arbitrary path made from lines and curves which defines the area on which drawing can occur. This is used to undo the effect of a clipping region defined with the begin path command. See the example CLIP.GLE in appendix B at the end of the manual. \item[{\sf begin length {\it var}}] \index{length} \index{begin!length} This block computes the total length of all the elements that are included in it and saves the result in the variable ``{\it var}''. See Fig.~\ref{beginlen:fig} for an example. \begin{figure}[tb] \centering \mbox{\input{primitives/fig/curve_length.inc}} \caption{\label{beginlen:fig}Compute the total length of a shape.} \end{figure} \item[{\sf begin name {\it name}}] \index{begin!name}\label{cmd:beginname} Saves the coordinates of what is inside the block for later use with among others the {\sf join} command. This command is equivalent to `{\sf begin box name} $\ldots$ {\sf nobox}'. \item[{\sf begin object {\it name} [{\it arg1}, \ldots, {\it argn}]}] \index{begin!object}\label{cmd:beginobject} Declares a new object (sub-figure) that can be drawn later with the `{\sf draw}' command. Section~\ref{sec:objblocks} explains in detail how this command works and how it can be used. Object blocks can have arguments if they are not defined inside a subroutine. Such object blocks are called `static objects'; they behave similar to subroutines. Object blocks can also be defined inside a subroutine. In that case, they are called `dynamic objects' and cannot have arguments. They may, however, refer to all arguments and local variables of the surrounding subroutine. \item[{\sf begin origin}] \index{amove (origin)} \index{begin!origin} This makes the current point the origin. This is good for subroutines or something which has been drawn using {\sf amove,aline}. Everything between the {\sf begin origin} and {\sf end origin} can be moved as one unit. The current point is also saved and restored. \item[{\sf begin path [stroke] [fill {\it pattern}] [clip]} ] \index{begin!path} \index{path} \index{end path} \index{stroke} Initialises the drawing of a filled shape. All the lines and curves generated until the next {\sf end path} command will be stored and then used to draw the shape. {\sf stroke} draws the outline of the shape, {\sf fill} paints the inside of the shape in the given colour and {\sf clip} defines the shape as a clipping region for all future drawing. Clipping and filling will only work on PostScript devices. \item[{\sf begin rotate {\it angle}}] \index{rotate} \index{angle} \index{begin!rotate} The coordinate system is rotated anti-clockwise about the current point by the angle {\it angle} (in degrees). For example, to draw a line of text running vertically up the page (as a Y axis label, say), type: \begin{minipage}[c]{8cm} \begin{Verbatim} begin rotate 90 text This is end rotate \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_rot1}} \end{minipage} \item[{\sf begin scale {\it x y}}] \index{scale} \index{begin!scale} Everything between the {\sf begin} and {\sf end} is scaled by the factors x and y. E.g., {\it scale 2 3} would make the picture twice as wide and three times higher. \begin{minipage}[c]{8cm} \begin{Verbatim} begin scale 3 1 begin rotate 30 text This is end rotate end scale \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_rot2}} \end{minipage} \pagebreak[2] \item[{\sf begin table }] \index{table} \index{begin!table} This module is an alternative to the TEXT module. It reads the spaces and tabs in the source file and aligns the words accordingly. A single space between two words is treated as a real space, not an alignment space. With a proportionally spaced font columns will line up on the left hand side but not on the right hand side. However with a fixed pitch font, like {\sf tt}, everything will line up. \begin{minipage}[c]{8cm} \begin{Verbatim} begin table Here is my table of text see how 22 44 55 33 0.1 999 1 .2 3 33 2 33 it lines up end table \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_table}} \end{minipage} \item[{\sf begin text [width {\it exp}] } ] \index{text (width)} \index{text (begin)} \index{justify (text)} \index{begin!text} This module displays multiple lines/paragraphs of text. The block of text is justified according to the current justify setting. See the {\sf set just} command for a description of justification settings. If a width is specified the text is wrapped and justified to the given width. If a width is not given, each line of text is drawn as it appears in the file. Remember that GLE treats text in the same way that \LaTeX \ does, so multiple spaces are ignored and some characters have special meaning. E.g, \verb#\ ^ _ & { }# \index{Greek characters} To include Greek characters in the middle of text use a backslash followed by the name of the character. E.g., \verb+ 3.3\Omega S+ would produce ``3.3$\Omega$S''. To put a space between the Omega and the S add a backslash space at the end. E.g., \verb+ 3.3\Omega\ S+ produces ``3.3$\Omega$ S'' Sometimes the space control characters (e.g. \verb+\:+) are also ignored, this may happen at the beginning of a line of text. In this case use the control sequence \verb+\glass+ which will trick GLE into thinking it isn't at the beginning of a line. E.g., \preglecode{} \begin{Verbatim} text \glass \:\: Indented text \end{Verbatim} \postglecode{} \begin{minipage}[c]{8cm} \begin{Verbatim} set hei 0.25 just tl font tt begin text width 5 This is my paragraph of text to see if it wraps things at four cm as I have told it to do. end text ... begin text Now some text without a width specified end text \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_text}} \end{minipage} \index{char} \index{chardef} \index{def} \index{movexy} \index{setfont} \index{sethei} \index{baselineskip} \index{setstretch} \index{lineskip} \index{mathchar} \index{mathchardef} \index{mathcode} \index{TeX} \index{\LaTeX} There are several \LaTeX \ like commands which can be used within text. The complete list can be found in Appendix \ref{ltxsym:sec}. A few examples are: \begin{Verbatim} \ \' \v \u \= \^ \. \H \~ \'' Implemented TeX accents ^{} _{} Superscript, subscript \\ \_ Forced Newline, underscore character \, \: \; 0.5em, 1em, 2em space (em = width of the letter `m') \tex{expression} Any LaTeX expression \char{22} Any character in current font \glass Makes move/space work on beginning of line \rule{2}{4} Draws a filled in box, 2cm by 4cm \setfont{rmb} Sets the current text font \sethei{0.3} Sets the font height (in cm) \setstretch{2} Scales the quantity of glue between words \lineskip{0.1} Sets the default distance between lines of text \linegap{-1} Sets the minimum required gap between lines {\rm ...}, {\it ...} Sets roman, and italic font {\bf ...}, {\tt ...} Sets bold, and typewriter (monospaced) font \alpha, \beta, ... Greek symbols \end{Verbatim} \item[{\sf begin translate {\it x y}}] \index{translate} \index{begin!translate} Everything between the {\sf begin} and {\sf end} is moved x units to the right and y units up. \item[{\sf bezier {\it x1 y1 x2 y2 x3 y3}}] \index{bezier} Draws a B\'{e}zier cubic section from the current point to the point {\it (x3,y3)} with B\'{e}zier cubic control points at the coordinates {\it (x1,y1)} and {\it (x2,y2)}. For a full explanation of B\'{e}zier curves see the PostScript Language Reference Manual. % \item[{\sf bigfile {\it filename.gle}}] % \index{bigfile} \index{include (bigfile)} % This command reads the file one line at a time, compiles each line and executes it. This means it can read any sized file. However, complex multi-line commands cannot be used. Subroutines can be used but not defined, inside the bigfile. Note: there is also a bigfile option in the graphing module for large datasets. \item[{\sf bitmap {\it filename width height} [type {\it type}]}] \index{bitmap} Imports the bitmap \textit{filename}. The bitmap is scaled to \textit{width}$\times$\textit{height}. If one of these is zero, it is computed based on the other one and the aspect ratio of the bitmap. GLE supports TIFF, JPEG, PNG and GIF bitmaps (depending on the compilation options). Bitmaps are compressed automatically by GLE using either the LZW or the JPEG compression scheme. \item[{\sf bitmap\_info {\it filename width height} [type {\it type}]}] \index{bitmap\_info} Returns the dimensions in pixels of the bitmap in the output parameters \textit{width} and \textit{height}. \item[{\sf box {\it x y} [justify {\it jtype}] [fill {\it color}] [name {\it xxx}] [nobox] [round {\it val}]} ] \index{box} \index{nobox} \index{justify (box)} \index{name (box)} Draws a box, of width {\it x} and height {\it y}, with its bottom left corner at the current point. If the justify option is used, the box will be positioned relative to the specified point. E.g., TL = top left, CC = center center, BL = bottom left, CENTER = bottom center, RIGHT = bottom right, LEFT = bottom left. See {\sf set just} for a description of justification settings. If a fill pattern is specified, the box will be filled. Remember that white fill is different from no fill pattern - white fill will erase anything that was inside the box. If the {\sf round} option is used, a box with rounded corners will be drawn. \item[{\sf circle {\it radius} [fill {\it pattern}]} ] \index{circle} \index{radius} Draws a circle at the current point, with radius {\it radius}. If a fill pattern is specified the circle will be filled. \item[{\sf closepath }] \index{closepath} \index{aline (closepath)} Joins the beginning of a line to the end of a line. I.e., it does an {\sf aline} to the end of the last {\sf amove}. \item[{\sf colormap {\it fct} {\it xmin} {\it xmax} {\it ymin} {\it ymax} {\it pixels-x} {\it pixels-y} {\it width} {\it height} [color] [palette {\it pal}]}] \index{colormap!command} Draws a colormap of the function {\it fct}$(x,y)$, in which $x$ ranges from {\it xmin} to {\it xmax}, and $y$ ranges from {\it ymin} to {\it ymax}. The size of the colormap is {\it width} by {\it height} centimeter and the resolution is {\it pixels-x} by {\it pixels-y} pixels. A colormap is grayscale by default; it is drawn in color if the option {\it color} is given. In the latter case, it is possible to specify a palette subroutine {\it pal} mapping the range $0 \ldots 1$ to a range of colors. This command is similar to the colormap command in a graph block (Sec.~\ref{colormap}). \item[{\sf curve {\it ix iy }[{\it x1 y1 x y x y ... xn yn}]{\it ex ey }} ] \index{curve} Draws a curve starting at the current point and passing through the points {\it (x1,y1)} \ldots {\it (xn,yn)}, with an initial slope of {\it (ix,iy)} to {\it (x1,y1)} and a final slope of {\it (ex,ey)}. All the vectors are relative movements from the vector before. \begin{minipage}[c]{8cm} \begin{Verbatim} amove 1 1 curve 1 0 0 1 1 0 0 -1 1 0 amove 3.6 1 curve 0 1 0 1 1 0 0 -1 0 -1 \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_curve}} \end{minipage} \item[{\sf define marker {\it markername subroutine-name}}] \index{define marker} This defines a new marker called {\it markername} which will call the subroutine {\it subroutine-name} whenever it is used. It passes two parameters, the first is the requested size of the marker and the second is a value from a secondary dataset which can be used to vary size or rotation of a marker for each point plotted. To define a character from the postscript ZapDingbats font as a marker you would use, e.g. \begin{Verbatim} sub subnamex size mdata gsave ! save font and x,y set just left font pszd hei size t$ = "\char{102}" rmove -twidth(t$)/2 -theight(t$)/2 ! centers marker write t$ grestore ! restores font and x,y end sub \end{Verbatim} The second parameter can be supplied using the {\it mdata} command when drawing a graph, this gives the marker subroutine a value from another dataset to use to draw the marker. For example the marker could vary in size, or angle, with every one plotted. \begin{Verbatim} d3 marker myname mdata d4 \end{Verbatim} \item[{\sf define {\it markername fontname scale dx dy}}] This command defines a new marker, from any font, it is automatically centered but can be adjusted using dx,dy. e.g. \begin{Verbatim} defmarker hand pszd 43 1 0 0 \end{Verbatim} \item[{\sf draw {\it name.point [{\it arg1} ... {\it argn}] [name {\it name}]}}] \label{cmd:draw} Draws a named object block that has been previously defined using a ``begin/end object'' (p.~\pageref{cmd:beginobject}) construct. The object is drawn such that the point indicated by the first argument of the draw command appears at the current position. The point can be any (hierarchically) named point on the object and may include the justify options defined for the join command (p.~\pageref{cmd:join}). If the object block has parameters (similar to a subroutine) then these parameters can be given as {\it arg1} \ldots {\it argn}. The ``draw'' command names the object using the same name as the name of the object block by default. An alternative name can be supplied using its ``name'' option. See Sec.~\ref{sec:objblocks} for a detailed explanation of this command with examples. \item[{\sf ellipse} {\it dx dy [options]}] \index{ellipse} This command draws an ellipse with the diameters {\it dx} and {\it dy} in the $x$ and $y$ directions, respectively. The {\it options} are the same as the {\sf circle} command. \item[{\sf elliptical\_arc} {\it dx dy theta1 theta2 [options]}] \index{elliptical\_arc}\index{elliptical\_narc} This command is similar to the {\sf arc} command except that it draws an elliptical arc in the clockwise direction with the diameters {\it dx} and {\it dy} in the $x$ and $y$ directions, respectively. {\it theta1} and {\it theta2} are the start and stop angle, respectively. The {\it options} are the same as for the {\sf arc} command. The command {\sf elliptical\_narc} is identical but draws the arc in the clockwise direction. This is important when constructing a path. \item[{\sf for {\it var} = {\it exp1} to {\it exp2} [step {\it exp3}] {\it command [...]} next {\it var}} ] \index{for} \index{next} \index{step} The {\sf for ... next} structure lets you repeat a block of statements a number of times. GLE sets {\sf var} equal to {\it exp1} and then repeats the following steps. \begin{itemize} \item If {\sf var} is greater than {\it exp2} then GLE commands are skipped until the line after the {\sf next} statement. \item The value {\it exp3} is added to {\sf var}. \item The statements between the {\sf for} and {\sf next} statement are executed. \end{itemize} If {\it exp1} is greater than {\it exp2} then the loop is not executed. \begin{minipage}[c]{8cm} \begin{Verbatim} for x = 1 to 4 step 0.5 amove x 1 aline 5-x 2 next x \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_for}} \end{minipage} \item[{\sf grestore}] \index{grestore} Restores the most recently saved graphics state. This is the simplest way to restore complicated transformations such as rotations and translations. It must be paired with a previous {\sf gsave} command. \item[{\sf gsave}] \index{gsave} Saves the current graphics transformation matrix and the current point and the current colour, font etc. \item[{\sf if {\it expression} then {\it command [...]} else {\it command [...]} end if}] \index{if}\index{else}\index{then}\index{end if} If {\it expression} evaluates to true, then execution continues with the statements up to the corresponding {\sf else}, otherwise the statements following the {\sf else} and up to the corresponding {\sf end if} are executed. \preglecode{} \begin{Verbatim} amove 3 3 if xpos()=3 then text We are at x=3 else text We are elsewhere end if \end{Verbatim} \postglecode{} Note: {\sf end if} is not spelt {\sf endif}. \item[{\sf include {\it filename}}]\label{incl:cmnd} \index{include} Includes the GLE script ``filename'' into the current script. This is useful for including library scripts with subroutines. GLE searches a number of predefined directories for include files. By default, this includes the current directory and the ``lib'' or ``gleinc'' subdirectory of the root directory (GLE\_TOP) of your GLE installation. The latter includes a number of subroutine files that are distributed with GLE (Table~\ref{inc:tab}). Additional include directories can be defined by means of the environment variable GLE\_USRLIB\index{GLE\_USRLIB}. \begin{table}[t] \centering \caption{\label{inc:tab}Include files distributed with GLE.} \begin{tabular}{ll} \hline barstyles.gle & Defines additional styles for bar plots.\\ color.gle & Defines functions for working with colors.\\ colors-gle-4.0.12.gle & Redefines all colors defined in GLE 4.0.12 and before.\\ contour.gle & Subroutines for drawing contour plots\\ electronics.gle & Subroutines for drawing electrical circuits\\ ellipse.gle & Draw text in an ellipse\\ feyn.gle & Subroutines for drawing Feynmann diagrams\\ graphutil.gle & Subroutines for drawing graphs\\ piesub.gle & Pie chart routines\\ polarplot.gle & Polar plotting routines\\ shape.gle & Drawing various shapes\\ simpletree.gle & Draw simple trees\\ stm.gle & Add labels to images\\ ziptext.gle & Draw zipped text\\ \hline \end{tabular} \end{table} % With a large include file GLE may run out of memory. If this happens, use the {\sf bigfile} command instead of {\sf include}. Note: there is also a bigfile option in the graphing module. \item[{\sf join {\it object1.just sep object2.just} [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]}] \index{name (join)}\index{join}\label{cmd:join} Draws a line between two named objects. An object is simply a point or a box which was given a name when it was drawn. \index{justify (join)} The justify qualifiers are the standard GLE justification abbreviations: \verb#.br# (bottom right), \verb#.bl# (bottom left), \verb#.bc# (bottom centre), \verb#.tr# (top right), \verb#.tc# (top centre), \verb#.tl# (top left), \verb#.cr# (centre right), \verb#.cc# (centre centre), and \verb#.cl# (centre left). In addition, \verb#.v# and \verb#.h# can be used to draw vertical or horizontal lines connecting to the object, \verb#.c# for drawing a line connecting to e circle or ellipse, and \verb#.box# for drawing a line to a rectangle. Fig.~\ref{joicur:fig} shows examples of the different cases. If {\it sep} is written as {\sf -}, a line is drawn between the named objects e.g. \begin{Verbatim} join fred.tr - mary.tl \end{Verbatim} Arrow heads can be included at both ends of the line by writing {\it sep} as \verb#<->#. Single arrow heads are produced by \verb#<-# and \verb#->#. Note that {\it sep} must be separated from object1.just and object2.just by white space. If the justification qualifiers are omitted, a line will be drawn between the centers of the two objects (clipped at the edges of the rectangles which define the objects). This is the same as using the \verb#.box# qualifier on both objects. The {\sf curve} option is explained with the {\sf aline} command. Fig.~\ref{joicur:fig} shows an example where the ``join'' command is used with the curve option. Sec.~\ref{join:sec} contains several examples of joining objects. \item[{\sf local {\it var}$_1$, $\ldots$, {\it var}$_n$}] \index{local} Defines a local variable inside a subroutine. It is possible to initialize the variable to a particular value with, e.g., `\texttt{local x = 3}', which defines the local variable `x' and assigns it the value 3. You can also define several local variables at once, e.g., `\texttt{local x, y}' defines the local variables `x' and `y'. \item[{\sf margins {\it top} {\it bottom} {\it left} {\it right}}] This command can be used to define the page margins. Margins are only relevant for making full-page figures (using the -fullpage command line option). See also the ``papersize'' command. \index{scale (marker)} \item[{\sf marker {\it marker-name} [{\it scale-factor}]} ] \index{wmarker} \index{marker} Draws marker {\it marker-name} at the current point. The size of the marker is proportional to the current font size, scaled by the value of {\it scale-factor} if present. Markers are referred to by name, eg. {\sf square}, {\sf diamond}, {\sf triangle} and {\sf fcircle}. Markers beginning with the letter {\sf f} are usually filled variants. Markers beginning with {\sf w} are filled with white so lines are not visible through the marker. For a complete list of markers refer to Fig.~\ref{mark:fig}. \begin{Verbatim} set just lc amove 0.5 2.5 marker diamond 1 rmove 0.6 0; text Diamond amove 0.5 2 marker triangle 1 rmove 0.6 0; text Triangle ... \end{Verbatim} \item[{\sf orientation {\it o}}]\index{orientation}\label{orient:cmd} Sets the orientation of the output in full-page mode. Possible values are ``portrait'' and ``landscape''. Fig.~\ref{fullpage:fig} illustrates these two cases. \begin{figure}[tb] \centering \mbox{\input{primitives/fig/gc_marker.inc}} \caption{\label{mark:fig}All markers supported by GLE. (The names that start with ``w'' are white filled.)} \end{figure} \item[{\sf papersize {\it size}}]\index{papersize}\label{papsiz:cmd} \item[{\sf papersize {\it width} {\it height}}] Sets the paper size of the output. This is used only when GLE is run with the option ``-fullpage'' or when the PostScript output device is used (i.e., ``-d ps''). The command either takes one argument, which should be one of the predefined paper size names or two numbers, which give the width and height of the output measured in cm. The following paper sizes are known by GLE: a0paper, a1paper, a2paper, a3paper, a4paper, and letterpaper. If a ``size'' command is given in the script, then the output is drawn centered on the page. If no size command is included in the script, then the output will appear relative to the bottom-left corner of the page, offset by the page margins (see ``margins'' command). Fig.~\ref{fullpage:fig} illustrates these two cases. The paper size can also be set in GLE's configuration file (Sec.~\ref{conffile:sec}). \begin{figure}[tb] \centering \mbox{\input{primitives/fig/fullpage.inc}} \caption{\label{fullpage:fig}Result of different combinations of the commands ``papersize'', ``margins'', ``size'', and ``orientation'' for fullpage graphics (gle -fullpage figure.gle).} \end{figure} \item[{\sf postscript {\it filename.eps width-exp height-exp}} ] \index{postscript} Includes an encapsulated postscript file into a GLE picture, the postscript picture will be scaled up or down to fit the width given. On the screen you will just see a rectangle. Only the {\it width-exp} is used to scale the picture so that the aspect ratio is maintained. The height is only used to display a rectangle of the right size on the screen. \item[{\sf print {\it string\$} $\ldots$}] \index{print} This command prints its argument to the console (terminal). \item[{\sf psbbtweak}] \index{psbbtweak} Changes the default behavior of the bounding box. The default behavior is to have the lower corner at (-1,-1), which for some interpreters (i.e., Photoshop) will leave a black line around the bottom and left borders. If this command is specified then the origin of the bounding box will be set to (0,0). This command must appear before the first {\sf size} command in the GLE file. \begin{figure} \centering \mbox{\input{primitives/fig/curve.inc}} \caption{\label{joicur:fig}Different ways of joining objects.} \end{figure} \item[{\sf pscomment} {\it exp}] \index{pscomment} Allows inclusion of {\it exp} as a comment in the preamble of the postscript file. Multiple {\sf pscomment} commands are allowed. This command must appear before the first {\sf size} command in the GLE file. \item[{\sf rbezier {\it x1 y1 x2 y2 x3 y3}}] \index{bezier (rbezier)} \index{rbezier} This command is identical to the BEZIER command except that the points are all relative to the current point. \begin{minipage}[c]{8cm} \begin{Verbatim} amove 0.5 2.8 rbezier 1 1 2 -1 3 1 amove 0.2 0.2 rbezier 1 1 2 1.2 1.8 0 \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_bezier}} \end{minipage} \item[{\sf return} {\it exp}] \index{return} The {\sf return} command is used inside subroutines to return a value. \item[{\sf reverse }] \index{reverse} Reverses the direction of the current path. This is used when filling multiple paths in order that the Non-Zero Winding Rule will know which part of the path is `inside'. With the Non-Zero Winding Rule an imaginary line is drawn through the object. Every time a line of the object crosses it from left to right, one is added to the counter; every time a line of the object crosses it from right to left, one is subtracted from the counter. Everywhere the counter is non-zero is considered to be the `inside' of the drawing and is filled. % \graphin{gc_nonzero.eps}{12.0cm}{3.0cm}{1.0} \psgraphin{primitives/fig/gc_nonzero}{12.0}{3.0}{\sf reverse} \item[{\sf rline {\it x y} [arrow end] [arrow start] [arrow both] [curve {\it $\alpha1$} {\it $\alpha2$} {\it d1} {\it d2}]}] \index{rline} \index{arrow} Draws a line from the current point to the relative coordinates {\it (x,y)}, which then become the new current point. If the current point is (5,5) then {\sf rline 3 -2} is equivalent to {\sf aline 8 3}. The optional qualifiers on the end of the command will draw arrows at one or both ends of the line, the size of the arrow head is proportional to the current font size. The {\sf curve} option is explained with the {\sf aline} command. \item[{\sf rmove {\it x y}}] \index{rmove} Changes the current point to the relative coordinate {\it (x,y)}. If the current point is (5,5) then {\sf rmove 3 -2} is equivalent to {\sf amove 8 3}. \item[{\sf save {\it objectname}} ] \index{save} \index{name (point)} This command saves a point for later use with the join command. \item[{\sf set alabeldist {\it d}}] \index{alabeldist} The spacing between the graph axis labels and the axis is set to {\it d}. \item[{\sf set alabelscale {\it s}}] \index{alabelscale} The graph axis label font size is set to `\texttt{alabelscale}' times `\texttt{hei}'. \item[{\sf set arrowangle {\sf angle}}] \index{arrowangle} Sets the opening angle of the arrow tips. (Actually, half of the opening angle.) \item[{\sf set arrowsize {\sf size}}] \index{arrowsize} Sets the length of the arrow tips in centimeter. \item[{\sf set arrowstyle {\sf simple | filled | empty}}] \index{arrowstyle} Sets the style of the arrow tips. There are three pre-defined styles: simple, filled, and empty (See Fig.~\ref{arrsty:fig}). It is also possible to create user-defined arrow tip styles. To do so, create a subroutine `{\sf arrow\_xxxx langle aangle asize}', with {\sf xxxx} the name of the new style. The parameter {\sf langle} is the angle of the line on which the arrow tip is to be drawn and the parameters {\sf aangle} and {\sf asize} are the current values of the settings {\sf arrowangle} and {\sf arrowsize}. The user-defined style can be enabled, in the same way as the built-in ones, with `{\sf set arrowstyle xxxx}'. Fig.~\ref{arrsty:fig} shows the three predefined styles and a user-defined tip style that is defined by the following subroutine: \begin{Verbatim} sub arrow_circle langle aangle asize circle 0.1 fill red end sub \end{Verbatim} \noindent{}More complex examples of user-defined arrow styles can be found in the GLE example repository. \begin{figure} \centering \includegraphics{primitives/fig/gc_arrstyle} \caption{\label{arrsty:fig}Different arrow tip styles.} \end{figure} \item[{\sf set atitledist {\it s}}] \index{atitledist} The spacing between the graph axis title and the axis labels is set to {\it d}. \item[{\sf set atitlescale {\it s}}] \index{atitlescale} The graph axis title font size is set to `\texttt{atitlescale}' times `\texttt{hei}'. \item[{\sf set background {it c}}] Set the background color for a pattern fill to $c$. (See `set fill'.) Note that ``set background'' must come after ``set fill'' because ``set fill'' resets the background color to the specified color. \item[{\sf set cap {\sf butt | round | square}}] \index{round (cap)} \index{cap} Defines what happens at the end of a wide line. \psgraphin{primitives/fig/gc_cap}{12.0}{3.0}{\sf set cap} \item[{\sf set color {\it col}}] \label{scol:cmd} \index{color} \index{rgb()} \index{rgb255()} Sets the current colour for all future drawing operations. GLE supports all SVG/X11 standard color names. These are listed in Appendix~\ref{colors:sec}, and include the following: black, white, red, green, blue, cyan, magenta, yellow, gray10, gray20, $\ldots$, gray90. It is also possible to specify a gray scale as a real number with 0.0 = black and 1.0 = white. Colors can also be set using the HTML notation, e.g., \#FF0000 = red. Finally, the functions rgb(red,green,blue) and rgb255(red,green,blue) may be used to create custom colors. Fig.~\ref{colex:fig} gives some examples. \begin{figure} \centering \mbox{\input{primitives/fig/setcolor.inc}} \caption{\label{colex:fig}Examples of setting the drawing color.} \end{figure} \vspace{0.25cm} \begin{minipage}[c]{8cm} \begin{Verbatim} mm$ = "blue" amove 0.5 0.5 for c = 0 to 1 step 0.05 box 0.2 2 fill (c) nobox rmove 0.2 0 next c amove 2 1 box 2 1 fill white nobox rmove -0.2 0.2 box 2 1 fill mm$ \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_color}} \end{minipage} \item[{\sf set dashlen {\it dashlen-exp}}] \index{dashlen} Sets the length of the smallest dash used for the line styles. This command MUST come before the {\sf set lstyle} command. This may be needed when scaling a drawing by a large factor. \begin{figure} \centering \mbox{\input{primitives/fig/grids.inc}} \caption{\label{filpat:fig}Patterns for painting shapes.} \end{figure} \item[{\sf set fill {\it fill color/pattern}}] \index{fill} Sets the color or pattern for filling shapes. This command works in combination with shapes such as circles, ellipses, and boxes. If the argument is a color, then shapes are filled with the given color (see ``set color''). If it is a pattern, then the shapes are painted with the given pattern in black ink. Fig.~\ref{filpat:fig} lists a number of pre-defined patterns. To paint a shape in a color different from black, first set the color, then the pattern. That is, \begin{Verbatim} set fill red set pattern shade set background yellow box 2 2 \end{Verbatim} \noindent{}will draw a box and paint is using the shade pattern and red ink on a yellow background. To draw shapes that are not filled, use the command ``set fill clear''. That is, \begin{Verbatim} set fill clear box 2 2 \end{Verbatim} \noindent{}will draw an empty box. \item[{\sf set font {\it font-name}}] \index{font} \index{font} Sets the current font to {\it font-name}. Valid {\it font-name}s are listed in Appendix A.2. There are three types of font: PostScript, \LaTeX \ and Plotter. They will all work on any device, however \LaTeX \ fonts are drawn in outline on a plotter, and so may not look very nice. PostScript fonts will be emulated by \LaTeX \ fonts on non-PostScript printers. \item[{\sf set fontlwidth {\it line-width}}] \index{font (line width)} \index{fontlwidth} \index{fontlwidth} This sets the width of lines to be used to draw the stroked (Plotter fonts) on a PostScript printer. This has a great effect on their appearance. \begin{minipage}[c]{8cm} \begin{Verbatim} set font pltr amove .2 .2 text Tester set fontlwidth .1 set cap round rmove 0 1.5 text Tester \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_fontlwidth}} \end{minipage} \item[{\sf set hei {\it character-size}}] \label{shei:cmd} \index{hei} \index{character size} Sets the height of text. For historical reasons, concerning lead type and printing conventions, a height of 10cm actually results in capital letters about 6.5cm tall. The default value of ``hei'' is 0.3633 (to mimic the default height of \LaTeX{} expressions). \item[{\sf set join {\sf mitre | round | bevel }}] \index{round (join)} \index{join} \index{join (set join)} \index{bevel} \index{round} \index{mitre} Defines how two wide lines will be joined together. With {\sf mitre}, the outside edges of the join are extended to a point and then chopped off at a certain distance from the intersection of the two lines. With {\bf round}, a curve is drawn between the outside edges. %%\graphin{gc_ljoin.eps}{12.0cm}{3.0cm}{1.0} \psgraphin{primitives/fig/gc_ljoin}{12.0}{3.0}{\sf set join} \item[{\sf set just left $|$ center $|$ right $|$ tl $|$ etc...} ] \label{sjust:cmd} \index{just} \index{justify (set)} Sets the justification which will be used for {\sf text} commands. \begin{minipage}[c]{8cm} \begin{Verbatim} amove 0.5 3 set just left box 1.5 0.6 text Justify left rmove 2 0 set just bl box 1.5 0.6 text Justify bl \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_justify}} \end{minipage} \item[{\sf set lstyle {\it line-style}}] \label{lstyle:cmd} \index{lstyle (set)} \index{lstyle} Sets the current line style to line style number {\sf line-style}. There are 9 predefined line styles (1--9). When a line style is given with more than one digit the first digit is read as a run length in black, the second a run length in white, the third a run length in black, etc. \begin{minipage}[c]{8cm} \begin{Verbatim} set just left for z = 0 to 4 set lstyle z rline 2 0 rmove 0.1 0 write z rmove -2.1 -0.4 next z \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_lstyle}} \end{minipage} \item[{\sf set lwidth {\it line-width}}] \index{lwidth} Sets the width of lines to {\it line-width} cm. A value of zero will result in the device default of about 0.02 cm, so a lwidth of .0001 gives a thinner line than an lwidth of 0. \item[{\sf set pattern {\it fill pattern}}] \index{pattern} Specifies the filling pattern. A number of pre-defined patterns is listed in Fig.~\ref{filpat:fig}. See the description of ``set fill'' for more information. Note that ``set pattern'' must come after ``set fill'' because ``set fill'' resets the pattern to solid. \item[{\sf set texscale {\it scale} $|$ {\it fixed} $|$ {\it none}}] \index{texscale} This setting controls the scaling of \LaTeX{} expressions (Sec.~\ref{latexexp:sec}): `\texttt{scale}' scales them to the value of `\texttt{hei}', `\texttt{fixed}' scales them to the closest \LaTeX{} default size to `\texttt{hei}', and `\texttt{none}' does not scale them. With `\texttt{none}', the font size in your graphics will be exactly the same as in your main document. \item[{\sf set titlescale {\it s}}] \index{titlescale} The graph title font size is set to `\texttt{titlescale}' times `\texttt{hei}'. \item[{\sf set ticksscale {\it s}}] \index{ticksscale} The size of the graph axis ticks is set to `\texttt{ticksscale}' times `\texttt{hei}'. \item[{\sf size {\it w} {\it h}}] Sets the size of GLE's output to {\it w} centimeter wide by {\it h} centimeter tall. This command usually appears at the top of a GLE script. That is, only commands that do not generate output can precede the `size' command. For example, the `include' command, subroutine definitions, and assignments to variables can appear before the `size' command. Commands like `aline', on the other hand, should appear after the `size' command. It is possible to omit the size command. In that case, the size of the output is determined by the `pagesize' command (see Fig.~\ref{fullpage:fig}). \item[{\sf sub {\it sub-name parameter1 parameter2 etc.}}] \index{sub} Defines a subroutine. The end of the subroutine is denoted with {\sf end sub}. Subroutines must be defined before they are used. Subroutines can be called inside any GLE expression, and can also return values. The parameters of a subroutine become local variables. Subroutines are re-entrant. \begin{Verbatim} sub tree x y a$ amove x y rline 0 1 write a$ return x/y end sub tree 2 4 "mytree" (Normal call to subroutine) slope = tree(2,4,"mytree") (Using subroutine in an expression) \end{Verbatim} \item[{\sf tex {\it string} [name {\it xxx}] [add {\it val}]}] \index{tex} Draw a \LaTeX{} expression at the current point using the current value of `justify'. See Sec.~\ref{latexexp:sec} for more information. Using the {\sf name} option, the \LaTeX{} expression can be named, just like a box. The size of the virtual named box can be increased with the {\sf add} option. \item[{\sf text {\it unquoted-text-string}}] \index{begin!text (single line)} \index{text} This is the simplest command for drawing text. The current point is unmodified after the text is drawn so following one text command with another will result in the second line of text being drawn on top of the first. To generate multiple lines of text, use the {\sf begin text} \ldots {\sf end text} construct. \preglecode{} \begin{Verbatim} text "Hi, how's tricks", said Jack! \end{Verbatim} \postglecode{} \item[{\sf write {\it string\$} $\ldots$}] \index{write} This command is similar to {\sf text} except that it expects a quoted string, string variable, or string expression as a parameter. If write has more than one parameter, it will concatenate the values of all the parameters. \begin{minipage}[c]{8cm} \begin{Verbatim} a$ = "Hello there " xx = sqrt(10) t$ = time$() c$ = a$+t$ write a$+t$ xx \end{Verbatim} \end{minipage} \hfill \begin{minipage}[c]{7cm} \mbox{\includegraphics{primitives/fig/gc_write}} \end{minipage} The built in functions {\sf sqrt()} and {\sf time\$()} are described in Appendix~\ref{fct:sec}. \end{commanddescription} @inproceedings{Chollet:2018:IID:3267851.3267874, author = { , }, title = {Influence of Individual Differences when Training Public Speaking with Virtual Audiences}, booktitle = {Proceedings of the 18th International Conference on Intelligent Virtual Agents}, series = {IVA '18}, year = {2018}, isbn = {978-1-4503-6013-5}, location = {Sydney, NSW, Australia}, pages = {1--7}, numpages = {7}, url = {http://doi.acm.org/10.1145/3267851.3267874}, doi = {10.1145/3267851.3267874}, acmid = {3267874}, publisher = {ACM}, address = {New York, NY, USA}, keywords = {Individual Differences, Social Skills Training, Virtual Audiences}, } Предполагается, что базовый алгоритм специализации должен быть хуже по скорости и, в некоторых случаях, по качеству специализации. Данная работа не затрагивает способы повышения эффективности процесса суперкомпиляции, однако демонстрирует некоторые продвижения в эту сторону из-за чисто прагматических соображений. Основное внимание уделяется подходам к суперкомпиляции и их влиянию на суперкомпиляционные эффекты. \subsubsection{Стратегии свёртки} % \textbf{Поиск узлов для переименования среди всех вычисленных поддеревьев} В базовом алгоритме суперкомпиляции поиск узлов на которые происходят переименования происходит среди родителей. Это напрямую соотносится с понятием символьных вычислений: по достижении узла, который является переименованием уже встреченного, вычисление переходит на родительский узел. Однако довольно часто встречается такая ситуация, что в разных поддеревьях дерева процессов встречаются одинаковые конфигурации, поддеревья которых оказываются полностью идентичными. В таком случае, кажется очевидной и несложной оптимизация, при которой запоминаются вычисленные поддеревья и в случае, когда встречается схожая конфигурация, поддерево не вычисляется заново, а добавляется ссылка на него. % \todo{Вопрос на засыпку меня же: может ли быть такое, что конфигурации % являются вариантами друг друга, но накопленные подстановки приведут к появлению % различных поддеревьев?}\\ %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsubsection{Стратегии развёртки} % \textbf{Стратегии развёртывания реляционных вызовов}\\ Как уже говорилось, разные стратегии развёртывания реляционных вызовов могут привести к разным эффектам специализации. К примеру, полная стратегия развёртывания, которая была принята за базовую, может приводить к дефорестации, описанной ранее. Основной недостаток базового подхода в том, что для получения всех возможных состояний он производит декартово произведение нормализованных состояний тел вызовов в конъюнкциях, что приводит к сильному разрастанию и дерева процессов и, как следствие, сильно требователен к вычислительным ресурсам и приводит к большой ветвистости программ. Последнее оказывает негативный эффект на процесс вычисления и может ухудшить производительность. Вследствие чего реализация новых стратегий развёртывания производится в исследовательских и прикладных целях. Для лёгкой подмены стратегий суперкомпиляции был разработан специальный интерфейс \lstinline{Unfoldable} (рисунок~\ref{fig:unfoldable}). \begin{figure}[h!] \begin{lstlisting} class Unfoldable a where initialize :: Conf $\rarrow$ a get :: a $\rarrow$ Conf unfold :: a $\rarrow$ Env $\rarrow$ List (Env, a) \end{lstlisting} \caption{Интерфейс для различных стратегий развёртывания.} \label{fig:unfoldable} \end{figure} Предоставляемые интерфейсом функции используются в алгоритме суперкомпиляции следующим образом: \begin{itemize} \item \lstinline{initialize} оборачивает конфигурацию в структуру, в которой может содержаться вспомогательная информация для процесса развёртывания; \item \lstinline{get} позволяет получить конфигурацию для её применения к операциям, не зависящим от стратегий; \item \lstinline{unfold} непосредственно проводит шаг вычисления на основе текущей конфигурации и её окружения, порождая новые конфигурации с соответствующими им состояниями. \end{itemize} В работе рассмотрен и реализован ряд стратегий, описанных ниже. \begin{itemize} \item {\bf Модифицированная полная стратегия развёртки}, при которой сначала из цели раскрываются все нерекурсивные вызовы. Нерекурсивность определяется лишь тем, содержит ли определение реляционный вызов самого себя. Более сложный анализ структуры функций не мог бы быть использован в силу того, что тогда было бы необходимо реализовать класс алгоритмов анализа, что совершенно отдельная задача. \item {\bf Стратегия развёртывания самого первого элемента}, при которой ожидается, что конфигурации не будут часто разбиваться на подконъюнкции, что может привести к оптимизационным эффектам, схожим с полным развёртыванием. \item {\bf Последовательная стратегия развёртывания}, при которой отслеживается, какой вызов был раскрыт на предыдущем шаге, чтобы на текущем раскрыть следующий за ним. Ожидается, что в этой стратегии при суперкомпиляции будут учтены, так или иначе, все или большинство конъюнтов конфигурации. \item {\bf Нерекурсивная стратегия развёртывания}, являющаяся модификацией последовательной стратегии, при которой в первую очередь раскрывается нерекурсивный вызов в конфигурации. Ожидается, что при нерекурсивной стратегии развёртывания из конфигураций будут как можно быстрее появляться выражения, которые могут быть сокращены или вовсе удалены из-за унификации (к примеру, отношения, кодирующие таблицы истинности, такие как \rel{and}) или привести к скорой свёрткке. \item {\bf Рекурсивная стратегия развёртывания}, являющаяся модификацией последовательной стратегии, при которой в первую очередь раскрывается рекурсивный вызов в конфигурации. Ожидается, что это может привести к скорому обобщению. \item {\bf Стратегия развёртывания вызовов с минимальным количеством ветвлений}, при которой на каждом шаге вычисления будет появляться минимально возможное количество конфигураций, что приведёт к минимальной ветвистости дерева. \item {\bf Стратегия развёртывания вызовов с максимальным количеством ветвлений}, при которой на каждом шаге вычисления будет появляться максимально возможное количество конфигураций, что, с одной стороны, увеличит количество возможных состояний, но потенциально может привести к скорому сворачиванию или обобщению. \end{itemize} % \subparagraph{Смешанная стратегия развёртывания} % \todo{Ещё нужно бы проработать} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsubsection{Стратегии обобщения} Для модификации стратегии обобщения были рассмотрены два подхода. %\textbf{Обобщение на все вычисленные узлы, не только на родительские} % \todo{Понять, почему это не противоречит методам суперкомпиляции} Во-первых, увеличение множества рассматриваемых конфигураций, на которые производится обобщение, до всех уже рассмотренных конфигураций. Покажем, что допустимо использовать обобщение на все вершины. Для этого рассмотрим конфигурацию $C_w$ и некоторую конфигурацию $C_p$ из множества конфигураций для обобщения, расширенное всеми уже обработанными конфигурациями. При обобщении $C_p$ и $C_w$ породится множество дочерних конфигураций, % $\{ C_w^i \}$, каждая из которых будет более общим подконъюнтом $C_w$ и, в силу своей обобщённости, будет содержать меньше информации, чем исходная конфигурация, однако не будет ей противоречить. В итоге, обобщение на все обработанные вершины влияет только на то, каким образом разбивается рассматриваемая конфигурация, что может привести к скорой свёртке, и как много информации о переменных теряется. %Если $C_p$ является предком $C_w$, тогда ничего не нарушается. %Иначе, % В ином случае, рассмотрим более внимательно эти конфигурации. % По определению, $C_p \embed^+ C_w$ и Обобщение на вычисленные узлы приводит к тому, что деревья конфигураций быстрее сходятся, однако остаётся под вопросом, ухудшает ли этот подход качество специализации. Во-вторых, допустимо ввести модификацию, которая запрещает прозводить операцию обобщения, если родительская конфигурация была получена с помощью обобщения. Такая эвристика может привести к том, что программы будут чаще развёртываться, однако в купе с другими модификациями может привести к интересным результатам. % \textbf{Реализация обобщения вверх} В-третьих, в суперкомпиляции, в отличие от методов частичной дедукции, для обобщения дополнительно может использоваться техника обобщения вверх, при которой происходит не подвешивание обобщённой конфигурации в качестве потомка конфигурации, которая обобщалась, но замена самого родителя на новую конфигурацию~\cite{scPos}. Старое же поддерево родителя уничтожается. Для определения необходимости обобщать вверх введём предикат $e_1 \genup e_2$, который определяет, что $e_1 \strictinst e_2$ и $e_2 \not\strictinst e_1$. Такое ограничение необходимо из-за того, что суперкомпилятор оперирует конъюнкциями выражений и делает операции разделения и обобщения вниз за один шаг с конъюнкциями возможно разной длины, однако для обобщения вверх необходимо удостовериться, что замена родительского дерева, во-первых, не добавит конфигураций, которых там не может быть, во-вторых, предоставит только более общую информацию. \begin{figure}[h!] \begin{lstlisting} else if $\exists$ parent: parent $\genup$ configuration then node $\larrow$ generalize(configuration, parent) addUp(env, tree, parent, node) \end{lstlisting} \caption{Расширение алгоритма суперкомпиляции.} \label{fig:scalgogenExtended} \end{figure} Наличие операции обобщения вверх предполагает, что необходимо умение передвигаться по дереву вверх и изменять его. Реализация в Haskell этой идеи --- задача крайне нетривиальная. Возможно представлять деревья в мутабельных массивах, однако при обобщении необходимо удалять целые поддеревья, что при таком подходе сложная операция. Классическим способом решения этой проблемы являются \emph{зипперы}~\origin{zipper}~\cite{zipper}, на основе которых, к примеру, создан система для суперкомпиляции, представленная в работе ~\cite{optimus}. Идиома зипперов предлагает рассматривать структуру данных как пару из элемента, на котором установлен фокус, и контекста, который представляется как структура данных с ``дыркой'', в котором сфокусированный элемент должен находиться. К примеру, зиппер для списка \lstinline{[1, 2, 3, 4, 5]} при фокусе на 3 представляется таким образом: \lstinline{(3, ([2, 1], [4, 5]))}. Тогда перефокусировка вправо или влево на один элемент происходит за константное время, как и замена элемента, для которой достаточно заменить первую компоненту пары. В то время как, в силу того, что операция взятия элемента в связном списке по индексу происходит за линейное время от длины списка, взятие элемента слева от 3 также будет происходить за линейное время, как и, соответственно, модификация списка. Для деревьев с произвольным количеством детей зиппер может выглядеть как пара из текущего узла и списка родителей, отсортированного в порядке близости к узлу (рисунок~\ref{fig:zipper}). \begin{figure}[h!] \begin{lstlisting}[mathescape,language=Haskell,extendedchars=\true,frame=single,basicstyle=\ttfamily] data Parent = Parent { children :: ListZipper Node } type TreeZipper = (Node, List Parent) \end{lstlisting} \caption{Пример структуры зиппера для деревьев} \label{fig:zipper} \end{figure} Родительский (структура \lstinline{Parent}) cписок детей представлен в виде зиппера (поле \lstinline{children}) для списка, в котором происходит фокус: у непосредственного родителя --- на элемент в фокусе, а у остальных родителей --- на предыдущего в порядке сортировки. Тогда передвижение вверх до первого подходящего узла происходит путём заполнения дыры при переходе на родителя, а передвижение в левого брата --- простой сдвиг в зиппере списка \lstinline{children}. При представлении дерева процессов в идиоме зипперов основа алгоритма суперкомпиляции принимает форму описания действий при смене состояния зиппера. Дерево всё ещё строится в глубину и происходит это следующим образом: если мы в узле, порождающем другие узлы (то есть \lstinline{Unfolding} или \lstinline{Abstraction}), то порождённые узлы добавляются в дерево с пометкой о том, что они не достроены, а алгоритм спускается в первого ребёнка. Когда же алгоритм приходит в листовой узел, то ему нужно подняться до родителя, у которого существует помеченный потомок, и спуститься в этого потомка. Алгоритм завершается, когда не осталось помеченных детей. Обобщение вверх приводит к тому, что происходит замена целого поддерева процессов предка, на которого обобщается конфигурация. Иногда это может приводить к потере связи между аргументами, из-за чего исчезает потенциал для возможных положительных эффектов, к примеру, протягивания констант. К примеру, на рисунке~\ref{fig:genup} представлено дерево процессов, при котором происходит обобщение вверх. \begin{figure}[h!] \center \begin{tikzpicture}[->,node distance=2cm, sibling distance=5cm] \tikzstyle{conf}=[rectangle,draw, rounded corners=.8ex] \node[conf] (root) {\rel{reverse}($a$, $a$)} ; \node[conf] (gen) [below of = root] {Generalizer: $\{ v_1 \mapsto a, v_2 \mapsto a \}$}; \node[conf] (node) [below of = gen] {\rel{reverse}($v_1$, $v_2$)}; \node (rest)[below of = node] {$\cdots$}; \path (root) edge (gen) (gen) edge (node) (node) edge (rest); \end{tikzpicture} \caption{Демонстрация потери информации при обобщении вверх.} \label{fig:genup} \end{figure} C одной стороны, теряется потенциал для генерации более оптимальной для цели \rel{reverse}($a$, $a$) программы, но c другой стороны рассмотрим следующие соображения. В данном примере процесс прогонки происходил следующим образом: некоторое время строился граф для конфигурации \rel{reverse}($a$, $a$), затем вывелась конфигурация \rel{reverse}($a$, $b$). Если бы обобщения вверх не происходило бы, то поиск ответов в результирующей программе малое время провёл бы в поддереве, оптимизированном под \rel{reverse}($a$, $a$), а остальное --- в обобщённом \rel{reverse}($a$, $b$). В общем случае это может происходить не только с корнем дерева, но и в каких-то его поддеревьях. Однако запрет на обобщение вверх в поддеревьях может сгенерировать слишком много частных случаев и привести к более неэффективным программам. Из того, что, во-первых, есть потенциал оптимизации при сохранении информации в корне дерева и, во-вторых, необходимо сдерживать разрастание дерева конфигураций, допускается рассмотрение алгоритма с обобщением вверх с запретом на обобщение к корню дерева. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \subsubsection{Расширение \ukanren оператором неэквивалентности} Множество операций в оригинальном \ukanren покрывает все нужды реляционного программирования, однако на ряде программ оно вычислительно допускает пути исполнения, которые не приводят к успеху, однако сообщить об этом не представляется возможным. К примеру, на рисунке~\ref{fig:lookup} изображена операция поиска значения по ключу в списке пар ключ-значения \rel{lookup}. \begin{figure}[h!] \begin{lstlisting} $\text{lookup}^o$ K L R = (K', V) :: L' $\equiv$ L $\land$ (K' $\equiv$ K $\land$ V $\equiv$ R $\lor$ $\text{lookup}^o$ K L' R) \end{lstlisting} \caption{Отношения поиска значения по ключу.} \label{fig:lookup} \end{figure} В соответствии с программой список \lstinline{L} должен иметь в голове пару из ключа и значения \lstinline{(K', V)} и либо этот ключ \lstinline{K'} унифицируется с искомым ключом \lstinline{K} и значение \lstinline{V} --- с результатом \lstinline{R}, либо поиск происходит в хвосте списка \lstinline{L'}. Проблема этой программы в том, что если унификация \lstinline{(K',V)::L' $\equiv$ L} прошла успешно и был найден результат, то поиск всё равно продёт во вторую ветку с рекурсивным вызовом и будет искать значение дальше, хотя по прагматике поиска ключа в списке должен вернуться лишь одно значение. Более того, суперкомпилятору тоже придётся учитывать и, возможно, проводить вычисления, которые не принесут никакой пользы. В miniKanren существует операция неэквивалентности $t_1 \not\equiv t_1$, вводящая ограничение неэквивалентности \origin{disequality contraints}\cite{mkConstr}. Операция неэквивалентности определяет, что два терма $t_1$ и $t_2$ никогда не должны быть равны, накладывая ограничения на возможные значения свободных переменных терма. Расширение синтаксиса \ukanren представлено на рисунке~\ref{fig:syntaxExt}. \begin{figure}[h!] \centering \[\begin{array}{ccll} \mathcal{G} & = & \hspace{1cm} \dots & \\ & & \hspace{1cm} \mathcal{T_X}\not\equiv\mathcal{T_X} \hspace{2cm} &\mbox{неэквивалентность} \\ \end{array}\] \caption{Расширение синтаксиса \ukanren относительно указанного на рисунке~\ref{fig:syntax}.} \label{fig:syntaxExt} \end{figure} Исправленная версия отношения \rel{lookup} представлена на рисунке~\ref{fig:lookupExt}. \begin{figure}[h!] \begin{lstlisting} $\text{lookup}^o$ K L R = (K', V) :: L' $\equiv$ L $\land$ (K' $\equiv$ K $\land$ V $\equiv$ R $\lor$ K' $\not\equiv$ K $\land$ $\text{lookup}^o$ K L' R) \end{lstlisting} \caption{Исправленное отношение поиска значения по ключу.} \label{fig:lookupExt} \end{figure} В такой реализации две по сути исключающие друг друга ветви исполнения будут исключать друг друга и при вычислении запросов, и при суперкомпиляции. Для реализации ограничения неэквивалентности вводится новая сущность под названием ``хранилище ограничений'' $\Omega$ \origin{constraints store}, которое используется для проверки нарушений неэквивалентности. Окружение расширяется хранилищем ограничений, которое затем используется при унификации и при добавлении новых ограничений. Тогда нужно ввести следующие модификации в алгоритм унификации конфигурации, который собирает все операции унификации в конъюнкции перед тем, как добавить её в множество допустимых конфигураций. \begin{itemize} \item При встрече операции неэквивалентности $t_1 \not\equiv t_2$ необходимо произвести следующие действия. Применить накопленную подстановку к термам $t_1 \theta = t_1'$ и $t_2 \theta = t_2'$ и унифицировать термы $t_1'$ и $t_2'$. Если получился пустой унификатор, значит, эти термы равны и ограничение нарушено. В таком случае суперкомпилятор покинет эту ветвь вычислений. Если же термы не унифицируются, значит, никакая подстановка в дальнейшем не нарушит ограничение. Иначе необходимо запомнить унификатор в хранилище. \item При встрече операции унификации $t_1 \equiv t_2$ необходимо получить их унификатор. Если его не существует или он пуст, то дополнительных действий производить не нужно. Иначе нужно проверить, не нарушает ли унификатор ограничения неэквивалентности. \end{itemize} Указанное расширение было добавлено в библиотеку с реализацией сопуствующих алгоритмов. % Выявление остаточной программы по дереву процессов --- \emph{резидуализация} --- % породит новые опеределения отношений. Больше одного отношения из дерева процессов может % появиться в случае, когда узлы \lstinline{Renaming} указывают на узлы, отличные от корня. % Поэтому первой фазой происходит пометка узлов, задающих таким образом отношения, % а также удаление поддеревьев, у которых все ветви вычисления пришли к неудаче. % Далее происходит обход дерева, во время которого генерируются узлы синтаксического дерева программы % в зависимости от типа текущего узла дерева процессов: % \begin{itemize} % \item \lstinline{Unfoldable} узел приводит к появлению дизъюнкций подпрограмм, которые задают дети этого узла. % Это обусловлено тем, что при прогонке в этом узле происходит ветвеление вычислений; % \item \lstinline{Abstraction} узел приводит к появлению конъюнкций подпрограмм, которые задают дети этого узла. % Это обусловлено тем, что хотя операция обобщения выявляет подконъюнкции из конфигурации и рассматривает их отдельно, % оба поддерева, задающиеся этими подконъюнкциями, должны выполнятся в одно и то же время; % \item \lstinline{Generalizer} задаёт обобщающий унификатор, который должен быть добавлен % перед своим поддеревом; % \item \lstinline{Renaming} формирует вызов реляционного отношения; % \item \lstinline{Success} представляет собой успешное вычисление, предоставляющее непротиворечивую подстановку. % \end{itemize} % latex table generated in R 3.3.3 by xtable 1.8-2 package % Wed Mar 15 22:06:10 2017 \begin{table}[H] \centering \caption[Percentage area of each LU class, NA]{Percentage area of each land use class of the study area in Kenya.} \label{table:lu_percentage_NA} \begin{tabular}{lr} \toprule Land use & Percentage (\%) \\ \midrule CCA & 1.98 \\ Government Land & 3.24 \\ Large Scale Farms & 0.94 \\ Mukogodo Group Ranches & 3.12 \\ Ranches & 59.99 \\ Rhino Sanctuary & 3.61 \\ Settlements & 19.22 \\ Swamp & 0.30 \\ Trust Land & 7.28 \\ Urban Settlements & 0.31 \\ \bottomrule \end{tabular} \end{table} \section{Additional Plots} tig3r66/math-plots % !TEX program = xelatex \documentclass{standalone} \usepackage{pgfplots} \pgfplotsset{compat=newest} \usepackage{tikz} \begin{document} % hyperboloid of one sheet \begin{tikzpicture} \begin{axis}[ axis on top, axis lines=center, set layers=default, xlabel={$x$}, ylabel={$y$}, zlabel={$z$}, xtick=\empty, ytick=\empty, ztick=\empty, domain=0:1, y domain=0:2*pi, colormap/viridis, opacity=0.75, xmin=-2, xmax=2, ymin=-2, ymax=2, view={135}{30} ] \addplot3 [surf,domain=0:360,y domain=-1:1] ({cosh(y)*cos(x)}, {cosh(y)*sin(x)}, {sinh(y)}); \end{axis} \end{tikzpicture} \end{document} % Encoding: UTF-8 @Article{CostabelGue2014VZJ, author = { and }, title = {{N}oninvasive {E}stimation of {W}ater {R}etention {P}arameters by {O}bserving the {C}apillary {F}ringe with {M}agnetic {R}esonance {S}ounding}, journal = {Vadose Zone Journal}, year = {2014}, volume = {13}, number = {6}, pages = {14}, doi = {10.2136/vzj2013.09.0163}, timestamp = {2014.06.06}, url = {https://dl.sciencesocieties.org/publications/vzj/abstracts/0/0/}, } @InProceedings{Guenther2013NSG, Title = {{O}n {I}nversion of {F}requency {D}omain {E}lectromagnetic {D}ata in {S}alt {W}ater {P}roblems - {S}ensitivity and {R}esolution}, Author = {}, Booktitle = {Ext. Abstr., 19th European Meeting of Environmental and Engineering Geophysics, Bochum, Germany}, Year = {2013}, Doi = {10.3997/2214-4609.20131387}, Timestamp = {2013.09.17} } @Article{GuentherMue2012HESS, author = {. and .}, title = {{H}ydraulic properties at the {N}orth {S}ea island of {B}orkum derived from joint inversion of magnetic resonance and electrical resistivity soundings}, journal = {Hydrology and Earth System Sciences}, year = {2012}, volume = {16}, number = {9}, pages = {3279--3291}, doi = {10.5194/hess-16-3279-2012}, owner = {Guenther.T}, timestamp = {2016.09.14}, } @Article{LoewerIgeWag2016, author = { and and }, title = {{S}pectral {D}ecomposition of {S}oil {E}lectrical and {D}ielectric {L}osses and {P}rediction of {I}n {S}itu {GPR} {P}erformance}, journal = {{IEEE} J. Sel. Top. Appl. Earth Observations Remote Sensing}, year = {2016}, volume = {9}, number = {1}, pages = {212--230}, doi = {10.1109/jstars.2015.2424152}, owner = {TG}, publisher = {Institute of Electrical {\&} Electronics Engineers ({IEEE})}, timestamp = {2016.09.14}, url = {http://dx.doi.org/10.1109/JSTARS.2015.2424152}, } @InProceedings{UllmannWieGueSie2015NSG, author = {. and . and . and .}, title = {{H}ydrogeophysics at the {E}lbe-{W}eser estuary prior to groundwater modelling}, booktitle = {Ext. Abstr., Near Surface Geoscience 2015, Turin, Italy}, year = {2015}, owner = {TG}, timestamp = {2016.09.14}, } @InProceedings{IgelStaGue2016SAGEEP, author = {. and . and .}, title = {{H}igh-resolution investigation of the capillary transition zone and its influence on {GPR} signatures}, booktitle = {Ext. Abstr., SAGEEP, Denver, USA}, year = {2016}, owner = {TG}, timestamp = {2016.06.28}, } @Article{CosciaGreLin+2010, author = { and and and and and and }, title = {3{D} crosshole apparent resistivity static inversion and monitoring of a coupled river-aquifer system}, journal = {Geophysics}, year = {2011}, volume = {76}, number = {2}, pages = {G49-59}, owner = {Guenther.T}, doi = {10.1190/1.3553003}, timestamp = {2010.08.09}, } @PhdThesis{rueckerdiss, author = {}, title = {{A}dvanced {E}lectrical {R}esistivity {M}odelling and {I}nversion using {U}nstructured {D}iscretization}, school = {University of Leipzig}, note = {http://nbn-resolving.de/urn:nbn:de:bsz:15-qucosa-69066}, year = {2011}, owner = {Guenther.T}, timestamp = {2010.09.17}, } @Article{CostabelGueDluMue2016, author = { and and and }, title = {{T}orus-nuclear magnetic resonance: {Q}uasicontinuous airborne magnetic resonance profiling by using a helium-filled balloon}, journal = {GEOPHYSICS}, year = {2016}, volume = {81}, number = {4}, pages = {WB119--WB129}, month = {jun}, doi = {10.1190/geo2015-0467.1}, file = {costabel2016G.pdf:costabel2016G.pdf:PDF}, owner = {TG}, publisher = {Society of Exploration Geophysicists}, timestamp = {2016.09.14}, url = {http://dx.doi.org/10.1190/geo2015-0467.1}, } @Article{WagnerGueSchMau2015GEO, author = {. and . and .}, title = {{C}onstructive optimization of electrode locations for target-focused resistivity monitoring}, journal = {Geophysics}, year = {2015}, volume = {80}, number = {2}, pages = {E29--E40}, doi = {10.1190/geo2014-0214.1}, owner = {TG}, timestamp = {2015.01.14}, } @InProceedings{GuentherRuecker2006SAGEEP-Joint, author = { and }, title = {{A} new joint inversion approach applied to the combined tomography of dc resistivity and seismic refraction data.}, booktitle = {Ext. Abstract, 19. EEGS annual meeting (SAGEEP), 02.-06.04.2006; Seattle, USA.}, year = {2006}, abstract = {We present a new joint inversion approach that allows for a combined inversion of independent physical parameters by exchanging structural information. The technique is based on the ideas of robust modeling. Occurring gradients of one parameter facilitate the development of gradients in the other but does not enforce this. In the presence of boundaries that can be seen by both methods it leads to sharp contrasted models. Finally, a combined image of the subsurface is obtained by cluster analysis. The technique is applied to the inversion of dc resistivity and seismic refraction data. Two synthetic data sets show how different boundary types are resolved with and without structural coupling. It is demonstrated how the quality of the inversion results is improved by the new approach.}, doi = {10.4133/1.2923578}, file = {GuentherRuecker2006SAGEEP-Joint.pdf:GuentherRuecker2006SAGEEP-Joint.pdf:PDF}, owner = {Guenther.T}, timestamp = {2015.12.21}, } @InProceedings{GuentherDluHolYar2010SAGEEP, author = { and and and }, title = {{A}quifer characterization using coupled inversion of {MRS} \& {DC}/{IP} data on a hydrogeophysical test-site}, booktitle = {Ext. Abstract, 23. EEGS annual meeting (SAGEEP), April 11-14, 2010; Keystone, CO.}, year = {2010}, volume = {23}, pages = {302-307}, abstract = {We present a new joint inversion approach that allows for a combined inversion of independent physical parameters by exchanging structural information. The technique is based on the ideas of robust modeling. Occurring gradients of one parameter facilitate the development of gradients in the other but does not enforce this. In the presence of boundaries that can be seen by both methods it leads to sharp contrasted models. Finally, a combined image of the subsurface is obtained by cluster analysis. The technique is applied to the inversion of dc resistivity and seismic refraction data. Two synthetic data sets show how different boundary types are resolved with and without structural coupling. It is demonstrated how the quality of the inversion results is improved by the new approach.}, doi = {10.4133/1.3445447}, owner = {Guenther.T}, timestamp = {2016.09.14}, } @Article{guenther2016JoAG, author = { and }, title = {{S}pectral two-dimensional inversion of frequency-domain induced polarisation data from a mining slag heap}, journal = {Journal of Applied Geophysics}, year = {2016}, volume = {135}, pages = {436-448}, doi = {10.1016/j.jappgeo.2016.01.008}, file = {guenther2016JoAG.pdf:guenther2016JoAG.pdf:PDF}, owner = {TG}, timestamp = {2015.12.21}, } @Article{Hupfer2016, author = { and and and \"{u}nther and and and }, title = {{P}olarization effects of unconsolidated sulphide-sand-mixtures}, journal = {Journal of Applied Geophysics}, year = {2016}, volume = {135}, pages = {456-465}, month = {dec}, doi = {10.1016/j.jappgeo.2015.12.003}, owner = {Dlugosch.R}, publisher = {Elsevier {BV}}, timestamp = {2016.12.16}, } @InProceedings{GuentherMarRue2016IPWS, author = {. and . and .}, title = {{S}pectral {I}nversion of {SIP} field data using py{GIML}i/{BERT}}, booktitle = {Ext. Abstr., 4th International Workshop on Induced Polarization, Aarhus, Denmark}, year = {2016}, owner = {TG}, timestamp = {2016.06.28}, } @Article{loewer2017GJI, author = { and and and and and }, title = {Ultra-broad-band electrical spectroscopy of soils and sediments{\textemdash}a combined permittivity and conductivity model}, journal = {Geophysical Journal International}, year = {2017}, volume = {210}, number = {3}, pages = {1360--1373}, month = {jun}, doi = {10.1093/gji/ggx242}, owner = {Dlugosch.R}, publisher = {Oxford University Press ({OUP})}, timestamp = {2017.07.27}, url = {https://doi.org/10.1093/gji/ggx242}, } @Article{ronczka2017SE, author = { and and \"{u}nther and {\'{e}}n and }, title = {Electric resistivity and seismic refraction tomography: a challenging joint underwater survey at \"{A}sp\"{o} Hard Rock Laboratory}, journal = {Solid Earth}, year = {2017}, volume = {8}, number = {3}, pages = {671--682}, month = {jun}, doi = {10.5194/se-8-671-2017}, owner = {Dlugosch.R}, publisher = {Copernicus {GmbH}}, timestamp = {2017.07.27}, url = {https://doi.org/10.5194/se-8-671-2017}, } @Article{hellman2017JoAG, author = { and and \"{u}nther and and \"{u}cker and }, title = {Structurally coupled inversion of {ERT} and refraction seismic data combined with cluster-based model integration}, journal = {Journal of Applied Geophysics}, year = {2017}, volume = {143}, pages = {169--181}, month = {aug}, doi = {10.1016/j.jappgeo.2017.06.008}, owner = {Dlugosch.R}, publisher = {Elsevier {BV}}, timestamp = {2017.07.27}, url = {https://doi.org/10.1016/j.jappgeo.2017.06.008}, } @Article{RueckerGueWag2017GaG, author = { and and }, title = {{pyGIMLi}: An open-source library for modelling and inversion in geophysics}, journal = {Computers \& Geosciences}, year = {2017}, owner = {Dlugosch.R}, timestamp = {2017.07.27}, } @Comment{jabref-meta: databaseType:bibtex;} ICML14MoMCompare/spectral-learn % GNUPLOT: plain TeX with Postscript \expandafter\ifx\csname GNUPLOTpicture\endcsname\relax \begingroup % Inner version of \newdimen from btxmac.tex \edef\innernewdimen{\expandafter\noexpand\csname newdimen\endcsname}% \innernewdimen\GNUPLOTunit \endgroup \gdef\GNUPLOTpicture(#1,#2){\vbox to#2\GNUPLOTunit\bgroup \def\put(##1,##2)##3{\unskip\raise##2\GNUPLOTunit \hbox to0pt{\kern##1\GNUPLOTunit ##3\hss}\ignorespaces}% \def\ljust##1{\vbox to0pt{\vss\hbox to0pt{##1\hss}\vss}}% \def\cjust##1{\vbox to0pt{\vss\hbox to0pt{\hss ##1\hss}\vss}}% \def\rjust##1{\vbox to0pt{\vss\hbox to0pt{\hss ##1}\vss}}% \def\stack##1{\let\\=\cr\tabskip=0pt\halign{\hfil ####\hfil\cr ##1\crcr}}% \def\lstack##1{\hbox to0pt{\vbox to0pt{\vss\stack{##1}}\hss}}% \def\cstack##1{\hbox to0pt{\hss\vbox to0pt{\vss\stack{##1}}\hss}}% \def\rstack##1{\hbox to0pt{\vbox to0pt{\stack{##1}\vss}\hss}}% \vss\hbox to#1\GNUPLOTunit\bgroup\ignorespaces}% \gdef\endGNUPLOTpicture{\hss\egroup\egroup}% \fi \GNUPLOTunit=0.1bp \special{! %!PS-Adobe-2.0 EPSF-2.0 %%Title: random-walk.tex %%Creator: gnuplot 3.5 (pre 3.6) patchlevel beta 347 %%CreationDate: Wed Jul 18 12:50:00 2001 %%DocumentFonts: %%BoundingBox: 0 0 251 216 %%Orientation: Landscape %%EndComments /gnudict 120 dict def gnudict begin /Color false def /Solid false def /gnulinewidth 5.000 def /userlinewidth gnulinewidth def /vshift -33 def /dl {10 mul} def /hpt_ 31.5 def /vpt_ 31.5 def /hpt hpt_ def /vpt vpt_ def /M {moveto} bind def /L {lineto} bind def /R {rmoveto} bind def /V {rlineto} bind def /vpt2 vpt 2 mul def /hpt2 hpt 2 mul def /Lshow { currentpoint stroke M 0 vshift R show } def /Rshow { currentpoint stroke M dup stringwidth pop neg vshift R show } def /Cshow { currentpoint stroke M dup stringwidth pop -2 div vshift R show } def /UP { dup vpt_ mul /vpt exch def hpt_ mul /hpt exch def /hpt2 hpt 2 mul def /vpt2 vpt 2 mul def } def /DL { Color {setrgbcolor Solid {pop []} if 0 setdash } {pop pop pop Solid {pop []} if 0 setdash} ifelse } def /BL { stroke gnulinewidth 2 mul setlinewidth } def /AL { stroke gnulinewidth 2 div setlinewidth } def /UL { gnulinewidth mul /userlinewidth exch def } def /PL { stroke userlinewidth setlinewidth } def /LTb { BL [] 0 0 0 DL } def /LTa { AL [1 dl 2 dl] 0 setdash 0 0 0 setrgbcolor } def /LT0 { PL [] 1 0 0 DL } def /LT1 { PL [4 dl 2 dl] 0 1 0 DL } def /LT2 { PL [2 dl 3 dl] 0 0 1 DL } def /LT3 { PL [1 dl 1.5 dl] 1 0 1 DL } def /LT4 { PL [5 dl 2 dl 1 dl 2 dl] 0 1 1 DL } def /LT5 { PL [4 dl 3 dl 1 dl 3 dl] 1 1 0 DL } def /LT6 { PL [2 dl 2 dl 2 dl 4 dl] 0 0 0 DL } def /LT7 { PL [2 dl 2 dl 2 dl 2 dl 2 dl 4 dl] 1 0.3 0 DL } def /LT8 { PL [2 dl 2 dl 2 dl 2 dl 2 dl 2 dl 2 dl 4 dl] 0.5 0.5 0.5 DL } def /Pnt { stroke [] 0 setdash gsave 1 setlinecap M 0 0 V stroke grestore } def /Dia { stroke [] 0 setdash 2 copy vpt add M hpt neg vpt neg V hpt vpt neg V hpt vpt V hpt neg vpt V closepath stroke Pnt } def /Pls { stroke [] 0 setdash vpt sub M 0 vpt2 V currentpoint stroke M hpt neg vpt neg R hpt2 0 V stroke } def /Box { stroke [] 0 setdash 2 copy exch hpt sub exch vpt add M 0 vpt2 neg V hpt2 0 V 0 vpt2 V hpt2 neg 0 V closepath stroke Pnt } def /Crs { stroke [] 0 setdash exch hpt sub exch vpt add M hpt2 vpt2 neg V currentpoint stroke M hpt2 neg 0 R hpt2 vpt2 V stroke } def /TriU { stroke [] 0 setdash 2 copy vpt 1.12 mul add M hpt neg vpt -1.62 mul V hpt 2 mul 0 V hpt neg vpt 1.62 mul V closepath stroke Pnt } def /Star { 2 copy Pls Crs } def /BoxF { stroke [] 0 setdash exch hpt sub exch vpt add M 0 vpt2 neg V hpt2 0 V 0 vpt2 V hpt2 neg 0 V closepath fill } def /TriUF { stroke [] 0 setdash vpt 1.12 mul add M hpt neg vpt -1.62 mul V hpt 2 mul 0 V hpt neg vpt 1.62 mul V closepath fill } def /TriD { stroke [] 0 setdash 2 copy vpt 1.12 mul sub M hpt neg vpt 1.62 mul V hpt 2 mul 0 V hpt neg vpt -1.62 mul V closepath stroke Pnt } def /TriDF { stroke [] 0 setdash vpt 1.12 mul sub M hpt neg vpt 1.62 mul V hpt 2 mul 0 V hpt neg vpt -1.62 mul V closepath fill} def /DiaF { stroke [] 0 setdash vpt add M hpt neg vpt neg V hpt vpt neg V hpt vpt V hpt neg vpt V closepath fill } def /Pent { stroke [] 0 setdash 2 copy gsave translate 0 hpt M 4 {72 rotate 0 hpt L} repeat closepath stroke grestore Pnt } def /PentF { stroke [] 0 setdash gsave translate 0 hpt M 4 {72 rotate 0 hpt L} repeat closepath fill grestore } def /Circle { stroke [] 0 setdash 2 copy hpt 0 360 arc stroke Pnt } def /CircleF { stroke [] 0 setdash hpt 0 360 arc fill } def /C0 { BL [] 0 setdash 2 copy moveto vpt 90 450 arc } bind def /C1 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 90 arc closepath fill vpt 0 360 arc closepath } bind def /C2 { BL [] 0 setdash 2 copy moveto 2 copy vpt 90 180 arc closepath fill vpt 0 360 arc closepath } bind def /C3 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 180 arc closepath fill vpt 0 360 arc closepath } bind def /C4 { BL [] 0 setdash 2 copy moveto 2 copy vpt 180 270 arc closepath fill vpt 0 360 arc closepath } bind def /C5 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 90 arc 2 copy moveto 2 copy vpt 180 270 arc closepath fill vpt 0 360 arc } bind def /C6 { BL [] 0 setdash 2 copy moveto 2 copy vpt 90 270 arc closepath fill vpt 0 360 arc closepath } bind def /C7 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 270 arc closepath fill vpt 0 360 arc closepath } bind def /C8 { BL [] 0 setdash 2 copy moveto 2 copy vpt 270 360 arc closepath fill vpt 0 360 arc closepath } bind def /C9 { BL [] 0 setdash 2 copy moveto 2 copy vpt 270 450 arc closepath fill vpt 0 360 arc closepath } bind def /C10 { BL [] 0 setdash 2 copy 2 copy moveto vpt 270 360 arc closepath fill 2 copy moveto 2 copy vpt 90 180 arc closepath fill vpt 0 360 arc closepath } bind def /C11 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 180 arc closepath fill 2 copy moveto 2 copy vpt 270 360 arc closepath fill vpt 0 360 arc closepath } bind def /C12 { BL [] 0 setdash 2 copy moveto 2 copy vpt 180 360 arc closepath fill vpt 0 360 arc closepath } bind def /C13 { BL [] 0 setdash 2 copy moveto 2 copy vpt 0 90 arc closepath fill 2 copy moveto 2 copy vpt 180 360 arc closepath fill vpt 0 360 arc closepath } bind def /C14 { BL [] 0 setdash 2 copy moveto 2 copy vpt 90 360 arc closepath fill vpt 0 360 arc } bind def /C15 { BL [] 0 setdash 2 copy vpt 0 360 arc closepath fill vpt 0 360 arc closepath } bind def /Rec { newpath 4 2 roll moveto 1 index 0 rlineto 0 exch rlineto neg 0 rlineto closepath } bind def /Square { dup Rec } bind def /Bsquare { vpt sub exch vpt sub exch vpt2 Square } bind def /S0 { BL [] 0 setdash 2 copy moveto 0 vpt rlineto BL Bsquare } bind def /S1 { BL [] 0 setdash 2 copy vpt Square fill Bsquare } bind def /S2 { BL [] 0 setdash 2 copy exch vpt sub exch vpt Square fill Bsquare } bind def /S3 { BL [] 0 setdash 2 copy exch vpt sub exch vpt2 vpt Rec fill Bsquare } bind def /S4 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt Square fill Bsquare } bind def /S5 { BL [] 0 setdash 2 copy 2 copy vpt Square fill exch vpt sub exch vpt sub vpt Square fill Bsquare } bind def /S6 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt vpt2 Rec fill Bsquare } bind def /S7 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt vpt2 Rec fill 2 copy vpt Square fill Bsquare } bind def /S8 { BL [] 0 setdash 2 copy vpt sub vpt Square fill Bsquare } bind def /S9 { BL [] 0 setdash 2 copy vpt sub vpt vpt2 Rec fill Bsquare } bind def /S10 { BL [] 0 setdash 2 copy vpt sub vpt Square fill 2 copy exch vpt sub exch vpt Square fill Bsquare } bind def /S11 { BL [] 0 setdash 2 copy vpt sub vpt Square fill 2 copy exch vpt sub exch vpt2 vpt Rec fill Bsquare } bind def /S12 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt2 vpt Rec fill Bsquare } bind def /S13 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt2 vpt Rec fill 2 copy vpt Square fill Bsquare } bind def /S14 { BL [] 0 setdash 2 copy exch vpt sub exch vpt sub vpt2 vpt Rec fill 2 copy exch vpt sub exch vpt Square fill Bsquare } bind def /S15 { BL [] 0 setdash 2 copy Bsquare fill Bsquare } bind def /D0 { gsave translate 45 rotate 0 0 S0 stroke grestore } bind def /D1 { gsave translate 45 rotate 0 0 S1 stroke grestore } bind def /D2 { gsave translate 45 rotate 0 0 S2 stroke grestore } bind def /D3 { gsave translate 45 rotate 0 0 S3 stroke grestore } bind def /D4 { gsave translate 45 rotate 0 0 S4 stroke grestore } bind def /D5 { gsave translate 45 rotate 0 0 S5 stroke grestore } bind def /D6 { gsave translate 45 rotate 0 0 S6 stroke grestore } bind def /D7 { gsave translate 45 rotate 0 0 S7 stroke grestore } bind def /D8 { gsave translate 45 rotate 0 0 S8 stroke grestore } bind def /D9 { gsave translate 45 rotate 0 0 S9 stroke grestore } bind def /D10 { gsave translate 45 rotate 0 0 S10 stroke grestore } bind def /D11 { gsave translate 45 rotate 0 0 S11 stroke grestore } bind def /D12 { gsave translate 45 rotate 0 0 S12 stroke grestore } bind def /D13 { gsave translate 45 rotate 0 0 S13 stroke grestore } bind def /D14 { gsave translate 45 rotate 0 0 S14 stroke grestore } bind def /D15 { gsave translate 45 rotate 0 0 S15 stroke grestore } bind def /DiaE { stroke [] 0 setdash vpt add M hpt neg vpt neg V hpt vpt neg V hpt vpt V hpt neg vpt V closepath stroke } def /BoxE { stroke [] 0 setdash exch hpt sub exch vpt add M 0 vpt2 neg V hpt2 0 V 0 vpt2 V hpt2 neg 0 V closepath stroke } def /TriUE { stroke [] 0 setdash vpt 1.12 mul add M hpt neg vpt -1.62 mul V hpt 2 mul 0 V hpt neg vpt 1.62 mul V closepath stroke } def /TriDE { stroke [] 0 setdash vpt 1.12 mul sub M hpt neg vpt 1.62 mul V hpt 2 mul 0 V hpt neg vpt -1.62 mul V closepath stroke } def /PentE { stroke [] 0 setdash gsave translate 0 hpt M 4 {72 rotate 0 hpt L} repeat closepath stroke grestore } def /CircE { stroke [] 0 setdash hpt 0 360 arc stroke } def /Opaque { gsave closepath 1 setgray fill grestore 0 setgray closepath } def /DiaW { stroke [] 0 setdash vpt add M hpt neg vpt neg V hpt vpt neg V hpt vpt V hpt neg vpt V Opaque stroke } def /BoxW { stroke [] 0 setdash exch hpt sub exch vpt add M 0 vpt2 neg V hpt2 0 V 0 vpt2 V hpt2 neg 0 V Opaque stroke } def /TriUW { stroke [] 0 setdash vpt 1.12 mul add M hpt neg vpt -1.62 mul V hpt 2 mul 0 V hpt neg vpt 1.62 mul V Opaque stroke } def /TriDW { stroke [] 0 setdash vpt 1.12 mul sub M hpt neg vpt 1.62 mul V hpt 2 mul 0 V hpt neg vpt -1.62 mul V Opaque stroke } def /PentW { stroke [] 0 setdash gsave translate 0 hpt M 4 {72 rotate 0 hpt L} repeat Opaque stroke grestore } def /CircW { stroke [] 0 setdash hpt 0 360 arc Opaque stroke } def /BoxFill { gsave Rec 1 setgray fill grestore } def end %%EndProlog } \GNUPLOTpicture(2519,2160) \special{" gnudict begin gsave 0 0 translate 0.100 0.100 scale 0 setgray newpath 1.000 UL LTb 300 300 M 1660 0 R 300 419 M 1660 0 R 300 537 M 1660 0 R 300 656 M 1660 0 R 300 774 M 1660 0 R 300 893 M 1660 0 R 300 1011 M 1660 0 R 300 1130 M 1660 0 R 300 1249 M 1660 0 R 300 1367 M 1660 0 R 300 1486 M 1660 0 R 300 1604 M 1660 0 R 300 1723 M 1660 0 R 300 1841 M 1660 0 R 300 1960 M 1660 0 R 300 300 M 0 1660 R 359 300 M 0 63 V 0 1597 R 0 -63 V 419 300 M 0 1660 R 478 300 M 0 63 V 0 1597 R 0 -63 V 537 300 M 0 1660 R 596 300 M 0 63 V 0 1597 R 0 -63 V 656 300 M 0 1660 R 715 300 M 0 63 V 0 1597 R 0 -63 V 774 300 M 0 1660 R 834 300 M 0 63 V 0 1597 R 0 -63 V 893 300 M 0 1660 R 952 300 M 0 63 V 0 1597 R 0 -63 V 1011 300 M 0 1660 R 1071 300 M 0 63 V 0 1597 R 0 -63 V 1130 300 M 0 1660 R 1189 300 M 0 63 V 0 1597 R 0 -63 V 1249 300 M 0 1660 R 1308 300 M 0 63 V 0 1597 R 0 -63 V 1367 300 M 0 1660 R 1426 300 M 0 63 V 0 1597 R 0 -63 V 1486 300 M 0 1660 R 1545 300 M 0 63 V 0 1597 R 0 -63 V 1604 300 M 0 1660 R 1664 300 M 0 63 V 0 1597 R 0 -63 V 1723 300 M 0 1660 R 1782 300 M 0 63 V 0 1597 R 0 -63 V 1841 300 M 0 1660 R 1901 300 M 0 63 V 0 1597 R 0 -63 V 1960 300 M 0 1660 R 1.000 UL LTa 300 1130 M 1660 0 V 1.000 UL LTa 1130 300 M 0 1660 V 1.000 UL LTb 300 300 M 1660 0 V 0 1660 V -1660 0 V 300 300 L 1.000 UP 1.000 UL LT0 1130 1130 M -118 -7 V 911 1060 L -105 55 V -94 73 V 58 103 V -75 -91 V 68 97 V 6 -119 V 1 -118 V 653 1044 L 477 86 R 109 47 V 41 -111 V 91 77 V -105 -56 V 0 119 V 19 117 V 44 110 V -115 32 V -113 -35 V -31 -114 V 60 -186 R 24 116 V 115 26 V 66 -99 V 119 5 V 104 -56 V 115 -30 V 118 8 V -73 -93 V 1686 893 L 97 68 V 1130 1130 Pls 1012 1123 Pls 911 1060 Pls 806 1115 Pls 712 1188 Pls 770 1291 Pls 695 1200 Pls 763 1297 Pls 769 1178 Pls 770 1060 Pls 653 1044 Pls 1130 1130 Pls 1239 1177 Pls 1280 1066 Pls 1371 1143 Pls 1266 1087 Pls 1266 1206 Pls 1285 1323 Pls 1329 1433 Pls 1214 1465 Pls 1101 1430 Pls 1070 1316 Pls 1130 1130 Pls 1154 1246 Pls 1269 1272 Pls 1335 1173 Pls 1454 1178 Pls 1558 1122 Pls 1673 1092 Pls 1791 1100 Pls 1718 1007 Pls 1686 893 Pls 1783 961 Pls stroke grestore end showpage } \put(1130,2060){\cjust{Random walk}} \put(1130,100){\cjust{$k$}} \put(0,1130){% \special{ps: gsave currentpoint currentpoint translate 270 rotate neg exch neg exch translate}% \cstack{$p(k)$}% \special{ps: currentpoint grestore moveto}% } \put(1960,200){\cjust{7}} \put(1841,200){\cjust{6}} \put(1723,200){\cjust{5}} \put(1604,200){\cjust{4}} \put(1486,200){\cjust{3}} \put(1367,200){\cjust{2}} \put(1249,200){\cjust{1}} \put(1130,200){\cjust{0}} \put(1011,200){\cjust{-1}} \put(893,200){\cjust{-2}} \put(774,200){\cjust{-3}} \put(656,200){\cjust{-4}} \put(537,200){\cjust{-5}} \put(419,200){\cjust{-6}} \put(300,200){\cjust{-7}} \put(250,1960){\rjust{7}} \put(250,1841){\rjust{6}} \put(250,1723){\rjust{5}} \put(250,1604){\rjust{4}} \put(250,1486){\rjust{3}} \put(250,1367){\rjust{2}} \put(250,1249){\rjust{1}} \put(250,1130){\rjust{0}} \put(250,1011){\rjust{-1}} \put(250,893){\rjust{-2}} \put(250,774){\rjust{-3}} \put(250,656){\rjust{-4}} \put(250,537){\rjust{-5}} \put(250,419){\rjust{-6}} \put(250,300){\rjust{-7}} \endGNUPLOTpicture \documentclass{standalone} % Packages \usepackage{tikz} \usetikzlibrary{calc} \begin{document} \begin{tikzpicture} \newcommand\radius{2} \draw[line width=1pt] (0,0) circle (\radius); \foreach \x in {15, 45, 200, 265}{% \draw[fill] ({\radius*cos(\x)}, {\radius*sin(\x)}) circle (2pt); } \end{tikzpicture} \end{document}\hypertarget{_touch_control_editor_8cs}{}\doxysection{C\+:/\+Users/3032f/\+Francois\+Sauce/\+Francois\+Sauce/\+Assets/\+\_\+\+\_\+\+\_\+\+\_\+\+Francois\+Sauce\+External\+Assets/\+In\+Control/\+Editor/\+Touch/\+Touch\+Control\+Editor.cs File Reference} \label{_touch_control_editor_8cs}\index{C:/Users/3032f/FrancoisSauce/FrancoisSauce/Assets/\_\_\_\_FrancoisSauceExternalAssets/InControl/Editor/Touch/TouchControlEditor.cs@{C:/Users/3032f/FrancoisSauce/FrancoisSauce/Assets/\_\_\_\_FrancoisSauceExternalAssets/InControl/Editor/Touch/TouchControlEditor.cs}} ale64bit/logic \outline{1} {Chapter 2 First-Order Theories} \centerline{\xmplbxi CHAPTER 2} \medskip \centerline{\xmplbx FIRST-ORDER THEORIES} \bigskip % ============================================================================== \outline{2} {Notation} \beginsection NOTATION {\leftskip 1in \item{\a, \b, \c, \d:} syntactical variables over terms. \item{\A, \B, \C, \D:} syntactical variables over formulas. \item{\e:} syntactical variables over constant symbols. \item{\f, \g:} syntactical variables over function symbols. \item{\i, \j:} syntactical variables over names. \item{\p, \q:} syntactical variables over predicate symbols. \item{\r, \s, \t:} syntactical variables over special constants. \item{\u, \v:} syntactical variables over expressions. \item{\x, \y, \z, \w:} syntactical variables over (individual) variables. \par} % ============================================================================== \outline{2} {Definitions} \beginsection DEFINITIONS \item{$\bullet$} A {\it first-order language} has as symbols: \itemitem{a)} the {\it variables}: $x$, $y$, $z$, $w$, $x'$, $y'$, $z'$, $w'$, $x''$, $y''$, $z''$, $w''$, \dots \itemitem{b)} for each $n$, the $n${\it-ary function symbols} and the $n${\it-ary predicate symbols}. \itemitem{c)} the symbols $\neg$, $\lor$ and $\exists$. \smallskip \item{$\bullet$} A {\it term} is defined inductively as: \itemitem{i)} $\x$ is a term; \itemitem{ii)} if $\f$ is $n$-ary, then $\f\a_1 \dots \a_n$ is a term. \smallskip \item{$\bullet$} A {\it formula} is defined inductively as: \itemitem{i)} if $\p$ is $n$-ary, then an atomic formula $\p\a_1 \dots \a_n$ is a formula; \itemitem{ii)} $\neg \A$ is a formula; \itemitem{iii)} $\lor\A\B$ is a formula; \itemitem{iv)} $\exists \x\A$ is a formula. \smallskip \item{$\bullet$} A {\it designator} is an expression which is either a term or a formula. \smallskip \item{$\bullet$} A {\it structure} $\cal A$ for a first-order language $L$ consist of: \itemitem{i)} A nonempty set $|{\cal A}|$, the {\it universe} and its {\it individuals}. \itemitem{ii)} For each $n$-ary function symbol $\f$ of $L$, an $n$-ary function $\f_{\cal A} : |{\cal A}|^n \to |{\cal A}|$. (In particular, for each constant $\e$ of $L$, $\e_{\cal A}$ is an individual of $\cal A$.) \itemitem{iii)} For each $n$-ary predicate symbol $\p$ of $L$ other than $=$, an $n$-ary predicate $\p_{\cal A}$ in $|{\cal A}|$. Also, ${\cal A}(\a)$ designates an individual and ${\cal A}(\A)$ designates a truth value. \smallskip \item{$\bullet$} A formula $\A$ is {\it valid} in a structure ${\cal A}$ if ${\cal A}(\A') = \top$ for every ${\cal A}$-instance $\A'$ of $\A$. In particular, a closed formula $\A$ is valid in ${\cal A}$ iff ${\cal A}(\A) = \top$. \smallskip \item{$\bullet$} A formula $\A$ is {\it logically valid} if it's valid in every structure. \smallskip \item{$\bullet$} A formula $\A$ is a {\it consequence} of a set $\Gamma$ of formulas if the validity of $\A$ follows from the validity of the formulas in $\Gamma$. \smallskip \item{$\bullet$} A formula $\A$ is a {\it logical consequence} of a set $\Gamma$ of formulas if $\A$ is valid in every structure for $L$ in which all of the formulas in $\Gamma$ are valid. \smallskip \item{$\bullet$} A {\it first-order theory} is a formal system $T$ such that \itemitem{i)} the language of $T$ is a first-order language; \itemitem{ii)} the axioms of $T$ are the logical axioms of $L(T)$ and certain further axioms, the {\it nonlogical axioms}; \itemitem{iii)} the rules of $T$ are Expansion, Contraction, Associative, Cut and $\exists$-Introduction. \smallskip \item{$\bullet$} A {\it model} of a theory $T$, is a structure for $L(T)$ in which all the nonlogical axioms of $T$ are valid. \item{$\bullet$} A formula $\A$ is {\it valid} in a theory $T$ if it is valid in every model of $T$. % ============================================================================== \outline{2} {Logical Axioms} \beginsection LOGICAL AXIOMS {\leftskip 1in \itemitem{{\bf Propositional}:} $\neg \A \lor \A$ \itemitem{{\bf Substitution}:} $\A_\x[\a] \to \exists \x\A$ \itemitem{{\bf Identity}:} $\x = \x$ \itemitem{{\bf Equality}:} $\x_1 = \y_1 \to \cdots \to \x_n = \y_n \to \f\x_1 \dots \x_n = \f\y_1 \dots \y_n$ \itemitem{} $\x_1 = \y_1 \to \cdots \to \x_n = \y_n \to \p\x_1 \dots \x_n \to \p\y_1 \dots \y_n$ \par} % ============================================================================== \outline{2} {Rules of Inference} \beginsection RULES OF INFERENCE {\leftskip 1in \itemitem{{\bf Expansion}.} Infer $\B \lor \A$ from $\A$. \itemitem{{\bf Contraction}.} Infer $\A$ from $\A \lor \A$. \itemitem{{\bf Associative}.} Infer $(\A \lor \B) \lor \C$ from $\A \lor (\B \lor \C)$. \itemitem{{\bf Cut}.} Infer $\B \lor \C$ from $\A \lor \B$ and $\neg\A \lor \C$. \itemitem{{\bf $\exists$-Introduction}.} If $\x$ is not free in $\B$, infer $\exists \x\A \to \B$ from $\A \to \B$. \par} % ============================================================================== \outline{2} {Results} \beginsection RESULTS \noindent {\bf \S2.4} \proclaim Lemma 1. If $\u_1, \dots, \u_n, \u_1', \dots, \u_n'$ are designators and $\u_1 \dots \u_n$ and $\u_1' \dots \u_n'$ are compatible, then $\u_i$ is $\u'_i$ for $i = 1, \dots , n$. \proclaim Formation Theorem. Every designator can be written in the form $\u\v_1\dots\v_n$, where $\bf u$ is a symbol of index $n$ and $\v_1, \dots, \v_n$ are designators, in one and only one way. \proclaim Lemma 2. Every occurrence of a symbol in a designator $\bf u$ begins an occurrence of a designator in $\bf u$. \proclaim Occurrence Theorem. Let $\bf u$ be a symbol of index $n$, and let $\v_1, \dots, \v_n$ be designators. Then any occurrence of a designator $\bf v$ in $\u\v_1 \dots \v_n$ is either all of $\u\v_1 \dots \v_n$ or a part of one of the $\v_i$. \noindent {\bf \S2.5} \proclaim Lemma. Let $\cal A$ be a structure for $L$; $\bf a$ a variable-free term in $L({\cal A})$; $\bf i$ the name of ${\cal A}(\a)$. If $\bf b$ is a term of $L({\cal A})$ in which no variable except $\bf x$ occurs, then ${\cal A}({\bf b_x}[\a]) = {\cal A}({\bf b_x}[{\bf i}])$. If $\A$ is a formula of $L({\cal A})$ in which no variable except $\bf x$ is free, then ${\cal A}({\bf A_x}[\a]) = {\cal A}({\bf A_x}[{\bf i}])$. \anchor{2.5.validity} \proclaim Validity Theorem. If $T$ is a theory, then every theorem of $T$ is valid in $T$. % ============================================================================== \outline{2} {Problems} \beginsection PROBLEMS % ------------------------------------------------------------------------------ \ans 1. \ansitem (a) Let $F(a_1, \dots, a_n)$ be any truth function. We can construct another function $$ F'(a_1, \dots, a_n) = H_{d,m}( H_{c,n}(a_1^1, \dots, a_n^1), \dots, H_{c,n}(a_1^m, \dots, a_n^m) ) $$ where the $a_1^i, \dots, a_n^i$ are all the tuples of truth values such that $F(a_1^i, \dots, a_n^i) = \top$. Thus, $a_j^i = a_j$ or $a_j^i = H_\neg(a_j)$, for some values of $i$ and $j$. Now, we can see that $F$ and $F'$ are the same function, since any truth assignment $a_1', \dots, a_n'$ that satisfies (falsifies) $F$, also satisfies (falsifies) $F'$, respectively. This is called {\it Disjunctive Normal Form (DNF)}. We can also construct a similar function $$ \eqalign{F''(a_1, \dots, a_n) &= H_{c,m}( H_\neg(H_{c,n}(a_1^1, \dots, a_n^1)), \dots, H_\neg(H_{c,n}(a_1^m, \dots, a_n^m)) )\cr &= H_{c,m}( H_{d,n}(H_\neg(a_1^1), \dots, H_\neg(a_n^1)), \dots, H_{d,n}(H_\neg(a_1^m), \dots, H_\neg(a_n^m)) )}$$ where the $a_1^i, \dots, a_n^i$ are all the tuples of truth values such that $F(a_1^i, \dots, a_n^i) = \bot$. It can be seen by a reasoning similar to above, that $F$ and $F''$ are the same function. This is called {\it Conjunctive Normal Form (CNF)}. \smallskip \ansitem (b) It can be seen that $$\eqalign{ H_{c,n} &= H_\land(a_1, H_\land(a_2, \dots))\cr H_{d,n} &= H_\lor(a_1, H_\lor(a_2, \dots)). }$$ This means we can define any truth function $F$ in terms of $H_\neg$, $H_\lor$ and $H_\land$, due to (a). Additionally, we can convert each instance of $H_\land(a, b)$ into $H_\neg(H_{\lor}(H_\neg(a), H_\neg(b)))$. Thus, every truth function is definable in terms of $H_\neg$ and $H_{\lor}$. \smallskip \ansitem (c) Since $H_\lor(a, b)$ can be defined as $H_\to(H_\neg(a), b)$, every truth function is definable in terms of $H_\neg$ and $H_\to$, due to (b). \smallskip \ansitem (d) Since $H_\lor(a, b)$ can be defined as $H_\neg(H_\land(H_\neg(a), H_\neg(b)))$, every truth function is definable in terms of $H_\neg$ and $H_\land$, due to (b). \smallskip \ansitem (e) Consider the following identities, which can be easily verified e.g. via their truth tables $$ H_\lor(a, a) = a, \quad H_\lor(a, \top) = \top $$ $$ H_\land(a, a) = a, \quad H_\land(a, \top) = a $$ $$ H_\to(a, a) = \top, \quad H_\to(a, \top) = \top, \quad H_\to(\top, a) = a $$ $$ H_\leftrightarrow(a, a) = \top, \quad H_\leftrightarrow(a, \top) = a, \quad H_\leftrightarrow(\top, a) = a. $$ Thus, any formula consisting of only those connectives and the free variable $a$ can be inductively reduced to either $a$ or $\top$ and can never define $H_\neg$. Those connectives can only define monotone functions while negation is not monotone. Note that allowing constants in the expression would allow to define negation as e.g. $H_\neg(a) = H_\to(a, \bot)$. \medskip % ------------------------------------------------------------------------------ \ans 2. \ansitem (a) Note that $H_d(a, b) = H_\land(H_\neg(a), H_\neg(b))$. We can then define $$\eqalign{H_\neg(a) &= H_d(a, a)\cr H_\lor(a, b) &= H_d(H_d(a, b), H_d(a, b))}$$ and thus every truth function is definable in terms of $H_d$ (using result from 1.1(b)). \smallskip \ansitem (b) Note that $H_s(a, b) = H_\neg(H_\land(a, b))$. We can then define $$\eqalign{H_\neg(a) &= H_s(a, a)\cr H_\lor(a, b) &= H_s(H_s(a, a), H_s(b, b))}$$ and thus every truth function is definable in terms of $H_s$ (using result from 1.1(b)). \smallskip \ansitem (c) Let $H$ be singulary with $H(a_1, \dots, a_n) = H'(a_i)$. The syntax of every truth function $F(a_1, \dots, a_m)$ definable in terms of $H$ can be inductively defined by $$ e ::= a_j \vert H(e_1, \dots, e_n) $$ where $1 \le j \le m$ and $e_1, \dots, e_n$ are valid expressions. We can then reduce every expression to an equivalent expression that involves a single $a_j$: as long as the expression has the form $H(e_1, \dots, e_n)$, we can replace it with $H'(e_i)$ and inductively reduce $e_i$. Thus, every truth function $F$ definable in terms of $H$ is singulary and furthermore $$ F(a_1, \dots, a_m) = {H'}^k(a_j) $$ for some integers $k \ge 0$ and $1 \le j \le m$. \smallskip \ansitem (d) Note that since any $n$-ary truth function is completely determined by its truth table, there are $2^{2^n}$ of them. So we know there are $2^{2^2}=16$ binary truth functions. Let's analyze them: \itemitem{-} Consider the four binary truth functions $H$ such that $$ H(a, a) = a. $$ It is easy to see that any function definable in terms of such $H$ can be inductively reduced to $a$, in a similar fashion as before. Thus, none of these four functions can define every truth function (e.g. negation $H_{\neg}$ cannot be defined). \itemitem{-} Consider the four binary truth functions $H$ such that $$ H(a,a) = \bot. $$ For each of these four functions, we have $$ H(a,\bot) \in \{a, \bot\}, \quad H(\bot, a) \in \{a, \bot\} $$ and thus none of these four functions can define every truth function (e.g. negation $H_{\neg}$ cannot be defined). \itemitem{-} Consider the four binary truth functions $H$ such that $$ H(a,a) = \top. $$ This case is symmetric to the previous one. For each of these four functions, we have $$ H(a,\top) \in \{a, \top\}, \quad H(\top, a) \in \{a, \top\} $$ and thus none of these four functions can define every truth function (e.g. negation $H_{\neg}$ cannot be defined). \itemitem{-} For the four remaining binary truth functions, we have $$ H(\top, \top) = \bot, \quad H(\bot, \bot) = \top. $$ Two of those functions $$\eqalign{ H_1(\top, \bot) &= \top, \quad H_1(\bot, \top) = \bot\cr H_2(\top, \bot) &= \bot, \quad H_2(\bot, \top) = \top} $$ are singulary and thus cannot define functions such as $H_{\lor}$, due to the result from 2.2(c). The two remaining functions are $H_d$ and $H_s$, presented in 2.2(a) and 2.2(b), respectively. \medskip % ------------------------------------------------------------------------------ \ans 3. If $\v$ is empty, then trivially neither $\u$ or $\v'$ are empty, and they are both designators. Let's assume that $\v$ is not empty and that the designator $\u\v$ has the form ${\bf t t}_1 \dots {\bf t}_n$. Since $\u\v$ and $\v\v'$ are designators, they both begin with a symbol: thus $\v$ also begins with a symbol, since it is a non-empty prefix of $\v\v'$. The occurrence of this symbol in $\u\v$ begins the occurrence of a designator $\u'$ in $\u\v$ (by Lemma 2), which is compatible with $\v$. Moreover, the occurrence of $\u'$ in $\u\v$ is either all of $\u\v$ or part of one of the ${\bf t}_i$ (by the Occurrence Theorem). In the former case, it means that $\v$ is a designator and $\u$ and $\v'$ are empty. On the other hand, if $\u'$ is part of one of the ${\bf t}_i$, it means that $\v\v'$ begins with $\u'$, and thus $\u'$ and $\v$ are the same (by the Formation Theorem) and $\v'$ is empty. \medskip % ------------------------------------------------------------------------------ \ans 4. If a term is: \itemitem{i)} a variable $\x'$, then the substitution result is $\x$ itself, which is also a term. \itemitem{ii)} a function application ${\bf fa}_1 \dots \a_n$, then $\a$ is one of the $\a_i$ and the substitution result is also a term, or $\a$ is substituted in one of the terms $\a_i$, and it remains a term, by the induction hypothesis. If a formula is: \itemitem{i)} an atomic formula ${\bf pa}_1 \dots \a_n$, then substituting $\a$ in any of the $\a_i$ results in a term, as previously shown. Thus it remains a formula. \itemitem{ii)} $\neg \A$, then substituting $\a$ in $\A$ remains a formula by the induction hypothesis. \itemitem{iii)} $\lor\A\B$, then substituting $\a$ in $\A$ or $\B$ remains a formula by the induction hypothesis. \itemitem{iv)} $\exists\y\A$, then substituting $\a$ in $\A$ remains a formula by the induction hypothesis. \medskip % ------------------------------------------------------------------------------ \ans 5. The sections of this problem show that each axiom and rule allows proving theorems that wouldn't be provable without it. The common strategy is finding a suitable mapping $f$ from formulas to truth values and a specific theorem $\B$, such that if a theorem $\A$ is provable without the axiom or rule in question, then $f(\A) = \top$, but $f(\B) = \bot$ and it is thus no provable without it. This means that none of the axioms or rules are redundant. \smallskip \ansitem (a) The hinted function is defined as: $$\eqalign{f(\A) &= \top, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \bot;\cr f(\A \lor \B) &= f(\B);\cr f(\exists \x\A) &= \top.\cr}$$ Let's prove that if $\A$ is provable without propositional axioms then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = f(\exists \x\A') = \top$; \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\A) = f(\A') = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\B' \lor \C') = f(\C') = f(\A) = \top$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case $f(\neg \A' \lor \C') = f(\C') = f(\A) = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case $f(\A' \to \B') = f(\neg \A' \lor \B') = f(\B') = f(\A) = \top$. Thus, if $\A$ is provable without propositional axioms, we have $f(\A) = \top$. But $f(\neg \neg (x=x) \lor \neg (x=x)) = f(\neg (x=x)) = \bot$ and so it is not provable without propositional axioms. \smallskip \ansitem (b) The hinted function is defined as: $$\eqalign{f(\A) &= \top, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \neg f(\A);\cr f(\A \lor \B) &= f(\A) \lor f(\B);\cr f(\exists \x\A) &= \bot.\cr}$$ Let's prove that if $\A$ is provable without substitution axioms then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \neg f(\A') \lor f(\A') = \top$; \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = \neg f(\x_1 = \y_1) \lor \neg f(\x_2 = \y_2) \lor \neg f(\x_1 = \x_2) \lor f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\B' \lor \A') = f(\B') \lor f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') \lor f(\A') = f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\A') \lor f(\B' \lor \C') = f(\A') \lor f(\B') \lor f(\C') = f(\A' \lor \B') \lor f(\C') = f((\A' \lor \B') \lor \C') = \top$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case we have $f(\B' \lor \C') = f(\B') \lor f(\C')$, $f(\A' \lor \B') = f(\A') \lor f(\B')$ and $f(\neg \A' \lor \C') = \neg f(\A') \lor f(\C')$. If $f(\A') = \top$, then $f(\C') = \top$. If $f(\A') = \bot$, then $f(\B') = \top$. Thus $f(\B') \lor f(\C') = f(\A) = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case $f(\A) = f(\neg \exists \x\A' \lor \B') = \neg f(\exists \x\A') \lor f(\B') = \top$. Thus, if $\A$ is provable without substitution axioms, we have $f(\A) = \top$. But $f(x=x \to \exists x(x=x)) = \neg f(x=x) \lor f(\exists x(x=x)) = \bot$ and so it is not provable without substitution axioms. \smallskip \ansitem (c) The hinted function is defined as: $$\eqalign{f(\A) &= \bot, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \neg f(\A);\cr f(\A \lor \B) &= f(\A) \lor f(\B);\cr f(\exists \x\A) &= f(\A).\cr}$$ Let's prove that if $\A$ is provable without identity axioms then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \neg f(\A') \lor f(\A') = \top$; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = \neg f(\A'_\x[\a]) \lor f(\A') = \top$ (see below for this case); \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = \neg f(\x_1 = \y_1) \lor \neg f(\x_2 = \y_2) \lor \neg f(\x_1 = \x_2) \lor f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\B' \lor \A') = f(\B') \lor f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') \lor f(\A') = f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\A') \lor f(\B' \lor \C') = f(\A') \lor f(\B') \lor f(\C') = f(\A' \lor \B') \lor f(\C') = f((\A' \lor \B') \lor \C') = \top$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case we have $f(\B' \lor \C') = f(\B') \lor f(\C')$, $f(\A' \lor \B') = f(\A') \lor f(\B')$ and $f(\neg \A' \lor \C') = \neg f(\A') \lor f(\C')$. If $f(\A') = \top$, then $f(\C') = \top$. If $f(\A') = \bot$, then $f(\B') = \top$. Thus $f(\B') \lor f(\C') = f(\A) = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case $f(\A) = f(\neg \exists \x\A' \lor \B') = \neg f(\A') \lor f(\B') = f(\neg \A' \lor \B') = f(\A' \to \B') = \top$. To treat substitution axioms, let's show that $f(\A_\x[\a]) = f(\A)$ by induction on the length of $\A$: \itemitem{$\bullet$} for $\A$ atomic with form ${\bf pb}_1 \dots \b_n$: we have $f(\A_\x[\a]) = f({{\bf pb}_1}_\x[\a] \dots {\b_n}_\x[\a]) = \bot$ and $f(\A) = f({\bf pb}_1 \dots \b_n) = \bot$. \itemitem{$\bullet$} for $\A$ with form $\neg \A'$: we have $f(\A_\x[\a]) = \neg f(\A'_\x[\a])$ and $f(\A) = \neg f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\A' \lor \B'$: we have $f(\A_\x[\a]) = f(\A'_\x[\a]) \lor f(\B'_\x[\a])$ and $f(\A) = f(\A') \lor f(\B')$ and $f(\A'_\x[\a]) = f(\A')$ and $f(\B'_\x[\a]) = f(\B')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\exists \y\A'$: we have $f(\exists \y\A'_\x[\a]) = f(\A'_\x[\a])$ and $f(\exists \y\A') = f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. Thus, if $\A$ is provable without identity axioms, we have $f(\A) = \top$. But $f(x=x) = \bot$ and so it is not provable without identity axioms. \smallskip \ansitem (d) The hinted function is defined as: $$\eqalign{f(\e_i = \e_j) &= \top \quad \hbox{iff $i \le j$;}\cr f(\neg \A) &= \neg f(\A);\cr f(\A \lor \B) &= f(\A) \lor f(\B);\cr f(\exists \x\A) &= \top \quad \hbox{iff $f(\A_\x[\e_i]) = \top$ for some $i$}.\cr}$$ Let's prove that if $\A$ is provable without equality axioms then $f(\A') = \top$ for every formula obtained from $\A$ by replacing each variable by some $\e_i$ at all its free occurrences, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \neg f(\A') \lor f(\A') = \top$ for every closed formula $\A''$ obtained from $\A'$ by replacing each variable by some $\e_i$ at all its free occurrences; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = \neg f(\A'_\x[\a]) \lor f(\exists \x\A')$. For every closed formula $\A''$ obtained from $\A'$ by replacing each variable (except $\x$) by some $\e_i$ at all its free occurrences: if $f(\A''_\x[\e_i]) = \top$ for some $i$, then $f(\exists \x\A'') = \top$ by the definition of $f$. Otherwise, $f(\A''_\x[\e_i]) = \bot$ for all $i$ and thus $\neg f(\A''_\x[\e_i]) = \top$; \itemitem{$\bullet$} an identity axiom: $f(\A) = f(\x = \x) = \top$ for any substitution of $\x$ by some $\e_i$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A') = f(\B') \lor f(\A')$ with $f(\A') = \top$ for every closed formula $\A''$ obtained from $\A'$ by replacing each variable by some $\e_i$ at all its free occurrences, by the induction hypothesis; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = f(\A') \lor f(\A') = f(\A') = \top$ for every closed formula $\A''$ obtained from $\A'$ by replacing each variable by some $\e_i$ at all its free occurrences, by the induction hypothesis; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C') = f(\A') \lor f(\B') \lor f(\C')$ with $f(\A' \lor (\B' \lor \C')) = f(\A') \lor f(\B') \lor f(\C') = \top$ for every closed formulas $\A''$, $\B''$ and $\C''$ obtained from $\A'$, $\B'$ and $\C'$, respectively, by replacing each variable by some $\e_i$ at all its free occurrences, by the induction hypothesis; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C') = f(\B') \lor f(\C')$ with $f(\A' \lor \B') = f(\A') \lor f(\B') = \top$ and $f(\neg \A' \lor \C') = \neg f(\A') \lor f(\C') = \top$ for every closed formulas $\A''$, $\B''$ and $\C''$ obtained from $\A'$, $\B'$ and $\C'$, respectively, by replacing each variable by some $\e_i$ at all its free occurrences, by the induction hypothesis. If $f(\A') = \top$, then $f(\C') = \top$. If $f(\A') = \bot$, then $f(\B') = \top$. Thus $f(\B') \lor f(\C') = f(\A) = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B') = \neg f(\exists \x\A') \lor f(\B')$ with $f(\A' \to \B') = \neg f(\A') \lor f(\B') = \top$ for every closed formula $\A''$ and $\B''$ obtained from $\A'$ and $\B'$, respectively, by replacing each variable by some $\e_i$ at all its free occurrences, by the induction hypothesis. If $f(\B') = \top$, then $f(\A) = \top$ follows trivially. Otherwise, we must have $f(\A') = \bot$ for all closed formulas $\A''$ obtained from $\A'$ as described above. This implies that $f(\exists \x\A') = \bot$ and thus $f(\A) = \top$. Thus, if $\A$ is provable without equality axioms, we have $f(\A') = \top$ for every formula $\A'$ obtained from $\A$ by replacing each variable by some $\e_i$ at all its free occurences. But $f(x=y \to x=z \to x=x \to y=z) = \neg f(x=y) \lor \neg f(x=z) \lor \neg f(x=x) \lor f(y=z) = \bot$ since it does not hold for the substitution $[\x, \y, \z] \to [\e_1, \e_3, \e_2]$ and so it is not provable without equality axioms. \smallskip \ansitem (e) The hinted function is defined as: $$\eqalign{f(\A) &= \top, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \neg f(\A);\cr f(\A \lor \B) &= f(\A) \leftrightarrow \neg f(\B);\cr f(\exists \x\A) &= f(\A).\cr}$$ Let's prove that if $\A$ is provable without the expansion rule then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \neg f(\A') \leftrightarrow \neg f(\A') = \top$; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = \neg f(\A'_\x[\a]) \leftrightarrow \neg f(\A') = \top$ (see below for this case); \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = \neg f(\x_1 = \y_1) \leftrightarrow \neg f(\x_2 = \y_2) \leftrightarrow \neg f(\x_1 = \x_2) \leftrightarrow \neg f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. However, this is a contradiction since $f(\A') \leftrightarrow \neg f(\A') = \bot$ for any $\bf A'$ so it's not possible to have a proof where the contraction rule is applied (???); \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\A') \leftrightarrow \neg f(\B' \lor \C') = f(\A') \leftrightarrow \neg (f(\B') \leftrightarrow \neg f(\C')) = f(\A') \leftrightarrow (f(\B') \leftrightarrow f(\C'))$ and $f((\A' \lor \B') \lor \C') = f(\A' \lor \B') \leftrightarrow \neg f(\C') = (f(\A') \leftrightarrow \neg f(\B')) \leftrightarrow \neg f(\C') = f(\A') \leftrightarrow f(\B') \leftrightarrow f(\C')$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case $f(\A' \lor \B') = f(\A') \leftrightarrow \neg f(\B')$ and $f(\neg \A \lor \C') = \neg f(\A') \leftrightarrow \neg f(\C') = f(\A') \leftrightarrow f(\C')$, and thus $f(\B') \leftrightarrow \neg f(\C') = f(\B' \lor \C') = f(\A) = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case $f(\neg \A' \lor \B') = f(\A') \leftrightarrow f(\B')$ and $f(\neg \exists \x\A' \lor \B') = f(\A') \leftrightarrow f(\B')$. To treat substitution axioms, let's show that $f(\A_\x[\a]) = f(\A)$ by induction on the length of $\A$: \itemitem{$\bullet$} for $\A$ atomic with form ${\bf pb}_1 \dots \b_n$: we have $f(\A_\x[\a]) = f({{\bf pb}_1}_\x[\a] \dots {\b_n}_\x[\a]) = \top$ and $f(\A) = f({\bf pb}_1 \dots \b_n) = \top$. \itemitem{$\bullet$} for $\A$ with form $\neg \A'$: we have $f(\A_\x[\a]) = \neg f(\A'_\x[\a])$ and $f(\A) = \neg f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\A' \lor \B'$: we have $f(\A_\x[\a]) = f(\A'_\x[\a]) \leftrightarrow \neg f(\B'_\x[\a])$ and $f(\A) = f(\A') \leftrightarrow \neg f(\B')$ and $f(\A'_\x[\a]) = f(\A')$ and $f(\B'_\x[\a]) = f(\B')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\exists \y\A'$: we have $f(\exists \y\A'_\x[\a]) = f(\A'_\x[\a])$ and $f(\exists \y\A') = f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. Thus, if $\A$ is provable without the expansion rule, we have $f(\A) = \top$. But $f(x=x \lor (\neg (x=x) \lor (x=x))) = f(x=x) \leftrightarrow \neg (\neg f(x=x) \leftrightarrow \neg f(x=x)) = \bot$ and so it is not provable without the expansion rule. \smallskip \ansitem (f) The hinted function is defined as: $$\eqalign{f(\A) &= \top, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \bot;\cr f(\A \lor \B) &= \top;\cr f(\exists \x\A) &= \bot.\cr}$$ Let's prove that if $\A$ is provable without the contraction rule then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \top$; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = \top$; \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\A) = f(\B' \lor \A') = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A) = f((\A' \lor \B') \lor \C') = \top$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case $f(\A) = f(\B' \lor \C') = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case $f(\A) = f(\neg \exists \x\A' \lor \B') = \top$. Thus, if $\A$ is provable without the contraction rule, we have $f(\A) = \top$. But $f(\neg \neg (x=x)) = \bot$ and so it is not provable without the contraction rule. \smallskip \ansitem (g) The hinted function is defined as: $$\eqalign{f(\A) &= 0, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= 1 - f(\A);\cr f(\A \lor \B) &= f(\A) \cdot f(\B) \cdot (1 - f(\A) - f(\B));\cr f(\exists \x\A) &= f(\A).\cr}$$ Let's prove that if $\A$ is provable without the associative rule then $f(\A) = 0$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = (1 - f(\A')) \cdot f(\A') \cdot (1 - (1 - f(\A')) - f(\A')) = (1 - f(\A')) \cdot f(\A') \cdot (f(\A') - f(\A')) = 0$; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = (1 - f(\A'_\x[\a])) \cdot f(\A') \cdot (1 - (1 - f(\A'_\x[\a])) - f(\A')) = 0$ (see below for this case); \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = 0$; \itemitem{$\bullet$} an equality axiom: we have $$\eqalign{f(\A) &= f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2)\cr &= f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2))\cr &= (1-f(\x_1=\y_1)) \cdot f(\A') \cdot (f(\x_1 = \y_1) - f(\A'))}$$ $$f(\A') = f(\x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = (1-f(\x_2 = \y_2)) \cdot f(\A'') \cdot (f(\x_2 = \y_2) - f(\A''))$$ $$f(\A'') = f(\x_1 = \x_2 \to \y_1 = \y_2) = (1-f(\x_1 = \x_2)) \cdot f(\y_1 = \y_2) \cdot (f(\x_1 = \x_2) - f(\y_1 = \y_2)) = 0;$$ \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = 0$ by the induction hypothesis. In this case $f(\A) = f(\B') \cdot f(\A') \cdot (1 - f(\B') - f(\A')) = 0$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = 0$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') \cdot f(\A') \cdot (1 - f(\A') - f(\A')) = 0$ and the only integer solution is $f(\A') = 0$ and thus $f(\A) = 0$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = 0$ and $f(\neg \A' \lor \C') = 0$ by the induction hypothesis. Consider the equations $$\eqalignno{f(\A' \lor \B') &= f(\A') \cdot f(\B') \cdot (1 - f(\A') - f(\B')) = 0&(1)\cr f(\neg \A' \lor \C') &= (1 - f(\A')) \cdot f(\C') \cdot (f(\A') - f(\C')) = 0&(2)\cr f(\A) = f(\B' \lor \C') &= f(\B') \cdot f(\C') \cdot (1 - f(\B') - f(\C')) = 0&(3) }$$ and the possible cases that satisfy equation (2). First, $(1-f(\A')) = 0$ implies that $f(\A') = 1$ and substituting in equation (1) we obtain $f(\B') \cdot (-f(\B')) = 0$ which means that $f(\B') = 0$ which satisfies equation (3). Second, $f(\C') = 0$, which trivially satifies equation (3). Third, $f(\A') - f(\C') = 0$ which implies $f(\A') = f(\C')$ and substituting in equation (1) we obtain $f(\C') \cdot f(\B') \cdot (1 - f(\C') - f(\B')) = 0$. Thus, equation (3) is satisfied in all cases; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = 0$ by the induction hypothesis. In this case $f(\A) = f(\neg \exists \x\A' \lor \B') = (1 -f(\A')) \cdot f(\B') \cdot (f(\A') - f(\B')) = f(\A' \to \B') = 0$. To treat substitution axioms, let's show that $f(\A_\x[\a]) = f(\A)$ by induction on the length of $\A$: \itemitem{$\bullet$} for $\A$ atomic with form ${\bf pb}_1 \dots \b_n$: we have $f(\A_\x[\a]) = f({{\bf pb}_1}_\x[\a] \dots {\b_n}_\x[\a]) = 0$ and $f(\A) = f({\bf pb}_1 \dots \b_n) = 0$. \itemitem{$\bullet$} for $\A$ with form $\neg \A'$: we have $f(\A_\x[\a]) = 1 - f(\A'_\x[\a])$ and $f(\A) = 1 - f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\A' \lor \B'$: we have $f(\A_\x[\a]) = f(\A'_\x[\a]) \cdot f(\B'_\x[\a]) \cdot (1 - f(\A'_\x[\a]) - f(\B'_\x[\a]))$ and $f(\A) = f(\A') \cdot f(\B') \cdot (1 - f(\A') - f(\B'))$ and $f(\A'_\x[\a]) = f(\A')$ and $f(\B'_\x[\a]) = f(\B')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\exists \y\A'$: we have $f(\exists \y\A'_\x[\a]) = f(\A'_\x[\a])$ and $f(\exists \y\A') = f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. Thus, if $\A$ is provable without the associative rule, we have $f(\A) = 0$. But $f(\neg (\neg (x=x) \lor \neg (x=x))) = 1 - f(\neg (x=x) \lor \neg (x=x)) = 1 - ((1-f(x=x))^2 \cdot (1 - 2 \cdot (1-f(x=x)))) = 1 - (1 - 2) = 2$ and so it is not provable without the associative rule. \smallskip \ansitem (h) The hinted function is defined as: $$\eqalign{f(\A) &= \top \quad \hbox{for $\A$ atomic};\cr f(\neg \A) &= \cases{\top, &if $f(\A) = \bot$ or $\A$ is atomic;\cr \bot, &otherwise.}\cr f(\A \lor \B) &= f(\A) \lor f(\B);\cr f(\exists \x\A) &= f(\A).\cr}$$ Let's prove that if $\A$ is provable without the cut rule then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = f(\neg \A') \lor f(\A') = \top$ (since if $f(\A') = \bot$, then $f(\neg \A') = \top$ from the definition of $f$); \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = f(\neg \A'_\x[\a]) \lor f(\A') = \top$ (since if $f(\A') = \bot$, then $f(\neg \A') = \top$ from the definition of $f$ and $f(\neg \A'_\x[\a]) = f(\neg \A')$. See below for this case); \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = f(\neg (\x_1 = \y_1)) \lor f(\neg (\x_2 = \y_2)) \lor f(\neg (\x_1 = \x_2)) \lor f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\A) = f(\B' \lor \A') = f(\B') \lor f(\A') = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') \lor f(\A') = \top$ and thus $f(\A) = f(\A') = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\A') \lor f(\B' \lor \C') = f(\A') \lor f(\B') \lor f(\C') = f(\A' \lor \B') \lor f(\C') = f((\A' \lor \B') \lor \C') = \top$; \itemitem{$\bullet$} the $\exists$-introduction rule: we have $f(\A) = f(\exists \x\A' \to \B')$ with $f(\A' \to \B') = \top$ by the induction hypothesis. In this case we have $f(\A) = f(\neg \exists \x\A' \lor \B') = f(\neg \exists \x\A') \lor f(\B')$ and $f(\neg \A') \lor f(\B') = \top$. So either $f(\neg \A') = \top$ or $f(\B') = \top$. In the latter case, it follows trivially that $f(\A) = \top$. In the former case, note that since $f(\exists \x\A) = f(\A)$ and $\exists \x\A$ is not atomic, then $f(\neg \exists \x\A') = f(\neg \A')$. To treat substitution axioms, let's show that $f(\A_\x[\a]) = f(\A)$ by induction on the length of $\A$: \itemitem{$\bullet$} for $\A$ atomic with form ${\bf pb}_1 \dots \b_n$: we have $f(\A_\x[\a]) = f({{\bf pb}_1}_\x[\a] \dots {\b_n}_\x[\a]) = \top$ and $f(\A) = f({\bf pb}_1 \dots \b_n) = \top$. \itemitem{$\bullet$} for $\A$ with form $\neg \A'$ with $\A'$ atomic: we have $f(\A_\x[\a]) = f(\neg \A'_\x[\a]) = \top$ and $f(\A) = f(\neg \A') = \top$. \itemitem{$\bullet$} for $\A$ with form $\neg \A'$ with $\A'$ not atomic: we have $f(\A_\x[\a]) = f(\neg \A'_\x[\a])$ and $f(\A) = f(\neg \A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\A' \lor \B'$: we have $f(\A_\x[\a]) = f(\A'_\x[\a]) \lor f(\B'_\x[\a])$ and $f(\A) = f(\A') \lor f(\B')$ and $f(\A'_\x[\a]) = f(\A')$ and $f(\B'_\x[\a]) = f(\B')$ by the induction hypothesis. \itemitem{$\bullet$} for $\A$ with form $\exists \y\A'$: we have $f(\exists \y\A'_\x[\a]) = f(\A'_\x[\a])$ and $f(\exists \y\A') = f(\A')$ and $f(\A'_\x[\a]) = f(\A')$ by the induction hypothesis. Thus, if $\A$ is provable without the cut rule, we have $f(\A) = \top$. But $f(\neg \neg (x=x)) = \bot$ since $f(\neg (x=x)) = \top$ and so it is not provable without the cut rule. \smallskip \ansitem (i) The hinted function is defined as: $$\eqalign{f(\A) &= \top, \quad \hbox{for $\A$ atomic;}\cr f(\neg \A) &= \neg f(\A);\cr f(\A \lor \B) &= f(\A) \lor f(\B);\cr f(\exists \x\A) &= \top.\cr}$$ Let's prove that if $\A$ is provable without the $\exists$-introduction rule then $f(\A) = \top$, by induction on theorems. In a proof of $\A$, if $\A$ was obtained from: \itemitem{$\bullet$} a propositional axiom: we have $f(\A) = f(\neg \A' \lor \A') = \neg f(\A') \lor f(\A') = \top$; \itemitem{$\bullet$} a substitution axiom: we have $f(\A) = f(\neg \A'_\x[\a] \lor \exists \x\A') = \neg f(\A'_\x[\a]) \lor f(\exists \x\A') = \top$; \itemitem{$\bullet$} an identity axiom: since it's an atomic formula, $f(\A) = f(\x = \x) = \top$; \itemitem{$\bullet$} an equality axiom: we have $f(\A) = f(\x_1 = \y_1 \to \x_2 = \y_2 \to \x_1 = \x_2 \to \y_1 = \y_2) = f(\neg (\x_1 = \y_1) \lor \neg (\x_2 = \y_2) \lor \neg (\x_1 = \x_2) \lor (\y_1 = \y_2)) = \neg f(\x_1 = \y_1) \lor \neg f(\x_2 = \y_2) \lor \neg f(\x_1 = \x_2) \lor f(\y_1 = \y_2) = \top$; \itemitem{$\bullet$} the expansion rule: we have $f(\A) = f(\B' \lor \A')$ with $f(\A') = \top$ by the induction hypothesis. In this case $f(\A) = f(\B' \lor \A') = f(\B') \lor f(\A') = \top$; \itemitem{$\bullet$} the contraction rule: we have $f(\A) = f(\A')$ with $f(\A' \lor \A') = \top$ by the induction hypothesis. In this case $f(\A' \lor \A') = f(\A') \lor f(\A') = f(\A') = f(\A) = \top$; \itemitem{$\bullet$} the associative rule: we have $f(\A) = f((\A' \lor \B') \lor \C')$ with $f(\A' \lor (\B' \lor \C')) = \top$ by the induction hypothesis. In this case $f(\A' \lor (\B' \lor \C')) = f(\A') \lor f(\B' \lor \C') = f(\A') \lor f(\B') \lor f(\C') = f(\A' \lor \B') \lor f(\C') = f((\A' \lor \B') \lor \C') = \top$; \itemitem{$\bullet$} the cut rule: we have $f(\A) = f(\B' \lor \C')$ with $f(\A' \lor \B') = \top$ and $f(\neg \A' \lor \C') = \top$ by the induction hypothesis. In this case we have $f(\B' \lor \C') = f(\B') \lor f(\C')$, $f(\A' \lor \B') = f(\A') \lor f(\B')$ and $f(\neg \A' \lor \C') = \neg f(\A') \lor f(\C')$. If $f(\A') = \top$, then $f(\C') = \top$. If $f(\A') = \bot$, then $f(\B') = \top$. Thus $f(\B') \lor f(\C') = f(\A) = \top$; Thus, if $\A$ is provable without the $\exists$-introduction rule, we have $f(\A) = \top$. But $f(\exists y \neg (x=x) \to \neg (x=x)) = \neg f(\exists y \neg (x=x)) \lor \neg f(x=x) = \bot$ and so it is not provable without the $\exists$-introduction rule. \smallskip \vfill \break aabor/textbooks % !TeX program = lualatex -synctex=1 -interaction=nonstopmode --shell-escape %.tex \documentclass[_fin_decisions_lectures.tex]{subfiles} \begin{document} \subsection{Риск: определение} \setbeamercovered{transparent} \begin{frame}[shrink=15]{История возникновения понятия риска} \begin{itemize}[<+->] \item \textbf{Французский \textit{\foreignlanguage{french}{risqué}}} - опастность или неудобство, как предсказуемое так и нет (1578 г., 1690 г. официальный термин). \item \textbf{Итальянский \textit{risco}} (1-ая половина XIV века) - возможность вреда, неприятных последствий и т.п. \item \textbf{Средневековая латынь \textit{resicum, risicum}} (сер. XII века) - слово использовалось в торговле в смысле ``угроза, опасность'', в морском деле - ``подводная скала, риф''. \item \textbf{Средневековый французский, испанский диалекты \textit{resicq, risicq, rezegue, risc, riesgo}} - возможность потери или повреждения товаров. \item \textbf{Арабский \textit{rizq }}- слово использовалось в разных смыслах ``надел данный Богом каждому человеку'', ``благодать, благославение (от Бога)'', ``богатство, собственность, доход'', ``доля, судьба, удача, шанс''. \end{itemize} \end{frame} \begin{frame}{Основные теории происхождения слова риск, выводы} \begin{itemize} \item В европейских языках слово ``риск'' изначально имело значение ``нечто могущее порезать'', иными словами, ``скала, риф, утес'', что имело отношение к морской торговле и в целом, к морскому делу. Как глагол использовалось в смысле ``хождение по морю в неизвестных водах или слишком близко от рифов''. \item Арабское и греческое происхождение слова ``риск'' указывает на связь с понятиями ``богатсво, доля, судьба, шанс''. \end{itemize} \end{frame} \begin{frame}[shrink=10]{Повседневное использование слова ``риск''} \begin{itemize} \item Подверженность возможности потерь, ущербу, повреждению или другим враждебным или неблагоприятным обстоятельствам; шанс или ситуация, заключающая в себе такую возможность. \item Опасное путешествие, предприятие, или образ действий, авантюра. \item Человек или вещь, способные произвести хороший или плохой результат в определенном смысле; также человек или вещь, рассматриваемые как угроза или источник ущерба. \end{itemize} Понятие ``риск'' может использоваться как в положительном, так и в отрицательном смысле, а также может быть как глаголом так и существительным. \end{frame} \begin{frame}[allowframebreaks]{Классификация определений понятия ``риск''} 1. Риск=Ожидаемая прибыль/убыток: риск это произведение вероятности убытка на размер потерь. 2. Риск=Вероятность ущерба или убытка. 3. Риск=Объективная неопределенность: риск это измеряемая неопределенность, т.е. неопределенность, для которой известно распределение исходов (статистически или на основе прошлого опыта). \pagebreak 4. Риск=Неопределенность: относительно издержек, потерь или ущерба. 5. Риск=Возможность/потенциал убытков: риск это возможность неприятных последствий от события. \pagebreak 6. Риск=Вероятность и сценарии/Последствия/тяжесть последствий: риск это комбинация угроз и вероятности их наступления, риск это единица измерения одновременно вероятности и тяжести негативных последствий. 7. Риск=Событие или последствие: риск это ситуация, когда есть угроза для имущества или жизни людей, и результат является неопределенным. \pagebreak 8. Риск=Последствия/Ущерб/и их тяжесть + неопределенность: риск это комбинация последствий от действий по отношению к каким-либо вещам, ценным для человека. Отклонения от целей, ожиданий. 9. Риск это влияние неопределенности на цели (ISO). 10. Риск это триада: события (сценарии), их вероятности, а также их последствия. \end{frame} \begin{frame}{Понятие риска} \begin{block}{Риск} \quad – это деятельность, связанная с преодолением неопределенности в ситуации неизбежного выбора, в процессе которой имеется возможность количественно и качественно оценить вероятность достижения предполагаемого результата, неудачи и отклонения от цели. \end{block} \end{frame} \begin{frame}[ allowframebreaks ]{Риск и неопределенность} \begin{itemize} \item Различают задачи принятия решений при риске и соответственно в условиях неопределенности. Если существует возможность качественно и количественно определить степень вероятности того или иного варианта, то это и будет ситуация риска. \pagebreak \item Разница между риском и неопределенностью относится к способу задания информации и определяется наличием (в случае риска) или отсутствием (при неопределенности) вероятностных характеристик неконтролируемых переменных. \end{itemize} \end{frame} \begin{frame}{} \begin{block}{Ситуация риска (рискованная ситуация)} это разновидность неопределенности, когда наступление событий вероятно и может быть определено, т.е. в этом случае объективно существует возможность оценить вероятность событий, возникающих в результате совместной деятельности партнеров по производству, контрдействий конкурентов или противников, влияние природной среды на развитие экономики, внедрение достижений науки в народное хозяйство и т.д. \end{block} \end{frame} \begin{frame}{Характеристики рисковой ситуации} \setbeamertemplate{itemize items}[circle] \begin{itemize}[<+->] \item случайный характер события, который определяет, какой из возможных исходов реализуется на практике (наличие неопределенности); \item наличие альтернативных решений; \item известны или можно определить вероятности исходов и ожидаемые результаты; \item вероятность возникновения убытков; \item вероятность получения дополнительной прибыли. \end{itemize} \end{frame} \begin{frame}[allowframebreaks]{Риск и его особенности}{} \begin{itemize} \item \textbf{Субъективная и объективная стороны риска.} Риск всегда связан с выбором определенных альтернатив (субъективная сторона) и расчетом вероятности их результата (объективная сторона). \item \textbf{Особенности предпринимательского риска.} Предприниматель проявляет готовность идти на риск в условиях неопределенности, поскольку наряду с риском потерь существует возможность дополнительных доходов. \pagebreak \item \textbf{Конструктивная форма предпринимательского риска.} Риск предпринимателя ориентирован на получение значимых результатов нетрадиционными методами. Тем самым он позволяет преодолеть консерватизм, догматизм, косность, психологические барьеры, препятствующие перспективным нововведениям. \pagebreak \item \textbf{Риск и авантюризм.} Вместе с тем риск может стать проявлением авантюризма, если решение принимается в условиях неполной информации, без должного учета закономерностей развития явления. В этом случае риск выступает в качестве дестабилизирующего фактора. \item \textbf{Риск и инновации.} Большинство фирм, компаний добиваются успеха, становятся конкурентоспособными на основе инновационной экономической деятельности, связанной с риском. \pagebreak \item \textbf{Риск и неудачи. }Инициативным, предприимчивым хозяйственникам нужны правовые, политические и экономические гарантии, исключающие в случае неудачи наказание и стимулирующие оправданный риск. Предприниматель должен быть уверен, что возможная ошибка (риск) не может скомпрометировать ни его дело, ни его имидж, так как она произошла вследствие не оправдавшего себя, хотя и рассчитанного риска. \end{itemize} \end{frame} \begin{frame}{Управление риском}{Выбор оптимального решения в условиях риска} \begin{itemize} \item Основной задачей предпринимателя является не отказ от риска вообще, а выборы решений, связанных с риском на основе объективных критериев, а именно: до каких пределов может действовать предприниматель, идя на риск. \item Использование специальных методов анализа в сложных ситуациях. \end{itemize} \end{frame} \subsection{Классификация рисков} % You can reveal the parts of a slide one at a time % with the \pause command: \begin{frame}[shrink=10]{Виды рисков по характеру потерь:}{Чистые риски} \setbeamercovered{transparent} \textbf{Чистые риски.} Несут в себе только потери для предпринимательской деятельности. Их причинами могут быть стихийные бедствия, несчастные случаи, недееспособность руководителей фирм и др.: \end{frame} \begin{frame}[ allowframebreaks ]{Чистые риски} \setbeamertemplate{itemize items}[circle] \begin{itemize} \item природно-естественные риски – это риски связанные с проявлением стихийных сил природы; \item экологические риски связаны с наступлением гражданской ответственности за нанесение ущерба окружающей среде; \pagebreak \item политические риски – это возможность возникновения убытков или сокращения размеров прибыли, являющихся следствием государственной политики; \item транспортные риски связаны с перевозками грузов различными видами транспорта. \end{itemize} \end{frame} \begin{frame}{Виды рисков по характеру потерь:}{Спекулятивные риски} \begin{block}{\textbf{Спекулятивные риски.}} Несут в себе либо потери, либо дополнительную прибыль для предпринимателя. Их причинами могут быть изменение курсов валют, изменение конъюнктуры рынка, изменение условий инвестиций и др. \end{block} \end{frame} \begin{frame}{Виды рисков по сфере возникновения} \begin{block}{Коммерческий риск} \quad– это риск потерь в процессе финансово-хозяйственной деятельности; его причинами могут быть снижение объемов реализации, непредвиденное снижение объемов закупок, повышение закупочной цены товара, повышение издержек обращения, потери товара в процессе обращения и др. \end{block} \end{frame} \begin{frame}{Виды коммерческих рисков} \begin{itemize} \item \textbf{имущественные риски }– это риски от потери имущества предпринимателя по причинам от него не зависящим; \item \textbf{имущественные риски }зависят от убытков по причине задержки платежей, не поставки товара, отказа от платежа и т.п; \item \textbf{риск упущенной выгоды }заключается в том, что возникает финансовый ущерб в результате неосуществления некоторого мероприятия \end{itemize} \end{frame} \begin{frame}{Финансовый риск} \begin{block}{\textbf{Финансовый риск}} \quad - это возможность потерь в результате изменения стоимости финансовых активов или других изменений в финансовом секторе экономики. Причинами здесь могут быть изменение покупательной способности денег, неосуществление платежей, изменение валютных курсов и т.п. \end{block} \end{frame} \begin{frame}[allowframebreaks]{Виды финансовых рисков} \begin{itemize} \item инфляционные риски, которые обусловлены обесцениванием реальной покупательной способности денег, при этом предприниматель несет реальные потери; \item дефляционный риск связан с тем, что при росте дефляции падает уровень цен и, следовательно, снижаются доходы; \item валютные риски связаны с изменением валютных курсов; \pagebreak \item риск ликвидности связан с потерями при реализации ценных бумаг или других товаров из-за изменения оценки их качества и потребительской стоимости; \item процентный риск, возникающий в результате превышения процентных ставок, выплачиваемых по привлеченным средствам, над ставками по предоставленным кредитам; \item кредитный риск, возникающий в случае неуплаты заемщиком основного долга и процентов, причитающихся кредитору; \pagebreak \item биржевые риски представляют собой опасность потерь от биржевых сделок; \item инвестиционные риски возникают из-за неправильного выбора направления вложения капиталов, вида ценных бумаг для инвестирования; \item риск банкротства связан с полной потерей предпринимателем собственного капитала из-за его неправильного вложения. \end{itemize} \end{frame} \subsection{Неопределенность} \begin{frame}{Понятие неопределенности} \begin{block}{\textbf{Неопределенность}} \quad – это неполное или неточное представление о значениях различных параметров в будущем, порождаемых различными причинами и, прежде всего, неполнотой или неточностью информации об условиях реализации решения, в том числе связанных с ними затратах и результатах. \end{block} \end{frame} \begin{frame}{Связь неопределенности и риска} \begin{block}{Неопределенность и риск} Неопределенность, связанная с возможностью возникновения в ходе реализации решения неблагоприятных ситуаций и последствий, характеризуется понятием риск. \end{block} \textbf{Ключевые моменты риска:} \begin{itemize} \item наличие неопределенного исхода события, вероятностные характеристики которого известны; \item возможность потерь, при неблагоприятном развитии событий. \end{itemize} \end{frame} \begin{frame}{Неопределенность как причина возникновения экономического риска} \textbf{Рискованная ситуация} связана со статистическими процессами и ей сопутствуют три сосуществующих условия: \begin{inparaenum}[\itshape a\upshape)] \item наличие неопределенности; \item необходимость выбора альтернативы и \item возможность качественной и количественной оценки вероятности осуществления того или иного варианта\end{inparaenum}. \end{frame} \begin{frame}[allowframebreaks]{Неопределенности по факторам возникновения }{Экономические (коммерческие) и политические} \begin{itemize} \item \textbf{Экономические неопределенности }обусловлены неблагоприятными изменениями в экономике предприятия или в экономике страны, к ним относятся: неопределенность рыночного спроса, слабая предсказуемость рыночных цен, неопределенность рыночного предложения, недостаточность информации о действиях конкурентов и т.д. \pagebreak \item \textbf{Политические неопределенности }обусловлены изменением политической обстановки, влияющей на предпринимательскую деятельность. \end{itemize} \end{frame} \begin{frame}[allowframebreaks]{Неопределенность в зависимости от вероятности выпадения событий} \begin{itemize} \item \textbf{Полная неопределенность }характеризуется близкой к нулю прогнозируемостью наступления события. \pagebreak \item \textbf{Полная определенность} Доверительная вероятность прогнозов развития компании и рыночных тенденций составляет 0,9 – 0,99. \pagebreak \item \textbf{Частичная неопределенность }отвечает таким событиям, прогнозируемость которых лежит в пределах от 0 до 1. \end{itemize} \end{frame} \begin{frame}[allowframebreaks]{Прочие виды неопределенностей} \begin{itemize} \item \textbf{Природная неопределенность }описывается совокупностью факторов, среди которых могут быть: климатические, погодные условия, различного рода помехи (атмосферные, электромагнитные и др.). \pagebreak \item \textbf{Неопределенность внешней среды.} Внутренняя среда предприятия включает факторы, обусловленные деятельностью самого предпринимателя и его контактами. Внешняя среда представлена факторами, которые не связаны непосредственно с деятельностью предпринимателя и имеют более широкий социальный, демографический, политический и иной характер. \pagebreak \item \textbf{Конфликтные ситуации}, в качестве которых могут быть: стратегия и тактика лиц, участвующих в том или ином конкурсе, действия конкурентов, ценовая политика олигополистов и т.п. Проблемы несовпадающих интересов и многокритериального выбора оптимальных решений в условиях неопределенности \end{itemize} \end{frame} \begin{frame}{Неопределенность и принятие экономических решений} На практике принимаются решения на основе детерминированных моделей. Поэтому политика выбора эффективных решений без учета неконтролируемых факторов во многих случаях приводит к значительным потерям экономического, социального и иного содержания. \end{frame} \subsection{Управление риском} \begin{frame}{Пределы риска} \begin{block}{Допустимый предел риска} - это уровень риска в пределах его среднего уровня, то есть среднего по отношению к другим видам деятельности и другим хозяйственным субъектам. \end{block} \end{frame} \begin{frame}[allowframebreaks]{Уровни риска} \textbf{Допустимый риск} \begin{align} R_DR_{max}. \end{align} \end{frame} \begin{frame}{Задачи системы управления риском} \begin{itemize}[<+->] \item обеспечить максимальную сохранность собственных средств; \item минимизировать отрицательное воздействие внешних и внутренних факторов; \item повысить ответственность перед клиентами, контрагентами и инвесторами. \end{itemize} \end{frame} \begin{frame}[ allowframebreaks ]{Принципы управления рисками} \begin{itemize} \item не рисковать, если есть такая возможность; \item не рисковать больше, чем это может позволить собственный капитал; \item думать о последствиях риска и не рисковать многим ради малого; \item не создавать рисковых ситуаций ради получения сверхприбыли; \item держать риски под контролем; \item снижать риски, распределяя их среди клиентов и участников по видам деятельности; \pagebreak \item создавать необходимые резервы для покрытия рисков; \item устанавливать постоянное наблюдение за изменением рисков; \item количественно измерять уровень принимаемых рисков; \item определять новые источники и критические зоны риска и (или) групп операций с повышенным уровнем риска. \end{itemize} \end{frame} \subsection{Анализ и контроль риска} \begin{frame}[shrink=10]{Схема анализа риска} \centering \includegraphics[scale=1 % trim={ } ,trim={1cm 2cm 3cm 0cm},clip] {tikz/risk_analysis_scheme} \end{frame} \begin{frame}[shrink=10]{Этапы анализа риска} \begin{itemize} \item Анализ риска, предполагающий численное определение отдельных рисков и риска проекта (решения) в целом и допустимого в уровень риска данного проекта. \item Результат: картина возможных рисковых событий, вероятность их наступления и последствий. \item Стратегия управления риском, предполагающая меры предотвращения и уменьшения риска. \end{itemize} \end{frame} \begin{frame}[ allowframebreaks ]{Подходы к анализу риска} \begin{itemize} \item \textbf{Качественный анализ. }Словесное описание уровня риска (например, «проекту присвоен высокий уровень риска») путем выявления негативной информации, на основании которой взвешиваются (оцениваются) негативные факторы, влияющие на величину риска. \pagebreak \item \textbf{Количественный анализ.} Величина риска в относительном выражении (например, «максимальные потери по проекту составят 50\% от суммы кредита»). Относительное выражение риска в виде установления допустимого уровня при совершении различных операций применяется при выработке финансовой политики предприятия. \end{itemize} \end{frame} \begin{frame}{Контроль риска } \quad включает в себя все меры, направленные на своевременное выявление риска с целью его снижения или исключения. \textbf{Способы контроля риска:} \begin{itemize}[<+->] \item Внутренний аудит; \item Внешний аудит; \item Внутренний контроль. \end{itemize} \end{frame} \begin{frame}[shrink=5]{Внутренний аудит} \begin{block}{Внутренний аудит} \quad осуществляется внутренним структурным подразделением коммерческой организации: ревизионные комиссии, внутренние аудиторы в рамках существующей на предприятии системы внутреннего контроля. \end{block} \end{frame} \begin{frame}{Внутренний аудит} \begin{itemize}[<+->] \item обеспечение эффективности финансово-хозяйственной деятельности; \item обеспечение достоверности, полноты, объективности и своевременности составления и представления отчетности, а также информационной безопасности; \item соблюдение законодательства; \item борьба с легализацией (отмыванием) доходов. \end{itemize} \end{frame} \begin{frame}[allowframebreaks]{Внешний аудит} Элементы надзора за предприятиями реального сектора выполняют налоговые органы, а также органы лицензирования и др. контролирующие органы. \pagebreak Надзор за деятельностью финансовых организаций осуществляется центральным аппаратом и территориальными учреждениями Банка России (надзорные органы). Ежегодный обязательный аудит в соответствии со статьей 5 Федерального закона №307-ФЗ «Об аудиторской деятельности». \end{frame} \begin{frame}[ allowframebreaks ]{Внутренний контроль} \begin{itemize} \item мониторинга и оценки эффективности политики управления рисками; \item расследования причин возникновения убытков, фактов наступления событий или обстоятельств, приводящих к убыткам (в т.ч. в ходе проведения мониторинга системы внутреннего контроля); \pagebreak \item участия в разработке предложений и мероприятий по оптимизации бизнес-процессов с целью минимизации рисков; \item контроля за рисками новых продуктов и совершением рисковых сделок. \end{itemize} \end{frame} \subsection{Способы снижения риска} \begin{frame}{Способы снижения риска} \begin{itemize} \item отказ от риска; \item снижение риска (резервирование, диверсификация, лимитирование, минимизация); \item передача риска третьему лицу (страхование, хеджирование, распределение). \end{itemize} \end{frame} \begin{frame}[ allowframebreaks ]{Меры по снижению риска} \begin{itemize} \item \textbf{Резервирование }является одним из основных способов управления финансовым риском. \pagebreak \item \textbf{Диверсификация }— это процесс распределения инвестируемых средств между различными объектами вложения капитала, которые непосредственно не связаны между собой, с целью снижения степени риска. \pagebreak \item \textbf{Лимитирование }- это установление предельных сумм расходов по различным операциям. Процесс установления лимитов допустимой величины риска должен быть гибким, основываеться на изучении рынка, суждении и опыте аналитиков. \pagebreak \item \textbf{Минимизация (нивелирование) риска} снижение вероятности наступления событий или обстоятельств, приводящих к убыткам и сокращение потенциальных убытков. Мера реализуется механизмами внутреннего контроля, эффективна до возникновения реальных убытков. \end{itemize} \end{frame} \begin{frame}[ allowframebreaks ]{Варианты передачи риска третьему лицу} \begin{itemize} \item \textbf{Страхование. }При страховании обеспечения от утраты или повреждения, риски передаются страховщику или гаранту. Данный метод наиболее подходит для снижения кредитных рисков. \pagebreak \item \textbf{Хеджирование} - это передача риска участникам финансового рынка путем заключения сделок с использованием производных финансовых инструментов (форварды, фьючерсы, опционы, свопы и т.д.). Как и при страховании, хеджирование требует отвлечения дополнительных ресурсов в виде уплаты опционной премии или внесения депозита, применяется для снижения рыночных рисков. \pagebreak \item \textbf{Распределение } риска между участниками кредитной сделки в виде его включения в стоимость услуг: в процентную ставку (рисковая надбавка), комиссию, штрафные санкции и т.д. \end{itemize} \end{frame} \subsection{Контрольные вопросы} \begin{frame}[allowframebreaks]{Контрольные вопросы} 1. История возникновения понятия риска. 2. Понятие и характеристики риска в современной экономике. 3. Классификация экономических рисков. 4. Виды финансовых рисков и их классификация. \pagebreak 5. Понятие неопределенности. Связь неопределенности и риска. 6. Виды неопределенностей. 7. Пределы и уровни риска. 8. Принципы управления рисками. \pagebreak 9. Анализ и контроль риска. 10. Способы снижения риска. \end{frame} \end{document}\hypertarget{structwchar__selector_3_014_01_4}{ \section{wchar\_\-selector$<$ 4 $>$ Struct Template Reference} \label{structwchar__selector_3_014_01_4}\index{wchar\_\-selector$<$ 4 $>$@{wchar\_\-selector$<$ 4 $>$}} } \subsection*{Public Types} \begin{DoxyCompactItemize} \item typedef uint32\_\-t \hyperlink{structwchar__selector_3_014_01_4_af45ac603ab6fefec66e5c29044b4eed6}{type} \item typedef \hyperlink{structutf32__counter}{utf32\_\-counter} \hyperlink{structwchar__selector_3_014_01_4_a7d7c585ae0819660112b8c8683971b97}{counter} \item typedef \hyperlink{structutf32__writer}{utf32\_\-writer} \hyperlink{structwchar__selector_3_014_01_4_a48042e7fe51c4661397ae7afe3905243}{writer} \end{DoxyCompactItemize} \subsubsection*{template$<$$>$ struct wchar\_\-selector$<$ 4 $>$} \subsection{Member Typedef Documentation} \hypertarget{structwchar__selector_3_014_01_4_a7d7c585ae0819660112b8c8683971b97}{ \index{wchar\_\-selector$<$ 4 $>$@{wchar\_\-selector$<$ 4 $>$}!counter@{counter}} \index{counter@{counter}!wchar_selector< 4 >@{wchar\_\-selector$<$ 4 $>$}} \subsubsection[{counter}]{\setlength{\rightskip}{0pt plus 5cm}typedef {\bf utf32\_\-counter} wchar\_\-selector$<$ 4 $>$::{\bf counter}}} \label{structwchar__selector_3_014_01_4_a7d7c585ae0819660112b8c8683971b97} \hypertarget{structwchar__selector_3_014_01_4_af45ac603ab6fefec66e5c29044b4eed6}{ \index{wchar\_\-selector$<$ 4 $>$@{wchar\_\-selector$<$ 4 $>$}!type@{type}} \index{type@{type}!wchar_selector< 4 >@{wchar\_\-selector$<$ 4 $>$}} \subsubsection[{type}]{\setlength{\rightskip}{0pt plus 5cm}typedef uint32\_\-t wchar\_\-selector$<$ 4 $>$::{\bf type}}} \label{structwchar__selector_3_014_01_4_af45ac603ab6fefec66e5c29044b4eed6} \hypertarget{structwchar__selector_3_014_01_4_a48042e7fe51c4661397ae7afe3905243}{ \index{wchar\_\-selector$<$ 4 $>$@{wchar\_\-selector$<$ 4 $>$}!writer@{writer}} \index{writer@{writer}!wchar_selector< 4 >@{wchar\_\-selector$<$ 4 $>$}} \subsubsection[{writer}]{\setlength{\rightskip}{0pt plus 5cm}typedef {\bf utf32\_\-writer} wchar\_\-selector$<$ 4 $>$::{\bf writer}}} \label{structwchar__selector_3_014_01_4_a48042e7fe51c4661397ae7afe3905243} The documentation for this struct was generated from the following file:\begin{DoxyCompactItemize} \item src/\hyperlink{pugixml_8cpp}{pugixml.cpp}\end{DoxyCompactItemize} @article{hastings1991, title = {Chaos in a {Three}-{Species} {Food} {Chain}}, volume = {72}, issn = {00129658}, url = {http://doi.wiley.com/10.2307/1940591}, doi = {10.2307/1940591}, abstract = {A continuous time model of a food chain incorporating nonlinear functional (and numerical) responses exhibits chaotic dynamics in long-term behavior when biologically reasonable parameter values are chosen. The appearance of chaos in this model suggests that chaotic dynamics may be common in natural food webs.}, language = {en}, number = {3}, urldate = {2019-02-28}, journal = {Ecology}, author = { and }, month = jun, year = {1991}, pages = {896--903}, file = {Hastings_Powell_1991_Chaos in a Three-Species Food.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\T4UCH8TA\\Hastings_Powell_1991_Chaos in a Three-Species Food.pdf:application/pdf} } @article{rosenzweig1963, title = {Graphical {Representation} and {Stability} {Conditions} of {Predator}-{Prey} {Interactions}}, volume = {97}, issn = {0003-0147, 1537-5323}, url = {https://www.journals.uchicago.edu/doi/10.1086/282272}, doi = {10.1086/282272}, language = {en}, number = {895}, urldate = {2019-04-28}, journal = {The American Naturalist}, author = {. and .}, month = jul, year = {1963}, pages = {209--223}, file = {Rosenzweig et MacArthur - 1963 - Graphical Representation and Stability Conditions .pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\EVR9LECB\\Rosenzweig et MacArthur - 1963 - Graphical Representation and Stability Conditions .pdf:application/pdf} } @article{rosenzweig1973, title = {Exploitation in {Three} {Trophic} {Levels}}, volume = {107}, issn = {0003-0147, 1537-5323}, url = {https://www.journals.uchicago.edu/doi/10.1086/282830}, doi = {10.1086/282830}, language = {en}, number = {954}, urldate = {2019-04-28}, journal = {The American Naturalist}, author = {Rosenzweig, .}, month = mar, year = {1973}, pages = {275--294}, file = {Rosenzweig - 1973 - Exploitation in Three Trophic Levels.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\F5S99TKY\\Rosenzweig - 1973 - Exploitation in Three Trophic Levels.pdf:application/pdf} } @article{price1980, title = {Interactions {Among} {Three} {Trophic} {Levels}: {Influence} of {Plants} on {Interactions} {Between} {Insect} {Herbivores} and {Natural} {Enemies}}, volume = {11}, issn = {0066-4162}, shorttitle = {Interactions {Among} {Three} {Trophic} {Levels}}, url = {http://www.annualreviews.org/doi/10.1146/annurev.es.11.110180.000353}, doi = {10.1146/annurev.es.11.110180.000353}, language = {en}, number = {1}, urldate = {2019-04-28}, journal = {Annual Review of Ecology and Systematics}, author = { }, month = nov, year = {1980}, pages = {41--65}, file = {Price et al. - 1980 - Interactions Among Three Trophic Levels Influence.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\RUGCV4YJ\\Price et al. - 1980 - Interactions Among Three Trophic Levels Influence.pdf:application/pdf} } @article{gakkhar2012, title = {Control of chaos due to additional predator in the {Hastings}–{Powell} food chain model}, volume = {385}, issn = {0022247X}, url = {https://linkinghub.elsevier.com/retrieve/pii/S0022247X11005956}, doi = {10.1016/j.jmaa.2011.06.047}, abstract = {A three species Hastings and Powell (HP) food chain model involving another predator of top prey is proposed and studied. The modified food web model is analyzed to obtain the different conditions for which system exhibits stability around the biological feasible equilibrium points. The permanence is established and global stability of boundary equilibrium point Ex is discussed. It is observed through numerical simulations, that four-dimensional model may show stable dynamics in contrast to chaotic dynamics that occurred in three species food chain. Varieties of dynamical behaviors in the food web are possible depending upon the sharing of food between the two predators of the top prey. The results demonstrate that the additional predator play the crucial role in reducing the complexity in the dynamical behavior of the system.}, language = {en}, number = {1}, urldate = {2019-04-28}, journal = {Journal of Mathematical Analysis and Applications}, author = { and }, month = jan, year = {2012}, pages = {423--438}, file = {Gakkhar et Singh - 2012 - Control of chaos due to additional predator in the.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\53BX6R4P\\Gakkhar et Singh - 2012 - Control of chaos due to additional predator in the.pdf:application/pdf} } @article{canale1970, title = {An analysis of models describing predator-prey interaction}, volume = {12}, issn = {0006-3592, 1097-0290}, url = {http://doi.wiley.com/10.1002/bit.260120305}, doi = {10.1002/bit.260120305}, abstract = {Mathematical models of the interaction between predator and host populations have been expressed aa systems of nonlinear ordinary differential equations. Solutions of such systems may be periodic or aperiodic. Periodic oscillatory solutions may depend on the initial conditions of the system or may be limit cycles. Aperiodic solutions can, but do not necessarily, exhibit oscillatory behavior. Therefore, it is important to characterize predatory-prey models on the basis of the possible types of solutions they may possess. This characterization can be accomplished using some well-known methods of nonlinear analysis. Examination of the system singular points and inspection of phase plane portraits have proved to be useful techniques for evaluating the effect of various modifications of early predator-prey models. Of particular interest is the existence of limit cycle oscillations in a model in which predator growth rate is a function of the concentration of prey.}, language = {en}, number = {3}, urldate = {2019-04-28}, journal = {Biotechnology and Bioengineering}, author = {.}, month = may, year = {1970}, pages = {353--378}, file = {Canale - 1970 - An analysis of models describing predator-prey int.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\4JLMP5P9\\Canale - 1970 - An analysis of models describing predator-prey int.pdf:application/pdf} } @article{blasius1999, title = {Complex dynamics and phase synchronization in spatially extended ecological systems}, volume = {399}, issn = {0028-0836, 1476-4687}, url = {http://www.nature.com/articles/20676}, doi = {10.1038/20676}, language = {en}, number = {6734}, urldate = {2019-03-21}, journal = {Nature}, author = { and }, month = may, year = {1999}, pages = {354--359}, file = {Blasius_et al_1999_Complex dynamics and phase synchronization in spatially extended ecological.pdf:/home/gdansereau/Zotero/storage/63H2CWRE/Blasius_et al_1999_Complex dynamics and phase synchronization in spatially extended ecological.pdf:application/pdf} } @article{hastings1993, title = {Chaos in {Ecology}: {Is} {Mother} {Nature} a {Strange} {Attractor}?}, volume = {24}, shorttitle = {Chaos in {Ecology}}, url = {https://doi.org/10.1146/annurev.es.24.110193.000245}, doi = {10.1146/annurev.es.24.110193.000245}, number = {1}, urldate = {2019-04-28}, journal = {Annual Review of Ecology and Systematics}, author = { Hom, . and . .}, year = {1993}, pages = {1--33}, file = {Hastings et al. - Chaos in Ecology Is Mother Nature a Strange Attra.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\435QAZ79\\Hastings et al. - Chaos in Ecology Is Mother Nature a Strange Attra.pdf:application/pdf} } @article{bezanson2017, title = {Julia: {A} {Fresh} {Approach} to {Numerical} {Computing}}, volume = {59}, issn = {0036-1445, 1095-7200}, shorttitle = {Julia}, url = {https://epubs.siam.org/doi/10.1137/141000671}, doi = {10.1137/141000671}, language = {en}, number = {1}, urldate = {2019-04-29}, journal = {SIAM Review}, author = {.}, month = jan, year = {2017}, pages = {65--98}, file = {Bezanson et al. - 2017 - Julia A Fresh Approach to Numerical Computing.pdf:C\:\\Users\\Gabriel\\Zotero\\storage\\HEMXXR59\\Bezanson et al. - 2017 - Julia A Fresh Approach to Numerical Computing.pdf:application/pdf} } @article{rackauckas2017, author = {Rackauckas, }, doi = {10.5334/jors.151}, journal = {The Journal of Open Research Software}, keywords = {Applied Mathematics}, note = {Exported from https://app.dimensions.ai on 2019/05/05}, number = {1}, pages = {}, title = {DifferentialEquations.jl – A Performant and Feature-Rich Ecosystem for Solving Differential Equations in Julia}, url = {https://app.dimensions.ai/details/publication/pub.1085583166 and http://openresearchsoftware.metajnl.com/articles/10.5334/jors.151/galley/245/download/}, volume = {5}, year = {2017} } \section{A Stream Based Approach} Streaming algorithms provide excellent solutions to many problems where data sets are large enough that we wish (or need) to sacrifice exactness for low memory usage and time consumption. \texttt{Ikonomovska et al.(2013)}\citep{ikonomovskazelke}: "Streaming algorithms drop the demand of random access to the input. Rather, the input is assumed to arrive in arbitrary order as an input stream. Moreover, streaming algorithms are designed to settle for a working memory that is much smaller than the size of the input." To be precise, they require that the size of the working memory is sublinear in both the cardinality of the stream and the universe. Due to this nature of streaming algorithms they are not commonly used for problems that require analysing parts of non-constant, non-parameterized size of the data set, for each given input. Hence an approach to solve our problem --- sorting --- based on streaming algorithms will provide some interesting trade-offs. The definition of our stream, $S \in (U\times R)^{|S|}$ is the same as the sequence described in the introduction. With this definition we get a strict turnstile stream --- in fact the delta $r$ for \textit{every} stream element is positive. The simplest algorithm to solve our problem is then simply calculating the normalized frequency vector for the stream, and sort it when queried. However, this is not very satisfactory. The working memory is sublinear in $|S|$, but linear in the universe size $|U|$. It does not provide a current solution either, as we have to sort the frequency vector when queried. On the positive side, the solution provided by the algorithm is exact. To be precise, this algorithm would require $O(m)$ working memory, and constant time for each stream item, $m$ being the number of distinct movies in the stream, which we assume to be, or closely approach, $|U|$. A query would then require $O(m \log m)$ time. The rest of this section discusses techniques that alleviate these problems with different trade-offs. \subsection{Order Maintenance} \ref{sec:ordermaint} In the simple algorithm, results were not \textit{current}, because every query required a sort of the frequency vector --- which is long. If we allow ourselves to use more than constant processing time per stream element, this problem can be solved by maintaining an \textit{always sorted} data structure with pointers into the frequency vector, such as a search tree. This algorithm is equivalent to maintaining an ordered set of running averages, and is thus the same as the online-sorting approach described previously. Knowing that for most stream items $(j,r)$, the movie represented by $j$ will already be in the ordered set, it might be possible to achieve insertion time linear in the number of inversions needed to reorder the set, though it is not clear that this should improve the $\log n$ insertion time in binary search trees. In summary, we get $O(\log n)$ processing time for each stream element, but queries can now be performed in $O(n)$. This a very natural change from the simple algorithm, that really only moves the required work from the time of querying, to the time of input. \subsection{Approximation based on sampling} In addition to the lack of currency, the simple algorithm requires a lot of memory. Not surprisingly, streaming algorithms lets us buy a lower memory requirement, at the cost of exactness. There seem to be two obvious approaches; normal reservoir sampling over the stream, or sampling over the movies, deliberately making sure that all movies are represented in the sample. The later obviously fails to improve memory consumption, and is only suggested because taking a random sample over the stream seems dangerous, as it might well discard movies from the stream, by not picking any of their ratings for the sample. As it turns out, this is not a big problem. Although a uniformly random sample of ratings is not an answer to our original problem, it does have some nice properties. As stated in \sloppy{\texttt{Ikonomovska et al.(2013)}\citep{ikonomovskazelke}: all $|S| \choose k$ possible samples, where $k$ is the sample size, are equally likely to be our result. It follows directly from this that popular ratings for a movie are more likely to occur than unpopular ones. However, the likelihood of a movie occurring in the sample similarly correlates to how many ratings it has, not -- as we would want -- how high it's average rating is. In other words, reservoir sampling gives us a random set of ratings, with no guarantee that the sample contains good movies. The common reservoir sampling algorithm described in \texttt{Ikonomovska et al.(2013)}\citep{ikonomovskazelke}: remembers $k$ samples $K_0..K_{k}$. After the first $k$, each stream element \raggedright{$\alpha_i, k < i \leq |S|$}, replaces one of the $k$ samples with probability $k/i$, choosing the sample to be replaced at random. The sampling approach solves our memory issues by parametrizing the memory consumption. The algorithm uses $O(k)$ memory, independent of both $|U|$ and $|S|$. We can modify the reservoir sampling algorithm to keep running averages instead of samples, modifying samples when observing a stream element that refers to a movie already being monitored, and only replacing a sample when observing an element that is not already being monitored (i.e. not currently in the set of remembered movie samples). We then no longer get sampled ratings, but estimates of the movie averages --- which is what we wanted. Maintaining running averages can be thought of as keeping a frequency vector, and changing the stream so that each element $(j, r)$ becomes $(j, r')$, where $r'$ is the change $r$ imposes on the kept average for movie $j$. Despite the possibility of $r'$ being negative, we are still in the strict turnstile model, as the average will never be negative, regardless of what subset if $S$ we look at. Direct application of this analogue is infeasible, as it would require $O(|U|)$ memory to keep track off the counters necessary to calculate $r'$ from $r$. However, it would be interesting to apply it with estimations of those counters. The problem of missing movies persists however. If wish achieve the $O(k)$ memory bound, we can not hope to find the exact solution using sampling in this way. However, if we limit our problem to find the top-l movies, we can. %As described in \texttt{Ikonomovska et al.(2013)}\citep{ikonomovskazelke}: we can adjust the quality of %the result by adjusting $k$ as following. If we alter our running-average sampling to replace the \textit{smallest} sample, instead picking one at random, the probability of our $k$ samples containing the top $l K_k$ must exist in the sample-set. I.e any movie with an actual frequency higher than the lowest estimated frequency in the sample, must be in the sample. \item[Metwally Theorem 2] \hfill \\ Whether or not the movie with actual rank $i$ occupies the $i^{th}$ position in $K$, $K_i \geq F_i$ \end{description} As mentioned, the \textit{space-saving} algorithm is intended for estimating frequencies, and needs modification to work with averages. A first thought might be to run two instances in parallel, estimating the count of ratings and sum of ratings respectively (equivalent to frequencies in the cash register and turnstile model respectively). This clearly does not work, as the set of most frequently rated movies is not necessarily similar to the set of movies with a high sum of ratings. Instead, we can adapt the algorithm to maintain running averages instead of frequencies. This presents a problem however. The proofs for the presented lemmas and theorems depend on the fact fact that each replacement in the sample-set increments the counter by $1$. But incrementing by $1$ will not provide us with running averages. Hence, keeping running averages will break the proofs. For this approach to be valuable in our setting, it remains to provide new error bounds, or modify the algorithm so that the ones from \texttt{Metwally} holds. \subsection{Approximation based on Sketching} \label{sec:sketching} Sketching lets us sacrifice exactness for lower memory consumption, without having to worry about loosing entire movies from the solution. We pay for this with an error bound that is linear in $|S|$. The count-min sketch algorithm as described in {\texttt{Ikonomovska et al.(2013)}\citep{ikonomovskazelke} estimates the frequency vector of the stream, using $O(\log(1/\delta)/\varepsilon + \log n \cdot \log(1/\delta))$ working memory. Where $\varepsilon$ determines the expected error $\varepsilon \cdot |S| /2$, of each frequency and $\delta$ is the probability of the actual error exceeding that bound. The error is guaranteed to be positive i.e. the algorithm cannot underestimate the actual frequencies. A simple adaption of the algorithm to our problem (estimating averages, not frequencies) is to run two instances of the algorithm in parallel, estimating respectively the sum and frequency of ratings for each movie. We denote the actual sum and frequency for a movie $j$ by $R_j$ and $F_j$, and the estimates produced by the algorithm $\bar{R_j}$ and $\bar{F_j}$ We can then get an estimate of the average rating for a movie $j$ by $\bar{R_j}/\bar{C_j}$. If we use the same set of hashing functions and vector lengths for sketching $R_j$ and $C_j$, we see that $R_j-\bar{R_j} \le F_j-\bar{F_j}$. Hence, the resulting error is positive. This guarantee only holds because our ratings are defined to be positive. This also shows $$\frac{R_j}{F_j} \le \frac{\bar{R_j}}{\bar{F_j}} \le \frac{\epsilon_r \cdot R_j}{\epsilon_f \cdot F_j}$$ Where $\epsilon_f$ is our error bound $\varepsilon \cdot |S|/2$, and $\epsilon_r$ is $\epsilon_f$ times the maximum rating (10, in our case). The count-min sketch algorithm can be modified as described in section \ref{sec:ordermaint}, to provide fast query times as well. matcdac/IETF_RFCs @misc{rfc5070, series = {Request for Comments}, number = 5070, howpublished = {RFC 5070}, publisher = {RFC Editor}, doi = {10.17487/RFC5070}, url = {https://rfc-editor.org/rfc/rfc5070.txt}, author = { and and }, title = {{The Incident Object Description Exchange Format}}, pagetotal = 92, year = 2007, month = dec, abstract = {The Incident Object Description Exchange Format (IODEF) defines a data representation that provides a framework for sharing information commonly exchanged by Computer Security Incident Response Teams (CSIRTs) about computer security incidents. This document describes the information model for the IODEF and provides an associated data model specified with XML Schema. {[}STANDARDS-TRACK{]}}, } % file: parts/family.tex %%%%%%%%%%%%%%% \begin{frame}{} \centerline{\teal{\Large The Boy/Girl Puzzle}} \vspace{0.30cm} \fignocaption{width = 0.60\textwidth}{figs/family} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \begin{exampleblock}{Both Girls (CS Problem $5.3-12$)} Mr. and Mrs. Smith have two children of different ages,\\ what is the probability that they has \red{two girls}, \begin{enumerate}[(a)] \item given that \violet{one of the children} is a girl? \item given that \violet{the older child} is a girl? \end{enumerate} \end{exampleblock} \fignocaption{width = 0.35\textwidth}{figs/girl} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \[ G_1: \text{the older child is a girl} \] \[ G_2: \text{the younger child is a girl} \] \begin{align*} \onslide<2->{\teal{\Pr\set{G_1 \land G_2 \mid G_1 \lor G_2}}} \onslide<3->{&= \frac{\Pr{\set{G_1 \land G_2}}}{\Pr{\set{G_1 \lor G_2}}} \\} \onslide<4->{&= \frac{\Pr\set{G_1 \land G_2}}{\Pr\set{G_1} + \Pr\set{G_2} - \Pr\set{G_1 \land G_2}} \\} \onslide<5->{&= \frac{1/4}{3/4} = \frac{1}{3}} \end{align*} \begin{align*} \onslide<6->{\teal{\Pr\set{G_1 \land G_2 \mid G_1}}} \onslide<7->{= \frac{\Pr{\set{G_1 \land G_2}}}{\Pr\set{G_1}}} \onslide<8->{= \frac{1/4}{1/2} = \frac{1}{2}} \end{align*} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \[ \boxed{(G_1, G_2), (G_1, B_2), (B_1, G_2), (B_1, B_2)} \] \pause \[ \teal{\Pr\set{G_1 \land G_2 \mid G_1 \lor G_2} = \frac{1}{3}} \] \pause \[ \red{(G_1, G_2)}, (G_1, B_2), (B_1, G_2), \lgray{(B_1, B_2)} \] \pause \[ \teal{\Pr\set{G_1 \land G_2 \mid G_1} = \frac{1}{2}} \] \pause \[ \red{(G_1, G_2)}, (G_1, B_2), \lgray{(B_1, G_2)}, \lgray{(B_1, B_2)} \] \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \fignocaption{width = 0.55\textwidth}{figs/wlog} \centerline{\red{\large Suppose that the older child is a girl:}} \pause \[ \Pr\set{G_1 \land G_2 \mid G_1 \lor G_2} = \Pr\set{G_1 \land G_2 \mid G_1} = \frac{1}{2} \] \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \[ \teal{\Pr\set{G_1 \land G_2 \mid G_1 \lor G_2} = \frac{1}{3}} \] \[ \teal{\Pr\set{G_1 \land G_2 \mid G_1} = \frac{1}{2}} \] \pause \[ \Pr\set{G_1 \land G_2 \mid G_1 \lor G_2} = \red{\boxed{\frac{2}{3}}} \Pr\set{G_1 \land G_2 \mid G_1} \] \pause \[ \Pr\set{G_1 \land G_2 \mid G_1 \lor G_2} = \red{\boxed{\Pr\set{G_1 \mid G_1 \lor G_2}}} \Pr\set{G_1 \land G_2 \mid G_1} \] \pause \[ \Pr\set{G_1 \mid G_1 \lor G_2} = \frac{\Pr\set{G_1 \land (G_1 \lor G_2)}}{\Pr\set{G_1 \lor G_2}} \pause = \frac{\Pr\set{G_1}}{\Pr\set{G_1 \land G_2}} = \frac{2}{3} \] \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \fignocaption{width = 0.40\textwidth}{figs/easy} \pause \fignocaption{width = 0.25\textwidth}{figs/keep-calm-not-finished} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \begin{exampleblock}{Both Girls (CS Problem $5.3-12$)} Mr. and Mrs. Smith have two children of different ages,\\ what is the probability that they has \red{two girls}, \begin{enumerate}[(a)] \item given that \violet{one of the children} is a girl? \end{enumerate} \end{exampleblock} \pause \vspace{0.50cm} \centerline{\red{\large $Q:$ {\bf How} do you know that \violet{``one of the children is a girl''}?}} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% % \begin{frame}{} % \centerline{\red{\large $Q:$ {\bf How} do you know that \violet{``one of the children is a girl''}?}} % % \pause % \vspace{0.30cm} % \begin{columns} % \column{0.25\textwidth} % \fignocaption{width = 0.85\textwidth}{figs/house1} % \column{0.25\textwidth} % \fignocaption{width = 0.75\textwidth}{figs/house0} % \column{0.25\textwidth} % \fignocaption{width = 0.75\textwidth}{figs/house2} % \column{0.25\textwidth} % \fignocaption{width = 0.90\textwidth}{figs/house3} % \end{columns} % % \pause % \vspace{0.80cm} % \begin{enumerate}[(I)] % \setlength{\itemsep}{8pt} % \item From all families with two children, at least one of whom is a girl, \\ % \red{a {\footnotesize (Smith's)} family is chosen at random}. % \pause % \item From all families with two children, \red{one child {\footnotesize (of Smith)} is selected} \\ % at random that happens to be a girl. % \end{enumerate} % \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \centerline{\red{\large $Q:$ {\bf How} do you know that \violet{``one of the children is a girl''}?}} \pause \vspace{0.30cm} \fignocaption{width = 0.45\textwidth}{figs/two-rooms} \pause \begin{enumerate}[(I)] \setlength{\itemsep}{8pt} \item I \purple{\textsc{know}} them well and I tell you that ``one of the children is a girl''. \pause \item I \purple{\textsc{don't know}} them. I just open a room door and see a girl. \end{enumerate} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \centerline{\red{\large $Q:$ {\bf How} do you know that \violet{``one of the children is a girl''}?}} \vspace{0.60cm} \begin{enumerate}[(I)] \setcounter{enumi}{1} \item \red{$g:$} I \purple{\textsc{don't know}} them. I just open a room door and see a girl. \end{enumerate} \pause \vspace{0.50cm} \[ \Pr\set{G_1 \land G_2 \mid g} = \frac{\set{G_1 \land G_2 \land g}}{\Pr\set{g}} \pause = \frac{1/4}{1/2} = \frac{1}{2} \] \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \begin{exampleblock}{After-class Exercise:} A new couple, \teal{known to have two children}, has just moved into town. \\ Suppose that the mother is \teal{encountered walking with one of her children}. \\ If this child is a girl, what is the probability that \red{both children are girls}? \end{exampleblock} \vspace{0.50cm} \fignocaption{width = 0.40\textwidth}{figs/mother-girl-walking.png} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \begin{frame}{} \fignocaption{width = 0.40\textwidth}{figs/qrcode-boy-girl-wiki} \vspace{-0.30cm} \centerline{Boy or Girl paradox (wiki)} \end{frame} %%%%%%%%%%%%%%% metaborg/declare-your-language \documentclass[a4paper, leqno]{exam} \usepackage{graphicx} \usepackage{palatino} \usepackage{fancyvrb} \usepackage{amsmath, amssymb, amsfonts} \usepackage{color} \usepackage{alltt} \pagestyle{headandfoot} \header{\textsf{IN4303: Compiler Construction}}{\textsf{\textsc{Exercises}}}{\textsf{Jan 24, 2018}} \headrule \footer{}{\textsf{Page \thepage\ of \numpages}}{} \footrule \begin{document} \qformat{\textbf{Exercise \thequestion:} \thequestiontitle\hfill} \begin{questions} \fvset{frame=lines} \fvset{framesep=8pt} \begin{Verbatim} l1: if y = 0 goto l2 q := x / y t := q * y r := x - t x := y y := r goto l1 l2: return x \end{Verbatim} \titledquestion{Control flow} \begin{parts} \part Construct the control flow graph for the intermediate code above. \begin{solution} By line number: \begin{Verbatim} 1 -> 2 -> 3 -> 4 -> 6 -> 7 -> 1 1 -> 8 \end{Verbatim} \end{solution} \part Show the control flow graph with basic blocks. \begin{solution} \begin{Verbatim} 1 -> [2,3,4,5,6,7] -> 1 1 -> 8 \end{Verbatim} \end{solution} \end{parts} \titledquestion{Live variable analysis} \begin{parts} \part When is a variable considered live? \begin{solution} A variable is live if it holds a value that may be observed (read) in the future of the program. The writer of the liveness analysis defines which variables are live at the end of the control flow of a program. \textbf{Alternative wording:} A variable is live in a program between the point where it was last assigned to and the point where it is read. A variable is also live between the start of the program and the first read of the variable is there is no assignment of that variable in between. The writer of the liveness analysis defines which variables are live at the end of the control flow of a program. \end{solution} \part What is the information that live variables analysis calculates and associates with each control flow graph node? \begin{solution} A set of names of variables that are live at that point in the program. \end{solution} \part Is live variable analysis a \emph{may} or \emph{must} analysis? \begin{solution} It is a \emph{may} analysis. \end{solution} \fvset{frame=lines} \fvset{framesep=8pt} \begin{Verbatim} int x = 0; if(y > 0) { x = 1; } else { y = 0; } return x; \end{Verbatim} \part Given the above MiniJava method body, provide the live variables analysis result per line of code. Explain where the analysis result demonstrates the analysis property (may or must) you answered above. \begin{solution} In this program variable \Verb+x+ is live after initialisation despite being immediately reassigned in one branch of the \Verb+if+ statement. There is a path where \Verb+x == 0+ is observed, namely when \Verb+y > 0+. \fvset{frame=lines} \fvset{framesep=8pt} \begin{Verbatim} // (in , out) int x = 0; // {y}, {x,y} if(y > 0) { // {x,y} , {x} \/ {} x = 1; // {} , {x} } else { // y = 0; // {x} , {x} } // return x; // {x} , {x} \end{Verbatim} \end{solution} \part What information does live variables expose at assignments, and how can this be used for optimisation? \begin{solution} At an assignment when the variable that's assigned isn't live, that assignment is not observed by the rest of the program, and is therefore dead code. This dead code can be eliminated in an optimisation step. \end{solution} \end{parts} \end{questions} \end{document} \hypertarget{_double_test_case_8php}{}\doxysection{vendor/phpunit/phpunit/tests/\+\_\+files/\+Double\+Test\+Case.php File Reference} \label{_double_test_case_8php}\index{vendor/phpunit/phpunit/tests/\_files/DoubleTestCase.php@{vendor/phpunit/phpunit/tests/\_files/DoubleTestCase.php}} \doxysubsection*{Data Structures} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{class_double_test_case}{Double\+Test\+Case}} \end{DoxyCompactItemize} sikouhjw/zhangyu1000 \section{多元函数微分学} \subsection{概念} \begin{ti} 设 $f(x,y) = \ee^{x + y} \Bigl[ x^{\frac{1}{3}} (y - 1)^{\frac{1}{3}} + y^{\frac{1}{3}} (x - 1)^{\frac{2}{3}} \Bigr]$，则在点 $(0,1)$ 处的两个偏导数 $f_{x}'(0,1)$ 和 $f_{y}'(0,1)$ 的情况为\kuo. \onech{两个偏导数均不存在}{$f_{x}'(0,1)$ 不存在，$f_{y}'(0,1) = \frac{4}{3}\ee$}{$f_{x}'(0,1) = \frac{\ee}{3}$，$f_{y}'(0,1) = \frac{4}{3}\ee$}{$f_{x}'(0,1) = \frac{\ee}{3}$，$f_{y}'(0,1)$ 不存在} \end{ti} \begin{ti} 函数 $z = f(x,y) = \sqrt{|xy|}$ 在点 $(0,0)$ \kuo. \onech{连续，但偏导数不存在}{偏导数存在，但不可微}{可微}{偏导数存在且连续} \end{ti} \begin{ti} 函数 $f(x,y) = \sqrt[3]{x^{2} y}$ 在点 $(0,0)$ 处： \begin{enumerate} \item 是否连续，说明理由; \item 偏导数是否存在，说明理由; \item 是否可微，说明理由. \end{enumerate} \end{ti} \begin{ti} 设 \begin{enumerate} \item $f(x,y) = \begin{cases} \frac{x^{2} y^{2}}{\left( x^{2} + y^{2} \right)^{3/2}}, & (x,y) \ne (0,0),\\ 0, & (x,y) = (0,0); \end{cases}$ \item $g(x,y) = \begin{cases} \bigl( x^{2} + y^{2} \bigr) \sin \frac{1}{x^{2} + y^{2}}, & (x,y) \ne (0,0),\\ 0, & (x,y) = (0,0). \end{cases}$ \end{enumerate} 讨论它们在点 $(0,0)$ 处的 \begin{enumerate} \item[\libcirc{1}] 偏导数的存在性; \item[\libcirc{2}] 函数的连续性; \item[\libcirc{3}] 方向导数的存在性; \item[\libcirc{4}] 函数的可微性. \end{enumerate} \end{ti} \begin{ti} 已知 $f(x,y) = \bigl( xy + xy^{2} \bigr) \ee^{x + y}$，则 $\frac{\partial^{10}f}{\partial x^{5} \partial y^{5}} = $\htwo. \end{ti} \begin{ti} \begin{enumerate} \item 设 $y = \frac{1}{x(1 - x)}$，求 $\frac{\dd^{n}y}{\dd{x^{n}}}$; \item 设 $z = \frac{y^{2}}{x(1 - x)}$，求 $\frac{\partial^{n}z}{\partial x^{n}}$. \end{enumerate} \end{ti} \begin{ti} 设 $z = y^{2} \ln \bigl( 1 - x^{2} \bigr)$，求 $\frac{\partial^{n}z}{\partial x^{n}}$. \end{ti} \begin{ti} 设 $z = x \ln \bigl[ \bigl( 1 + y^{2} \bigr) \ee^{x^{2} \sin y} \bigr]$，则 $\frac{\partial^{4}z}{\partial y^{2} \partial x^{2}} = $\htwo. \end{ti} \begin{ti} 设函数 $f(x,y)$ 的一阶偏导数连续，在点 $(1,0)$ 的某邻域内有 \[ f(x,y) = 1 - x - 2y + o\left( \sqrt{(x - 1)^{2} + y^{2}} \right) \] 成立. 记 $z(x,y) = f\bigl( \ee^{y}, x + y \bigr)$，则 $\dd{[z(x,y)]}|_{(0,0)} = $\htwo. \end{ti} \begin{ti} 设函数 $f(x,y)$ 及它的二阶偏导数在全平面连续，且 $f(0,0) = 0$，$\Bigl| \frac{\partial f}{\partial x} \Bigr| \leq 2 \bigl|x - y\bigr|$，$\Bigl| \frac{\partial f}{\partial y} \Bigr| \leq 2 \bigl|x - y\bigr|$. 求证：$\bigl|f(5,4)\bigr| \leq 1$. \end{ti} \begin{ti} 二元函数 $f(x,y) = x^{y}$ 在点 $(\ee,0)$ 处的二阶(即 $n = 2$) 泰勒展开式(不要求写出余项)为\htwo. \end{ti}varqox/img2texmain-untexed/2070.tex n \leq 300\,000 data/results/tables/Table1.tex { \def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi} \begin{tabular}{l*{1}{cccccc}} \hline\hline &\multicolumn{6}{c}{} \\ & Mean& Median& sd& Min& Max& Obs\\ \hline \textbf{Prices} & & & & & & \\ Unit buyer price & 3.6& 3.5& 1.3& 2.4& 9.3& 152\\ Subscription price per issue& 2.8& 2.7& 0.7& 1.9& 5.6& 148\\ Display ad rate (listed price)& 121.1& 114.5& 81.0& 17.5& 274.2& 121\\ \textbf{Revenues \& journalists}& & & & & & \\ Total revenues (million \euro)& 425& 271& 403& 19& 1482& 162\\ Revenues from advertising (million \euro)& 228& 103& 258& 7& 864& 161\\ Revenues from sales (million \euro)& 199& 145& 181& 12& 657& 162\\ Share of advertising in total revenues (\%)& 47.4& 51.1& 21.3& 8.0& 81.0& 162\\ Number of journalists& 117& 85& 81& 21& 326& 158\\ \textbf{Circulation}& & & & & & \\ Total circulation & 295,210& 181,574& 292,838& 16,112& 1,143,676& 162\\ Share of subscribers (\%)& 25.6& 18.5& 26.3& 0.7& 92.3& 163\\ \textbf{Content} & & & & & & \\ Number of pages & 19& 17& 7& 8& 38& 138\\ News hole (nonadvertising space)& 13& 13& 4& 6& 25& 138\\ Advertising space & 5& 4& 4& 0& 16& 138\\ \hline\hline \end{tabular} } % % 82 % \chapter{The Fundamental Properties of Analytic Functions; Taylor's, Laurent's and Liouville's Theorems} \Section{5}{1}{Property of the elementary functions.} The reader will be already familiar with the term elementary function, as used (in text-books on Algebra, Trigonometry, and the Differential Calculus) to denote certain analytical expressions* depending on a variable z, the symbols involved therein being those of elementary algebra together with exponentials, logarithms and the trigonometrical functions; examples of such expressions are 1 § Z-, e~, iogz, arcsm '-. Such combinations of the elementary functions of analysis have in common a remarkable property, which will now be investigated. Take as an example the function e . Write e'=f z). Then, if 2 be a fixed point and if z' be any other point, we have f z')-f z) \ e~ -e' \, e' '- ' - 1 z -z z - z z - z f, z' - z (z -z)- and since the last series in brackets is uniformly convergent for all values of it follows \hardsectionref{3}{7}) that, as z'->z, the quotient z - z tends to the limit e, uniformly for all values of arg z - z). This shews that tlie limit of f z:)-f z) z - z is in this case independent of the path by which the point z tends towards coincidence witJt z. It wall be found that this property is shared by many of the well-known elementary functions; namely, that iif(z) be one of these functions and h. be * The reader will observe that this is uot the sense in which the term function is defined \hardsectionref{3}{1}) in this work. Thus e.g. .r - hj and | z \ are functions of z (-x + iy) in the sense of \hardsectionref{3}{1}, but are not elementary functions of the type under consideration. % % 83 % any complex number, the limiting value of exists and is independent of the mode in luhich h fends to zero. The reader will, however, easily prove that, ii f(z)=x -iy, where z = x- iy, then lim' - - - - IXJ- is not independent of the mode in which A- >0. \Subsection{5}{1}{1}{Occasional failure of the jwoperty.} For each of the elementary functions, however, there will be certain points z at which this property will cease to hold good. Thus it does not hold for the function l/( - a) at the point z = a, since i Qh\ z - a+h z-a does not exist when z = a. Similarly it does not hold for the functions log z and z at the point z = Q. These exceptional points are called singular points or singulaHties of the function f(z) under consideration; at other points f(z) is said to be analytic. The property does not hold good at any point for the function \ z. \Subsection{5}{1}{2}{Cauchy's* definition of an analytic function of a complex variable.} The property considered in § b\ \ will be taken as the basis of the definition of an analytic function, which may be stated as follows. Let a two-dimensional region in the -plane be given; and let w be a function of z defined uniquely at all points of the region. Let z, z- hz be values of the variable z at two points, and u, u + Bu the corresponding values of u. Then, if, at any point z within the area, - tends to a limit when 8x->0, By-*0, independently (where 8z = 8x + iBy), u is said to be a function of z which is monogenic or analytic j" at the point. If the function is analytic and one-valued at all points of the region, we say that the function is analytic throughout the region. We shall frequently use the word ' function ' alone to denote an analytic function, as the functions studied in this work will be almost exclusively analytic functions. * See the memoir cited in § 5 "2. t The words ' regular ' and ' holomorphic ' are sometimes used. A distinctiou has been made by Borel between ' monogenic ' and ' analytic ' functions in the case of functions with an infinite number of singularities. See \hardsubsectionref{5}{5}{1}. X See \hardsectionref{5}{2} cor. 2, footnote. 6-2 % % 84 % In the foregoing definition, the function u has been defined only within a certain region in the -plane. As will be seen subsequently, however, the function u can generally be defined for other values of 2 not included in this region; and (as in the case of the elementary functions already discussed) may have singularities, for which the fundamental property no longer holds, at certain points outside the limits of the region. We shall now state the definition of analytic functionality in a more arithmetical form. Let f(z) be analytic at z, and let e be an arbitrary positive number; then we can find numbers I and h, h depending on e) such that z -z I whenever \ z' - z\ < h. Vi f(z) is analytic at all points of a region, I obviously depends on z; we consequently write 1 = f z). Hence /(/) = f(z) + - z) f(z) + v z'- z\ where v is a function of z and z such that ! v < e when \ z -z\ < t>. Example 1. Find the points at which the following functions are not analytic : z - \ (i) 2 . (ii) cosec2 (2 = /itt, w any integer). (iii) - - - - - (3 = 2,3). 1 (iv) ez (j = 0). (V) z- ) zf (2 = 0,1). Example 2. If z = x - iy, f(z) = u ii\ where u, v, x, y are real and / is an analytic function, shew that 8m "bv cu cv /T - s :r =, : =-; . \addexamplecitation{Kiemann.} ex oy oy ex \Subsection{5}{1}{3}{An application of the modified Heine-Borel theorem.} Let f(z) be analytic at all points of a continuum; and on any point z of the boundar ' of the continuum let numbers f (z), S (S depending on z) exist such that \ f(z')-f(z)- z'-z)Mz)e (An + In Xi)\ \ '2, Avhere An is the area of the complete square of which Dn is part, In is the side of this square and \ n is the length of the part of G which lies inside this square. Hence, if \ be the whole length of G, while I is the side of a square which encloses all the squares Gn and Dn, f(z)dz k S f z)dz + t \ f z)dz |J(0 I n = \'J(Cn) I n = \ \ J Dn) | ( M N iV ) <4eV2 An+ S An' + l tXn\ (m = 1 m = 1 n=l ) < 4e V2 . I' + X). Now e is arbitrarily small, and I, \ and 1 f z)dz are independent of e. I iC)' It therefore follows from this inequality that the only value which I f(z) dz can have is zero; and this is Cauchy's result. % % 87 % Corollary . If there are two jjaths z AZ and Zq,BZ from 2o to Z, and if /(z) is a function of z analytic at all points on these curves and throughout the domain enclosed by these two paths, then / f(z) dz has the same value whether the path of integration is Z(iAZ (jv ZqBZ. This follows from the fact that ZqAZBzq is a contour, and so the integral taken round it (which is the difference of the integrals along zqAZ and ZoBZ) is zero. Thus, if /(2) be an analytic function of z, the value of / f z)dz is to a certain extent J AB independent of the choice of the arc AB, and depends only on the terminal points A and B. It must be borne in mind that (his is only the case whenf z) is an analytic function in the sense of \hardsubsectionref{5}{1}{2}. Corollary 2. Suppose that two simple closed curves (7 and Cj are given, such that Co completely encloses Cj, as e.g. would be the case if Cq and Ci were confocal ellipses. Suppose moreover that/ (2) is a function which is analytic* at all points on Cq and Cj and throughout the ring-shaped region contained between Cq and C . Then by drawing a network of intersecting lines in this ring-shaped space, we can shew, exactly as in the theorem just proved, that the integral //( dz is zero, ivhere the integration is taken round the whole boundary of the nng-shaped space; this boundary consisting of two curves Co and C , the one described in the counter-clochvise direction aiid the other described in the qlockwise direction. Corollary 3. In general, if any connected region be given in the 3-plane, bounded by any number of simple closed curves Co, Ci, Co, ..., and if /(z) be any function of z which is analytic and one-valued everywhere in this region, then I f z)dz is zero, where the integral is taken round the whole boundary of the region; this boundary consisting of the curves Co, Ci, ..., each described in such a sense that the region is kept either uhvays on the right or always on the left of a person walking in the sense in question round the boundary. An extension of Cauchy's theorem I f(z) dz = 0, to curves lying on a cone whose vertex is at the origin, has been made by Ravut (N'ouv. Annales de Math. (3) xvi. (1897), pp. 365-7). Morera, Moid, del 1st. Lombardo, xxii. (1889), p. 191, and . Amer. Math. Soc. II. (1896), pp. 296-302, have shewn that the property f z)dz = may be taken as the property defining an analytic function, the other properties being deducible from it. (See p. 110, example 16.) Example. A ring-shaped region is bounded by the two circles | s | = 1 and | z | = 2 in the 2-plane. Verify that the value of I -, where the integral is taken round the boundary of this region, is zero. * The phrase 'analytic throughout a region' implies one-valuedness (§ 5-12); that is to say that after z has described a closed path surrounding Co, f(z) has returned to its initial value. A function such as log z considered in the region 1 | | 2 will be said to be ' analytic at all points of the region. ' % % 88 % For the boundary consists of the circumference ]2| = 1, described in the clockwise direction, together with the circumference |s| = 2, described in the counter-clockwise direction. Thus, if for points on the iirst circumference we write 2 = e, and for points on the second circumference we write z = '2e<'t>, then 6 and cp are real, and the integral becomes jo e' jo 2e'* \Subsection{5}{2}{1}{The value of an analytic function at a 'point, expressed as an integral taken round a contour enclosing the point.} Let C he a contour within and on which /( ) is an analytic function of z. Then, if a be any point within the contour, z - a is a function of z, which is analytic at all points within the contour G except the point z = a. Now, given e, we can find 3 such that \ fi z)-f a)- z-a)f' a)\ \ \ z-a\ whenever | - a | < S; with the point a as centre describe a circle 7 of radius 7' < 8, r being so small that 7 lies wholly inside C Then in the space between 7 and G f z)\ \ {z - a) is analytic, and so, by \hardsectionref{5}{2} corollary 2, we have f z)dz\ [f z)dz c z - a Jy z - a where I and I denote integrals taken counter-clockwise along the curves G and 7 respectively. But, since 1 2 - a | < S on 7, we have r f(z) dz\ f f a)+iz- a)f (a) + v(z-a) J y z - a J y z - a where \ v\ < e; and so r m y j f dz f ( ) dz -\ vdz. J C Z-a \ lyZ -a - Jy Jy Now, if z be on 7, we may write z - a = re, where r is the radius of the circle 7, and consequently "2t ire dO jyZ - a Jo = I I de = 2771, y j, re'" Jo and dz= ire' dd = 0; also, by \hardsubsectionref{4}{6}{2}, vdz < e . 27rr. i % % 89 % r f(z)d2 \ 2 if( a) I = f €dz\ \ 2irre. J c z - a \ J y 1 ( r ( 5: 1 n.7: i /' Thus c z - a But the left-hand side is independent of e, and so it must be zero, since e is arbitrary; that is to say 2771 ] c z - a This remarkable result expresses the value of a function f z\ (which is analytic on and inside (7) at any point a within a contour C, in terms of an integral which depends only on the value of/( ) at points on the contour itself. Corollary. If f(z) is an analytic one-valued function of 2 in a ring-shaped region bounded by two curves C and C", and a is a point in the region, then ' 2ni J c -(i "tti c' Z - a where C is the outer of the ciu-ves and the integi-als are taken counter-clockwise. \Subsection{5}{2}{2}{The derivates of an analytic function $f(z)$.} The function/' (2'), which is the limit of f z + h)-f z) h as h tends to zero, is called the derivate of / z). We shall now shew that f' z) is itself an analytic function of z, and consequently itself possesses a derivate. For if be a contour surrounding the point a, and situated entirely within the region in which f(z) is analytic, we have f(a+h)-f(a) f (a) = hm --, h o n ,, 0 2TTih [J c z-a-h (j z-a ) = Km -I- \ /( ) dz h Q liri J c z-a) z- a~h) Itti J c z - af h 27ri J c z - ay z-a- h) Now, on C, f(z) is continuous and therefore bounded, and so is (z - a)~; while we can take | h less than the upper bound of U - a |. \ % % 90 % Therefore (z - a)- z - a - h) Then, if I be the length of C, h f f z)dz [chap. V is bounded; let its upper bound be K. TODO:missingpagenum? lim 0 unle.ss/(2) = or/' z) = Q. \addexamplecitation{Trinity, 1910.} \Section{5}{3}{Analytic functions represented by uniformly convergent series.} X) Let X fn (z) be a series such that (i) it converges uniformly along a =o contour C, (ii) / (z) is analytic throughout C and its interior. 00 Then 2 fni ) converges, and the sum* of the series is an analytic n = function throughout C and its interior. 00 For let a be any point inside C; on C, let S fn z) = (z). Then -. - dz = j- \ fn( )\ 27ri J c z - a 27ri j c Im=o ] z -a .o \ 2mlc 2-a \ ' 00 by* \hardsectionref{4}{7}. But this last series, by \hardsubsectionref{5}{2}{1}, is S fiid)', the series under n = consideration therefore converges at all points inside C; let its sum inside G (as well as on C) be called (z). Then the function is analytic if it ha,s a unique differential coefficient at all points inside G. But if a and a + h be inside G, (a + h)- (a) \ J f ( ) ( A 27rt J c z - a) z - a - h)' and hence, as in \hardsubsectionref{5}{2}{2}, lim W a- h) - a)] A~ ] exists and is equal to 7t-*0 * Since | 2 - a |~i is bounded when a is fixed and z is on C, the uniformity of the convergence of S / z)l[z - a) follows from that of 2 / [z). n=0 n=0 % % 92 % - : 7 - dz; and therefore (z) is analytic inside G. Further, by 27rt J c z- a) transforming the last integral in the same way as we transformed the first 00 ao one, we see that ' (a) = S / ' (a), so that 2 fn (a) may be ' differentiated n=0 n=0 term by term.' If a series of analytic functions converges only at points of a curve which is not closed nothing can be inferred as to the convergence of the derived series*. 'COS 7hJ7 Thus 2 ( - )" - 2 - converges uniformly for real values of x \hardsubsectionref{3}{3}{4}). But the derived H = l 'i sin ij series 2 ( - )""* converges non-uniformly near A' = (2m+1) tt, (m any integer); and H = l 'ii the derived series of this, viz. 2 ( - )"~ cos n.v, does not converge at all. (1=1 Corollary. By \hardsectionref{3}{7}, the sum of a power series is analytic inside its circle of con- vergence. \Subsection{5}{3}{1}{Analytic functions represented by integrals.} Let f t, z) satisfy the following conditions when t lies on a certain path of integration (a, h) and z is any point of a region >S' : (i) f and ~ are continuous functions of t. dz (ii) / is an analytic function of z. df * (in) The continuity of ~- qua function of z is uniform with respect to the variable t. rb . Then I f(t, z)dt is an analytic function of z. For, by \hardsectionref{4}{2}, it has the , !* f* dt\ t, z) . unique derivate - - - - dt. \Subsection{5}{3}{2}{Analytic functions represented by infinite integrals.} From \hardsubsectionref{4}{4}{4} (II) corollary, it follows that I f (t, z) dt is an analytic J a function of z at all points of a region >S' if (i) the integral converges, (ii) f t, z) is an analytic function of z when t is on the path of integration and z is on S, (iii) - : ' is a continuous function of both variables, (iv) - - - dt dz J a OZ converges uniformly throughout 8. For if these conditions are satisfied f t, z) dt has the unique derivate J a J a dz * This might have been anticipated as the main theorem of this section deals with uniformity of convergence over a two-dimensional region. % % 93 % A case of very great importance is afforded by the integral I e~' /(0 dt, Jo where /(t) is continuous and \ f t)\ < Ke' where K, r are independent of t; it is obvious from the conditions stated that the integral is an analytic function of z when R z) r, > r. [Condition (iv) is satisfied, by \hardsubsubsectionref{4}{4}{3}{1} (I), r since I fe"""''*' rf converges.] Jo \Section{5}{4}{Taylor's Theorem*TODO.} Consider a function f z), which is analytic in the neighbourhood of a point z = a. Let (7 be a circle with a as centre in the -plane, which does not have any singular point of the function f(z) on or inside it ; so that f(z) is analytic at all points on and inside C. Let z = a + h be any point inside the circle C. Then, by \hardsubsectionref{5}{2}{1}, we have Itti J c z-a-h -rriJc Xz-a iz-af ' ' ' z - ay- ' z - a)'' ' z - a - h)] f z) But when z is on C, the modulus of - - -j is continuous, and so, z - a - h by \hardsubsectionref{3}{6}{1} cor. (ii), will not exceed some finite number M. Therefore, by \hardsubsectionref{4}{6}{2}, 1 f f(z)dz.h''+ 27riJc z-ay'+' z-a-h) " 27r [rJ ' where R is the radius of the circle C, so that 'IttR is the length of the path of integration in the last integral, and R = \ z - a\ for points z on the cir- cumference of C. The right-hand side of the last inequality tends to zero as ?? - > oo . We have therefore /(a + /0=/(a) + / /'(a)-h|,/"(a)+...-f |/-'(a) + ..., which we can write f(z)=f a) + (z - a)/' (a) + ~ ~ff" a) + ... + ri /< ) (a) + .... This result is known as Taylors Theorem; and the proof given is due to Cauchy. It follows that the radius of convergence of a poiuer series is always * The formal expansion was first published by Dr (1715) in his Methodus Incrementonim. % % 94 % at least so large as only just to exclude from the interior of the circle of con- vergence the nearest singularity of the function represented by the series. And by \hardsectionref{5}{3} corollary, it follows that the radius of convergence is not larger than the number just specified. Hence the radius of convergence is just such as to exclude from the interior of the circle that singularity of the function which is nearest to a. At this stage we may introduce some terms which will be frequently used. If f(a) = 0, the function f(z) is said to have a zero at the point z = a. If at such a point f (a) is different from zero, the zero of f(a) is said to be simple; if, however,/' a),f" a), .../"*"'' (a) are all zero, so that the Taylor's expansion of f(z) at z = a begins with a term in (z - a)", then the function f(z) is said to have a zero of the nth. order at the point z = a. Example 1. Find the function / (s), which is analytic throughout the circle C and its interior, whose centre is at the origin and whose radius is unity, and has the value a - cos 6 . sin 6 a -2acos6 + l a -2acos0 + l (where a> 1 and 6 is the vectorial angle) at points on the circumference of C. [We have f z) dz /( )(0) = .f - - ' ' 2mjc z" n ! /"St 27nJo \ n\ n e~"i9d6 \ n \ f dz f d"" 1 "1 ~2ffjo a-e<9 ~2iriJcz''(a-z)~~\ \ ds a-zJ e- 'O.idd. -7- - ~;r ., (puttuig z = e*e) a- -2a cos + 1 ' ° ' Therefore by Maclaurin's Theorem*, )l=0 " or/(2) = (a-2)~' for all points within the circle. This example raises the interesting question, Will it still be convenient to define f(z) as (a-2)~ at points outside the circle ? This will be discussed in \hardsubsectionref{5}{5}{1}.] Example 2. Prove that the arithmetic mean of all values of 2"" 2 cv, for points z on the circumference of the circle \ z\ = l, is a, if Sc? " is analytic throughout the circle and its interior. / (") (0) [Let 2 v2''=/(2), so that a, =;- . Then, writing z = e', and calling C the circle K=0 " 277 jo 2" ~ 2ni j c 2"* ~ n\ """ -' z'i * The re8ult/ 2) =/(0) +2/' (0) + -/" (0) + ..., obtained By putting a = in Taylor's Theorem, is usually called Maclaurin's Theorem; it was discovered by Stirling (1717) and published by Maclaurin (1742) in his Fluxions. % % 95 % Example 3. Let $f(z) = z^{r}$; then $f(z+h)$ is an analytic function of $h$ when $\absval{h} < \absval{z}$ for all values of $r$; and so $(z + h)^{r} = z^{r} + rz^{r-1} h + \frac{ r (r-1) }{2} z^{r-2} h^{2} + \cdots, $ this series converging when $\absval{h} < \absval{z}$. This is the binomial theorem.\index{Binomial theorem} Example 4. Prove that if h is a positive constant, and (1 - 2zh- h?) ~ in expanded in the form \ + hP z) + h''P.2 z) + h P z) + (A), (where P (2) is easily seen to be a polynomial of degree n in z), then this series converges so long as z is in the interior of an ellipse whose foci are the points z = \ and 2= -1, and whose semi-major axis is (A + A"'). Let the series be first regarded as a function of A. It is a power series in A, and therefore converges so long as the point A lies within a circle in the /; -plane. The centre of this circle is the point A = 0, and its circumference will be such as to pass through that singularity of (1 - 2zh- h' )~ which is nearest to A = 0. But 1 - 22A -1- A2 = A - 2 + (22 - 1 )5>. /i \ 2 \ ( 2 \ 1 )ij, so the singularities of (1 - 22A-|-A-)~2 are the points h=z - z' - ) ' and h=z + z - ) . [These singularities are branch points (see \hardsectionref{5}{7}).] Thus the series (A) converges so long as | A | is less than both |2-(22-l) | and |2-f(22\ l) |. Draw an ellipse in the 2-plane passing through the point 2 and having its foci at +L Let a be its semi-major axis, and 6 the eccentric angle of 2 on it. Then 2 = a cos -f i (a - 1 ) sin 9, which gives 2 ± (22 - 1 )i = a + (a2 \ i) (cos + 1 sin 6), so i2±(22-l)i | = a + (a2\ i)4. Thus the series (A) converges so long as A is less than the smaller of the numbers a-|-(a2- 1) and a- a - 1)2, i.e. so long as A is less than a-(a2\ i)5. But A = a - (a2- 1) when a = |(A-f-A~i). Therefore the series (A) converges so long as 2 is within an ellipse whose foci are 1 and - 1, and whose semi-major axis is h h + h~ ). \Subsection{5}{4}{1}{Forms of the remainder in Taylor's series.} Let f(x) be a real function of a real variable; and let it have continuous differential coefficients of the first n orders when a x a + h. If O i l, we have fl (71-1 hm ) /i n -f\ n-\ It li. 7! (1 - ') "/'"" ( + *> = ifr /'"' ( + "'> - ''/' ( + "') Integrating this between the limits and 1, we have n-l hm /! hn /I \ f\ n-i f(a + h)=f a)+ -,f (a)+ ] ', f'Ha + th)dt. ni = im: Jo n-L)l Let Rn = J~iyi ] \ l - 0' - V'"* ( + ih) dt; and let j-j be a positive integer such that p n. % % 96 % Then Rn = r - f ' ( " y~" ' ( " O"" /'"* ( + th) dt. Let U, L be the upper and lower bounds of (1 - )' -p/("' (a + th). Then f ' X (1 - 0 "' dt<\ \ t)P-' . (1 - O' -P/"'' (a + th) dt<[ U(l- t)P-' dt. Jo Jo Jo Since (1 - t)' ~P / '" (a + th) is a continuous function it passes through all values between U and L, and hence we can find 6 such that 1, and [ I- ) -y <' ' (a + th) dt = -1(1- ey-Pf " (a + Oh). Jo Therefore R = .J \ y;(1 - f- /*"' (a + 6h). A" Writing p = n, we get Rn = - /"" (a + Oh), which is Lagrange s form for A" Me remainder; and writing j9 = 1, we get Rn = - \, y, (1 - )''-'/''" (a + A), which is Cauchys form for the remainder. Taking n = \ in this result, we get f a h)-f a) = hf a + 6h) \ i f(x) is continuous when a x a + h; this result is usually known as the First Mean Value Theorem (see also § 4-14). Darboux gave in 1876 Journal de Math. (3) ii. p. 291) a form for the remainder in Taylor's Series, which is applicable to complex variables and resembles the above form given by Lagrange for the case of real variables. \Section{5}{5}{The Process of Continuation.} Near every point P, Zq, in the neighbourhood of which a function f z) is analytic, we have seen that an expansion exists for the function as a sei'ies of ascending positive integral powers of z - Zq), the coefficients in which involve the successive derivates of the function at z . Now let A be the singularity of f(z) which is nearest to P. Then the circle within which this expansion is valid has P for centre and PA for radius. Suppose that we are merely given the values of a function at all points of the circumference of a circle slightly smaller than the circle of convergence and concentric with it together with the condition that the function is to be analytic throughout the interior of the larger circle. Then the preceding theorems enable us to find its value at all points within the smaller circle and to determine the coefficients in the Taylor series proceeding in powers of z - Zq. The question arises, Is it possible to define the function at points outside the circle in such a way that the function is analytic throughout a larger domain than the interior of the circle ?, % % 97 % In other words, given a potver series which converges and represents a function only at poiiits within a circle, to define hy means of it the values of the function at points outside the circle. For this purpose choose any point TODO within the circle, not on the line PA. We know the value of the function and all its derivates at Pj, from the series, and so we can form the Taylor series (for the same function) with Pi as origin, which will define a function analytic throughout some circle of centre Pj. Now this circle will extend as far as the singularity* which is nearest to Pi, which may or may not be A; but in either case, this- new circle will iisuall '! lie partly outside the old circle of convergence, and for jjoints in the region which is included in the new circle but not in the old circle, the new series may he used to define the values of the function, although the old series failed to do so. Similarly we can take any other point Po, in the region for which the values of the function are now known, and form the Taylor series with P as origin, which will in general enable us to define the function at other points, at which its values were not previously known; and so on. This process is called continuation . By means of it, starting from a representation of a function by any one power series we can find any number of other power series, which between them define the value of the function at all points of a domain, any point of which can be reached from P without passing through a singularity of the function; and the aggregate § of all the power series thus obtained constitutes the analytical expression of the function. It is important to know whether continuation by two different paths fBQ, PB'Q will give the same final power series; it will be seen that this is the case, if the function have no singularity inside the closed curve PBQB'P, in the following way : Let P be any point on PBQ, inside the circle C' with centre P; obtain the continuation of the function with Pi as origin, and let it converge inside a circle Ci; let P be any point inside both circles and also inside the curve PBQB'P; let S, Si, Si be the power series with P, Pi, Pi as origins; then|| *S'i = S'i' over a certain domain which will contain Pi, if Pi' be taken sufficiently near Pi; and hence Si will be the continuation of Si; for if Ti were the continuation of Si, we have Ti = Si over a domain containing Pj, and so \hardsubsectionref{3}{7}{3}) corresponding coefficients in i and Ti are the same. By carrying out such a process a sufficient number of times, we deform the path PBQ into the path PB'Q if no singular point is inside PBQB'P. The reader will convince himself by drawing a figure that the process can be carried out in a finite number of steps. * Of the function defined by the new sei'ies. + The word ' usually ' must be taken as referring to the cases which are likely to come under the reader's notice while studying the less advanced parts of the subject. X French, prolongement; German, Fortsetzung. § Such an aggregate of power series has been obtained for various functions by , by purely algebraical processes, Proc. London Math. Soc. xxxv. (1903), pp. 388-416. II Since each is equal to S. . 7 % % 98 % Example. The series 1, Z 22 3 a a a" a* represents the function /('-) = -- a - z only for points z within the circle | I = | a | . But any number of other power series exist, of the type 1 z-h z-hf z-hf a-b' a-hf" a-bf' a-bf '-' ' if b/a is not real and positive these converge at points inside a circle which is partly inside and partly outside | s | = | a j; these series represent this same function at points outside this circle. \Subsubsection{5}{5}{0}{1}{On functions to which the continuation-process cannot be applied.} It is not always possible to carry out the process of continuation. Take as an example the function /(2) defined by the power series which clearly converges in the interior of a circle whose radius is unity and w hose centre is at the origin. Now it is obvious that, as -1-0, /(2) +qc; the point +1 is therefore a singularity of/ (2). But /(2)=22+/( 2) and if z-- 0, f(z )- x and so /(s)- x, and hence the points for which z- = l are singularities oi f z); the point 2= - 1 is therefore also a singularity oif z). Similarly since we see that if 2 is such that 2* = 1, then z is a singularity of/ (2) ; and, in general, any root of any of the equations 22=1, 2* = 1, 28 = 1, 2l =l, ..., is a singularity of f z). But these points all lie on the circle | 2 | = 1; and in any arc of this circle, however small, there are an unlimited number of them. The attempt to carry out the process of continuation will therefore be frustrated by the existence of this imbroken front of singularities, beyond which it is impossible to pass. In such a case the function f(z) cannot be continued at all to points 2 situated outside the circle i 2 | = 1; such a function is called a lacuvary f miction, and the circle is said to be a limiting circle for the function. \Subsection{5}{5}{1}{The identity of two functions. .,} The two series 1 + 2 + ' + 2' + . . . and - 1 + ( - 2) - ( - 2y- + ( - 2) - (2 - 2 + ... do not both converge for any value, of z, and are distinct expansions. Nevertheless, we generally say that they represent the same function, on the strength of the fact that they can both be represented by the same rational 1 expression . % % 99 % This raises the question of the identity of two functions. When can two different expansions be said to represent the same function ? We might define a function (after Weierstrass), by means of the last article, as consisting of one power series together with all the other power series which can be derived from it by the process of continuation. Two different analytical expressions will then define the same function, if they represent power series derivable from each other by continuation. Since if a function is analytic (in the sense of Cauchy,\hardsubsectionref{5}{1}{2}) at and near a point it can be expanded into a Taylor's series, and since a convergent power series has a unique differential coefficient (§ 5'3), it follows that the definition of Weierstrass is really equivalent to that of Cauchy. It is important to observe that the limit of a combination of analytic functions can represent different analytic functions in different parts of the plane. This can be seen by considering the series 5(' + j)\!,('"i)(TT7.-r;i ) The sum of the first n + 1 terms of this series is 1 / 1\ 1 z \ zJ' I + z' ' The series therefore converges for all values of z (zero excepted) not on the circle j 2 | = 1. But, as w - > oo, | 2:'* | - > or j £" i - oo according as | j is less or greater than unity; hence we see that the sum to infinity of the series is z when \ z\ < 1, and - when | j > 1. This series therefore represents one function at points in the interior of the circle | j = 1, and an entirely different function at points outside the same circle. The reader will see from \hardsectionref{5}{3} that this result is connected with the non-uniformity of the convergence of the series near | j = 1. It has been shewn by Borel* that if a region C is taken and a set of points S such that points of the set S are arbitrarily near every point of 6', it may be possible to define a function which has a unique differential coefficient (i.e. is monogenic) at all points of C which do not belong to *S'; but the function is not analytic in C in the sense of Weierstrass. Such a function is .,, 00 w n exp(-expw*) f z)= 2 2 2; . . n=ip=oq=o z- p + qi)jn * Proc. Math. Congress, Cambridge (1912), i. pp. 137-liJ8. Leqons sur les fonctions mono- genes (1917). The functions are not monogenic strictly in the sense of \hardsectionref{5}{1} because, in the example quoted, in working out f z + h) -f(z)]jli, it must be su Dposed that R z + h) and I z + ]i) are not botli rational fractions. % % 100 % \Section{5}{6}{Laurent's Theorem.} A very important theorem was published in 1843 by Laurent*; it relates to expansions of functions to which Taylor's Theorem cannot be applied. Let G and C be two concentric circles of centre a, of which C is the inner; and let f(z) be a function which is analytic i* at all points on G and G' and throughout the annulus between G and C. Let a + A be any point in this ring-shaped space. Then we have \hardsubsectionref{5}{2}{1} corollary) ZTTt j cz - a - h liTi j c z - a - h where the integrals are supposed taken in the positive or counter-clockwise direction round the circles. This can be written We find, as in the proof of Taylor's Theorem, that f(z)dz.h- r f(z)dz(z-ar+ c(z-a)"+' z-a-h) Jc z-a-h)h +' tend to zero as n- cc; and thus we have /(a + h) = cio + a h + ajr + ... + -~ + j + ..., where + a = - / K, and 6 = 5- . I z - ay'-'f(z) dz. This result is Laurent's Theorem; changing the notation, it can be expressed in the following form: If f(z) be analytic on the concentric circles G and G' of centre a, and throughout the annulus between them, then at any point z of the annidus f(z) can he expanded in the form $$ TODO $$ where a = . j a.nd b = . [ (t - aT' f(t) dt. An important case of Laurent's Theorem arises when there is only one singularity within the inner circle G', namely at the centre a. In this case the circle G' can be taken as small as we please, and so Laurent's expansion is valid for all points in the interior of the circle C, except the centre a. * Comptes Rendus, xvii. (1843), pp. 348-349. t See \hardsectionref{5}{2} corollary 2, footnote. X We cannot write a- = fW (a)ln ! as in Taylor's Theorem since /(2) is not necessarily analytic inside C. % % 101 % Example 1. Prove that e ' = J x)+zJ x) + z' J x) + ...-irz-J,, x)- ... 1 /"St where Jn (*) = s~ I ' ~ " ) TT J [For the function of z under consideration is analytic in any domain which does not include the point z=Q; and so by Laurent's Theorem, g2V z) = a,, + axZ + aoz'-+... + - ->r + ..., where n = s - I - n and G = i, - . I e z dz, and where 6' and C" are any circles with the origin as centre. Taking C to be the circle of radius unity, and writing = e', we have 1 Z"' "" 1 /' -'f -=--, / e:' sin .e- ' irf = jr- / COS (n6 - X H\ n 0) dB, 27ri. Jo 27r y I sin (/i - i'sin ) tZ vanishes, as may be seen by writing in-cf) for 9. ThiLS suice a = J (A'), and \& = ( -)", since the function expanded is unaltered if -z l)e written for z, so that 6 = ( - )" Ai(.:i ), and the proof is complete.] Example 2. Shew that, in the annulus defined by|a|<|3|i-l) /a\' " ' = ,!o 2 .ll l + n)l [b) The function is one- valued and analytic in the annulus (see \hardsectionref{5}{7}), for the branch-points 0, a neutralise each other, and so, by Laurent's Theorem, if C denote the circle \ z\=r, where | o ' < / < | 6 |, the coefficient of 2" in the required expansion is 1 f dz ( bz 1 27riJ c '- \ \ {z-a) b-z) ' Putting z = re', this becomes 1 [-' .,, 1.3... (2i-l) ?*<*" 1.3... (2i-l)a' -'' the series being absolutely convergent and uniformly convergent with regard to 6. The only terms which give integrals different from zero are those for which k = l + n. So the coefficient of z" is (2 -1) 1 . 3 ...(2 + 2/i-l) a \ Sn 1 r 'T * 1 2ir J 1=0 2Kl\ 2* + ™.(/+m) ! ¥*" b" ' Similarly it can be shewn that the coefficient of - is S a'\ % % 102 % Example 3. Shew that Z Z'' 1 rsT where = '" '' "> '''" cos ( - v) sin (9 - ?i(9 o?, ZTT / 1 /"-" and i = 5- / e'" + ")'=°' cos (i'-?Osi" - '<9 < <9- ZTry \Subsection{5}{6}{1}{T ie nature of the singularities of one-valued functions.} Consider first a function f(z) which is analytic throughout a closed region 8, except at a single point a inside the region. Let it be possible to define a function z) such that (i) z) is analj tic throughout S, (ii) when a, /( ) = < ( ) + - 4-A . + ...+ " z - a (z - a)- z - ay- Then f(z) is said to have a 'pole of order n at a'; and the terms h, -T + ... + -. are called the principal part of f(z) near a. z - a. (z - a) (z - aY r r j \ / By the definition of a singularity \hardsubsectionref{5}{1}{2}) a pole is a singularity. If n = 1, the singularity is called a simple pole. Any singularity of a one-valued function other than a pole is called an essential singularity. If the essential singularity, a, is isolated (i.e. if a region, of which a is an interior point, can be found containing no singularities other than a), then a Laurent expansion can be found, in ascending and descending powers of a valid when dk>\ z - a h, where A depends on the other singularities of the function, and 8 is arbitrarily small. Hence the ' principal part ' of a function near an isolated essential singularity consists of an infinite series. It should be noted that a pole is, by definition, an isolated singularity, so that all singularities which are not isolated (e.g. the limiting point of a sequence of poles) are essential singularities. There does not exist, in general, an expansion of a function valid near a non-isolated singularity in the way that Laurent's expansion is valid near an isolated singularity. Corollary. If f(z) has a pole of order n at a, and z) = z - aYf z) z a), i\ r a)= lim z-a)' f z), then y\ r z) is analytic at a. Example 1. A function is not bounded near an isolated essential singularity. [Prove that if the function were bounded near z=a, the coefficients of negative powers of 2 - a would all vanish.] % % 103 % c z\ Example 2. Find the singularities of the function e ~ 7 e - 1 . At 2 = 0, the numerator is analytic, and the denominator has a simple zero. Hence the function has a simple pole at 2 = 0. Similarly there is a simple pole at each of the points mria ( = + 1, +2, +3, ...); the denominator is analytic and does not vanish for other values of z. A.t z = a, the numerator has an isolated singularity, so Laurent's Theorem is applicable, and the coefficients in the Laurent expansion may be obtained from the quotient c c z- a 2 I (z - aV Ml+'--" + ...)-l which gives an expansion involving all positive and negative powers of z - a). So there is an essential singularity at 2 = a. Example 3. Shew that the function defined by the series I wg"- (l +)t- )"-l =1 (2 -l) 2 -(l+/i-l) has simple poles at the points 2 = (1 + ~i)e- ''' / ( =0, 1, 2, ... n - \; ?i = l, 2, 3, ...). y \addexamplecitation{Math. Trip. 1899.} \Subsection{5}{6}{2}{The 'point at infinity.'} The behaviour of a function /C ') as | ' - oo can be treated in a similar way to its behaviour as z tends to a finite limit. If we write z = -,, so that large values of z are represented by small values of z' in the '-plane, there is a one-one correspondence between z and z, provided that neither is zero; and to make the correspondence complete it is sometimes convenient to say that when z is the origin, z is the * point at infinity.' But the reader must be careful to observe that this is not a definite point, and any proposition about it is really a proposition concerning the point / = 0. Let/(2 ) = 4> z'). Then < z') is not defined at z = 0, but its behaviour near z = is determined by its Taylor (or Laurent) expansion in powers of z \ and we define < (0) as lim (/) if that limit exists. For instance the function (/) may have a zero of order m at the point 2' =; in this case the Taylor expansion of ( ') will be of the form and so the expansion of f(z) valid for sufficiently large values of | .0 j will be of the form In this case,/(0) is said to have a zero of order m at ' infinity.' % % 104 % Again, the function (f)(2') may have a pole of order m at the point z' = 0; in this case and so, for sufficiently large values of \ z\, f(z) can be expanded in the form N P f z) = Az"' + Bz'''-' + Cz'"-- +...+Lz + M+- + - + .... In this case,/(2 ) is said to have a pole of order m at ' infinity. ' Similarly f(z) is said to have an essential singularity at infinity, if z) has an essential singularity at the point / = 0. Thus the function e' has an essential singularity at infinity, since the function e~' or 1 \ 1\ 1 has an essential singularity at z = 0. Example. Discuss the function represented by the series 2 -, :; 5-5, ( >1). Z 1 The function represented by this series has singularities at 2=- and 2= - i n=\, 2, 3, ...), since at each of these points the denominator of one of the terms in the series is zero. These singularities are on the imaginary axis, and have 3 = as a limiting point; so no Taylor or Laurent expansion can be foi med for the function valid throughout any region of which the origin is an interior point. For values of z, other than these singularities, the series converges absolutely, since the limit of the ratio of the (?i + l)th term to the ?ith is lim (?i+ l)~i a~ = 0. The function is an even function of z (i.e. is unchanged if the sign of z be changed), tends to zero as I 2 I - Qc, and is analytic on and outside a circle C of radius greater than unity and centre at the origin. So, for points outside this circle, it can be expanded in the form h + j + b+ where, by Laurent's Theorem, TODO This double .series converges absolutely when | 2 | > 1, and if it be rearranged in powers of 2 it converges uniformly. Since the coefficient of 2 ~ Ms 2 -; and the only term which furnishes a non- H=o n ! zero integral is the term in z~, we have (\ )fc-ia-2A- dz b'k 1 - . Itti J c =o, fo n\ a2* % % 105 % Therefore, when | 2 | > 1, the function can be expanded in the form ill e"' e * e The function has a zero of the second order at infinity, since the expansion begins with a term in z~' . \Subsection{5}{6}{3}{Liouville's Theorem*.} Let f(z) he analytic for all values of z and let \ f z)\ < K for all values of z, where K is a constant (so that \ f(z) is bounded as | r | - > x ). Then f(z) is a constant. Let z, z' be any two points and let C be a contour such that z, z are inside it. Then, by \hardsubsectionref{5}{2}{1}, take C to be a circle whose centre is z and whose radius is p 2 | / - s |; on C write \ \ = z pe' \ since jf - /| 2P when is on C it follows from § 4-62 that = 2\ z -z\ Kp-K Make p- cc keeping z and z' fixed; then it is obvious that/(/) -f(z) = 0; that is to say, f(z) is constant. As will 1)0 seen in the next article, and again frequently in the latter half of this volume (Chapters xx, xxi and xxii), Liouville's theorem furnishes short and convenient proofs of some of the most important results in Analysis. \Subsection{5}{6}{4}{Functions with no essential singularities.} We shall now shew that the only one-valued functions luhich have no singidarities, except poles, at any jjoint (including oo ) are rational functions. For let f(z) be such a function; let its singularities in the finite part of the plane be at the points Cj, c-j, ... Ck'. and let the principal part \hardsubsectionref{5}{6}{1}) of its expansion at the pole Cr be + 1 ::; + . . . + - 7 - Z - Cr (Z - Crf '" (Z - CyJ Let the principal part of its expansion at the pole at infinity be a- z + a-iZ + ... + anZ \ if there is not a pole at infinity, then all the coefficients in this expansion will be zero. * This theorem, which is really due to Cauchy, Comptes Rendus, xix. (1844), pp. 1377, 1378, was given this name by Borchardt, Journal fvr Math, lxxxviii. (1880), pp. 277-310, who heard it in Liouville's lectures in 1847. % % 106 % Now the function has clearly no singularities at the points c, c.j, ... Ck, or at infinity; it is therefore analytic everywhere and is bounded as \ z - >cc, and so, by Liou\ dlle's Theorem, is a constant; that is, /(.) = + .. + a..= + . . . +,,.. + I l i + ( . + + (i J where C is constant; f(z) is therefore a rational function, and the theorem is established. It is evident from Liouville's theorem (combined with \hardsubsectionref{3}{6}{1} corollary (ii)) that a function which is analytic everywhere (including oc ) is merely a constant. Functions which are analytic everywhere except at oc are of considerable importance; they are known as integral functions*. Examples of such functions are e, sin z, e . From \hardsectionref{5}{4} it is apparent that there is no finite radius of convergence of a Taylor's series which represents an integral function; and from the result of this section it is evident that all integral functions (except mere polynomials) have essential singularities at oo . \Section{5}{7}{Many-valued functions.} In all the previous work, the functions under consideration have had a unique value (or limit) corresponding to each value (other than singularities) of . But functions may be defined which have more than one value for each value of z; thus if 2- = r (cos 6 + i sin 6), the function 2- has the two values r* (cos 1(9 + t sin (9), r jcos \ 0 + lir) + i sin \ (6 + 27r)|; and the function arc tan x (x real) has an unlimited number of values, viz. Arc tan x + mr, where - tt < Arc tan x <- 'tt and n is any integer; further examples of many- valued functions are log z, z, sin z~). Either of the two functions which z represents is. however, analytic except at = 0, and we can apply to them the theorems of this chapter; and the two functions are called ' branches of the many-valued function z'-.' There will be certain points in general at which two or more branches coincide or at which one branch has an infinite limit; these points are called ' branch-points.' Thus z- has a branch-point at; and, if we consider the change in z as z describes a circle counter-clockwise round 0, we see that 6 * Vxench, fonction entilre; Germa,n, game Funktion. % % 107 % increases by 27r, r remains unchanged, and either branch of the function passes over into the other branch. This will be found to be a general characteristic of branch-points. It is not the purpose of this book to give a full discussion of the properties of many-valued functions, as we shall always have to consider particular branches of functions in regions not containing branch- points, so that there will be comparatively little difficulty in seeing whether or not Cauchy's Theorem may be applied. Thus we cannot apply Cauchy's Theorem to such a function as z when the path of iutegi-ation is a circle surrounding tlie origin; but it is permissible to apply it to one of the branches of z when the path of integration is like that shewn in § 6 '24, for through- out the contour and its interior the function has a single definite value. Example. Prove that if the different values of a, corresponding to a given value of z, are represented on an Argand diagram, the representative points will be the vertices of an equiangular polygon inscribed in an equiangular spiral, the angle of the spiral being independent of a. \addexamplecitation{Math. Trip. 1899.} The idea of the different braiickes of a function helps us to understand such a paradox as the following. Consider the function y = ' i for which ~=x log x). AVhen x is negative and real, is not real. But if x is negative and of the form P (where p and q are positive or negative integers), y is real. Z'j -f- 1 If therefore we draw the real c\ irve we have for negative values of ./ a set of conjugate points, one point corresponding to each rational value of x with an odd denominator; and then we might think of proceeding to form the tangent as the limit of the chord, just as if the curve were continuous; and thus --, when derived from the inclination of the tangent to the axis of x, would appear to be real. The question thus arises, Why does the ordinary process of differentiation give a non-real value for - ? The explanation is, that these conjugate points do not all arise from the same branch of the function y = x' . We have in fact y - 5 where k is any integer. To each value of k corresponds one branch of the function y. Now in order to get a real value of y when x is negative, we have to choose a suitable value for k : and this value of k varies as we go from one conjugate point to an adjacent one. So the conjugate points do not represent values of y arising from the same branch of the function y=x, and consequently we cannot expect the value of - when evaluated for a definite branch to be given by the tangent of the inclination to the axis of x of the line joining two arbitrarily close members of the series of conjugate points. % % 108 % REFERENCES. , Cours dJ Analyse, ii. (Paris, 1911), Chs. xiv and xvi. , et son prolongement analytique (Scientia, 1901). , (Paris, 1905). , Covrs d' Analyse Infinitesimale, i. (Paris and Louvain, 1914), Ch. X. , Lecons stir les Fonctions Entieres (Paris, 1900). , Complex Integration and Cauchy's Theorem (Camb. Math. Tracts, no. 15, 1914). Miscellaneous Examples. 1. Obtain the expansion /(.)=/( ) + 2 1 -/ (- + -3-3-r/ ( -2-j + -25751- ' \ \ r-] ' and determine the circumstances and range of its vaUdity. 2. Obtain, under suitable circumstances, the expansion + .... (Corey, Ann. of Math. (2), i. (1900), p. 77.) 3. Shew that for the series 1 )i=o -r~ the region of convergence consists of two distinct areas, namely outside and inside a circle of radius unity, and that in each of these the series represents one function and represents it completely. (Weierstrass, ., 1880, p. 731; , 11. (1895), p. 227.) 4. Shew that the function 2 s"' tends to infinity as 2- -exp i-niplm !) along the radius through the point; where m is any integer and p takes the values 0, 1, 2, ..'. (ni I - 1). Deduce that the function cannot be continued beyond the unit circle. (. BOhm. Acad., 1885-6, pp. 571-582.) 5. Shew that, if z-- 1 is not a positive real number, then -". .s:)"<'-")- /:'=-('-'- '-**- \addexamplecitation{Jacobi and Scheibner.} % % 109 % 6. Shew that, if s - 1 is not a positive real number, then ,,,, m, m(m+ ) to (m + 1) ... (to + ?i - 1) (1-2) - =l+-2 + - 2 "" +-+ -1 . n : J \addexamplecitation{Jacobi and Scheibner.} 7. Shew that, if z and 1-2 are not negative real numbers, then Jo m+l [ m + 3 (m+3) ... (m + 27i- 1) J + "') (m+l)(w + 3)...(m + 27i-l)Jo ' - \addexamplecitation{Jacobi and Scheibner.} 8. If, in the expansion of (a + i2 + a22 )'" by the multinomial theorem, the remainder after n terms be denoted by R (2), so that (a4- i2 + a2S-)'" = o + - i2 + -'-l2-"" + --- + n-iS"'"' + (s), shew that /4(.) = ( + a.. + ..r j, - (a + .M+a, r " ' 9. If (ao + i2 + a22 )""'"' /'(ao + i< + 2 -)'"o?< y \addexamplecitation{Scheibner.} be expanded in ascending powers of 2 in the form Jl2 + .4222+..., shew that the remainder after n-\ terms is (ao + ai2 + a22 )~'"" I (ao + a, + a2 )'" o .i-(2?'i + ? + 1) 2 -i ""'c C \addexamplecitation{Scheibner*.} 10. Shew that the series where X (2)= - 1 +2- f, + I",- - + (- )", ' and where (2) is analytic near 2 = 0, is convergent near the point 2 = ; and shew that if the sum of the series be denoted by/(2), then/(2) satisfies the differential equation /'(--)=/( )-0(4 ( (5), v. (1896), p. 27.) 11. Shew that the arithmetic mean of the squares of the moduli of all the values of the series "2 0, on a circle |2| = r, situated within its circle of convergence, is equal i=0 to the sum of the squares of the moduli of the separate terms. (Gutzmer, Math. Ann. xxxii. (1888), pp. .596-600.) * The results of examples 5, 6 and 7 are special cases of formulae contained in Jacobi's dissertation (Berlin, 182.5) published in bis Ges. Werke, in. (1884), pp. 1-44. Jacobi's formulae were generalised by Scheibner, , xlv. (1893), pp. 432-443, % % 110 % 12. Shew that the series 2 e-2(am) -m-l m = l converges when | 2 | < 1; and that, when a > 0, the function which it represents can also be represented when | s ! < 1 by the integral /ay f" e-"" dx n-J Jo ex\ 2 x ' and that it has no singularities except at the point z=l. (Lerch, Monatshefte fiir Math, und Phys. viii.) 13. Shew that the series 2 \ \,, 2 r z\ zj \ z + z )-V- 2 ( i\ 2 \ 2y'zi)(2v + iv'zif' iv-2v'z- i) 2v + 2v'z-' xf]' in which the summation extends over all integral values of v, v', except the combination (i' = 0, v' = 0), converges absolutely for all values of z except purely imaginary values; and that its sum is + 1 or - 1, according as the real part of z is positive or negative. (Weierstrass, , 1880, p. 735.) 14. Shew that sin \ u ( + -)[ can be expanded in a series of the type a, + aiZ + aoy-+...+- + j, + ..., in which the coefficients, both of s" and of z~'\ are 1 / 2t -- I sm (2m cos ) cos Ji o? . 2ir J ) n=i n-z' + a- shew that/ (2) is finite and continuous for all real values of z, but cannot be expanded as a Maclaurin's series in ascending powers of z; and explain this apparent anomaly. [For other cases of failure of Maclaurin's theorem, see a posthumous memoir by . des Set. Math. (2), xiv. (1890), pp. 145-599; Lerch, Journal fiir Math. cm. (1888), pp. 126-138; Pringsheim, Math. Ann. XLii. (1893), pp. 153-184; and , , vi. (1876), p. 235.] 16. If f(z) be a continuous one- valued function of z throughout a two-dimensional region, and if f z)dz = h for all closed contours C lying inside the region, then f(z) is an analytic function of z throughout the interior of the region. [Let a be any point of the region and let F z)=\'f z)dz J a It follows from the data that F(z) has the unique derivate f z). Hence F z) is analytic \hardsectionref{5}{1}) and so \hardsubsectionref{5}{2}{2}) its derivate /(z) is also analytic. This important converse of Cauchy's theorem is due to Morera, . 1st. Lomhardo Milano), xxil. (1889), p. 191.] iowaguy/starter-academic @inproceedings{Pacheco_Hippel_Weintraub_Goldwasser_Nita-Rotaru_2022, author = {Pacheco, Hippel, , , , Cristina}, booktitle = {IEEE Security & Privacy 2022}, pages = {18}, title = {Automated Attack Synthesis by Extracting Finite State Machines from Protocol Specification Documents}, year = {2022} } \documentclass{article} \usepackage[utf8]{inputenc} \usepackage{ctex} \usepackage{amsmath} %% This is package for typesetting derivations. \title{test for \LaTeX\ document} \author{thwzjx} \date{\today} \begin{document} \maketitle \tableofcontents \section{chap1} \subsection{Introduction for Real Analysis} \end{document}serg-ios/pub-loc \chapter{Introducción} En este capítulo se introducirá el proyecto. Hablaremos sobre el contexto y la motivación por la cual se decidió llevarlo a cabo. Se hará un resumen de los objetivos que se esperaban alcanzar y las soluciones propuestas para alcanzarlos. Y también se llevará a cabo en esta sección un breve resumen de los siguientes apartados de la memoria. %%% \section{Motivación} Estamos en plena revolución digital, durante la cual hemos presenciado un desarrollo increíble de los dispositivos móviles. Desde los primeros que sólo permitían llamar, hasta los que tenemos hoy en día, que nos permiten consultar el correo, subir fotos a las redes sociales, decirnos como llegar a los sitios, etcétera. Pero lo más importante es la difusión que han tenido estos aparatos, en Mayo de 2017 se registraron más de 5.000 millones de líneas activas y no sería de extrañar que en unos años hayan muchas más líneas que personas en el mundo. Debido a su facilidad de uso y a su gran utilidad los sistemas operativos más utilizados son \textit{Android} e \textit{iOS}. \begin{figure}[t] \begin{center} \includegraphics[scale=0.6]{figures/ios_android.jpg} \caption{\textit{Android} vs \textit{iOS} en diferentes países.\label{fig:and_vs_ios}} \end{center} \end{figure} Como vemos en la figura~\ref{fig:and_vs_ios}, \textit{Android} tiene una mayor aceptación entre el público, sobre todo porque estos móviles tienen precios más asequibles que los que tienen \textit{iOS}. \textit{Android} es utilizado por diversos fabricantes de teléfonos móviles, por esto es que hay mucha competencia y los precios bajan, pero la parte mala de esto es que tenemos un amplio abanico de \textit{Hardware} con diferentes requisitos, complicando la tarea al programador. Los terminales de \textit{Apple} son más caros pero su \textit{Hardware} está especialmente diseñado y optimizado para funcionar de la mano con su propio sistema operativo. Los terminales móviles poseen una gran cantidad de sensores, entre ellos se encuentra un receptor \textit{GPS}, el cual se puede utilizar para obtener nuestra posición en cualquier punto del globo. Pero esta tecnología tiene un punto débil: los espacios interiores, ya que en ellos la señal no penetra adecuadamente y no puede darnos nuestra ubicación con exactitud. Aquí aparece una nueva necesidad, a la que se ha intentado dar solución en los últimos años de diferentes maneras: con \textit{WiFi}, \textit{Bluetooth}, y otras tecnologías inalámbricas. Pero todavía hay muchas barreras, como la obligación de calibrar el entorno previamente o la necesidad de infraestructura auxiliar como transmisores \textit{Bluetooth} o \textit{WiFi} auxiliares. En nuestro caso, utilizaremos esta tecnología para la obtención de publicidad basada en la ubicación del cliente dentro de centros comerciales. Y los datos sobre la posición del usuario serán proporcionados por la plataforma \textit{Situm}, que nos da la localización de un usuario en un entorno previamente calibrado con una precisión bastante buena. Gracias a esta aplicación, las tiendas podrán emitir publicidad por un nuevo canal y los consumidores podrán disfrutar de ofertas, descuentos y estar al tanto de productos de su interés que se encuentren en su entorno. \section{Objetivos} El objetivo principal de este proyecto es crear una red publicitaria que se distribuirá entre los potenciales clientes según su ubicación en tiempo real. Se utilizarán técnicas de posicionamiento en interiores para solventar las limitaciones de cobertura propias del \textit{GPS}, pero sin dejarlo de lado, de tal manera que la aplicación podrá utilizarse en espacios abiertos y cerrados. Los objetivos principales serán los siguientes: \begin{itemize} \item Autenticación de los usuarios y asignación por roles. \item Publicación de ofertas. \item Disfrute de dichas ofertas por parte de los clientes. \end{itemize} %% \section{Propuesta} Proponemos desarrollar una plataforma de publicidad geolocalizada divida en tres componentes: \begin{enumerate} \item \textbf{Sistema de localización de interiores.} Se utilizarán los servicios ofrecidos por la plataforma \textit{Situm}. \item \textbf{Sistema de almacenamiento en la nube.} Para almacenar toda la información relativa a usuarios, ofertas, tiendas, edificios, roles, permisos e imágenes. Nos hemos decantado por la \textit{API} de \textit{Firebase Google}, que nos ofrece gran cantidad de servicios en la nube, aunque hay otros muchos disponibles. Los servicios de \textit{Firebase} que usaremos en este proyecto serán: base de datos no relacional, \textit{API REST}, \textit{Cloud Functions} y autenticación. \item\textbf{ Aplicación móvil \textit{iOS}.} Será la parte de cliente, que se comunicará con los servidores de \textit{Situm} y con \textit{Firebase} a través de su \textit{API REST}. La interfaz y las acciones que podrá llevar a cabo el usuario, dependerán de su rol. \end{enumerate} \section{Estructura de la memoria} En los capítulos siguientes, se profundizará en las múltiples fases del desarrollo de este proyecto. A continuación, un breve resumen del contenido de estos: \begin{itemize} \item \textbf{Estado del arte.} Aplicaciones de la localización en interiores en la actualidad. \item \textbf{Fundamentos teóricos.} Bases teóricas sobre las que se asienta la tecnología de posicionamiento en interiores. \item \textbf{Fundamentos tecnológicos.} Este capítulo explicará las tecnologías empleadas para el desarrollo de este proyecto en concreto. \item \textbf{Análisis.} Este capítulo tratará el análisis de requisitos, actores y casos de uso. \item \textbf{Metodología.} Aquí se explicará la elección de la metodología de desarrollo y su adaptación a este proyecto concreto. \item \textbf{Gestión de proyecto.} En este capítulo se hablará sobre la planificación de las tareas, estimaciones de costes y valoración de riesgos. \item \textbf{Arquitectura.} Se expondrá la arquitectura del proyecto, como se estructuraron los diferentes componentes. \item \textbf{Implementación.} Se tratarán aquí los detalles de la implementación, a partir de la arquitectura comentada en el capítulo anterior. \item \textbf{Pruebas.} Se hablará sobre las pruebas realizadas a lo largo de la realización del proyecto y al final del mismo. \item \textbf{Conclusiones.} Como cierre de la memoria, hablaremos sobre las conclusiones, lecciones aprendidas a lo largo de todo el desarrollo y del trabajo que nos queda por delante en caso de continuar con el proyecto. \item{}\textbf{Instalación y uso.} En este apéndice, se guiará al usuario a través del proceso de instalación de la aplicación y se le indicará como utilizar las funcionalidades básicas. \end{itemize} \begin{exercise} \topic{grupos!cíclicos} Se $G$ é um grupo de ordem $n$, $G$ é cíclico sse $G$ tem um elemento de ordem $n$. \end{exercise} \begin{exercise} \topic{grupos!cíclicos} Todo subgrupo cíclico é abeliano. \end{exercise} \begin{exercise} \topic{grupos!cíclicos} Seja $G$ grupo e sejam $a,b \in G$. Prove que se $a$ é uma potência de $b$, isto é, $a = b^k$, então $\langle a \rangle \subseteq \langle b \rangle$. \end{exercise} \begin{exercise} \topic{grupos} Seja $G$ grupo e $\emptyset \not = H \subseteq G$. Prove que $$ H \subgroup G \iff (\forall a,b \in H)[ab^{-1} \in H] $$ \end{exercise} \begin{exercise} \topic{grupos} Seja $G$ grupo e $H \subgroup G$. Defina: $$ a \sim b \iffdf ab^{-1} \in H $$ \begin{enumerate}[(i)] \item Prove que $\sim$ é uma relação de equivalência. \item Prove que para todo $a,b \in G$ \begin{enumerate}[(a)] \item Se $a \in H$ e $b \in H$, então $a \sim b$. \item Se $a \in H$ e $b \not \in H$, então $a \not \sim b$. \end{enumerate} \end{enumerate} \end{exercise} 1-10 \documentclass[a4paper,12pt]{article} %this enables IPA characters for linguistics %\usepackage{tipa} %this changes the way citations look and how they are organized in the bibliography %\usepackage[authoryear, round]{natbib} %\bibliographystyle{apa-good-ampersand} \usepackage[numbers]{natbib} \bibliographystyle{unsrtnat} %enables the use of images \usepackage{graphicx} %enables hyperlinks \usepackage[colorlinks=false]{hyperref} %enables some special characters, needed for writing Scandinavian \usepackage[utf8]{inputenc} %fixes the margin \usepackage[margin=2.5cm]{geometry} %removes page number which PDF readers have anyway. \pagestyle{empty} %removes ligatures \usepackage{microtype} \DisableLigatures{encoding = *, family = * } %for making tables stay in place, use [H] \usepackage{float} \begin{document} \noindent Submitted to \textit{Open Differential Psychology} March 24th, 2014\\ Published in \textit{Open Differential Psychology} DATE, YEAR\\ \\ \begin{center} \textbf{{\Huge The global hereditarian hypothesis and the National Longitudinal Survey of Freshman}} \end{center} \noindent\Large \footnote{Corresponding author: }\\ \noindent\Large \\ \\ \normalsize\textbf{Abstract}\\ We discuss the global hereditarian hypothesis of race differences in \textit{g} and test it on data from the NLSF. We find that migrants country of origin's IQ predicts GPA and SAT/ACT. \noindent\textbf{Keywords}: National IQs, race differences, country of origin, NLSF \section{Introduction} In terms of race differences in \textit{g} no other pair of races is as well studied as the 1.1 \textit{d} IQ gap between African Americans and European Americans\cite{roth2001ethnic}. For this reason, most discussion about the causes of race differences has centered on this gap\cite{rushton2005thirty}. has advanced the hypothesis that the gap is likely due to involve genetic factors, and this has been called "the hereditarian hypothesis"\cite{jensen1969much,jenseneducability,jensen1998g,rushton2005thirty}. Since the publication of and 's 2002 book \textit{IQ and the wealth of nations} and follow-up books the discussion of racial differences has moved to the global perspective\cite{lynn2002iq,lynn2006iq,lynn2006race,lynn2008global,lynn2012intelligence}. While there was initially much hostility towards the data, even researchers previously critical of the idea has themselves done research on the data\cite{hunt2006sorry,hunt2012makes} and others have called it a "productive research paradigm"\cite{rindermann2013intelligence}. Lynn and Vanhanen have also advanced the hypothesis that not just the African American European American \textit{IQ} gap, but more generally the gaps between different races living in various countries are due in part to genetic factors. We call this "the global hereditarian hypothesis". Much research since 2002 has centered on the country level correlations between national IQs, scholastic achievement scores, economic, health, governmental, biological, environmental variables and theories about their origin (see overview of findings in \cite{lynn2012intelligence}). We think it is time to approach the topic from another angle, that of immigrant populations in other countries. The idea is that if the global hereditarian hypothesis is true to a significant degree, for instance 50\% of the differences in national IQs are due to genetic factors, then persons who travel to other countries should be similar in their IQ to their home country, everything else equal (e.g. no selective migration and IQ-environment interaction effects). Since IQ is a known predictor and cause of many social and economic variables at the personal level, it should also be a predictor at the group level inside host countries. We call this hypothesis "the spatial transferability hypothesis". Many countries now have large and growing immigrant population which make it possible to test this hypothesis. Other researchers have already found evidence for the hypothesis. Jones and Schneider have shown that one can predict wages among immigrant groups in the U. S. by their country of origin\cite{jones2010iq} and Vinogradov and Kolvereid have shown that one can predict self-employment in Norway among immigrant groups\cite{vinogradov2010home}. One of us (JF) has previously shown that national IQs highly correlate with GMAT (Graduate Management Admission Test), GRE (Graduate Record Examinations), and TOEFL (Test of English as a Foreign Language) scores.\cite{JFuerst1}. As these tests have often been shown to be predictive of student performance irrespective of national citizenship\cite{talento-miller2009validity}, it is implied that national IQs predict, to some extent, migrant academic performance. JF further showed that migrant PISA scores are predictable from their country of origin's IQ and PISA score and not from the destination country\cite{JFuerst2}. Results are summarized in Table \ref{JFresults}. \begin{table}[H] \centering \begin{tabular}{|l|l|} \hline \textbf{Variable (N)} & \textbf{Correlation with national IQ} \\ \hline GMAT (143) & .720 \\ \hline TOEFL (157) & .625 \\ \hline GRE-50N (173) & .740 \\ \hline GRE-M (144) & .764 \\ \hline GRE-R (144) & .552 \\ \hline GRE-T (144) & .765 \\ \hline PISA (60) & .589 \\ \hline \end{tabular} \caption{Correlations of national IQs with scholastic tests and English abilities.}\label{JFresults} \end{table} The other of us (EOWK) previously tested the hypothesis with data from Denmark and Norway\cite{kirkegaard2013DK,kirkegaard2014DK,kirkegaard2014NO}. In the first study, EOWK found a military study from 2005 which reported raw score data from testing of immigrants in 2003. EOWK then found information about the immigrant population's composition by country of origin from the official statistics agency. Then, EOWK used Lynn and Vanhanen's national IQs with the composition data to estimate the mean score of the immigrant population. The estimate came very close to the observed score from the army study, 86.3 was found while 86.7 was estimated -- a discrepancy of a mere .4 IQ. In the second and third studies EOWK found data for criminality, fertility and country of origin from Denmark and criminality, employment rate and citizenship in Norway. EOWK then gathered information about the countries of origin in an attempt to predict these rates: IQs, GDP, Islam belief, height, murder rate, fertility. The data from Denmark and Norway were highly consistent as shown in Table \ref{kirkegaardresults}. Furthermore, the correlation between the crime rate in Denmark and Norway for the countries of origin in both samples is very high about .7-.8 (N=20). In this paper we further test the hypothesis by examining data from The National Longitudinal Survey of Freshmen. \begin{table}[H] \centering \begin{tabular}{|l|l|l|l|} \hline \textbf{Predictor} & \textbf{Variable} & \textbf{Denmark (N=71)} & \textbf{Norway (N=21)} \\ \hline Islam & Crime & .593 to .787 & .695 to .805 \\ \hline IQ & Crime & -.467 to -.653 & -.620 \\ \hline GDP & Crime & -.287 to -.414 & -.449 to -.512 \\ \hline Height & Crime & -.036 to -.218 & -.287 to -.300 \\ \hline Murder rate & Crime & .058 to .242 & .059 to .101 \\ \hline Islam & Fertility & .671 & \\ \hline IQ & Fertility & -.514 & \\ \hline GDP & Fertility & -.316 & \\ \hline Fertility & Fertility & .504 & \\ \hline Islam & Employment rate & & -.764 \\ \hline IQ & Employment rate & & .507 \\ \hline GDP & Employment rate & & .598 \\ \hline \end{tabular} \caption{Predictors and socio-economic variables in Denmark and Norway.}\label{kirkegaardresults} \end{table} \section{Methods} The National Longitudinal Survey of Freshman is described as follows\cite{NLSF}: \begin{quote} The National Longitudinal Survey of Freshmen (NLSF) follows a cohort of first-time freshman at selective colleges and universities through their college careers. Equal numbers of whites, blacks, Hispanics, and Asians were sampled at each of the 28 participating schools. \end{quote} As this sample is not representative of migrants to the U.S. and as relevant variables were self reported or interviewee assessed, this is not an ideal sample for testing our hypothesis. Nonetheless, these problematic factors will tend to attenuate any true association and, thus, make for a more robust test of the ST hypothesis. A detailed description of variables is provided in the supplementary material. \section{Analysis 1} We inspected scores by nativity status and U.S. defined race. This analysis yielded few interesting results. First and second generation self defined Blacks, Hispanics, Whites, and Asians performed similar to their respective third generation racial peers; the same pattern of score differences as found in the U.S. population as a whole was found in this selective college sample; the relationship between cumulative GPA and tests scores did not vary by nativity within each racial group, though it did vary significantly between racial groupings. Results are shown in Table \ref{analysis1a} and \ref{analysis1b}. \begin{table}[H] \includegraphics[width=\linewidth]{analysis1a} \caption{Results from analysis 1a.}\label{analysis1a} \end{table} \begin{table}[H] \centering \includegraphics[width=300px]{analysis1b} \caption{Results from analysis 1b.}\label{analysis1b} \end{table} \section{Analysis 2} We next looked at the association between three measures of national cognitive ability, a measure of national skin reflectance, combined first and second generation test scores, and combined first and second generation skin color ratings. With regards to migrants, we computed test scores, GPA, and skin color means separately by biological mother's and biological father's nation of origin; we then n-weight averaged the (mother and father) nation of origin means. In the vast majority of instances, both parents hailed from the same country; when not, though, we effectively split their representation. Since we were concerned with migrant scores, the U.S. was not included as a migrant sending country. When computing correlations, we weighted scores by the number of migrants per country. The correlations are shown below. All were statistically significant at the .05 level. Results are shown in Table \ref{analysis2a}. All three measures of national cognitive ability were significantly and similarly correlated with migrant test performance. The correlation between migrant skin color and national skin reflectance was .891, implying that our migrants were relatively ethnically representative of their nation of origin populations. Since our national cognitive measures were similarly predictive, for the remainder of the discussion, we simply report result based on L\&V's (2012) national IQs. \begin{table}[H] \centering \includegraphics[width=350px]{analysis2a} \caption{Results from analysis 2a.}\label{analysis2a} \end{table} We further looked at the association between L\&V's (2012) national IQs, migrant first and second generation test scores, and migrant second generation GPA scores. The values were weighted by the migrant sample sizes. The correlation results are shown in Table \ref{analysis2b}. The national IQ x test score association did not vary much by generation. Also, L\&V's (2012) National IQs predicted migrant national group GPA about as well as migrant test scores did on the individual level and half as well as migrant test scores did on the national level. \begin{table}[H] \centering \includegraphics[width=350px]{analysis2b} \caption{Results from analysis 2b.}\label{analysis2b} \end{table} \section{Analysis 3} We looked to see if the association between national cognitive measures and second generation migrant GPA scores was mediated by second generation migrant test scores. The results for migrant GPA (dependent), L\&V's (2012) national IQ (independent), and migrant test scores (covariant) are shown below in Table \ref{analysis3a}. The national cognitive ability x GPA associations were significantly mediated by migrant test scores. \begin{table}[H] \centering \includegraphics[width=350px]{analysis3a} \caption{Results from analysis 3a.}\label{analysis3a} \end{table} Similarly, we checked if the association between migrant skin color and migrant test scores was mediated by national cognitive measures. (We also included national skin reflectance values as a covariate.) This was found to be the case for first, second, and the combined generation samples. The results for the combined sample are shown below in Table \ref{analysis3b}. \begin{table}[H] \centering \includegraphics[width=350px]{analysis3b} \caption{Results from analysis 3b.}\label{analysis3b} \end{table} \section{Analysis 4} As a robustness check, we reran analysis 2 after taking into account parental migrant selectivity. To compute selectivity we took the difference between the parents' standardized mean educational levels as reported in the NLSF survey and the standardized average schooling years for the origin countries. For the country of origin values, we used age 25-29 data for year 1980, as this would have been the approximate cohort which birthed the NLSF students; the data came from Barro-Lee's dataset. The results are shown in Table \ref{analysis4a}. Controlling for migrant selectivity substantially increased the \textit{r} (national cognitive measures x migrant test score correlations). \begin{table}[H] \centering \includegraphics[width=350px]{analysis4a} \caption{Results from analysis 4a.}\label{analysis4a} \end{table} Taking into account selection also increased the National IQ x GPA association from \textit{r}(806) = .266, \textit{p} $ < $ .01 to a first-order partial correlation of r(668) = .441, \textit{p} $ < $ .01. Another way to approach this matter is to control for, instead of selectivity, parental educational levels. When doing so, the National IQ x Test Score and National IQ x GPA associations, respectively \textit{r}(803) = .612, \textit{p} $ < $ .01 and \textit{r}(803) = .351, \textit{p} $ < $ .01 are strong and significant. \section{General discussion and conclusion} We have found more evidence in support of the spatial transferability hypothesis (ST) which posits the first and second generation migrants carry their nation and region of origin IQs with them and that these IQs have predictive value. In the sample analyzed, National IQs predicted migrant tests scores and GPA; the association does not seem to be explainable in terms of migrant unrepresentativeness with regards to ethnicity or human capital. This finding is consistent with previous ones. Further analyses need to be conducted to see if the ST hypothesis holds up in more representative samples and in other countries. As we found that migrant selectivity substantially moderated the association between national cognitive ability and migrant cognitive ability, future analyses should attempt to take into account migrant selectivity. We demonstrated one method by which this could be done. We further note that NLSF is a selected sample for cognitive ability and so there is restriction of range. We did not correct for this because we did not know how strong the restriction was. This restriction of range attenuates the correlations reported here. \section{Acknowledgments} We thank for providing multiple datasets. \section{Detailed methods and data} Detailed methods and data can be found in the supplementary material. \bibliography{refs} \end{document} implementation.tex \section{Optimizing a Leapfrog Triejoin in Scala}\label{sec:lftj-optimizations} A simple, idiomatic Scala implementation of the Tributary join is not able to beat Spark's \textit{BroadcastHashjoin} on any other query than the triangle query. Hence, we report on how to optimize the join. After, we are able to beat Spark's \textit{BroadcastHashjoin} on nearly all queries and datasets. We report measured run-times for the unfiltered 5-clique on the Amazon0601 dataset for different optimizations in~\cref{table:lftj-optimizations}. In total, we improved the \textsc{WCOJ} running time from 316.5 seconds to 54.1 seconds. We discuss the optimization in categories: Leapfrog Triejoin specific, binary search specific, Spark related, Scala related and general. We conclude the section with some changes we tried that do not improve performance. Binary search specific optimizations become a category on its own because the sorted search is the most expensive operation in the Tributary join. According to profiler sessions, the join spends more than 70\% of its time in this method. This result is in line with the observation that `in the Tributary join algorithm, the most expensive step is the binary search' from~\cite{myria-detailed}. \begin{table} \begin{tabular}{llr} \toprule Category & Optimization & Runtime \\ \midrule NA & Baseline & 316.5 \\ Scala & Custom insertion sort instead of Scala's \textit{sort} method & 310.6 \\ Scala & Maps instead of linear lookup of \textit{LJ}'s and \textit{TI}'s in \textit{LFTJ} & 171.6 \\ General & Factor out computed values which are reused in \textit{TI} & 153.7 \\ Binary Search & Linear search after galloping and binary search & 138.0 \\ Scala & Arrays instead of maps for \textit{LJ}'s and \textit{TI}'s in \textit{LFTJ} & 100.5 \\ Scala & while loop instead of foreach loop in \textit{LFTJ} & 85.7 \\ Scala & use of \textit{private[this]} & 84.3 \\ Scala & use of \textit{@inline} annotation & 82.4 \\ Spark & direct array access of input relationships & 76.9 \\ General & strength reduction of modulo operations & 68.9 \\ Binary search & linear search shortcut before galloping search and binary search & 64.0 \\ Binary search & less branches in binary search & 58.9 \\ Binary search & removing galloping search & 56.4 \\ LFTJ & no sorting in \textit{LF} \textit{init} method & 54.1 \\ \bottomrule \end{tabular} \caption{Optimizations to the \textsc{LFTJ} algorithm in Scala and their runtimes on the unfiltered 5-clique query on the Amazon0601 dataset. \textit{LFTJ}, \textit{LF} and \textit{TI} refers to the \textit{LeapfrogTriejoin}, \textit{LeapfrogJoin} and \textit{TrieIterator} component of the Leapfrog Triejoin algorithm. } \label{table:lftj-optimizations} \end{table} We applied one \textsc{LFTJ} specific optimization. The \textit{LeapfrogJoin.init} method is originally described to sort its \textit{TrieIterators} so the method \textit{leapfrogSearch} knows the position of the largest and smallest iterator (see~\cref{subsubsec:leapfrog-triejoin}). However, the method can be improved by avoiding to sort the \textit{TrieIterators}. We can start moving the \textit{TrieIterator} without sorting them and arrive at an ordered array in $\mathcal{O} (n)$ steps with $n$ defined as the size of the array. This approach improves over the original algorithm in two ways: (1) it starts moving the \textit{TrieIterators} to their next intersection immediately without sorting them first and (2) orders the array in fewer steps than traditional sorting algorithms. To implement this we find the maximum value for all iterators and store the index in $p$. Then we move the \textit{TrieIterator} at $p + 1$ to the least upper bound of the maximum value (by calling \textit{seek}) and store the result as the new maximum. We proceed with this process, wrapping \textit{p} around when it reaches \textit{iterators.length}, until \textit{p} equals the original maximum index. Now, we are either in a state in which all \textit{TrieIterators} point to the same value or we arrived at a state in which the iterators are sorted by their key value. In the first state, the \textit{LeapfrogJoin} is initialized; the array is sorted and the first value of the intersection found. In the other possible state, the array of \textit{TrieIterators} is sorted and we can use the original \textit{leapfrogSearch} (\cref{alg:leapfrogSearch}) to find the first key in the intersection. can proceed as in the original \textit{LeapfrogJoin.init} method. \Cref{table:lftj-optimizations} mentions two optimizations for the sorting in the \textit{LeapfrogJoin} \textit{init} method. The one described above is the second one. For the first one, which is no completely replaced by the second, we used a self-written insertion sort which is faster than Scala's array sort. Scala's array sort is slow because it copies the array twice and casts the values to \textit{Java.Object} such that it can use Java's sorting methods. An insertion sort is an asymptotical suboptimal algorithm but a good option given that a \textit{LeapfrogJoin} normally operates on less than 20 \textit{TrieIterators}. The binary search is the most expensive operation of the Leapfrog Triejoin. Hence, special attention needs to be paid while implementing it. Our most important optimization is to change to a linear search once we narrowed the search space to a certain threshold. We experiment with different thresholds and show the results in~\cref{subsec:linear-search-threshold}. We directly perform a linear search if the search space is smaller than the threshold from the beginning (see 12th optimization in \cref{table:lftj-optimizations}). Another important optimization is to avoid unnecessary if-statements in the loop of the binary search, e.g. the implementation on Wikipedia and many other example implementations use an if-statement with three branches for smaller, bigger and equal but two branches for greater than and less-or-equal suffice for a least upper bound search. A similar optimization can be applied to a linear search on a sorted array: intuitively one would use the while-loop condition \textit{array(i) > key $\wedge$ i < end} with \textit{key} being the key to find the least upper bound for, \textit{i} the loop invariant and \textit{end} the exclusive end of the search space. Anyhow, it is faster to check for \textit{key > array(end - 1)} once before the loop and return if this is the case because the value cannot be found in the search space. This obviously circumvents the main loop of the linear search; additionally, it simplifies the loop condition to \textit{array(i) > key}. The impact of this optimization is shown in the 13th row of \cref{table:lftj-optimizations}. The Spark infrastructure uses the interface \textit{ColumnVector} to represent columns of relationships. The implementation \textit{OnHeapColumnVector} is a simple wrapper around an array of the correct type with support for \textit{null} values and \textit{append} operations. First, we used this data structure to represent our columns but we could see a clear increase in performance by replacing it by an implementation that exposes the array to allow the binary search to run on the array directly. This is likely due to saving virtual function calls in the hottest part of our code. \Cref{table:lftj-optimizations} shows the results of this change in row 11. We found many standard optimizations and Scala specific optimizations to be really useful. These are the optimizations that brought the biggest performance improvements. However, they are well-known, so we mention them only in tabular form~\ref{table:lftj-optimizations}. For Scala-specific optimizations one can find good explanations at~\cite{databricks-scala-guide}. Apart from the aforementioned very useful optimizations, we investigated multiple other avenues in hope for performance improvements which did not succeed, we list these approaches here to save others the work of investigating: \begin{itemize} \item reimplement in Java \item use of a Galloping search before the binary search \item unrolling the while-loop in \textit{LeapfrogTriejoin} state machine (see \cref{alg:leapfrogTrieJoin-state-machine}) \item predicating the \textit{action} variable in \textit{LeapfrogTriejoin} state machine \end{itemize} Finally, we believe that code generation for specific queries that combines the functionality of \textit{LeapfrogTriejoin}, \textit{LeapfrogJoin} and \textit{TrieIterator} into one query-specific function would lead to noticeable performance improvements. The reason for this belief is that our implementation takes about 3.46 seconds for a triangle query on the Twitter social circle dataset while a triangle query-specific Julia implementation, of a colleague of ours, needs only half a second. The main difference between our implementation and his are: the language used (Julia is a high-performance, compiled language) and the fact that his implementation has no query interpretation overhead but cannot handle any other query than the triangle query. However, a code generated Leapfrog Triejoin is out of scope for this thesis, also, we are aware of efforts by RelationalAi to write a paper about this specific topic. % filter? % distinct filter does not help % but smaller than does - a lot % variable ordering coolmaksat/academic-kickstart0 @article{evaluating2017, abstract = {Background Ontologies are widely used as metadata in biological and biomedical datasets. Measures of semantic similarity utilize ontologies to determine how similar two entities annotated with classes from ontologies are, and semantic similarity is increasingly applied in applications ranging from diagnosis of disease to investigation in gene networks and functions of gene products. Results Here, we analyze a large number of semantic similarity measures and the sensitivity of similarity values to the number of annotations of entities, difference in annotation size and to the depth or specificity of annotation classes. We find that most similarity measures are sensitive to the number of annotations of entities, difference in annotation size as well as to the depth of annotation classes; well-studied and richly annotated entities will usually show higher similarity than entities with only few annotations even in the absence of any biological relation. Conclusions Our findings may have significant impact on the interpretation of results that rely on measures of semantic similarity, and we demonstrate how the sensitivity to annotation size can lead to a bias when using semantic similarity to predict protein-protein interactions.}, author = { and Hoehndorf, Robert}, doi = {10.1186/s13326-017-0119-z}, issn = {2041-1480}, journal = {Journal of Biomedical Semantics}, month = {feb}, number = {1}, pages = {7}, title = {Evaluating the effect of annotation size on measures of semantic similarity}, url = {https://doi.org/10.1186/s13326-017-0119-z}, urldate = {2018-11-05}, volume = {8}, year = {2017} } \begin{table}[htbp] \footnotesize \centering \small %\caption{Model geometries summary.} \begin{tabularx}{\textwidth}{cc} \toprule \midrule \textbf{Name} & \textbf{Number of phases}\\ bounded-RVE&3\\ \midrule \multicolumn{2}{X}{\textbf{Description}}\\ \multicolumn{2}{X}{Circular fiber inside a square matrix domain, bounded by two UD rectangular domains on the upper and lower side.}\\ \midrule \multicolumn{2}{X}{\textbf{Geometry of each phase}}\\ \multicolumn{2}{X}{Fiber: circular; matrix: square with circular inclusion at its center; UD: rectangular.}\\ \midrule \multicolumn{2}{X}{\textbf{Boundary conditions}}\\ \multicolumn{2}{X}{Free surface at $z=\pm l$.}\\ \midrule \multicolumn{2}{X}{\textbf{Imposed conditions}}\\ \multicolumn{2}{X}{Constant displacement $u_{x}|_{z=\pm l}=\bar{u}_{x}=\bar{\varepsilon}_{x}\cdot l$ at $x=\pm l$.}\\ \midrule \bottomrule \end{tabularx}% \label{tab:geom_tab2}% \end{table}%\hypertarget{memide_8c}{}\doxysection{memide.\+c File Reference} \label{memide_8c}\index{memide.c@{memide.c}} {\ttfamily \#include \char`\"{}types.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}defs.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}param.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}mmu.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}proc.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}x86.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}traps.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}spinlock.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}sleeplock.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}fs.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}buf.\+h\char`\"{}}\newline Include dependency graph for memide.\+c\+: % FIG 0 \doxysubsection*{Functions} \begin{DoxyCompactItemize} \item void \mbox{\hyperlink{memide_8c_aefb190a6104cb58c0bc1f8fec88d1307}{ideinit}} (void) \item void \mbox{\hyperlink{memide_8c_a709693afdb9b89d848e684e7acde1f8f}{ideintr}} (void) \item void \mbox{\hyperlink{memide_8c_a7f36b008f02088c86f76e98e05b55af5}{iderw}} (struct \mbox{\hyperlink{structbuf}{buf}} $\ast$b) \end{DoxyCompactItemize} \doxysubsection*{Variables} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{types_8h_a65f85814a8290f9797005d3b28e7e5fc}{uchar}} \mbox{\hyperlink{memide_8c_af73b74f3a51ba441c3b68b1cf3e276c3}{\+\_\+binary\+\_\+fs\+\_\+img\+\_\+start}} \mbox{[}$\,$\mbox{]} \item \mbox{\hyperlink{types_8h_a65f85814a8290f9797005d3b28e7e5fc}{uchar}} \mbox{\hyperlink{memide_8c_a10b9652e23b65245bbc2126350c915d3}{\+\_\+binary\+\_\+fs\+\_\+img\+\_\+size}} \mbox{[}$\,$\mbox{]} \end{DoxyCompactItemize} \doxysubsection{Function Documentation} \mbox{\Hypertarget{memide_8c_aefb190a6104cb58c0bc1f8fec88d1307}\label{memide_8c_aefb190a6104cb58c0bc1f8fec88d1307}} \index{memide.c@{memide.c}!ideinit@{ideinit}} \index{ideinit@{ideinit}!memide.c@{memide.c}} \doxysubsubsection{\texorpdfstring{ideinit()}{ideinit()}} {\footnotesize\ttfamily void ideinit (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{memide_8c_a709693afdb9b89d848e684e7acde1f8f}\label{memide_8c_a709693afdb9b89d848e684e7acde1f8f}} \index{memide.c@{memide.c}!ideintr@{ideintr}} \index{ideintr@{ideintr}!memide.c@{memide.c}} \doxysubsubsection{\texorpdfstring{ideintr()}{ideintr()}} {\footnotesize\ttfamily void ideintr (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the caller graph for this function\+: % FIG 1 \mbox{\Hypertarget{memide_8c_a7f36b008f02088c86f76e98e05b55af5}\label{memide_8c_a7f36b008f02088c86f76e98e05b55af5}} \index{memide.c@{memide.c}!iderw@{iderw}} \index{iderw@{iderw}!memide.c@{memide.c}} \doxysubsubsection{\texorpdfstring{iderw()}{iderw()}} {\footnotesize\ttfamily void iderw (\begin{DoxyParamCaption}\item[{struct \mbox{\hyperlink{structbuf}{buf}} $\ast$}]{b }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 2 Here is the caller graph for this function\+: % FIG 3 \doxysubsection{Variable Documentation} \mbox{\Hypertarget{memide_8c_a10b9652e23b65245bbc2126350c915d3}\label{memide_8c_a10b9652e23b65245bbc2126350c915d3}} \index{memide.c@{memide.c}!\_binary\_fs\_img\_size@{\_binary\_fs\_img\_size}} \index{\_binary\_fs\_img\_size@{\_binary\_fs\_img\_size}!memide.c@{memide.c}} \doxysubsubsection{\texorpdfstring{\_binary\_fs\_img\_size}{\_binary\_fs\_img\_size}} {\footnotesize\ttfamily \mbox{\hyperlink{types_8h_a65f85814a8290f9797005d3b28e7e5fc}{uchar}} \+\_\+binary\+\_\+fs\+\_\+img\+\_\+size\mbox{[}$\,$\mbox{]}} \mbox{\Hypertarget{memide_8c_af73b74f3a51ba441c3b68b1cf3e276c3}\label{memide_8c_af73b74f3a51ba441c3b68b1cf3e276c3}} \index{memide.c@{memide.c}!\_binary\_fs\_img\_start@{\_binary\_fs\_img\_start}} \index{\_binary\_fs\_img\_start@{\_binary\_fs\_img\_start}!memide.c@{memide.c}} \doxysubsubsection{\texorpdfstring{\_binary\_fs\_img\_start}{\_binary\_fs\_img\_start}} {\footnotesize\ttfamily \mbox{\hyperlink{types_8h_a65f85814a8290f9797005d3b28e7e5fc}{uchar}} \+\_\+binary\+\_\+fs\+\_\+img\+\_\+start\mbox{[}$\,$\mbox{]}} srl295/keyman ##Package: Core ##Status: Completed (I) ---------------------------------------------------------------------------------------------------- @@JvAppRegistryStorage.pas Summary Contains the TJvAppRegistryStorage component. Author ---------------------------------------------------------------------------------------------------- @@TJvAppRegistryStorage.RegRoot Summary Specifies the registry root HKEY to use. Description RegRoot specifies the registry root HKEY to use. It defaults to HKEY_CURRENT_USER. ---------------------------------------------------------------------------------------------------- @@TJvAppRegistryStorage.StorageOptions Summary Write here a summary (1 line) Description Write here a description See Also List here other properties, methods (comma seperated) Remove the 'See Also' section if there are no references ---------------------------------------------------------------------------------------------------- @@TJvAppRegistryStorage #JVCLInfo Summary Registry based application data storage. Description TJvAppRegistryStorage is a data storage backend that stores the data in the registry. As a descendant of TJvCustomAppStorage, it provides in the standard interface for storing data (such as settings, form positions and sizes, etc). This data will be stored in the registry. The component publishes the property and introduces a new property RegRoot to specify which registry root key should be used (HKEY_CURRENT_USER, HKEY_LOCAL_MACHINE, etc). All the abstract access methods are implemented. ---------------------------------------------------------------------------------------------------- @@TJvAppRegistryStorageOptions Summary Write here a summary (1 line) Description Write here a description ---------------------------------------------------------------------------------------------------- @@TJvAppRegistryStorage.UseOldDefaultRoot Summary Write here a summary (1 line) Description Write here a description See Also List here other properties, methods (comma seperated) Remove the 'See Also' section if there are no references 0 \vssub \subsection{~Licensing terms} \vssub Starting with model version 3.14, \ws\ is distributed under the following licensing terms: \vspace \baselineskip \noindent \centerline{ \rule[1mm]{47mm}{.5mm} {\rm start of licensing terms} \rule[1mm]{47mm}{.5mm}} \vspace \baselineskip \noindent Software, as understood herein, shall be broadly interpreted as being inclusive of algorithms, source code, object code, data bases and related documentation, all of which shall be furnished free of charge to the Licensee. Corrections, upgrades or enhancements may be furnished and, if furnished, shall also be furnished to the Licensee without charge. NOAA, however, is not required to develop or furnish such corrections, upgrades or enhancements. NOAA's software, whether that initially furnished or corrections or upgrades, are furnished “as is.” NOAA furnishes its software without any warranty whatsoever and is not responsible for any direct, indirect or consequential damages that may be incurred by the Licensee. Warranties of merchantability, fitness for any particular purpose, title, and non-infringement, are specifically negated. The Licensee is not required to develop any software related to the licensed software. However, in the event that the Licensee does so, the Licensee is required to offer same to NOAA for inclusion under the instant licensing terms with NOAA's licensed software along with documentation regarding its principles, use and its advantages. This includes changes to the wave model proper including numerical and physical approaches to wave modeling, and boundary layer parameterizations embedded in the wave model The Licensee is encouraged but not obligated to provide pre-and post processing tools for model input and output. The software required to be offered shall not include additional models to which the wave model may be coupled, such as oceanic or atmospheric circulation models. The software provided by the Licensee shall be consistent with the latest model version available to the Licensee, and interface routines to the software provided shall conform to programming standards as outlined in the model documentation. The software offered to NOAA shall be offered as is, without any warranties whatsoever and without any liability for damages whatsoever. NOAA shall not be required to include a Licensee's software as part of its software. Licensee's offered software shall not include software developed by others. A Licensee may reproduce sufficient software to satisfy its needs. All copies shall bear the name of the software with any version number as well as replicas of any applied copyright notice, trademark notice, other notices and credit lines. Additionally, if the copies have been modified, e.g. with deletions or additions, this shall be so stated and identified. All of Licensee's employees who have a need to use the software may have access to the software but only after reading the instant license and stating, in writing, that they have read and understood the license and have agreed to its terms. Licensee is responsible for employing reasonable efforts to assure that only those of its employees that should have access to the software, in fact, have access. The Licensee may use the software for any purpose relating to sea state prediction. No disclosure of any portion of the software, whether by means of a media or verbally, may be made to any third party by the Licensee or the Licensee's employees The Licensee is responsible for compliance with any applicable export or import control laws of the United States. \vspace \baselineskip \noindent \centerline{ \rule[1mm]{48mm}{.5mm} {\rm end of licensing terms} \rule[1mm]{48mm}{.5mm}} \vspace \baselineskip \noindent The software will be distributed through our web site after the Licensee has agreed to the license terms.cwimpy/ggplot2-tutorial %% This BibTeX bibliography file was created using BibDesk. %% http://bibdesk.sourceforge.net/ %% Created for at 2019-05-21 13:26:05 -0400 %% Saved with string encoding Unicode (UTF-8) @book{Wickham:2019, Address = {Boca Raton}, Author = {}, Date-Added = {2019-05-21 13:23:35 -0400}, Date-Modified = {2019-05-21 13:25:58 -0400}, Publisher = {Chapman & Hall/CRC}, Title = {Advanced R}, Year = {2019}} @book{Wickham:2016, Address = {New York}, Author = {}, Date-Added = {2019-05-21 10:58:10 -0400}, Date-Modified = {2019-05-21 10:58:10 -0400}, Publisher = {Springer}, Title = {ggplot2: Elegant Graphics for Data Analysis}, Year = {2016}, Bdsk-Url-1 = {http://ggplot2.org}} @book{Wickham:2017, Address = {Sebastopol}, Author = {Wickham, Hadley and }, Date-Added = {2019-05-21 10:58:10 -0400}, Date-Modified = {2019-05-21 10:58:10 -0400}, Keywords = {https://r4ds.had.co.nz}, Publisher = {O'Reilly Media}, Title = {R for Data Science: Import, Tidy, Transform, Visualize, and Model Data}, Year = {2017}} @book{Wilkinson:2006, Address = {New York, NY}, Author = {}, Date-Added = {2019-05-21 10:58:10 -0400}, Date-Modified = {2019-05-21 10:58:10 -0400}, Publisher = {Springer Science \& Business Media}, Title = {The Grammar of Graphics}, Year = {2006}} \chapter*{Abstract} {\small\indent abstract\ldots...} \vspace{4ex} {\small graph theory, graphs, networks, complex networks, data mining} algs/gfp2_algs.tex \begin{algorithm}[tp] \caption{Legendre symbol for $\F_{p^{2}}$; Alg.~8 from~\cite{adj2012sqrtEvenExt}} \label{alg:legendre-fp2} \begin{algorithmic}[1] \Function{LegendreFP2}{$a$} \Comment{$a = a_{0} + a_{1}i\in\F_{p^{2}}$} \State $\alpha = a_{0}^{2} + a_{1}^{2}$ \State \Return $\chi_{p}(\alpha)$ \EndFunction \end{algorithmic} \end{algorithm} \begin{algorithm}[tp] \caption{Square Root in $\F_{p^{2}}$; Alg.~9 from~\cite{adj2012sqrtEvenExt}} \label{alg:sqrt-fp2} \begin{algorithmic}[1] \Function{SqrtFP2}{$a$} \Comment{$a\in\F_{p^{2}}$; assuming $a$ is quadratic residue} \State $t = a^{\frac{p-3}{4}}$ \State $y = ta$ \State $\alpha = ty$ \If {$\alpha = -1$} \State $x = iy$ \Else \State $b = \parens{1+\alpha}^{\frac{p-1}{2}}$ \State $x = by$ \EndIf \State \Return $x$ \EndFunction \end{algorithmic} \end{algorithm} % Table created by stargazer v.5.2.2 by , Harvard University. E-mail: hlavac at fas.harvard.edu % Date and time: Tue, Feb 18, 2020 - 14:21:03 \begin{table}[!htbp] \centering \caption{} \label{nmds_envfit} \begin{tabular}{@{\extracolsep{5pt}} ccccc} \\[-1.8ex]\hline \hline \\[-1.8ex] var\_plot & NMDS1 & NMDS2 & r2 & p\_format \\ \hline \\[-1.8ex] MAT & 0.489 & -0.872 & 0.75 & p\textless 0.01 \\ MAP & 0.736 & 0.677 & 0.4 & p\textless 0.01 \\ MAT SD & -0.168 & 0.986 & 0.46 & p\textless 0.01 \\ CWD & 0.992 & 0.129 & 0.54 & p\textless 0.01 \\ \hline \\[-1.8ex] \end{tabular} \end{table} @article{xiong2020intelligent, title={Intelligent Thermal Control Algorithm Based on Deep Deterministic Policy Gradient for Spacecraft}, author={ and and }, journal={Journal of Thermophysics and Heat Transfer}, volume={34}, number={4}, pages={683--695}, year={2020}, publisher={American Institute of Aeronautics and Astronautics} }1-10 \hypertarget{classsinsy_1_1Mode}{\section{sinsy\-:\-:\-Mode \-Class \-Reference} \label{classsinsy_1_1Mode}\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}} } \subsection*{\-Public \-Member \-Functions} \begin{DoxyCompactItemize} \item \hypertarget{classsinsy_1_1Mode_a098246baedeee64fa0ed3b53547cf55f}{\hyperlink{classsinsy_1_1Mode_a098246baedeee64fa0ed3b53547cf55f}{\-Mode} ()}\label{classsinsy_1_1Mode_a098246baedeee64fa0ed3b53547cf55f} \begin{DoxyCompactList}\small\item\em constructor \end{DoxyCompactList}\item \hyperlink{classsinsy_1_1Mode_a1b2d5d264bc2de44e9efc27b98fb482c}{\-Mode} (const std\-::string \&str) \begin{DoxyCompactList}\small\item\em constructor \end{DoxyCompactList}\item \hypertarget{classsinsy_1_1Mode_ae7f0ff16a094082d2f9896c077fd6ddf}{\hyperlink{classsinsy_1_1Mode_ae7f0ff16a094082d2f9896c077fd6ddf}{\-Mode} (const \hyperlink{classsinsy_1_1Mode}{\-Mode} \&obj)}\label{classsinsy_1_1Mode_ae7f0ff16a094082d2f9896c077fd6ddf} \begin{DoxyCompactList}\small\item\em copy constructor \end{DoxyCompactList}\item \hypertarget{classsinsy_1_1Mode_a6b8809388084822aeac00adbcbe107f7}{virtual \hyperlink{classsinsy_1_1Mode_a6b8809388084822aeac00adbcbe107f7}{$\sim$\-Mode} ()}\label{classsinsy_1_1Mode_a6b8809388084822aeac00adbcbe107f7} \begin{DoxyCompactList}\small\item\em destructor \end{DoxyCompactList}\item \hypertarget{classsinsy_1_1Mode_ac8122253680690c29f4a48e01d7fef2a}{\hyperlink{classsinsy_1_1Mode}{\-Mode} \& \hyperlink{classsinsy_1_1Mode_ac8122253680690c29f4a48e01d7fef2a}{operator=} (const \hyperlink{classsinsy_1_1Mode}{\-Mode} \&obj)}\label{classsinsy_1_1Mode_ac8122253680690c29f4a48e01d7fef2a} \begin{DoxyCompactList}\small\item\em assignment operator \end{DoxyCompactList}\item bool \hyperlink{classsinsy_1_1Mode_a9a90d9c77d25798d8b4dd3c427a7cd3f}{operator==} (const \hyperlink{classsinsy_1_1Mode}{\-Mode} \&obj) const \begin{DoxyCompactList}\small\item\em equal \end{DoxyCompactList}\item bool \hyperlink{classsinsy_1_1Mode_a585d85c933d17df58e76a5ce5770a5b8}{operator!=} (const \hyperlink{classsinsy_1_1Mode}{\-Mode} \&obj) const \begin{DoxyCompactList}\small\item\em not equal \end{DoxyCompactList}\item void \hyperlink{classsinsy_1_1Mode_a406a7630df242c2e2cc0d751d684af32}{set} (const std\-::string \&str) \begin{DoxyCompactList}\small\item\em set mode \end{DoxyCompactList}\item const std\-::string \& \hyperlink{classsinsy_1_1Mode_af6cddd424f3303817245cdc324ef1589}{get} () const \begin{DoxyCompactList}\small\item\em get mode \end{DoxyCompactList}\end{DoxyCompactItemize} \subsection*{\-Static \-Public \-Attributes} \begin{DoxyCompactItemize} \item \hypertarget{classsinsy_1_1Mode_ae0408cf44be2a33f0da76c9f9500dadf}{static const \hyperlink{classsinsy_1_1Mode}{\-Mode} {\bfseries \-M\-A\-J\-O\-R}}\label{classsinsy_1_1Mode_ae0408cf44be2a33f0da76c9f9500dadf} \item \hypertarget{classsinsy_1_1Mode_a2e0a735ec05a389546ef8c1a2ff8d125}{static const \hyperlink{classsinsy_1_1Mode}{\-Mode} {\bfseries \-M\-I\-N\-O\-R}}\label{classsinsy_1_1Mode_a2e0a735ec05a389546ef8c1a2ff8d125} \end{DoxyCompactItemize} \subsection{\-Constructor \& \-Destructor \-Documentation} \hypertarget{classsinsy_1_1Mode_a1b2d5d264bc2de44e9efc27b98fb482c}{\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}!\-Mode@{\-Mode}} \index{\-Mode@{\-Mode}!sinsy::Mode@{sinsy\-::\-Mode}} \subsubsection[{\-Mode}]{\setlength{\rightskip}{0pt plus 5cm}{\bf \-Mode\-::\-Mode} ( \begin{DoxyParamCaption} \item[{const std\-::string \&}]{str} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily \mbox{[}explicit\mbox{]}}}}\label{classsinsy_1_1Mode_a1b2d5d264bc2de44e9efc27b98fb482c} constructor constructor \begin{DoxyParams}{\-Parameters} {\em str} & mode \\ \hline \end{DoxyParams} \subsection{\-Member \-Function \-Documentation} \hypertarget{classsinsy_1_1Mode_af6cddd424f3303817245cdc324ef1589}{\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}!get@{get}} \index{get@{get}!sinsy::Mode@{sinsy\-::\-Mode}} \subsubsection[{get}]{\setlength{\rightskip}{0pt plus 5cm}const std\-::string \& {\bf \-Mode\-::get} ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} ) const}}\label{classsinsy_1_1Mode_af6cddd424f3303817245cdc324ef1589} get mode get mode \begin{DoxyReturn}{\-Returns} mode \end{DoxyReturn} \hypertarget{classsinsy_1_1Mode_a585d85c933d17df58e76a5ce5770a5b8}{\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}!operator!=@{operator!=}} \index{operator!=@{operator!=}!sinsy::Mode@{sinsy\-::\-Mode}} \subsubsection[{operator!=}]{\setlength{\rightskip}{0pt plus 5cm}bool \-Mode\-::operator!= ( \begin{DoxyParamCaption} \item[{const {\bf \-Mode} \&}]{obj} \end{DoxyParamCaption} ) const}}\label{classsinsy_1_1Mode_a585d85c933d17df58e76a5ce5770a5b8} not equal not equal operator \hypertarget{classsinsy_1_1Mode_a9a90d9c77d25798d8b4dd3c427a7cd3f}{\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}!operator==@{operator==}} \index{operator==@{operator==}!sinsy::Mode@{sinsy\-::\-Mode}} \subsubsection[{operator==}]{\setlength{\rightskip}{0pt plus 5cm}bool \-Mode\-::operator== ( \begin{DoxyParamCaption} \item[{const {\bf \-Mode} \&}]{obj} \end{DoxyParamCaption} ) const}}\label{classsinsy_1_1Mode_a9a90d9c77d25798d8b4dd3c427a7cd3f} equal equal operator \hypertarget{classsinsy_1_1Mode_a406a7630df242c2e2cc0d751d684af32}{\index{sinsy\-::\-Mode@{sinsy\-::\-Mode}!set@{set}} \index{set@{set}!sinsy::Mode@{sinsy\-::\-Mode}} \subsubsection[{set}]{\setlength{\rightskip}{0pt plus 5cm}void {\bf \-Mode\-::set} ( \begin{DoxyParamCaption} \item[{const std\-::string \&}]{str} \end{DoxyParamCaption} )}}\label{classsinsy_1_1Mode_a406a7630df242c2e2cc0d751d684af32} set mode set mode \begin{DoxyParams}{\-Parameters} {\em str} & mode \\ \hline \end{DoxyParams} \-The documentation for this class was generated from the following files\-:\begin{DoxyCompactItemize} \item lib/score/\-Mode.\-h\item lib/score/\-Mode.\-cpp\end{DoxyCompactItemize} capitulo11.tex \chapter{Autómatas con Transiciones Epsilon} Extenderemos la definición de los AFND para incluir transiciones de un estado a otro que no dependen de ninguna entrada. \textbf{Notación: }AFND-$\varepsilon$ \textbf{Definición: }Formalmente un AFND-$\varepsilon$ es una quíntupla $N=(S,I,\delta,s^*,F)$ donde $S,I,s^*,F$ se definen como en los AFND. $\delta$ es una función total de $(S\times I \cup\{\varepsilon\} \rightarrow P(s))$\\ \textbf{Ejemplo: }El siguiente AFND-$\varepsilon$ (Figura \ref{img_11_1}), $I=\{0,1,2\}$ %grafico1 \begin{figure}[h!] \centering \includegraphics[width=0.5\textwidth]{img_11_1.png} \caption{Diagrama de Transición} \label{img_11_1} \end{figure} $$L(N)=\{0^m1^n2^p/m\geq0,n\geq 0,p\geq 0\}$$ \textbf{Ejemplo: }Averiguar si la cadena $w=002$ es aceptada por el AFND-$\varepsilon$ $N$ (Figura \ref{img_11_1}), usando la definición recursiva. \begin{align*} \widehat{\delta}(s_0,\downlegend{0}{a}\downlegend{02}{u}) &=\widehat{\delta}(\delta(s_0,0),\downlegend{0}{a}\downlegend{2}{u})=\widehat{\delta}(s_0,02)=\widehat{\delta}(\delta(s_0,0),2) \\ &=\widehat{\delta}(s_0,\downlegend{2}{$\varepsilon 2$})=\widehat{\delta}(\delta(s_0,\varepsilon),2)=\widehat{\delta}(s_1,\downlegend{2}{$\varepsilon 2$})=\widehat{\delta}(\delta(s_1,\varepsilon),2)\\ &=\widehat{\delta}(s_2,2)=\delta(s_2,2)=s_2 \end{align*} $w$ es aceptado por $N$, pues $s_2\in F$. El camino que va de $s_0$ hacia $s_2$ es: $$s_0\rightarrow \downlegend{s_0}{0}\rightarrow \downlegend{s_0}{0}\rightarrow \downlegend{s_1}{$\varepsilon$}\rightarrow \downlegend{s_2}{$\varepsilon$} \rightarrow \downlegend{s_2}{2}$$ \textbf{Representación de un AFND}-$\varepsilon$: $\delta$ toma un estado de $S$ y un elemento de $I\cup \{\varepsilon\}$. \textbf{Ejemplo: }Dibuje la tabla de transición para $N$ \begin{center} $\begin{array}{c|cccc} s &0 &1 &2 & \varepsilon \\ \hline s_0 &s_0 &\phi &\phi &s_1 \\ s_1 &\phi &s_1 &\phi &s_2 \\ \sharp s_2 &\phi &\phi &s_2 &\phi \end{array}$ \end{center} \textbf{Definición: }Para todo estado $s\in S$ definimos la $\varepsilon$-clausura de $s$ como: $$clausura_\epsilon(s)=\{ q/q \mbox{ es accesible desde }s\mbox{ sin consumir nada en la entrada} \}$$ \textbf{Nota: }El estado $s$ pertenece a $clausura_\epsilon(s)$. \textbf{Ejemplo: }En el AFND-$\varepsilon$ $N$ (Figura \ref{img_11_1}) determine la clausura $\varepsilon$ para: \begin{itemize} \item $s=s_0$ \item $s=s_1$ \item $s=s_2$ \end{itemize} \textbf{Solución: } \begin{itemize} \item $clausura_\varepsilon(s_0)=\{s_0,s_1,s_2\}$ \item $clausura_\varepsilon(s_1)=\{s_1,s_2\}$ \item $clausura_\varepsilon(s_2)=\{s_2\}$ \end{itemize} Podemos extender la $\varepsilon$-clausura a un conjunto de estados $Q$ como sigue: \begin{align*} clausura_\varepsilon(Q)&=\bigcup_{q\in Q}clausura_\varepsilon(q)\\ clausura_\varepsilon(\{q_1,q_2,...,q_{i_N}\})&= \bigcup_{k=1}^N clausura_\varepsilon(q_{i_k})\\ Q\subseteq S \end{align*} \textbf{Ejemplo: }Dado el siguiente diagrama (Figura \ref{img_11_2}): %grafico2 \begin{figure}[h!] \centering \includegraphics[width=0.5\textwidth]{img_11_2.png} \caption{Diagrama de Transición} \label{img_11_2} \end{figure} Determine: \begin{itemize} \item $clausura_\varepsilon(s_1)$ \item El camino seguido para llegar a cada estado \end{itemize} \textbf{Solución: } \begin{itemize} \item $clausura_\varepsilon(s_1)=\{s_1,s_2,s_3,s_4,s_6\}$ \item $\begin{array}{c|c} s &\mbox{camino} \\ \hline s_1 &s_1 \\ s_2 &s_1\rightarrow s_2 \\ s_3 &s_1\rightarrow s_2\rightarrow s_3 \\ s_4 &s_1\rightarrow s_4 \\ s_6 &s_1\rightarrow s_2\rightarrow s_3\rightarrow s_6 \end{array}$ \end{itemize} \textbf{Definición: }La función de transición extendida a cadenas $\widehat{\delta}$ se define recursivamente: \begin{itemize} \item $\widehat{\delta}(s,\varepsilon)=clausura_\varepsilon(s)$ \item $\widehat{\delta}(s,ua)=clausura_\varepsilon(Q)$ Donde: $Q=\{q/\exists r\in \widehat{\delta}(s,u)\land q\in \delta(r,a)\}$ $a\in I;u\in I^*;r,s\in S$ %%revisar \end{itemize} \textbf{Definición: }El lenguaje aceptado por un AFND-$\varepsilon$ $N=(S,I,\delta,s^*,F)$ es el conjunto. $$L(N)=\{w/\widehat{\delta}(s^*,w)\land F\not= \phi\}$$ \textbf{Ejemplo: }Determine si $N$ acepta a la cadena $w=01$ en (Figura \ref{img_11_1}). \textbf{Solución: }Hallamos primero $\widehat{\delta}(s_0,\varepsilon)$ y $\widehat{\delta}(s_0,0)$, $F=\{s_2\}$. \begin{itemize} \item $\widehat{\delta}(s_0,\varepsilon)=clausura_\varepsilon(s_0)=\{s_0,s_1,s_2\}$ \item \begin{align*} \widehat{\delta}(s_0,\underbrace{0}_{\varepsilon 0})&=clausura_\varepsilon(\delta(\widehat{\delta}(s_0,\underbrace{\varepsilon}_{u}),\underbrace{0}_{a}))\\ &=clausura_\varepsilon(\delta(\{\underbrace{s_0,s_1,s_2}_{Q}\},0))\\ &=clausura_\varepsilon(\delta(s_0,0)\cup \delta(s_1,0)\cup \delta(s_2,0))\\ &=clausura_\varepsilon(s_0)=\{s_0,s_1,s_2\} \end{align*} Así: \begin{align*} \widehat{\delta}(s_0,\underbrace{0}_{u}\underbrace{1}_{a})&=clausura_\varepsilon(\delta(\widehat{\delta}(s_0,0),1))\\ &=clausura_\varepsilon(\delta(\{\underbrace{s_0,s_1,s_2}_{Q}\},1))\\ &=clausura_\varepsilon(\underbrace{\delta(s_0,1)}_{\phi}\cup \underbrace{\delta(s_1,1)}_{s_1}\cup \underbrace{\delta(s_2,1)}_{\phi})\\ &=clausura_\varepsilon(s_1)=\{s_1,s_2\} \end{align*} Como $\{s_1,s_2\}\cap F=\{s_2\}$, se acepta $w$. \end{itemize} \textbf{Definición: }Se define el conjunto de estado que siguen a $p$, pasando por $\sigma$, mediante: $$d(p,\sigma)=\{q/\mbox{hay una transición de }p\mbox{ a }q\mbox{ etiquetada por }\sigma\}\qquad \sigma\in I$$ \textbf{Extensión} $$d(\{\underbrace{p_{i_1},p_{i_2},...,p_{i_k},a}_{Q}\})=\bigcup_{j=1}^{k}d(p_{i_j},a)$$ \textbf{Ejemplo: }Sea el AFND-$\varepsilon$ dado el DT (Figura \ref{img_11_3}): %grafico3 \begin{figure}[h!] \centering \includegraphics[width=0.4\textwidth]{img_11_3.png} \caption{Diagrama de Transición} \label{img_11_3} \end{figure} $I=\{a,b\}$ $F=\{s_2\}$ Obtener: \begin{itemize} \item $\mbox{clausura}_\varepsilon(s_3) \qquad d(s_0,a)$ \item $\mbox{clausura}_\varepsilon(s_0) \qquad d(s_0,b)$ \item $\mbox{clausura}_\varepsilon(s_4) \qquad d(\{s_3,s_4\},b)$ \end{itemize} \textbf{Solución: } \begin{align*} \mbox{clausura}_\varepsilon(s_3) &=\{s_3\} \\ \mbox{clausura}_\varepsilon(s_0) &=\{s_0,s_1,s_2\} \\ \mbox{clausura}_\varepsilon(s_4) &=\{s_4,s_1,s_2\} \\ d(s_0,a) &=\{s_3\} \\ d(s_0,b) &=\phi \\ d(\{s_3,s_4\},b) &=d(s_3,b)\cup d(s_4,b)=\{s_4,s_0\} \end{align*}@misc{rfc6247, series = {Request for Comments}, number = 6247, howpublished = {RFC 6247}, publisher = {RFC Editor}, doi = {10.17487/RFC6247}, url = {https://rfc-editor.org/rfc/rfc6247.txt}, author = {}, title = {{Moving the Undeployed TCP Extensions RFC 1072, RFC 1106, RFC 1110, RFC 1145, RFC 1146, RFC 1379, RFC 1644, and RFC 1693 to Historic Status}}, pagetotal = 4, year = 2011, month = may, abstract = {This document reclassifies several TCP extensions that have never seen widespread use to Historic status. The affected RFCs are RFC 1072, RFC 1106, RFC 1110, RFC 1145, RFC 1146, RFC 1379, RFC 1644, and RFC 1693. This document is not an Internet Standards Track specification; it is published for informational purposes.}, } Atalay-Ileri/Disksec-Doc \begin{BVerbatim}[commandchars=\\\{\},codes={\catcode`\$=3\catcode`\^=7\catcode`\_=8},fontsize=\footnotesize] \PY{k+kn}{Theorem} \PY{n}{unseal\PYZus{}public\PYZus{}to\PYZus{}state\PYZus{}noninterference} \PY{o}{:} \PY{k}{forall} \PY{o}{`}\PY{o}{(}\PY{n}{p} \PY{o}{:} \PY{n}{proc} \PY{n}{T}\PY{o}{)}\PY{o}{,} \PY{n}{unseal\PYZus{}public} \PY{n}{p} \PY{o}{\PYZhy{}\PYZgt{}} \PY{n}{state\PYZus{}noninterference} \PY{n}{p}\PY{o}{.} \end{BVerbatim} \chapter{Electrodynamics} In this chapter, we would like to present an important example of field theory. We want to focus on the Hamiltonian description of electrodynamics and see that it yields the classical results. The Lagrangian of the free electromagnetic field without charges is \begin{align} L[A^{\mu}(\bar{x}), \dot{A}^{\mu}(\bar{x})] = - \frac{1}{4} \displaystyle\int d^3 x \ F_{\mu \nu}(\bar{x}) F^{\mu \nu}(\bar{x}), \end{align} where we use natural units and \begin{align} F_{\mu \nu}(\bar{x}) = \partial_{\mu} A_{\nu}(\bar{x}) - \partial_{\nu} A_{\mu}(\bar{x}) \end{align} is the antisymmetric field tensor. Since it is expressed through the vector potential $A^{\mu}(\bar{x})$, the vector potential takes the role of our field $\varphi(\bar{x})$. Note that we have four different fields, since $\mu = 0,1,2,3$. That's why we define the generalized momenta to be \begin{align} \pi_{\rho}(\bar{y}) = \frac{\delta L}{\delta \dot{A}^{\rho}(\bar{y})}. \end{align} Inserting the Lagrangian, one gets \begin{align} \pi_{\rho}(\bar{y}) &= \frac{\delta L}{\delta \dot{A}^{\rho}(\bar{y})} = - \frac{1}{4} \int d^3 x \ \frac{\delta \left( F_{\mu \nu}(\vec{x}) F^{\mu \nu}(\bar{x}) \right)}{\delta \dot{A}^{\rho}(\bar{y})} \notag \\ &= - \frac{1}{2} \int d^3 x \ F_{\mu \nu}(\bar{x}) \ \frac{\delta \left(\partial^{\mu} A^{\nu}(\bar{x}) - \partial^{\nu} A^{\mu}(\bar{x}) \right)}{\delta \dot{A}^{\rho}(\bar{y})} \notag \\ &= - \frac{1}{2} \int d^3 x \ F_{\mu \nu}(\bar{x}) \left( \delta_0^{\mu} \ \frac{\delta \dot{A}^{\nu}(\bar{x})}{\delta \dot{A}^{\rho}(\bar{y})} - \delta_0^{\nu} \ \frac{\delta \dot{A}^{\mu}(\bar{x})}{\delta \dot{A}^{\rho}(\bar{y})} \right) \notag \\ &= - \frac{1}{2} \int d^3 x \ F_{\mu \nu}(\bar{x}) \left( \delta_0^{\mu} \delta_{\rho}^{\nu} \ \delta^3(\bar{x} - \bar{y}) - \delta_0^{\nu} \delta_{\rho}^{\mu} \ \delta^3(\bar{x} - \bar{y}) \right) \notag \\ &= - \frac{1}{2} \int d^3 x \ \left( F_{0 \rho}(\bar{x}) \ \delta^3(\bar{x} - \bar{y}) - F_{\rho 0}(\bar{x}) \ \delta^3(\bar{x} - \bar{y}) \right) \notag \\ &= - \frac{1}{2} \left( F_{0 \rho}(\bar{y}) - F_{\rho 0}(\bar{y}) \right) = F_{\rho 0}(\bar{y}) . \end{align} Using the explicit formula for $F_{\mu \nu}$, the generalized momenta can be written as \begin{align} \pi_{\rho}(\bar{y}) = \partial_{\rho} A_0(\bar{y}) - \dot{A}_{\rho}(\bar{y}). \end{align} \label{sec:electrodynamics_primary_constraints} To find the Hamiltonian, one has to invert this equation and express $\dot{A}_{\rho}(\vec{y})$ through $\pi_{\rho}(\vec{y})$ and the field components. We see that this won't be possible for $\rho = 0$, we have a primary constraint: \begin{align} \pi_0(\bar{y}) = \partial_0 A_0(\bar{y}) - \dot{A}_0(\bar{y}) = 0 \ \ \ \Longrightarrow \ \ \ \phi_1 = \pi_0(\bar{y}) = 0 \ \ \forall \bar{y} \in \mathbb{R}^3. \end{align} More exactly, we have infinitly many constraints because the momentum $\pi_0(\bar{y})$ vanishes in every point in space. This corresponds to the gauge freedom of $A^0(\bar{x})$ as we will see later. No matter how we choose it, the conjugated momentum vanishes identically. For the other components, everything is fine and no more constraints arise. \\ \label{sec:electrodynamics_hamiltonian} The Hamiltonian is therefore \begin{align} H[A^{\mu}(\bar{x}), \pi_{\mu}(\bar{x})] &= \int d^3 x \ \dot{A}^{\mu}(\bar{x}) \pi_{\mu}(\bar{x}) \ - \ L[A^{\mu}(\bar{x}), \dot{A}^{\mu}(\bar{x})] \notag \\ &= \int d^3 x \ \dot{A}^{\mu}(\bar{x}) \pi_{\mu}(\bar{x}) \ + \ \frac{1}{4} \displaystyle\int d^3 x \ F_{\mu \nu}(\bar{x}) F^{\mu \nu}(\bar{x}) \notag \\ &= \int d^3 x \left( \dot{A}^{i}(\bar{x}) \pi_{i}(\bar{x}) + \frac{1}{2} F_{i 0}(\bar{x}) F^{i 0}(\bar{x}) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) \right) \notag \\ &= \int d^3 x \left( \left( \pi_i(\bar{x}) - \partial_i A_0(\bar{x}) \right) \pi_i(\bar{x}) + \frac{1}{2} \pi_i(\bar{x}) (- \pi_i(\bar{x})) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) \right) \notag \\ &= \int d^3 x \left( \frac{1}{2} \pi_i(\bar{x})\pi_i(\bar{x}) - \partial_i A_0(\bar{x}) \pi_i(\bar{x}) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) \right) \notag \\ &= \int d^3 x \left( \frac{1}{2} \pi_i(\bar{x})\pi_i(\bar{x}) + A_0(\bar{x}) \partial_i \pi_i(\bar{x}) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) \right), \end{align} where in the last step we integrated by parts and assumed that the fields vanish at infinity. Terms like $\pi_i(\bar{x})\pi_i(\bar{x})$, where both indices are down or up, denote usual summation. \\ The first term contains the momenta and is equivalent to the kinetic energy. The last term contains spatial derivatives of the fied components and is therefore equivalent to a potential energy. The second term is a mix of kinetic and potential energy and should not appear in a regular Hamiltonian. We have to check whether the consistency condition for our primary constraint is fulfilled or we have secondary constraints. \\ The total Hamiltonian is \begin{align} H_T = H + \int d^3 x \ u(\bar{x}) \pi_0(\bar{x}). \end{align} And the equation of motion is given by \begin{align} \dot{g} = \left \{ g,H_T \right \} = \left \{ g,H \right \} + \int d^3 x \ u(\bar{x}) \left \{ g,\pi_0(\bar{x}) \right \}, \end{align} where the Poisson bracket is given by \begin{align} \left \{ f,g \right \} = \int d^3 z \left( \frac{\delta f}{\delta A^{\mu}(\bar{z})} \frac{\delta g}{\delta \pi_{\mu}(\bar{z})} - \frac{\delta g}{\delta A^{\mu}(\bar{z})} \frac{\delta f}{\delta \pi_{\mu}(\bar{z})} \right), \end{align} analog to the definition in field theory. It follows that \begin{align} 0 \overset{!}{=} \dot{\pi}_0(\bar{y}) &= \left \{ \pi_0(\bar{y}),H(\bar{y}) \right \} + \int d^3 x \ u(\bar{x}) \left \{ \pi_0(\bar{y}),\pi_0(\bar{x}) \right \} \notag \\ &= - \int d^3 z \ \frac{\delta \pi_0(\bar{y})}{\delta \pi_{\mu}(\bar{z})} \frac{\delta H(\bar{y})}{\delta A^{\mu}(\bar{z})} = - \frac{\delta H(\bar{y})}{\delta A^0(\bar{y})} \notag \\ &= - \int d^3 x \ \partial_i \pi_i(\bar{x}) \frac{\delta A^0(\bar{x})}{\delta A^0(\bar{y})} = - \partial_i \pi_i(\bar{y}). \end{align} \label{sec:electrodynamics_secondary_constraints} So we really get a secondary constraint which is \begin{align} \phi_2 = \partial_i \pi_i = \text{div} \ \bar{\pi} = 0. \end{align} Again we have to proof that this constraint is fulfilled at every moment (if not, then the primary constraint wouldn't be fulfilled at every moment). \\ It follows that \begin{align} 0 \overset{!}{=} \frac{d}{dt} \Big( \partial_i \pi_i(\bar{y}) \Big) &= \left \{ \partial_i \pi_i(\bar{y}),H(\bar{y}) \right \} + \int d^3 x \ u(\bar{x}) \left \{ \partial_i \pi_i(\bar{y}),\pi_0(\bar{x}) \right \} \notag \\ &= - \int d^3 z \ \frac{\delta (\partial_i B_i(\bar{y}))}{\delta B_{\mu}(\bar{z})} \frac{\delta H(\bar{y})}{\delta A^{\mu}(\bar{z})} = - \int d^3 z \ \partial_i \left(\delta^3(\bar{y} - \bar{z})\right) \frac{\delta H(\bar{y})}{\delta A^i(\bar{z})} \notag \\ &= - \partial^i \left( \frac{\delta H(\bar{y})}{\delta A^i(\bar{y})} \right), \end{align} where the variation is given by \begin{align} \frac{\delta H(\bar{y})}{\delta A^i(\bar{y})} &= \frac{\delta}{\delta A^i(\bar{y})} \left( \frac{1}{4} \int d^3 x \ F_{lk}(\bar{x})F^{lk}(\bar{x}) \right) \notag \\ &= \frac{1}{2} \int d^3 x \ F_{lk}(\bar{x}) \frac{\delta F^{lk}(\bar{x})}{\delta A^i(\bar{y})} \notag \\ &= \frac{1}{2} \int d^3 x \ F_{lk}(\bar{x}) \frac{\delta (\partial^l A^k(\bar{x}) - \partial^k A^l(\bar{x}))}{\delta A^i(\bar{y})} \notag \\ &= \frac{1}{2} \int d^3 x \ F_{lk}(\bar{x}) \left( \delta_i^k \partial^l \left(\delta^3(\bar{x} - \bar{y})\right) - \delta_i^l \partial^k \left(\delta^3(\bar{x} - \bar{y})\right) \right) \notag \\ &= \frac{1}{2} \int d^3 x \ \left( - \partial^l F_{li}(\bar{x}) + \partial^k F_{ik}(\bar{x}) \right) \delta^3(\bar{x} - \bar{y}) \notag \\ &= - \partial^k F_{ki}(\bar{y}). \end{align} In the end, we get \begin{align} \frac{d}{dt} \Big( \partial_i \pi_i(\bar{y}) \Big) = \partial^i \partial^k F_{ki}(\bar{y}), \end{align} which is clearly zero since we have a contraction of a symmetric with an antisymmetric tensor. So the secondary constraint is automatically conserved and we get no more constraints. \\ Now, that our procedure is finished, we can classify the constraints and say which is first-class and which is second-class. Since both constraints depends only on the momentum, they are first-class. So one can write the generalized Hamiltonian as \begin{align} H_E = H + \int d^3 x \ v(\bar{x}) \pi_0(\bar{x}) + \int d^3 x \ V(\bar{x}) \partial_i \pi_i(\bar{x}), \end{align} where $v(\bar{x})$ and $V(\bar{x})$ are arbitrary functions. \\ \label{sec:electrodynamics_transformations} Since we have two undetermined functions in our Hamiltonian, this corresponds to two degrees of gauge freedom. Let's see what transformations $g \longrightarrow g' = g + \delta g$ these first-class constraints do generate. \\ We calculate the small shift $\delta g$ analog to the classical discrete case: \begin{align} \delta g = \varepsilon_m \left\{ g,\phi_m \right\} \ \ \ \Longrightarrow \ \ \ \delta g = \int d^3 y \ \varepsilon(\bar{y}) \left\{ g,\phi(\bar{y}) \right\}. \end{align} We also have a kind of canonical commutation relation \begin{align} \left\{ A^{\mu}(\bar{x}), \pi_{\nu}(\bar{y}) \right\} = \delta_{\nu}^{\mu} \ \delta^3(\bar{x} - \bar{y}), \end{align} analog to $\left\{ q_i, p_k \right\} = \delta_{ik}$. Now, we are able to determine the gauge transformations generated by the two constraints. \begin{itemize} \item Choosing $g = A^{\mu}(\bar{x})$, we get for the first constraint: \begin{align} \delta A^{\mu}(\bar{x}) &= \int d^3 y \ \varepsilon(\bar{y}) \left\{ A^{\mu}(\bar{x}), \pi_0(\bar{y}) \right\} \notag \\ &= \int d^3 y \ \varepsilon(\bar{y}) \ \delta_0^{\mu} \ \delta^3(\bar{x} - \bar{y}) \notag \\ &= \delta_0^{\mu} \ \varepsilon(\bar{x}). \end{align} So the first constraint only generates transformations \begin{align} A_0(\bar{x}) \ \longrightarrow \ A'_0(\bar{x}) = A_0(\bar{x}) + \varepsilon(\bar{x}) \end{align} and leaves the momentum unchanged since $\left\{ \pi_{\mu}(\bar{x}), \pi_0(\bar{y}) \right\} = 0$. \item The second constraint leaves the momentum unchanged too because of the same reason. Furthermore it leaves $A_0$ unchanged because the constraint contains only spatial terms and the Poisson bracket vanishes. So we have \begin{align} \delta A^k(\bar{x}) &= \int d^3 y \ \tilde{\varepsilon}(\bar{y}) \left\{ A^k(\bar{x}), \partial_i \pi_i(\bar{y}) \right\} \notag \\ &= - \int d^3 y \ \tilde{\varepsilon}(\bar{y}) \ \delta_i^k \ \partial^i \left( \delta^3(\bar{x} - \bar{y}) \right) \notag \\ &= \partial^k \tilde{\varepsilon}(\bar{x}). \end{align} So the second constraint generates transformations \begin{align} A^k(\bar{x}) \ \longrightarrow \ A'^k(\bar{x}) = A^k(\bar{x}) + \partial^k \tilde{\varepsilon}(\bar{x}), \end{align} or in vector notation \begin{align} \bar{A}(\bar{x}) \ \longrightarrow \ \bar{A}'(\bar{x}) = \bar{A}(\bar{x}) + \nabla \tilde{\varepsilon}(\bar{x}). \end{align} \end{itemize} So one can write the generalized Hamiltonian as \begin{align} H_E &= \int d^3 x \left( \frac{1}{2} \pi_i(\bar{x})\pi_i(\bar{x}) + (A_0(\bar{x}) + V(\bar{x})) \partial_i \pi_i(\bar{x}) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) + v(\bar{x}) \pi_0(\bar{x}) \right) \notag \\ &= \int d^3 x \left( \frac{1}{2} \pi_i(\bar{x})\pi_i(\bar{x}) + V(\bar{x}) \partial_i \pi_i(\bar{x}) + \frac{1}{4} F_{i k}(\bar{x}) F^{i k}(\bar{x}) + v(\bar{x}) \pi_0(\bar{x}) \right), \end{align} because of the gauge freedom of $A_0$, and we see that no mixed terms with momenta and coordinates appear. The Hamiltonian is regular and can be devided in a kinetic and potential energy part. \\ Let's analyse the spatial part of the Hamiltonian and see what it means in terms of the electric field $\bar{E}$ and the magnetic field $\bar{B}$. \\ Remembering that the electromagnetic field tensor in natural units is \begin{align} F^{\mu \nu} = \left( \arraycolsep=1.4pt\def\arraystretch{1.2} \begin{array}{cccc} 0 & - E^1 & - E^2 & - E^3 \\ E^1 & 0 & - B^3 & B^2 \\ E^2 & B^3 & 0 & - B^1 \\ E^3 & - B^2 & B^1 & 0 \end{array} \right) \ \ \ \text{and} \ \ \ \bar{B} = \nabla \times \bar{A}, \end{align} the momentum is nothing more than the electric field: \begin{align} \pi^i = F^{i 0} = E^i. \end{align} Moreover, if one notices that \begin{align} \bar{B}^2 &= \left( \varepsilon_{ijk} \partial^j A^k \right) \left( \varepsilon_{ilm} \partial^l A^m \right) \notag \\ &= \left( \delta_{jl} \delta_{km} - \delta_{jm} \delta_{kl} \right) \partial^j A^k \partial^l A^m \notag \\ &= \partial_l A_m(\bar{x}) \partial^l A^m(\bar{x}) - \partial_m A_l(\bar{x}) \partial^l A^m(\bar{x}), \end{align} we can rewrite \begin{align} F_{i k}(\bar{x}) F^{i k}(\bar{x}) &= \big( \partial_i A_k(\bar{x}) - \partial_k A_i(\bar{x}) \big) \left( \partial^i A^k(\bar{x}) - \partial^k A^i(\bar{x}) \right) \notag \\ &= 2 \left( \partial_i A_k(\bar{x}) \partial^i A^k(\bar{x}) - \partial_i A_k(\bar{x}) \partial^k A^i(\bar{x}) \right) \notag \\ &= 2 \bar{B}^2. \end{align} So the spatial part of the Hamiltonian can be written as \begin{align} H_E = \int d^3 x \left( \frac{\bar{E}^2(\bar{x})}{2} + \frac{\bar{B}^2(\bar{x})}{2} \right) + \int d^3 x \ V(\bar{x}) \ \nabla \cdot \bar{E}(\bar{x}), \end{align} which is the classical result for the energy of the electromagnetic field. The first term is the kinetic term, the second term is the potential and the third term generates the gauge transformation $\bar{A}' = \bar{A} + \nabla \varepsilon$ that leaves the magnetic field $B$ unchanged.\documentclass[11pt]{article} \usepackage{a4wide} \renewcommand{\baselinestretch}{1.6} \setlength{\parskip}{11pt} \begin{document} \pagestyle{empty} \large \begin{center} \textbf{CONSENT TO COPYRIGHT LICENSING UNDER OPEN SOURCE INITIATIVE APPROVED LICENSES} \end{center} \normalsize \vspace{1cm} On behalf of \underline{\hspace{12.5cm}} \\ (hereafter referred to as ``Organization'') I hereby give my consent to the redistribution and/or modification of Organization contributions to the below specified list of FEniCS Project components (hereafter referred to as ``Components''), made during the below specified time period, under the terms of a license approved by the Open Source Initiative (https://opensource.org/licenses/category). Each of the Components contains file(s) (LICENSE, COPYING, COPYING.LESSER) explicitly stating the license under which the Organization contributions can be modified and/or redistributed. In the case a component specifies the GPL or LGPL licenses as published by the Free Software Foundation; either version 3 of the License, or (at each distributor's option) any later version., \bigskip \begin{tabular}{ll} Components: & \underline{\hspace{9cm}} \\[1.5cm] Time period: & \underline{\hspace{9cm}} \\[3cm] Signature: & \underline{\hspace{9cm}} \\[1.5cm] Printed name: & \underline{\hspace{9cm}} \\[1.5cm] Position: & \underline{\hspace{9cm}} \\[1.5cm] Date signed: & \underline{\hspace{9cm}} \end{tabular} \end{document} ref.bib @inproceedings{shah2016towards, title={Towards manipulation planning for multiple interlinked deformable linear objects}, author={ and }, booktitle={Robotics and Automation (ICRA), 2016 IEEE International Conference on}, year={2016}, organization={IEEE} } @article{shah2018planning, title={Planning for Manipulation of Interlinked Deformable Linear Objects With Applications to Aircraft Assembly}, author={ and and }, journal={IEEE Transactions on Automation Science and Engineering}, year={2018}, publisher={IEEE} } @inproceedings{shah2018nips, title = {Bayesian Inference of Temporal Task Specifications from Demonstrations}, author = { and and }, booktitle = {Conference on Neural Information Processing Systems}, year = {2018}, } @inproceedings{shah2018rss, title = {Towards Specification Learning from Demonstrations}, author = { and }, booktitle = {Robotics: Science and Systems, Workshop on Learning From Demonstrations for High-Level Robotics Tasks}, year = {2018}, } @inproceedings{kim2019ijcai, title = {Bayesian Inference of Temporal Specifications to Explain How Plans Differ}, author = { and and and and }, booktitle = {International Joint Conference on Artificial Intelligence}, year = {2019}, } @inproceedings{shah2019ijrr, title = {Supervised Bayesian Specification Inference from Demonstrations}, author = { and and and and and and }, booktitle = {(under review)}, year = {2019}, } @inproceedings{kim2019xaip, title = {Bayesian Inference of Temporal Specifications to Explain How Plans Differ}, author = { and Muise, Christian and and and }, booktitle = {ICAPS 2019 Workshop on explainable AI in planning}, year = {2019}, } @inproceedings{gombolay2016appraisal, title={Appraisal of Statistical Practices in HRI vis-a-vis the T-Test for Likert Items/Scales}, author={ and }, booktitle={2016 AAAI Fall Symposium Series}, year={2016} } @mastersthesis{shah2016thesis, title = {Planning for Manipulation of Interlinked Deformable Linear Objects With Applications to Aircraft Assembly}, school = {Massachusetts Institure of Technology}, author = {}, year = {2016}, } @article{2019arXiv190603218S, author = {{Shah}, Ankit and {Li}, Shen and {Shah}, Julie}, title = "{Planning With Uncertain Specifications (PUnS)}", journal = {arXiv e-prints}, keywords = {Computer Science - Robotics}, year = "2019", month = "Jun", eid = {arXiv:1906.03218}, pages = {arXiv:1906.03218}, archivePrefix = {arXiv}, eprint = {1906.03218}, primaryClass = {cs.RO}, } @inproceedings{shah2019rss, title = {Planning with Uncertain Specifications}, author = { and }, booktitle = {Robotics: Science and Systems, Workshop on Combining Learning and Reasoning -- Towards Human-Level Robot Intelligence}, year = {2019}, } content/publication/brunswicker-creating-2016/cite.bib @article{brunswicker_creating_2016, abstract = {Modern science has become collaborative and digital. The Internet has supported the emergence of scientific digital platforms that globally connect programmers and users of novel digital scientific products such as scientific interactive software tools. These digital scientific innovations complement traditional text-based products like journal publications. This article is focused on the scientific impact of a platform’s programming community that produces these digital scientific innovations. The article’s main theoretical argument is that beyond an individual’s contribution efforts to these innovations, a new social structure affects his scientific recognition through citations of his tools in text-based publications. Taking a practice theory lens, we introduce the concept of a digital practice structure that emerges from the digital innovation work practice, performed by programmers who jointly work on a tool. This digital practice creates dependence forces among the community members in an analogy to Newton’s gravity concept. Our model represents such dependencies in a spatial autocorrelative model. We empirically estimate this model using data of the programming community of nanoHUB in which 477 nanotechnology tool programmers have contributed more than 715 million lines of code. Our results show that a programmer’s contributions to digital innovations may have positive effects, while the digital practice structure creates negative dependency effects. Colloquially speaking, being surrounded by star performers can be harmful. Our findings suggest that modeling scientific impact needs to account for a scientist’s contribution to programming communities that produce digital scientific innovations and the digital work structures in which these contributions are embedded.}, author = {, Klimeck, Gerhard}, issn = {0138-9130, 1588-2861}, journal = {Scientometrics}, language = {en}, month = {September}, pages = {1--26}, shorttitle = {Creating impact in the digital space}, title = {Creating impact in the digital space: digital practice dependency in communities of digital scientific innovations}, url = {http://link.springer.com/article/10.1007/s11192-016-2106-z}, urldate = {2016-09-14}, volume = {Online (09 September 2016)}, year = {2016} } \hypertarget{sysfile_8c}{}\doxysection{sysfile.\+c File Reference} \label{sysfile_8c}\index{sysfile.c@{sysfile.c}} {\ttfamily \#include \char`\"{}types.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}defs.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}param.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}stat.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}mmu.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}proc.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}fs.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}spinlock.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}sleeplock.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}file.\+h\char`\"{}}\newline {\ttfamily \#include \char`\"{}fcntl.\+h\char`\"{}}\newline Include dependency graph for sysfile.\+c\+: % FIG 0 \doxysubsection*{Functions} \begin{DoxyCompactItemize} \item int \mbox{\hyperlink{sysfile_8c_a8f8b9e2d98e8b444f23a39ae8992feff}{sys\+\_\+dup}} (void) \item int \mbox{\hyperlink{sysfile_8c_a54bf714d9e898cbdcbc061b280bbfae0}{sys\+\_\+read}} (void) \item int \mbox{\hyperlink{sysfile_8c_a687d939a9e4792af15db96f2c2f34378}{sys\+\_\+write}} (void) \item int \mbox{\hyperlink{sysfile_8c_a32945488fd39bc405757177b37cd2250}{sys\+\_\+close}} (void) \item int \mbox{\hyperlink{sysfile_8c_ac243c8f20f5fb2e3e257b5007af2c204}{sys\+\_\+fstat}} (void) \item int \mbox{\hyperlink{sysfile_8c_a759600870314007ac558871239122fb7}{sys\+\_\+link}} (void) \item int \mbox{\hyperlink{sysfile_8c_ae1e58ee11d41f643929520d8c1640da7}{sys\+\_\+unlink}} (void) \item int \mbox{\hyperlink{sysfile_8c_a74e45efc661ca17c068bc283b3842e6d}{sys\+\_\+open}} (void) \item int \mbox{\hyperlink{sysfile_8c_a057e5bce2de7a87ebfd2dc33967bca4a}{sys\+\_\+mkdir}} (void) \item int \mbox{\hyperlink{sysfile_8c_a25697aa3d828b5878d38170d724adb27}{sys\+\_\+mknod}} (void) \item int \mbox{\hyperlink{sysfile_8c_ad1c5f8693cb35b9605fee09eebdda640}{sys\+\_\+chdir}} (void) \item int \mbox{\hyperlink{sysfile_8c_aeaa813ddeb6a5fac3c45714c7351c526}{sys\+\_\+exec}} (void) \item int \mbox{\hyperlink{sysfile_8c_a9a70db941def46ec25939e6c2d30e399}{sys\+\_\+pipe}} (void) \end{DoxyCompactItemize} \doxysubsection{Function Documentation} \mbox{\Hypertarget{sysfile_8c_ad1c5f8693cb35b9605fee09eebdda640}\label{sysfile_8c_ad1c5f8693cb35b9605fee09eebdda640}} \index{sysfile.c@{sysfile.c}!sys\_chdir@{sys\_chdir}} \index{sys\_chdir@{sys\_chdir}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_chdir()}{sys\_chdir()}} {\footnotesize\ttfamily int sys\+\_\+chdir (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 1 \mbox{\Hypertarget{sysfile_8c_a32945488fd39bc405757177b37cd2250}\label{sysfile_8c_a32945488fd39bc405757177b37cd2250}} \index{sysfile.c@{sysfile.c}!sys\_close@{sys\_close}} \index{sys\_close@{sys\_close}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_close()}{sys\_close()}} {\footnotesize\ttfamily int sys\+\_\+close (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{sysfile_8c_a8f8b9e2d98e8b444f23a39ae8992feff}\label{sysfile_8c_a8f8b9e2d98e8b444f23a39ae8992feff}} \index{sysfile.c@{sysfile.c}!sys\_dup@{sys\_dup}} \index{sys\_dup@{sys\_dup}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_dup()}{sys\_dup()}} {\footnotesize\ttfamily int sys\+\_\+dup (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{sysfile_8c_aeaa813ddeb6a5fac3c45714c7351c526}\label{sysfile_8c_aeaa813ddeb6a5fac3c45714c7351c526}} \index{sysfile.c@{sysfile.c}!sys\_exec@{sys\_exec}} \index{sys\_exec@{sys\_exec}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_exec()}{sys\_exec()}} {\footnotesize\ttfamily int sys\+\_\+exec (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 2 \mbox{\Hypertarget{sysfile_8c_ac243c8f20f5fb2e3e257b5007af2c204}\label{sysfile_8c_ac243c8f20f5fb2e3e257b5007af2c204}} \index{sysfile.c@{sysfile.c}!sys\_fstat@{sys\_fstat}} \index{sys\_fstat@{sys\_fstat}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_fstat()}{sys\_fstat()}} {\footnotesize\ttfamily int sys\+\_\+fstat (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{sysfile_8c_a759600870314007ac558871239122fb7}\label{sysfile_8c_a759600870314007ac558871239122fb7}} \index{sysfile.c@{sysfile.c}!sys\_link@{sys\_link}} \index{sys\_link@{sys\_link}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_link()}{sys\_link()}} {\footnotesize\ttfamily int sys\+\_\+link (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 3 \mbox{\Hypertarget{sysfile_8c_a057e5bce2de7a87ebfd2dc33967bca4a}\label{sysfile_8c_a057e5bce2de7a87ebfd2dc33967bca4a}} \index{sysfile.c@{sysfile.c}!sys\_mkdir@{sys\_mkdir}} \index{sys\_mkdir@{sys\_mkdir}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_mkdir()}{sys\_mkdir()}} {\footnotesize\ttfamily int sys\+\_\+mkdir (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 4 \mbox{\Hypertarget{sysfile_8c_a25697aa3d828b5878d38170d724adb27}\label{sysfile_8c_a25697aa3d828b5878d38170d724adb27}} \index{sysfile.c@{sysfile.c}!sys\_mknod@{sys\_mknod}} \index{sys\_mknod@{sys\_mknod}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_mknod()}{sys\_mknod()}} {\footnotesize\ttfamily int sys\+\_\+mknod (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 5 \mbox{\Hypertarget{sysfile_8c_a74e45efc661ca17c068bc283b3842e6d}\label{sysfile_8c_a74e45efc661ca17c068bc283b3842e6d}} \index{sysfile.c@{sysfile.c}!sys\_open@{sys\_open}} \index{sys\_open@{sys\_open}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_open()}{sys\_open()}} {\footnotesize\ttfamily int sys\+\_\+open (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 6 \mbox{\Hypertarget{sysfile_8c_a9a70db941def46ec25939e6c2d30e399}\label{sysfile_8c_a9a70db941def46ec25939e6c2d30e399}} \index{sysfile.c@{sysfile.c}!sys\_pipe@{sys\_pipe}} \index{sys\_pipe@{sys\_pipe}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_pipe()}{sys\_pipe()}} {\footnotesize\ttfamily int sys\+\_\+pipe (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 7 \mbox{\Hypertarget{sysfile_8c_a54bf714d9e898cbdcbc061b280bbfae0}\label{sysfile_8c_a54bf714d9e898cbdcbc061b280bbfae0}} \index{sysfile.c@{sysfile.c}!sys\_read@{sys\_read}} \index{sys\_read@{sys\_read}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_read()}{sys\_read()}} {\footnotesize\ttfamily int sys\+\_\+read (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{sysfile_8c_ae1e58ee11d41f643929520d8c1640da7}\label{sysfile_8c_ae1e58ee11d41f643929520d8c1640da7}} \index{sysfile.c@{sysfile.c}!sys\_unlink@{sys\_unlink}} \index{sys\_unlink@{sys\_unlink}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_unlink()}{sys\_unlink()}} {\footnotesize\ttfamily int sys\+\_\+unlink (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} Here is the call graph for this function\+: % FIG 8 \mbox{\Hypertarget{sysfile_8c_a687d939a9e4792af15db96f2c2f34378}\label{sysfile_8c_a687d939a9e4792af15db96f2c2f34378}} \index{sysfile.c@{sysfile.c}!sys\_write@{sys\_write}} \index{sys\_write@{sys\_write}!sysfile.c@{sysfile.c}} \doxysubsubsection{\texorpdfstring{sys\_write()}{sys\_write()}} {\footnotesize\ttfamily int sys\+\_\+write (\begin{DoxyParamCaption}\item[{void}]{ }\end{DoxyParamCaption})} pisnicky/zelenefrancouzskeplane.tex \hyt{zelenefrancouzskeplane} \song{Zelené francouzské pláně} \interpret{asonance}{Asonance} \vers{1}{ Tak \chord{D}jakpak se máš, Willy, \chord{G}příteli \chord{Em}můj?\\ Dovo\chord{A}líš, jen si sednu do \chord{G}stínu na hrob \chord{A}tvůj.\\ \chord{D}Horkým letním sluncem jsem \chord{G}rozpále\chord{Em}ná,\\ \chord{A}chodím celý den a jsem \chord{G}unave\chord{D}ná.\\ Podle \chord{D}náhrobku vidím, slavils \chord{Em}osmnáct let,\\ když jsi v \chord{A}zbytečné válce o\chord{D}pustil tenhle \chord{A}svět.\\ \chord{D}Doufám, že zemřel jsi \chord{Em}čistý a hned, že jsi \chord{A}nestřílel první, když \chord{G}přišel rozkaz \chord{D}vpřed. } \refrain{ Hrály \chord{A}píšťaly k tomu, nesli \chord{G}tě zpátky \chord{D}domů a \chord{Em}zazněly salvy, když tě \chord{A}spouštěli \chord{D}níž.\\ A když \chord{G}kněz dělal kříž nad tvým \chord{A}tělem, začly \chord{G}dudy zvolna \chord{Em}hrát tesknou \chord{A}pí\chord{D}seň. } \vers{2}{ Zanechal jsi tu dívku nebo snad ženu svou?\\ Vzpomínky v jejím srdci nikdy nevyblednou.\\ I když zemřel jsi dávno, před mnoha lety,\\ v jejích očích jsi zůstal osmnáctiletý.\\ Nebo jsi pouhý cizinec beze jména,\\ v dlouhém seznamu padlých, jehož připomíná\\ zažloutlá fotka chlapce s hnědýma očima, která nikoho dávno už nezajímá? } \refsm{} \vers{3}{ Na francouzských pláních vlčí máky kvetou,\\ barvy ze strání svítí pestrou paletou.\\ Horký letní vítr začal od moře vát,\\ žádné pušky a plyn, žádný ostnatý drát.\\ Jenom tisíce křížů tady v písku stojí\\ a svým tichým hlasem k oblakům žalují,\\ že z touhy člověka vládnout vzešlo utrpení, že zničil a proklel celá pokolení. } \refsm{} \cod{ A když kněz dělal kříž nad tvým tělem, začly dudy zvolna hrát tesknou píseň. } \newpage @Article{VanVuuren2017c, Author = { and and and and and and and and and and and and and and and and and and }, Title = {Energy, land-use and greenhouse gas emissions trajectories under a green growth paradigm}, Journal = {Global Environmental Change}, Year = {2017}, Volume = {42}, Pages = {237--250}, Doi = {10.1016/j.gloenvcha.2016.05.008}, Url = {https://linkinghub.elsevier.com/retrieve/pii/S095937801630067X} } 0 Histoire racontée par Cherie, la maman d'Everest, à : Un jour après avoir reçu le vaccin Pfizer, Everest est tombé malade. Cinq jours après avoir été vacciné, le jeune homme de 17 ans a été hospitalisé lorsque les médecins ont découvert deux caillots de sang dans son cerveau. Everest, un joueur de basket-ball universitaire à la Corner Canyon High School de Draper, dans l'Utah, était un enfant athlétique en très bonne santé qui était en pleine saison de recrutement pour l'équipe de basket-ball de son école. Aujourd'hui, l'adolescent peut à peine marcher! Moins de 24 heures après avoir été vacciné, Everest a commencé à ressentir une “quantité exorbitante” de douleur et d'enflure au cou, du même côté que celui où il a été vacciné. Cherie a emmené son fils chez le pédiatre, qui a diagnostiqué un muscle froissé et lui a mis une minerve. Le médecin a été “plutôt dédaigneux”, a déclaré Cherie. “On nous a renvoyés à la maison.” Mais dans les 24 heures suivant la visite du médecin, Everest a eu “l'une des pires migraines qu'il ait jamais eues”. Cela a duré sans interruption pendant plusieurs jours avant que Cherie n'emmène son fils aux urgences. 10-100 \newcommand{\ReferenceDESThroughput}{12.01} \newcommand{\ReferenceDESSLOC}{1053} \newcommand{\UsubaDESThroughput}{11.47} \newcommand{\UsubaDESSLOC}{655} \newcommand{\DESAbsoluteSpeedup}{+4.50} \newcommand{\ReferenceAESKSThroughput}{7.77} \newcommand{\ReferenceAESKSSLOC}{272} \newcommand{\UsubaAESKSThroughput}{7.92} \newcommand{\UsubaAESKSSLOC}{218} \newcommand{\AESKSAbsoluteSpeedup}{-1.93} \newcommand{\ReferenceAESKiviThroughput}{5.59} \newcommand{\ReferenceAESKiviSLOC}{339} \newcommand{\UsubaAESKiviThroughput}{5.71} \newcommand{\UsubaAESKiviSLOC}{218} \newcommand{\AESKiviAbsoluteSpeedup}{-2.15} \newcommand{\ReferenceChachaAVXtwoThroughput}{1.03} \newcommand{\ReferenceChachaAVXtwoSLOC}{20} \newcommand{\UsubaChachaAVXtwoThroughput}{1.02} \newcommand{\UsubaChachaAVXtwoSLOC}{24} \newcommand{\ChachaAVXtwoAbsoluteSpeedup}{+0.97} \newcommand{\ReferenceChachaAVXThroughput}{2.09} \newcommand{\ReferenceChachaAVXSLOC}{134} \newcommand{\UsubaChachaAVXThroughput}{2.07} \newcommand{\UsubaChachaAVXSLOC}{24} \newcommand{\ChachaAVXAbsoluteSpeedup}{+0.96} \newcommand{\ReferenceChachaSSEThroughput}{2.72} \newcommand{\ReferenceChachaSSESLOC}{134} \newcommand{\UsubaChachaSSEThroughput}{2.31} \newcommand{\UsubaChachaSSESLOC}{24} \newcommand{\ChachaSSEAbsoluteSpeedup}{+15.07} \newcommand{\ReferenceChachaGPThroughput}{5.64} \newcommand{\ReferenceChachaGPSLOC}{26} \newcommand{\UsubaChachaGPThroughput}{5.65} \newcommand{\UsubaChachaGPSLOC}{24} \newcommand{\ChachaGPAbsoluteSpeedup}{-0.18} \newcommand{\ReferenceSerpentAVXtwoThroughput}{4.33} \newcommand{\ReferenceSerpentAVXtwoSLOC}{300} \newcommand{\UsubaSerpentAVXtwoThroughput}{4.53} \newcommand{\UsubaSerpentAVXtwoSLOC}{214} \newcommand{\SerpentAVXtwoAbsoluteSpeedup}{-4.62} \newcommand{\ReferenceSerpentAVXThroughput}{8.36} \newcommand{\ReferenceSerpentAVXSLOC}{300} \newcommand{\UsubaSerpentAVXThroughput}{8.66} \newcommand{\UsubaSerpentAVXSLOC}{214} \newcommand{\SerpentAVXAbsoluteSpeedup}{-3.59} \newcommand{\ReferenceSerpentSSEThroughput}{11.48} \newcommand{\ReferenceSerpentSSESLOC}{300} \newcommand{\UsubaSerpentSSEThroughput}{11.29} \newcommand{\UsubaSerpentSSESLOC}{214} \newcommand{\SerpentSSEAbsoluteSpeedup}{+1.66} \newcommand{\ReferenceSerpentGPThroughput}{30.37} \newcommand{\ReferenceSerpentGPSLOC}{300} \newcommand{\UsubaSerpentGPThroughput}{25.78} \newcommand{\UsubaSerpentGPSLOC}{214} \newcommand{\SerpentGPAbsoluteSpeedup}{+15.11} \newcommand{\ReferenceRectangleGPThroughput}{28.61} \newcommand{\ReferenceRectangleGPSLOC}{115} \newcommand{\UsubaRectangleGPThroughput}{25.88} \newcommand{\UsubaRectangleGPSLOC}{31} \newcommand{\RectangleGPAbsoluteSpeedup}{+9.54} \newcommand{\ReferenceRectangleAVXThroughput}{4.92} \newcommand{\ReferenceRectangleAVXSLOC}{115} \newcommand{\UsubaRectangleAVXThroughput}{4.21} \newcommand{\UsubaRectangleAVXSLOC}{31} \newcommand{\RectangleAVXAbsoluteSpeedup}{+14.43} \newcommand{\ReferenceRectangleSSEThroughput}{14.51} \newcommand{\ReferenceRectangleSSESLOC}{115} \newcommand{\UsubaRectangleSSEThroughput}{11.18} \newcommand{\UsubaRectangleSSESLOC}{31} \newcommand{\RectangleSSEAbsoluteSpeedup}{+22.95} \newcommand{\ReferenceRectangleAVXtwoThroughput}{2.45} \newcommand{\ReferenceRectangleAVXtwoSLOC}{115} \newcommand{\UsubaRectangleAVXtwoThroughput}{2.10} \newcommand{\UsubaRectangleAVXtwoSLOC}{31} \newcommand{\RectangleAVXtwoAbsoluteSpeedup}{+14.29} \documentclass[11pt,letterpaper]{article} %\setlength{\textwidth}{14cm} \usepackage[margin=3cm]{geometry} \usepackage[utf8]{inputenc} \usepackage{colortbl} \usepackage[spanish]{babel} \usepackage{amsmath} \usepackage{etoolbox} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{graphicx} \usepackage{pdflscape} \usepackage{afterpage} \usepackage{cite} \usepackage{footnote} \usepackage{capt-of} \usepackage{float} %\usepackage{times} %\usepackage{dsfont} \usepackage{hyperref} \hypersetup{ bookmarks=false, % show bookmarks bar unicode=false, % non-Latin characters in Acrobat’s bookmarks pdftoolbar=true, % show Acrobat’s toolbar? pdfmenubar=true, % show Acrobat’s menu? pdffitwindow=false, % window fit to page when opened pdfstartview={FitH}, % fits the width of the page to the window pdftitle={Anteproyecto: Trabajo de grado}, % title pdfauthor={}, % author pdfsubject={Anteproyecto}, % subject of the document %pdfcreator={Creator}, % creator of the document %pdfproducer={Producer}, % producer of the document %pdfkeywords={Phase Diversity} {Fase} {Aberración} {Vórtice Óptico}, % list of keywords pdfnewwindow=true, % links in new window colorlinks=True, % false: boxed links; true: colored links linkcolor=red, % color of internal links (change box color with linkbordercolor) citecolor=blue, % color of links to bibliography filecolor=magenta, % color of file links urlcolor=cyan % color of external links } %\setlength{\parindent}{0pt} \newcommand{\etal}{\emph{et al.} } \newcommand{\invivo}{\emph{in-vivo} } \newcommand{\exvivo}{\emph{ex-vivo} } \begin{document} \renewcommand{\listfigurename}{Lista de Figuras} \renewcommand{\listtablename}{Lista de Tablas} \renewcommand{\contentsname}{Contenido} \renewcommand{\tablename}{Tabla} \renewcommand{\bibname}{Referencias} \input{front/Cubierta} \input{front/Contenido} \input{chap/chapter01} \input{chap/chapter02} \input{chap/chapter03} \input{chap/chapter04} \addcontentsline{toc}{section}{Referencias} \bibliography{bib/Referencias} \bibliographystyle{unsrt} \end{document}vs-team/Papers %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % BENCHMARKS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% - Windows, Xbox, Wp7 (, iPad?) - memory recycling - parallel execution - query optimization 0 % By , Beerware licence is acceptable... \documentclass[tikz,border=10pt]{standalone} \usepackage{tikz} \usetikzlibrary{positioning, fit, arrows.meta, shapes} \begin{document} \begin{tikzpicture}[ % GLOBAL CFG font=\sf \scriptsize, >=LaTeX, % Styles cell/.style={% For the main box rectangle, rounded corners=5mm, draw, very thick, }, operator/.style={%For operators like + and x circle, draw, inner sep=-0.5pt, minimum height =.2cm, }, function/.style={%For functions ellipse, draw, inner sep=1pt }, ct/.style={% For external inputs and outputs circle, draw, line width = .75pt, minimum width=1cm, inner sep=1pt, }, gt/.style={% For internal inputs rectangle, draw, minimum width=4mm, minimum height=3mm, inner sep=1pt }, mylabel/.style={% something new that I have learned font=\scriptsize\sffamily }, ArrowC1/.style={% Arrows with rounded corners rounded corners=.25cm, thick, }, ArrowC2/.style={% Arrows with big rounded corners rounded corners=.5cm, thick, }, ] %Start drawing the thing... % Draw the cell: \node [cell, minimum height =4cm, minimum width=6cm] at (0,0){} ; % Draw inputs named ibox# \node [gt] (ibox1) at (-2,-0.75) {$\sigma$}; \node [gt] (ibox2) at (-1.5,-0.75) {$\sigma$}; \node [gt, minimum width=0.7cm] (ibox3) at (-0.5,-0.75) {tanh}; \node [gt] (ibox4) at (0.5,-0.75) {$\sigma$}; % Draw opérators named mux# , add# and func# \node [operator] (mux1) at (-2,1.5) {$\times$}; \node [operator] (add1) at (-0.5,1.5) {+}; \node [operator] (mux2) at (-0.5,0) {$\times$}; \node [operator] (mux3) at (1.5,0) {$\times$}; \node [gt, minimum width=0.7cm] (func1) at (1.5,0.75) {tanh}; % Draw External inputs? named as basis c,h,x \node[ct, label={[mylabel]Cell}] (c) at (-4,1.5) {$c_{t-1}$}; \node[ct, label={[mylabel]Hidden}] (h) at (-4,-1.5) {$h_{t-1}$}; \node[ct, label={[mylabel]left:Input}] (x) at (-2.5,-3) {$x_t$}; % Draw External outputs? named as basis c2,h2,x2 \node[ct, label={[mylabel]Label1}] (c2) at (4,1.5) {$c_t$}; \node[ct, label={[mylabel]Label2}] (h2) at (4,-1.5) {$h_t$}; \node[ct, label={[mylabel]left:Label3}] (x2) at (2.5,3) {$h_t$}; % Start connecting all. %Intersections and displacements are used. % Drawing arrows \draw [ArrowC1] (c) -- (mux1) -- (add1) -- (c2); % Inputs \draw [ArrowC2] (h) -| (ibox4); \draw [ArrowC1] (h -| ibox1)++(-0.5,0) -| (ibox1); \draw [ArrowC1] (h -| ibox2)++(-0.5,0) -| (ibox2); \draw [ArrowC1] (h -| ibox3)++(-0.5,0) -| (ibox3); \draw [ArrowC1] (x) -- (x |- h)-| (ibox3); % Internal \draw [->, ArrowC2] (ibox1) -- (mux1); \draw [->, ArrowC2] (ibox2) |- (mux2); \draw [->, ArrowC2] (ibox3) -- (mux2); \draw [->, ArrowC2] (ibox4) |- (mux3); \draw [->, ArrowC2] (mux2) -- (add1); \draw [->, ArrowC1] (add1 -| func1)++(-0.5,0) -| (func1); \draw [->, ArrowC2] (func1) -- (mux3); %Outputs \draw [-, ArrowC2] (mux3) |- (h2); \draw (c2 -| x2) ++(0,-0.1) coordinate (i1); \draw [-, ArrowC2] (h2 -| x2)++(-0.5,0) -| (i1); \draw [-, ArrowC2] (i1)++(0,0.2) -- (x2); \end{tikzpicture} \end{document}texdocs/figs/run_38/run_38_tw_contour.tex \begin{figure}[H] \centering \includegraphics[width=4.25in]{figs/run_38/run_38_tw_contour} \caption{Contour plot of $\overline{v_{\theta}^{\prime} v_{z}^{\prime}}$ at $z/c$=8.50, $V_{free}$=28.81, station 4.} \label{fig:run_38_tw_contour} \end{figure} @article{MTMT:2220872, author = { Murinkó, }, doi = {10.4054/DemRes.2013.28.1}, eissn = {1435-9871}, journal = {DEMOGRAPHIC RESEARCH}, journal-iso = {DEMOGR RES}, keywords = {Europe; WOMEN; RATES; TRANSITION; DETERMINANTS; fertility decline}, orcid-numbers = {Bartus, Tamás/0000-0002-8356-9408; Murinkó, Lívia/0000-0002-4145-2412}, pages = {1-32}, title = {The effect of education on second births in Hungary}, unique-id = {2220872}, volume = {28}, year = {2013} } Doxygen/latex/df/d41/namespace_t_h_b___plugin___lesson.tex \hypertarget{namespace_t_h_b___plugin___lesson}{}\section{T\+H\+B\+\_\+\+Plugin\+\_\+\+Lesson Namespace Reference} \label{namespace_t_h_b___plugin___lesson}\index{T\+H\+B\+\_\+\+Plugin\+\_\+\+Lesson@{T\+H\+B\+\_\+\+Plugin\+\_\+\+Lesson}} \subsection*{Namespaces} \begin{DoxyCompactItemize} \end{DoxyCompactItemize} \subsection*{Classes} \begin{DoxyCompactItemize} \item class \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_assesments}{Assesments}} \begin{DoxyCompactList}\small\item\em \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_assesments}{Assesments}} \end{DoxyCompactList}\item class \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_attendance_log}{Attendance\+Log}} \begin{DoxyCompactList}\small\item\em \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_attendance_log}{Attendance\+Log}} \end{DoxyCompactList}\item class \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_lesson_board}{Lesson\+Board}} \begin{DoxyCompactList}\small\item\em \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_lesson_board}{Lesson\+Board}} \end{DoxyCompactList}\item class \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_repository}{Repository}} \begin{DoxyCompactList}\small\item\em \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_repository}{Repository}} \end{DoxyCompactList}\item class \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_user_control1}{User\+Control1}} \begin{DoxyCompactList}\small\item\em \mbox{\hyperlink{class_t_h_b___plugin___lesson_1_1_user_control1}{User\+Control1}} \end{DoxyCompactList}\end{DoxyCompactItemize} nevenjovanovic/modruski-temrezah %-*-coding:utf-8;-*- %&LaTeX \documentclass[a5paper,twoside]{article} \usepackage{polyglossia} \setmainlanguage{croatian} \setotherlanguage{latin} \defaultfontfeatures{Ligatures=TeX} \usepackage{indentfirst,titling,enumitem,tabto} % prilagodba dimenzija stranice \usepackage[a5paper,pdftex]{geometry}%a4paper,pdftex %\usepackage[a4paper,pdftex]{geometry} \geometry{scale=1,paperwidth=161mm,paperheight=230mm, bindingoffset=0in,top=27mm,inner=21mm,outer=26mm,width=117mm,height=180mm}% \parindent=7mm %\usepackage[small,center,rm]{titlesec} %\titlespacing{\section}{0pt}{11mm}{7mm} \usepackage{verse} \usepackage[medium,center,rm]{titlesec} \titlelabel{\thesection.\enspace} \titlelabel{\thesubsection.\enspace} % \chapter, \subsection...: no additional code \titleformat{\section} {\normalfont\large\bfseries\uppercase} {\thesection.\enspace}{0em}{} \titlespacing{\section} {7mm}{3.5ex plus .1ex minus .2ex}{1.5ex minus .1ex} \titleformat{\subsection} {\normalfont\large\bfseries} {\thesubsection.\enspace}{0em}{} \titlespacing{\subsection} {7mm}{3.5ex plus .1ex minus .2ex}{1.5ex minus .1ex} % titling \pretitle{\begin{center}\vskip 5mm} \posttitle{\par\end{center}} \preauthor{} \postauthor{} \predate{} \postdate{} % slova \setmainfont{Times New Roman} %\setsansfont{Old Standard TT} % makro za regum gesta \newcommand{\kratica}{\emph{Regum gesta}}% % reference bold \newcommand{\refb}[1]{\textbf{\ref{#1}}} \renewcommand{\labelenumi}{\Alph{enumi}} %tekuće glave %tekuće glave \usepackage{fancyhdr} \pagestyle{fancy} \fancyhead{} % clearall \fancyhead[RO,LE]{\thepage} \fancyhead[EC]{\MakeUppercase{: Djela u službi pape Siksta IV.}} \fancyhead[OC]{\MakeUppercase{Govor za (1474)}} \fancyfoot{} \renewcommand{\headrulewidth}{0.4pt} \makeatletter \renewcommand{\@makefnmark}{\mbox{% \textsuperscript{\normalfont\@thefnmark\ }}} \makeatother \renewcommand\footnotesize{\fontsize{10}{10.5} \selectfont} \renewcommand\tiny{\fontsize{9}{9.5} \selectfont} \renewcommand\large{\fontsize{11}{11.5} \selectfont} \renewcommand\Large{\fontsize{12.5}{13} \selectfont} % fusnote i crta \renewcommand{\footnoterule}{\vspace*{-7pt} \noindent\rule{2in}{0.4pt}\vspace*{6.6pt}} \setlength{\headheight}{12.6pt} % makro za naziv programa \newcommand{\makeup}[1]{\MakeUppercase{\emph{#1}}} % prilagodba quote % prilagođavamo citat propozicijama \renewenvironment{quote} {\list{}{\rightmargin 0mm \leftmargin 7mm \itemindent 0em}% \item\relax} {\endlist} \newenvironment{opisi}{% \let\olditem\item% \renewcommand\item[2][]{\olditem ##1 ##2 \itemsep 3pt}% \begin{description}}{\end{description}% } %siročići i udovice % \widowpenalty=10000 % \clubpenalty=10000 \begin{document} %%tth:\begin{html}\end{html} \hyphenation{his-to-ri-o-graf-ske po-li-tič-ko po-li-tič-ko-his-to-ri-o-graf-ske} \hyphenation{quae-sti-o-nes quin-qua-gin-ta quo-que DXXXVIII Zuo-ni-me-ro} \hyphenation{quad-ra-ta quo-mo-do Schwand-tne-ro-vi cro-a-tiae I-su-krs-ta} \hyphenation{Pas-qua-li-go XVIII U-di-sla-va in-tel-le-xis-set kra-ljev-stvo} \hyphenation{in-ua-se-runt quod pes-si-mo pe-dan-tno-an-ti-kvar-ski pas-cha-les} \hyphenation{Pa-lu-šom MDCCXLVIII} % No extra spacing after periods \frenchspacing % Font sizing \fontsize{11}{13.2} \selectfont % veća čitljivost \linespread{1.1} %uvlaka paragrafa \setlength{\parindent}{7mm} % počinje na s. 5 \setcounter{page}{169} \title{\Large{\MakeUppercase{Govor za Pietra Riarija (1474)}}} \date{}%Datum ove verzije: \today} \maketitle %\texttt{Završna verzija, uredniku na čitanje.} \thispagestyle{empty} % [ova stranica namjerno prazna] \section{Uvodne napomene} Latinski nadgrobni govor modruškog biskupa Nikole za kardinala ja (preminulog 5. siječnja, sahranjenog 18. siječnja 1474. u Rimu) igrom je slučaja prva poznata nam tiskana knjiga nekog hrvatskog autora, svojevrstan latinski pandan devet godina kasnije tiskanog glagoljskog \textit{Misala po zakonu Rimskoga dvora.} Autograf djela nije sačuvan. Priređujući kritičko izdanje i prvi hrvatski prijevod Nikolina govora,\footnote{Znanstveno je izdanje, kako ga je priredio autor ove studije, dostupno od 2009.\ u digitalnoj zbirci \textit{\textlatin{Croatiae auctores Latini}}; prema tom izdanju govor je nedavno preveden na engleski: i , »Princely Ambiguity: A Translation of Nikolaus of Modruš’ Funeral Oration for Cardinal Pietro Riario: Oratio in funere Petri Cardinalis Sancti Sixti (1474)«, \textit{Royal Studies Journal,} 5 (2018), 2, str.~92–128.} željeli smo uspostaviti najbolji mogući tekst, omogućiti uvid u tekstualne inačice trinaest svjedoka predaje, te djelo učiniti današnjem čitaocu jezično i stvarno što razumljivijim. Boljem razumijevanju namijenjen je i ovaj uvod. Predstavit ćemo najprije povijesni kontekst samog govora za Riarija i njegovih izdanja, potom retorički zadatak govora, tematsku strukturu, žanrovsku pripadnost i stilska obilježja djela, da bismo na koncu prikazali svjedoke predaje teksta, njihov međusobni odnos i načela ovog izdanja. \section{Povijesni kontekst} Kad je preminuo, dvadesetdevetogodišnji (1445–1474), franjevac i kardinal Sv. Siksta, nećak pape Siksta IV. (; papa 1471-1484), bio je prelat izniman po dobi, bogatstvu i moći. Odigrao je važnu ulogu već pri izboru za papu, a tijekom prve četiri godine Sikstova pontifikata upravo je osmišljavao i vodio papinu vanjsku politiku. U službi te politike bili su i Riarijeva raskoš i rasipnost, spektakularnost i sponzoriranje umjetnosti (njegov je dvor brojio petsto uglednika; u četiri je godine potrošio 300.000 dukata, ostavio 60.000 duga); njegove su ekstravagancije pokazivale moć Siksta IV. i Crkvene Države kako stranim vladarima i poslanicima, tako i pred samim žiteljima Rima. \section{Tematska i retorička struktura govora} Nikolin govor za Riarija ima jasnu i jednostavnu strukturu (\textit{\textlatin{dispositio}} antičke retorike): uvod, glavni dio i završetak. Ključ za razumijevanje strukture daje sam govornik. Već u sredini prve rečenice (MR 1, prema oznakama odlomaka u kritičkom izdanju) kao svrhu govora on sugerira veličanje dostignuća i vrlina pokojnog prijatelja: \begin{quote} \begin{latin} quo extremum amici munus rebus ab eo bene gestis uirtutumque ipsius copia ac splendore amplissimis laudibus exornarent \end{latin} uveličati posljednju počast prijatelju opsežnom pohvalom njegovih dostignuća te obilja i sjaja njegovih vrlina \end{quote} U posljednjim rečenicama MR 3, pak, govornik najavljuje kojih će točno šest duhovnih vrlina preminulog kardinala biti predmet hvale. \begin{quote} \begin{latin} habeo tamen et alias immortales ac propemodum diuinas animi ipsius laudes (ut fortunae corporisque quaelibet ingentia bona tamquam aliena relinquam): pietatem, magnitudinem animi, munificentiam, prudentiam, modestiam, atque iustitiam; quae quales in eo fuerint, breuiter explicare conabor. \end{latin} ipak znam i druge besmrtne i gotovo božanske odlike njegova duha (da i ne spominjem dobra zadobivena srećom ili tjelesnim sposobnostima; ma kako ona bila silna, ipak su u stanovitoj mjeri tuđa): pobožnost, veličanstvenost duha, darežljivost, razboritost, skromnost i pravednost. Kakve su one bile u pokojniku, pokušat ću kratko izložiti. \end{quote} Strukturu tematski raščlanjujemo navodeći brojeve odlomaka u kritičkom izdanju (uz siglu MR) i dodatno citirajući početke i krajeve pojedinih cjelina. Napomene u zagradama standardni su retorički termini za pojedini dio govora ili pojedine ključne riječi. \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[1. MR 1–2 Uvod \textlatin{(exordium, prooimion)}] Potreba utjehe i veličina boli: \textlatin{Cum in omni funebri celebratione\dots\ quam ingratitudinis notam subire uerebor.} \item[2. MR 3–22 Pripovjedni dio (narratio)] Riarijev život, vrline, smrt: \textlatin{Dicturus igitur de laudibus reuerendissimi domini\dots\ ad dominum suum confestim euolauit.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.1. MR 3] Domovina i rod: \textlatin{eas laudes, quas uel a parentibus uel a patria\dots\ omnibus futuris seculis non desinent celebrari.} \item[2.2. MR 3] Vrline: \textlatin{Quae quidem tametsi satis grandis eius gloria sit\dots\ quae quales in eo fuerint, breuiter explicare conabor.} \item[2.3. MR 4–5] Dječaštvo: \textlatin{Qua igitur pietate primum erga deum\dots\ aut dicere illi quare sic fecisti.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.3.1. MR 4] Dječaštvo – stupanje u franjevački red i pobožnost \textlatin{(pietas): Qua igitur pietate primum erga deum\dots\ seruatisque pro more religionis rite caerimoniis uestem Christi induit.} \item[2.3.2. MR 4] Dječaštvo – školovanje i mudrost \textlatin{(prudentia): Qua assumpta ita omnia tyrocinii rudimenta libens promptusque\dots\ ita memoriter recitabat ut ea illum heri aut nudiustertius memoriae mandasse existimares.} \item[2.3.3. MR 5] Dječaštvo – dolazak u Rim i proročka moć \textlatin{(divinitas)}; Francesco della Rovere postaje papa Siksto; komentar o božjoj providnosti: \textlatin{Perfectis igitur quam celerrime omnium bonarum artium studiis\dots\ aut dicere illi quare sic fecisti.} \end{description} \item[2.4. MR 6–11] Kardinalska čast – dužnosti kardinala \textlatin{(officia): Petrus igitur per hunc modum in principatu constitutus\dots\ in uiros praestantes ac bene meritos officiosum esse.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.4.1. MR 6] Dužnosti – darežljivost \textlatin{(munificentia): Turpe enim et indecorum merito ducebat\dots\ culpam magis dolens quam damnum.} \item[2.4.2. MR 7] Dužnosti – ustrajnost \textlatin{(perseverantia): A suscepto semel negotio nullo metu\dots\ magnanimitatis eius testimonia.} \item[2.4.3. MR 7] Dužnosti – milostivost \textlatin{(clementia): Sed illud mea sententia uincit uniuersa\dots\ ut inimicis suis benefacere gauderet.} \item[2.4.4. MR 8] Dužnosti – iskrenost \textlatin{(sinceritas): Nihil in se fictum, nihil subdolum\dots\ sed integritate uiros superare.} \item[2.4.5. MR 8–9] Dužnosti – kardinalski dvor \textlatin{(familia)} i darežljivost \textlatin{(munificentia et liberalitas); komentar o dužnostima kardinala: Sane munificentiae liberalitatisque eius\dots\ bonorum morum gratia diuersari licet.} \item[2.4.6. MR 10] Dužnosti – primanje darova: \textlatin{Accipiebat praeterea munera, non auaritiae\dots\ multa insuper dona tulerunt.} \item[2.4.7. MR 11] Dužnosti – optužbe za simoniju: \textlatin{Vidi illum quodam uesperi non sine graui stomacho\dots\ in uiros praestantes ac bene meritos officiosum esse.} \end{description} \item[2.5. MR 12] Patetični ekskurs protiv zavidnika: \textlatin{Vbi nunc sunt rubiginosa illa maliuolorum pectora\dots\ tota die concinnabant dolos.} \item[2.6. MR 12–13] Povratak na glavnu temu (hipostrofa): \textlatin{Sed haec illi uiderunt\dots\ magna ex parte potuit esse manifesta.} \item[2.7. MR 13–14] Mudrost (prudentia) na dužnosti: \textlatin{Quamobrem tantum eam partem attingam\dots\ ut socium crederes, non dominum.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.7.1. MR 13–14] Mudrost u politici (diplomatska misija u Italiji, prihvaćanje »uresa«, zalaganje za mir u Italiji): \textlatin{Ostendit id nouissima haec ipsius legatio\dots\ ni eum nobis haec dira atque crudelis mors tam repente praeripuisset.} \item[2.7.2. MR 14] Zauzetost za opće dobro \textlatin{(omnium salus): Vincebat ingenio humana consilia\dots\ ut socium crederes, non dominum.} \end{description} \item[2.8. MR 15–16] Umjerenost \textlatin{(moderatio): Tanta uero animi moderatione erat\dots\ urbem illam magis ecclesiae quam fratris gratia uoluit uendicare.} \item[2.9. MR 17] Pravičnost \textlatin{(iustitia): Porro iustitiam ipsius ex illo spectare licet\dots\ pro sua benignitate dissoluat.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.9.1. MR 17] Pravičnost pri povratu grada Imole: \textlatin{Porro iustitiam ipsius ex illo spectare licet\dots\ neque Imolam nisi eo uolente redimere uoluit.} \item[2.9.2. MR 17] Pravičnost u ophođenju s dužnosnicima: \textlatin{Magistratus hortabatur ius suum absque ullo respectu\dots\ non nisi iustum honestumque fieri permaxime uelle.} \item[2.9.3. MR 17] Pravičnost u izdavanju isprava: \textlatin{Hinc et rescripta non nisi sanctissima\dots\ apud notarios exstare iusserat.} \item[2.9.4. MR 17] Pravičnost u sastavljanju oporuke: \textlatin{Huius praeclarissimae uirtutis nec moriens\dots\ pro sua benignitate dissoluat.} \end{description} \item[2.10. MR 18] Izostavljanje \textlatin{(praetermissio): Cogor hoc loco potissimas eius praetermittere\dots\ felicem meritorum ipsius copiam pauca dicendo uitiare.} \item[2.11. MR 18–19] Pobožnost kao posljednja spomenuta i posebna vrlina \textlatin{(pietas): Illud unum dicam, cunctis, qui eius consuetudinem nouerunt\dots\ omnium scientiarum libris egregie refertam.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.11.1. MR 18] Pobožnost prema papi: \textlatin{Illud unum dicam, cunctis, qui eius consuetudinem nouerunt\dots\ nunc mortui desiderio adeo moueatur.} \item[2.11.2. MR 19] Pobožnost pri obnavljanju i opremanju crkava i samostana: \textlatin{Finem dicendi faciam si prius illud summum eius pietatis munus\dots\ omnium scientiarum libris egregie refertam.} \end{description} \item[2.12. MR 19–22] Smrt: \textlatin{Sed dei uoluntate nobis tam repente ademptus est\dots\ ad dominum suum confestim euolauit.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[2.12.1. MR 20] Neustrašivost i strpljivost u bolesti \textlatin{(patientia): Cuius miserationis dilectionisque certa indicia\dots\ dolores mira patientia pertulit.} \item[2.12.2. MR 20] Ispovijest i pokajanje: \textlatin{Delicta, quae uel aetatis uel fortunae\dots\ diuinam uoluntatem accinctus praestolabatur.} \item[2.12.3. MR 20–21] Govor pred smrt \textlatin{(novissima verba): Iamque uicinus morti\dots\ uel meo exemplo discite.} \item[2.12.4. MR 22] Opis smrtnog časa: \textlatin{His atque aliis huiusmodi plerisque summa cum religione\dots\ ad dominum suum confestim euolauit.} \end{description} \end{description} \item[3. MR 23 Završetak (peroratio)] pohvala i utjeha: \textlatin{O felix atque iterum felix\dots\ sit nomen domini benedictum. Amen.} \end{description} \bigskip Iz ovog se prikaza vidi Nikolina glavna ideja: upravo su vrline Riariju omogućile da uspješno izvršava dužnosti kardinala. No, pripovjedni dio govora Nikola ne organizira ne samo tematski, po vrlinama, nego i po još jednom načelu, kronološkom, prateći Riarijev put od njegova porijekla (MR 3) do smrtnog časa (MR 22). Govor tako ima dvije usporedne okosnice, životopis i vrline; autor ih vješto isprepleće te jedno postaje dokaz drugome. \section{Retorički zadatak} Povijesne okolnosti jasno sugeriraju kakav je izazovan zadatak imao kao autor nadgrobnog govora. Trebalo je hvaliti jednog od najistaknutijih, ali i najkontroverznijih dužnosnika Papinske Države, usto i papina nećaka (o čijem su načinu života, pa i o porijeklu, kolale najrazličitije glasine); pritom je trebalo izbjeći i opasnost laskanja i izlaganje optužbi za neistinu. Napokon, govor je morao biti ne samo stilski prihvatljiv, nego i primjeren liku onoga koga hvali. Trebalo je, dakle, ostvariti uzoran prikaz kardinala.% Dodatno, iako govor (najvjerojatnije) nije održan, trebalo ga je jasno smjestiti na mjesto i u vrijeme Riarijeva pogreba. Koliko je retorička strategija koju je odabrao Nikola bila u skladu s razmišljanjima renesanse, pokazuje usporedba govora za Riarija s uputama za hvaljenje koje u djelu \textit{De ratione scribendi} (nastalo vjerojatno oko 1485, prvi put tiskano 1549. u Baselu) daje mlađi suvremenik Modruškoga Aurelio ini (oko 1454. – 1497), i sam vješt epideiktičkom govorništvu i, od oko 1480, blizak papinskoj kuriji: \begin{quote} \begin{latin} Defuncti ita mihi laudandi esse uidentur ut consolari magis eos apud quos dicimus quam illos laudare uelle uideamur. Quod contra faciunt plerique eorum qui nostro tempore oratores haberi uolunt. Ita enim ad laudandum accedunt ut aperte se laudatores atque assentatores esse fateantur. Ita uero laudant, non ut orationem habere, sed ut eius uiri recitare ab se scriptam historiam existimentur. Mihi uero funebrium orationum ea seruanda ratio uidetur: primum ut iustissimum esse eorum apud quos dicimus dolorem nostrumque cum illis communem esse ostendamus. Deinde ut eos illorum qui defuncti sunt uirtutibus, rebus gestis, gloria, felicitate, et memoria consolemur. (\dots) Laudantur tamen non immerito ii, qui, quum ignoti obscurique sint, summas tum opes, tum dignitates sunt uirtutibus et rebus gestis consecuti. (\dots) Illa uero, de qua ueteres scribunt, uirtutum quatuor in partes distributio, in excogitando quidem apud animum semper est, meo iudicio, proponenda, in scribendo aut dicendo non semper explicanda. Omnes enim uirtutes in omnibus commemorare si uelimus, irridere eos potius, quam laudare uideamur. In milite enim religionem, in sacerdote bellicam disciplinam laudare non conuenit. (\dots) In iustitiae partibus quaedam nostra aetate laudantur quae ueteres non commemorabant. Et obseruantia in Deum ac religionem nostram, et obedientia, in sacerdotibus praesertim, ac religiosis uiris magnam, si assit, laudem affert: si absit, summam uituperationem. Quid pars illa liberalitatis, quae egenis succurrit, nobis ex diuina institutione tradita, quam nos recepto Graeco uocabulo eleemosynam appellamus, quantam nunc hominibus laudem affert, quae ueteribus pene incognita fuerat? Clementia erga inimicos, quae nobis item ex diuino instituto seruanda est, si assit, magna laude digna erit: quod praecipue in Stephano laudamus martyre. Si absit, sacerdotibus praesertim, uitio dari poterit. Patientiam quoque religio nostra laudabiliorem atque illustriorem facit. Castitas apud antiquos tantum in mulieribus laudabatur: nunc uero etiam in uiris ita ut intactum se ab omni coitu mundumque seruasse et ut quidem nunc loquimur uirginem esse, quod de Ioanne apostolo scribitur, summae laudi ac sanctimoniae tribuatur.\footnote{, \textit{ De ratione scribendi libri tres, numquam antea in lucem editi\dots}, Basel 1549, »De laudandis hominibus. Cap. IX.«, s. 106–110.} \end{latin} Pokojnike treba, smatram, hvaliti tako da više pokažemo da tješimo one pred kojima govorimo nego da želimo hvaliti same preminule. Suprotno postupa većina onih koji u naše doba žele da ih smatraju govornicima. Hvaljenju pristupaju tako da otvoreno izjavljuju da su hvalitelji i povlađivači. A hvale tako da se ne čini da drže govor, nego da čitaju povjesnicu tog čovjeka koju su sami napisali. No, po meni za nadgrobne govore vrijedi sljedeće načelo: prvo, pokazati da je bol onih pred kojima govorimo posve opravdan, i da ga s njima dijelimo. Potom, utješimo ih vrlinama preminulih, njihovim podvizima, slavom, srećom, i uspomenom. (\dots) Ipak, opravdano je hvaliti one koji su, mada su bili neznatna i tamna porijekla, postigli velika bogatstva i časti svojim vrlinama i junačkim djelima. (\dots) A ona podjela vrlina na četiri dijela, o kojoj pišu stari, treba, po mom sudu, pri planiranju uvijek biti pred očima, ali pri pisanju ili govorenju ne treba je uvijek poštivati. Ako bismo, naime, htjeli kod svakoga spomenuti sve vrline, činilo bi se prije da ih ismijavamo, a ne da ih hvalimo. Kod vojnika ne priliči hvaliti pobožnost, kao ni kod svećenika ratnu vještinu. (\dots) Kod nas poštivanje Boga i vjere, i pokornost, osobito kod svećenika i duhovnika, donosi veliku slavu, ako ih ima; ako ih nema, slijedi najveća pokuda. A tek ona darežljivost kojom se pomaže potrebitima, kojoj nas je poučio božanski nauk, koju preuzevši grčku riječ zovemo \textit{eleemosyna}, koliku ona sada ljudima donosi slavu, a starima je bila gotovo nepoznata? Milosrđe prema neprijateljima, koje također moramo pokazivati prema božanskom nauku, bit će dostojno velike hvale, ima li ga; posebno ga hvalimo kod Stjepana prvomučenika. Ako ga nema, to se može smatrati porokom, osobito kod svećenika. I strpljivost je naša vjera učinila hvalevrednijom i uglednijom. Tjelesna se čistoća kod starih hvalila samo kod žena, a sada vrijedi i kod muškaraca, tako da se najvećom slavom i bogobojaznošću smatraju uzdržavanje od snošaja i čuvanje čistoće i, kako se barem sada to naziva, djevičanstvo, o čemu piše kod Ivana apostola. \end{quote} I Nikola počinje govor kao tješitelj, pokazujući vlastitu bol, bol okupljenih i samoga Siksta, a govor strukturira oko Riarijevih vrlina (pritom vrijedi primijetiti da od vrlina koje spominje Brandolini Nikola izostavlja samo \textit{castitas}, dok je \textit{liberalitas} prikazana isključivo u odnosu na članove kućanstva, klijente i crkve), prateći istovremeno put Riarija – i njegova ujaka – od skromnih početaka do najviših časti. Dojam da pripovijeda Riarijevu »povjesnicu« Nikola izbjegava raznolikošću tona i izlaganja: anegdotama (uključujući i one ispričane u prvom licu), konkretnim primjerima, komentarima i patetičnim ekskursima. Među ekskursima se posebno ističe lirsko-biblijski napad na »zavidnike i zlobnike« koji su Riarija optuživali za prodaju kardinalskih časti (MR 12). On svjedoči o daljnjoj Nikolinoj retoričkoj strategiji. Pohvala je, naime, izrazitija kad je suprotstavljena pokudi; ali kako kuditi u pohvalnom govoru? Nikola pokudu rezervira za Riarijeve \textit{kritičare}, za sve one koji su širili (dakako, lažne i zlobne) neistine o moćnom papinu nećaku. I ostale aspekte Riarijeva života koji su dali povoda kritikama Nikola spominje, ali ih reinterpretira pozitivno. Rodbinska povezanost s budućim papom, Riarijeva mlada dob, njegova uloga u izboru za papu protumačeni su u Nikolinu govoru kao čudesni događaji u službi Božjeg nauma (MR 4-5). Raskoš i sjaj Riarijeva dvora (zgrade, gozbe, namještaj) Nikola vidi kao obavezu prema stranim »visokim gostima«, koji su, uostalom, sami nudili sredstva papi i njegovoj kuriji (MR 6). Kao dio dužnosti kardinala i očitovanje Riarijevih vrlina prikazani su i dvor od petsto ljudi (MR 9) i primanje darova (MR 10). Na optužbe za lijenost odgovaraju čak dva prikaza Riarijeva radnog dana (MR 14, 16). Govornik priznaje Riarijevo zalaganje za brata, ali u interesu Crkve (MR 16), i da je kardinal intervenirao u korist »većeg broja osoba«, ali pritom nije dopuštao ni pomisao na djelovanje mimo zakona i pravde (MR 17). \section{Žanrovska pripadnost} Nikolino djelo možemo razmatrati iz najmanje pet žanrovskih aspekata.\footnote{Žanrovski i književnopovijesni kontekst uspostavljamo prema O'Malley, . \textit{Praise and Blame in Renaissance Rome: Rhetoric, Doctrine, and Reform in the Sacred Orators of the Papal Court, C. 1450-1521.} : Duke University Press, 1979. te , »The Ideal Renaissance Pope: Funeral Oratory from the Papal Court«, \textit{Archivum Historiae Pontificiae}, 14 (1976), 9-70.} Najprije, prema antičkoj podjeli vrsta govora (na sudske, političke, pohvalne), riječ je o pohvalnom, odnosno epideiktičkom govoru, čiji je cilj hvaljenjem djela i vrlina izazvati u publici divljenje, pa i želju da se ta djela i vrline nasljeduju. Situacijski, budući da je osoba koju govor hvali preminula, radi se o nadgrobnom govoru; dodatno ga određuju činjenice da je smješten u renesansni crkveni kontekst – pripada, dakle, duhovnoj, a ne svjetovnoj retorici – i to u kontekst samog vrha katoličke hijerarhije, kao nastup pred papom i članovima Rimske kurije, a u počast preminulom kardinalu.\footnote{Primjerom pokazujući što čini dobra kardinala, »kneza Crkve«, Nikola anticipira traktat učenika Pomponija Leta i službenika Rimske kurije Paola Cortesija (Rim, 1465. – San Gimignano, 1510) \textit{De cardinalatu (O kardinalskoj časti}, djelo posmrtno objavljeno 1510).} Prema sačuvanim govorima i propovijedima pred papom O'Malley i McManamon rekonstruirali su retoričke trendove na papinskom dvoru, smještajući na jedan kraj dijapazona skolastici bliske tzv. tematske propovijedi (koje polaze od navoda Svetog pisma te ga raščlanjuju i tumače dio po dio, pri čemu im je glavni cilj poučavanje). Na suprotnoj je strani humanističko epideiktičko govorništvo, skup estetskih i poetičkih načela čiji se utjecaj u književnom stvaranju pri Svetoj stolici opaža od pedesetih godina Quattrocenta i pape Nikole V. Humanistički govori na publiku djeluju prvenstveno umjetničkim oblikovanjem, izazivajući jak dojam i užitak – \textit{movere} i \textit{delectare} antičke retorike. Već pregled sadržaja pokazuje da Nikolino djelo nije oblikovano kao tematska propovijed; radi se o humanističkom ostvarenju. \section{Stil} %Nikolin nadgrobni govor za Riarija bio je, sudeći po broju izdanja i prijepisa, popularno djelo. Popularnost jednim dijelom mora da duguje temi, preranoj smrti moćnog i slavnog kardinala, papina nećaka, čija je kratka i ekstravagantna karijera dala povoda kritikama i ogovaranjima. Dio razloga za uspjeh Nikolina nadgrobnog govora morao je, osim u predmetu i temi, biti i u samom oblikovanju. Delikatan zadatak pohvale kontroverzne osobe Nikola je, po svemu sudeći, obavio na zadovoljavajuć način. Zato vrijedi, čitajući ovaj govor, razmišljati što je u njemu renesansna publika smatrala zadovoljavajućim, u čemu su čitaoci nalazili spomenuto \textit{movere} i \textit{delectare}. Ovdje ćemo iznijeti nekoliko poticaja za takvo stilističko čitanje Nikolina govora. Pritom ne pretendiramo na iscrpnost i ne priređujemo potpun katalog retoričkih i stilskih postupaka; donosimo tek primjere posebno uočljivih strategija. \subsection{Gradnja rečenice: concinnitas i variranje} Ciceronovski termin \textit{concinnitas} označava sintaktički simetrično oblikovanje koordiniranih (paralelnih) rečenica ili dijelova rečenice.\footnote{Leumann s. 813, §~49.} Simetrija i koordinacija mogu se unutar rečenice ostvarivati na različite načine, a dužina koordiniranih članaka ne mora biti identična. Kako uspostavlja simetričnost i varira odnose i dužine članaka i rečenica, pokazat ćemo najprije na mjestu koje nije posebno stilski obilježeno (mada iznosi dojmljiv podatak o broju članova Riarijeva kućanstva) – na kraju MR 8. Autor komentira Riarijev duhovit odgovor i hvali njegovu darežljivost (liberalitas), te prelazi na sljedeću temu (riječi i izraze koji ostvaruju simetriju istakli smo kurzivom i masnim tiskom). \begin{quote} \begin{latin} O praeclaram \textit{uocem}, \\ o \textit{sententiam} summo principe dignam; \\ \textit{nec impunitatem} \textbf{erratorum} laudauit, \\ \textit{nec liberalitatem suam} ullis \textbf{male merentium} factis occludi passus est. \textbf{Felices}\\ \textit{quibus} illa perfrui licuit, \\ et nunc omnium \textbf{infortunatissimos} \\ \textit{quibus} tam crudeli fato erepta est! % quĭbūs tām crūdēlī fāt[o] ērēpt[a] ēst \textit{Quingentos} ferme pascebat \textit{familiares}, \\ partim \textbf{illustri}, \\ partim \textbf{nobili}, \\ omnes \textbf{honesto} loco natos; \\ praelatos, milites, doctores, oratores, poetas, \\ aut alicui alii \textit{honestae} arti deditos; \\ \textit{nullis} tantorum \textbf{sumptibus}, \\ \textit{nullis} grauabatur \textbf{impensis}. % nūllīs grăvābātŭr īmpēnsīs (kretik + spondej?) \end{latin} \end{quote} Prva se rečenica sastoji od dva para simetričnih članaka. U prvom su paru uzvici; njihovi su atributi i imenice postavljeni hijastički (a \textit{a, b} b); sadržajno, drugi član para ne donosi ništa novo, ista se misao izriče drugim riječima. Drugi je par koordiniran veznicima \textlatin{(nec\dots\ nec\dots).} Položaj imenica i atributa u člancima je istovjetan, pri čemu je paralelizam imenica dodatno naglašen homeoteleutom \textlatin{(impun\textit{itatem}, liberal\textit{itatem}),}\footnote{I drugi homeoteleuti u ovom govoru javljaju se na mjestima koja su stilski višestruko obilježena: \textlatin{cuius suauissimam consuetudinem, comi\textit{tatem,} benigni\textit{tatem,} liberalitatemque quotidie experie\textit{bamini,} cuius ingenii dexteritatem et incredibilem consilii prudentiam indies magis admira\textit{bamini}} (MR 2); \textlatin{ut coepta eius declarant aedific\textit{ia,} totque magnificentissimo cultu celebrata conuiu\textit{ia,} et supellex imperialibus fastibus digna} (MR 6, prva dva članka trikolona); \textlatin{Vnde sua illi negotia cred\textit{ebant} et rerum omnium summam fidei ipsius ultro committ\textit{ebant}} (MR 13, uz sinonimiju); \textlatin{nemini uim attul\textit{it,} nullum uiolenter oppress\textit{it}} (MR, 17 opet sinonimi); \textlatin{Haec quoque sacra apostolorum aedes beneficentiam eius testari potu\textit{isset} si tantum quattuor mensibus superstitem uid\textit{isset}} (MR 19).} a drugi je članak duži od prvoga.\footnote{U skladu sa »zakonom dužih članaka«, \textit{Gesetz der wachsenden Glieder,} kako ga je 1923. – 1932. formulirao lingvist : »od dva rečenična članka, kraći će, po mogućnosti, prethoditi dužemu«; Leumann II.16, s. 722-724.} I druga rečenica donosi par akuzativa uzvika, antitetički suprotstavljenih \textlatin{(felices, infortunatissimos);} struktura je paralelna – svaki akuzativ otvara mjesto odnosnoj rečenici \textlatin{(quibus\dots\ quibus\dots).}\footnote{Uzvik sličan ovom, pritom kombiniran s hijazmom i antitezom \textlatin{(uita – mors)}, nalazimo i na kraju prikaza Riarijeve smrti, MR 23: \textlatin{O felix atque iterum felix cui et uita summam \textit{dedit} \textbf{gloriam} et mors ipsa meritam \textbf{diuinitatem} \textit{non denegauit!}}} Druga je polovica rečenice i ovdje duža od prve. Kako smo već mogli naslutiti iz ponavljanja uzvika, korespondencija postoji i među čitavim rečenicama; ojačava je okolnost da svaka rečenica ima po dva predikata, da u svakoj najprije dolazi aktivni, potom pasivni, odnosno deponentni glagolski oblik \textlatin{(laudauit – passus est; licuit – erepta est).} Trećom rečenicom počinje nova tema. Fantastičan broj članova Riarijeva kućanstva dodatno je istaknut hiperbatom – broj je odvojen od imenice koju opisuje. Slijedi trikolon pridjeva modificiranih prilozima, a svi su ovisni o istoj imenici (zeugma); dok su prva dva članka brojem slogova jednaki, treći je članak duži; struktura je, dakle, 2{}\verb!+!1. Srednji dio rečenice jest asindetsko nabrajanje zanimanja, rezimirano opisom čiji particip uspostavlja paralelu s participom s kraja trikolona \textlatin{(loco natos\dots\ arti deditos);} pridjev \textit{honestus} ponavlja se kao ključan termin. Rečenicu zaključuje par kontaktnih sinonima \textlatin{(sumptibus, impensis)} asindetski koordiniranih niječnim atributom \textlatin{(nullis,} anaforično);\footnote{Anafora s asindetom kao sredstvo uspostave paralelizma susreće se i drugdje u govoru: \textlatin{\textit{omnibus} carus, \textit{omnibus} dilectus esse coepit (MR 4, sinonimi); \textit{coepit} auunculum suum hortari, \textit{coepit} importunius compellere (MR 5, sinonimi); \textit{nullo} metu absterreri potuerat, \textit{nullo} periculo depelli (MR 7, sinonimi); \textit{hoc} ipsum et senatori Vrbis interrogatus respondit, \textit{hoc} gubernatoribus, \textit{hoc} praefectis, \textit{hoc} omnibus legationis suae iudicibus saepe mandabat (MR 17).}} atribut i imenice u oba članka ostvaruju hiperbat. I ovdje je par značenjski redundantan, svrha mu je retorička, estetska, ekspresivna (ističe se Riarijeva spremnost na trošenje). Kako bismo pokazali da se ovako građene rečenice susreću u svim dijelovima govora, analizirat ćemo još jedan, gotovo nasumce odabran odlomak (s početka MR 18, gdje autor prekida nabrajanje vrlina da bi ipak istaknuo još odanost papi kao najvažniju Riarijevu odliku). \begin{quote} \begin{latin} Cogor hoc loco potissimas eius praetermittere laudes, \\ \textit{cum} temporis \textbf{exclusus} \textit{angustia,} \\ \textit{tum} ipsarum rerum \textit{multitudine} \textbf{superatus}. \textit{Qualem} se \textit{erga} amicos, \\ \textit{qualem erga} parentes, \\ et praesertim \textit{qualem erga} ipsum summum pontificem gesserit, \\ \textbf{malo} haec omnia \textit{in aliud tempus differre} \\ \textbf{quam adeo} felicem meritorum ipsius copiam \textit{pauca dicendo uitiare.} Illud unum dicam, \\ cunctis, \\ qui eius consuetudinem nouerunt, \\ attestantibus, \\ \textit{nullum} fuisse \textbf{tam piissimum} filium, \\ \textit{nullum} \textbf{adeo parenti deditum}, \\ aut \textbf{cui} maior et salutis et honoris \textit{genitoris sui} cura fuit, \\ quam \textbf{huic uni} \textit{summi pontificis nostri} \\ \tabto{2em} \textbf{a prima} eius familiaritatis die \\ \tabto{2em} \textbf{usque ad ultimum} uitae exitum; \\ \textit{nullos} pro eo \textit{recusauit} subire \textbf{labores}, \\ \textit{nulla} \textbf{pericula} \textit{deuitauit;} \\ laborantem, aegrotantem, peregrinantem \textit{nunquam deseruit,} \\ \textit{nunquam} ab officio \textit{decessit,} \\ \textit{semperque,} ut datus a domino Tobiae angelus, \\ \textit{lateri haesit;} \\ aduersa \textit{procurauit,} \\ \textit{accersiuit} prospera; \\ pia sedulitate fouit, coluit, ueneratus est, \\ ut nemini mirum uideri debeat \\ si \textbf{aut uiuentem} \textit{tantum dilexit} \\ \textbf{aut} nunc \textbf{mortui} desiderio \textit{adeo moueatur.} \end{latin} \end{quote} U prvoj od citiranih rečenica opet zapažamo hiperbat \textlatin{(potissimas eius\dots\ laudes)}, koordinaciju \textlatin{(cum\dots\ tum\dots)} te hijazam imenica i participa koji ih opisuju \textlatin{(exclusus angustia – multitudine superatus).} Sljedeća je rečenica duža, i počinje trikolonom \textlatin{(qualem se erga\dots\ qualem erga\dots\ qualem erga\dots),} opet organiziranim u simetričan par i treći, duži članak (2{}\verb!+!1). Drugi je dio rečenice antiteza \textlatin{(malo\dots\ quam\dots)}, a završava sintaktički identičnim konstrukcijama, infinitivima \textlatin{(differre – uitiare)} kojima prethode njihove priložne oznake. Posljednji članak rečenice obogaćen je dodatnom antitezom \textlatin{(felicem meritorum\dots\ copiam – pauca dicendo).} Trikolon organiziran po načelu 2{}\verb!+!1 susrećemo i u trećoj, sintaktički najrazvijenijoj rečenici. Nakon najave \textlatin{(illud unum dicam\dots),} prva su dva članka koordinirana ponavljanjem uz varijaciju \textlatin{(nullum\dots\ tam – nullum adeo).} Treći, kao kulminacija, opet je, sa svoje strane, sastavljen od dva članka \textlatin{(aut cui maior\dots\ – quam huic uni\dots);} drugi od tih članaka sadrži i antitezu »od prvog do posljednjeg dana života«. U drugoj polovici rečenice na parove se nadovezuju trodjelni članci. Nakon asindetskog para \textlatin{nullos\dots\ – nulla\dots,} s hijazmom \textlatin{recusauit\dots\ labores – pericula deuitauit,} slijedi tročlani asindet participa \textlatin{laborantem, aegrotantem, peregrinantem.} Sljedeći je trikolon oblikovan u sve duže i duže članke \textlatin{nunquam deseruit – nunquam\dots\ decessit – semperque\dots\ lateri haesit} (također struktura 2{}\verb!+!1). Zatim se javlja vrlo dojmljiv par u hijazmu \textlatin{aduersa procurauit, accersiuit prospera,} a nakon njega opet asindetski trikolon glagola bliskih značenja \textlatin{fouit, coluit, ueneratus est,} sličan onom ranijem asindetu (s participima). Kao i prva polovica rečenice, i ova završava parom koji je istovremeno i antiteza \textlatin{aut uiuentem tantum dilexit – aut nunc mortui desiderio adeo moueatur.} Izmjenjivanje dužih i kraćih rečenica te variranje jezičnih sredstava koja uspostavljaju paralelizam osnovne su autorove strategije u čitavom djelu; smatramo ih temeljem elegancije nadgrobnog govora za Riarija. \subsection{Stilski posebno dojmljiva mjesta} TBA \subsection{Hijazmi} TBA \subsection{Predikat na početku} TBA \subsection{Slikovitost, mudre misli, deiksa} TBA \subsection{Razine intertekstualnosti} citati: Biblija, ostali neidentificirani početak kao Ciceron dokaz utemeljenosti u živoj tradiciji pisanja na latinskom \section{Predaja teksta} Nikolin govor za imao je sedam tiskanih izdanja i šest prijepisa. Zbog teme djela i poznatih podataka o aktivnosti tiskara, može se pretpostaviti da je \textit{terminus post quem non} tiskanih izdanja, čak i tamo gdje nije naznačen datum, smrt pape Siksta IV, 12. kolovoza 1484.\footnote{Opširniji izvještaj o tiskarima vidi u Neven Jovanović, »Nadgrobni govor og za « \textit{Colloquia Maruliana} XXVII (2018), str.~123–141.} Prikazat ćemo najprije tiskana izdanja, potom rukopise. \subsection{Tiskana izdanja} U kronologiji tiskanja skloni smo razlikovati dvije faze; prva bi bila neposredno povezana s Riarijevom smrću, dok drugu smještamo u godine oko 1482, kada se pojavljuje jedino datirano izdanje, ono padovansko Cerdonisovo. Važno je primijetiti da je tada već pokojan (umro je prije 29. svibnja 1480) te u ovoj fazi nije mogao imati udjela. Sljedeći popis izdanja navodi kronološkim redom, prema datumima potvrđenima u kolofonu ili pretpostavljenima (uglate zagrade označavaju da je godina objavljivanja pretpostavljena), s mjestima i tiskarima. Izdanja označavamo siglom korištenom u kolaciji i kritičkom aparatu. Oznaka iza GW upućuje na zapis u računalnoj bazi podataka \textit{Gesamtkatalog der Wiegendrucke}.\footnote{Staatsbibliothek zu Berlin, Gesamtkatalog der Wiegendrucke / Inkunabelsammlung, www.gesamtkatalogderwiegendrucke.de/. Pristupljeno 29. studenoga 2019.} \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[V\hphantom{e}] GW M26710, Rim [1474], In domo Antonii et Raphaelis de Vulterris (tiskara djelovala 1472–1474) \item[V1] GW M26711, Rim [1474], In domo Antonii et Raphaelis de Vulterris\footnote{ ustanovio je 1976, uspoređujući prijelom kolofona i posljednjeg lista primjerka koji se čuva u Trevisu s onim opisanim u katalogu British Library, da postoje dva različita izdanja: »Primjerak iz Trevisa također je proizašao iz oficine Antonii et Raphaelis Vulterris i ono je stvarno drugo izdanje, jer je očito, da tako skladan prelom djela mogao je nastati tek nakon već otisnutog izdanja, a koje je poslužilo kao predložak«. , »Dvije značajne hrvatske knjižice«, \textit{Hrvatska revija: Jubilarni zbornik 1951–1975}, Knjižnica Hrvatske revije, Barcelona, 1976, 590–598 i , »Zwei unbekannte Inkunabelausgaben«, \textit{Beiträge zur Inkunabelkunde} 3 (1983), 8, 91–93. Bošnjakova je identifikacija prihvaćena u \textit{Gesamtkatalog der Wiegendrucke}.} \item[Ge] GW M26707, Rim [1474], (djelovao 1473–1474) \item[R\hphantom{o}] GW M26712, Rostock [1476], ad S.~Michaelem (tiskara djelovala 1476–1531) \item[C\hphantom{d}] GW M26706, Padova 30.~kolovoza 1482, (djelovao 1482-1487) \item[P\hphantom{d}] GW M26709, Rim [1482], (djelovao 1479-1501) \item[Gd] GW M26708, Rim [1482], (djelovao 1475–1488) \end{description} Većini tiskara zajednička je orijentacija na mala, ponajviše prigodna izdanja; u taj se profil Nikolin govor savršeno uklapa. Znatan dio angažiranih poduzeća bio je kratka trajanja; tiskare de Vulterris, Gensbergova i Cerdonisova djelovale su pet godina ili manje. Antonio, jedan od braće Maffei, potvrđeno je bio povezan s moćnom obitelji Riario, baš kao i – možda se prvo izdanje govora za Riarija pojavilo upravo \textit{in domo de Vulterris}.\footnote{ radila su na papinskom dvoru kao pisari; bili su aktivni i kao sakupljači rukopisa i kao izdavači. sudjelovao je u takozvanoj uroti Pazzijevih. Na vezu tiskara i urotnika upozorila je Farenga, n. dj, 214, bilj. 95.} \subsection{Rukopisi} Svih šest danas poznatih rukopisnih prijepisa potječe iz posljednje četvrtine XV.~stoljeća. Četiri se čuvaju u Italiji – u Vatikanu, Rimu, Veneciji i Palermu – dok je po jedan u Njemačkoj (München), odnosno u Češkoj (Olomouc). Svi su prijepisi u humanističkim zbornicima (u kojima prevladavaju prozni sastavci). Svi su pisani humanistikom osim onog u Olomoucu, čije je pismo bastarda (iznimno raširena u Češkoj tijekom XIV. i XV. st). Nijedan prijepis nema bitnih naknadnih ispravaka ili marginalnih bilježaka. Također, prema onome što dosad znamo o ruci škog,\footnote{Luka Špoljarić, »Ex libris Nicolai episcopi Modrussiensis: knjižnica Nikole Modruškoga«, \textit{Colloquia Maruliana} XXI, str.~25–63.} nijedan od prijepisa nije autograf i ne nosi tragove autorskih intervencija. Donosimo osnovne podatke o prijepisima uz sigle kojima ćemo kodekse citirati u kritičkom aparatu. \begin{description}[nolistsep,itemsep=3pt,font=\rmfamily] \item[ve] Venecija, Biblioteca Marciana; Marc. Lat. cl. XIV, 180 (4667), XV. st, ff.~9r–19v. \item[va] Vatikan, Bibliotheca apostolica Vaticana; Vat. lat. 8750; kraj XV. – početak XVI. st; ff.~205r–212v. Nekoliko folija ispred govora Modruškog (ff.~162r–172v) u ovom je kodeksu prepisan i nadgrobni govor (1429–1480) za kardinala Riarija. \item[co] Rim, Accademia Nazionale dei Lincei, Biblioteca dell'Accademia dei Lincei e Corsiniana, fondo principale, cart. misc., XV. st, Corsin. 583 (45 C 18), ff. 117r–123r. Prepisivač je (Pistoia, 1451–1511).\footnote{Identificiran u , »Alcuni codici Corsiniani di mano di Tommaso e «, \textit{Rendiconti dell' Accademia nazionale dei Lincei, classe di scienze morali, storiche e filologiche}, serie 8, XI (1956), str.~252–263.} Prije Nikolina govora, na ff.~113v–116v, prijepis je govora pred Sikstom IV. (1436 – nakon 1513), a nakon Nikolina slijedi govor (1429–1478) pred istim papom, ff.~123v–125v. \item[pa] Palermo, Biblioteca centrale della Regione siciliana, I.B.6, oko 1474–1480, ff.~32r–54r. Datacija u katalogu određena je upravo prema Nikolinu djelu. \item[m] München, , CLM 461, rukom (1440–1514), XV. st; ff.~129r–138r. Kodeks, u kojem najmlađi tekst potječe iz 1492, sadrži još nekoliko govora pred Sikstom IV.\ i govora povodom smrti drugih kardinala; radi se o djelima (1408–1489), Lá (r.~oko 1460), Lodovica da Imola (djelovao 1462–1479) u smrt kardinala (u.~1478), o anonimnom govoru u smrt kardinala Spoleta, te o govoru Gianantonija de San Giorgio (1439–1509) u smrt Ferrija de Clugny, kardinala i biskupa Tournaija (u.~1483). \item[o] Olomouc, na, \textit{Textus uarii; Historia Bohemica,} sign. M I 159, oko 1476–1480, pisan u Litoměřicama;\footnote{́ř, \textit{Počátky humanismu v Cechách}, V Praze: Nákladem České akademie císaře Františka Josefa pro vědy, slovesnost a umění, str.~507.} ff.~170r–174v. \end{description} \subsection{Kolacija svjedoka i stemma codicum} Kolacija tiskanih i rukopisnih svjedoka predaje otkriva podjelu u dvije glavne obitelji. Izvan tih dviju skupina ostaju četiri izvora za koja nismo spremni tvrditi da pripadaju samostalnoj obitelji. Obitelj 1 čine četiri svjedoka. To su izdanja (\textit{Ge}, Rim, nakon 18. siječnja 1474) i (\textit{C}, Padova, 30. kolovoza 1482) te rukopisi \textit{va} i \textit{o}. Oba su rukopisa, čini se, prepisana iz Gensbergova izdanja. I u obitelji 2 četiri su svjedoka: dva izdanja in domo de Vulterris (\textit{V, V1;} Rim, nakon 18. siječnja 1474), te njima bliska izdanja i a (\textit{P, Gd;} oba u Rimu, oko 1482). Rukopis \textit{m} zasigurno je prijepis Plannckova izdanja. Nešto bliže potonjoj obitelji – ali ne bismo se usudili tvrditi da se radi o izravnoj vezi – stoji izdanje i viridis ad (\textit{R}, Rostock, 1474. ili kasnije) te rukopisi \textit{ve, pa} i \textit{co; pa} je bliži \textit{ve} nego \textit{co.} Ovi su međusobni odnosi shematski prikazani kao \textit{stemma codicum} na priloženom dijagramu (slika 1). U nastavku navodimo najvažnije tekstualne inačice na osnovi kojih smo predložili ovakvu podjelu svjedoka. Brojevi odlomaka slijede podjelu u ovom kritičkom izdanju.\footnote{Opširniji popis različitih čitanja vidi u aparatu, te u , »Rukopisi, kolacija svjedoka predaje i paratekstovi nadgrobnog govora og za kardinala (1474)« (rad u postupku ocjenjivanja).} \subsection{Obitelj 1 (V Gd P m)} \begin{enumerate}[label=\alph*)] \item MR 4 \textlatin{inter cenandum de uariis disciplinarum studiis frequenter disserere consueuerat adeo acute adeoque prompte ac subtiliter de quaestione proposita ut eum putares die noctuque nulli adeo alii rei quam euoluendis theologorum philosophorumque libris uacare.} – sva četiri donose sintaktički nelogično »putare« nasuprot »putares« \textit{(R va ve pa co),} odnosno »putaret« \textit{(Ge C o)} \item MR 4 \textlatin{ex perceptis semel principiis difficillima quaeuis uel philosophiae uel theologiae problemata summa cum omnium admiratione absoluebat} – sva četiri ispuštaju »uel theologiae« \item MR 11 \textlatin{seque ab inuidis atque malignis impie ac flagitiose eius criminis insimulari persancte iurabat.} – umjesto »inuidis« \textit{V P m} imaju neobično »infidiis«, \textit{Gd} nešto prihvatljivije »inuidiis« (svi ostali »inuidis«, kao paralelu pridjevu »malignis«) \item MR 17 \textlatin{Et licet nonnumquam, domesticorum amicorumque euictus precibus, aut litteris aut nuntiis multos iudicibus commendaret, id tamen citra cuiusque iniuriam fieri uolebat.} – umjesto »citra«, \textit{V Gd P} imaju »circa«; u \textit{m} je zabilježena i inačica: »circa al. citra« (svi ostali imaju »citra«) \item MR 19 \textlatin{testatur Taruisii maior basilica non paruis ditata uectigalibus} – sva četiri donose neobično »Taruixit« umjesto »Taruisii« \textit{(co} ima »Taruixij«); nadalje, \textit{V Gd P} imaju »non parua« (što je zalihosno), ostali čuvaju litotu čitanjem »non paruis« \item MR 20 \textlatin{Delicta, quae uel aetatis uel fortunae uitio pro fragilitate humana contraxerat, pia confessione saepius diligentiusque purgauit et munitus caelesti uiatico, quod summa cum deuotione acceperat, diuinam uoluntatem accinctus praestolabatur.} – nasuprot svim ostalima, \textit{V Gd P} imaju nezgodno čitanje »minutus« (kao da je umirući »umanjen za nebesku popudbinu«, tj.\ da je ostao bez nje) \item MR 20 \textlatin{non fortunam accusauit nec se in medio iuuentutis flore ex tanto imperio et ex talibus opibus subtrahi uel leuiter indoluit} – sva četiri imaju »operibus«, nasuprot »opibus« kod ostalih; »operibus« je ponešto laskavije po Riarija, kojeg smrt u tom slučaju ne bi odvlačila od bogatstva, nego od (započetih) pothvata \item MR 20 \textlatin{»Sentio,« inquit, »filii fratresque mei, manum Domini super me aggrauari; uolens lubensque eius praesto sum uoluntati, eo quidem libentior quo me et famae et gloriae meae satis uixisse scio.«} – sva četiri (također i \textit{pa)} imaju »nolens«, a ne »uolens«, što dramatično izvrće značenje Riarijevih riječi \end{enumerate} \subsection{Veza Gd P m} \begin{enumerate}[label=\alph*)] \item MR 2 \textlatin{Extinctus iacet optimarum artium deditissimus amator} – Gd i P imaju gramatički neumjesno »Extinctis«, svi ostali svjedoci »Extinctus« \item MR 2 \textlatin{Nolite igitur, nolite expectare, praestantissimi patres} – u »Nolite igitur, nolite« ova su tri svjedoka ispustila drugo »nolite«, poništavajući tako dijakopu \item MR 4 \textlatin{Clarescebat autem iam tunc nomen religiosissimi doctissimique uiri, magistri Francisci, conciuis et auunculi sui} – ova tri svjedoka imaju sintaktički neobično »Clarescat« (ostali »Clarescebat«, osim \textit{C} »Clare sciebat«) \item MR 4 \textlatin{Quem ubi conspexisset Franciscus iam religionis ueste indutum} – umjesto »ueste«, \textit{Gd} i \textit{P} imaju »iuste« (svi ostali »ueste«); \textit{m} je očito zapazio nelogičnost pri prijepisu \item MR 5 \textlatin{Quibus uarie sollicitatus coepit auunculum suum hortari, coepit importunius compellere Romam peteret} – umjesto »compellere« sva tri imaju »complere« (uz potonji bismo glagol, upotrijebljen u prenesenom smislu, očekivali oznaku \textit{čime} je nećak »prekrivao« ili »zasipavao« ujaka) \item MR 10 \textlatin{Accipiebat praeterea munera, non auaritiae, sed honoris comparandaeque beniuolentiae gratia} – umjesto »comparandaeque« \textit{Gd} ima »comperandique«, \textit{P m} »comparandique«; uz gerund bi trebalo biti »comparandique beniuolentiam«, no tada se ne bi poštovala težnja biranog latinskog stila da gerund zamijeni gerundivom \item MR 11 \textlatin{Vidi illum quodam uesperi non sine graui stomacho lacrimis suffusum oculos et cum multa indignatione Deum optimum maximumque testem citare} – umjesto »et (cum multa\dots)« sva tri imaju gramatički neumjesno »te« \item MR 14 \textlatin{Omnium saluti die noctuque inseruiebat, et tamen a nonnullis negligentiae accusabatur} – umjesto »accusabatur« \textit{Gd} i \textit{P} imaju sintaktički nelogično »accusabantur«, kao da su »negligentiae« subjekt \item MR 16 \textlatin{excepto hoc piissimo fratre comite Hieronymo; quem quoniam ab inclito duce Mediolani connubio filiae dignatum cernebat} – umjesto »connubio« sva tri imaju »connubia«, kao da je objekt predikata »cernebat« \item MR 19 \textlatin{Proinde non cessabat ecclesias suae curae commendatas collapsas erigere, exornare deformes; praedia occupata uendicare, bona priorum rectorum distracta negligentia propriis pecuniis recuperare; uestimenta, libros, uasa sacra et caetera ad splendorem diuini cultus spectantia maximis sumptibus coemere} – umjesto »uasa« \textit{Gd} i \textit{P} imaju gramatički neodrživo »uaso« \item MR 20 \textlatin{ut uos horter atque obtester ne huius mundi illecebris atque lenociniis animum uestrum inducatis neue in luxu ac inanibus eius diuitiis spem ullam ponatis} – umjesto »inanibus« (\textit{va} »in anibus«) \textit{Gd} i \textit{P} imaju »manibus«, riječ koja je sama za sebe smislena, ali u rečenici neumjesna \end{enumerate} \subsection{Tijesna veza P m} \begin{enumerate}[label=\alph*)] \item MR 3 \textlatin{quoniam et honestissimis nobilissimisque ciuitatis suae parentibus est ortus et celeberrimo uetustoque Ligurum oppido Saona} – oba svjedoka imaju »Soana«, ostali »Saona« \item MR 4 \textlatin{uersus complures multosque grammaticae textus, quos olim puer edidicerat} – oba svjedoka imaju »edicerat«, ostali »edidicerat« \item MR 5 \textlatin{Cerno hic nonnullos praelatos et ex aliis ordinibus uiros praestantes a quibus magna cum attestatione audiui} – oba svjedoka imaju gramatički nemotivirano »praestante«, ostali »praestantes« \item MR 5 \textlatin{dicens in manu solius omnipotentis Dei esse omnia regna terrarum atque illa, quibus ipse uoluerit, tradi} – oba imaju sintaktički neopravdano »uoluerint«, ostali »uoluerit« \item MR 6 \textlatin{simulque cum eis magnanimitatem, clementiam, munificentiam, et ceteras, quas prius commemorauimus, imperantium uirtutes} – \textit{V P m} imaju neobično »prius quas prius«, ostali samo »quas prius« \item MR 8 \textlatin{Mali quippe et iniqui hominis esse dicebat meliorem se foris ostendere quam gerere domi} – \textit{V P m} imaju (nelogično) »fortis«, nasuprot »foris« kod ostalih (osim »fori« kod \textit{Ge C)} \item MR 8 \textlatin{liberius argueretur quod nimia indulgentia et largitate domesticos faceret insolentiores} – \textit{P m} imaju negramatično »indulgenti et a«, ostali »indulgentia et« \item MR 8 \textlatin{placida uoce respondit} – \textit{P m} imaju »placenda«, ostali »placida« \item MR 17 \textlatin{Quattuordecim enim milia ducatorum accepit et opulentissimum oppidum Bosti, ex quo et aliis a duce adiectis praediis plus quam quinque milia ducatorum quotannis capere poterit} – umjesto »et opulentissimum« \textit{P m} imaju negramatično »hec opulentissimum«, \textit{V} »ec opulentissimum« (svi ostali »et opulentissimum«) \item MR 17 \textlatin{»Illud« inquit »ut iustitiam mearum precum gratia minime uioles, nec secus feceris etiam si te germanus meus Hieronymus\dots«} – \textit{P m} imaju uiolet (kao da se Riario ne obraća izravno upravitelju grada Rima, bez obzira na »secus feceris« koje neposredno slijedi), ostali »uioles« \item MR 20 \textlatin{Iamque uicinus morti domesticos ac familiares accersiri iubet, quibus praesto existentibus in nullos prorupit fletus, nullis mundanarum cupiditatum desideriis ingemuit} – dok ostali imaju »nullis«, u \textit{P m} stoji vrlo nezgodno »multis« (što surečenici daje upravo suprotan smisao) \item MR 21 \textlatin{Minus quippe meae iuuentutis potens fui et nonnunquam partim oculos, partim aures uestras in multis offendi. Sed eorum tanto facilius me a Domino misericordiam consecuturum confido quanto uos modestius uiuentes pro me Dominum deprecabimini. Ego quoque, si quis mortuis erit sensus, idem pro uobis me spondeo facturum.} – \textit{P m}, opet sadržajno neugodno, ispuštaju čitavu rečenicu »Sed\dots\ deprecabimini«, što dramatično iskrivljuje smisao odlomka (»i vaše oči, i uši umnogome sam povrijedio. A i ja ću, tvrdo obećajem, i za vas učiniti isto\dots«) \item MR 21 \textlatin{Viuite mei memores et, quam caduca sit huius mundi felicitas, uel meo exemplo discite} – gdje ostali imaju »Viuite«, \textit{P m} imaju »Uite« (što se možda može shvatiti kao genitiv imenice \textit{uita,} »pamteći moj život«) \end{enumerate} \subsection{Obitelj 2 (Ge C va o)} \begin{enumerate}[label=\alph*)] \item MR 3 \textlatin{eas laudes, quas uel a parentibus uel a patria ipsius colligere poteram, hoc loco praetermittendas putaui; non quod illas aut obscuras aut tenues fore duxerim} – \textit{Ge C va o} imaju »duxerim«, nasuprot »dixerim« kod ostalih (obje su varijante jezično i smisleno prihvatljive) \item MR 6 \textlatin{Assumpsit enim cum sublimi magistratu sublimes animos et spiritus tanti imperii maiestate dignos simulque cum eis magnanimitatem} – ova četiri svjedoka imaju zalihosno »et simulque« nasuprot »simulque« kod ostalih \item MR 19 \textlatin{Haec quoque sacra apostolorum aedes beneficentiam eius testari potuisset si tantum quattuor mensibus superstitem uidisset.} – sva četiri imaju »et si (tantum quattuor\dots)«, ostali si; et si bi rečenici dalo neželjen dopusni prizvuk (»čak i da ga je bazilika vidjela kako je još četiri mjeseca poživio«) \item MR 22 \textlatin{Ad hanc uocem illa dilecta Deo anima ueluti certo accepto signo ad Dominum suum confestim euolauit.} – u opisu Riarijeve smrti, sva četiri svjedoka ispuštaju čitavu ovu rečenicu, čime se (možda iz razloga dobrog ukusa) ublažava inzistiranje na svetosti kardinalovih posljednjih trenutaka; ovo smatramo ključnom razlikom za podjelu na grane u predaji \end{enumerate} \subsection{Veza Ge va o} \begin{enumerate}[label=\alph*)] \item MR 2 \textlatin{Extinctus iacet optimarum artium deditissimus amator, interiit omnium studiosorum praecipuus fautor} – umjesto »studiosorum« ova tri svjedoka imaju »studiorum«, i time unekoliko mijenjaju značenje (u toj verziji Riario nije zaštitnik znanstvenika, nego znanosti) \item MR 5 \textlatin{ut discerent uniuersi ueram certamque esse illam Nabuchdenosoris confessionem in quam et regno et sensui restitutus supplex prorupit dicens in manu solius omnipotentis Dei esse omnia regna terrarum} – \textit{Ge} i \textit{o} imaju gramatički neodrživo »eā« (skraćenicu za »eam«), \textit{va} piše »ea«, kao da se odnosi na »regna«; ostali svjedoci imaju »esse« \end{enumerate} \subsection{Poseban položaj R ve pa co} \begin{enumerate}[label=\alph*)] \item MR 1 \textlatin{Cum in omni funebri celebratione duo praecipue dicendi genera a maioribus nostris usurpari soleant} – samo ova četiri svjedoka ispuštaju »omni« \item MR 23 \textlatin{nec sibi nec gloriae suae parum uixit qui, quaecumque unius hominis fortuna capere potuit, abunde consecutus est. Nobis forsan amplius uiuere poterat, et nimirum magno et ornamento et utilitati.} – samo ova četiri iz druge rečenice ispuštaju »uiuere«, kao da se podrazumijeva prema prvoj \end{enumerate} \subsection{Veza ve pa} \begin{enumerate}[label=\alph*)] \item MR 4 \textlatin{ut eum putares die noctuque nulli adeo alii rei quam euoluendis theologorum philosophorumque libris uacare} – oba svjedoka ispuštaju »adeo« i time poništavaju hiperbat \end{enumerate} \subsection{Amplifikacije u rostočkom izdanju} \subsection{Paratekstovi u mletačkom kodeksu} Samo mletački rukopis \textit{(ve)} na posljednjoj stranici prijepisa Nikolina govora (f.~19v), odmah nakon završetka glavnog teksta, donosi četiri pohvalna epigrama od po dva stiha (odnosno, po jedan elegijski distih). Pjesme su očito bile predviđene kao paratekstovi tiskanog izdanja. Prvi epigram hvali Nikolin govor – pohvala potvrđuje i kontroverznost Riarijeve reputacije i planiranu apologetsku funkciju govora – dok su ostala tri varijante epitafa za pokojnog kardinala. Sve pjesme donosimo ovdje, uz naš prijevod. \poemtitle*{In laudem libelli} \begin{verse} \begin{altverse} Ęloquio uires quantę sint, aspice, lector\\ Quem prius odisti, protinus hunc peramas. \end{altverse} \end{verse} \poemtitle*{U slavu knjižice} \begin{verse} \begin{altverse} Čitaoče, uvjeri se kakva je u riječima snaga:\\ kog si nekoć mrzio, sad si odmah zavolio. \end{altverse} \end{verse} \bigskip \poemtitle*{Epithaphion (!)} \begin{verse} \begin{altverse} Nemo magis docuit perituri temnere sęcli\\ Diuitias, fastum, luxuriemque simul. \end{altverse} \end{verse} \poemtitle*{Epitaf} \begin{verse} \begin{altverse} Nitko nas nije poučio bolje prezirati prolaznog svijeta\\ raskoš, bogatstvo i sva blaga. \end{altverse} \end{verse} \bigskip \poemtitle*{Aliud.} \begin{verse} \begin{altverse} Quam celeris fugiat ruituri gloria mundi\\ Me speculum cernas quisque uiator ades. \end{altverse} \end{verse} \poemtitle*{Još jedan.} \begin{verse} \begin{altverse} Kako brzo nestaje slava krhkog svijeta,\\ vidi u meni, ogledalu svojem, putniče, ma tko bio. \end{altverse} \end{verse} \bigskip \poemtitle*{Aliud.} \begin{verse} \begin{altverse} Sorte humili natum qui me cognouerit ante\\ Fortunę uarios rideat ille iocos. \end{altverse} \end{verse} \poemtitle*{Još jedan.} \begin{verse} \begin{altverse} Tko je znao kako skromnog sam roda, nek se smije\\ šalama raznovrsnim što ih zbija hir sudbine. \end{altverse} \end{verse} Ne znamo ni autora ni vrijeme nastanka ovih stihova. Pjesme variraju motivima i formom, kako i dolikuje nadgrobnome ciklusu. Stihovi su složeni metrički besprijekorno. Odabir riječi i sintaksa ne odstupaju od uzusa rimske književnosti. Nepoznati autor ne »reciklira« dijelove antičkih stihova.\footnote{Usporedba s korpusom sačuvane rimske poezije tek u trećoj pjesmi pronalazi paralele sa stihom Lukrecijeva epa \textit{O prirodi} i Ovidijevih \textit{Fasta}: \textlatin{quisque uiator ades: Ovid. F. 3, 351–352: »At certe credemur« ait »si uerba sequetur / Exitus: en audi crastina, quisquis ades,«} (govori Numa Pompilije). – \textlatin{Quam celeris: Lucret. 4, 210–213: Quam celeri motu rerum simulacra ferantur, / Quod simul ac primum sub diu splendor aquai / Ponitur, extemplo caelo stellante serena / Sidera respondent in aqua radiantia mundi} (o brzini kojom se odraz zvijezda pojavljuje na vodi).} Nešto su učestalije paralele s latinskom poezijom talijanskih suvremenika kog – no, i ovdje su odjeci daleki, više u sličnostima nego u identičnosti.\footnote{Višekratno nailazimo na paralele s poezijom Landinove zbirke \textit{Xandra} (dovršena 1460) i Pontanova \textit{Eridana} (zbirka započeta 1483). Evo popisa paralela: \textlatin{aspice, lector: \textit{Anthologia Latina} 855d, 1–2: Volue tuos oculos: metuendum hunc aspice, lector, / Armorum bellique ducem} (natpis pod likom ; autor ovog epigrama je , oko 1350–1421). – \textlatin{Nemo magis docuit: (Firenca, 1424–1498), \textit{Xandra} 3, 17 \textit{(Ad Petrum Medicem de laudibus Poggi,} 1456–1458), 37–38: Nemo magis dubiis potuit cognoscere rebus / Utile nec docto promptius ore loqui. – Diuitias, fastum, luxuriemque simul: Bonvesin da la Riva (Milano, oko 1240 – oko 1315), \textit{Vita scolastica,} 187–188: Privat, sternit opes, viciat, scelus omne ministrat, / Furta docet, predas luxuriamque simul; isto, 773–774: A viciis caveat, virtutibus hereat, absint / Fastus avaricie luxurieque fimus.} (Bonvesinova je poučna pjesma bila vrlo popularna u Quattrocentu, postoje i brojna tiskana izdanja). – \textlatin{Me speculum cernas quisque uiator ades: (Cerreto di Spoleto 1429 – Napulj 1503), \textit{Eridanus} 2, 2 \textit{(Puerum alloquitur faculam nocturnam praeferentem,} oko 1490), 11–12: Quisquis ades Stellamque vides, mea pectora cerne / In speculo; speculum pectoris illa mei est – Sorte humili natum: Cristoforo Landino, \textit{Xandra} 3, 6 \textit{(Ad Paulum ne timeat bellum Aragonense,} 1452), 69–70: Per quos ah casus, per quanta pericula cernes / Ex humili hunc summum iam tetigisse gradum. – Fortunę uarios rideat ille iocos: (Padova 1261 – Chioggia 1329), \textit{De obsidione} 1 \textit{(De obsidione domini Canis Grandis de Verona circa menia Paduane civitatis),} 730–731: Victores victique pari non mente rotatum / Fortune videre iocum fatique profundi; \textit{De obsidione} 3 \textit{(De conflictu domini Canis apud Paduam),} 218–219: O nunquam stabilis, vario diversa rotatu, / Rebus in humanis quantum, Fortuna, iocaris! Petrarca (Arezzo 1304 – Padova 1374), \textit{Africa} 7, 419–421: Scimus: et hinc maior nostris speratur ab armis / Gloria. Nec regnum Fortune ignoro iocantis / Rebus in humanis. – Ugolino Verrino (Firenca, 1438–1516), \textit{Flametta} (1463) 1, 4 \textit{(Ad Flamettam),} 35–36: O si mixta viris spartanae more palestrae / Tractaret varios tusca puella iocos} (Verrino je Landinov učenik, \textit{Flametta} slijedi uzor zbirke Xandra). – \textlatin{Pontano, \textit{Eridanus} 2, 31 \textit{(Ad Marcum Antonium Sabellicum scriptorem historiarum,} nakon 1487), 25–26 Stant et opes animi validae; ridemus iniquas / Fortunae insidias instabilisque vices.}} \section{Načela ovog izdanja} \end{document} \section*{1.\thinspace LIBRI IMPRESSI} \section*{2.\thinspace CODICES} \bigskip \begin{description}[noitemsep,itemsep=3pt,labelsep=5pt,font=\rmfamily] \item[ve] -- Codex Venetus, \emph{Marc.~Lat.\ cl.~XIV, 180 (4667),} saec.~XV, ff.~9r–19v. \item[va] -- Codex Vaticanus: Bibliotheca apostolica Vaticana \emph{Vat.~lat.\ 8750;} saec.\ XV exeunte – XVI ineunte; ff.~205r–212v; e \textit{Ge} descriptus esse uidetur. \item[pa] -- Codex Panormitanus: Panormi, Biblioteca centrale della Regione siciliana, \emph{I.B.6,} c.~1474–1480; ff.~32r–54r. \item[co] -- Codex Corsinianus: cart.~misc., saec.~XV; Romae, Accademia Nazionale dei Lincei, Biblioteca dell'Accademia dei Lincei e Corsiniana, fondo principale, \emph{Corsin. 583 (45 C 18),} ff.~117r–123r. \item[m\hphantom{i}] -- Codex Monacensis: Bibliotheca Bauarica \emph{CLM 461}, ab Hartmanno Schedelio scriptus, saec.~XV; ff.~129r–138r; descriptio \textit{P} uidetur esse. \item[o\hphantom{o}] -- Codex Olomoucensis, sub fine saec. XV; Olomoucii, Vědecká knihovna; Textus uarii; Historia Bohemica \emph{sign. M I 159}, ff.~170r–174v; e \textit{Ge} descriptus esse uidetur. \end{description} \clearpage \thispagestyle{empty} \hfill \clearpage wangjunxiao/NFVCloud A minimal matching utility. \href{http://travis-ci.org/isaacs/minimatch}{\tt } This is the matching library used internally by npm. It works by converting glob expressions into Java\+Script {\ttfamily Reg\+Exp} objects. \subsection*{Usage} \begin{DoxyCode} var minimatch = require("minimatch") minimatch("bar.foo", "*.foo") // true! minimatch("bar.foo", "*.bar") // false! minimatch("bar.foo", "*.+(bar|foo)", \{ debug: true \}) // true, and noisy! \end{DoxyCode} \subsection*{Features} Supports these glob features\+: \begin{DoxyItemize} \item Brace Expansion \item Extended glob matching \item \char`\"{}\+Globstar\char`\"{} {\ttfamily $\ast$$\ast$} matching \end{DoxyItemize} See\+: \begin{DoxyItemize} \item {\ttfamily man sh} \item {\ttfamily man bash} \item {\ttfamily man 3 fnmatch} \item {\ttfamily man 5 gitignore} \end{DoxyItemize} \subsection*{Minimatch Class} Create a minimatch object by instantiating the {\ttfamily minimatch.\+Minimatch} class. \begin{DoxyCode} var Minimatch = require("minimatch").Minimatch var mm = new Minimatch(pattern, options) \end{DoxyCode} \subsubsection*{Properties} \begin{DoxyItemize} \item {\ttfamily pattern} The original pattern the minimatch object represents. \item {\ttfamily options} The options supplied to the constructor. \item {\ttfamily set} A 2-\/dimensional array of regexp or string expressions. Each row in the array corresponds to a brace-\/expanded pattern. Each item in the row corresponds to a single path-\/part. For example, the pattern {\ttfamily \{a,b/c\}/d} would expand to a set of patterns like\+: \begin{DoxyVerb} [ [ a, d ] , [ b, c, d ] ] \end{DoxyVerb} If a portion of the pattern doesn\textquotesingle{}t have any \char`\"{}magic\char`\"{} in it (that is, it\textquotesingle{}s something like {\ttfamily \char`\"{}foo\char`\"{}} rather than {\ttfamily fo$\ast$o?}), then it will be left as a string rather than converted to a regular expression. \item {\ttfamily regexp} Created by the {\ttfamily make\+Re} method. A single regular expression expressing the entire pattern. This is useful in cases where you wish to use the pattern somewhat like {\ttfamily fnmatch(3)} with {\ttfamily F\+N\+M\+\_\+\+P\+A\+TH} enabled. \item {\ttfamily negate} True if the pattern is negated. \item {\ttfamily comment} True if the pattern is a comment. \item {\ttfamily empty} True if the pattern is {\ttfamily \char`\"{}\char`\"{}}. \end{DoxyItemize} \subsubsection*{Methods} \begin{DoxyItemize} \item {\ttfamily make\+Re} Generate the {\ttfamily regexp} member if necessary, and return it. Will return {\ttfamily false} if the pattern is invalid. \item {\ttfamily match(fname)} Return true if the filename matches the pattern, or false otherwise. \item {\ttfamily match\+One(file\+Array, pattern\+Array, partial)} Take a {\ttfamily /}-\/split filename, and match it against a single row in the {\ttfamily reg\+Exp\+Set}. This method is mainly for internal use, but is exposed so that it can be used by a glob-\/walker that needs to avoid excessive filesystem calls. \end{DoxyItemize} All other methods are internal, and will be called as necessary. \subsubsection*{minimatch(path, pattern, options)} Main export. Tests a path against the pattern using the options. \begin{DoxyCode} var isJS = minimatch(file, "*.js", \{ matchBase: true \}) \end{DoxyCode} \subsubsection*{minimatch.\+filter(pattern, options)} Returns a function that tests its supplied argument, suitable for use with {\ttfamily Array.\+filter}. Example\+: \begin{DoxyCode} var javascripts = fileList.filter(minimatch.filter("*.js", \{matchBase: true\})) \end{DoxyCode} \subsubsection*{minimatch.\+match(list, pattern, options)} Match against the list of files, in the style of fnmatch or glob. If nothing is matched, and options.\+nonull is set, then return a list containing the pattern itself. \begin{DoxyCode} var javascripts = minimatch.match(fileList, "*.js", \{matchBase: true\})) \end{DoxyCode} \subsubsection*{minimatch.\+make\+Re(pattern, options)} Make a regular expression object from the pattern. \subsection*{Options} All options are {\ttfamily false} by default. \subsubsection*{debug} Dump a ton of stuff to stderr. \subsubsection*{nobrace} Do not expand {\ttfamily \{a,b\}} and {\ttfamily \{1..3\}} brace sets. \subsubsection*{noglobstar} Disable {\ttfamily $\ast$$\ast$} matching against multiple folder names. \subsubsection*{dot} Allow patterns to match filenames starting with a period, even if the pattern does not explicitly have a period in that spot. Note that by default, {\ttfamily a/$\ast$$\ast$/b} will {\bfseries not} match {\ttfamily a/.d/b}, unless {\ttfamily dot} is set. \subsubsection*{noext} Disable \char`\"{}extglob\char`\"{} style patterns like {\ttfamily +(a$\vert$b)}. \subsubsection*{nocase} Perform a case-\/insensitive match. \subsubsection*{nonull} When a match is not found by {\ttfamily minimatch.\+match}, return a list containing the pattern itself if this option is set. When not set, an empty list is returned if there are no matches. \subsubsection*{match\+Base} If set, then patterns without slashes will be matched against the basename of the path if it contains slashes. For example, {\ttfamily a?b} would match the path {\ttfamily /xyz/123/acb}, but not {\ttfamily /xyz/acb/123}. \subsubsection*{nocomment} Suppress the behavior of treating {\ttfamily \#} at the start of a pattern as a comment. \subsubsection*{nonegate} Suppress the behavior of treating a leading {\ttfamily !} character as negation. \subsubsection*{flip\+Negate} Returns from negate expressions the same as if they were not negated. (Ie, true on a hit, false on a miss.) \subsection*{Comparisons to other fnmatch/glob implementations} While strict compliance with the existing standards is a worthwhile goal, some discrepancies exist between minimatch and other implementations, and are intentional. If the pattern starts with a {\ttfamily !} character, then it is negated. Set the {\ttfamily nonegate} flag to suppress this behavior, and treat leading {\ttfamily !} characters normally. This is perhaps relevant if you wish to start the pattern with a negative extglob pattern like {\ttfamily !(a$\vert$B)}. Multiple {\ttfamily !} characters at the start of a pattern will negate the pattern multiple times. If a pattern starts with {\ttfamily \#}, then it is treated as a comment, and will not match anything. Use {\ttfamily \textbackslash{}\#} to match a literal {\ttfamily \#} at the start of a line, or set the {\ttfamily nocomment} flag to suppress this behavior. The double-\/star character {\ttfamily $\ast$$\ast$} is supported by default, unless the {\ttfamily noglobstar} flag is set. This is supported in the manner of bsdglob and bash 4.\+1, where {\ttfamily $\ast$$\ast$} only has special significance if it is the only thing in a path part. That is, {\ttfamily a/$\ast$$\ast$/b} will match {\ttfamily a/x/y/b}, but {\ttfamily a/$\ast$$\ast$b} will not. If an escaped pattern has no matches, and the {\ttfamily nonull} flag is set, then minimatch.\+match returns the pattern as-\/provided, rather than interpreting the character escapes. For example, {\ttfamily minimatch.\+match(\mbox{[}\mbox{]}, \char`\"{}\textbackslash{}\textbackslash{}$\ast$a\textbackslash{}\textbackslash{}?\char`\"{})} will return {\ttfamily \char`\"{}\textbackslash{}\textbackslash{}$\ast$a\textbackslash{}\textbackslash{}?\char`\"{}} rather than {\ttfamily \char`\"{}$\ast$a?\char`\"{}}. This is akin to setting the {\ttfamily nullglob} option in bash, except that it does not resolve escaped pattern characters. If brace expansion is not disabled, then it is performed before any other interpretation of the glob pattern. Thus, a pattern like {\ttfamily +(a$\vert$\{b),c)\}}, which would not be valid in bash or zsh, is expanded {\bfseries first} into the set of {\ttfamily +(a$\vert$b)} and {\ttfamily +(a$\vert$c)}, and those patterns are checked for validity. Since those two are valid, matching proceeds. RFC-cite-refs/bib/rfc8468.bib @misc{rfc8468, series = {Request for Comments}, number = 8468, howpublished = {RFC 8468}, publisher = {RFC Editor}, doi = {10.17487/RFC8468}, url = {https://rfc-editor.org/rfc/rfc8468.txt}, author = { and and and and }, title = {{IPv4, IPv6, and IPv4-IPv6 Coexistence: Updates for the IP Performance Metrics (IPPM) Framework}}, pagetotal = 15, year = 2018, month = nov, abstract = {This memo updates the IP Performance Metrics (IPPM) framework defined by RFC 2330 with new considerations for measurement methodology and testing. It updates the definition of standard-formed packets to include IPv6 packets, deprecates the definition of minimal IP packet, and augments distinguishing aspects, referred to as Type-P, for test packets in RFC 2330. This memo identifies that IPv4-IPv6 coexistence can challenge measurements within the scope of the IPPM framework. Example use cases include, but are not limited to, IPv4-IPv6 translation, NAT, and protocol encapsulation. IPv6 header compression and use of IPv6 over Low-Power Wireless Area Networks (6LoWPAN) are considered and excluded from the standard-formed packet evaluation.}, } DeepRoboticsResearch/OpenRoboticsdb/v29.bib @proceedings{ACML2013, booktitle = {Asian Conference on Machine Learning}, shortname = {ACML}, year = {2013}, editor = { and }, volume = {29}, sections = {preface=Front Matter|default=Accepted Papers} } h3ct0r/itv_report_latex_templatebibliography.bib @article{goertz1954mechanical, title={Mechanical master-slave manipulator}, author={}, journal={Nucleonics (US) Ceased publication}, volume={12}, year={1954}, publisher={Argonne National Lab., Lemont, Ill.} } @book{Siciliano2016Springer, author = { and }, title = {Springer Handbook of Robotics}, year = {2016}, isbn = {3319325507, 9783319325507}, edition = {2nd}, publisher = {Springer Publishing Company, Incorporated}, } @article{lichiardopol2007survey, title={A survey on teleoperation}, author={}, journal={Technische Universitat Eindhoven, DCT report}, year={2007} } @article{steuer1992defining, title={Defining virtual reality: Dimensions determining telepresence}, author={}, journal={Journal of communication}, volume={42}, number={4}, pages={73--93}, year={1992}, publisher={Wiley Online Library} } @inproceedings{varkonyi2014survey, title={Survey on the control of time delay teleoperation systems}, author={ and and and }, booktitle={Intelligent Engineering Systems (INES), 2014 18th International Conference on}, pages={89--94}, year={2014}, organization={IEEE} } @article{teng2010customization, title={Customization, immersion satisfaction, and online gamer loyalty}, author={}, journal={Computers in Human Behavior}, volume={26}, number={6}, pages={1547--1554}, year={2010}, publisher={Elsevier} } @article{stanney1998aftereffects, title={Aftereffects and sense of presence in virtual environments: Formulation of a research and development agenda}, author={ and }, journal={International Journal of Human-Computer Interaction}, volume={10}, number={2}, pages={135--187}, year={1998}, publisher={Taylor \& Francis} } @techreport{witmer1994measuring, title={Measuring immersion in virtual environments}, author={ and Singer, }, year={1994}, institution={Tech. Rep. 1014, US Army Research Institute for the Behavioral and Social Sciences} } @article{riley2004situation, title={Situation awareness and attention allocation measures for quantifying telepresence experiences in teleoperation}, author={ and and Draper, }, journal={Human Factors and Ergonomics in Manufacturing \& Service Industries}, volume={14}, number={1}, pages={51--67}, year={2004}, publisher={Wiley Online Library} } @article{lombard2000concept, title={The concept of presence: Explication statement}, author={}, journal={Available from International Society for Presence Research Web site: http://www. temple. edu/ispr/index. htm}, year={2000} } @inproceedings{biocca2001networked, title={The networked minds measure of social presence: Pilot test of the factor structure and concurrent validity}, author={ }, booktitle={4th annual international workshop on presence, Philadelphia, PA}, pages={1--9}, year={2001} } @article{lessiter2001cross, title={A cross-media presence questionnaire: The ITC-Sense of Presence Inventory}, author={ }, journal={Presence: Teleoperators and virtual environments}, volume={10}, number={3}, pages={282--297}, year={2001}, publisher={MIT Press} } @inproceedings{lombard2004literature, title={A literature-based presence measurement instrument the temple presence inventory (tpi)(beta)}, author={ and }, booktitle={The seventh international workshop on presence (PRESENCE2004)}, year={2004} } @article{kendon1990spatial, title={Spatial Organization in Social Encounters: the F-formation System. A. Kendon, Ed}, author={}, journal={Conducting interaction: Patterns of behavior in focused encounters}, pages={209--237}, year={1990} } @inproceedings{marshall2011using, title={Using F-formations to analyse spatial patterns of interaction in physical environments}, author={ and and }, booktitle={Proceedings of the ACM 2011 conference on Computer supported cooperative work}, pages={445--454}, year={2011}, organization={ACM} } @article{kristoffersson2013towards, title={Towards measuring quality of interaction in mobile robotic telepresence using sociometric badges}, author={ }, journal={Paladyn, Journal of Behavioral Robotics}, volume={4}, number={1}, pages={34--48}, year={2013} } @inproceedings{lombard2009measuring, title={Measuring presence: the temple presence inventory}, author={ and }, booktitle={Proceedings of the 12th Annual International Workshop on Presence}, pages={1--15}, year={2009} } @article{kendon1990spatial, title={Spatial Organization in Social Encounters: the F-formation System. A. Kendon, Ed}, author={}, journal={Conducting interaction: Patterns of behavior in focused encounters}, pages={209--237}, year={1990} } @inproceedings{marshall2011using, title={Using F-formations to analyse spatial patterns of interaction in physical environments}, author={ }, booktitle={Proceedings of the ACM 2011 conference on Computer supported cooperative work}, pages={445--454}, year={2011}, organization={ACM} } @article{kristoffersson2013towards, title={Towards measuring quality of interaction in mobile robotic telepresence using sociometric badges}, author={ }, journal={Paladyn, Journal of Behavioral Robotics}, volume={4}, number={1}, pages={34--48}, year={2013} } @article{sheridan1992musings, title={Musings on telepresence and virtual presence}, author={}, journal={Presence: Teleoperators \& Virtual Environments}, volume={1}, number={1}, pages={120--126}, year={1992}, publisher={MIT Press} } @article{sheridan1989telerobotics, title={Telerobotics}, author={}, journal={Automatica}, volume={25}, number={4}, pages={487--507}, year={1989}, publisher={Elsevier} } @article{sheridan1993space, title={Space teleoperation through time delay: Review and prognosis}, author={}, journal={IEEE Transactions on robotics and Automation}, volume={9}, number={5}, pages={592--606}, year={1993}, publisher={IEEE} } @article{slater1997framework, title={A framework for immersive virtual environments (FIVE): Speculations on the role of presence in virtual environments}, author={ and }, journal={Presence: Teleoperators and virtual environments}, volume={6}, number={6}, pages={603--616}, year={1997}, publisher={MIT Press} } @article{barfield1993sense, title={The sense of presence within virtual environments: A conceptual framework}, author={ and }, journal={Advances in Human Factors Ergonomics}, volume={19}, pages={699--699}, year={1993} } @inproceedings{prothero1995towards, title={Towards a robust, quantitative measure for presence}, author={ and and and }, booktitle={Proceedings of the conference on Experimental Analysis and Measurement of Situation Awareness}, pages={359--366}, year={1995} } @article{draper1998telepresence, title={Telepresence}, author={ and Kaber, and }, journal={Human factors}, volume={40}, number={3}, pages={354--375}, year={1998}, publisher={SAGE Publications Sage CA: Los Angeles, CA} } @article{nichols2000measurement, title={Measurement of presence and its consequences in virtual environments}, author={ and }, journal={International Journal of Human-Computer Studies}, volume={52}, number={3}, pages={471--491}, year={2000}, publisher={Elsevier} } @article{Park2006, doi = {10.1177/0278364906065385}, url = {https://doi.org/10.1177/0278364906065385}, year = {2006}, month = {may}, publisher = {{SAGE} Publications}, volume = {25}, number = {5-6}, pages = {575--591}, author = { and }, title = {A Haptic Teleoperation Approach Based on Contact Force Control}, journal = {The International Journal of Robotics Research} } @article{Owada2017, author = { and }, isbn = {9781509067596}, keywords = {Humanoid robot,Index Terms—Teleoperation,RHI,Telexistance}, pages = {1602--1607}, title = {{Direction Consistent Approach for Telepresence System with Miniature Humanoid Robot}}, year = {2017} } @book{dragan2012formalizing, title={Formalizing assistive teleoperation}, author={ and }, year={2012}, publisher={MIT Press, July} } @inproceedings{steinfeld2006common, title={Common metrics for human-robot interaction}, author={, Lewis, , Jean and Schultz, Alan and }, booktitle={Proceedings of the 1st ACM SIGCHI/SIGART conference on Human-robot interaction}, pages={33--40}, year={2006}, organization={ACM} } @inproceedings{ralston1998numbat, title={The Numbat: A remotely controlled mine emergency response vehicle}, author={Ralston, and Hainsworth, }, booktitle={Field and Service Robotics}, pages={53--59}, year={1998}, organization={Springer} } @inproceedings{you2012assisted, title={Assisted teleoperation strategies for aggressively controlling a robot arm with 2d input}, author={ and }, booktitle={Robotics: science and systems}, volume={7}, pages={354}, year={2012} } @article{Gaemperle2015, author = { Popovic, , Yusuf}, doi = {10.1109/ISM.2014.62}, isbn = {9781479943111}, journal = {Proceedings - 2014 IEEE International Symposium on Multimedia, ISM 2014}, keywords = {Head-Mounted Display,Omnidirectional Video,Real-Time,Streaming,Telepresence}, pages = {175--178}, title = {{An immersive telepresence system using a real-time omnidirectional camera and a virtual reality head-mounted display}}, year = {2015} } @inproceedings{park1995augmented, title={Augmented reality for mining teleoperation}, author={ and Kazman, }, booktitle={Telemanipulator and Telepresence Technologies}, volume={2351}, pages={119--130}, year={1995}, organization={International Society for Optics and Photonics} } @article{ralston2001recent, title={Recent advances in remote coal mining machine sensing, guidance, and teleoperation}, author={Ralston, and Hainsworth, and Reid, and Anderson, DL and McPhee, }, journal={Robotica}, volume={19}, number={5}, pages={513--526}, year={2001}, publisher={Cambridge University Press} } @inproceedings{reid2010real, title={Real-world automation: New capabilities for underground longwall mining}, author={Reid, and Dunn, and Reid, and Ralston, }, booktitle={Proc. Australasian Conf. Robotics and Automation 2010}, year={2010} } @inproceedings{kwitowski1989computer, title={Computer-based, teleoperation of a new highwall mining system}, author={Kwitowski, and Lewis, and Mayercheck, }, booktitle={Robotics and Automation, 1989. Proceedings., 1989 IEEE International Conference on}, pages={1478--1483}, year={1989}, organization={IEEE} } @article{kim2012autonomy, title={How autonomy impacts performance and satisfaction: Results from a study with spinal cord injured subjects using an assistive robot}, author={ and and Culver-Godfrey, Heather and Rucks, Greta and and Portee, David and Bricout, John and Wang, Zhao and Behal, Aman}, journal={IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans}, volume={42}, number={1}, pages={2--14}, year={2012}, publisher={IEEE} } @inproceedings{baier2000benefits, title={Benefits of combined active stereo vision and haptic telepresence}, author={ Freyberger, , G{\"u}nther}, booktitle={Intelligent Robots and Systems, 2000.(IROS 2000). Proceedings. 2000 IEEE/RSJ International Conference on}, volume={1}, pages={702--707}, year={2000}, organization={IEEE} } @article{mcintire2014stereoscopic, title={Stereoscopic 3D displays and human performance: A comprehensive review}, author={McIntire, and Havig, and Geiselman, }, journal={Displays}, volume={35}, number={1}, pages={18--26}, year={2014}, publisher={Elsevier} } @article{johan2010evaluation, title={An Evaluation of Local Autonomy Applied to Teleoperated Vehicles in Underground Mines}, author={, and }, journal={IEEE International Conference on Robotics and Automation}, pages={1745--1752}, year={2010}, publisher={IEEE} } @article{kennedy1993simulator, title={Simulator sickness questionnaire: An enhanced method for quantifying simulator sickness}, author={Kennedy, and Lane, and Berbaum, and Lilienthal, }, journal={The international journal of aviation psychology}, volume={3}, number={3}, pages={203--220}, year={1993}, publisher={Taylor \& Francis} } @article{lambooij2009visual, title={Visual discomfort and visual fatigue of stereoscopic displays: A review}, author={Lambooij, Marc and Fortuin, and }, journal={Journal of Imaging Science and Technology}, volume={53}, number={3}, pages={30201--1}, year={2009}, publisher={Society for Imaging Science and Technology} } @inproceedings{hayward1996performance, title={Performance measures for haptic interfaces}, author={ and Astley, }, booktitle={ROBOTICS RESEARCH-INTERNATIONAL SYMPOSIUM-}, volume={7}, pages={195--206}, year={1996}, organization={MIT PRESS} } @inproceedings{Miller, doi = {10.1109/ichr.2004.1442670}, url = {https://doi.org/10.1109/ichr.2004.1442670}, publisher = {{IEEE}}, author = { and and and }, title = {Motion capture from inertial sensing for untethered humanoid teleoperation}, booktitle = {4th {IEEE}/{RAS} International Conference on Humanoid Robots, 2004.} } @article{kobayashi2016human, title={Human motion caption with vision and inertial sensors for hand/arm robot teleoperation}, author={ and and and Nakamoto, Hiroyuki and }, journal={International Journal of Applied Electromagnetics and Mechanics}, volume={52}, number={3-4}, pages={1629--1636}, year={2016}, publisher={IOS Press} } @article{dejing2016local, doi = {10.1109/ICSensT.2016.7796325}, title={Local Implicit Surface Based Virtual Fixtures in Point Cloud Virtual Environment for Teleoperation}, author={, , , }, journal={Tenth International Conference on Sensing Technology}, pages={1--6}, year={2016}, publisher={IEEE} } @article{moeslund2006survey, title={A survey of advances in vision-based human motion capture and analysis}, author={ and and }, journal={Computer vision and image understanding}, volume={104}, number={2}, pages={90--126}, year={2006}, publisher={Elsevier} } @inproceedings{muller2007documentation, title={Documentation mocap database hdm05}, author={ }, year={2007}, publisher={Citeseer}, booktitle={Proc. Australasian Conference on Robotics and Automation} } @inproceedings{stanton2012teleoperation, title={Teleoperation of a humanoid robot using full-body motion capture, example movements, and machine learning}, author={ and Bogdanovych, Anton and Ratanasena, Edward}, booktitle={Proc. Australasian Conference on Robotics and Automation}, year={2012} } @techreport{begault2000sound, title={3-D Sound for Virtual Reality and Multimedia}, author={Begault, and Trejo, }, year={2000}, organization={NASA Ames Research Center, Moffett Field, CA United States }, } @article{moeslund2001survey, title={A survey of computer vision-based human motion capture}, author={Moeslund, and }, journal={Computer vision and image understanding}, volume={81}, number={3}, pages={231--268}, year={2001}, publisher={Elsevier} } @article{Sian2005switching, title={A switching command-based whole-body operation method for humanoid robots}, author={, , , and }, journal={IEEE/ASME Transactions on Mechatronics}, volume={10}, pages={546--559}, year={2005}, publisher={IEEE Journals \& Magazines} } @book{earnshaw2014virtual, title={Virtual reality systems}, author={}, year={2014}, publisher={Academic press} } @inproceedings{kirner2007realidade, title={Realidade virtual e aumentada: conceitos, projeto e aplica{\c{c}}{\~o}es}, author={ and }, booktitle={Livro do IX Symposium on Virtual and Augmented Reality, Petr{\'o}polis (RJ), Porto Alegre: SBC}, year={2007} } @article{zhang2017head, title={Head-mounted display-based intuitive virtual reality training system for the mining industry}, author={}, journal={International Journal of Mining Science and Technology}, year={2017}, publisher={Elsevier} } @article{azuma1997survey, title={A survey of augmented reality}, author={}, journal={Presence: Teleoperators and virtual environments}, volume={6}, number={4}, pages={355--385}, year={1997}, publisher={MIT Press} } @article{azuma2001recent, title={Recent advances in augmented reality}, author={}, journal={IEEE computer graphics and applications}, volume={21}, number={6}, pages={34--47}, year={2001}, publisher={IEEE} } @article{yew2017immersive, title={Immersive Augmented Reality Environment for the Teleoperation of Maintenance Robots}, author={Yew, AWW and Ong, SK and Nee, AYC}, journal={Procedia CIRP}, volume={61}, pages={305--310}, year={2017}, publisher={Elsevier} } @article{chintamani2010improved, title={Improved telemanipulator navigation during display-control misalignments using augmented reality cues}, author={ }, journal={IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans}, volume={40}, number={1}, pages={29--39}, year={2010}, publisher={IEEE} } @article{xin20113d, title={3D augmented reality teleoperated robot system based on dual vision}, author={ and and and and }, journal={The Journal of China Universities of Posts and Telecommunications}, volume={18}, number={1}, pages={105--112}, year={2011}, publisher={Elsevier} } @article{sanchez1999virtual, title={Virtual environment interaction through 3D audio by blind children}, author={}, journal={CyberPsychology \& Behavior}, volume={2}, number={2}, pages={101--111}, year={1999} } hortont424/orbitals \documentclass{acmsiggraph} \usepackage{parskip} \usepackage{graphicx} \usepackage{footmisc} \usepackage{amsmath} \usepackage{url} \onlineid{0} \title{\Large Plotting the Single-Electron Solution to Schr\"{o}dinger's Equation with OpenCL} \author{\thanks{e-mail: }\\Rensselaer Polytechnic Institute} \begin{document} \maketitle \begin{figure} \includegraphics[width=84.5mm]{320.png} \caption{The 3-2-0 orbital of the hydrogen atom} \end{figure} \section{Abstract} The realm of physics provides many readily parallelizable algorithms --- often involving the simple evaluation of a function at an enormous number of points in space --- which also happen to produce attractive visualizations. One such visualization is that of the probability distribution of the single electron within a hydrogen atom (or He$^+$, Li$^{2+}$, and so on). By evaluating Schr\"{o}dinger's equation at millions of points, one can construct a representation of the likely location of the electron --- the orbital cloud, if you will. Luckily, the evaluation of this function at a particular point is entirely independent of nearby points, so it is effectively perfect for parallelization. We will use OpenCL to develop and benchmark (on a varied array of hardware, from dated CPUs to very modern GPUs) an implementation of this algorithm which will produce a simple --- but attractive --- image of the atomic orbital of the single-electron hydrogen atom. \section{Physics} Quantum mechanics --- since Louis de Broglie's paper {\it Research on Quantum Theory} --- tells us that matter behaves sometimes as a wave, and other times as a particle. The electron in a hydrogen atom is no different: when it isn't being observed, it doesn't exist at a particular point in space, but instead expands as a wave. The atomic orbital cloud which we're plotting here is actually simply a plot of the density of that wave at all points surrounding the atom's nucleus. All of the equations below are borrowed from \cite{quantumBook}, but have been slightly modified to fit our purposes, as discussed in section \ref{simpSection}. \subsection{Schr\"{o}dinger's Equation} From \cite{quantumBook}, we find the single-electron solution to Schr\"{o}dinger's equation: \begin{equation}\label{psi} \psi\left(n, l, m\right)=\psi_c \left(\mathit{e}^{-r/n}\right) \left(\frac{2r}{n}\right)^l \left[L_{n-l-1}^{2l+1} \left(\frac{2r}{n}\right)\right] Y_l^m\left(\theta,\phi\right), \end{equation} \begin{equation}\label{psiConstant} \psi_c=\sqrt{\left(\frac{2}{n}\right)^3 \frac{\left(n-l-1\right)!}{2n\left[\left(n+l\right)!\right]^3}}. \end{equation} This is the fundamental formula involved in our computation --- it is this equation which we evaluate at each sample. The result of this equation gives the probability that the single electron of our hydrogen atom is at the particular chosen location in the spherical coordinate system, $\left(\theta, \phi, r\right)$, which is the information that we eventually aim to plot. One will notice that $\psi_c$ is independent of the coordinates of evaluation, $\left(\theta, \phi, r\right)$, so it --- as an obvious optimization --- can be computed once and cached for the entire image. Indeed, while all of the benchmarks in this paper were done with $\psi_c$ included, one might notice that --- since it's a constant term across the entire image, and we're scaling the image by {\it another} constant scale factor --- that it doesn't actually affect the visualization, and can be entirely discarded. It appears that discarding the overhead required to pass this value in as an argument to each of the kernel instances and to multiply it into the value of $\psi$ gives a performance increase of between 6 and 10 percent: quite a significant improvement for no change in output! Also, $\psi$ depends on two external functions: $Y$, the spherical harmonic equation (equation \ref{yFunc}), and $L$, the Laguerre polynomial (equation \ref{laguerre}), which will be detailed in sections \ref{sphericalHarmonics} and \ref{laguerreLegendre}, respectively. \subsection{Spherical Harmonics} \label{sphericalHarmonics} $Y$, a solution to Laplace's spherical harmonic function, provides the angular component of $\psi$: \begin{equation}\label{yFunc} Y_l^m\left(\theta,\phi\right)=\epsilon \sqrt{\frac{\left(2l+1\right)}{4\pi} \frac{\left(l-\left|m\right|\right)!}{\left(l+\left|m\right|\right)!}} \mathit{e}^{{\rm i}m\phi} P^m_l\left(\cos\theta\right), \end{equation} \begin{equation}\label{yEpsilon} \epsilon=\begin{cases} \left(-1\right)^m & \text{$m\ge0$} \\ 1 & \text{$m<0$} \end{cases}. \end{equation} $Y$ depends on an additional external function, $P$, the Legendre polynomial (equation \ref{legendre}). \subsection{Laguerre and Legendre} \label{laguerreLegendre} Key to the evaluation of Schr\"{o}dinger's equation are the Laguerre and Legendre polynomials. Unfortunately, the generation of both requires symbolically solving differential equations. \cite{legendreCite} Since OpenCL (rightfully) lacks a symbolic differential equation solver (and the implementation of such a program is far outside of the scope of this project), we have added to our kernel a simple table of the first few polynomials, computed with Mathematica. This limits the range of input $n, l, m$ parameters which can be used with this program, but extending the tables is very simple and requires only patience. The generating equations for these polynomials are as follows: \begin{equation}\label{laguerre} L^\alpha_n\left(x\right)=\frac{x^{-\alpha}e^x}{n!}\frac{d^n}{dx^n}\left(e^{-x}x^{n+\alpha}\right), \end{equation} \begin{equation}\label{legendre} P^u_v\left(z\right)=\frac{\left(1+z\right)^{\mu/2}}{\left(1-z\right)^{\mu/2}} \tilde{F}_{2,1} \left(-v,v+1;1-\mu;\frac{1-z}{2}\right), \end{equation} where $\tilde{F}$ is Gauss' hypergeometric function, which we don't need to bother ourselves with since we're simply allowing Mathematica to evaluate it and using the comparatively simple symbolic results in our table. \section{Implementation} \subsection{Overview} Our implementation is twofold: an OpenCL kernel (written in the C-like OpenCL kernel language) which performs the evaluation of all of the Schr\"{o}dinger's equation samples required for a single pixel of the output image, and a Python script which parses command line options, uses PyOpenCL to load and run the kernel and PIL to create the output image, and performs timed benchmarks. \subsection{Simplifications} \label{simpSection} The decision was made during the implementation phase to redefine the Bohr radius so that $a=1$; this was done in order to keep intermediate numbers within the range of single-precision floating point values, as OpenCL doesn't currently support native double-precision math on any of the GPUs available for benchmarking. Since we're only using this software to generate visualizations, this doesn't affect the output; it simply scales the distances away from the atomic level. The equations above have $a=1$ already substituted in, for brevity. To keep the implementation of this algorithm within the scope of this project, we also decided to use a fixed-camera orthogonal projection. This worked to significantly simplify the process of iteration over all of the samples, as with this restriction, the program simply has to iterate over all of the pixels in the image, evaluating many points in the $z$ dimension for each pixel, without worrying about complex transformations between coordinate systems. \subsection{Evaluation} After arguments have been parsed and an OpenCL context has been created, we create a buffer large enough to store the entirety of the resultant image, and also compute the constant term of $\psi$, both of which are passed to each instance of the kernel. For each pixel in the image, we spawn an instance of our OpenCL kernel which evaluates 2000 samples in the $z$ dimension. Each sample represents the value of the density function, $\psi$, at that point in space (the cartesian coordinates of the image have to be converted into spherical coordinates in order to evaluate $\psi$). All 2000 samples are summed and stored into their respective pixel in the image buffer. Once computation has completed, the buffer is copied back from working memory (either video memory or system memory) to the final image buffer in system memory. Once there, pixel values are scaled linearly so that the brightest pixel is 100\% white, and the image is compressed to PNG and saved to the filesystem. \subsection{Complex Math} The OpenCL specification unfortunately currently does not include complex math primitives, which are necessary for the evaluation of $\psi$. It does reserve the {\bf complex} keyword, which suggests that perhaps support for something similar to GCC's complex type is coming in a future version of OpenCL, which would put it on ground more similar to CUDA. In order to solve this problem, we implemented a small library of complex math functions which make use of the OpenCL {\bf float2} type to store complex numbers. This library includes various complex number constructors, as well as exponentiation, multiplication, square root, the exponential function, and conjugation. These functions are used extensively within the evaluation of $\psi$. \section{Hardware} Benchmarks will be performed on a number of different computation devices across a few different computers: \subsection{GPU} \begin{itemize} \item ATI Radeon 4890, 800$\times$850MHz, 1GB, 250\$\footnote{All hardware prices listed are approximate launch prices. Prices of older hardware, especially the Core 2 Quad, have dropped significantly since introduction.\label{fn:prices}}\footnote{Tested on Windows, with ATI Stream SDK\label{fn:windows}} \end{itemize} \subsection{CPUs} \begin{itemize} \item Intel Core i7 620M, 2$\times$3333MHz, 4GB, 332\$\footref{fn:prices}\footnote{Tested on Mac OS X, with Apple OpenCL\label{fn:osx}} \item Intel Core 2 Quad Q6600, 4$\times$3000MHz, 4GB, 851\$\footref{fn:prices}\footnote{Tested on Linux, with ATI Stream SDK\label{fn:linux}} \item Intel Core 2 Duo E7200, 2$\times$2530MHz, 8GB, 133\$\footref{fn:prices}\footref{fn:osx} \end{itemize} These machines run a variety of different operating systems (including Mac OS X, Windows, and Linux). Comparisons made later in this paper assume that each OS and the drivers and OpenCL implementation used within them are created equally; this is only somewhat reasonable, and should be kept in mind when interpreting results. Also, it should be noted that while each core on a given CPU could potentially be evaluating samples in parallel, a GPU's cores are much more restricted (less general-purpose) and must work in small groups to accomplish their work. Therefore, given the kernel used for this project, the 4890 listed above only has 50 compute units (16 stream processors work together to compute a single sample). \section{Results} Three computers have spent countless hours performing calculations to bring you the following results, as each benchmark point is the minimum of five individual trials. \subsection{Communication Overhead} \begin{figure} \includegraphics[width=84.5mm]{overheadPlot.pdf} \caption{Overhead from image buffer copy as a percentage of the total runtime; smaller values are better} \label{fig:overheadPlot} \end{figure} The algorithm we implemented for this project required no communication between cores whatsoever during the course of the computation. The primary overhead involved in the entire process is the single copy of the $n\times n$ output buffer from working memory into system memory, where it is then normalized, compressed, and output (we don't consider these parts of the process when measuring performance or overhead, as they're written in Python, a language not known for its performance characteristics). The overhead incurred during this copy can be seen in the chart in figure \ref{fig:overheadPlot} to be negligible --- in the range of a tenth of a percent of the total runtime when working on the GPU, and a one-hundred-thousandth of a percent when working on the CPU, where the data has a shorter distance to travel. \subsection{Strong Scaling} \label{strongScaling} \begin{figure} \includegraphics[width=84.5mm]{strongPlotOne.pdf} \caption{Runtime as core count increases on the 4890 with a fixed resolution ($400\times400$); smaller values are better} \label{fig:strongPlotOne} \end{figure} \begin{figure} \includegraphics[width=84.5mm]{strongPlotTwo.pdf} \caption{Speedup as core count increases (measured against the single-core run) on the 4890 with a fixed resolution ($400\times400$); the orange line is the theoretical maximum speedup, given linear scaling and 50 execution units (800 cores / 16 cores per unit); the red line is a linear speedup curve; larger values are better} \label{fig:strongPlotTwo} \end{figure} In order to measure the strong scalability of our implementation, we fix the problem size and vary the number of processing units used to compute it. For this benchmark, we will fix the image at $400\times400$ pixels, requiring $320,000,000$ total samples, and we will vary the number of cores of the 4890 that we use. One can see in figure \ref{fig:strongPlotTwo} that this implementation is strongly scalable; the speedup is almost perfectly linear as the core count increases. It should be noted that since --- for our kernel --- the 4890 only has 50 compute units, the linear increase is only maintained until we reach 50 cores; after that, the performance gains plateau as they should. \subsection{Large- vs. Small-scale Parallelism} \begin{figure} \includegraphics[width=84.5mm]{runtimePlot.pdf} \caption{Runtime vs. Resolution on all hardware; smaller values are better} \label{fig:runtimePlot} \end{figure} As one can see in figure \ref{fig:runtimePlot}, the GPU significantly outperforms all of the CPUs. This is not surprising, as it can perform more than an order of magnitude more parallel computations, but it does validate the premise of this experiment. \begin{figure} \includegraphics[width=84.5mm]{speedupPlot.pdf} \caption{Speedup of fastest GPU (4890) over fastest CPU (Core 2 Quad) as resolution increases; larger values are better} \label{fig:speedupPlot} \end{figure} Figure \ref{fig:speedupPlot} shows that our implementation converges to an approximately 14.7x speedup (between the fastest of each class of hardware, namely, the 4890 and Core 2 Quad) as the resolution of the output image increases. Noise in the speedup values with smaller images is likely due to measurement error and the overhead and unpredictability involved when copying data to/from video memory. It should be noted that these speedup values are across different pieces of hardware at different core clock speeds, and, as such, do not represent the actual parallel speedup gained (for that, look at section \ref{strongScaling}). This is, instead, a representation of the performance potentially gained by moving computation to the GPU. \subsection{Cost Effectiveness of Hardware} \begin{figure} \includegraphics[width=84.5mm]{dollarPlot.pdf} \caption{Samples per second per dollar with a fixed resolution ($500\times500$); larger values are better} \label{fig:costEffectiveness} \end{figure} \begin{figure} \includegraphics[width=84.5mm]{wattPlot.pdf} \caption{Samples per second per heavy-load watt with a fixed resolution ($500\times500$); larger values are better} \label{fig:wattPlot} \end{figure} One oft-cited measure of the performance of a computational device is its {\it performance-per-dollar}; this is especially of interest for consumers with budgets or people building large computation clusters. In order to measure performance-per-dollar, we have to define {\it performance} in the context of our problem. In all of the plots in this section, performance is measured as the number of samples per second. Figures \ref{fig:costEffectiveness} and \ref{fig:wattPlot} all use a 500$\times$500 image, with 2000 samples per pixel, for $500,000,000$ total samples. Figure \ref{fig:costEffectiveness} illustrates the performance-per-dollar of our four devices. As is quite evident already, the 4890 --- though in the same price range as most of the other devices, and significantly cheaper than the Core 2 Quad --- manages to dominate the chart. It's safe to say that if an institution is looking into buying a large number of a particular device for scientific computation, they should seriously consider looking into high-end graphics cards. Indeed; each dollar spent on a 4890 goes 18 times farther, performance-wise, than it would if spent on the next-most-cost-effective chip, the Core 2 Duo E7200. Another important point when considering the cost of various pieces of hardware is how much it will cost to run. One way to measure this is to consider the power efficiency of the chip, in terms of performance-per-watt. Again using the same measure of performance from our performance-per-dollar comparison, we can see in figure \ref{fig:wattPlot} that the GPU continues to dominate, even though it requires three times more energy to run than its closest competitor, the Core 2 Quad. As one might expect, the older --- thus, less power efficient --- Core 2 Duo falls behind the other CPUs on this scale: though it was the performance-per-dollar winner out of the three CPUs, it would cost significantly more than the other two to run for any period of time. \section{Future Work} There are quite a few things which have come up while developing this project that would be interesting to implement if we were to continue work on the codebase. We will discuss a few of these below: \subsection{Camera Transformation} \label{cameraTransformation} One key feature that is missing from the current implementation is the ability to manipulate the "camera" location, changing your viewport on the orbital cloud. Implementing this would add a bit of complexity, but would allow for interesting video rendering --- for example, video of the camera spinning around the atom, providing a better idea of the actual shape of the cloud. It can be hard --- without prior knowledge --- to comprehend the shape of the cloud from our current visualization. In addition, if the number of samples was sufficiently reduced, or a fast enough video card was obtained, one could implement {\it live} manipulation of the camera's view, which could assist in inspection of the orbital cloud to an even greater extent. \subsection{Slicing} \begin{figure} \includegraphics[width=84.5mm]{320-slice.png} \caption{A slice through the 3-2-0 orbital of the hydrogen atom} \label{fig:slice} \end{figure} Section \ref{cameraTransformation} covered the ideal solution to the problem of comprehending the shape of the orbital from our 2D representation --- however, if that solution proved to be too challenging, one could instead generate an animation of slices through the cloud, each slice being similar to figure \ref{fig:slice}. This would provide another form of interesting visualization of the resulting data, rather similar to the data collected by an MRI device. \subsection{Automatic Scaling} Our OpenCL kernel currently contains an arbitrary zoom factor which needs to be adjusted based on the orbital configuration --- higher energy electron states lead to larger probability clouds. Automatically generating this zoom factor would lead to significantly improved ease of use when changing between different configurations. This could most effectively be implemented by finding a formula for a reasonable scale factor given a set of $n, l, m$ values. \subsection{Coloring} A third way to make the visualization more appealing (or even more informative) would be to provide some manner of colorization. One potential algorithm would be to colorize each sample based on its depth in $z$; the resulting image would then have more depth information, which might increase the ease with which it is interpreted. A variety of other coloring functions are possible, but none were pursued during this project, for time's sake. \subsection{Generalization of Parameters} Equations \ref{laguerre} and \ref{legendre} provide a generalized way to construct Laguerre and Legendre polynomials, respectively. However, our implementation does not make use of these functions, instead using a table of simplified solutions, generated with Mathematica, which is limited in scope. The length of the table determines the range of $n, l, m$ values which can be passed into our program, and is the only restriction on generality within our implementation. If one were to construct these functions on the fly from their generating equations, the algorithm would be completely generalized in terms of $n, l, m$, and would be significantly more useful from an exploratory standpoint. However, as discussed in section \ref{laguerreLegendre}, this would require (at least) the implementation of a symbolic differential equation solver in OpenCL, which could potentially be a very time-consuming task (it could also increase the number of GPU stream processors per kernel required, significantly decreasing the speedup gained). Alternatively, one could write a translation program to automatically generate the table from Mathematica, removing the slow, error-prone human translation step, and very quickly expand the size of the table to something more useful. \subsection{Multiple Electrons} Another feature that would be incredibly cool to implement would be simulation of atoms with more than one electron. In this case, interference between the electrons changes the pattern to be significantly more complex (and much more interesting, as well). It's also a good bit more complex to compute: there is no exact solution of the Schr\"{o}dinger's equation for multiple electrons. Instead, an implementation would depend on a numerical differential equation solver, which would significantly affect performance. \section{Conclusion} After consuming many computer-hours computing the 3-2-0 orbital repeatedly, we've come to a few clear conclusions. Firstly, scientific computation --- the calculation of atomic orbitals, at the very least --- should clearly be a prime target for porting to parallel computation on the GPU, as the field takes off. Projects like Folding@Home are already taking advantage of this (besides being massively parallel, as a web-distributed project) by providing ATI Stream SDK and CUDA ports of their client, and the last few years have seen an explosion in other such projects. We've also found that --- for problems which are embarrassingly easy to parallelize, like this one --- it's very easy to implement your problem using OpenCL. We've used MPI and pthreads in the past, and it seems that OpenCL (perhaps because it's a more modern API) is a bit easier to use in terms of implementation. The added benefit of kernels running on a wider array of hardware (not just the CPU, like MPI, but on the GPU as well) means that OpenCL is a no-brainer --- at least for this sort of simple project. It's quite likely that a problem requiring a large amount of inter-kernel communication would be better suited by something with more communications primitives, like MPI. Indeed, besides {\it barrier}, OpenCL doesn't seem to have any manner of inter-kernel communications functions. \section{Code} All of the code developed for this project is available under the two-clause BSD license, and is hosted on GitHub: \url{http://github.com/hortont424/orbitals} \url{git://github.com/hortont424/orbitals.git} The code has only been tested with the Apple and ATI OpenCL compilers, but should work with few to no changes on NVIDIA's SDK. \bibliographystyle{acmsiggraph} \nocite{*} \bibliography{paper} \end{document}CrazyHsu/wanglab_cau @article{Li_2008_feb, author = { and and and and and and and and and and and and }, doi = {10.1105/tpc.107.056879}, journal = {The Plant Cell}, month = {feb}, number = {2}, pages = {259--276}, publisher = {Oxford University Press (OUP)}, title = {High-Resolution Mapping of Epigenetic Modifications of the Rice Genome Uncovers Interplay between DNA Methylation, Histone Methylation, and Gene Expression}, url = {https://doi.org/10.1105%2Ftpc.107.056879}, volume = {20}, year = {2008} } docs/proposal/Dissertation/Chapters/Intro/sec:techniques.tex0 %% SECTION HEADER ///////////////////////////////////////////////////////////////////////////////////// \section{Chosen \ac{shm} techniques using \acp{pzt}} \label{sec:techniques} %% SECTION CONTENT //////////////////////////////////////////////////////////////////////////////////// \subsection{Guided waves based techniques} \Acp{gw} are mechanical vibrations being a superposition of shear and longitudinal waves propagating in a bounded elastic medium, e.g., bars, beams, rods, plates and shells. Guided waves are multi-modal and dispersive, i.e. more than one mode travels simultaneously through the medium with the phase velocity depending on the frequency. Fig.~\ref{fig:dispersion} shows an example of dispersion curves generated by the Dispersion Calculator~\cite{huber2021dispersion} software tool for a 1 mm thick \ac{cfrp} plate in the frequency range 0-2000 kHz. \Ac{a0} and \ac{s0}, considering the distribution of particle displacements on the upper and lower free surface relative to a central surface, are observed for low frequencies. The mode shapes are pictured in Fig.~\ref{fig:mode_shape}, with the \ac{s0} particle displacements being dominant in-plane, while the \ac{a0} is dominated by out-of-plane. Moreover, higher harmonic modes appear over the cut-off frequency, as shown in Fig.~\ref{fig:dispersion}. \begin{figure} % \begin{center} \includegraphics[width=1\linewidth]{Intro/dispersion} % \end{center} \caption{Dispersion diagram for a 1 mm \ac{cfrp} plate (adopted from Dispersion Calculator~\cite{huber2021dispersion}). Red and blue solid curves represent symmetric and antisymmetric modes, respectively; a black dashed line indicates the cut-off frequency for higher modes.} \label{fig:dispersion} \end{figure} \begin{figure} % \begin{center} \includegraphics[width=1\linewidth]{Intro/mode_shape} % \end{center} \caption{Mode shape of the \textbf{(a)} \ac{a0} and \textbf{(b)} \ac{s0} at 100 kHz for \(\phi=0^{\circ}\) in 2 mm composite plate (exported from Dispersion Calculator~\cite{huber2021dispersion}).} \label{fig:mode_shape} \end{figure} Detection schemes based on \acp{gw} exploit reflection, attenuation, and mode conversion when the propagating wave encounters a discontinuity in the structure \cite{alleyne1992interaction}. Thus, this technique is efficient in detecting various types of defects, such as delamination \cite{sohn2011delamination,tian2015delamination}, adhesive disbonds \cite{rucka2018damage,balasubramaniam2021ultrasonic}, corrosion changes \cite{alleyne1995long,lowe1998defect}, cracks \cite{tua2004detection,lu2006crack,zima2020detection} and failures occurring in \acp{hsc} \cite{mustapha2011assessment, sikdar2016guided, sikdar2016ultrasonic,radzienski2016assessment, yu2019core}. Many techniques based on \ac{gw} propagation have been developed for damage detection and localization. A pitch-catch technique \cite{ihn2008pitch, sikdar2017structural} uses a pair of detached sensors, one excites, and the other receives a signal. If the wave encounters a defect between the sensors, it will scatter, and the recorded signal will be distorted. In the case of the pulse-echo technique \cite{guo1993interaction, kudela2008damage}, there is one sensor that excites the wave and, at the same time, registers possible echoes from the damage. The damage localization can be determined if the wave speed is known and the time of flight is measured. The radar principles were utilized in a phased array technique for plate inspection \cite{giurgiutiu2004embedded, ostachowicz2008elastic, kudela2018structural}. The technique uses an array of transducers, each excited with an appropriate time offset, to focus all the waves at a single grid point of the area to be inspected. A damage map is determined once the signals are obtained and processed for the entire grid. Fink proposed a different approach, developing what he called a time-reversal mirror \cite{fink1992time}. In this method, the wave propagates from one sensor to another, and then after time-reversal and dispersion compensation, the wave is re-emitted to the origin sensor. The resulting signal will be a mirror image of the forcing signal only if the wave does not encounter damage along the way \cite{park2007time, eremin2016analytically}. The \acp{pzt} can be used mutually as actuator-receiver pairs or as a single actuator with other types of devices, e.g. \ac{sldv}, \ac{fbg} sensors. The \acp{pzt} generate high forces with broadband frequency, so methods based on \ac{gw} can detect various damage types of different sizes in a large inspected area. Moreover, specific algorithms do not require a baseline model, and the method implementation is economically efficient. \subsection{Electromechanical impedance methods} \Ac{emi} spectroscopy is also an effective and powerful technique in \ac{shm} for real-time structural damage assessment \cite{park2003overview}. The basis of this method is the influence of the mechanical impedance of the inspected host structure on the electrical impedance of the \ac{pzt} attached to the structure. Assuming that the mechanical property of the sensor remains unchanged over the monitoring period, any changes in measurements of the electrical impedance can be considered a difference in the structure stiffness, which in turn can indicate that a defect has occurred. Fundamentals of the \ac{emi} method were introduced by Liang et al. \cite{liang1994impedance}. An analytical model of a \ac{pzt} actuator bonded to one end of a single degree of freedom mass-spring-damper system was presented in this pioneering work. In the early papers, the authors adopted quasi-static sensor approximation until Giurgiutiu and Zagrai \cite{giurgiutiu2000characterization} derived an expression where the sensor dynamic was incorporated. The dynamics of a single \ac{pzt} with various boundary conditions (free, clamped and elastically constrained) and sensor attached to a beam were considered. Further investigation was performed for the sensor bonded to the host structures was performed \cite{zagrai2001electro, giurgiutiu2005damage}. Damage detection is realized by comparing the state of the structure with the reference state using overall statistical damage indices, e.g., the \ac{rmsd}, the \ac{mapd}, \ac{ccd} and \ac{pnn}. Malinowski et al. \cite{malinowski2014characterisation, malinowski2015use} investigated the effects of \ac{emi} changes related to the state of the adhesive layer between two composite plates. The technique has been used to evaluate weak bonds due to inadequate adhesive curing temperature, release agent and moisture contamination. This type of damage is not detectable using the method based on \ac{gw} propagation. Experimental testing was conducted on weakened samples and compared with a reference. The \ac{rms} of the conductance in the range of 3-5 MHz and the first thickness resonant frequency shift were considered for bond-line assessment. An and Sohn \cite{an2012integrated} proposed a new damage detection technique that combines \ac{emi} and \ac{gw} advantages. In the method, measured admittance characteristic is separated into two parts: active and passive. \Ac{di} is a weighted sum of two indicators obtained from \ac{gw} signal and active admittance. Because passive impedance is only sensitive to temperature variation, it is used for temperature compensation on both mentioned signals. Instead of two \acp{di}, Sevillano et al. \cite{sevillano2016damage} proposed more integrated \ac{di} based on the electromechanical power dissipation of the \ac{pzt} sensor. The \ac{emi} technique can detect damage, such as delamination or cracks, but is also sensitive to changes, such as weak bonds, which the \ac{gw} method is ineffective at detecting. However, the \ac{emi} is a local method for the high frequency. Giurgiutiu et al. \cite{giurgiutiu2001electro} obtained consistent results for crack detection in distances up to 40 mm in the frequency range 300-450 kHz. This method is also sensitive to environmental conditions such as temperature and humidity fluctuations\cite{bhalla2002practical} or loading variations \cite{lim2011impedance}. %\subsection{Modal analysis techniques}capitulos/6-gerencia.tex \chapter{Tópicos de Gerência de Requisitos} \label{management} \section{Rastreabilidade} \subsection{Introdução ao Conceito de Rastreabilidade} A rastreabilidade advém da necessidade de identificar as dependências entre os artefatos, já que eles relacionam-se e afetam-se ao longo do desenvolvimento do \textit{software}. É notável que os requisitos mantém também dependência entre si de forma que alguns só podem ser implementados após outros. Sendo assim, é necessário identificar e controlar essas relações entre os requisitos com o intuito de evitar possíveis erros catastróficos e complicações para o desenvolvimento. A rastreabilidade é precisamente o que permite obter essa identificação e controle. ~\cite{gotel} definem a rastreabilidade de requisitos como “a capacidade de descrever e seguir a vida de um requisito, em ambas as direções para frente e para trás(\textit{forward and backward direction}), isso é, desde as suas origens, através do seu desenvolvimento e especificação, à sua subsequente implementação e utilização”. Rastrear um requisito para frente (\textit{forward traceability}) significa “seguir as ligações de rastreabilidade para os artefatos que foram derivados do artefacto em consideração”.~\cite{winkler} Rastrear um requisito para trás (\textit{backward traceability}) “refere-se à capacidade de seguir as ligações de rastreabilidade de um artefato específico de volta às suas origens de onde foi derivado”.~\cite{winkler} A figura \ref{back-and-foward} demonstra as pré-rastreabilidades e pós-rastreabilidades, introduzidas por ~\cite{gotel}. \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.5] {figuras/rastreabilidade_frente_tras.eps} \caption{Rastreabilidade pra frente e pra trás ~\cite{dahlstedt}} \label{back-and-foward} \end{figure} \clearpage{} \textbf{Pré-rastreabilidade}: “refere-se aos aspetos da existência de um requisito antes de ser incluído na especificação de requisitos” ~\cite{gotel} e “está focada em permitir uma melhor compreensão dos requisitos” ~\cite{persson}. “A pré-rastreabilidade é a base para gerir a evolução de um sistema, porque permite o levantamento das partes de especificação que são afetadas por uma mudança específica no pedido suscitado”.~\cite{persson} \textbf{Pós-rastreabilidade}: “refere-se aos aspetos da existência de um requisito a partir do momento em que foi incluído na especificação de requisitos” ~\cite{gotel} e “está focada em permitir uma melhor compreensão e aceitação do atual \textit{software} de sistema”. ~\cite{persson} Dentre os estudos da rastreabilidade de \textit{software}, ~\cite{ramesh} introduziram a rastreabilidade horizontal e vertical. \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.5] {figuras/rastreabilidade_horizontal.eps} \caption{Rastreabilidade horizontal e vertical ~\cite{genvigir}} \label{sideways} \end{figure} \clearpage{} \textbf{Rastreabilidade Horizontal}: trata de relacionar versões ou variantes do mesmo tipo de informação, por exemplo, entre requisitos ou entre componentes do sistema. ~\cite{persson}~\cite{winkler} \textbf{Rastreabilidade Vertical}: preocupa-se em rastrear informação entre anteriores e subsequentes fases no processo de desenvolvimento, isto é, entre objetos de informação de diferentes tipos ~\cite{persson}~\cite{winkler}. Por exemplo, uma relação entre um requisito e um elemento da conceção.~\cite{winkler} \subsection{Sobre a Rastreabilidade no RUP} Pelo própria definição de rastreabilidade do processo unificado: “A rastreabilidade é a capacidade de rastrear um elemento de projeto para outros elementos de projeto relacionados, especialmente aqueles relacionados a requisitos. Os elementos de projeto envolvidos na rastreabilidade são chamados de itens de rastreabilidade. Os itens de rastreabilidade típicos incluem diferentes tipos de requisitos, elementos de modelo de análise e \textit{design}, produtos de trabalho de teste e materiais de treinamento e documentação de suporte ao usuário final”.~\cite{rup1} A figura \ref{common-itens} a seguir demonstra esses itens normalmente utilizados na rastreabilidade de um \textit{software}. \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.7] {figuras/itens_comuns.eps} \caption{Itens normalmente utilizados na rastreabilidade.~\cite{rup1}} \label{common-itens} \end{figure} \clearpage{} Os itens tipicamente utilizados são: \begin{itemize} \item Necessidades dos envolvidos, documentados no Documento de Visão. \item Recurso do produto, documentados no Documento de Visão. \item Requisito suplementar, documentados em Especificações Suplementares. \item Casos de uso, documentados no Documento de Casos de Uso. \item Seção de caso de uso, documentados no Documento de Casos de Uso. \item Elemento de Design documentados no documento de Modelo de \textit{Design}. \item Conjunto de Teste, documentados no documento de Caso de Teste. \end{itemize} Uma rastreabilidade típica pode ser vista pela figura \ref{common-tracebility} abaixo: \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.9] {figuras/rastreabilidade_tipica.eps} \caption{Rastreabilidade típica utilizada no RUP.~\cite{rup1}} \label{common-tracebility} \end{figure} \clearpage{} Existem várias técnicas para realizar a rastreabilidade no processo unificado, dentre as mais comuns estão as referências cruzadas e o uso de documentos. As técnicas que utilizam as referências cruzadas são mais simples de serem implementadas e compreendidas, podendo utilizar numerações, indexação, \textit{tags} e matrizes de rastreabilidade. Já as técnicas que utilizam documentos, costumam usar \textit{templates} para os documentos e artefatos. No caso de referências cruzadas, pode-se utilizar uma matriz de rastreabilidade, gerada por um \textit{software} de uso geral como uma planilha no Excel® ou em um \textit{software} específico como o Rational® RequisitePro®. Um exemplo de matriz de rastreabilidade pode ser visto pela tabela \ref{table1}. \begin{table}[h] \centering \begin{tabular}{|l|l|l|l|l|} \hline \multicolumn{5}{|c|}{\textbf{Projeto \textless nome\textgreater - Matriz de Rastreabilidade}} \\ \hline \textbf{Requisito} & \textbf{Documento Fonte} & \multicolumn{1}{l|}{\textbf{Arquitetura}} & \textbf{Componente} & \textbf{Caso de Teste} \\ \hline & & & & \\ \hline & & & & \\ \hline \end{tabular} \caption{Rastreabilidade entre requisitos e seus correspondentes artefatos gerados} \label{table1} \end{table} O requisito pode ser expresso em linguagem natural e numerado sequencialmente. Nas demais colunas, são indicados os artefatos relacionados ao requisito, onde a ordem correspondente é sempre de 1 para 1. A mesma técnica pode ser utilizada para referenciar dependência entre requisitos, como pode ser observado pela figura \ref{matrix_requirement}. \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.7] {figuras/matriz_rastreabilidade2.eps} \caption{Matriz de rastreabilidade entre Requisitos Funcionais e Não Funcionais.} \label{matrix_requirement} \end{figure} \clearpage{} \subsection{Rastreabilidade Adotada no Projeto} Considerando a necessidade de cada requisito ser rastreável à sua fonte e seu uso durante o projeto em implementações e testes, por exemplo, foi escolhida a utilização da rastreabilidade vertical especificada abaixo. Distribuição dos artefatos gerados: \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.7] {figuras/arvore_rastreabilidade.eps} \caption{Árvore de rastreabilidade} \label{matrix_requirement} \end{figure} Identificação dos itens dos artefatos: \begin{table}[] \centering \label{artefacts-identifier} \begin{tabular}{|l|l|} \hline \textbf{Identificador} & \textbf{Descrição} \\ \hline RN\textless número\textgreater & Regra de Negócio \\ \hline RF\textless número\textgreater & Requisito Funcional. Ex: RF0014 \\ \hline RNF\textless número\textgreater & Requisito Não Funcional. EX:RNF0365 \\ \hline UC\textless número\textgreater & Caso de Uso. Ex:US1234 \\ \hline UCD\textless número\textgreater & Diagrama de Caso de Uso \\ \hline TC\textless número\textgreater & Caso de Teste. \\ \hline \end{tabular} \caption{Identificação dos artefatos} \end{table} \begin{itemize} \item \textbf{Atributos de Requisitos}: Características utilizadas para avaliar um requisito. Serão utilizados: \begin{itemize} \item Prioridade \item \textit Status \item Dificuldade \end{itemize} \item \textbf{Prioridade}: Define o quão importante é o requisito para o sistema em relação aos demais. Os de maior prioridade serão implementados antes em relação aos demais, isso para todos os níveis de prioridade, que são: \begin{itemize} \item Alta \item Média \item Baixa \end{itemize} \item \textbf{\textit{Status}}: Seguindo o mesmo exemplo de prioridade, em \textit{status} também serão utilizados 3 níveis que irão ditar o progresso das atividades relacionadas à um requisito: \begin{itemize} \item Completado \item Em andamento \item Não inicializado \end{itemize} \item \textbf{Dificuldade}: Serve de indicativo para uma possível dificuldade na implementação dos requisitos. É um atributo fundamental para o planejamento das iterações. Também será dividido em 3 níveis: \begin{itemize} \item Alta \item Média \item Baixa \end{itemize} \end{itemize} Para servir de exemplo, segue-se uma tabela de atributos: \begin{table}[h] \centering \label{requirements-qualities} \begin{tabular}{|l|l|l|l|l|} \hline \textbf{Identificador} & \textbf{Nome} & \textbf{Prioridade} & \textbf{Dificuldade} & \textbf{Status} \\ \hline & & & & \\ \hline \end{tabular} \caption{Atributos dos Requisitos} \end{table} Para ter um maior controle sobre o fluxo do relacionamento entre os artefatos e os requisitos, será utilizada uma matriz de rastreabilidade como o modelo abaixo: \begin{table}[h] \centering \label{itens-traceability} \begin{tabular}{|l|l|l|l|l|} \hline \multicolumn{5}{|c|}{\textbf{Alvos}} \\ \hline \multirow{4}{*}{\textbf{Fontes}} & F/A & A1 & A2 & A3 \\ \cline{2-5} & F1 & X & & X \\ \cline{2-5} & F2 & & X & \\ \cline{2-5} & F3 & & & X \\ \hline \end{tabular} \caption{Modelo de como será a disposição dos itens na matriz de rastreabilidade} \end{table} Segue um exemplo também de uma matriz de rastreabilidade entre casos de uso e requisitos: \begin{figure}[htb] \centering \includegraphics[keepaspectratio=true,scale=0.5] {figuras/casos_de_uso.eps} \caption{Casos de Uso e Requisitos.~\cite{dev1}} \label{uc-requirements} \end{figure} \section{Modelos de Maturidade} \subsection{Introdução à Modelos de Maturidade} Um modelo de maturidade é responsável por analisar e avaliar cada processo de uma estrutura organizacional de forma a melhorar o mesmo em cada nível da empresa avaliada. Em outras palavras, ele é responsável por avaliar a capacidade de processos na realização de seus objetivos, localizar oportunidades de melhoria de produtividade, redução de custos e planejar e monitorar as ações de melhoria contínua dos processos empresariais. \subsection{CMMI} “O CMMI é uma metodologia criada pela \textit{Software Engineering Institute} (SEI) para ser um guia destinado a melhorar os processos organizacionais e habilidade desses em gerenciar o desenvolvimento, a aquisição e a manutenção de produtos e serviços. O CMMI organiza as práticas, que já são consideradas efetivas, em uma estrutura que visa auxiliar a organização a estabelecer prioridades para melhoria e também fornece um guia para a implementação dessas melhorias".~\cite{dev3} O CMMI traz consigo grandes possibilidades para melhoria em projetos, elevando o nível da qualidade e cumprindo melhor os prazos estipulados. O CMMI se divide em 5 estágios: \begin{itemize} \item Otimização \item Quantitativamente Gerenciado \item Definido \item Gerenciado \item Inicial \end{itemize} Os níveis em que a engenharia de requisitos está presente são Gerenciado e Definido. Nesses níveis, a ER subdivide em 6 processos: \begin{enumerate} \item Gerência de Requisitos \item Desenvolvimento de Requisitos \item Solução Técnica \item Integração do Produto \item Verificação \item Validação \end{enumerate} \subsubsection{Vantagens e Desvantagens} \begin{itemize} \item Vantagens \begin{itemize} \item Reconhecimento internacional \item Eliminação de inconsistências \item Utilização de terminologia comum \item Desenvolvimento de \textit{software} com qualidade, garantindo cumprimento de prazos e atendendo às necessidades dos clientes \item Consistência com a norma ISO-15504 (norma que define o processo de desenvolvimento de \textit{software} de acordo com os níveis de capacidade para cada processo de acordo com o CMMI~\cite{cortes} \item Integração de sistemas \item Métricas e análises \item Identificação de riscos \end{itemize} \item Desvantagens \begin{itemize} \item Não há nenhuma empresa brasileira certificada pelo SEI para auditoria, sendo necessário que a certificação seja criada nos Estados Unidos e trazida por um \textit{lead assessor} \item Os preços cobrados pela avaliação do CMMI não são acessíveis. Por esse motivo, pequenas e médias empresas tem problema em adotar esse modelo. \end{itemize} \end{itemize} \subsection{MPS-BR} “É um movimento de melhoria de software voltado para a realidade brasileira. O programa é coordenado pela Associação para promoção de \textit{Software} brasileiro (SOFTEX) com início de desenvolvimento em 2003.”~\cite{softex} É voltado para a realidade brasileira, além disso, é adequado para micro, pequenas e médias empresas alcançarem melhorias no processo de construção de um \textit{software}. Os níveis definidos no MPS-BR são: \begin{itemize} \item A - Em otimização \item B - Gerenciado Quantitativamente \item C - Definido \item D - Largamento Definido \item E - Parcialmente Definido \item F - Gerenciado \item G - Parcialmente Gerenciado \end{itemize} O foco da ER está nos níveis G e D. \subsubsection{Vantagens e Desvantagens} \begin{itemize} \item Vantagens \begin{itemize} \item Adequado à realidade brasileira (implementação acessível à micro, pequenas e médias empresas) \item Possui compatibilidade com o CMMI (baseado nas normas ISO 12207 e 15504) \item Integração universidade-empresa \item Avaliações feitas por empresas brasileiras \item Custo relativamente pequeno (cerca de 72 mil reais para uma empresa nível G ser habilitada como nível F) \item Avaliação rápida (feitas em até 5 dias) \end{itemize} \item Desvantagens \begin{itemize} \item A certificação não é competitiva o suficiente para tornar a empresa reconhecida internacionalmente \item Foco principal em pequenas empresas \end{itemize} \end{itemize} \subsection{Por que utilizar o MPS-BR para modelar o processo da Engrena?} Enquanto o CMMI oferece competitividade e reconhecimento internacional, ele é um modelo extremamente caro para utilizar em um ambiente de Empresa Júnior. Já o MPS-BR é adaptado à realidade brasileira, facilitando a implementação gradual dos processos de gerência, focado na gerência de requisitos e desenvolvimento de requisitos. Implementação essa que conta com preços mais acessíveis e objetivos esperados, com divisões claras que facilitam a visibilidade dos resultados da melhoria de processos em prazos mais curtos.\ProvidesPackage{UofRdissertation} % these were all the packages I needed to write my own dissertation in Physics. You might need more, I recommend including them directly in the following list: \usepackage[table,xcdraw]{xcolor} \usepackage{eso-pic, graphicx} \usepackage{amsmath} \usepackage{amssymb} \usepackage{bm} \usepackage{setspace} \usepackage{float} \usepackage{subfiles} \usepackage{lipsum} \usepackage[]{algorithm2e} \usepackage{nicefrac} \usepackage{wrapfig} \usepackage[hidelinks]{hyperref} % University of Rochester requires main body to have the equivalent text size of 12 point Times New Roman or 10 point Arial. \usepackage{fontspec} %\setmainfont{Arial} \setmainfont[SizeFeatures={Size=12}]{Times New Roman} \newfontfamily\titlefont[SizeFeatures={Size=32}]{Times New Roman} % Fonts in figures and tables may be smaller than 11 point, but all text must remain legible if reduced to 50% size. % University of Rochester requires the page numbers to appear in the top-margin, preferably in the top-right at least 0.75 inches from each edge \usepackage{fancyhdr} \fancyhf{} % make sure header and footer format is empty \renewcommand{\headrulewidth}{0pt} % get rid of horizontal line that appears by default \fancypagestyle{uofrthesis}{ \rhead{\thepage} } \let\ps@plain\ps@uofrthesis % override "plain" page style which is used by default in the table of contents and elsewhere with our custom style % Bibliography Style \bibliographystyle{unsrt} % University of Rochester requires at least 1.25 inch margins on all sides \usepackage[letterpaper,top=1.25in, bottom=1.25in, outer=1.25in, inner=1.25in]{geometry} %Rename "Contents" to "Table of Contents" \renewcommand{\contentsname}{Table of Contents} %So we can build chapters independently if desired \usepackage{subfiles} \subsection{Syntax} The syntax of networking in Casanova is rather simple. In the following we only provide an intuitive illustration of the terms that can be used, and a first description of their purpose. The first series of supported keywords are those that are used for determining ownership of entities. The keywords below are all used to delimit the \textit{scope} within which a given set of rules is valid: \begin{lstlisting} local { ... } remote { ... } \end{lstlisting} Every entity in Casanova is duplicated across all the current instances of the game. Only one of this instances has ownership of the entity, that is acts as the authoritative instance which updates the entity for all other instances. The rules that perform such updates, and which also send the updates to the other instances, are all defined inside a \texttt{local} block. The rules that are executed in all the other, remote, instances are all defined inside a \texttt{remote} block. Another keyword, which is used nested in \texttt{local} or \texttt{remote}, is: \begin{lstlisting} connect { ... } \end{lstlisting} When a new instance of the game connects, then we also run, just once, all the rules inside the \texttt{connect} block. When a new instance is run, then its \texttt{local connect} rules are run once. When existing instances, on the other hand, witness the start of a new instance, then their \texttt{remote connect} rules are run once. There are only four primitives for transmitting data across the network. Two are for sending data, and two are for receiving. Sending simply takes as input a value of any type, and returns nothing. Sending may also be done reliably, thereby trying to ensure that the other party has indeed received the message. Reliable sending may fail, for example if the receiver disconnects during the transmission. For this reason, sending reliably returns a boolean value that will be \texttt{true} if the transmission was successful, and \texttt{false} if the transmission failed for some reason: \begin{lstlisting} send : T -> Unit send_reliable -> bool \end{lstlisting} As a convenience, it is possible to, at the same time, locally store a value and send it across the network. For this purpose, we can use the \texttt{+send} syntax, which both \textit{sends and returns} the expression that is passed as an argument to \texttt{+send}: \begin{lstlisting} yield +send(x) \end{lstlisting} Receiving may also be done in two different manners. Simple reception of a message is done with the \texttt{receive} primitive that returns a value of some type \texttt{T}. Receiving may also be done by all other instances at the same time, for example when voting or for other kinds of global synchronization. In this case, we wait until all other instances have each sent their value of some type \texttt{T}, and then we return all said values in a list of \texttt{T}s: \begin{lstlisting} receive : Unit -> T receive_many : Unit -> List \end{lstlisting} \subsection{Semantics} In the following we discuss the semantics of Casanova networking, but not in formal terms. Rather, we suggest a translation from Casanova with networking into Casanova without networking, assuming that one of the external libraries that Casanova is using is providing some low-level networking service. Networking in Casanova is based on two separate systems. The first such system is the underlying networking library that is accessed as an external service to be orchestrated. This is directly \textit{linked} to all programs that need networking functionalities. Multiple versions of this service may exist for different networking libraries, but in general we can assume that not many such services need to be built, and that there is a one-to-many relationship between networking services and actual games. The second system is the Casanova compiler itself, which modifies entity declarations and even defines whole new entities. The compiler will, effectively, translate away all networking operations and keywords (even \texttt{local}, \texttt{remote}, and \texttt{connect}) and turn them into much simpler operations on lists. The only assumption made that really has anything to do with networking is that some special memory locations are actually written to or read from the network. \subsubsection{Common primitives} We now present the common primitives that are provided by the networking service. The core of the networking service is the \texttt{NetManager}. The \texttt{NetManager} maintains the connections between the local instance of the game and the remote instances: \begin{lstlisting} entity NetManager = { \end{lstlisting} The \texttt{NetManager} maintains a list of \texttt{NetPeer}s. Each \texttt{NetPeer} represents a remote instance of the game. The \texttt{NetManager} also store the unique \texttt{Id} associated with the local instance: \begin{lstlisting} Peers : List Id : PeerId \end{lstlisting} The \texttt{NetManager} also manages two flags, which will be used to determine when the \texttt{local connect} and the \texttt{remote connect} rules are run: \begin{lstlisting} localConnect : bool remoteConnect : bool \end{lstlisting} The \texttt{localConnect} flag may run only once, at the beginning of the game. The game world will then reset the flag to \texttt{false} when the \texttt{local connect} rules are all terminated. Whenever a new connection is established with a new \texttt{NetPeer}, then we add that peer to the list of \texttt{Peers}, and we set \texttt{remoteConnect} to \texttt{true} so that the local \texttt{remote connect} rules may be run. Notice that we wait for \texttt{remoteConnect} to be set to false, which is only done by the game world when the current \texttt{remote connect} rules all terminate. This allows us to process all new connections one at a time: \begin{lstlisting} rule Peers, remoteConnect = wait_until(remoteConnect = false) let new_peer = wait_some(NetPeer.NewPeer()) yield Peers + new_peer, true \end{lstlisting} Every few seconds, we check which peers disconnected and remove them from the list of peers. The disconnection is all managed internally by the underlying networking library: \begin{lstlisting} rule Peers = wait 5.0f from p in Peers where p.Channel.Connected select p \end{lstlisting} We initialize the \texttt{NetManager} by finding all reachable peers across the network. We use their current \texttt{Id} values to find an \texttt{Id} for this instance that is unique to this game session. We also set \texttt{localConnect} to true, since we need to send the locally managed values to the other instances, and \texttt{remoteConnect} to false: \begin{lstlisting} Create() = let peers = NetChannel.FindPeers() { Id = from p in peers max_by p.Id + 1 Peers = peers localConnect = true remoteConnect = false } } \end{lstlisting} Another, remote instance of the game is represented by the \texttt{NetPeer}. A \texttt{NetPeer} is responsible for handling the actual communication to the other instances of the game: \begin{lstlisting} entity NetPeer = { \end{lstlisting} The \texttt{NetPeer} contains a channel, which is an instance of a data-type supplied by some network library and which will, automatically, send and receive messages. The \texttt{NetPeer} also contains an \texttt{Id} which uniquely identifies it among the various instances of the game, and a list of messages received so far: \begin{lstlisting} Channel : NetChannel Id : PeerId ReceivedMessages : List \end{lstlisting} The list of messages received so far is constantly refreshed with the list of received messages automatically populated by the channel: \begin{lstlisting} rule ReceivedMessages = yield Channel.ReceivedMessages \end{lstlisting} The \texttt{NetPeer} also looks for all the messages that need to be sent across the game world, both reliably and unreliably. These messages are then written into the \texttt{SentMessages} and \texttt{ReliablySentMessages} lists of the underlying channel: \begin{lstlisting} rule Channel.SentMessages = from (m:OutMessage) in * where exists(Id, m.Targets) || m.Targets = [] select m rule Channel.ReliablySentMessages = from (m:ReliableOutMessage) in * where exists(Id, m.Targets) || m.Targets = [] select m } \end{lstlisting} The underlying networking library is also expected to provide a series of data types which represent messages and channels. We do not care about the concrete shape of the data types, as long as they contain the required properties. The simple message only needs to handle the \textit{Casanova header}, which stores which instance of the game sent this message, what type of data the message contains, and the entity from which this message was sent: \begin{lstlisting} interface Message Sender : PeerId ContentType : TypeId OwnerEntity : EntityId \end{lstlisting} An outgoing message inherits from \texttt{Message}. It also has a list of target instances to which this message is addressed. The list of targets may also be empty, in which case we wish to send the message to all reachable peers. An \texttt{OutMessage} also offers a series of low-level write methods to send various primitive values such as integers, floating-point numbers, strings, etc.: \begin{lstlisting} interface OutMessage inherit Message Targets : List member WriteInt : int -> Unit member WriteFloat : float32 -> Unit member WriteString : string -> Unit member WriteT : T -> Unit // only for elementary data-types \end{lstlisting} Almost identical to the \texttt{OutMessage} is the \texttt{ReliableOutMessage}. A reliable outgoing message only differs from a simple outgoing message in that it also has properties that tells us whether or not the message has been received or the transmission has failed: \begin{lstlisting} interface ReliableOutMessage inherit Message Targets : List member WriteInt : int -> Unit member WriteFloat : float32 -> Unit member WriteString : string -> Unit ... member Received : bool member Failed : bool \end{lstlisting} A received message inherits from the simple \texttt{Message}, and also offers a series of low-level write methods to read various primitive values such as integers, floating-point numbers, strings, etc.: \begin{lstlisting} interface InMessage inherit Message member ReadInt : Unit -> int member ReadFloat : Unit -> float32 member ReadString : Unit -> string ... \end{lstlisting} The final data-type that is provided by the networking library is the communication channel itself. Casanova requires a channel to expose the messages which were just received, and lists where the messages to be sent can be put. Also, the channel should provide a (static) mechanism to find those peers that just connected: \begin{lstlisting} interface NetChannel member ReceivedMessages : List member SentMessages : List member ReliablySentMessages : List static member FindPeers : Unit -> List \end{lstlisting} Notice that in the listings above we have slightly abused the notion of \textit{interface}. We have used a notion that resembles more closely that of a \textit{type-trait} or a \textit{type-class}, but the abuse is quite minor and we believe the idea of an interface to capture the essence of what Casanova expects from the underlying library. \subsubsection{Chat sample translated} Inside an application, the compiler generates a series of additional entities and modifies the game rules in order to accommodate networking primitives. The generated entities are all wrappers for messages, both incoming and outgoing. A pair of incoming/outgoing message entities is created for each type \texttt{T} such that a \texttt{send} and a \texttt{receive} appear in the game rules. In the case of the chat sample, we only ever send strings, so only one such pair is generated. One generated entity inherits from \texttt{InMessage} and contains a string value which was just received: \begin{lstlisting} entity InMessageString = { inherit InMessage Value : string \end{lstlisting} When creating an \texttt{InMessageString}, we take a simpler \texttt{InMessage}, ``parse''\footnote{\textit{Parsing} in this context is a bit of an excess.} it by invoking \texttt{ReadString} once, and then store message and string: \begin{lstlisting} Create(msg : InMessage) = let value = msg.ReadString() { InMessage = msg Value = value } } \end{lstlisting} The dual of the entity we have just seen is the \texttt{ReliableOutMessageString}, which inherits from the simpler \texttt{ReliableOutMessage}: \begin{lstlisting} entity ReliableOutMessageString = { inherit ReliableOutMessage \end{lstlisting} When we create a \texttt{ReliableOutMessageString}, in reality we create a \texttt{ReliableOutMessage} and write the content of the message (a string) to it with \texttt{WriteString}: \begin{lstlisting} Create(value : string, targets : List, sender : ConnectionId, owner_entity : EntityId) = let m = new ReliableOutMessage(sender, StringTypeId, owner_entity, targets) do m.WriteString(m) { ReliableOutMessage = m } } \end{lstlisting} At this point we can move on to the definition of the game world. The first two fields are identical to the sample as we have seen it in previously. Local rules that do not \texttt{send} or \texttt{receive} are unchanged: \begin{lstlisting} world Chat = { Text : string Line : string ... \end{lstlisting} The compiler also adds a series of additional fields. One of the fields is a network manager, which will manage the various connections. Two lists, one for incoming and one for outgoing messages are also declared. Finally, an \texttt{Id} is used to store a unique identifier for this specific entity: \begin{lstlisting} Network : NetManager Inbox : List Outbox : List Id : EntityId \end{lstlisting} We automatically empty the list of outgoing messages, in the assumption that the \texttt{NetPeer} instances have already stored them and are handling them: \begin{lstlisting} rule Outbox = yield [] \end{lstlisting} We fill the list of incoming messages from the incoming messages found in the channels of the various peers. We filter those messages, so that only those that were specifically aimed towards this entity (and contain data of the expected type, in our case \texttt{string}) are kept: \begin{lstlisting} yield Inbox + from c in Network.Peers from m in c.ReceivedMessages where m.ContentType = StringTypeId && m.OwnerEntity = Id select InMessageString.Create(m.Value) \end{lstlisting} When we want to send a string, we also create a message with the string we wish to send, add it to the \texttt{Outbox} list, and wait until the message is received. The rest of the rule is unchanged: \begin{lstlisting} rule Line, Text, Outbox = wait_until(IsKeyDown(Keys.Enter)) let msg = ReliableOutMessageString.Create(Line, [], Network.Id, Chat.TypeId, Id).ReliableOutMessage yield Outbox <- Outbox + msg wait_until(msg.Received) yield Line <- "", Text <- Text + "\n" + Line \end{lstlisting} Essentially, \texttt{send\_reliably} turns into the following lines: \begin{lstlisting} let msg = ReliableOutMessageString.Create(Line, [], Network.Id, Id).ReliableOutMessage yield Outbox <- Outbox + msg wait_until(msg.Received) \end{lstlisting} In particular, we create the output message by also specifying: \begin{itemize} \item the message recipients, which are the empty list \texttt{[]} which means that the message will be sent to all other peers \item the peer that is the sender (and owner) of the message, which is \texttt{Network.Id} \item the entity that the message was sent from, which is simply the world \texttt{Id} \end{itemize} We consider a message received when it appears in the \texttt{Inbox} list. When we find one, we remove it from the list, and process it as the result of the \texttt{receive} function: \begin{lstlisting} rule Text, Inbox = wait_until (Inbox.Length > 1) yield Text + "\n" + Inbox.Head.Value, Inbox.Tail \end{lstlisting} Essentially, \texttt{receive} has turned into: \begin{lstlisting} wait_until (Inbox.Length > 1) Inbox.Head.Value // the received string \end{lstlisting} The creation of the world now simply initializes the network manager, creates a new unique id for the entity, and then initializes the various other fields with empty values: \begin{lstlisting} Create() = let network = NetManager.Create() let id = NetManager.NextId() { Text = "" Line = "" Id = id Network = network Inbox = [] Outbox = [] StringsInbox = [] } } \end{lstlisting} As we can see from the sample, it is possible to translate away the networking primitives, provided a very small component capable of sending and receiving messages created from an aggregation of elementary data structures. \subsubsection{Lobby sample translated} In this more complete example we also see how local and remote blocks are handled, both at connection time and during the main run. The lobby sample generates four entities for (reliably) sending and receiving \texttt{bool} and \texttt{LobbyPlayer} values. The \texttt{bool} message entities are almost identical to the \texttt{string} ones that we have seen in the chat sample, and so we omit them. The \texttt{LobbyPlayer} message entities, on the other hand, are more articulated. A received \texttt{LobbyPlayer} will have an underlying incoming message and the \texttt{LobbyPlayer} itself: \begin{lstlisting} entity InMessageLobbyPlayer = { inherit InMessage Value : LobbyPlayer \end{lstlisting} When we create the \texttt{InMessageLobbyPlayer}, we parse its contents. First we read the player \texttt{name}, then its \texttt{ready} status, and then the \texttt{X} and \texttt{Y} of its starting position. The \texttt{id} of the player is stored in the underlying message, so we do not need to read it again and can reuse it directly. Finally, the owner of the entity is its sender, and the received entity will thus be remoted to the underlying message sender: \begin{lstlisting} Create(msg : InMessage) = let name = msg.ReadString() let ready = msg.ReadBool() let start_pos = Vector2(msg.ReadFloat(), msg.ReadFloat()) let id = msg.EntityId let ownership = remote(msg.Sender) { InMessage = msg Value = { Name = name Ready = ready StartPosition = start_pos Id = id InboxBool = [] Outbox = [] Ownership = ownership } } } \end{lstlisting} Similarly, we define the \texttt{ReliableOutMessageLobbyPlayer} as a wrapper over the simpler \texttt{ReliableOutMessage}: \begin{lstlisting} entity ReliableOutMessageLobbyPlayer = { inherit ReliableOutMessage \end{lstlisting} When we send a \texttt{LobbyPlayer} we first create the underlying message with the Casanova header of sender, owner entity, etc. Then, we perform a series of \texttt{write} operations that mirror the \texttt{read} operations in the \texttt{InMessageLobbyPlayer}: \begin{lstlisting} Create(value : LobbyPlayer, targets : List, sender : ConnectionId, owner_entity : EntityId) = let m = new ReliableOutMessage(sender, LobbyPlayerTypeId, owner_entity, targets) do m.WriteString(value.Name) do m.WriteBool(value.Ready) do m.WriteFloat(value.StartPosition.X) do m.WriteFloat(value.StartPosition.Y) { ReliableOutMessage = m } } \end{lstlisting} The \texttt{Lobby} itself contains the same fields that store the various players: \begin{lstlisting} world Lobby = { Self : LobbyPlayer Others : List \end{lstlisting} Additionally, the compiler generates a series of networking-related fields. The entity has a networking manager, and \texttt{id}, and a series of lists for storing incoming and outgoing messages. In particular, it may seem as we receive \texttt{LobbyPlayers} twice, but we actually store the received lobby players for both \texttt{receive} and \texttt{receive\_many}: \begin{lstlisting} Network : NetManager Id : int InboxLobbyPlayer : List InboxListLobbyPlayer : List Outbox : List \end{lstlisting} Just like we did for the chat, we empty the list of outgoing messages, and fill in the lists of \texttt{LobbyPlayer} messages from the various peers: \begin{lstlisting} rule Outbox = yield [] rule InboxListLobbyPlayer = yield InboxListLobbyPlayer + from c in Network.Peers from m in c.ReceivedMessages where m.ContentType = LobbyPlayerTypeId && m.OwnerEntity = Id select InMessageLobbyPlayer.Create(m.Value) rule InboxLobbyPlayer = yield InboxLobbyPlayer + from c in Network.Peers from m in c.ReceivedMessages where m.ContentType = LobbyPlayerTypeIdTypeId && m.OwnerEntity = Id select InMessageLobbyPlayer.Create(m.Value) \end{lstlisting} The \texttt{local connect} waits until the network manager sets its \texttt{localConnect} flag to \texttt{true}. This will only allow the rule to run once upon first connection, and then stop: \begin{lstlisting} rule Self, Others, InboxListLobbyPlayer, Outbox, Network.localConnect = wait_until(Network.localConnect = true) \end{lstlisting} We now perform a \texttt{receive\_many} by waiting until all peers have sent us something. We then take one received \texttt{LobbyPlayer} per peer: \begin{lstlisting} wait_until(from m in InboxListLobbyPlayer select m group_by m.Sender count = Network.Peers.Length) let others = // take one item per peer from m in InboxListLobbyPlayer select m group_by m.Sender as g select g.Elements.Head.Value yield InboxListLobbyPlayer <- [] \end{lstlisting} We then send our own \texttt{LobbyPlayer}, and wait until it has been received by all: \begin{lstlisting} let max_x = maxby p in others select p.StartPosition.X let start_position = Vector2(max_x + 5.0f, 0.0f) let self = { Self with Position = start_position } let msg = ReliableOutMessageLobbyPlayer.Create(Self, [], Network.Id, Id).ReliableOutMessage yield Outbox <- Outbox + msg wait_until(msg.Received) \end{lstlisting} Finally, we store the players locally, and reset \texttt{localConnect} to \texttt{false}. A very important notice is that, in case of multiple \texttt{local connect} rules, then we need to wait until all of them are done before resetting \texttt{localConnect}. This will require additional boolean flags: \begin{lstlisting} yield Self <- self, Others <- others, Network.localConnect <- false \end{lstlisting} The \texttt{remote connect} waits until the network manager sets its \texttt{remoteConnect} flag to \texttt{true}. This will only allow the rule to run once for every new connection, and then stop: \begin{lstlisting} rule Others, InboxLobbyPlayer, Outbox, Network.remoteConnect = wait_until(Network.remoteConnect = true) \end{lstlisting} We send the local player to the new instance, and wait for the message to be received: \begin{lstlisting} let msg = ReliableOutMessageLobbyPlayer.Create(Self, [], Network.Id, Id).ReliableOutMessage yield Outbox <- Outbox + msg wait_until(msg.Received) \end{lstlisting} We then received the local player of the new instance: \begin{lstlisting} let new_player = receive() wait_until(InboxLobbyPlayer.Length > 1) let others = InboxLobbyPlayer.Head.Value yield InboxLobbyPlayer <- InboxLobbyPlayer.Tail \end{lstlisting} Finally, we add the new player to the list of players, and reset \texttt{remoteConnect} to \texttt{false}. A very important notice is that, in case of multiple \texttt{remote connect} rules, then we need to wait until all of them are done before resetting \texttt{remoteConnect}. This will require additional boolean flags: \begin{lstlisting} yield Others <- Others + new_player, Network.remoteConnect <- false \end{lstlisting} Starting the game waits for all players to be ready. This rule is unchanged: \begin{lstlisting} rule CurrentWorld = wait_until(Self.Ready && from p in Others select p.Ready) yield Arena.Create(Self) \end{lstlisting} We create the game like we did for the chat sample. We initialize the network and local fields, and start the various inbox and outbox lists to empty lists: \begin{lstlisting} Create(own_name) = let network = NetManager.Create() let id = network.NextId { Self = LobbyPlayer.Create(own_name, Vector2.Zero, network) Others = [] Network = network Id = id InboxLobbyPlayer = [] InboxListLobbyPlayer = [] Outbox = [] } } \end{lstlisting} We can now present the \texttt{LobbyPlayer} itself. The entity contains its original fields of name, readiness, and initial position for the game: \begin{lstlisting} entity LobbyPlayer = { Name : string Ready : bool StartPosition : Vector2 \end{lstlisting} The entity also has a network \texttt{id}, plus an \texttt{ownership} value that indicates whether or not the entity is owned by the local instance (in which case \texttt{Ownership = local}) or whether it is owned by a remote peer (in which case \texttt{Ownership = remote(Peer)} where \texttt{Peer} is the \texttt{Id} of the owner). The entity also contains a list of received and unprocessed \texttt{bool} messages, and a list of outgoing messages that will be sent: \begin{lstlisting} Id : int Ownership : NetOwnership InboxBool : List Outbox : List \end{lstlisting} We automatically empty the list of outgoing messages, in the assumption that the \texttt{NetPeer} instances have already stored them and are handling them, and we store locally the received messages of type \texttt{bool} and destined to this entity: \begin{lstlisting} rule Outbox = yield [] rule InboxBool = yield InboxBool + from c in Network.Peers from m in c.ReceivedMessages where m.ContentType = BoolTypeId && m.OwnerEntity = Id select InMessageBool.Create(m.Value) \end{lstlisting} The local rules all run exclusively if the entity is locally owned. For this reason we wait, just once, that \texttt{Ownership = local}, and then we loop forever the body of the rule because ownership does not change: \begin{lstlisting} rule Ready, Outbox = wait_until(Ownership = local) while(true) \end{lstlisting} Whenever the \texttt{Enter} key is pressed, we create a reliable outgoing boolean message and then put it in the outgoing queue of messages. We then wait until the message is received: \begin{lstlisting} wait_until(IsKeyDown(Enter)) let msg = ReliableOutMessageBool.Create(true, [], world.Network.Id, Id).ReliableOutMessage yield Outbox <- Outbox + msg wait_until(msg.Received) \end{lstlisting} Finally, we set the local value of \texttt{Ready} to \texttt{true}: \begin{lstlisting} yield Ready <- true \end{lstlisting} The remote rules all run exclusively if the entity is remotely owned. For this reason we wait, just once, that \texttt{Ownership <> local}, and then we loop forever the body of the rule because ownership does not change: \begin{lstlisting} rule Ready, InboxBool = wait_until(Ownership <> local) while(true) \end{lstlisting} We wait until a boolean message appears in the incoming queue. We then return this message, and discard it from the queue as it has now been processed: \begin{lstlisting} wait_until(InboxBool.Length > 1) yield InboxBool.Head.Value, InboxBool.Tail \end{lstlisting} We create the entity from its original parameters, but we also need the local \texttt{NetManager} in order to obtain a unique \texttt{Id} for the entity. We also initialize the various message lists to empty lists. Finally, when creating an entity locally we will always set its \texttt{Ownership} to \texttt{local}, because the entity is owned by the peer that creates it: \begin{lstlisting} Create(name, p, network : NetManager) = { Name = name Ready = false StartPosition = p Id = network.NextId InboxBool = [] Outbox = [] Ownership = local } } \end{lstlisting} \subsection{Formal semantics} We now present the semantics in a more formal and compact framework. To describe the way multi-player games work, we consider the various instances of the game running in lock-step. Each instance has its own game world: $$\omega_1, \omega_2, \dots$$ The game world is structured like a tree of entities. Each entity has some fields and some rules. Each rule acts on a subset of the fields of the entity by defining their new value after one (or more) ticks of the simulation. For simplicity, in the following we assume that each rule updates all fields together\footnote{\texttt{rule X = yield 10} is equivalent to \texttt{rule X,Y,Z = yield 10,Y,Z}}: $$E = { f_1 \dots f_n \ \ r_1 \dots r_m }$$ An entity is updated by evaluating, in order, all the rules to the fields: \begin{align*} \tick(e:E, dt) = { \tick(f_1',dt) \dots \tick(f_n',dt) \ \ r_1' \dots r_m' }\\ f_1',\dots,f_n',r_1',\dots,r_m' = \step(\dots \step(f_1,\dots,f_n,r_1), \dots, r_m) \end{align*} We define the $\step$ function as a function that recursively evaluates the body of a rule. The function evaluates until it encounters either a $\wait$ or a $\yield$ statement. $\step$ also returns the remainder of the rule body, so that the rule will effectively be resumed where it left off, at the next evaluation of $\step$: \begin{align*} \step(f_1, \dots, f_n, \letlet x = y \letin z) = \step(f_1, \dots, f_n, z[x:=y]) \\ \vdots \\ \step(f_1, \dots, f_n, \yield x; b) = x, b \end{align*} In order to add networking, we assume that each entity has two new fields, which are $\id$ and $\owner$. $\id$ is a simple (numeric) identifier that, inside a single instance of the game world, uniquely denotes a specific entity. $\owner$ may be either $\local$ or $\remote$. Given a set of entities (one per game world) that share the same type and the same $\id$, exactly one of them will be $\local$, all the others will be $\remote$. First of all we ``translate away'' the $\local$ and $\remote$ scopes. This means that given an entity with some rules defined inside such scopes: $$E = { f_1 \dots f_n \ \ r_1 \dots \ \ \local{ r_j \dots } \remote{ r_l \dots } }$$ we transform the rules $r_j$ inside the $\local$ scope into: $$\wait(\owner = \local); r_j$$ and we transform the rules $r_j$ inside the $\remote$ scope into: $$\wait(\owner = \remote); r_l$$ This will prevent those rules from running when they should not. Similarly, by adding the global $\localconnect$ and $\remoteconnect$ flags, we can have some rules only execute when a new instance is added to the game. All existing instances will enable $\remoteconnect$, while the new instance will enable $\localconnect$. When evaluating a rule with the $\step$ function, we stop at $\receive[T]$ statements. Assume that we are updating entities inside a specific game instance, the one with the world $\omega_i$. We look for some entity, $e'$, belonging to \textit{another} game world, with the same $\id$ as the entity we are updating, and which is sending a value $v$ of the type we are expecting ($T$): $$\step(f_1, \dots, \letlet x = \receive[T](); b) = \left\{ \begin{matrix} f_1, \dots, b[x:=v'] \ \ \text{when } \exists r_l \in e' : r_l = \send[T](v) \\ f_1, \dots, \letlet x = \receive[T](); b \ \ \text{otherwise} \end{matrix} \right. $$ Notice that when we receive a value $v$, we turn it into $v'$ in the receiving instance of the game. $v'$ is identical to $v$, but all of its $\owner$ fields are changed to $\remote$. This is needed in order to reflect the fact that even if the transmitted entity was owned by the sender, it is certainly not owned by the receiver. Similarly, when performing a $\receiveMany$, then we look for another entity $e'_j$ with the same type and $\id$ of the current entity \textit{for each other game world}: $$\step(f_1, \dots, \letlet x = \receiveMany[T](); b) = \left\{ \begin{matrix} f_1, \dots, b[x:=[v'_1 \dots ]] \ \ \text{when } \forall e'_j : \exists r^j_l \in e'_j : r^j_l = \send[T](v^j) \\ f_1, \dots, \letlet x = \receiveMany[T](); b \ \ \text{otherwise} \end{matrix} \right. $$ $\receiveMany$ then binds the list of \textit{all the received values} to $x$. Both $\receive$ and $\receiveMany$ also work with $\sendRel$, and not just with $\send$. The main difference is that $\send$ does not wait to be evaluated, while $\sendRel$ waits until someone performs a $\receive$ or a $\receiveMany$. Thus, $\send$ is eagerly stepped: $$\step(f_1, \dots, \send[T](v); b) = f_1, \dots, b$$ while $\sendRel$ needs to wait: $$\step(f_1, \dots, \sendRel[T](v); b) = \left\{ \begin{matrix} f_1, \dots, b \ \ \text{when } \exists r_l \in e' : r_l = \receive[T]() \\ f_1, \dots, \sendRel[T](v); b \ \ \text{otherwise} \end{matrix} \right. $$ The semantics of sending and receiving are thus easily explained in terms of changes to the values (and the control flow) of other instances, with some added machinery to make sure that only the appropriate rules (\texttt{local}, \texttt{remote}, etc.) are run on each instance. \subsection{Reliability} One aspect that has not been covered so far is handling of transmission failures. All primitives but $\sendRel$ never fail. $\sendRel$, on the other hand, may fail when the receiver is either disconnected or unreachable. After a certain time-out, $\sendRel$ will return \texttt{false} to denote transmission failure, and inform the other instances that the receiver should be considered disconnected. This means that $\sendRel$ are implicitly used for forcing disconnection of unreachable instances. \subsection{Asymmetry and load sharing} The ability to determine which instance of the game runs which entities locally allows us to take advantage of asymmetry for performance purposes. By estimating the network and CPU performance of an instance, we can determine its current overall performance. Overall performance of an instance could be used to have that instance create, and handle, more performance-intensive entities, such as those with complex AI, physics, etc. This way instances that have more processing power (for example because of better hardware) or bandwidth (because of a better local network infrastructure) would be used to lighten the load of the ``weaker'' instances. This topic is quite broad and complex, and it is outside the scope of the current work. For this reason we do not expand it in depth, but leave it sketched. \subsection{Closing remarks} This concludes the presentation of Casanova syntax and semantics. The syntax may look, at a first glance, deceptively simple, but in reality it drives a complex translation. All networking operations are then translated away in a series of list operations on incoming or outgoing mailboxes. The various mailboxes then handle messages, which are responsible for the low-level sending and receiving of data through some simple, low-level, underlying networking library that is required to provide little more than sockets and connections. % preliminaries.tex %%%%%%%%%%%%%%% \begin{frame}{} \fig{width = 0.50\textwidth}{figs/preliminary} \end{frame} %%%%%%%%%%%%%%% %%%%%%%%%%%%%%% \input{parts/notations} \input{parts/rules} %%%%%%%%%%%%%%%% Copyright 2020 % % Licensed under the Apache License, Version 2.0 (the "License"); % you may not use this file except in compliance with the License. % You may obtain a copy of the License at % % http://www.apache.org/licenses/LICENSE-2.0 % % Unless required by applicable law or agreed to in writing, software % distributed under the License is distributed on an "AS IS" BASIS, % WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. % See the License for the specific language governing permissions and % limitations under the License. \NeedsTeXFormat{LaTeX2e} \newcommand{\@ttNameLg}{ Elementary Foundations --- Logic Macros } \newcommand{\@ttVersionLg}{ v1.5 } \newcommand{\@ttDateLg}{ 2020/08/26 } \newcommand{\@ttDescriptionLg}{ This package provides macros available at EF: An Introduction To Topics in Discrete Mathematics. Package file was named in lieu of Jeremy Sylvestre. } \ProvidesPackage{lg_jsylvest}[\@ttDateLg \@ttVersionLg \@ttNameLg] % Macros pulled from https://sites.ualberta.ca/~jsylvest/books/EF/ % [TODO] Make sure that all of the macros work % [TODO] Warn user if they used any predefined logical symbols % in 'problems.sty' \RequirePackage{cancel} \RequirePackage{mathrsfs} \newcommand{\nth}[1][n]{{#1}^{\mathrm{th}}} \newcommand{\bbrac}[1]{\bigl(#1\bigr)} \newcommand{\Bbrac}[1]{\Bigl(#1\Bigr)} \newcommand{\correct}{\boldsymbol{\checkmark}} \newcommand{\incorrect}{\boldsymbol{\times}} \newcommand{\inv}[2][1]{{#2}^{-{#1}}} \newcommand{\leftsub}[3][1]{\mathord{{}_{#2\mkern-#1mu}#3}} \newcommand{\N}{\mathbb{N}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\Q}{\mathbb{Q}} \newcommand{\R}{\mathbb{R}} \newcommand{\I}{\mathbb{I}} \newcommand{\abs}[1]{\left\lvert #1 \right\rvert} \DeclareMathOperator{\sqrtop}{sqrt} % Logic-related operators \newcommand{\lgcnot}{\neg} \newcommand{\lgcand}{\wedge} \newcommand{\lgcor}{\vee} \newcommand{\lgccond}{\rightarrow} \newcommand{\lgcbicond}{\leftrightarrow} \newcommand{\lgcimplies}{\Rightarrow} \newcommand{\lgcequiv}{\Leftrightarrow} \newcommand{\lgctrue}{\mathrm{T}} \newcommand{\lgcfalse}{\mathrm{F}} % Boolean \newcommand{\boolnot}[1]{{#1}'} \newcommand{\boolzero}{\mathbf{0}} \newcommand{\boolone}{\mathbf{1}} % Sets and functions \newcommand{\setdef}[2]{\left\{\mathrel{}#1\mathrel{}\middle|\mathrel{}#2\mathrel{}\right\}} \newcommand{\inlinesetdef}[2]{\{\mathrel{}#1\mathrel{}\mid\mathrel{}#2\mathrel{}\}} \let\emptyword\emptyset \renewcommand{\emptyset}{\varnothing} \newcommand{\relcmplmnt}{\smallsetminus} \newcommand{\union}{\cup} \newcommand{\intersection}{\cap} \newcommand{\cmplmnt}[1]{{#1}^{\mathrm{c}}} \newcommand{\disjunion}{\sqcup} \newcommand{\cartprod}{\times} \newcommand{\words}[1]{{#1}^\ast} \newcommand{\length}[1]{\abs{#1}} \newcommand{\powsetbare}{\mathscr{P}} \newcommand{\powset}[1]{\powsetbare(#1)} \newcommand{\funcdef}[4][\to]{#2\colon #3 #1 #4} \newcommand{\ifuncto}{\hookrightarrow} \newcommand{\ifuncdef}[3]{\funcdef[\ifuncto]{#1}{#2}{#3}} \newcommand{\sfuncto}{\twoheadrightarrow} \newcommand{\sfuncdef}[3]{\funcdef[\sfuncto]{#1}{#2}{#3}} \newcommand{\funcgraphbare}{\Delta} \newcommand{\funcgraph}[1]{\funcgraphbare(#1)} \newcommand{\relset}[3]{#1_{{} #2 #3}} \newcommand{\gtset}[2]{\relset{#1}{\gt}{#2}} \newcommand{\posset}[1]{\gtset{#1}{0}} \newcommand{\geset}[2]{\relset{#1}{\ge}{#2}} \newcommand{\nnegset}[1]{\geset{#1}{0}} \newcommand{\neqset}[2]{\relset{#1}{\neq}{#2}} \newcommand{\nzeroset}[1]{\neqset{#1}{0}} \newcommand{\ltset}[2]{\relset{#1}{\lt}{#2}} \newcommand{\leset}[2]{\relset{#1}{\le}{#2}} \newcommand{\natnumlt}[1]{\ltset{\N}{#1}} \DeclareMathOperator{\id}{id} \newcommand{\inclfunc}[2]{\iota_{#1}^{#2}} \newcommand{\projfunc}[1]{\rho_{#1}} \DeclareMathOperator{\proj}{proj} \newcommand{\funcres}[2]{\left.{#1}\right\rvert_{#2}} \newcommand{\altfuncres}[2]{\left.{#1}\right\rvert{#2}} \DeclareMathOperator{\res}{res} \DeclareMathOperator{\flr}{flr} \newcommand{\floor}[1]{\lfloor {#1} \rfloor} \newcommand{\funccomp}{\circ} \newcommand{\funcinvimg}[2]{\inv{#1}\left({#2}\right)} \newcommand{\card}[1]{\left\lvert #1 \right\rvert} \DeclareMathOperator{\cardop}{card} \DeclareMathOperator{\ncardop}{\#} \newcommand{\EngAlphabet}{\{ \mathrm{a}, \, \mathrm{b}, \, \mathrm{c}, \, \dotsc, \, \mathrm{y}, \, \mathrm{z} \}} \newcommand{\ShortEngAlphabet}{\{ \mathrm{a}, \, \mathrm{b}, \, \dotsc, \, \mathrm{z} \}} \newcommand{\eqclass}[1]{\left[#1\right]} \newcommand{\partorder}{\preceq} \newcommand{\partorderstrict}{\prec} \newcommand{\npartorder}{\npreceq} % Graph theory \newcommand{\subgraph}{\preceq} \newcommand{\subgraphset}[1]{\mathcal{S}(#1)} \newcommand{\connectedsubgraphset}[1]{\mathcal{C}(#1)} % Combinatorics \newcommand{\permcomb}[3]{{#1}(#2,#3)} \newcommand{\permcombalt}[3]{{#1}^{#2}_{#3}} \newcommand{\permcombaltalt}[3]{{\leftsub{#2}{#1}}_{#3}} \newcommand{\permutation}[2]{\permcomb{P}{#1}{#2}} \newcommand{\permutationalt}[2]{\permcombalt{P}{#1}{#2}} \newcommand{\permutationaltalt}[2]{{\permcombaltalt{P}{#1}{#2}}} \newcommand{\combination}[2]{\permcomb{C}{#1}{#2}} \newcommand{\combinationalt}[2]{\permcombalt{C}{#1}{#2}} \newcommand{\combinationaltalt}[2]{{\permcombaltalt{C}{#1}{#2}}} \newcommand{\choosefuncformula}[3]{\frac{#1 !}{#2 ! \, #3 !}} \DeclareMathOperator{\matrixring}{M} \newcommand{\uvec}[1]{\mathbf{#1}} \newcommand{\zerovec}{\uvec{0}} % The following were used to input escaped characters < (<), % > (%gt;), and & (&). % I'll be replacing them back to its LaTeX equivalents % because it's not necessary anymore. % % Future write-ups will have the following conversions done % using find & replace. \newcommand{\lt}{<} \newcommand{\gt}{>} \newcommand{\amp}{&} \endinputaneeshnaik/aneeshnaik.github.io \section*{Run \#2} \begin{figure}[H] \centering \includegraphics[width=0.5\textwidth]{images/2/stream_image} \end{figure} \begin{figure}[H] \centering \includegraphics[width=\textwidth]{images/2/trajectories} \end{figure} CarlkD/GraduationProjectChapter_Examples/chapter_04.tex %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \chapter{REFERENCES, QUOTINGS AND FOOTNOTES}\label{Ch4} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% In this section, information will be given about how citations, quotings and footnotes should be. \section{Citing (indication of references in main text body)} \subsection{Citing according to surname of author} References are cited with the surname of author and year. In the references section, the references are listed alphabetically according to the surname of the author. Citing of a reference at the beginning of or within a sentence must be as Boran (2003), whereas a citation at the end of a sentence must be as (Boran, 2003). The full-stop is placed directly after the citation. A reference with two authors must be cited as (2004) at the beginning of or within a sentence, or as (Yı, 2004) at the end of a sentence. A reference with more than two authors must be cited as Yılmaz et al. (2004) at the beginning of or within a sentence, or as (Yılmaz et al, 2004) at the end of a sentence. Different publications of an author published in the same year must be cited as Feray (2005a), Feray (2005b). While citing a part of a publication; the number of the page the cited material (chapter, table, figure, or equation) is on must be indicated. While citing, the expression “page” must be abbreviated, but “chapter” must not. For example; (Centers for Disease Control and Prevention, 2005, p. 10), (Shimamura, 1989, Chapter 3). Citing multiple publications in one pair of brackets; (Berndt, 2002; Harlow, 1983). Citing personal communication in main text body; (, personal communication, September 28, 1998), (, personal communication, August 15, 2009). In the references section, reference tags must be listed according to the surname of author. For citing of secondary references (In case the reference cites another reference), the secondary reference must be cited in brackets. In the references section, the reference tag is organized according to the secondary reference, the original reference must not be used as a tag. For example; In his e-mails, Smith argued that asynchronous line dancing would be the next Internet meme (as cited in Jones, 2010). \subsection{Citing according to order of appearance} References are cited by numbering and indicating the number in square brackets ([]) in the main text body. The first reference cited in a thesis is numbered [1] and the following references are numbered according to the order of appearance. In the main text body, references must be cited as specified below: \vspace*{-12pt} \begin{tabbing} \hspace*{1.5cm}\= \kill [1] \> Reference no. 1\\ [1--3] \> References from no.1 to 3 (thus, references 1,2 and 3)\\ [1,3] \> References no. 1 and 3\\ [1,3,8] \> References no.1, 3 and 8\\ [1,3--8] \> References no.1, and from no.3 to 8 (thus, references 1, 3, 4, 5, 6, 7 and 8) \end{tabbing} \vspace*{-12pt} Different volumes of a reference must be cited and numbered individually. \section{Quoting} Generally, quoting is done by remaining faithful to the original text in terms of words, spelling and punctuation. In case there is a mistake, the correct version is written in square brackets in the quoted text. Short quotations (not longer than 40 words) must be given in quotation marks. Following the text quoted, the reference must be written and a full-stop must be placed afterwards. Quotations longer than 40 words must not be shown in quotation marks. Instead, they must be indented 1 tab space (1.27 cm) from the left side of the page. The font size for long quotations indented from the left must be 2 pt smaller than the font size used in main text body. However, it is not advised to quote very long texts and to quote very frequently. Unlike short quotations, references of long quotations must be placed after the full stop. (i.e., .(p.196)) Example for a quotation at the beginning of a sentence; According to Jones (1998), "Students often had difficulty using APA style, especially when it was their first time" (p. 199). Example for a quotation in the middle of a sentence; Interpreting these results, Robbins et al. (2003) suggested that the “therapists in dropout cases may have inadvertently validated parental negativity about the adolescent without adequately responding to the adolescent’s needs or concerns” (p. 541) contributing to an overall climate of negativity. Example for a quotation at the end of a sentence; Confusing this issue is the overlapping nature of roles in palliative care, whereby “medical needs are met by those in the medical disciplines; nonmedical needs may be addressed by anyone on the team” (Csikai \& Chaitin, 2006, p. 112). Detailed information on quoting could be found on websites of Graduate Schools and associated links. \section{Footnotes} Footnotes could be used in theses to add content-expanding, content-enhancing, or additional information. Footnote numbers must be placed directly after a quotation. In case the quotation is a paragraph, the footnote numbers must be placed directly after the last word of the paragraph (as superscript). In case the quotation is a concept or a noun, footnote numbers must be placed directly after that concept or noun (as superscript). Footnote numbers in the main text body must be indicated as superscript, as shown\footnotemark. A punctuation mark must not be placed after the number. Footnotes must be written with a font size 2 pt smaller than the main text body font size. 1 space must be set between footnote line and footnote number, 1/2 space must be set between footnote number and the first line of the footnote. Footnotes must be separated from the main text body with a thin horizontal line. Detailed information on footnotes could be found on the websites of Graduate Schools and associated links. \footnotetext{~Reference display can not be done with footnotes.~Footnotes could be used in theses to add content-expanding, content-enhancing, or additional information.~If these information must include references, these references must be indicated in References section.} \section{Second Level Title: First Letters Capital} Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut labore et dolore magna aliquyam erat, sed diam voluptua. At vero eos et accusam et justo duo dolores et ea rebum. Stet clita kasd gub rgren, no sea. \subsection{Third level title: Only first letter capital} Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut labore et dolore magna aliquyam erat, sed diam voluptua. At vero eos et accusam et justo duo dolores et ea rebum. Stet clita kasd gub rgren, no sea. \subsubsection{Fourth level title: Only first letter capital} Stet clita kasd gub rgren, no sea takimata sanctus est Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut lab ore sit et dolore magna. \subsubsubsection{Fifth level title: No numbering after fourth level titles} Stet clita kasd gub rgren, no sea takimata sanctus est Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut lab ore sit et dolore magna\footnotemark. % Include tilda to provide one letter spacing between the foot number and the text at the bottom - SBÖ \footnotetext{~~Footnotes must be written with a font size 2 pt smaller than the main text body font size.} \begin{figure}[t] \centering \includegraphics[width=230pt,keepaspectratio=true]{./fig/sekil6} % sekil6.eps: 0x0 pixel, 300dpi, 0.00x0.00 cm, bb=14 14 555 489 \caption{Example figure.} \label{Figure4.1} \end{figure} This indicates that the ANN is accurate at base flow and flow height values lower then 3 m. \begin{table*}[h] {\setlength{\tabcolsep}{14pt} \caption{Example table.} \begin{center} \vspace{-6mm} \begin{tabular}{cccc} \hline \\[-2.45ex] \hline \\[-2.1ex] Column A & Column B & Column C & Column D \\ \hline \\[-1.8ex] Row A & Row A & Row A & Row A \\ Row B & Row B & Row B & Row B \\ Row C & Row C & Row C & Row C \\ [-0ex] \hline \end{tabular} \vspace{-6mm} \end{center} \label{Table4.1}} \end{table*} Stet clita kasd gub rgren, no sea takimata sanctus est Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut lab ore sit et dolore magna. Stet clita kasd gub rgren, no sea takimata sanctus est Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut lab ore sit et dolore magna. Stet clita kasd gub rgren, no sea takimata sanctus est Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor invidunt ut lab ore sit et dolore magna. \cleardoublepage % Analitic index in new odd page, set page num in index \phantomsection % Because we included hyperref ??? \addcontentsline{toc}{chapter} {\indexname} % Add the analitic index item to index list \printindex % Print analitic indexbody/08-crabbe.tex \chapter[Mr. Campbell and Mr. Crabbe] {mr. campbell {\Large and} mr. crabbe} Mr. Campbell may be said to hold a place (among modern poets) between Lord Byron and Mr. Rogers. With much of the glossy splendour, the pointed vigour, and romantic interest of the one, he possesses the fastidious refinement, the classic elegance of the other. Mr. Rogers, as a writer, is too effeminate, Lord Byron too extravagant: Mr. Campbell is neither. The author of the \emph{Pleasures of Memory} polishes his lines till they sparkle with the most exquisite finish; he attenuates them into the utmost degree of trembling softness: but we may complain, in spite of the delicacy and brilliancy of the execution, of a want of strength and solidity. The author of the \emph{Pleasures of Hope}, with a richer and deeper vein of thought and imagination, works it out into figures of equal grace and dazzling beauty, avoiding on the one hand the tinsel of flimsy affectation, and on the other the vices of a rude and barbarous negligence. His Pegasus is not a rough, skittish colt, running wild among the mountains, covered with bur-docks and thistles, nor a tame, sleek pad, unable to get out of the same ambling pace, but a beautiful \emph{manege}-horse, full of life and spirit in itself, and subject to the complete controul of the rider. Mr. Campbell gives scope to his feelings and his fancy, and embodies them in a noble and naturally interesting subject; and he at the same time conceives himself called upon (in these days of critical nicety) to pay the exactest attention to the expression of each thought, and to modulate each line into the most faultless harmony. The character of his mind is a lofty and self-scrutinising ambition, that strives to reconcile the integrity of general design with the perfect elaboration of each component part, that aims at striking effect, but is jealous of the means by which this is to be produced. Our poet is not averse to popularity (nay, he is tremblingly alive to it)\textemdash but self-respect is the primary law, the indispensable condition on which it must be obtained. We should dread to point out (even if we could) a false concord, a mixed metaphor, an imperfect rhyme in any of Mr. Campbell's productions; for we think that all his fame would hardly compensate to him for the discovery. He seeks for perfection, and nothing evidently short of it can satisfy his mind. He is a \emph{high finisher} in poetry, whose every work must bear inspection, whose slightest touch is precious\textemdash not a coarse dauber who is contented to impose on public wonder and credulity by some huge, ill-executed design, or who endeavours to wear out patience and opposition together by a load of lumbering, feeble, awkward, improgressive lines\textemdash on the contrary, Mr. Campbell labours to lend every grace of execution to his subject, while he borrows his ardour and inspiration from it, and to deserve the laurels he has earned, by true genius and by true pains. There is an apparent consciousness of this in most of his writings. He has attained to great excellence by aiming at the greatest, by a cautious and yet daring selection of topics, and by studiously (and with a religious horror) avoiding all those faults which arise from grossness, vulgarity, haste, and disregard of public opinion. He seizes on the highest point of eminence, and strives to keep it to himself\textemdash he ``snatches a grace beyond the reach of art,'' and will not let it go\textemdash he steeps a single thought or image so deep in the Tyrian dyes of a gorgeous imagination, that it throws its lustre over a whole page\textemdash every where vivid \emph{ideal} forms hover (in intense conception) over the poet's verse, which ascends, like the aloe, to the clouds, with pure flowers at its top. Or to take an humbler comparison (the pride of genius must sometimes stoop to the lowliness of criticism) Mr. Campbell's poetry often reminds us of the purple gilliflower, both for its colour and its scent, its glowing warmth, its rich, languid, sullen hue, \begin{verse} \vleftofline{``}Yet sweeter than the lids of Juno's eyes, \\ Or Cytherea's breath!'' \end{verse} There are those who complain of the little that has done in poetry, and who seem to insinuate that he is deterred by his own reputation from making any further or higher attempts. But after having produced two poems that have gone to the heart of a nation, and are gifts to a world, he may surely linger out the rest of his life in a dream of immortality. There are moments in our lives so exquisite that all that remains of them afterwards seems useless and barren; and there are lines and stanzas in our author's early writings in which he may be thought to have exhausted all the sweetness and all the essence of poetry, so that nothing farther was left to his efforts or his ambition. Happy is it for those few and fortunate worshippers of the Muse (not a subject of grudging or envy to others) who already enjoy in their life-time a foretaste of their future fame, who see their names accompanying them, like a cloud of glory, from youth to age, \begin{verse} \vleftofline{``}And by the vision splendid,\\ Are on their way attended''\textemdash \end{verse} and who know that they have built a shrine for the thoughts and feelings, that were most dear to them, in the minds and memories of other men, till the language which they lisped in childhood is forgotten, or the human heart shall beat no more! The \emph{Pleasures of Hope} alone would not have called forth these remarks from us; but there are passages in the \emph{Gertrude of Wyoming} of so rare and ripe a beauty, that they challenge, as they exceed all praise. Such, for instance, is the following peerless description of Gertrude's childhood:\textemdash \begin{verse} % verse 90-120 stanza xi-xiii 1809 \vleftofline{``}A lov'd bequest\textemdash and I may half impart\textemdash\\ To those that feel the strong paternal tie,\\ How like a new existence in his heart\\ That living flow'r uprose beneath his eye,\\ Dear as she was, from cherub infancy,\\ From hours when she would round his garden play,\\ To time when as the ripening years went by,\\ Her lovely mind could culture well repay,\\ And more engaging grew from pleasing day to day.\\[\stanzaskip] \vleftofline{``}I may not paint those thousand infant charms\\ (Unconscious fascination, undesign'd!)\\ The orison repeated in his arms,\\ For God to bless her sire and all mankind;\\ The book, the bosom on his knee reclined,\\ Or how sweet fairy-lore he heard her con\\ (The play-mate ere the teacher of her mind):\\ All uncompanion'd else her years had gone,\\ Till now in Gertrude's eyes their ninth blue summer shone.\\[\stanzaskip] \vleftofline{``}And summer was the tide, and sweet the hour,\\ When sire and daughter saw, with fleet descent,\\ An Indian from his bark approach their bow'r,\\ Of buskin'd limb and swarthy lineament;\\ The red wild feathers on his brow were blent,\\ And bracelets bound the arm that help'd to light\\ A boy, who seem'd, as he beside him went,\\ Of Christian vesture and complexion bright,\\ Led by his dusky guide like morning brought by night.''\\ \end{verse} In the foregoing stanzas we particularly admire the line\textemdash \begin{verse} ``Till now in Gertrude's eyes their ninth blue summer shone.'' \end{verse} It appears to us like the ecstatic union of natural beauty and poetic fancy, and in its playful sublimity resembles the azure canopy mirrored in the smiling waters, bright, liquid, serene, heavenly! A great outcry, we know, has prevailed for some time past against poetic diction and affected conceits, and, to a certain degree, we go along with it; but this must not prevent us from feeling the thrill of pleasure when we see beauty linked to beauty, like kindred flame to flame, or from applauding the voluptuous fancy that raises and adorns the fairy fabric of thought, that nature has begun! Pleasure is ``scattered in stray-gifts o'er the earth''\textemdash beauty streaks the ``famous poet's page'' in occasional lines of inconceivable brightness; and wherever this is the case, no splenetic censures or ``jealous leer malign,'' no idle theories or cold indifference should hinder us from greeting it with rapture.\textemdash There are other parts of this poem equally delightful, in which there is a light startling as the red-bird's wing; a perfume like that of the magnolia; a music like the murmuring of pathless woods or of the everlasting ocean. We conceive, however, that Mr. Campbell excels chiefly in sentiment and imagery. The story moves slow, and is mechanically conducted, and rather resembles a Scotch canal carried over lengthened aqueducts and with a number of \emph{locks} in it, than one of those rivers that sweep in their majestic course, broad and full, over Transatlantic plains and lose themselves in rolling gulfs, or thunder down lofty precipices. But in the centre, the inmost recesses of our poet's heart, the pearly dew of sensibility is distilled and collects, like the diamond in the mine, and the structure of his fame rests on the crystal columns of a polished imagination. We prefer the \emph{Gertrude} to the \emph{Pleasures of Hope}, because with perhaps less brilliancy, there is more of tenderness and natural imagery in the former. In the \emph{Pleasures of Hope} had not completely emancipated himself from the trammels of the more artificial style of poetry\textemdash from epigram, and antithesis, and hyperbole. The best line in it, in which earthly joys are said to be\textemdash \begin{verse} ``Like angels' visits, few and far between''\textemdash \end{verse} is a borrowed one.\footnote{\begin{quote} ``Like angels' visits, short and far between.''\textemdash \emph{Blair's Grave}. \end{quote} } But in the \emph{Gertrude of Wyoming} ``we perceive a softness coming over the heart of the author, and the scales and crust of formality that fence in his couplets and give them a somewhat glittering and rigid appearance, fall off,'' and he has succeeded in engrafting the wild and more expansive interest of the romantic school of poetry on classic elegance and precision. After the poem we have just named, Mr. Campbell's \textsc{Songs} are the happiest efforts of his Muse:\textemdash breathing freshness, blushing like the morn, they seem, like clustering roses, to weave a chaplet for love and liberty; or their bleeding words gush out in mournful and hurried succession, like ``ruddy drops that visit the sad heart'' of thoughtful Humanity. The \emph{Battle of Hohenlinden} is of all modern compositions the most lyrical in spirit and in sound. To justify this encomium, we need only recall the lines to the reader's memory. \begin{verse} \vleftofline{``}On Linden, when the sun was low,\\ All bloodless lay th' untrodden snow,\\ And dark as winter was the flow\\ Of Iser, rolling rapidly.\\[\stanzaskip] But Linden saw another sight,\\ When the drum beat at dead of night,\\ Commanding fires of death to light\\ The darkness of her scenery.\\[\stanzaskip] By torch and trumpet fast array'd,\\ Each horseman drew his battle blade,\\ And furious every charger neigh'd,\\ To join the dreadful revelry.\\[\stanzaskip] Then shook the hills with thunder riv'n,\\ Then rush'd the steed to battle driv'n,\\ And louder than the bolts of heav'n\\ Far flash'd the red artillery.\\[\stanzaskip] But redder yet that light shall glow\\ On Linden's hills of stained snow,\\ And bloodier yet the torrent flow\\ Of Iser, rolling rapidly.\\[\stanzaskip] 'Tis morn, but scarce yon level sun\\ Can pierce the war-clouds, rolling\footnote{Is not this word, which occurs in the last line but one, (as well as before) an instance of that repetition, which we so often meet with in the most correct and elegant writers?} dun,\\ Where furious Frank and fiery Hun\\ Shout in their sulph'rous canopy.\\[\stanzaskip] The combat deepens. On, ye brave,\\ Who rush to glory, or the grave!\\ Wave, Munich! all thy banners wave!\\ And charge with all thy chivalry!\\[\stanzaskip] Few, few shall part, where many meet!\\ The snow shall be their winding-sheet,\\ And every turf beneath their feet\\ Shall be a soldier's sepulchre."\\ \end{verse} Mr. Campbell's prose-criticisms on contemporary and other poets (which have appeared in the New Monthly Magazine) are in a style at once chaste, temperate, guarded, and just. Mr. Crabbe presents an entire contrast to Mr. Campbell:\textemdash the one is the most ambitious and aspiring of living poets, the other the most humble and prosaic. If the poetry of the one is like the arch of the rainbow, spanning and adorning the earth, that of the other is like a dull, leaden cloud hanging over it. Mr. Crabbe's style might be cited as an answer to Audrey's question\textemdash ``Is poetry a true thing?'' There are here no ornaments, no flights of fancy, no illusions of sentiment, no tinsel of words. His song is one sad reality, one unraised, unvaried note of unavailing woe. Literal fidelity serves him in the place of invention; he assumes importance by a number of petty details; he rivets attention by being tedious. He not only deals in incessant matters of fact, but in matters of fact of the most familiar, the least animating, and the most unpleasant kind; but he relies for the effect of novelty on the microscopic minuteness with which he dissects the most trivial objects\textemdash and for the interest he excites, on the unshrinking determination with which he handles the most painful. His poetry has an official and professional air. He is called in to cases of difficult births, of fractured limbs, or breaches of the peace; and makes out a parochial list of accidents and offences. He takes the most trite, the most gross and obvious and revolting part of nature, for the subject of his elaborate descriptions; but it is Nature still, and Nature is a great and mighty Goddess! It is well for the Reverend Author that it is so. Individuality is, in his theory, the only definition of poetry. Whatever \emph{is}, he hitches into rhyme. Whoever makes an exact image of any thing on the earth, however deformed or insignificant, according to him, must succeed\textemdash and he himself has succeeded. Mr. Crabbe is one of the most popular and admired of our living authors. That he is so, can be accounted for on no other principle than the strong ties that bind us to the world about us, and our involuntary yearnings after whatever in any manner powerfully and directly reminds us of it. His Muse is not one of \emph{the Daughters of Memory}, but the old toothless, mumbling dame herself, doling out the gossip and scandal of the neighbourhood, recounting \emph{totidem verbis et literis}, what happens in every place of the kingdom every hour in the year, and fastening always on the worst as the most palatable morsels. But she is a circumstantial old lady, communicative, scrupulous, leaving nothing to the imagination, harping on the smallest grievances, a village-oracle and critic, most veritable, most identical, bringing us acquainted with persons and things just as they chanced to exist, and giving us a local interest in all she knows and tells. Mr. Crabbe's Helicon is choked up with weeds and corruption; it reflects no light from heaven, it emits no cheerful sound: no flowers of love, of hope, or joy spring up near it, or they bloom only to wither in a moment. Our poet's verse does not put a spirit of youth in every thing, but a spirit of fear, despondency, and decay: it is not an electric spark to kindle or expand, but acts like the torpedo's touch to deaden or contract. It lends no dazzling tints to fancy, it aids no soothing feelings in the heart, it gladdens no prospect, it stirs no wish; in its view the current of life runs slow, dull, cold, dispirited, half under ground, muddy, and clogged with all creeping things. The world is one vast infirmary; the hill of Parnassus is a penitentiary, of which our author is the overseer: to read him is a penance, yet we read on! Mr. Crabbe, it must be confessed, is a repulsive writer. He contrives to ``turn diseases to commodities,'' and makes a virtue of necessity. He puts us out of conceit with this world, which perhaps a severe divine should do; yet does not, as a charitable divine ought, point to another. His morbid feelings droop and cling to the earth, grovel where they should soar; and throw a dead weight on every aspiration of the soul after the good or beautiful. By degrees we submit, and are reconciled to our fate, like patients to the physician, or prisoners in the condemned cell. We can only explain this by saying, as we said before, that Mr. Crabbe gives us one part of nature, the mean, the little, the disgusting, the distressing; that he does this thoroughly and like a master, and we forgive all the rest. Mr. Crabbe's first poems were published so long ago as the year 1782, and received the approbation of Dr. Johnson only a little before he died. This was a testimony from an enemy; for Dr. Johnson was not an admirer of the simple in style or minute in description. Still he was an acute, strong-minded man, and could see truth when it was presented to him, even through the mist of his prejudices and his foibles. There was something in Mr. Crabbe's intricate points that did not, after all, so ill accord with the Doctor's purblind vision; and he knew quite enough of the petty ills of life to judge of the merit of our poet's descriptions, though he himself chose to slur them over in high-sounding dogmas or general invectives. Mr. Crabbe's earliest poem of the \emph{Village} was recommended to the notice of Dr. Johnson by Sir ; and we cannot help thinking that a taste for that sort of poetry, which leans for support on the truth and fidelity of its imitations of nature, began to display itself much about that time, and, in a good measure, in consequence of the direction of the public taste to the subject of painting. Book-learning, the accumulation of wordy common-places, the gaudy pretensions of poetical fiction, had enfeebled and perverted our eye for nature. The study of the fine arts, which came into fashion about forty years ago, and was then first considered as a polite accomplishment, would tend imperceptibly to restore it. Painting is essentially an imitative art; it cannot subsist for a moment on empty generalities: the critic, therefore, who had been used to this sort of substantial entertainment, would be disposed to read poetry with the eye of a connoisseur, would be little captivated with smooth, polished, unmeaning periods, and would turn with double eagerness and relish to the force and precision of individual details, transferred, as it were, to the page from the canvas. Thus an admirer of Teniers or Hobbima might think little of the pastoral sketches of Pope or Goldsmith; even Thompson describes not so much the naked object as what he sees in his mind's eye, surrounded and glowing with the mild, bland, genial vapours of his brain:\textemdash but the adept in Dutch interiors, hovels, and pig-styes must find in Mr. Crabbe a man after his own heart. He is the very thing itself; he paints in words, instead of colours: there is no other difference. As Mr. Crabbe is not a painter, only because he does not use a brush and colours, so he is for the most part a poet, only because he writes in lines of ten syllables. All the rest might be found in a newspaper, an old magazine, or a county-register. Our author is himself a little jealous of the prudish fidelity of his homely Muse, and tries to justify himself by precedents. He brings as a parallel instance of merely literal description, Pope's lines on the gay Duke of Buckingham, beginning ``In the worst inn's worst room see Villiers lies!'' But surely nothing can be more dissimilar. Pope describes what is striking, Crabbe would have described merely what was there. The objects in Pope stand out to the fancy from the mixture of the mean with the gaudy, from the contrast of the scene and the character. There is an appeal to the imagination; you see what is passing in a poetical point of view. In Crabbe there is no foil, no contrast, no impulse given to the mind. It is all on a level and of a piece. In fact, there is so little connection between the subject-matter of Mr. Crabbe's lines and the ornament of rhyme which is tacked to them, that many of his verses read like serious burlesque, and the parodies which have been made upon them are hardly so quaint as the originals. Mr. Crabbe's great fault is certainly that he is a sickly, a querulous, a uniformly dissatisfied poet. He sings the country; and he sings it in a pitiful tone. He chooses this subject only to take the charm out of it, and to dispel the illusion, the glory, and the dream, which had hovered over it in golden verse from Theocritus to Cowper. He sets out with professing to overturn the theory which had hallowed a shepherd's life, and made the names of grove and valley music to our ears, in order to give us truth in its stead; but why not lay aside the fool's cap and bells at once? Why not insist on the unwelcome reality in plain prose? If our author is a poet, why trouble himself with statistics? If he is a statistic writer, why set his ill news to harsh and grating verse? The philosopher in painting the dark side of human nature may have reason on his side, and a moral lesson or remedy in view. The tragic poet, who shews the sad vicissitudes of things and the disappointments of the passions, at least strengthens our yearnings after imaginary good, and lends wings to our desires, by which we, ``at one bound, high overleap all bound'' of actual suffering. But does neither. He gives us discoloured paintings of life; helpless, repining, unprofitable, unedifying distress. He is not a philosopher, but a sophist, a misanthrope in verse; a namby-pamby Mandeville, a Malthus turned metrical romancer. He professes historical fidelity; but his vein is not dramatic; nor does he give us the \emph{pros} and \emph{cons} of that versatile gipsey, Nature. He does not indulge his fancy, or sympathise with us, or tell us how the poor feel; but how he should feel in their situation, which we do not want to know. He does not weave the web of their lives of a mingled yarn, good and ill together, but clothes them all in the same dingy linsey-woolsey, or tinges them with a green and yellow melancholy. He blocks out all possibility of good, cancels the hope, or even the wish for it as a weakness; check-mates Tityrus and Virgil at the game of pastoral cross-purposes, disables all his adversary's white pieces, and leaves none but black ones on the board. The situation of a country clergyman is not necessarily favourable to the cultivation of the Muse. He is set down, perhaps, as he thinks, in a small curacy for life, and he takes his revenge by imprisoning the reader's imagination in luckless verse. Shut out from social converse, from learned colleges and halls, where he passed his youth, he has no cordial fellow-feeling with the unlettered manners of the \emph{Village} or the \emph{Borough}; and he describes his neighbours as more uncomfortable and discontented than himself. All this while he dedicates successive volumes to rising generations of noble patrons; and while he desolates a line of coast with sterile, blighting lines, the only leaf of his books where honour, beauty, worth, or pleasure bloom, is that inscribed to the Rutland family! We might adduce instances of what we have said from every page of his works: let one suffice\textemdash \begin{verse} \vleftofline{``}Thus by himself compelled to live each day,\\ To wait for certain hours the tide's delay;\\ At the same times the same dull views to see,\\ The bounding marsh-bank and the blighted tree;\\ The water only when the tides were high,\\ When low, the mud half-covered and half-dry;\\ The sun-burnt tar that blisters on the planks,\\ And bank-side stakes in their uneven ranks;\\ Heaps of entangled weeds that slowly float,\\ As the tide rolls by the impeded boat.\\ When tides were neap, and in the sultry day,\\ Through the tall bounding mud-banks made their way,\\ Which on each side rose swelling, and below\\ The dark warm flood ran silently and slow;\\ There anchoring, Peter chose from man to hide,\\ There hang his head, and view the lazy tide\\ In its hot slimy channel slowly glide;\\ Where the small eels, that left the deeper way\\ For the warm shore, within the shallows play;\\ Where gaping muscles, left upon the mud,\\ Slope their slow passage to the fall'n flood:\\ Here dull and hopeless he'd lie down and trace\\ How side-long crabs had crawled their crooked race;\\ Or sadly listen to the tuneless cry\\ Of fishing gull or clanging golden-eye;\\ What time the sea-birds to the marsh would come,\\ And the loud bittern, from the bull-rush home,\\ Gave from the salt ditch-side the bellowing boom:\\ He nursed the feelings these dull scenes produce\\ And loved to stop beside the opening sluice;\\ Where the small stream, confined in narrow bound,\\ Ran with a dull, unvaried, saddening sound;\\ Where all, presented to the eye or ear,\\ Oppressed the soul with misery, grief, and fear.'' \end{verse} This is an exact \emph{fac-simile} of some of the most unlovely parts of the creation. Indeed the whole of Mr. Crabbe's \emph{Borough}, from which the above passage is taken, is done so to the life, that it seems almost like some sea-monster, crawled out of the neighbouring slime, and harbouring a breed of strange vermin, with a strong local scent of tar and bulge-water. Mr. Crabbe's \emph{Tales} are more readable than his \emph{Poems}; but in proportion as the interest increases, they become more oppressive. They turn, one and all, upon the same sort of teazing, helpless, mechanical, unimaginative distress;\textemdash and though it is not easy to lay them down, you never wish to take them up again. Still in this way, they are highly finished, striking, and original portraits, worked out with an eye to nature, and an intimate knowledge of the small and intricate folds of the human heart. Some of the best are the \emph{Confidant}, the story of \emph{Silly Shore}, the \emph{Young Poet}, the \emph{Painter}. The episode of \emph{} in the \emph{Village}, is one of the most tender and pensive; and the character of the methodist parson who persecutes the sailor's widow with his godly, selfish love, is one of the most profound. In a word, if Mr. Crabbe's writings do not add greatly to the store of entertaining and delightful fiction, yet they will remain ``as a thorn in the side of poetry,'' perhaps for a century to come! \documentclass{article} \usepackage[utf8]{inputenc} \title{PS5 Hoffman} \author{ } \date{March 2021} \usepackage{natbib} \usepackage{graphicx} \begin{document} \maketitle \section{Webscraping National Parks} My Grandpa and I are planning a trip to in May to see Mount Rushmore. It is one of the national monuments on his bucket list still yet to be crossed off. I decided I want to use some of my free time throughout my lifespan to travel to see all the national parks since my grandparents couldn't before my Grandma got sick. With 63 parks I have to get started soon! The data I parsed helps identify how many are located in each state/ territory and helps narrow the travel to 30 states. I used the online tutorial from , I really liked the gifs which helped remind me which steps to go through to webscrape the WIKIPEDIA page. \section{Webscraping with API for Federal Reserve of St. Louis} I found data for Washington's minimum wage rates from 1976 to 2021 from the St. Louis Federal Reserve website. As with government date, it required my API key to ensure no foul play. I found the way minimum wage has raised from 1.6 to an average 13.70 an hour. It appeared that big spikes coincided with political years as well. Washington is worth continued watching because of the 15 an hour mandate set back in 2014. I used the following packages: library(ggplot2) library(fredr) library(tidyverse) library(quantmod). \begin{figure}[h!] \centering \includegraphics[scale=0.75]{PS5_HOFFMAN.png} \caption{State Minimum Wage Rate for Washington} \label{fig:PS5_HOFFMAN} \end{figure} \end{document} J'ai passé deux jours à l'hôpital avec mon fils, les 19 et 20 mai 2021.NOTE : le fils de Leigh-Ann, âgé de 18 ans, a également souffert d'embolies pulmonaires à cause du vaccin de Johnson \& Johnson ! On lui a diagnostiqué des embolies pulmonaires bilatérales et un infarctus du poumon gauche dus au vaccin Johnson \& Johnson. Six jours plus tard, le 26 mai, l'arrière de ma jambe gauche était si douloureux que je ne pouvais pas marcher dessus. Mon fils m'a déposée aux urgences où l'on m'a fait une prise de sang et mon Dimer D était de 101630 (la normale est de 0-500). On m'a fait passer une échographie Doppler et on a découvert une TVP (caillot de sang) de 2,5 pieds de long dans ma jambe gauche. On m'a emmené en ambulance dans un autre hôpital pour le faire enlever chirurgicalement. Ils ont fait un scanner de contraste à l'hôpital suivant et ont trouvé des embolies pulmonaires. Ce qui a empêché l'opération. J'ai passé une semaine à l'hôpital sous perfusion d'héparine et 7 jours après sous injections de tinzaperin. Je suis maintenant sous anticoagulants à vie. Je suis retournée à l'hôpital une semaine plus tard avec une infection massive des tissus mous dans mon bras, mon aisselle, ma poitrine et mon côté. Mes symptômes ont commencé 19 jours après la vaccination par Pfizer, mais j'étais tellement distraite par mon fils que je n'ai été traitée qu'une semaine plus tard. Depuis, j'ai été hospitalisée deux fois pour des hémorragies menstruelles et je dois maintenant subir une hystérectomie éventuelle. J'ai un échocardiogramme, un moniteur holter, un rendez-vous en cardiologie, un rendez-vous en chirurgie vasculaire, un rendez-vous en hématologie et j'ai été envoyée en urologie pour un problème rénal. Je vis chaque jour brisée, tant physiquement que mentalement. Personne dans les hôpitaux ne se soucie de ce qui vous arrive si vous n'avez pas Covid. Pas un seul des médecins qui m'ont traité à l'hôpital n'a signalé ma blessure ! J'ai depuis signalé mon cas au ministère de la Santé de la Saskatchewan, où un médecin local a refusé ma demande de remboursement en raison de la blessure de mon fils et d'un `` trouble sanguin antérieur non diagnostiqué ``. Mon hématologue envisage de faire appel de cette décision. J'ai fait un rapport à Pfizer, qui a l'obligation de faire un rapport à Santé Canada. Aucune réponse. Je ne peux pas faire de demande d'indemnisation en raison du refus de l'OHM de ma demande initiale en juin. NOTE : le fils de Leigh-Ann, âgé de 18 ans, a également souffert d'embolies pulmonaires à cause du vaccin de Johnson \& Johnson ! \documentclass{slides} \usepackage{graphicx} \usepackage{slashbox} \usepackage{multirow} \usepackage[margin=3mm]{geometry} \usepackage[table]{xcolor} \begin{document} \rowcolors{2}{white}{blue!10} {\renewcommand{\arraystretch}{1.60}\rotatebox{90}{\resizebox{260mm}{!}{\begin{tabular}{|c|c|c|c|c|c|c|c|c|c|c|} \multicolumn{11}{c}{Each cell gives probabilities for getting x or more net hits on opposing roll, where x is 0-4, as following: $\frac{\textbf{0/1}}{2/3/4}$}\\\hline \rowcolor{blue!15} \backslashbox{Them}{You} & 21 & 22 & 23 & 24 & 25 & 26 & 27 & 28 & 29 & 30 \\ \hline 1 & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 2 & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 3 & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 4 & $\frac{\textbf{100/100}}{100/100/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 5 & $\frac{\textbf{100/100}}{100/99.9/99.5}$ & $\frac{\textbf{100/100}}{100/100/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 6 & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/100/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 7 & $\frac{\textbf{100/100}}{99.9/99.6/98.7}$ & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/100/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 8 & $\frac{\textbf{100/100}}{99.8/99.3/98.1}$ & $\frac{\textbf{100/100}}{99.9/99.5/98.7}$ & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 9 & $\frac{\textbf{100/99.9}}{99.6/98.9/97.2}$ & $\frac{\textbf{100/100}}{99.8/99.3/98.1}$ & $\frac{\textbf{100/100}}{99.9/99.5/98.8}$ & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 10 & $\frac{\textbf{100/99.8}}{99.4/98.3/96.1}$ & $\frac{\textbf{100/99.9}}{99.6/98.9/97.3}$ & $\frac{\textbf{100/100}}{99.7/99.3/98.2}$ & $\frac{\textbf{100/100}}{99.8/99.5/98.8}$ & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ & $\frac{\textbf{100/100}}{100/100/100}$ \\ \hline 11 & $\frac{\textbf{99.9/99.6}}{99.0/97.6/94.7}$ & $\frac{\textbf{100/99.8}}{99.4/98.4/96.3}$ & $\frac{\textbf{100/99.9}}{99.6/98.9/97.5}$ & $\frac{\textbf{100/100}}{99.7/99.3/98.3}$ & $\frac{\textbf{100/100}}{99.8/99.5/98.8}$ & $\frac{\textbf{100/100}}{100/99.7/99.2}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ & $\frac{\textbf{100/100}}{100/100/99.9}$ \\ \hline 12 & $\frac{\textbf{99.8/99.4}}{98.5/96.6/93.0}$ & $\frac{\textbf{99.9/99.6}}{99.0/97.7/95.0}$ & $\frac{\textbf{100/99.8}}{99.4/98.4/96.5}$ & $\frac{\textbf{100/99.9}}{99.6/99.0/97.6}$ & $\frac{\textbf{100/100}}{99.7/99.3/98.3}$ & $\frac{\textbf{100/100}}{99.8/99.6/98.9}$ & $\frac{\textbf{100/100}}{99.9/99.7/99.3}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ & $\frac{\textbf{100/100}}{100/100/99.8}$ \\ \hline 13 & $\frac{\textbf{99.7/99.1}}{97.9/95.4/91.0}$ & $\frac{\textbf{99.8/99.4}}{98.6/96.8/93.5}$ & $\frac{\textbf{99.9/99.6}}{99.1/97.8/95.3}$ & $\frac{\textbf{100/99.8}}{99.4/98.5/96.7}$ & $\frac{\textbf{100/99.9}}{99.6/99.0/97.7}$ & $\frac{\textbf{100/100}}{99.7/99.3/98.4}$ & $\frac{\textbf{100/100}}{99.8/99.6/98.9}$ & $\frac{\textbf{100/100}}{99.9/99.7/99.3}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ & $\frac{\textbf{100/100}}{100/99.9/99.7}$ \\ \hline 14 & $\frac{\textbf{99.5/98.7}}{97.1/94.0/88.7}$ & $\frac{\textbf{99.7/99.2}}{98.0/95.7/91.6}$ & $\frac{\textbf{99.8/99.4}}{98.6/97.0/93.9}$ & $\frac{\textbf{99.9/99.6}}{99.1/97.9/95.6}$ & $\frac{\textbf{100/99.8}}{99.4/98.6/96.9}$ & $\frac{\textbf{100/99.9}}{99.6/99.0/97.8}$ & $\frac{\textbf{100/100}}{99.7/99.3/98.5}$ & $\frac{\textbf{100/100}}{99.8/99.6/99.0}$ & $\frac{\textbf{100/100}}{99.9/99.7/99.3}$ & $\frac{\textbf{100/100}}{100/99.8/99.5}$ \\ \hline 15 & $\frac{\textbf{99.2/98.2}}{96.0/92.2/86.1}$ & $\frac{\textbf{99.5/98.8}}{97.2/94.3/89.5}$ & $\frac{\textbf{99.7/99.2}}{98.1/95.9/92.2}$ & $\frac{\textbf{99.8/99.5}}{98.7/97.1/94.3}$ & $\frac{\textbf{99.9/99.6}}{99.1/98.0/95.9}$ & $\frac{\textbf{100/99.8}}{99.4/98.6/97.0}$ & $\frac{\textbf{100/99.9}}{99.6/99.1/97.9}$ & $\frac{\textbf{100/100}}{99.7/99.4/98.6}$ & $\frac{\textbf{100/100}}{99.8/99.6/99.0}$ & $\frac{\textbf{100/100}}{99.9/99.7/99.3}$ \\ \hline 16 & $\frac{\textbf{98.9/97.5}}{94.8/90.2/83.1}$ & $\frac{\textbf{99.3/98.3}}{96.3/92.7/87.0}$ & $\frac{\textbf{99.5/98.8}}{97.4/94.7/90.2}$ & $\frac{\textbf{99.7/99.2}}{98.2/96.2/92.7}$ & $\frac{\textbf{99.8/99.5}}{98.7/97.3/94.6}$ & $\frac{\textbf{99.9/99.7}}{99.1/98.1/96.1}$ & $\frac{\textbf{100/99.8}}{99.4/98.7/97.2}$ & $\frac{\textbf{100/99.9}}{99.6/99.1/98.0}$ & $\frac{\textbf{100/100}}{99.7/99.4/98.6}$ & $\frac{\textbf{100/100}}{99.8/99.6/99.0}$ \\ \hline 17 & $\frac{\textbf{98.4/96.6}}{93.2/87.8/79.9}$ & $\frac{\textbf{98.9/97.6}}{95.1/90.8/84.3}$ & $\frac{\textbf{99.3/98.3}}{96.5/93.2/87.9}$ & $\frac{\textbf{99.5/98.9}}{97.5/95.0/90.9}$ & $\frac{\textbf{99.7/99.2}}{98.3/96.4/93.2}$ & $\frac{\textbf{99.8/99.5}}{98.8/97.4/95.0}$ & $\frac{\textbf{99.9/99.7}}{99.2/98.2/96.3}$ & $\frac{\textbf{100/99.8}}{99.4/98.7/97.4}$ & $\frac{\textbf{100/99.9}}{99.6/99.1/98.1}$ & $\frac{\textbf{100/100}}{99.8/99.4/98.7}$ \\ \hline 18 & $\frac{\textbf{97.8/95.5}}{91.4/85.2/76.4}$ & $\frac{\textbf{98.5/96.8}}{93.7/88.7/81.3}$ & $\frac{\textbf{99.0/97.7}}{95.4/91.4/85.4}$ & $\frac{\textbf{99.3/98.4}}{96.7/93.6/88.7}$ & $\frac{\textbf{99.5/98.9}}{97.6/95.3/91.5}$ & $\frac{\textbf{99.7/99.3}}{98.3/96.6/93.6}$ & $\frac{\textbf{99.8/99.5}}{98.8/97.6/95.3}$ & $\frac{\textbf{99.9/99.7}}{99.2/98.3/96.6}$ & $\frac{\textbf{100/99.8}}{99.5/98.8/97.5}$ & $\frac{\textbf{100/99.9}}{99.6/99.2/98.2}$ \\ \hline 19 & $\frac{\textbf{97.0/94.1}}{89.3/82.3/72.7}$ & $\frac{\textbf{97.9/95.7}}{92.0/86.2/78.0}$ & $\frac{\textbf{98.5/96.9}}{94.1/89.4/82.6}$ & $\frac{\textbf{99.0/97.8}}{95.7/92.0/86.4}$ & $\frac{\textbf{99.3/98.5}}{96.9/94.0/89.5}$ & $\frac{\textbf{99.5/98.9}}{97.8/95.6/92.0}$ & $\frac{\textbf{99.7/99.3}}{98.4/96.8/94.0}$ & $\frac{\textbf{99.8/99.5}}{98.9/97.7/95.6}$ & $\frac{\textbf{99.9/99.7}}{99.2/98.4/96.8}$ & $\frac{\textbf{100/99.8}}{99.5/98.9/97.7}$ \\ \hline 20 & $\frac{\textbf{96.0/92.5}}{87.0/79.1/68.8}$ & $\frac{\textbf{97.2/94.5}}{90.1/83.5/74.5}$ & $\frac{\textbf{98.0/96.0}}{92.5/87.1/79.5}$ & $\frac{\textbf{98.6/97.1}}{94.4/90.1/83.7}$ & $\frac{\textbf{99.0/97.9}}{95.9/92.5/87.3}$ & $\frac{\textbf{99.3/98.6}}{97.0/94.4/90.2}$ & $\frac{\textbf{99.6/99.0}}{97.9/95.9/92.5}$ & $\frac{\textbf{99.7/99.3}}{98.5/97.0/94.4}$ & $\frac{\textbf{99.8/99.5}}{98.9/97.8/95.8}$ & $\frac{\textbf{99.9/99.7}}{99.3/98.5/96.9}$ \\ \hline 21 & $\frac{\textbf{94.9/90.7}}{84.4/75.7/64.8}$ & $\frac{\textbf{96.3/93.0}}{87.9/80.5/70.9}$ & $\frac{\textbf{97.3/94.8}}{90.7/84.6/76.2}$ & $\frac{\textbf{98.1/96.2}}{93.0/88.0/80.9}$ & $\frac{\textbf{98.7/97.3}}{94.8/90.8/84.8}$ & $\frac{\textbf{99.1/98.1}}{96.2/93.0/88.1}$ & $\frac{\textbf{99.4/98.6}}{97.2/94.8/90.8}$ & $\frac{\textbf{99.6/99.0}}{98.0/96.1/93.0}$ & $\frac{\textbf{99.7/99.3}}{98.6/97.2/94.7}$ & $\frac{\textbf{99.8/99.5}}{99.0/97.9/96.1}$ \\ \hline 22 & $\frac{\textbf{93.5/88.6}}{81.5/72.1/60.8}$ & $\frac{\textbf{95.2/91.3}}{85.4/77.3/67.0}$ & $\frac{\textbf{96.5/93.5}}{88.7/81.8/72.7}$ & $\frac{\textbf{97.5/95.1}}{91.3/85.6/77.8}$ & $\frac{\textbf{98.2/96.4}}{93.5/88.8/82.1}$ & $\frac{\textbf{98.7/97.4}}{95.1/91.4/85.8}$ & $\frac{\textbf{99.1/98.2}}{96.4/93.5/88.9}$ & $\frac{\textbf{99.4/98.7}}{97.4/95.1/91.4}$ & $\frac{\textbf{99.6/99.1}}{98.1/96.4/93.5}$ & $\frac{\textbf{99.7/99.4}}{98.6/97.3/95.1}$ \\ \hline 23 & $\frac{\textbf{91.8/86.2}}{78.3/68.3/56.6}$ & $\frac{\textbf{93.9/89.3}}{82.7/73.9/63.1}$ & $\frac{\textbf{95.5/91.9}}{86.4/78.8/69.1}$ & $\frac{\textbf{96.7/93.9}}{89.4/83.0/74.5}$ & $\frac{\textbf{97.6/95.4}}{91.9/86.6/79.2}$ & $\frac{\textbf{98.3/96.7}}{93.9/89.5/83.3}$ & $\frac{\textbf{98.8/97.6}}{95.4/91.9/86.8}$ & $\frac{\textbf{99.2/98.3}}{96.6/93.9/89.6}$ & $\frac{\textbf{99.4/98.8}}{97.5/95.4/92.0}$ & $\frac{\textbf{99.6/99.1}}{98.2/96.6/93.9}$ \\ \hline 24 & $\frac{\textbf{90.0/83.6}}{75.0/64.4/52.5}$ & $\frac{\textbf{92.4/87.1}}{79.8/70.3/59.1}$ & $\frac{\textbf{94.3/90.0}}{83.8/75.5/65.3}$ & $\frac{\textbf{95.8/92.4}}{87.3/80.2/71.0}$ & $\frac{\textbf{96.9/94.3}}{90.1/84.1/76.1}$ & $\frac{\textbf{97.8/95.7}}{92.4/87.5/80.5}$ & $\frac{\textbf{98.4/96.9}}{94.3/90.2/84.4}$ & $\frac{\textbf{98.9/97.7}}{95.7/92.5/87.6}$ & $\frac{\textbf{99.2/98.3}}{96.8/94.3/90.3}$ & $\frac{\textbf{99.4/98.8}}{97.7/95.7/92.5}$ \\ \hline 25 & $\frac{\textbf{87.8/80.7}}{71.5/60.4/48.4}$ & $\frac{\textbf{90.6/84.7}}{76.6/66.6/55.1}$ & $\frac{\textbf{92.9/88.0}}{81.1/72.1/61.5}$ & $\frac{\textbf{94.6/90.7}}{84.9/77.1/67.4}$ & $\frac{\textbf{96.0/92.9}}{88.1/81.4/72.8}$ & $\frac{\textbf{97.1/94.6}}{90.8/85.2/77.6}$ & $\frac{\textbf{97.9/96.0}}{92.9/88.3/81.8}$ & $\frac{\textbf{98.5/97.0}}{94.6/90.8/85.4}$ & $\frac{\textbf{98.9/97.8}}{96.0/92.9/88.4}$ & $\frac{\textbf{99.2/98.4}}{97.0/94.6/90.9}$ \\ \hline 26 & $\frac{\textbf{85.5/77.6}}{67.8/56.4/44.5}$ & $\frac{\textbf{88.6/82.0}}{73.3/62.7/51.1}$ & $\frac{\textbf{91.2/85.7}}{78.1/68.6/57.6}$ & $\frac{\textbf{93.3/88.8}}{82.3/73.9/63.7}$ & $\frac{\textbf{95.0/91.3}}{85.9/78.6/69.3}$ & $\frac{\textbf{96.3/93.3}}{88.9/82.6/74.5}$ & $\frac{\textbf{97.2/95.0}}{91.4/86.1/79.0}$ & $\frac{\textbf{98.0/96.2}}{93.4/89.0/83.0}$ & $\frac{\textbf{98.6/97.2}}{95.0/91.4/86.3}$ & $\frac{\textbf{99.0/98.0}}{96.2/93.4/89.2}$ \\ \hline 27 & $\frac{\textbf{82.8/74.3}}{64.0/52.4/40.6}$ & $\frac{\textbf{86.4/79.1}}{69.7/58.9/47.2}$ & $\frac{\textbf{89.4/83.2}}{74.9/64.9/53.7}$ & $\frac{\textbf{91.8/86.6}}{79.5/70.5/59.9}$ & $\frac{\textbf{93.7/89.5}}{83.5/75.5/65.8}$ & $\frac{\textbf{95.3/91.9}}{86.8/79.9/71.2}$ & $\frac{\textbf{96.5/93.8}}{89.6/83.7/76.0}$ & $\frac{\textbf{97.4/95.3}}{91.9/87.0/80.3}$ & $\frac{\textbf{98.1/96.5}}{93.8/89.7/84.0}$ & $\frac{\textbf{98.6/97.4}}{95.3/92.0/87.2}$ \\ \hline 28 & $\frac{\textbf{80.0/70.9}}{60.1/48.5/36.9}$ & $\frac{\textbf{84.0/76.0}}{66.1/55.0/43.3}$ & $\frac{\textbf{87.3/80.4}}{71.6/61.2/49.8}$ & $\frac{\textbf{90.1/84.2}}{76.5/67.0/56.1}$ & $\frac{\textbf{92.3/87.5}}{80.8/72.3/62.1}$ & $\frac{\textbf{94.1/90.2}}{84.5/77.0/67.8}$ & $\frac{\textbf{95.6/92.4}}{87.7/81.2/72.9}$ & $\frac{\textbf{96.7/94.2}}{90.3/84.8/77.5}$ & $\frac{\textbf{97.6/95.6}}{92.4/87.8/81.5}$ & $\frac{\textbf{98.2/96.7}}{94.2/90.4/85.0}$ \\ \hline 29 & $\frac{\textbf{77.0/67.3}}{56.3/44.6/33.4}$ & $\frac{\textbf{81.3/72.7}}{62.4/51.1/39.6}$ & $\frac{\textbf{85.0/77.5}}{68.1/57.4/46.0}$ & $\frac{\textbf{88.1/81.7}}{73.3/63.3/52.3}$ & $\frac{\textbf{90.7/85.3}}{77.9/68.9/58.4}$ & $\frac{\textbf{92.8/88.3}}{82.0/73.9/64.2}$ & $\frac{\textbf{94.5/90.8}}{85.5/78.4/69.6}$ & $\frac{\textbf{95.8/92.9}}{88.4/82.3/74.5}$ & $\frac{\textbf{96.9/94.5}}{90.9/85.7/78.9}$ & $\frac{\textbf{97.7/95.8}}{92.9/88.6/82.7}$ \\ \hline 30 & $\frac{\textbf{73.8/63.6}}{52.4/40.9/30.0}$ & $\frac{\textbf{78.4/69.3}}{58.6/47.3/36.1}$ & $\frac{\textbf{82.5/74.4}}{64.5/53.6/42.3}$ & $\frac{\textbf{86.0/78.9}}{70.0/59.7/48.6}$ & $\frac{\textbf{88.9/82.8}}{74.9/65.4/54.7}$ & $\frac{\textbf{91.3/86.2}}{79.3/70.7/60.7}$ & $\frac{\textbf{93.3/89.0}}{83.1/75.5/66.2}$ & $\frac{\textbf{94.9/91.4}}{86.4/79.7/71.4}$ & $\frac{\textbf{96.1/93.3}}{89.2/83.4/76.0}$ & $\frac{\textbf{97.1/94.9}}{91.5/86.6/80.1}$ \\ \hline \end{tabular}}}} \end{document} 0 We separate future work into two different categories: data augmentation and fine-tuning. Given the small size of the dataset, we would like to experiment with more data augmentation methodologies. In particular, \cite{feng-etal-2021-survey} describes promising rules-based and model-based techniques. We’d also like to experiment with more novel methods. We’d like to generate headline embeddings from the labeled FIQA dataset and use them to fit a k-nearest neighbors aspect model on the FIQA data. The model could label Reuters headlines in an unsupervised manner, substantially increasing the size of the labeled dataset. Recent advancements in language model fine-tuning also show promise for increasing performance on ABSA tasks. Our attempt to adapt the DistilBERT model to the financial domain did not yield better results. \cite{felbo2017} has shown that the ‘chain-thaw’ methodology, which sequentially unfreezes and fine-tunes a single layer at a time, can improve performance on an NLP target task. \cite{howard-ruder-2018-universal} also shows that discriminative fine-tuning and slanted triangular learning rates can reduce catastrophic forgetting and improve language model accuracy on small datasets. Andrew telling his Mums Story: Mum went for her first vaccine on 3rd June 2021. Everything seemed fine except for the sore arm. On 4th June, at around 7pm she was even chatting with my Dad, but around 7:15pm, she collapsed. Dad called 995 and they guided him on the phone to do chest compressions. Fifteen minutes later, paramedics arrived, they did some emergency procedures and took her to the hospital. I rushed down to the hospital and waited outside the resuscitation room, with hope that she’ll be fine, praying and sending positive intentions. One hour passed. The doctor came out, when he started talking, I already knew… tears rolled uncontrollably. Mum passed on at 9pm. Can you imagine how that felt? Subsequently, the police investigation officer started to ask about foul play and pre-existing conditions, saying that these case notes would determine if the coroner decides to have an autopsy. I then said to the officer, why isn’t the focus of the inquiry the Pfizer Vaccine, which my mum had yesterday? Thankfully, I’m glad that the points I raised in the report led to the coroner doing the autopsy. But… the question in my mind was still, will they conclude that my mum’s death is caused by the vaccine? When the death certificate came out, the counter staff simply wanted me to sign off, take the certificate and leave. Then I asked, what was the cause of death? The counter staff pointed to the certificate saying it’s Ischaemic Heart Disease. I asked, what does that mean? She asked another senior officer to attend to me and that officer answered me saying it means it’s a “heart attack”. But I questioned again, is the heart attack caused by the Pfizer Vaccine? She answered, that can only be answered by the pathologist. I then asked, can I speak to the pathologist? She answered, you need to talk to your investigation officer to submit an inquiry to get the court to approve an inquiry. It looked like I wasn’t going to get my answer from the counter staff. I then called the investigation officer, he said he’ll submit the inquiry on my behalf but I need to be prepared to wait because Covid related inquiries currently have four months waiting time. Even though the anomaly was the vaccine, I knew I was not going to get my answer. At my mum’s wake, a friend of mine said his uncle also passed away shortly after taking the vaccine. I was starting to get suspicious. Is there a pattern? My dad also posted his story on the government’s REACH feedback unit and several netizens also reported cases of death or permanent disability after the vaccine. But here’s the thing I’m asking now. Is it possible that Covid Vaccine related injuries or deaths are classified as something else right now? That’s why, if you know of more of such cases of death or disability after the vaccine they need to be reported. I just want to seek the truth and hope that in my efforts, I can help other affected parties seek their truths, together with me. Please share this. Thank you for your kind attention. Please also look at the Story of the ‘Young Family’ as their mother also died within 8 hours of the Pfizer Vaccine in Singapore. ArashMassoudieh/GIFMod_ \hypertarget{struct_param}{}\section{Param Struct Reference} \label{struct_param}\index{Param@{Param}} \subsection*{Public Attributes} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{struct_param_ad5961716f06b01baa058c6f48a79f749}\label{struct_param_ad5961716f06b01baa058c6f48a79f749}} int {\bfseries param\+\_\+\+ID} \item \mbox{\Hypertarget{struct_param_a7f7569bbf0acf8464807758ca9f43fb5}\label{struct_param_a7f7569bbf0acf8464807758ca9f43fb5}} int {\bfseries type} \item \mbox{\Hypertarget{struct_param_ae50bd0586a30bb65f7cee49300ad9bb8}\label{struct_param_ae50bd0586a30bb65f7cee49300ad9bb8}} double {\bfseries low} \item \mbox{\Hypertarget{struct_param_a62b827f85e668ccfc873f0292041a9f5}\label{struct_param_a62b827f85e668ccfc873f0292041a9f5}} double {\bfseries high} \item \mbox{\Hypertarget{struct_param_ae8336c14ddbc6074f03e6e27774794ef}\label{struct_param_ae8336c14ddbc6074f03e6e27774794ef}} bool {\bfseries loged} \item \mbox{\Hypertarget{struct_param_a1edadf529bb1ed8c05cdbd18b02557a3}\label{struct_param_a1edadf529bb1ed8c05cdbd18b02557a3}} double {\bfseries mean} \item \mbox{\Hypertarget{struct_param_aa3bb4b0b558b550aad6aa78eb192bef7}\label{struct_param_aa3bb4b0b558b550aad6aa78eb192bef7}} double {\bfseries std} \end{DoxyCompactItemize} \subsection{Detailed Description} Definition at line 26 of file M\+C\+M\+C.\+h. The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item G\+U\+I/M\+C\+M\+C.\+h\end{DoxyCompactItemize} nmheim/thesis \section{Tikhonov Regularization} \label{sec:tikhonov_regularization} The goal of the optimization of the ESN output weights is to find a $\wmatr{out}$ such that for a given state $x_t$ we can produce a prediction of the next time step: \begin{equation} y_t = \wmatr{out} x_t \end{equation} By collecting a number of states $ \matr{X} = (x_1, ..., x_T)$, that are produced by the inputs $u_1, ..., u_T$, we can write \begin{equation} \matr{D} = \wmatr{out} \matr{X}, \end{equation} with $\matr{D} = (d_1, ..., d_T)$ as the conactenated desired outputs. The optimal output weights can then be found by simply solving the overdetermined system: \begin{equation} \wmatr{out} = \matr{D} \XT (\matr{X} \XT)^{-1} \end{equation} With this simple least squares solution, the output weights are very prone to overfitting, which is why a regularized version, also known as Tikhonov Regularization, often yields more general results, that let the ESN perform much better in the freely running mode. Tikhonov Regularization minimizes the function $\Phi$: \begin{equation} \label{eq:phi} \Phi(\wmatr{out}) = || \matr{D} - \wmatr{out} \matr{X} ||^2 + \beta^2 || \wmatr{out} ||^2, \end{equation} where the first term represents the \emph{misfit} of the outputs to the target and the second term introduces a the regularization. A larger coefficient $\beta$ will favor smaller output weights in the solution, which prevents them from overfitting. Eq.~\ref{eq:phi} can be written as: \begin{equation} \Phi(\wmatr{out}) = \bigg | \bigg| \vect{\matr{D}}{0} - \wmatr{out} \vect{\matr{X}}{\beta \matr{I}} \bigg | \bigg| ^2, \end{equation} which can be solved by: \begin{align} \vect{\matr{D}}{0} &= \wmatr{out} \XbI \\ \vect{\matr{D}}{0} \XbI^{\text{T}} &= \wmatr{out} \XbI \XbI^{\text{T}}\\ \vect{\matr{D}}{0} \XbI^{\text{T}} \bigg[\XbI \XbI^{\text{T}}\bigg]^{-1} &= \wmatr{out} \\ \wmatr{out} &= \matr{D} \XT (\matr{X}\XT + \beta^2 \matr{I})^{-1} \end{align} rugezhao/fiveMinuteStatsanalysis/citations.bib1-10 @article{Sellke.et.al.01, author={, , }, title={Calibration of $p$ values for Testing Precise Null Hypothesis}, journal={The American Statistician}, volume={55}, number={1}, year=2001 } @article{wasser.etal.07, author={, , , , , , and }, title={Using DNA to track the origin of the largest ivory seizure since the 1989 trade ban}, journal={Proceedings of the National Academy of Sciences}, volume={104}, number={10}, year=2007, month=3, pages={4228-4233} } @article{wasser.etal.08, author={, , , , , , and }, title={Combating the illegal trade in African elephat ivory with DNA forensics}, journal={Conservations Biology}, volume={22}, number={4}, year=2008, month=8, pages={1065–1071} } likhitam/Midterm \hypertarget{classtesting_1_1internal_1_1_do_both_action}{}\section{testing\+:\+:internal\+:\+:Do\+Both\+Action$<$ Action1, Action2 $>$ Class Template Reference} \label{classtesting_1_1internal_1_1_do_both_action}\index{testing\+::internal\+::\+Do\+Both\+Action$<$ Action1, Action2 $>$@{testing\+::internal\+::\+Do\+Both\+Action$<$ Action1, Action2 $>$}} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{classtesting_1_1internal_1_1_do_both_action_a55727c4dbdc1816ba6f1fe124e96088b}\label{classtesting_1_1internal_1_1_do_both_action_a55727c4dbdc1816ba6f1fe124e96088b}} {\bfseries Do\+Both\+Action} (Action1 action1, Action2 action2) \item \mbox{\Hypertarget{classtesting_1_1internal_1_1_do_both_action_a35733e2f117daad110bfbd3de84634a6}\label{classtesting_1_1internal_1_1_do_both_action_a35733e2f117daad110bfbd3de84634a6}} {\footnotesize template$<$typename F $>$ }\\{\bfseries operator Action$<$ F $>$} () const \end{DoxyCompactItemize} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item vendor/googletest/googlemock/include/gmock/gmock-\/actions.\+h\end{DoxyCompactItemize} % !TEX root = ../presentation.tex \textframe{How do I continue?} \begin{slide}{Resources} \begin{itemize} \item Deep Learning by Google @ Udacity \item http://colah.github.io \item http://cs231n.github.io \item http://www.deeplearningbook.org \item https://www.tensorflow.org \end{itemize} \end{slide} \begin{slide}{} {\huge Stay in Touch!}\\ \vspace{1cm} \begin{itemize} \item \texttt{} \item \texttt{linkedin.com/in/petergoldsborough} \item \texttt{github.com/goldsborough} \end{itemize} \pause \vspace{1cm} \texttt{github.com/peter-can-talk/pydata-london} \end{slide} \textframe{Q \& A} % AM-CV : Latex CV % Version 1.0 % Copyright 2016 <> % % Created using a combination of these two freely available CV templates % https://www.sharelatex.com/templates/cv-or-resume/professional-cv % https://www.sharelatex.com/templates/cv-or-resume/kjh-vita \ProvidesPackage{info} \def\myauthor{} \def\mytitle{Curriculum Vitae} \def\mycopyright{\myauthor} \def\mykeywords{} \def\mybibliostyle{plain} \def\mybibliocommand{} \def\mysubtitle{} \def\myaffiliation{University of Latex} \def\mydepartment{School of CV Writing} \def\mydepartmentemail{http://www.cv.latex.ac.uk/} \def\myaddrone{Address One} \def\myaddrtwo{Address Two} \def\myaddrthree{Address Three} \def\myemail{} \def\myweb{johnsmith.com} \def\myphone{00000 000000} \def\mytwitter{@jsmith} \def\myversion{} \def\myrevision{}ML121-Laboratorio de Circuitos Eléctricos/Informe 4/informef.tex \documentclass[a4paper,12pt]{report} \usepackage[spanish,mexico]{babel} \usepackage[utf8]{inputenc} \usepackage[T1]{fontenc} \usepackage{amsmath} \usepackage{amssymb} \usepackage{wasysym} \usepackage[dvipsnames,pdftex]{color} \usepackage[x11names]{xcolor} \usepackage{tikz, tkz-euclide} \usepackage[american]{circuitikz} \usepackage{siunitx} \usetikzlibrary{arrows} \usepackage[colorinlistoftodos]{todonotes} %\usepackage[left=2cm,right=1.5cm,top=1cm,bottom=1cm]{geometry} %\usepackage{helvet} %\renewcommand{\familydefault}{\sfdefault} \setlength{\oddsidemargin}{0in} \usepackage{geometry} \geometry{a4paper, total = {180mm,270mm}, left = 25mm, top = 20mm, right=15mm,bottom=20mm,% footskip=10mm} \usepackage{float} % \setlength{\topmargin}{0in} % \setlength{\voffset}{-0.5in} % \setlength{\hoffset}{0.3in} % \setlength{\textheight}{700pt} % \setlength{\textwidth}{440pt} % \setlength{\topskip}{0in} % \setlength{\parskip}{2ex} \renewcommand{\baselinestretch}{1.5} \usepackage{diagbox} \usepackage{array} \usepackage{listings} \usepackage{caption} %%% comandos definidos por el usuario \begin{document} \setcounter{page}{1} \pagenumbering{roman} \thispagestyle{empty} \begin{center} {\huge UNIVERSIDAD NACIONAL DE INGENIERÍA}\\[0.9cm] {\Large FACULTAD DE INGENIERÍA MECÁNICA}\\[0.6in] \end{center} \begin{figure}[h] \begin{center} \includegraphics[scale=0.33]{logoUNI.png} \vspace{0cm} \end{center} \end{figure} \vspace{0.5cm} \begin{center} INFORME DE LABORATORIO\\ LABORATORIO DE CIRCUITOS ELÉCTRICOS\\[5mm] {\large USO DEL GENERADOR DE ONDAS Y DEL OSCILOSCOPIO: VALORES CARACTERÍSTICOS DE ONDAS PERIÓDICAS}\\[10mm] \vfill LIMA - PERÚ \hfill OCTUBRE 2019 \end{center} \newpage \thispagestyle{empty} \begin{center} {\Large USO DEL GENERADOR DE ONDAS Y DEL OSCILOSCOPIO: VALORES CARACTERÍSTICOS DE ONDAS PERIÓDICAS}\\[0.7cm] \small ENTREGADO:\\[0.05cm] \small 09 OCTUBRE 2019\\[1.2cm] \end{center} \begin{flushleft} {\large ALUMNOS:}\\[2cm] \end{flushleft} \begin{center} \begin{tabular}{c@{\hspace{0.5in}}c} \rule[1pt]{2.6in}{1pt}&\rule[1pt]{2.6in}{1pt}\\ , 20130509E & , 20174070\\[1.5cm] \rule[1pt]{2.6in}{1pt}&\rule[1pt]{2.6in}{1pt}\\ , 20172024J & , 20170270C \\[1.5cm] %Sotelo , 20172125K & Nombre 5, 2017 \\[1.5cm] \end{tabular} \end{center} %\begin{center} %\begin{tabular}{c@{\hspace{0.6in}}c} %\rule[1pt]{3.14in}{1pt}\\ %, 20174070I \\[2cm] %\rule[1pt]{3.14in}{1pt}\\ %, 20174070I \\[2cm] %\rule[1pt]{3.14in}{1pt}\\ %, 20172125K \\[2cm] %\rule[1pt]{3.14in}{1pt}\\ %, 20172125K \\[2cm] %\end{tabular} %\end{center} \begin{center} \begin{tabular}{c} \rule[1pt]{3.14in}{1pt}\\ , 20172125K\\[2.5cm] \end{tabular} \end{center} %\rule[1pt]{3.14in}{1pt}\\ %, 20172019F \\[3cm] %\rule[1pt]{3.14in}{1pt}\\ %, 19774147I \\[3cm] %\rule[1pt]{3.14in}{1pt}\\ %, 20172125K %\end{tabular} %\end{center} %\\[0.7cm] {\large PROFESOR:} \\[2cm] \begin{center} \begin{tabular}{c} \rule[3pt]{4.8in}{1pt}\\[1pt] , FRANCISCO \end{tabular} \end{center} \vfill %\newpage %\begin{center} %{\Large \bf{RESUMEN}} %\end{center} \newpage \tableofcontents %\listoffigures %\addcontentsline{toc}{chapter}{Índice de figuras} \newpage \pagenumbering{arabic} %%% esto es para regresar el modo de numeración a numeración arábiga \setcounter{page}{1} %%% empezamos en página 1 %\part{Introducción} \chapter{Objetivos} \begin{enumerate} \item Interactuar con los diodos y apreciar su característica restrictiva aplicada a la modificación de una onda. \item Aprender a utilizar el osciloscopio digital. \item Comparar los valores medios y eficaces visualizados por el multímetro y osciloscopio con los calculados teóricamente. \end{enumerate} \chapter{Marco teórico} \section{Osciloscopio} Un osciloscopio es un instrumento de visualización electrónico para la representación gráfica de señales eléctricas que pueden variar en el tiempo. Es muy usado en electrónica de señal, frecuentemente junto a un analizador de espectro. Presenta los valores de las señales eléctricas en forma de coordenadas en una pantalla, en la que normalmente el eje $X$ (horizontal) representa tiempos y el eje $Y$ (vertical) representa tensiones. La imagen así obtenida se denomina oscilograma. Suelen incluir otra entrada, llamada ``eje THRASHER'' o ``Cilindro de Wehnelt'' que controla la luminosidad del haz, permitiendo resaltar o apagar algunos segmentos de la traza.\\ Los osciloscopios, clasificados según su funcionamiento interno, pueden ser tanto analógicos como digitales, siendo el resultado mostrado idéntico en cualquiera de los dos casos. \subsection{Osciloscopio digital} En la actualidad los osciloscopios analógicos están siendo desplazados en gran medida por los osciloscopios digitales, entre otras razones por la facilidad de poder transferir las medidas a una computadora personal o pantalla LCD. En el osciloscopio digital la señal es previamente digitalizada por un conversor analógico digital. Al depender la fiabilidad de la visualización de la calidad de este componente, esta debe ser cuidada al máximo.\\ Las características y procedimientos señalados para los osciloscopios analógicos son aplicables a los digitales. Sin embargo, en estos se tienen posibilidades adicionales, tales como el disparo anticipado (pre-triggering) para la visualización de eventos de corta duración, o la memorización del oscilograma transfiriendo los datos a un PC. Esto permite comparar medidas realizadas en el mismo punto de un circuito o elemento. Existen asimismo equipos que combinan etapas analógicas y digitales.\\ La principal característica de un osciloscopio digital es la velocidad de muestreo, la misma determinara el ancho de banda máximo que puede medir el instrumento basándose en el Teorema de Nyquist. Viene expresada en MS/s (millones de samples /muestras/ por segundo).\\ La mayoría de los osciloscopios digitales en la actualidad están basados en control por FPGA (del inglés Field Programmable Gate Array), el cual es el elemento controlador del conversor analógico a digital de alta velocidad del aparato y demás circuitería interna, como memoria, buffers, entre otros.\\ Estos osciloscopios añaden prestaciones y facilidades al usuario imposibles de obtener con circuitería analógica, como los siguientes: \begin{itemize} \item Medida automática de valores de pico, máximos y mínimos de señal. Verdadero valor eficaz. \item Medida de flancos de la señal y otros intervalos. \item Captura de transitorios. \item Cálculos avanzados, como la FFT para calcular el espectro de la señal. También sirve para medir señales de tensión. \item Dentro del osciloscopio digital existen dos tipos que se diferencian claramente: \end{itemize} \begin{figure}[H] \centering \includegraphics[scale=0.35]{osciloscopio.jpg} \end{figure} \section{Función peripodica} En matemática, una función es periódica si verifica la condición $f(x+T)=f(x)$; el número $T$ se llama periodo de la función. Generalmente, se llama período fundamental al menor número real positivo $T$ que satisface la condición. Las funciones trigonométricas son ejemplos sencillos de una función periódica, que en combinaciones adecuadas se emplean en el análisis armónico. De la misma manera, pero en un contexto físico, las ondas periódicas son aquellas ondas que muestran periodicidad respecto del tiempo, es decir, describen ciclos repetitivos. En una onda periódica se cumple: $$ x_{a}(t) = x_{a}(t+T_{p}) = x_{a}(t+nT_{p}) $$ donde el periodo propio fundamental $T_{p} = 1/F$, $F$ es la frecuencia de la componente fundamental de la onda periódica y $n$ un número entero. Toda onda periódica es, por definición, una onda determinista, por cuanto puede ser descrita matemáticamente (mediante un modelo matemático). \begin{figure}[H] \centering \includegraphics[scale=1.1]{periodica.png} \end{figure} \section{Rectificador} Un rectificador es el dispositivo electrónico que permite convertir la corriente alterna en corriente continua. Esto se realiza utilizando diodos rectificadores, ya sean semiconductores de estado sólido, válvulas al vacío o válvulas gaseosas como las de vapor de mercurio (actualmente en desuso). Dependiendo de las características de la alimentación en corriente alterna que emplean, se les clasifica en monofásicos, cuando están alimentados por una fase de la red eléctrica, o trifásicos cuando se alimentan por tres fases. Atendiendo al tipo de rectificación, pueden ser de media onda, cuando solo se utiliza uno de los semiciclos de la corriente, o de onda completa, donde ambos semiciclos son aprovechados. El tipo más básico de rectificador es el rectificador monofásico de media onda, constituido por un único diodo entre la fuente de alimentación alterna y la carga. \begin{figure}[H] \centering \includegraphics[scale=0.2]{rectificador.jpg} \end{figure} \chapter{Cuestionario} \begin{enumerate} \item Explicar el principio de funcionamiento del osciloscopio y el generador de ondas. Asimismo enumerar sus diversos usos.\\ Los osciloscopios comprueban y muestran las señales de tensión como formas de onda y como representaciones visuales de la variación de tensión en función del tiempo. Las señales se representan en un gráfico, que muestra cómo cambia la señal. El eje vertical (Y) representa la medición de la tensión, y el eje horizontal (X) representa el tiempo. El muestreo es el proceso de convertir una parte de una señal de entrada en un número de valores eléctricos discretos con el propósito de almacenarla, procesarla y visualizarla. La magnitud de cada punto de muestra es igual a la amplitud de la señal de entrada en el momento en que la señal es muestreada. La forma de onda de entrada aparece como una serie de puntos en la pantalla. Si los puntos se encuentran espaciados ampliamente y son difíciles de interpretar como una forma de onda, se pueden conectar usando un proceso llamado interpolación, que conecta los puntos con líneas, o vectores. Los controles de disparo le permiten estabilizar y mostrar una forma de onda repetitiva. El disparo por flanco es la forma más común de disparo. En este modo, el nivel de disparo y los controles de pendiente proporcionan la definición básica del punto de disparo. El control de pendiente determina si el punto de disparo se encuentra en el flanco ascendente o descendente de una señal, y el control de nivel determina en qué lugar del flanco se produce el punto de disparo. Cuando se trabaja con señales complejas, como una serie de pulsos, puede ser necesario usar el disparo por ancho de pulso. Con esta técnica, el ajuste de nivel de disparo y el próximo flanco descendente de la señal deben ocurrir dentro de un lapso de tiempo especificado. Una vez alcanzadas estas dos condiciones, el osciloscopio disparará. Otra técnica es la de disparo único, en la que el osciloscopio solo mostrará un trazo cuando la señal de entrada cumpla con las condiciones establecidas de disparo. Una vez cumplidas las condiciones de disparo, el osciloscopio adquiere y actualiza la pantalla, y luego congela la pantalla para mantener el trazo. \subsection*{Uso del osciloscopio} \begin{enumerate} \item Determinar directamente el periodo y el voltaje de una señal. \item Determinar indirectamente la frecuencia de una señal. \item Determinar que parte de la señal es DC y cual AC. \item Localizar averías en un circuito. \item Medir la fase entre dos señales. \item Determinar que parte de la señal es ruido y como varia este en el tiempo. \end{enumerate} \item Explicar el principio de funcionamiento del diodo y del puente de diodos y su aplicación en electricidad. \\ Los primeros diodos que aparecieron eran válvulas o tubos vacíos llamados válvulas termoiónicas y que se encontraban construidos por medio de dos electrodos rodeados de vació en un tubo de cristal, muy similares a las lámparas incandescentes.\\ El que es polarizado directamente permite el flujo a través de él de los electrones, o lo que es lo mismo permite el paso de la corriente eléctrica, en polarización inversa no permite el paso de los electrones por él. Para tensiones con polarización directa del diodo, según aumentamos la tensión en los bornes del diodo (patillas o extremos) va aumentando la corriente que circula por él.\\ Lógicamente el diodo tendrá una tensión máxima de trabajo que no se podrá sobrepasar porque se quemaría, para tensiones con polarización negativa no conduce y, por lo tanto, por mucho que aumentemos la tensión no se producirá corriente alguna a través del diodo.\\ Como vemos los diodos semiconductores tienen la valiosa propiedad de que los electrones solamente fluyen en una dirección a través de ellos y, como resultado, actúan como unos rectificadores. Son la estructura fundamental de los semiconductores y muchos otros componentes electrónicos se fabrican teniendo como base a los diodos. Los diodos tienen una estructura electrónica llamada Unión PN, es decir son la unión de un material semiconductor llamado N con otro llamado P.\\ \textbf{Puente de diodos} Es también llamado puente rectificador de diodos, es un dispositivo formado por cuatro diodos ensamblados de forma que una corriente alterna (AC) conectada a dos de los diodos produce una corriente continua (DC) de salida en los dos diodos restantes. Es un componente eléctrico utilizado en muchos aparatos tanto a nivel industrial como a nivel doméstico, por ejemplo, en los cargadores de los teléfonos móviles.\\ Para entender cómo funciona un puente rectificador de diodos es necesario primero conocer las diferencias básicas entre corriente alterna y corriente continua y como funciona un diodo. La mayoría de la gente está más que habituada a utilizar baterías en electrodomésticos, juguetes, teléfonos, dispositivos multimedia y otros muchos objetos cotidianos. Una batería es un buen ejemplo de fuente de alimentación de corriente contínua (CC o DC – Direct Current) pues tienen un polo positivo y un polo negativo que nunca cambian, presenta una polaridad continua. Por el contrario, la corriente alterna (CA o AC – Alternating Current) tiene una polaridad que se invierte aproximadamente entre 50 y 60 veces por segundo.\\ Para hacer que funcionen aparatos CC con un suministro de CA es necesario rectificar la polaridad alternante de la corriente CA y producir una corriente con polaridad estable. Sin esta rectificación, la corriente CA podría provocar daños graves en el aparato. En la mayoría de aparatos la rectificación se consigue con un puente de diodos o con un inversor de voltaje.\\ Los diodos son unos dispositivos que permiten el flujo eléctrico en un sólo sentido y cada puente rectificador lleva al menos cuatro. Para que la rectificación de la corriente alterna se produzca hay que conectar los cuatros diodos en una disposición específica, llamada configuración de rectificación, que efectivamente termina con una mitad del ciclo de la corriente alterna de entrada y deja pasar sólo la otra mitad a través del puente, pero siempre con el polo negativo y el positivo saliendo por los mismos diodos. \begin{figure}[H] \centering \includegraphics[scale=0.5]{sergod1.png} \end{figure} La señal eléctrica generada es forma de pulsos, la llamada media onda de rectificación. Antes de utilizarse es necesario estabilizarla para formar una señal de corriente DC completa, lo que se hace principalmente utilizando condensadores. Finalmente, y si es necesario, la señal puede pasar por un amplificador antes de abandonar el puente. Los puentes de diodos se utilizan para una amplia gama de aplicaciones. Podemos encontrarlos desde en pequeñas fuentes de alimentación de circuitos electrónicos hasta en grandes aplicaciones industriales que suministran corriente continua a motores y electroimanes. El tamaño de los diodos y los demás componentes del puente cambian según las necesidades de cada aplicación, pero la disposición y construcción permanece muy similar. Aunque existen otras alternativas para transformar corriente alterna en corriente continua, como los inversores de voltaje, los puentes de diodos siguen siendo la alternativa más barata. \begin{figure}[H] \centering \includegraphics[scale=0.5]{sergod2.png} \end{figure} \item Explicar el método empleado para hallar el desfasaje entre voltajes de la fuente y del condensador en un circuito R-C. ¿Qué otros métodos existen? Para hallar el desfasaje entre voltajes usamos el ociloscopio digital de manera que usamos los 2 canales para identificar las ondas periodicas. En el primer canal conectamos los terminales del condensador y en el segundo conectamos los terminales del voltaje. Acomodamos y superponemos las graficas y obtenemos el valor del desfasaje.\\ \item Elaborar un cuadro de los valores eficaces y medios visualizados en el multímetro, osciloscopio y los calculados teóricamente por fórmulas, indicando \% de error. \begin{figure}[H] \centering \includegraphics[scale=0.7]{kgda1.png} \end{figure} \begin{figure}[H] \centering \includegraphics[scale=0.45]{kgda2.png} \end{figure} \end{enumerate} \chapter{Conclusiones y recomendaciones} \begin{enumerate} \item Se concluyó que el puente de diodos usado tiene aplicaciones óptimas. \item En el laboratorio el primer puente de diodos que obtuvimos no estuvo en buenas condiciones, al cambiar de puente se pudo apreciar las gráficas demandadas. \item El osciloscopio y generador de ondas deben estar en buen estado para que no exista un problema con las frecuencias que se manden uno al otro. \item El puente de diodos al estar malogrados afecta las frecuencias emitidas en el osciloscopio y no muestra la onda que se necesita por ello debe observar si los puentes de diodos estan en buen estado. \item Los cables deben engancharse bien entre si sino se perdera las graficas sinusoidales en el osciloscopio. \end{enumerate} \begin{thebibliography}{99} %%%este es un contador para el número de bibliografías utilizados. \addcontentsline{toc}{chapter}{Bibliograf\'{\i}a} %%% Para introducir la bibliografía en el índice. %\bibitem{Rahman}{Rahman,Aminur y Doe, Hidekazu; ``Ion transfer of tetraalkylammonium cations at an interface between %frozen aqueous solution and 1,2-dichloroethane".{\em{Journal of Electroanalytical Chemistry}} {\bfseries 424},159,(1997).} \bibitem{Gro}{Boylestad, . ``Introducción al análisis de circuitos''. {\em{Pearson}}} \bibitem{Gro}{Sadiku, . ``Fundamemtos de circuitos eléctricos''. {\em{}}} %\bibitem{Ding}{. ``Spectroelectrochemistry and photoelectrochemistry of charge transfer at liquid/liquid %interfaces". {\em {Tesis, EPFL,}}(1999).} %\bibitem{AL}{. \em{Técnicas de mecanizado 1}} %\bibitem{AL}{Alonso, . ``Técnicas de mecanizado 1". {\em{Paraninfo}} {\bfseries España-Madrid}, 6-20, (2001).} %\bibitem{Samec2}{., ., ., . ``Interfacial tension and impedance measurements %of interfaces between two inmiscible electrolyte solutions". {\em{Journal of Electroanalytical Chemistry}} {\bfseries %43}, 47, (2000).} %\bibitem{Day}{Day R.A. y Underwood A.L. {\textit{Química Analítica Cuantitativa}},5ºed. Prentice-Hall, México, 1998. 45-48.} %\bibitem{Keyser}{, Michel. ``Determinación de la vida útil en herramientales de corte endurecido por el proceso de borurización en pasta''. {\em{Instituto tecnológico y de estudios superiores de Monterrey}}} %\bibitem{Zolotorevski}{Escalona, I. ``Máquinas: herramientas por arranque de viruta.''.{\em{El Cid Editor.}}} %\bibitem{Lasheras}{Lasheras. ``Tecnología de los Materiales Industriales''.} %\bibitem{Dieter}{Dieter. ``Metalurgia mecánica''.} %\bibitem{Apraiz}{. ``Tratamiento Térmico de los Aceros''.} %\bibitem{Smith}{Smith, . y ``Ciencia e ingeniería de materiales". {\em{ %Madrid: McGraw-Hill, Interamericana de España.}} 570, (2004).} %\bibitem{Callister}{Callister, . y Rethwisch, . ``Introducción a la ingeniería de los materiales''. %{\em{Barcelona Reverté.}}, 960, (2007).} %\bibitem{Askeland}{Askeland, ., y . ``Ciencia e ingeniería de los materiales''.{\em{México, D.F. Internacional Thomson Editores.}} {\textit{$6^{ta}$ edición}}, 1004, (2012).} %\bibitem{HARDBANDING}{Tabla de conversión de escala de durezas. \begin{verbatim}http://%hardbandingsolutions.com/postle_sp/hardness.php %\end{verbatim}} \bibitem{HE}{Apuntes circuitos transitorios. \begin{verbatim} http://users.df.uba.ar/moreno/cursos/lab3/apuntes/transitorios.pdf \end{verbatim}} %\bibitem{ASTM}{Normas ASTM.} %\bibitem{NTP}{Normas NTP.} \end{thebibliography} \end{document}\subsection{pipes -- Interface to shell pipelines} To be done .... % sjgknight/starter-hugo-research-group0 @article{shumThereNothingGood2007, author = {, Simon}, citation = {https://scholar.google.com/citations?view\textsubscriptop=view\textsubscriptcitation&hl=en&user=t3774iUAAAAJ&cstart=200&pagesize=100&citation\textsubscriptfor\textsubscriptview=t3774iUAAAAJ:t6usbXjVLHcC}, doi = {10.1109/MS.2007.148}, journal = {IEEE Software}, number = {5}, pages = {21--23}, title = {There's Nothing like a Good Argument...}, type = {Journal Article}, volume = {24}, year = {2007} } 1-10 % lang.tex % % programming languages to learn % % \subsection{Bash, perl, lisp, Lua} \subsection{Python data science: scikit-learn, matplotlib, pandas, numpy, statsmodels} \subsection{Julia $\leftarrow$ (Matlab, Octave, C, Python)} \subsection{Haskell} \subsection{JavaScript} \subsection{TensorFlow} \subsection{Theano} \subsection{Keras} \subsection{PyTorch} \subsection{Robotics frameworks: OpenCV, PCL, ROS} @article{gomez-ramirez_dont_2013, abstract = {The seriousness of the current crisis urgently demands new economic thinking that breaks the austerity vs. deficit spending circle in economic policy. The core tenet of the paper is that the most important problems that natural and social science are facing today are inverse problems, and that a new approach that goes beyond optimization is necessary. The approach presented here is radical in the sense that it identifies the roots in key assumptions in economic theory such as optimal behavior and stability to provide an inverse thinking perspective to economic modeling, of use in economic and financial stability policy. The inverse problem provides a truly multidisciplinary platform where related problems from different disciplines can be studied under a common approach with comparable results.}, author = {}, doi = {10.1007/s40309-013-0013-6}, file = {Full Text PDF:/Users/jagomez/Library/Application Support/Firefox/Profiles/4otj4fuq.default/zotero/storage/XNB2TX8N/Gomez-Ramirez - 2013 - Don’t blame the economists. It is an inverse probl.pdf:application/pdf;Snapshot:/Users/jagomez/Library/Application Support/Firefox/Profiles/4otj4fuq.default/zotero/storage/GV7D4HHQ/s40309-013-0013-6.html:text/html}, issn = {2195-4194, 2195-2248}, journal = {European Journal of Futures Research}, keywords = {Bio-inspired homeostasis, Complexity economics, History of Science, Inverse problem, Science, general, Social Sciences, general, Subjective expected utility}, language = {en}, number = {1}, pages = {1--7}, title = {Don't blame the economists. It is an inverse problem!}, url = {http://link.springer.com/article/10.1007/s40309-013-0013-6}, urldate = {2013-08-06}, volume = {1}, year = {2013} } \begin{center} \thispagestyle{plain} \pagenumbering{gobble} \vspace*{5cm} \textsc{\huge\textbf{Ubuntu/Debian Help Guide}}\\[2.5cm] \hrule \vspace{0.5cm} {\Huge \textbf{Detailed set-up, and usage instructions for various Ubuntu configurations}} \\[0.5cm] \hrule \vspace{2.5cm} \LARGE \emph{Author:}\\ \vfill \large Version 0.8.1 published \today ~using \LaTeX \end{center} \newpage \begin{center} \vspace*{1cm} {\Huge\textsc{Acknowledgements}}\\[4.5cm] \textit{To all those that have helped me compile this document over the years, thank you.}\\[1.5cm] \begin{flushright} \rule{5cm}{0.25mm}\\[0.1cm] {\large\textsl{Adam}} \end{flushright} \vfill \large \textit{(First compiled December 2011)} \end{center} \pagenumbering{roman} \tableofcontents \listoftables \listoffigures\documentclass{report} \usepackage[T2A]{fontenc} \usepackage[utf8]{luainputenc} \usepackage[english, russian]{babel} \usepackage[pdftex]{hyperref} \usepackage[14pt]{extsizes} \usepackage{listings} \usepackage{color} \usepackage{geometry} \usepackage{enumitem} \usepackage{multirow} \usepackage{graphicx} \usepackage{indentfirst} \geometry{a4paper,top=2cm,bottom=3cm,left=2cm,right=1.5cm} \setlength{\parskip}{0.5cm} \setlist{nolistsep, itemsep=0.3cm,parsep=0pt} \lstset{language=C++, basicstyle=\footnotesize, keywordstyle=\color{blue}\ttfamily, stringstyle=\color{red}\ttfamily, commentstyle=\color{green}\ttfamily, morecomment=[l][\color{magenta}]{\#}, tabsize=4, breaklines=true, breakatwhitespace=true, title=\lstname, } \makeatletter \renewcommand\@biblabel[1]{#1.\hfil} \makeatother \begin{document} \begin{titlepage} \begin{center} Министерство науки и высшего образования Российской Федерации \end{center} \begin{center} Федеральное государственное автономное образовательное учреждение высшего образования \\ Национальный исследовательский Нижегородский государственный университет им. Н.И. Лобачевского \end{center} \begin{center} Институт информационных технологий, математики и механики \end{center} \vspace{4em} \begin{center} \textbf{\LargeОтчет по лабораторной работе} \\ \end{center} \begin{center} \textbf{\Large«Умножение разреженных матриц. Элементы комплексного типа. Формат хранения матрицы – столбцовый (CCS).»} \\ \end{center} \vspace{4em} \newbox{\lbox} \savebox{\lbox}{\hbox{text}} \newlength{\maxl} \setlength{\maxl}{\wd\lbox} \hfill\parbox{7cm}{ \hspace*{5cm}\hspace*{-5cm}\textbf{Выполнила:} \\ студентка группы 381806-2 \\ Киселева А.С.\\ \\ \hspace*{5cm}\hspace*{-5cm}\textbf{Проверил:}\\ доцент кафедры МОСТ, \\ кандидат технических наук \\ Сысоев А. В.\\ } \vspace{\fill} \begin{center} Нижний Новгород \\ 2021 \end{center} \end{titlepage} \setcounter{page}{2} % Содержание \tableofcontents \newpage % Введение \section*{Введение} \addcontentsline{toc}{section}{Введение} Термин « матрица » имеет много значений. Например, в математике матрицей называется система элементов, имеющая вид прямоугольной таблицы, в программировании матрица – это двумерный массив, в электронике – набор проводников, которые можно замкнуть в точках их пересечений.Матрицы широко применяются в математике для компактной записи систем линейных алгебраических или дифференциальных уравнений. В этом случае, количество строк матрицы соответствует числу уравнений, а количество столбцов — количеству неизвестных. Обычно матрицы хранятся в двумерных массивах, но данный способ неособо удобен в случае разреженных матриц (матрицы, в которых большая часть элементов нули). В данном случае выгоднее хранить матрицу в CCS (столбцовый) или CRS (строчный). \par В данной лабораторной работе будет рассмотрен столбцовый формат хранения (CCS) разреженной матрицы и операция умножения таких матриц. Также будут рассмотрены два способа распараллеливания для уменьшения скорости вычислений. \newpage % Постановка задачи \section*{Постановка задачи} \addcontentsline{toc}{section}{Постановка задачи} В лабораторной работе необходимо реализовать последовательную и паралелльные версии умножения разреженных матриц с элементами комплексного типа в формате хранения CCS. \par Для реализации параллельных версий необходимо использовать средства OpenMP и TBB, а для проверки работы алгоритмов требуется использовать Google C++ Testing Framework. \newpage % Метод решения \section*{Метод решения} \addcontentsline{toc}{section}{Метод решения} В данной лабораторной работе используются разреженные матрицы с комплексными элементами. Для удобства используется typedef std::vector> ComplexMatr. Но сначала создаются матрицы, хранящиеся в одномерных векторах, с заданным числом ненулевых элементов (n). Далее они перезаписываются в CCS матрицы. \par Алгоритм создания CCS матрицы. \begin{itemize} \item Создаются 2 вектора размером равным количеству ненулевых элементов (n) : Values,Rows. И один вектор размером n+1 : ColumnIndexes. \item Записываем значения всех ненулевых элементов в Values. Порядок записи: столбец за столбцом. \item Записываем номер строки в которой находится каждый ненулевой элемент в Value. \item Элементы столбца i в векторе Values находятся по индексам от ColumnIndexes[i] до ColumnIndexes[i+ 1] –1 включительно. При этом количество ненулевых элементов в i столбце равен ColumnIndexes[i+1] - ColumnIndexes[i]. \end{itemize} \par Алгоритм умножения CCS матриц. \begin{itemize} \item Проверка условия возможности умножения матриц : количество столбцов левой = количество строк правой. \item Нахождение ненулевого элемента в обоих матрицах. Проводится проверка векторов ColumnIndexes: ColumnIndexes[i+1] - ColumnIndexes[i] не равен нулю. Если равен нулю, то ненулевых элементов нет, тогда в результирующую матрицу записывается 0. \item Если на данной позиции в векторах Rows совпадают значения, то производится сложение элементов из соответсвующих позиций Values. \end{itemize} \newpage % Схема распараллеливания \section*{Схема распараллеливания} \par В данной задаче берется цикл по строкам левой матрицы и внутренний по столбцам правой матрицы. Для распараллеливания целесообразнее распараллелить внешний цикл, то есть мы независимо рассматриваем строки левой матрицы. Данный способ более удобен, так как при умножении складываются элементы определенной строки левой матрицы. \par Для OpenMР данный цикл выглядит следующим образом: \begin{lstlisting} std::complex tmp1(std::complex(0, 0)); #pragma omp parallel { #pragma omp for private(tmp1) schedule(static) for (int i = 0; i < ma; i++) { for (int j = 0; j < mb; j++) \end{lstlisting} \par Для TBB: \begin{lstlisting} tbb::parallel_for(tbb::blocked_range(0, ma, ma/100), [&](const tbb::blocked_range& r) { std::complex tmp1(std::complex(0, 0)); for (int i = r.begin(); i < r.end(); i++) { for (int j = 0; j < mb; j++) { \end{lstlisting} \newpage % Описание программной реализации \section*{Описание программной реализации} \addcontentsline{toc}{section}{Описание программной реализации} Матрица подается в виде одномерного вектора, в котором элементы хранятся столбец за столбцом. Для удобства и эффективности сначала нужно правую матрицу транспонировать, то есть строки сделать столбцами, а столбцы строками. \par Функция для создания обычной матрицы : \begin{lstlisting} ComplexMatr rand(int n, int m, int nz); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item n-количество строк, \item m-количество столбцов, \item nz-количество ненулевых элементов. \end{itemize} \par Функция транспонирования : \begin{lstlisting} ComplexMatr transp(ComplexMatr a, int n, int m); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr a - матрица, \item n-количество строк, \item m-количество столбцов. \end{itemize} \par Функция создания вектора Values : \begin{lstlisting} ComplexMatr value(ComplexMatr matr, int n, int m); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr matr - матрица, \item n-количество строк, \item m-количество столбцов, \end{itemize} \par Функция создания вектора Rows: \begin{lstlisting} std::vector rows(ComplexMatr matr, int n, int m); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr matr - матрица, \item n-количество строк, \item m-количество столбцов. \end{itemize} \par Функция создания вектора ColumnIndexes: \begin{lstlisting} std::vector index(ComplexMatr matr, int n, int m); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr matr - матрица, \item n-количество строк, \item m-количество столбцов. \end{itemize} \par Функция последовательного умножения разреженных матриц: \begin{lstlisting} ComplexMatr umn_posled(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr valA - вектор Values левой матрицы, \item ComplexMatr valB - вектор Values правой матрицы, \item std::vector rowsA - вектор Rows левой матрицы, \item std::vector rowsB - вектор Rows правой матрицы, \item std::vector indexA - вектор ColumnIndexes левой матрицы, \item std::vector indexB - вектор ColumnIndexes правой матрицы, \item na-количество строк левой матрицы, \item ma-количество столбцов левой матрицы, \item nb-количество строк правой матрицы, \item mb-количество столбцов правой матрицы. \end{itemize} \par Функция параллельного умножения разреженных матриц (OpenMP): \begin{lstlisting} ComplexMatr umn_parallel(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr valA - вектор Values левой матрицы, \item ComplexMatr valB - вектор Values правой матрицы, \item std::vector rowsA - вектор Rows левой матрицы, \item std::vector rowsB - вектор Rows правой матрицы, \item std::vector indexA - вектор ColumnIndexes левой матрицы, \item std::vector indexB - вектор ColumnIndexes правой матрицы, \item na-количество строк левой матрицы, \item ma-количество столбцов левой матрицы, \item nb-количество строк правой матрицы, \item mb-количество столбцов правой матрицы. \end{itemize} \par Функция параллельного умножения разреженных матриц (TBB): \begin{lstlisting} ComplexMatr umn_parallel_tbb(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); \end{lstlisting} \par В качестве входных параметров передаются: \begin{itemize} \item ComplexMatr valA - вектор Values левой матрицы, \item ComplexMatr valB - вектор Values правой матрицы, \item std::vector rowsA - вектор Rows левой матрицы, \item std::vector rowsB - вектор Rows правой матрицы, \item std::vector indexA - вектор ColumnIndexes левой матрицы, \item std::vector indexB - вектор ColumnIndexes правой матрицы, \item na-количество строк левой матрицы, \item ma-количество столбцов левой матрицы, \item nb-количество строк правой матрицы, \item mb-количество столбцов правой матрицы. \end{itemize} \newpage % Тестирование \section*{Тестирование} \addcontentsline{toc}{section}{Тестирование} Для подтверждения работоспособности и эффективности в программе представлен набор тестов, разработанных с помощью использования Google C++ Testing Framework. \par Данные тесты проверяют правильность выполнения всех функций, а также эффективность выполнения последовательного и параллельных методов. \newpage % Результаты экспериментов \section*{Результаты экспериментов} \addcontentsline{toc}{section}{Результаты экспериментов} Тестирования проводились на компьютере обладающим, следующими характеристиками: \begin{itemize} \item Процессор: Intel® Core™ i7-9750H CPU @ 2.60GHz × 12. \item Оперативная память: 8 ГБ DDR4; \item ОС: Microsoft Windows 10 Домашняя для одного языка. \end{itemize} \par Проведем тест для матриц 25000x10000 и 10000x25000, в которых 110 и 80 ненулевых элементов соответственно. Количество используемых потоков определяется автоматически. \par Результаты : \begin{table}[!h] \caption{Результаты теста} \centering \begin{tabular}{p{2cm} p{5cm} p{4cm} p{4cm}} Номер & Последовательно & OpenMP & TBB \\ 1 & 8.0011 & 3.552471 & 2.684977 \\ 2 & 7.92055 & 3.68566 & 3.264525 \\ 3 & 9.07078 & 3.95639 & 3.80332 \\ Среднее & 8.33081 & 3.731507 & 3.05145 \\ \end{tabular} \end{table} \begin{table}[!h] \caption{Результаты теста} \centering \begin{tabular}{p{2cm} p{5cm} p{4cm} p{4cm}} Номер & Ускорение OpenMP & Ускорение TBB \\ 1 & 2.26 & 2.98 \\ 2 & 2.15 & 2.43 \\ 3 & 2.3 & 2.39\\ Среднее & 2.23 & 2.73 \\ \end{tabular} \end{table} \par Исходя из полученных данных, можно сделать вывод, что эффективность программы зависит от выбранного способа распараллеливания и от размера матриц.Реализации OpenMP и TBB показывают практически одинаковые результаты.Но TBB незначительлно медленнее. Также стоит заметить, что вычисления умножения матриц в обычном представлении заняло бы значительно больше времени. \newpage % Заключение \section*{Заключение} \addcontentsline{toc}{section}{Заключение} В данной лабораторной работе были реализованы последовательная и параллельные версии умножения разреженных матриц со столбцовой формой хранения (CCS). Также важно, что параллельная реализация является более эффективной, чем последовательная. Чтобы сделать выполнение параллельных программы еще эффективнее можно добавить еще потоков, которые будут выполняться одновременно. \newpage % Список литературы \begin{thebibliography}{1} \addcontentsline{toc}{section}{Список литературы} \bibitem{Gergel} . Теория и практика параллельных вычислений. – 2007. \bibitem{Reinders} (2007, July). Intel Threading Building Blocks: Outfitting C++ for Multi-core Processor Parallelism \bibitem{Voss} ., , . (2019) «Pro TBB. C++ Parallel Programming with Threading Building Blocks» \bibitem{Kunle} . Chip Multiprocessor Architecture - Techniques to Improve Throughput and Latency. — Morgan and Claypool Publishers, 2007. \bibitem{Mario} , . Multithreading Architecture. — Morgan and Claypool Publishers, 2013. \bibitem{Reginald} . Sparse Matrices. — Academic Press, 1973. \end{thebibliography} \newpage % Приложение \section*{Приложение} \addcontentsline{toc}{section}{Приложение} \begin{lstlisting} // umnrazr.h // Copyright 2021 #ifndef MODULES_TASK_3_KISELEVA_RAZR_UMN_TBB_UMNRAZR_H_ #define MODULES_TASK_3_KISELEVA_RAZR_UMN_TBB_UMNRAZR_H_ #include #include #include #include #include #include #include #include "tbb/blocked_range.h" #include "tbb/parallel_for.h" #include "tbb/tick_count.h" #define SIZE_ERROR -2 typedef std::vector> ComplexMatr; ComplexMatr rand(int n, int m, int nz); ComplexMatr transp(ComplexMatr a, int n, int m); ComplexMatr value(ComplexMatr matr, int n, int m); std::vector rows(ComplexMatr matr, int n, int m); std::vector index(ComplexMatr matr, int n, int m); ComplexMatr umn_posled(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); ComplexMatr umn_parallel(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); ComplexMatr umn_parallel_tbb(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb); #endif // MODULES_TASK_3_KISELEVA_RAZR_UMN_TBB_UMNRAZR_H_ \end{lstlisting} \begin{lstlisting} // umnrazr.cpp // Copyright 2021 #include "../../../modules/task_3/kiseleva_razr_umn_tbb/umnrazr.h" ComplexMatr rand(int n, int m, int nz) { ComplexMatr matr; std::vector tmp; std::vector num; for (int i = 0; i < n * m; i++) { tmp.push_back(i); } std::random_device rd; std::mt19937 g(rd()); shuffle(tmp.begin(), tmp.end(), g); for (int i = 0; i < nz; i++) { num.push_back(tmp[i]); } std::mt19937 gen; gen.seed(static_cast(time(0))); for (int i = 0; i < n * m; i++) { int s = 0; for (int j = 0; j < nz; j++) { if (i == num[j]) { s = 1; } } if (s == 0) { matr.push_back(std::complex(0, 0)); } else { matr.push_back(std::complex(gen(), gen())); } } return matr; } ComplexMatr transp(ComplexMatr a, int n, int m) { ComplexMatr res; for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { res.push_back(a[i + j * n]); } } return res; } ComplexMatr value(ComplexMatr matr, int n, int m) { ComplexMatr val; for (int i = 0; i < n * m; i++) { if (matr[i] != std::complex(0, 0)) { val.push_back(matr[i]); } } return val; } std::vector rows(ComplexMatr matr, int n, int m) { std::vector row; std::vector num; int p = 0; for (int j = 0; j < n * m; j++) { if (p < n) { num.push_back(p); p++; } else { p = 0; num.push_back(p); p++; } } for (int i = 0; i < n * m; i++) { if (matr[i] != std::complex(0, 0)) { row.push_back(num[i]); } } return row; } std::vector index(ComplexMatr matr, int n, int m) { std::vector ind; std::vector tmp; ind.push_back(0); int p = 0; int k = 1; for (int i = 0; i < n * m; i++) { if (i % n == 0) { p = 0; } if (matr[i] != std::complex(0, 0)) { p++; } if (i != 0) { if (i % (n * k - 1) == 0) { ind.push_back(p); k++; } } } tmp = ind; int s = 1; for (int i = 1; i < m; i++) { for (int j = s + 1; j < m + 1; j++) { ind[j] += tmp[s]; } s++; } return ind; } ComplexMatr umn_posled(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb) { ComplexMatr C; if (ma == nb) { int tmp; tmp = na; na = ma; ma = tmp; std::complex tmp1(std::complex(0, 0)); for (int i = 0; i < ma; i++) { for (int j = 0; j < mb; j++) { if ((indexA[i + 1] - indexA[i] == 0) || (indexB[j + 1] - indexB[j] == 0)) { C.push_back(std::complex(0, 0)); } else { for (int y = indexA[i]; y < indexA[i + 1]; y++) { for (int x = indexB[j]; x < indexB[j + 1]; x++) { if (rowsA[y] == rowsB[x]) { tmp1 += valA[y] * valB[x]; } } } C.push_back(tmp1); tmp1 = std::complex(0, 0); } } } } else { throw SIZE_ERROR; } return C; } ComplexMatr umn_parallel(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb) { ComplexMatr C(na * mb); if (ma == nb) { int tmp; tmp = na; na = ma; ma = tmp; std::complex tmp1(std::complex(0, 0)); #pragma omp parallel { #pragma omp for private(tmp1) schedule(static) for (int i = 0; i < ma; i++) { for (int j = 0; j < mb; j++) { if ((indexA[i + 1] - indexA[i] == 0) || (indexB[j + 1] - indexB[j] == 0)) { C[i * mb + j] = std::complex(0, 0); } else { for (int y = indexA[i]; y < indexA[i + 1]; y++) { for (int x = indexB[j]; x < indexB[j + 1]; x++) { if (rowsA[y] == rowsB[x]) { tmp1 += valA[y] * valB[x]; } } } C[i * mb + j] = tmp1; tmp1 = std::complex(0, 0); } } } } } else { throw SIZE_ERROR; } return C; } ComplexMatr umn_parallel_tbb(ComplexMatr valA, ComplexMatr valB, std::vector rowsA, std::vector rowsB, std::vector indexA, std::vector indexB, int na, int ma, int nb, int mb) { ComplexMatr C(na * mb); if (ma == nb) { int tmp; tmp = na; na = ma; ma = tmp; tbb::parallel_for(tbb::blocked_range(0, ma, ma/100), [&](const tbb::blocked_range& r) { std::complex tmp1(std::complex(0, 0)); for (int i = r.begin(); i < r.end(); i++) { for (int j = 0; j < mb; j++) { if ((indexA[i + 1] - indexA[i] == 0) || (indexB[j + 1] - indexB[j] == 0)) { C[i * mb + j] = std::complex(0, 0); } else { for (int y = indexA[i]; y < indexA[i + 1]; y++) { for (int x = indexB[j]; x < indexB[j + 1]; x++) { if (rowsA[y] == rowsB[x]) { tmp1 += valA[y] * valB[x]; } } } C[i * mb + j] = tmp1; tmp1 = std::complex(0, 0); } } } }); } else { throw SIZE_ERROR; } return C; } \end{lstlisting} \begin{lstlisting} // main.cpp // Copyright 2021 #include #include #include #include #include "./umnrazr.h" TEST(CCR_UMN, razm1020x100_100_10000) { int na = 1020; int ma = 100; int nb = 100; int mb = 100; int nza = 4; int nzb = 8; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } TEST(CCR_UMN, razm100x100_100x100_4_8) { int na = 100; int ma = 100; int nb = 100; int mb = 100; int nza = 4; int nzb = 8; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } TEST(CCR_UMN, razm113x85_85x21_7_1) { int na = 113; int ma = 85; int nb = 85; int mb = 21; int nza = 7; int nzb = 1; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } TEST(CCR_UMN, razm25000x10000_10000x25000_110_80) { int na = 25000; int ma = 10000; int nb = 10000; int mb = 25000; int nza = 110; int nzb = 80; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } TEST(CCR_UMN, razm10000x10000_10000x10000_10_8) { int na = 10000; int ma = 10000; int nb = 10000; int mb = 10000; int nza = 10; int nzb = 8; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } TEST(CCR_UMN, razm1000x500_500x200_10_0) { int na = 1000; int ma = 500; int nb = 500; int mb = 200; int nza = 10; int nzb = 0; ComplexMatr a = rand(na, ma, nza); ComplexMatr b = rand(nb, mb, nzb); a = transp(a, na, ma); int tmp; tmp = na; na = ma; ma = tmp; ComplexMatr valA = value(a, na, ma); ComplexMatr valB = value(b, nb, mb); std::vector rowsA = rows(a, na, ma); std::vector rowsB = rows(b, nb, mb); std::vector indexA = index(a, na, ma); std::vector indexB = index(b, nb, mb); tmp = na; na = ma; ma = tmp; tbb::tick_count start_time1, finish_time1, start_time, finish_time; start_time = tbb::tick_count::now(); ComplexMatr CCR = umn_posled(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time = tbb::tick_count::now(); start_time1 = tbb::tick_count::now(); ComplexMatr umn = umn_parallel_tbb(valA, valB, rowsA, rowsB, indexA, indexB, na, ma, nb, mb); finish_time1 = tbb::tick_count::now(); for (int i = 0; i < na * mb; i++) { ASSERT_EQ(CCR[i], umn[i]); } printf("\tPosled = %f\n", (finish_time - start_time).seconds()); printf("\tParalel = %f\n", (finish_time1 - start_time1).seconds()); } \end{lstlisting} \end{document}\begin{tiny}(Cfc09)\end{tiny} \begin{enumerate} \item On écrit des développements limités pour $x$ en $0$ à l'ordre $1$. Un $1$ se simplifie, on divise par $x>0$ et on termine par un passage à la limite dans une inégalité. \item On écrit l'inégalité de convexité pour $f^\alpha$ avec $u$, $v$ et $\lambda$. On utilise le a. avec $\alpha$ dans le rôle de $x$, $\ln(f(\lambda u+(1-\lambda)v)$ dans celui de $a$, $\ln(f(u))$ dans celui de $b$ et $\ln(f(v))$ dans celui de $c$. \end{enumerate} book/chapter3/chapter3.tex \chapter{Selection Statements} Oftentimes, programs will need to perform different actions in different scenarios. For instance, a program might need to test input to see if it is erroneous, or check which menu option a user selected. Many algorithms also rely on checking conditions to achieve their results. If we continue the programs as recipes metaphor, we can consider recipes that call for different baking temperatures depending on whether the pan you are using is metal or glass, or even depending on whether you live in a high altitude area. In this chapter, we will learn how to write programs to test for conditions and do different things based on whether they have occurred or not. \section{If statements} Tetra uses the keyword \texttt{if} to check if a condition has occurred. For instance, if we want the user to enter a positive number, we can use an \texttt{if} statement for this purpose: \def \codelabel {ch3if1} \def \codecaption{Using if to check input} \input{chapter3/if1.ttr.tex} The output of this program can be seen below: \input{chapter3/if1.ttr.out} \section{Conditions} \section{Boolean Variables} \section{Boolean Expressions} \section{Else Statements} \section{Elif Statements} \section{Summary} \section{Exercises} @book{aho.86.dragon, author = {. and .}, title = {Compilers: Principles, Techniques and Tools}, publisher = {Addison-Wesley}, year = 1986 } @article{aho.89.acmtpls, author = { and and }, title = {Code generation using tree matching and dynamic programming}, journal = {ACM Trans. Program. Lang. Syst.}, volume = 11, number = 4, year = 1989, issn = {0164-0925}, pages = {491--516}, doi = {http://doi.acm.org/10.1145/69558.75700}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Article{aiken.96.sigplan, author = {}, title = {Cool: A Portable Project for Teaching Compiler Construction}, journal = {ACM SIGPLAN Notices}, month = {July}, volume = 31, number = 7, year = 1996, issn = {0362-1340}, pages = {19--24}, doi = {http://doi.acm.org/10.1145/381841.381847}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{algol.star, author = {The Electronic Sky}, title = {Algol Demon Star, Beta Persei, 26 Persei}, howpublished = {\url{http://www.glyphweb.com/esky/stars/algol.html}}, year = 2005 } @Misc{antiquecomputers, oldkeys = {AntiqueComputers}, author = {}, title = {{Facts and stories about Antique (lonesome) Computers}}, howpublished = {\url{http://ed-thelen.org/comp-hist/index.html}}, year = 2003 } @TechReport{appel.87.concise, author = {}, title = {Concise specifications of locally optimal code generators}, institution = {Princeton University, Dept. of Computer Science, Princeton, New Jersey}, number = {CS-TR-080-87}, year = 1987, month = {February}, url = {citeseer.nj.nec.com/appel87concise.html}, abstract = {Dynamic programming allows locally optimal instruction selection for expression trees. More importantly, the algorithm allows concise and elegant specification of code generators. Aho, Ganapathi, and Tjiang have built the Twig code-generator-generator, which produces dynamic-programming code-generators from grammar-like specifications. Encoding a complex architecture as a grammar for a dynamic-programming codegenerator -generator shows the expressive power of the technique. Each instruction, addressing mode, register and class can be expressed individually in the grammar. The grammar can be factored much more readily than with the Graham-Glanville LR(1) algorithm, so it can be much more concise. Twig specifications for the VAX and MC68020 are described, and the corresponding code generators select very good (and under the right assumptions, optimal) instruction sequences. Limitations and possible improvements to the specification language are discussed.}, } @book{appel.98.modern, author = {}, title = {Modern Compiler Implementation in {C}, {Java}, {ML}}, publisher = {Cambridge University Press}, year = 1998 } @InProceedings{appelbe.79.sigcse, author = {be}, title = {Teaching Compiler Development}, booktitle = {{SIGCSE'79}: Proceedings of the tenth {SIGCSE} technical symposium on Computer science education}, year = 1979, pages = {23--27}, doi = {http://doi.acm.org/10.1145/800126.809546}, publisher = {ACM Press}, address = {New York, NY, USA}, } @misc{apple-history.www, title = {Apple History}, author = {}, howpublished = {\url{http://www.apple-history/.com/}}, year = 2006 } @manual{arm, oldkeys = {ARM}, key = {ARM}, title = {Ada Reference Manual}, year = 1983 } @Misc{artmuseum, oldkeys = {ArtMuseum}, author = {ArtMuseum}, title = {{In Depth}}, howpublished = {\url{http://www.artmuseum.net/w2vr/archives/archives.html}}, year = 2003 } @Misc{asn1con, author = {ASN.1 Consortium}, title = {{The ASN.1 Consortium}}, howpublished = {\url{http://www.asn1.org/}}, year = 2003 } @Misc{asn1info, author = {}, title = {{The ASN.1 Information Site}}, howpublished = {\url{http://asn1.elibel.tm.fr/en/}}, year = 2003 } @Misc{aterm, author = {{Centrum voor Wiskunde en Informatica}}, title = {{The ATerm Library}}, howpublished = {\url{http://www.cwi.nl/htbin/sen1/twiki/bin/view/SEN1/ATermLibrary}}, year = 2004 } @Misc{autoconf.www, author = { and and }, title = {{GNU Autoconf}}, howpublished = {\url{http://www.gnu.org/software/autoconf/}}, year = 2003, } @Misc{automake.www, oldkeys = {automake}, author = { and }, title = {{GNU Automake}}, howpublished = {\url{http://www.gnu.org/software/automake/}}, year = 2003, } @Article{aycock.02.crossroads, author = {}, title = {Why {Bison} is Becoming Extinct}, journal = {ACM Crossroads}, year = 2002, volume = 7, number = 5, } @InProceedings{aycock.03.sigplan, author = {}, title = {The {ART} of Compiler Construction Projects}, booktitle = {{ACM SIGPLAN Notices}}, pages = {28--32}, year = 2003, volume = 38, number = 12, month = {Dec}, } @TechReport{aycock.95.mbl, author = {}, title = {{MBL}: A Language for Teaching Compiler Construction}, institution = {Department of Computer Science, University of Calgary}, year = 1995, number = {1995-574-26}, } @InProceedings{baldwin.03.sigcse, author = {}, title = {A Compiler for Teaching about Compilers}, booktitle = {Proceedings of the 34th {SIGCSE} technical symposium on Computer science education ({SIGCSE}'03)}, pages = {19--23}, year = 2003, address = {Reno, Nevada, USA}, month = {February}, abstract = {Projects in which students write a small compiler are common in compiler design courses, but even a small compiler can be daunting to students with no prior compiler-writing experience. When I recently taught compiler design, I developed a very small language with a highly modular compiler, focusing the project on implementing the core parts of a compiler without requiring students to build all the infrastructure from scratch. This paper describes the language and its compiler, and their successes (and occasional limitations) for teaching compiler design.}, } @inproceedings{baxter.04.icse, author = { and and }, title = {{DMS}\textregistered: Program Transformations for Practical Scalable Software Evolution}, booktitle = {Proceedings of the 26th International Conference on Software Engineering (ICSE)}, year = 2004, isbn = {0-7695-2163-0}, pages = {625--634}, publisher = {IEEE Computer Society}, address = {Washington, DC, USA}, } @Misc{ben-kiki.05.yaml, author = { and and }, title = {{YAML} Specification Index}, howpublished = {\url{http://www.yaml.org/spec/}}, year = 2005 } @Misc{bordeleau.03, oldkeys = {Bordeleau}, author = {}, title = {{L'Histoire des Technologies Informatiques et quelques-unes de leurs Applications en \'Education}}, howpublished = {\url{http://www.scedu.umontreal.ca/sites/histoiredestec/}}, year = 2003 } @PhdThesis{briggs.92.phd, oldkeys = {Briggs92, briggs92, briggs.92}, author = {}, title = {Register Allocation via Graph Coloring}, school = {Rice University, Houston, Texas}, year = 1992, month = {April}, } @Article{cardelli.96.csur, oldkeys = {cardelli.96.acmcs}, author = {}, title = {Bad engineering properties of object-oriented languages}, journal = {ACM Computing Surveys (CSUR)}, volume = 28, number = {4es}, year = 1996, issn = {0360-0300}, pages = 150, doi = {http://doi.acm.org/10.1145/242224.242415}, publisher = {ACM Press}, howpublished = {\url{http://lucacardelli.name/Papers/BadPropertiesOfOO.pdf}}, lrdedoc = {comp/prog/oo/cardelli.96.csur.pdf}, } @article{cattell.80.acmtpls, author = {}, title = {Automatic Derivation of Code Generators from Machine Descriptions}, journal = {ACM Trans. Program. Lang. Syst.}, volume = 2, number = 2, year = 1980, issn = {0164-0925}, pages = {173--190}, doi = {http://doi.acm.org/10.1145/357094.357097}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{chroncomphist, oldkeys = {ChronCompHist}, author = {HCS Virtual computer history museum}, title = {{A Chronology of Computer History}}, howpublished = {\url{http://www.cyberstreet.com/hcs/museum/chron.htm}}, year = 2003 } @Misc{computer50, oldkeys = {Computer50}, author = {Computer50}, title = {Computer 50 Web Site}, howpublished = {\url{http://www.computer50.org/}}, year = 2002 } @Misc{computermuseum, author = {Computer Museum of America}, howpublished = {\url{http://www.computer-museum.org}}, year = 2002 } @article{conway.63.jacm, author = {}, title = {Design of a separable transition-diagram compiler}, journal = {Commun. ACM}, volume = {6}, issue = {7}, month = {July}, year = {1963}, issn = {0001-0782}, pages = {396--408}, numpages = {13}, url = {http://doi.acm.org/10.1145/366663.366704}, doi = {http://doi.acm.org/10.1145/366663.366704}, acmid = {366704}, publisher = {ACM}, address = {New York, NY, USA}, abstract = {A {COBOL} compiler design is presented which is compact enough to permit rapid, one-pass compilation of a large sub- set of COBOL on a moderately large computer. Versions of the same compiler for smaller machines require only two work- ing tapes plus a compiler tape. The methods given are largely applicable to the construction of {ALGOL} compilers.} } @Book{cooper.04.engineering, oldkeys = {cooper.2004.engineering}, author = { and }, title = {Engineering a Compiler}, publisher = {Morgan Kaufmann}, year = 2004, isbn = {1-55860-698-X}, } @InProceedings{debray.02.sigcse, author = {}, title = {Making compiler design relevant for students who will (most likely) never design a compiler}, booktitle = {Proceedings of the 33rd SIGCSE technical symposium on Computer science education}, year = 2002, isbn = {5-58113-473-8}, pages = {341--345}, location = {Cincinnati, Kentucky}, doi = {http://doi.acm.org/10.1145/563340.563473}, publisher = {ACM Press}, } @Article{dewar.08.crosstalk, author = { and }, title = {Computer Science Education: Where Are the Software Engineers of Tomorrow?}, journal = {CrossTalk, The Journal of Defense Software Engineering}, year = 2008, volume = 21, number = 1, pages = {28--30}, month = {January}, lrdedoc = {comp/edu/dewar.08.crosstalk.pdf}, url = {http://www.stsc.hill.af.mil/CrossTalk/2008/01/0801DewarSchonberg.html}, } @article{dijkstra.72, author = {}, title = {The Humble Programmer.}, journal = {Commun. ACM}, volume = 15, number = 10, year = 1972, pages = {859-866}, howpublished = {\url{http://www.cs.utexas.edu/users/EWD/ewd03xx/EWD340.PDF}} , bibsource = {DBLP, http://dblp.uni-trier.de} } @InProceedings{diwan.04.jeric, author = { and and and }, title = {{PL}-detective: A system for teaching programming language concepts}, journal = {Journal on Educational Resources in Computing ({JERIC})}, volume = 4, number = 4, year = 2004, issn = {1531-4278}, pages = 1, doi = {http://doi.acm.org/10.1145/1086339.1086340}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{duret-lutz.06.autotools, author = {}, title = {{The Autotools Tutorial}}, howpublished = {\url{http://www-src.lip6.fr/homepages/Alexandre.Duret-Lutz/dl/autotools.pdf/}}, year = 2006, } @Misc{edwards.03.ln, oldkeys = {EdwardsLN, edwardsln}, author = {}, title = {{COMS W4115 Programming Languages and Translators}}, howpublished = {\url{http://www.cs.columbia.edu/~sedwards/classes/2003/w4115/}}, year = 2003 } @Manual{flex.07.manual, oldkeys = {flex}, title = {The Flex Manual}, author = { and and }, organization = {The Flex Project}, month = {September}, year = 2007, note = {\url{http://flex.sourceforge.net/manual/index.html}} } @misc{fourmilab.07.univac, title = {Univac Memories}, author = {}, howpublished = {\url{http://www.fourmilab.ch/documents/univac/}}, year = 2007, } @Misc{franklinbio, oldkeys = {FranklinBio}, author = {Corrosion Doctors}, Title = {}, howpublished = {\url{http://www.corrosion-doctors.org/Biographies/FranklinBio.htm}}, year = 2004, } @Misc{franklinelec, oldkeys = {FranklinElec}, author = {}, Title = {Franklin's Electricity}, howpublished = {\url{http://www.uh.edu/engines/epi510.htm}}, year = 2004, } @InProceedings{gagnon.98.tools, author = { and }, title = {{SableCC}, an Object-Oriented Compiler Framework}, booktitle = {Proceedings of the Technology of Object-Oriented Languages and Systems (TOOLS)}, year = 1998, isbn = {0-8186-8482-8}, pages = {140--154}, publisher = {IEEE Computer Society}, address = {Washington, DC, USA}, } @Misc{gccxml, author = {}, Title = {GCC-XML}, howpublished = {\url{http://gccxml.org}}, year = 2004, } @Manual{glibc, title = {The {GNU} {C} Library Reference Manual}, author = { and and and and }, organization = {Free Software Foundation}, address = {59 Temple Place -- Suite 330, Boston, MA 02111-1307 USA}, edition = {0.10}, year = 2003, } @Article{goldberg.91.csur, author = {}, title = {What every computer scientist should know about floating-point arithmetic}, journal = {ACM Computing Surveys (CSUR)}, volume = 23, number = 1, year = 1991, issn = {0360-0300}, pages = {5--48}, doi = {http://doi.acm.org/10.1145/103162.103163}, publisher = {ACM Press}, address = {New York, NY, USA}, howpublished = {\url{http://docs.sun.com/source/806-3568/ncg_goldberg.html}}, } @Manual{gpc, title = {The GNU Pascal Compiler Manual}, organization = {Free Software Foundation}, address = {59 Temple Place -- Suite 330, Boston, MA 02111-1307 USA}, edition = 20030830, year = 2003, } @InProceedings{guimar.84.sigcse, author = {\~aes}, title = {Testing student-made compilers}, journal = {{SIGCSE} Bulletin}, volume = 26, number = 3, year = 1994, issn = {0097-8418}, pages = {37--44}, note = {\url{http://doi.acm.org/10.1145/187387.187402}}, publisher = {ACM Press}, address = {New York, NY, USA}, } @misc{guy.71.ladybird, author = {}, title = {How It Works...The Computer}, howpublished = {\url{http://davidguy.brinkster.net/computer/}}, year = 1971 } @Misc{havm.www, author = {}, Title = {\textsc{Havm}}, howpublished = {\url{http://tiger.lrde.epita.fr/Havm}}, year = 2003, } @inproceedings{haynes.98.sigcse, author = {}, title = {Experience with an analytic approach to teaching programming languages}, booktitle = {{SIGCSE '98}: Proceedings of the twenty-ninth {SIGCSE} technical symposium on Computer science education}, year = 1998, isbn = {0-89791-994-7}, pages = {350--354}, location = {Atlanta, Georgia, United States}, doi = {http://doi.acm.org/10.1145/273133.274328}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{helloworld.www, oldkeys = {HelloWorld, helloworld}, author = {ACM}, title = {{ACM ``Hello World'' Project}}, howpublished = {\url{http://www2.latech.edu/~acm/HelloWorld.shtml/}}, year = 2003 } @article{hildum.90.tse, author = { and }, title = {A Language for Specifying Program Transformations}, journal = {IEEE Trans. Softw. Eng.}, volume = 16, number = 6, year = 1990, issn = {0098-5589}, pages = {630--638}, doi = {http://dx.doi.org/10.1109/32.55091}, publisher = {IEEE Press}, address = {Piscataway, NJ, USA}, } @Misc{historycomputing.www, author = {The History of Computing Foundation}, title = {The History of Computing Project}, howpublished = {\url{http://www.thocp.net/}}, year = 2000 } @Misc{holmevik.03.simula, oldkeys = {Holmevik, holmevik}, author = {}, title = {{The History of Simula}}, howpublished = {\url{http://java.sun.com/people/jag/SimulaHistory.html}}, year = 2003 } @misc{holt.72, author = {}, title = {Teaching the Fatal Disease (or) Introductory Computer Programming Using {PL/I}}, institution = {Dept. of Computer Science and Computer Systems Research Group, University of Toronto}, year = 1972, month = {November}, howpublished = {\url{http://plg.uwaterloo.ca/~holt/papers/fatal_disease.html }} } @Misc{hpmuseum, oldkeys = {HPMuseum, hpmuseum}, author = {}, title = {{The Museum of HP Calculators}}, howpublished = {\url{http://hpmuseum.com/}}, year = 2003 } @misc{huggins.06.hopper, author = {}, title = {{Grace Murray Hopper}}, howpublished = {\url{http://www.jamesshuggins.com/h/tek1/grace_hopper.htm}}, year = 2006 } @article{hughes.89.cj, oldkeys = {Hughes89, hughes.89}, author = {}, title = {Why functional programming matters}, journal = {The Computer Journal}, year = 1989, volume = 32, number = 2, month = Apr, pages = {98--107}, note = {\url{http://www.math.chalmers.se/~rjmh/Papers/whyfp.html}} } @Misc{ibm.06.pl/i, author = {IBM}, title = {Enterprise {PL/I} for {z/OS}, Version 3.6, Language Reference, Arithmetic operations}, url = {\url{http://publib.boulder.ibm.com/infocenter/pdthelp/v1r1/index.jsp?topic=/com.ibm.debugtool7.doc/eqa7ug0202.htm}}, } @Misc{ieee.98.float, author = {ANSI/IEEE}, title = {{ANSI/IEEE} Std 854-1987 -- {IEEE} standard for radix-independent floating-point arithmetic}, year = 1987, } @Misc{infoage, oldkeys = {InfoAge, infoage}, author = {{National Museum of American History}}, title = {{Information Age: People, Information \& Technology}}, howpublished = {\url{http://photo2.si.edu/infoage/infoage.html}}, year = 2003 } @Misc{intel4004, oldkeys = {Intel4004, intel4004}, author = {{National Museum of American History}}, title = {{Intel 4004 Microprocessor}}, howpublished = {\url{http://micro.magnet.fsu.edu/micro/gallery/computers/computer2.html}}, year = 2003 } @Misc{intel4004it, oldkeys = {Intel4004It, intel4004it}, author = {{'informatica}}, title = {{Intel 4004}}, howpublished = {\url{http://www.khet.net/gmc/docs/museum/Intel4004.html}}, year = 2003 } @Misc{introspector, oldkeys = {Introspector, introspector}, author = {{}}, title = {{The Introspector Project}}, howpublished = {\url{http://introspector.sourceforge.net/}}, year = 2004 } @Misc{josuttis.99, author = {}, title = {The {C++} Standard Library: A Tutorial and Reference}, publisher = {Add}, year = 1999 } @Article{kay.77, author = { and }, title = {Personal Dynamic Media}, journal = {Computer}, year = 1977, volume = 31, month = {March} } @Misc{king.www, author = {}, year = 1993, title = {Photos from History Of Programming Languages II}, howpublished = {\url{http://www2.gsu.edu/~matknk/hopl.html}} } @article{kuipers.03.scp, author = { and }, title = {Object-oriented tree traversal with {JJForester}}, Journal = {Science of Computer Programming}, volume = 47, number = 1, year = 2003, issn = {0167-6423}, pages = {59--87}, doi = {http://dx.doi.org/10.1016/S0167-6423(02)00108-9}, publisher = {Elsevier North-Holland, Inc.}, address = {Amsterdam, The Netherlands}, lrdedoc = {comp/compilers/kuipers.03.scp.pdf}, } @TechReport{larus.90.spim, oldkeys = {LarusSpim}, author = {}, title = {\textsc{SPIM S20}: A \textsc{MIPS R2000} Simulator}, institution = {Computer Sciences Department, University of Wisconsin--Madison}, year = 1990, number = {TR966}, } @InProceedings{lattner.03.GCCSummit, oldkeys = {lattner.2003.GCCSummit}, Author = { and }, Title = {Architecture for a Next-Generation GCC}, Booktitle = {Proc. First Annual GCC Developers' Summit}, Address = {Ottawa, Canada}, Month = {May}, Year = 2003, URL = {http://llvm.cs.uiuc.edu/pubs/2003-05-01-GCCSummit2003.html} } @Misc{lavine, oldkeys = {Lavine}, author = {}, title = {{Computer Museum of America: Arthur Lavine Exhibition}}, howpublished = {\url{http://www.computer-museum.org/exhibits/lavine/index.html}}, year = 2003 } @TechReport{leung.96.prop, author = {}, title = {{Prop}: A {C++}-based Pattern Matching Language}, institution = {Courant Institute of Mathematical Sciences}, year = 1996, url = {http://www.cs.nyu.edu/leunga/papers/research/prop.ps} } @article{lewis.68.jacm, author = {. and .}, title = {Syntax-Directed Transduction}, journal = {J. ACM}, volume = {15}, issue = {3}, month = {July}, year = {1968}, issn = {0004-5411}, pages = {465--488}, numpages = {24}, url = {http://doi.acm.org/10.1145/321466.321477}, doi = {http://doi.acm.org/10.1145/321466.321477}, acmid = {321477}, publisher = {ACM}, address = {New York, NY, USA}, abstract = {A transduction is a mapping from one set of sequences to another. A syntax directed transduction is a particular type of transduction which is defined on the grammar of a context free language and which is meant to be a model of part of the translation process used in many compilers. The transduction is considered from an automata theory viewpoint as specifying the input-output relation of a machine. There is a close relationship between a subclass of these transductions, the simple syntax directed transductions, and push-down machines. This relationship is investigated in detail and conditions for such transductions to be performed on push-down machines are obtained. In addition, some time bounds for transductions on Turing machines are derived.} } @misc{lileks.07.compupromo, author = {}, title = {Compu-Promo}, url = {http://www.lileks.com/institute/compupromo/}, } @Book{lohr.02.goto, author = {}, title = {Go To: The Story of the Math Majors, Bridge Players, Engineers, Chess Wizards, Maverick Scientists and Iconoclasts --- The Programmers Who Created the Software Revolution}, publisher = {Basic Books}, year = 2002, ibsn = 0465042260 } @Misc{lrde.tiger, author = {}, title = {The {Tiger} Compiler Project Home Page}, howpublished = {\url{http://tiger.lrde.epita.fr}}, } @Article{markoff.02.sjmn, oldkeys = {Markoff}, author = {}, title = {Spacewar pioneers --- First video game programmers had a blast but didn't cash in}, pages = {3E}, month = {March}, year = 2002, journal = {San Jose Mercury News}, howpublished = {\url{http://ed-thelen.org/comp-hist/PDP-1-SpaceWar-Article.html}} } @inproceedings{matz.03, author = {}, title = {{Design and Implementation of a Graph Coloring Register Allocator for \textsc{gcc}}}, pages = {151--169}, year = 2003, } @misc{mccarthy.www, author = {}, howpublished = {\url{http://www-formal.stanford.edu/jmc/index.html}}, title = {jmc's web page}, year = 2006 } @InProceedings{mead.06.sigcse, author = {}, title = {A compiler tutorial scaled for the programming languages course}, booktitle = {{SIGCSE '06}: Proceedings of the 37th {SIGCSE} technical symposium on Computer science education}, year = 2006, isbn = {1-59593-259-3}, pages = {32--36}, location = {Houston, Texas, USA}, doi = {http://doi.acm.org/10.1145/1121341.1121354}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{menhir.www, author = {Fran\c{c} and }, title = {Menhir {LR(1)} Parser Generator --- Home page}, howpublished = {\url{http://pauillac.inria.fr/~fpottier/menhir/}}, year = 2005 } @misc{mohr.04.370, author = {}, title = {Computing Science 370 Programming Languages}, howpublished = {\url{http://www.augustana.ca/~mohrj/courses/2004.fall/csc370/index.html}}, year = 2004 } @Misc{mono.www, title = {{Mono} Home Page}, howpublished = {\url{http://www.mono-project.com}}, } @misc{mouse.www, title = {Mouse Site}, author = {}, url = {\url{http://sloan.stanford.edu/MouseSite/MouseSitePg1.html}}, } @Book{muchnick.97.advanced, oldkeys = {Muchnick97, muchnick.97}, author = {}, title = {Advanced Compiler Design and Implementation}, publisher = {Morgan Kaufmann Publishers}, year = 1997 } @Misc{numlock.05.scans, author = {numlock}, title = {Game/Computer Magazine Scans - 1982/83}, url = {\url{http://www.flickr.com/photos/numlok/sets/1394109/}}, } @Misc{ooshape, author = { and }, title = {{OO Example Code}}, howpublished = {\url{http://onestepback.org/articles/poly/}}, year = 2003, } @Misc{owen.03.www, author = {}, title = {Planet Sinclair}, howpublished = {\url{http://www.nvg.ntnu.no/sinclair/contents.htm}}, year = 2003, } @article{parr.95.spe, author = { and }, title = {{ANTLR}: A predicated-{LL}($k$) parser generator}, pages = {789--810}, year = 1995, volume = 25, number = 7, journal = {Software, Practice and Experience}, url = {http://citeseer.nj.nec.com/12770} } @article{partsch.83.csur, author = { and }, title = {Program Transformation Systems}, journal = {ACM Comput. Surv.}, volume = 15, number = 3, year = 1983, issn = {0360-0300}, pages = {199--236}, doi = {http://doi.acm.org/10.1145/356914.356917}, publisher = {ACM}, address = {New York, NY, USA}, } @Misc{pcmuseum:Windows, author = {Fortune City}, title = {Windows History}, howpublished = {\url{http://members.fortunecity.com/pcmuseum/windows.htm}} } @Misc{peoplebehindinformatics, title = {{People behind Informatics --- In memory of , and }}, howpublished = {\url{http://cs-exhibitions.uni-klu.ac.at/}}, year = 2003 } @Misc{pigott.06.hopl, author = {}, title = {{HOPL}: an interactive Roster of Programming Languages}, url = {\url{http://hopl.murdoch.edu.au/}}, } @Misc{pixel.07.syntax, author = {Pixel}, title = {Syntax Across Languages}, url = {\url{http://merd.sourceforge.net/pixel/language-study/syntax-across-languages/}}, } @Misc{round.06.quiz, title = {Programming Language Inventor or Serial Killer Quiz}, author = {}, howpublished = {\url{http://www.malevole.com/mv/misc/killerquiz/}}, } @InProceedings{schwartzbach.08.cc, author = {}, title = {Design Choices in a Compiler Course or How to Make Undergraduates Love Formal Notation}, booktitle = {Compiler Construction}, year = 2008, pages = {1-15}, ee = {http://dx.doi.org/10.1007/978-3-540-78791-4_1}, bibsource = {DBLP, http://dblp.uni-trier.de} } @Misc{scrivener.03, oldkeys = {Scrivener03}, author = {}, title = {{SIGKIDS 2003 Influences Timeline}}, howpublished = {\url{http://san-diego.siggraph.org/sigkids/Influences.html}}, year = 2003 } @book{sebesta.02, author = {}, title = {Concepts of programming languages (5th ed.)}, year = 2002, isbn = {0-321-17645-6}, publisher = {Addison Wesley Longman Publishing Co., Inc.}, address = {Redwood City, CA, USA}, } @InProceedings{shapiro.76.sigcse, author = { and }, title = {A new approach to teaching a first course in compiler construction}, booktitle = {Proceedings of the {ACM} {SIGCSE}-{SIGCUE} technical symposium on Computer science and education}, year = 1976, pages = {158--166}, doi = {http://doi.acm.org/10.1145/800107.803467}, publisher = {ACM Press}, address = {New York, NY, USA}, } @Misc{sharples.98, oldkeys = {Sharples98}, author = {{}}, title = {{Why did you bother to write a book?}}, howpublished = {\url{http://www.eee.bham.ac.uk/sharplem/Routledge/article.htm}}, year = 1998 } @Article{shopyrin.06.ieeesoft, author = {}, title = {Multimethods Implementation in {C++} Using Recursive Deferred Dispatching}, journal = {IEEE Software}, volume = 23, number = 3, year = 2006, issn = {0740-7459}, pages = {62--73}, doi = {http://dx.doi.org/10.1109/MS.2006.77}, publisher = {IEEE Computer Society Press}, address = {Los Alamitos, CA, USA}, lrdedoc = {comp/lang/c++/shopyrin.06.ieeesoft.pdf}, } @misc{smith.06, author = {}, title = {{PDP-1} Music in 1964}, howpublished = {\url{http://www.dpbsmith.com/pdp1music/}}, year = 2006 } @misc{smith.jed, title = {Univac Memories}, author = {}, howpublished = {\url{http://www.stewdio.org/jed/}}, year = 2007, } @misc{snopes.04.rand, author = {Snopes.com}, title = {Does Not Compute}, url = {\url{http://www.snopes.com/inboxer/hoaxes/computer.asp}}, year = 2004, } @misc{spacewar.www, author = { and and }, title = {The original {Spacewar!} for {PDP/1} emulated in {Java}}, year = 1996, howpublished = {\url{http://lcs.www.media.mit.edu/groups/el/projects/spacewar/}} } @Misc{speakandspell, author = {}, title = {{The Texas Instruments Speak \& Spell}}, howpublished = {\url{http://www.99er.net/spkspell.html}}, year = 2004 } @Misc{stephenson.03, title = {The Roman Hand-Abacus}, year = 2003, author = {}, howpublished = {\url{http://www.mat.uc.pt/~jaimecs/maistecno.html}} } @Misc{stroustrup.08.faq, author = {}, title = {{Bjarne} {Stroustrup}'s {FAQ}}, howpublished = {\url{http://www.research.att.com/~bs/bs_faq.html}}, month = {September}, year = 2008, } @Misc{stroustrup.97, author = {}, title = {The {C++} programming language third edition}, publisher = {Addison Wesley}, year = 1997 } @Misc{subversion, author = { and and }, title = {{Version Control With Subversion}}, howpublished = {\url{http://svnbook.red-bean.com}}, year = 2004 } @Misc{sunfortran77, oldkeys = {SunFortran77}, author = {{Sun Microsystems}}, title = {{Fortran 77 Language Reference (for f77 4.2)}}, howpublished = {\url{http://www.ictp.trieste.it/~manuals/programming/sun/fortran/f77rm/index.html}}, year = 1996 } @Misc{sutter.09.blog, author = {}, title = {Answer to "16 Technologies": {Engelbart} and the Mother of All Demos}, howpublished = {\url{http://herbsutter.com/2009/01/08/answer-to-16-technologies-engelbart-and-the-mother-of-all-demos/}}, month = jan, year = 2009, note = {Herb Sutter's blog post}, } @misc{tanenbaum.04.linux, title = {Some Notes on the {"Who wrote Linux"} Kerfuffle, Release 1.5}, author = {}, howpublished = {\url{lecture/compiler-construction/names/}}, note = {An interesting description of an attempt to discredit {Linux} {Torvalds}.}, year = 2004, month = {June}, } @misc{time.83.computer, author = {{Time Magazine}}, title = {The Computer, Machine of the Year}, howpublished = {\url{http://www.time.com/time/covers/0,16641,1101830103,00.html}}, year = 1983, month = {January} } @Manual{treecc, title = {Tree Compiler-Compiler}, organization = {The DotGNU Project}, note = {\url{http://dotgnu.org/treecc/treecc.html}} } @Misc{ucla-cg.05.jtb, author = {{UCLA Compilers Group}}, title = {{Java} {Tree} {Builder} ({JTB})}, howpublished = {\url{http://compilers.cs.ucla.edu/jtb/}}, year = 2005 } @Book{vanrossum.06.python, author = { and , Jr.}, title = {The {Python} Language Reference Manual (version 2.5)}, publisher = {Network Theory Ltd.}, year = 2006, month = {November}, isbn = {0-9541617-8-5}, howpublished = {http://www.network-theory.co.uk/python/language/} } @Misc{vaucher.99, oldkeys = {Vaucher}, author = {}, title = {Introduction \`a {SIMULA}}, howpublished = {\url{http://www.iro.umontreal.ca/~simula/Standard/intro.html}}, year = 1999 } @Misc{vintagecomputercollection, author = {}, title = {{Vintage Computer Collection}}, howpublished = {\url{http://www.vintage-computer.com/}}, year = 2004 } @Article{visser.01.sigplan, oldkeys = {Visser2001a}, author = {}, title = {Visitor Combination and Traversal Control}, pages = {270--282}, volume = 36, number = 11, month = nov, year = 2001, journal = {ACM SIGPLAN Notices}, note = {{OOPSLA}~2001 Conference Proceedings: Object-Oriented Programming Systems, Languages, and Applications.}, lrdedoc = {comp/compilers/visser.01.sigplan.pdf}, } @InProceedings{vollmar.06.sigcse, author = { and }, title = {{MARS}: An Education-Oriented {MIPS} Assembly Language Simulator}, booktitle = {Proceedings of the 37th {SIGCSE} technical symposium on Computer science education ({SIGCSE'06})}, year = 2006, address = {Houston, Texas, USA}, month = {March}, isbn = {1-59593-259-3}, pages = {239--243}, publisher = {ACM Press}, address = {New York, NY, USA}, doi = {http://doi.acm.org/10.1145/1121341.1121415}, abstract = {We describe the implementation of "MARS," a GUI, Java-based simulator for the MIPS assembly language. MIPS, the computer architecture underlying the simulated assembly language, is widely used in industry and is the basis of the popular textbook Computer Organization and Design [6], used at over 400 universities. The MARS simulator has been implemented with characteristics that are especially useful to undergraduate computer science students and their instructors.}, } @InProceedings{waite.06.iticse, author = { and and and }, title = {Design and Implementation of a Modern Compiler Course}, booktitle = {Proceedings of the Eleventh Annual Conference on Innovation and Technology in Computer Science Education (ITICSE)}, year = 2006, month = {June}, pages = {18--22}, address = {University of {B}ologna, {I}taly}, abstract = {Current literature states that the undergraduate curriculum can no longer afford the luxury of a traditional compiler construction course. Nevertheless, there is an increasing need for an understanding of how to design and implement domain-specific languages. This paper presents a modern course in compiler construction, designed to provide a student with the capability of quickly constructing robust processors for a variety of language-related applications.}, } @InProceedings{waite.06.sigcse, author = {}, title = {The Compiler Course in Today's Curriculum: Three Strategies}, booktitle = {Proceedings of the 37th {SIGCSE} technical symposium on Computer science education ({SIGCSE'06})}, year = 2006, address = {Houston, Texas, USA}, month = {March}, isbn = {1-59593-259-3}, pages = {87--91}, location = {Houston, Texas, USA}, doi = {http://doi.acm.org/10.1145/1121341.1121371}, publisher = {ACM Press}, address = {New York, NY, USA}, abstract = {The broadening of computer science education has called into question the roles of many traditional core courses. In order to remain viable, courses such as compiler construction must provide a coherent view of their subject matter that fits with the rest of the institution's curriculum. Three strategies have evolved for this course. As described in this paper, each strategy provides a model that a professor can use to design an appropriate course for their situation.}, } @misc{watchmen.www, author = {, , }, title = {The Faces in Front of the Monitors}, note = {Many picture of people who made the computer history.}, howpublished = {\url{http://www.wbglinks.net/pages/watchmen/}}, year = 2006 } @misc{wbg.06.www, title = {A Computer Geek's History of the Internet}, author = {{WBGLinks}}, howpublished = {\url{http://www.wbglinks.net/pages/history/}}, year = 2006, } @Misc{weatherley.02, author = {}, title = {{Treecc}, the {Tree Compiler-Compiler}}, howpublished = {\url{http://www.southern-storm.com.au/treecc.html}}, year = 2002 } @Misc{weatherley.02.www, author = {}, title = {{Treecc}: An Aspect-Oriented Approach to Writing Compilers}, howpublished = {\url{http://dotgnu.org/treecc_essay.html}}, year = 2002 } @article{werner.03.jcsc, author = {}, title = {A parser project in a programming languages course}, journal = {J. Comput. Small Coll.}, volume = 18, number = 5, year = 2003, pages = {184--192}, publisher = {Consortium for Computing Sciences in Colleges}, address = {, USA}, abstract = {This experience report describes a programming exercise designed to reinforce concepts of parsing, regular and context-free grammars, first sets, and abstract syntax trees. Students build a parser for a language with real applications. To make this feasible tools including a compiler compilers and automated support for the visitor design pattern, are used. The exercise is extensible.} } @Misc{wikipedia, author = {Wikipedia}, title = {Wikipedia, free encyclopedia}, year = 2005, howpublished = {\url{http://en.wikipedia.org/wiki/Main_Page}} } @misc{wikipedia.07.atari-2600, author = {Wikipedia}, title = {Atari 2600 --- Wikipedia{,} The Free Encyclopedia}, year = 2007, url = {http://en.wikipedia.org/w/index.php?title=Atari_2600&oldid=111764801}, note = "[Online; accessed 5-March-2007]" } @Book{wilson.93.lpc, oldkeys = {Wilson1993, wilson.93}, author = { and }, editor = {?}, title = {Langages de Programmation Compar\'es}, publisher = {Addison-Wesley}, year = 1993, edition = {2nd}, month = {November}, } @Misc{wirth.www, oldkeys = {Wirth}, author = {}, title = {{Miklaus Wirth Home Page}}, howpublished = {\url{http://www.cs.inf.ethz.ch/~wirth/}}, year = 1999 } @Misc{yaxx.www, key = {Yaxx}, author = { and }, title = {{YAXX}: {YA}cc e{X}tension to {XML}, a User Manual}, howpublished = {\url{http://yaxx.sourceforge.net/}}, year = 2003 } @Article{zendra.97.sigplan, author = { and and }, title = {Efficient dynamic dispatch without virtual function tables: the {SmallEiffel} compiler}, journal = {{ACM SIGPLAN Notices}}, volume = {32}, number = {10}, year = {1997}, issn = {0362-1340}, pages = {125--141}, doi = {http://doi.acm.org/10.1145/263700.263728}, publisher = {ACM}, address = {New York, NY, USA}, url = {http://smarteiffel.loria.fr/papers/oopsla97.pdf}, abstract = {SmallEiffel is an Eiffel compiler which uses a fast simple inference mechanism to remove most last binding calls, replacing them by static bindings. Starting from the system's entry point, it compiles only statically living code, which saves compiling and then removing dead code. As the whole system is analyzed at compile time, multiple inheritance and genericity do not cause any overhead. SmallEiffel features a coding scheme which eliminates the need for virtual function tables. Dynamic dispatch is implemented without any array access but uses a simple static binary branch code. We show that this implementation makes it possible to use modern hardware very efficiently. It also allows to inline more calls even when dynamic dispatch is required. Some more dispatch sites are removed after the type inference algorithm has been performed, if the different branches of a dispatch site lead to the same code. The advantage of this approach is that it greatly speeds up execution time and considerably decreases the amount of generated code. } } @misc{zuse.04, title = {The Life and Work of {Konrad Zuse}}, author = {}, howpublished = {\url{http://www.epemag.com/zuse/default.htm#index}}, year = 2004, } @book{gosavi_2014_sop, author = {}, title = {Simulation-Based Optimization: Parametric Optimization Techniques and Reinforcement Learning}, year = {2014}, isbn = {1489974903, 9781489974907}, edition = {2nd}, publisher = {Springer Publishing Company, Incorporated}, } Set x { constraint contains(x, x.shape.label) } Set `A` { shape = Circle { color = computeColor() } -- TODO: doesn't work; can't find identifier -- constraint at(A, 100, 100) } Set `B` { shape = Circle { color = computeColor2() } } Subset x y { constraint contains(y, x) constraint smallerThan(x, y) constraint outsideOf(y.shape.label, x) } @techreport{carter_spurring_2014, author = {Carter, . and Lybbert, . and Mathenge, . and Tjernstrom, Emilia}, copyright = {All rights reserved}, institution = {BASIS Assets and Market Access Innovation Lab}, keywords = {my_pubs, wtp_kenya}, month = {March}, number = {2014-04}, title = {Spurring Technological Innovation and Poverty Reduction? Evaluating the Impact of a New Seed Market Actor in Kenya}, type = {BASIS Brief}, url = {http://basis.ucdavis.edu/wp-content/uploads/2014/04/Carter_WesternSeed1.pdf}, urldate = {2014-09-10}, year = {2014} } 0 @inproceedings{Achananuparp2009a, abstract = {We propose a ranking model to diversify answers of non-factoid questions based on an inverse notion of graph connectivity. By representing a collection of candidate answers as a graph, we posit that novelty, a measure of diversity, is inversely proportional to answer vertices' connectivity. Hence, unlike the typical graph ranking models, which score vertices based on the degree of connectedness, our method assigns a penalty score for a candidate answer if it is strongly connected to other answers. That is, any redundant answers, indicated by a higher inter-sentence similarity, will be ranked lower than those with lower inter-sentence similarity. At the end of the ranking iterations, many redundant answers will be moved toward the bottom on the ranked list. The experimental results show that our method helps diversify answer coverage of non-factoid questions according to F-scores from nugget pyramid evaluation. Copyright 2009 ACM.}, author = { Yang, . and }, booktitle = {Proceedings of the 18th ACM international conference on Information and knowledge management - CIKM '09}, doi = {10.1145/1645953.1646203}, isbn = {9781605585123}, keywords = {Answer ranking,Negative voting,Non-factoid question answering}, title = {Using Negative Voting to Diversify Answers in Non-factoid Question Answering}, year = {2009} } src/pre-textuais/ficha-catalografica.tex1-10 % --- % Inserir a ficha bibliografica % --- % Isto é um exemplo de Ficha Catalográfica, ou ``Dados internacionais de % catalogação-na-publicação''. Você pode utilizar este modelo como referência. % Porém, provavelmente a biblioteca da sua universidade lhe fornecerá um PDF % com a ficha catalográfica definitiva após a defesa do trabalho. Quando estiver % com o documento, salve-o como PDF no diretório do seu projeto e substitua todo % o conteúdo de implementação deste arquivo pelo comando abaixo: % % \begin{fichacatalografica} % \includepdf{fig_ficha_catalografica.pdf} % \end{fichacatalografica} \begin{fichacatalografica} \sffamily \vspace*{\fill} % Posição vertical \begin{center} % Minipage Centralizado \fbox{\begin{minipage}[c][8cm]{13.5cm} % Largura \small \imprimirautor %Sobrenome, Nome do autor \hspace{0.5cm} \imprimirtitulo / \imprimirautor. -- \imprimirlocal, \imprimirdata- \hspace{0.5cm} \pageref{LastPage} p. : il. (algumas color.) ; 30 cm.\\ \hspace{0.5cm} \imprimirorientadorRotulo~\imprimirorientador\\ \hspace{0.5cm} \parbox[t]{\textwidth}{\imprimirtipotrabalho~--~\imprimirinstituicao, \imprimirdata.}\\ \hspace{0.5cm} 1. nodejs. 2. arduino. 2. meteorologia. I. Prof. Dr. . II. Instituto Federal de Educação, Ciência e Tecnologia de São Paulo. III. Tecnologia em análise e desenvolvimento de sistemas. IV. Solus, um software para monitoramento de painéis solares \end{minipage}} \end{center} \end{fichacatalografica} % --- 10-100 \documentclass[compress]{beamer} \usetheme{Frankfurt} \usepackage{lipsum} % for dummy text only % allowframebreaks numbering in the title \newcounter{cont} \makeatletter \setbeamertemplate{frametitle continuation}{% \setcounter{cont}{\beamer@endpageofframe}% \addtocounter{cont}{1}% \addtocounter{cont}{-\beamer@startpageofframe}% (\insertcontinuationcount/\arabic{cont})% } \makeatother % for the title page \title[Header, Navigation symbols and Footer]{Beamer Application Notes: Header, Navigation symbols and Footer} \author[]{ %\resizebox{\textwidth}{!}{Spark \& Shine} Spark \& Shine } \institute{ http://sparkandshine.net\\ } \date[Nov. 16]{Nov. 16\textsuperscript{th}, 2016} \thispagestyle{empty} % header \makeatletter \setbeamertemplate{headline}{% \leavevmode% \hbox{% \begin{beamercolorbox}[wd=\paperwidth,ht=2.5ex,dp=1.125ex]{palette quaternary}% % Flush the bar to the left \insertsectionnavigationhorizontal{\paperwidth}{}{\hskip0pt plus1filll} % Flush the bar to the left %\insertsectionnavigationhorizontal{\paperwidth}{\hskip0pt plus1filll}{} % Flush the bar to the center %\insertsectionnavigationhorizontal{\paperwidth}{\hskip0pt plus1filll}{\hskip0pt plus1filll} \insertsectionnavigationhorizontal{\paperwidth}{left}{right} \end{beamercolorbox}% } } \makeatother % footer \makeatletter \setbeamertemplate{footline} { \leavevmode% \hbox{% \begin{beamercolorbox}[wd=.15\paperwidth,ht=2.25ex,dp=1ex,center]{institute in head/foot}% \usebeamerfont{title in head/foot}% \insertshortauthor \end{beamercolorbox}% \begin{beamercolorbox}[wd=.6\paperwidth,ht=2.25ex,dp=1ex,center]{institute in head/foot}% \usebeamerfont{title in head/foot}% \insertshorttitle \end{beamercolorbox}% \begin{beamercolorbox}[wd=.15\paperwidth,ht=2.25ex,dp=1ex,center]{institute in head/foot}% \usebeamerfont{title in head/foot}% \insertshortdate \end{beamercolorbox}% \begin{beamercolorbox}[wd=.1\paperwidth,ht=2.25ex,dp=1ex,right]{institute in head/foot}% \usebeamerfont{title in head/foot} \insertframenumber{} / \inserttotalframenumber\hspace*{2ex} \end{beamercolorbox}}% } \makeatother % Get rid of % header navigation bar %\setbeamertemplate{headline}{} % bottom navigation symbols %\setbeamertemplate{navigation symbols}{} % footer %\setbeamertemplate{footline}{} % begin document \begin{document} \begin{frame}[noframenumbering]%[shrink=20] \titlepage \end{frame} \section{One Day At A Time} \subsection{Lyrics} \begin{frame}[allowframebreaks]{One Day At A Time, Cristy Lane} \begin{center} I'm only human, I'm just a woman. \\ Help me believe in what I could be \\ And all that I am. \\ Show me the stairway, I have to climb. \\ Lord for my sake, teach me to take \\ One day at a time. \\ Chorus: \\ One day at a time sweet Jesus\\ That's all I'm asking from you. \\ Just give me the strength \\ To do everyday what I have to do. \\ Yesterday's gone sweet Jesus \\ And tomorrow may never be mine. \\ Lord help me today, show me the way \\ One day at a time. \\ Do you remember, when you walked among men? \\ Well Jesus you know if you're looking below \\ It's worse now, than then. \\ Cheating and stealing, violence and crime \\ So for my sake, teach me to take \\ One day at a time. \\ \end{center} \end{frame} \section{Take Me to Church} \subsection{Lyrics} \begin{frame}[allowframebreaks]{Take Me to Church} \begin{center} My lover's got humour \\ She's the giggle at a funeral \\ Knows everybody's disapproval \\ I should've worshipped her sooner \\ If the Heavens ever did speak \\ She is the last true mouthpiece \\ Every Sunday's getting more bleak \\ A fresh poison each week \\ 'We were born sick, ' you heard them say it \\ My church offers no absolutes \\ She tells me 'worship in the bedroom' \\ The only heaven I'll be sent to \\ Is when I'm alone with you \\ I was born sick, but I love it \\ Command me to be well \\ Amen. Amen. Amen \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \\ If I'm a pagan of the good times \\ My lover's the sunlight \\ To keep the Goddess on my side \\ She demands a sacrifice \\ To drain the whole sea \\ Get something shiny \\ Something meaty for the main course \\ That's a fine looking high horse \\ What you got in the stable? \\ We've a lot of starving faithful \\ That looks tasty \\ That looks plenty \\ This is hungry work \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \\ No masters or kings when the ritual begins \\ There is no sweeter innocence than our gentle sin \\ In the madness and soil of that sad earthly scene \\ Only then I am human \\ Only then I am clean \\ Amen. Amen. Amen \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \\ Take me to church \\ I'll worship like a dog at the shrine of your lies \\ I'll tell you my sins and you can sharpen your knife \\ Offer me that deathless death \\ Good God, let me give you my life \end{center} \end{frame} \end{document} 100-1000 % 解决自动换行问题 https://github.com/annProg/PanBook/issues/26#issuecomment-500458739 \definecolor{shaderulecolor}{rgb}{0.66,0.66,0.66} \makeatletter \@ifpackageloaded{float}{}{\usepackage{float}} \@ifpackageloaded{caption}{}{\usepackage{caption}} \floatstyle{plaintop} \@ifundefined{c@chapter}{\newfloat{codelisting}{h}{lop}}{\newfloat{codelisting}{h}{lop}[chapter]} \captionsetup[plaintop]{skip=0pt} \@ifundefinedcolor{shadecolor}{\definecolor{shadecolor}{RGB}{246,248,250}} \makeatother %\usepackage{tcolorbox} %\tcbuselibrary{most} \usepackage[framemethod=tikz]{mdframed} \usetikzlibrary{shadows} \ifcsmacro{Shaded}{ \let\endShaded\undefined% %\renewenvironment{Shaded}{\begin{tcolorbox}[shadow={0.8mm}{-0.8mm}{0mm}{red!20!green!20!blue!20},enhanced,boxrule=0.5pt,boxsep=-1pt,breakable,colback=shadecolor,colframe=shaderulecolor,arc=0.1mm]}{\end{tcolorbox}} \renewenvironment{Shaded}{\begin{mdframed}[ linecolor=shaderulecolor, backgroundcolor=shadecolor, roundcorner=2pt, shadow=true, shadowsize=4pt, ]}{\end{mdframed}} }{} \usepackage{fvextra} \DefineVerbatimEnvironment{Highlighting}{Verbatim}{breaklines,breakanywhere,fontsize=\small,commandchars=\\\{\}} \documentclass[25pt, a0paper, landscape]{tikzposter} % Packages required for tikzposter class \usepackage{tikz, calc, ifthen, ae, xstring, etoolbox, xkeyval} \usepackage{amsmath} \usepackage{hyperref} \usepackage{graphicx} \usepackage{biblatex} \addbibresource{draft.bib} \definelayouttheme{Stankus}{ \usecolorstyle[colorPalette=GreenGrayViolet]{Australia} \usebackgroundstyle{Default} \usetitlestyle{Empty} \useblockstyle{Basic} \useinnerblockstyle{Default} \usenotestyle{Default} } \usetheme{Stankus} % turns off the comment on how the poster was created in the lower right corner of the poster \tikzposterlatexaffectionproofoff % How can we remove the white bar at the top of the poster? % \titlegraphic{Logo} \title{Recognizing Overlapping Handwritten Digits Using Neural Networks} \institute{Frost Research 2018} \author{Professor \textsuperscript{*}, \textsuperscript{*}, \textsuperscript{*}, \textsuperscript{*}, and \textsuperscript{*}} \begin{document} \maketitle \begin{columns} \column{0.35} \block{Our Goal }{ We wanted to use a machine learning concept called a neural network to learn how to identify handwritten overlapping and non overlapping digits to a high degree of accuracy. % image of a '4', a nonoverlapping '42', and a overlapping 58 \begin{center} \includegraphics[scale=6.0]{4.png} \includegraphics[scale=6.0]{42.png} \includegraphics[scale=6.0]{58.png} \end{center} } \block{What is a neural network made of? }{ A neuron is a real-valued function which takes as input a vector $\mathbf{x}$, a vector of weights $\mathbf{w}$ and a bias $b$ which outputs $\sigma(\mathbf{x}\bullet\mathbf{w}+b)$. The function $\sigma$ is called an activation function whose output is usually within $[-1,1]$ or $[0,1]$. A commonly used activation function with a range of $[0,1]$ is the sigmoid function: \[ \sigma(z) = \frac{1}{1+e^{-z}} \, . \] A densely-connected layer of neurons is a function which encapsulates a finite number of neurons, where each neuron in the layer is connected to every neuron in the next layer. % neural network image \begin{center} \input{simplenn.input} \end{center} A layer is a vector-valued function which takes as input a vector $\mathbf{x}$, a set of weight vectors $\mathbf{w_1},\ldots,\mathbf{w_m}$ and a set of biases $b_1,\ldots,b_m$ which outputs \[ \begin{bmatrix} \sigma(\mathbf{x}\bullet\mathbf{w_1}+b_1)\\ \vdots\\ \sigma(\mathbf{x}\bullet\mathbf{w_m}+b_m)\\ \end{bmatrix} \] } \block{Cost Function for Classifying Single Digits }{ \begin{itemize} \item A neural network used to classify single digit images has an output vector $\mathbf{\hat{y}}_0$ with 10 entries. Each entry is a value between 0 and 1 and can be interpreted as a probability that the given image is of a particular digit. The 1st entry corresponds to a '0', the 2nd entry to a '1' and the 10th entry to a '9'. \bigskip \item For example, we might have $\mathbf{\hat{y}}_0 = [ 0 \ \ 0 \ \ 0 \ \ 0.2 \ \ 0.5 \ \ 0.1 \ \ 0.1 \ \ 0 \ \ 0 \ \ 0 \ \ 0]^T$. \bigskip \item Suppose the weights and biases of our neural network are $\mathbf{w}_0$ and $\mathbf{b}_0$ and $\mathbf{x}_0$ is an image of a `4'. The label of $\mathbf{x}_0$ would be $\mathbf{y}_0 = [ 0 \ \ 0 \ \ 0 \ \ 0 \ \ 1 \ \ 0 \ \ 0 \ \ 0 \ \ 0 \ \ 0 \ \ 0]^T$. \bigskip \item The cost function $C$ describes how close our prediction is to the label of the image and is defined by \( C_{\mathbf{x}_0} (\mathbf{w}_0,\mathbf{b}_0) = \|\mathbf{y}_0-\mathbf{\hat{y}}_0\| \). \end{itemize} } \column{0.35} \block{How does a network learn? }{ \begin{itemize} \item For a fixed $\mathbf{x}$, the graph of the function $C_{\mathbf{x}}$ is a surface \item To train a neural network we want to go ``downhill'' along that surface by following the negative of the gradient of $C_{\mathbf x}$ \item We want to find $\mathbf{w}_0$ and $\mathbf{b}_0$ so that $C_{\mathbf{x}}(\mathbf{w}_0,\mathbf{b}_0)$ is small for each training vector $\mathbf{x}_0$ \item The neural network ``has learned'' if $C_{\mathbf{x}}(\mathbf{w}_0,\mathbf{b}_0)$ is also small for each test vector $\mathbf{x}$ \end{itemize} } \block{Training a Neural Network }{ \begin{enumerate} \item Convert black and white images to matrices of $0$'s and $1$'s, then flatten them into vectors \item Separate theses flattened vectors into training, validataion, and testing data \item Adjust the weights and biases as the training data is fed to the the neural network \item Validate those adjustments by testing the network on the validation data \item Train on several epochs, completing steps 3 and 4 is refered to as ``1 epoch'' \item Feed in the test data to see how well the training generalizes to unseen data. \item Save the neural network (the adjusted weights and biases) \item Now we can use the trained network to classify future data \end{enumerate} } \block{Image Recognition Solutions }{ % The capsule network will predict the existence of a face if and % only if the mouth prediction of the face, the eye prediction of the face, % and the rest of the features all have the same prediction for the % location and orientation of the entire face. % The capsule network will predict the location of the mouth and then % predict the location of the face based on the location and orientation % of the mouth Before 2017, the most common type of neural network used in image recognition was the convolutional neural network (cnn). A cnn classifies images by determining the existence of \textbf{certain features} in an image. It does not consider the relative positions and orientations of these features. A cnn might recognize a nose, an eye, or a mouth, but classify both of the following as faces. \begin{center} \cite{pechyonkin_2017}\includegraphics[scale=0.5]{cnn_problem.png} \end{center} A capsule network \cite{NIPS2017_6975} classifies an image as a face if and only if positions and orientations of the the features relative to the location of the face are correct. In the picture below on the right, recognizing the mouth would lead the network to predict that the rest of the face would resemble the highest of the three faces shown and recognizing the eye would lead the network to predict that the rest of the face would resemble the lowest of the three faces shown. Since the face predictions do not agree, the image is not classified as a face. \begin{center} \cite{bourdakos_2018}\includegraphics[scale=0.3]{capsule_network_solution.png} \end{center} } \column{0.3} \block{Our Solution }{ {\Large\underline{\textbf{Our Data:}}} \begin{itemize} \item MNIST is a collection of single digit handwritten numbers. \item With Python code and MNIST we created a dataset of single digit and overlapping double digit images. \end{itemize} {\Large\underline{\textbf{Our Neural Networks:}}} \begin{itemize} \item We used TensorFlow an open source machine learning framework created by Google to code our neural networks. \item We created and trained a densely connected neural network called choice\_net to classify each data sample in our randomly shuffled dataset as a single digit or double digit image \item The output of choice\_net is fed to one of 2 capsule networks named single\_caps and double\_caps which were trained on single digit and double digit images respectively. \end{itemize} } \block{Results }{ All of our neural networks were trained on 2 epochs of 60,000 training samples and 5,000 validation samples. Our testing dataset contained 10,000 samples. \begin{itemize} \item choice\_net achieved a test accuracy of 99.9+\% \item The single\_digit\_caps achieved a test accuracy of 99.1987\% \item The double\_digit\_caps achieved a test accuracy of 98.8381\% \end{itemize} The code is publically available at \url{https://github.com/mstankus/Frost-Neural-Network-Research-2018}. } \block{References }{ \printbibliography[heading=none] (This is a partial list containing the resources used to create the poster) } \block{Acknowledgments }{ \textsuperscript{*} Frost Research Fellows, funded by the Bill and Linda Frost Fund, Cal Poly, San Luis Obispo } \end{columns} \end{document} \unnumberedchapter{Glossary} \chapter*{Glossary} Please refer to \url{https://groups.oist.jp/grad/academic-program-policies} for specifications. Here is an example: % Break up this table into several ones if it takes up more than one page \begin{center} \begin{longtable}{r p{0.58 \textwidth}} Dipole Blockade & Phenomenon in which the simultaneous excitation of two atoms is inhibited by their dipolar interaction. \\ Cavity Induced Transparency & Phenomenon in which a cavity containing two atoms excited with light at a frequency halfway between the atomic frequencies contains the number of photons an empty cavity would contain. \\ \end{longtable} \end{center} silvaCarlosE/cardiomon \subsection{Criar conta} \begin{center} \begin{tabular}{ |p{7cm}|p{7cm}| } \hline \textbf {Nome do caso de uso} & Criar conta\\ \hline \textbf{Descrição geral} & Caso de uso para permitir cadastro de novos usuário \\ \hline \textbf{Ator Principal} & Usuário comum \\ \hline \textbf{Atores Secundários} & Nenhum \\ \hline \textbf{Resumo} & Caso de uso responsável por inserir novos responsáveis na aplicação. \\ \hline \textbf{Pré-Condições} & Nenhuma \\ \hline %-----------------------------------FLUXO DE EVENTOS \multicolumn{2}{|c|}{\textbf{Fluxo pricipal de eventos} } \\ \hline \textbf{Ações do ator} & \textbf{Ações do sistema} \\ \hline \multicolumn{2}{|c|}{\textbf{Fluxo de envio cadastro} } \\ \hline Abrir a aplicação e clicar cadastrar-se & Apresentar tela de cadastro de usuário \\ \hline Preencher com dados & Verificar se usuário existe no banco de dados\\ \hline & Verificar veracidade de e-mail\\ \hline & Salvar dados do novo usuário no banco \\ \hline & Redirecionar usuário para a tela de boas vindas \\ \hline \textbf{Pós condições} & Nenhuma \\ \hline \textbf{Observações} & Os dados inseridos necessitam ser verídicos, caso contrário será exibida uma tela de erro explicando como contornar.\\ \hline \end{tabular} \end{center}qwtb/qwtb1-10 \hypertarget{adctest_8m}{\section{adctest\-\_\-v43/adctest.m File Reference} \label{adctest_8m}\index{adctest\-\_\-v43/adctest.\-m@{adctest\-\_\-v43/adctest.\-m}} } \subsection*{Functions} \begin{DoxyCompactItemize} \item function \hyperlink{adctest_8m_ac924d2617db837cca185eca7235ba4dc}{adctest} () \item function \hyperlink{adctest_8m_a9749a59d1cd2fd5165133b4cbc504239}{Help\-Getting\-Started\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a248ad4d717ff781209732b831d7e1559}{Help\-Users\-Manual\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_abec88860dd6825301dd63e4eeb6056b0}{Help\-About\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a87bbe818617b124d4acf56e85b40c71b}{Import\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_aba3d245a7e26ae49629f5c5210fc6a0a}{Export\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a045c7449c039542e1893641f786c6dff}{Load\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_af6719bfe739f333dc9dd12b79c03b5e7}{Save\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a51bbd2db9dcde551543562202d3124ee}{Delete\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_ada8b7d37c0b844e6258ff838deeb5a2e}{New\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a7adc1a3fe0a3779e742c5597d4626c3f}{Edit\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a43ae11c0aba337455a05e53d655dd124}{Quit\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a3058413c430ab5f71262ee6ba5c0b76e}{Classify\-And\-Process\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a47e144482f670c0c42572c5680955c28}{Process\-L\-S4p\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_a5a4d3fbe44cbdda045d6ff71c1ffe0a6}{Process\-M\-L\-Fit\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_ab906051457cbcc82a27babe199023dde}{Process\-Histogram\-Test\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_acb6914975807c311b25fea68917cef02}{Process\-F\-F\-T\-Test\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_ac39777b38d9aa1636c90c4c136d6f0c3}{Descriptor\-Selector\-\_\-callback} (in source, in eventdata) \item function \hyperlink{adctest_8m_ae84478ad2b78f0751c55bc59c1813618}{Update\-Display} () \end{DoxyCompactItemize} \subsection{Function Documentation} \hypertarget{adctest_8m_ac924d2617db837cca185eca7235ba4dc}{\index{adctest.\-m@{adctest.\-m}!adctest@{adctest}} \index{adctest@{adctest}!adctest.m@{adctest.\-m}} \subsubsection[{adctest}]{\setlength{\rightskip}{0pt plus 5cm}function adctest ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )}}\label{adctest_8m_ac924d2617db837cca185eca7235ba4dc} \hypertarget{adctest_8m_a3058413c430ab5f71262ee6ba5c0b76e}{\index{adctest.\-m@{adctest.\-m}!Classify\-And\-Process\-\_\-callback@{Classify\-And\-Process\-\_\-callback}} \index{Classify\-And\-Process\-\_\-callback@{Classify\-And\-Process\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Classify\-And\-Process\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Classify\-And\-Process\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a3058413c430ab5f71262ee6ba5c0b76e} \hypertarget{adctest_8m_a51bbd2db9dcde551543562202d3124ee}{\index{adctest.\-m@{adctest.\-m}!Delete\-\_\-callback@{Delete\-\_\-callback}} \index{Delete\-\_\-callback@{Delete\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Delete\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Delete\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a51bbd2db9dcde551543562202d3124ee} \hypertarget{adctest_8m_ac39777b38d9aa1636c90c4c136d6f0c3}{\index{adctest.\-m@{adctest.\-m}!Descriptor\-Selector\-\_\-callback@{Descriptor\-Selector\-\_\-callback}} \index{Descriptor\-Selector\-\_\-callback@{Descriptor\-Selector\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Descriptor\-Selector\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Descriptor\-Selector\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_ac39777b38d9aa1636c90c4c136d6f0c3} \hypertarget{adctest_8m_a7adc1a3fe0a3779e742c5597d4626c3f}{\index{adctest.\-m@{adctest.\-m}!Edit\-\_\-callback@{Edit\-\_\-callback}} \index{Edit\-\_\-callback@{Edit\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Edit\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Edit\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a7adc1a3fe0a3779e742c5597d4626c3f} \hypertarget{adctest_8m_aba3d245a7e26ae49629f5c5210fc6a0a}{\index{adctest.\-m@{adctest.\-m}!Export\-\_\-callback@{Export\-\_\-callback}} \index{Export\-\_\-callback@{Export\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Export\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Export\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_aba3d245a7e26ae49629f5c5210fc6a0a} \hypertarget{adctest_8m_abec88860dd6825301dd63e4eeb6056b0}{\index{adctest.\-m@{adctest.\-m}!Help\-About\-\_\-callback@{Help\-About\-\_\-callback}} \index{Help\-About\-\_\-callback@{Help\-About\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Help\-About\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Help\-About\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_abec88860dd6825301dd63e4eeb6056b0} \hypertarget{adctest_8m_a9749a59d1cd2fd5165133b4cbc504239}{\index{adctest.\-m@{adctest.\-m}!Help\-Getting\-Started\-\_\-callback@{Help\-Getting\-Started\-\_\-callback}} \index{Help\-Getting\-Started\-\_\-callback@{Help\-Getting\-Started\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Help\-Getting\-Started\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Help\-Getting\-Started\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a9749a59d1cd2fd5165133b4cbc504239} \hypertarget{adctest_8m_a248ad4d717ff781209732b831d7e1559}{\index{adctest.\-m@{adctest.\-m}!Help\-Users\-Manual\-\_\-callback@{Help\-Users\-Manual\-\_\-callback}} \index{Help\-Users\-Manual\-\_\-callback@{Help\-Users\-Manual\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Help\-Users\-Manual\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Help\-Users\-Manual\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a248ad4d717ff781209732b831d7e1559} \hypertarget{adctest_8m_a87bbe818617b124d4acf56e85b40c71b}{\index{adctest.\-m@{adctest.\-m}!Import\-\_\-callback@{Import\-\_\-callback}} \index{Import\-\_\-callback@{Import\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Import\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Import\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a87bbe818617b124d4acf56e85b40c71b} \hypertarget{adctest_8m_a045c7449c039542e1893641f786c6dff}{\index{adctest.\-m@{adctest.\-m}!Load\-\_\-callback@{Load\-\_\-callback}} \index{Load\-\_\-callback@{Load\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Load\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Load\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a045c7449c039542e1893641f786c6dff} \hypertarget{adctest_8m_ada8b7d37c0b844e6258ff838deeb5a2e}{\index{adctest.\-m@{adctest.\-m}!New\-\_\-callback@{New\-\_\-callback}} \index{New\-\_\-callback@{New\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{New\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function New\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_ada8b7d37c0b844e6258ff838deeb5a2e} \hypertarget{adctest_8m_acb6914975807c311b25fea68917cef02}{\index{adctest.\-m@{adctest.\-m}!Process\-F\-F\-T\-Test\-\_\-callback@{Process\-F\-F\-T\-Test\-\_\-callback}} \index{Process\-F\-F\-T\-Test\-\_\-callback@{Process\-F\-F\-T\-Test\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Process\-F\-F\-T\-Test\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Process\-F\-F\-T\-Test\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_acb6914975807c311b25fea68917cef02} \hypertarget{adctest_8m_ab906051457cbcc82a27babe199023dde}{\index{adctest.\-m@{adctest.\-m}!Process\-Histogram\-Test\-\_\-callback@{Process\-Histogram\-Test\-\_\-callback}} \index{Process\-Histogram\-Test\-\_\-callback@{Process\-Histogram\-Test\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Process\-Histogram\-Test\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Process\-Histogram\-Test\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_ab906051457cbcc82a27babe199023dde} \hypertarget{adctest_8m_a47e144482f670c0c42572c5680955c28}{\index{adctest.\-m@{adctest.\-m}!Process\-L\-S4p\-\_\-callback@{Process\-L\-S4p\-\_\-callback}} \index{Process\-L\-S4p\-\_\-callback@{Process\-L\-S4p\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Process\-L\-S4p\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Process\-L\-S4p\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a47e144482f670c0c42572c5680955c28} \hypertarget{adctest_8m_a5a4d3fbe44cbdda045d6ff71c1ffe0a6}{\index{adctest.\-m@{adctest.\-m}!Process\-M\-L\-Fit\-\_\-callback@{Process\-M\-L\-Fit\-\_\-callback}} \index{Process\-M\-L\-Fit\-\_\-callback@{Process\-M\-L\-Fit\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Process\-M\-L\-Fit\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Process\-M\-L\-Fit\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a5a4d3fbe44cbdda045d6ff71c1ffe0a6} \hypertarget{adctest_8m_a43ae11c0aba337455a05e53d655dd124}{\index{adctest.\-m@{adctest.\-m}!Quit\-\_\-callback@{Quit\-\_\-callback}} \index{Quit\-\_\-callback@{Quit\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Quit\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Quit\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_a43ae11c0aba337455a05e53d655dd124} \hypertarget{adctest_8m_af6719bfe739f333dc9dd12b79c03b5e7}{\index{adctest.\-m@{adctest.\-m}!Save\-\_\-callback@{Save\-\_\-callback}} \index{Save\-\_\-callback@{Save\-\_\-callback}!adctest.m@{adctest.\-m}} \subsubsection[{Save\-\_\-callback}]{\setlength{\rightskip}{0pt plus 5cm}function Save\-\_\-callback ( \begin{DoxyParamCaption} \item[{in}]{source, } \item[{in}]{eventdata} \end{DoxyParamCaption} )}}\label{adctest_8m_af6719bfe739f333dc9dd12b79c03b5e7} \hypertarget{adctest_8m_ae84478ad2b78f0751c55bc59c1813618}{\index{adctest.\-m@{adctest.\-m}!Update\-Display@{Update\-Display}} \index{Update\-Display@{Update\-Display}!adctest.m@{adctest.\-m}} \subsubsection[{Update\-Display}]{\setlength{\rightskip}{0pt plus 5cm}function Update\-Display ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )}}\label{adctest_8m_ae84478ad2b78f0751c55bc59c1813618} %%% Работа с русским языком \usepackage{cmap} % поиск в PDF \usepackage{mathtext} % русские буквы в формулах \usepackage[T2A]{fontenc} % кодировка \usepackage[utf8]{inputenc} % кодировка исходного текста \usepackage[russian]{babel} % локализация и переносы %%% Пакеты для работы с математикой \usepackage{amsmath,amsfonts,amssymb,amsthm,mathtools} \usepackage{icomma} %% Номера формул %\mathtoolsset{showonlyrefs=true} % Показывать номера только у тех формул, на которые есть \eqref{} в тексте. %\usepackage{leqno} % Немуреация формул слева %% Шрифты \usepackage{euscript} % Шрифт Евклид \usepackage{mathrsfs} % Красивый матшрифт %% Поля (геометрия страницы) \usepackage[left=3cm,right=2cm,top=2cm,bottom=2cm,bindingoffset=0cm]{geometry} %% Русские списки \usepackage{enumitem} \makeatletter \AddEnumerateCounter{\asbuk}{\russian@alph}{щ} \makeatother %%% Работа с картинками \usepackage{caption} \captionsetup{justification=centering} % центрирование подписей к картинкам \usepackage{graphicx} % Для вставки рисунков \graphicspath{{images/}{images2/}} % папки с картинками \setlength\fboxsep{3pt} % Отступ рамки \fbox{} от рисунка \setlength\fboxrule{1pt} % Толщина линий рамки \fbox{} \usepackage{wrapfig} % Обтекание рисунков и таблиц текстом %%% Работа с таблицами \usepackage{array,tabularx,tabulary,booktabs} % Дополнительная работа с таблицами \usepackage{longtable} % Длинные таблицы \usepackage{multirow} % Слияние строк в таблице %% Красная строка \setlength{\parindent}{2em} %% Интервалы \linespread{1} \usepackage{multirow} %% TikZ \usepackage{tikz} \usetikzlibrary{graphs,graphs.standard} %% Верхний колонтитул \usepackage{fancyhdr} \pagestyle{fancy} %% Перенос знаков в формулах (по Львовскому) \newcommand*{\hm}[1]{#1\nobreak\discretionary{}{\hbox{$\mathsurround=0pt #1$}}{}} %% дополнения \usepackage{float} % Добавляет возможность работы с командой [H] которая улучшает расположение на странице \usepackage{gensymb} % Красивые градусы \usepackage{caption} % Пакет для подписей к рисункам, в частности, для работы caption* % подключаем hyperref (для ссылок внутри pdf) \usepackage[unicode, pdftex]{hyperref}nicolair/math-exos \begin{tiny}(Csc16)\end{tiny} Formons l'encadrement de définition de la partie entière \[ kx - 1 < \lfloor kx \rfloor \leq kx \] et sommons de $1$ à $n$. \begin{multline*} \left( \sum_{k=1}^{n} k\right) x - n < \sum_{k=1}^{n} \lfloor kx \rfloor \leq \left( \sum_{k=1}^{n} k\right) x \\ \Rightarrow \frac{n+1}{n} \frac{x}{2} - \frac{1}{n} < \frac{1}{n^2}\,\sum_{k=1}^{n} \lfloor kx \rfloor \leq \frac{n+1}{n} \frac{x}{2} \end{multline*} car $\sum_{k=1}^{n} k = \frac{n(n+1)}{2}$. On conclut par le théorème d'encadrement car $\left( \frac{n+1}{n}\right)_{n\in \N}\rightarrow 1$. \begin{figure}[h] \centering \includegraphics[scale=0.4]{traffic_profile/images/site_profile.png} \caption[CPU-Site Profile]{Plot of EnergyPlus CPU Profiles with the Network Coefficients .} \label{cpu_site} \end{figure}%% %%================================================================= %% %% %% %% 北航学位论文模板使用样例 %% %% 请将以下文件与此LaTeX文件放在同一目录中. %% %%----------- %% %% buaa.cls : LaTeX宏模板文件 %% %% bst/GBT7714-2005.bst : 国标参考文献BibTeX样式文件2005(https://github.com/Haixing-Hu/GBT7714-2005-BibTeX-Style) %% %% bst/GBT7714-2015.bst : 国标参考文献BibTeX样式文件2015(https://github.com/zepinglee/gbt7714-bibtex-style) %% %% pic/logo-buaa.eps : 论文封皮北航字样 %% %% pic/head-doctor.eps : 论文封皮学术博士学位论文标题(华文行楷字体替代解决方案) %% %% pic/head-prodoctor.eps : 论文封皮专业博士学位论文标题(华文行楷字体替代解决方案) %% %% pic/head-master.eps : 论文封皮学术硕士学位论文标题(华文行楷字体替代解决方案) %% %% pic/head-professional.eps : 论文封皮专业硕士学位论文标题(华文行楷字体替代解决方案) %% %% tex/*.tex : 本模板样例中的独立章节 %% %%----------- %% %% 请统一使用UTF-8编码. %% %%================================================================= %================================================================= % buaa基于ctexbook模板 % 论文样式参考自《研究生手册--二〇二〇年七月》 %====================== % 模板导入: % \documentclass[thesis,permission,printtype,ostype,]{buaa} %====================== % 模板选项: %====================== % I.论文类型(thesis) %-------------------- % a.学术硕士论文（master）[缺省值] % b.专业硕士论文（professional） % c.学术博士论文（doctor） % d.专业博士论文（prodoctor） %-------------------- % II.密级(permission) %-------------------- % a.公开（public）[缺省值] % b.内部（privacy） % c.秘密（secret=secret3） % c.1.秘密3年（secret3） % c.2.秘密5年（secret5） % c.3.秘密10年（secret10） % c.4.秘密永久（secret*） % d.机密（classified=classified5） % d.1.机密3年（classified3） % d.2.机密5年（classified5） % d.3.机密10年（classified10） % d.4.机密永久（classified*） % e.绝密（topsecret=topsecret10） % e.1.绝密3年（topsecret3） % e.2.绝密5年（topsecret5） % e.3.绝密10年（topsecret10） % e.4.绝密永久（topsecret*） %-------------------- % III.打印设置(printtype) %-------------------- % a.单面打印（oneside）[缺省值] % b.双面打印（twoside） %-------------------- % IV.系统类型(ostype) %-------------------- % a.win（oneside）[缺省值] % b.linux (linux) % c.mac (mac) %-------------------- % V.ctexbook设置选项() %-------------------- % ... %====================== % 其他说明: % 1. Mac系统请使用mac选项，并使用XeLaTeX编译。 % 2. 可加入额外ctexbook文档类的选项，其将会被传递给ctexbook。 % 例如：\documentclass[fontset=founder]{buaa} % 3. CTeX在Linux下默认使用Fandol字体，为避免某些生僻字无法显示，在系统已安装方正 % 字体的前提下可通过fontset=founder选项常用方正字体。 %================================================================= \documentclass[master,privacy,twoside,win]{buaa} %================================================================= % 开启/关闭引用编号颜色：参考文献，公式，图，表，算法 等…… \refcolor{on} % 开启: on[默认]; 关闭: off; % 摘要和正文从右侧页开始 \beginright{on} % 开启: on[默认]; 关闭: off; % 空白页留字 \emptypagewords{[ -- This page is a preset empty page -- ]} %================================================================= % buaa模板已内嵌以下LaTeX工具包: %-------------------- % ifthen, etoolbox, titletoc, remreset, % geometry, fancyhdr, setspace, % float, graphicx, subfigure, epstopdf, % array, enumitem, % booktabs, longtable, multirow, caption, % listings, algorithm2e, amsmath, amsthm, % hyperref, pifont, color, soul, % --- % For Win: times % For Lin: newtxtext, newtxmath % For Mac: times, fontspec %-------------------- % 请在此处添加额外工具包>> %================================================================= % buaa模板已内嵌以下LaTeX宏: %-------------------- % \highlight{text} % 黄色高亮 %-------------------- % 请在此处添加自定义宏>> %%================================================================= % 论文题目及副标题-{中文}{英文} \Title{北航硕博士学位论文~\LaTeX{}模板\BUAAThesis{}}{\LaTeX{} Template of Beihang University Thesis \BUAAThesis{}} \Subtitle{版本 \BUAAThesisVer{}}{Version \BUAAThesisVer{}} % 学科大类,默认工学 % \Branch{工学} % 院系,专业及研究方向 \Department{宇航学院} \Major{控制科学与工程} \Feild{模式识别与智能系统} % 导师信息-{中文名}{英文名}{职称} \Tutor{导师姓名}{Tutor}{教授} \Cotutor{副导师姓名}{Cotutor}{高工} % 学生姓名-{中文名}{英文名} \Author{学生姓名}{Student} % 学生学号 \StudentID{ID123456} % 中图分类号 \CLC{TP391.4} % 时间节点-{月}{日}{年} \DateEnroll{09}{01}{2015} \DateGraduate{03}{31}{2018} \DateSubmit{01}{10}{2018} \DateDefence{03}{01}{2018} %%================================================================= % 摘要-{中文}{英文} \Abstract{% 摘要是学位论文内容的简短陈述，应体现论文工作的核心思想。论文摘要应力求语言精炼准确。博士学位论文的中文摘要一般约800$\sim$1200字；硕士学位论文的中文摘要一般约500字。摘要内容应涉及本项科研工作的目的和意义、研究思想和方法、研究成果和结论。博士学位论文必须突出论文的创造性成果，硕士学位论文必须突出论文的新见解。 关键字是为用户查找文献，从文中选取出来揭示全文主体内容的一组词语或术语，应尽量采用词表中的规范词（参考相应的技术术语标准）。关键词一般3$\sim$5个，按词条的外延层次排列（外延大的排在前面）。关键词之间用逗号分开，最后一个关键词后不打标点符号。 为了国际交流的需要，论文必须有英文摘要。英文摘要的内容及关键词应与中文摘要及关键词一致，要符合英语语法，语句通顺，文字流畅。英文和汉语拼音一律为Times New Roman体，字号与中文摘要相同。 }{% What were you doing 500 years ago? Oh, that's right nothing, because you didn't exist yet. In fact, several generations of your family had yet to leave their mark on the world, but one very special shark may already have been swimming in the chilly North Atlantic at that time, and the incredible animal is somehow still alive today. Scientists studying Greenland sharks observed the particularly old specimen just recently, and after studying it they've determined that the creature is approximately 272 to 512 years old. That's an absolutely insane figure, and if its age lands towards the higher end, it makes the animal the oldest observed living vertebrate on the entire planet. Greenland sharks are an incredible species in a number of ways, but most notable is its longevity. The sharks are well over 100 years old before even reaching sexual maturity, and regularly live for centuries. This particularly old specimen, along with 27 others, were analyzed using radiocarbon dating. The reading came back at around 392 years, but potential margin of error means the animal's true age is somewhere between 272 and 512. The shark, which is a female, measures an impressive 18 feet long. That's pretty large, but it might not sound particularly large for an ocean-dwelling creature that lives hundreds of years. That is, until you consider that the Greenland shark only grows around one centimeter per year. With that in mind, 18 feet is actually downright massive. As for how this particular shark species manages to live so incredibly long, scientists attribute a lot of its longevity to its sluggish metabolism, as well as its environment. The frigid waters where the sharks thrive is thought to increase overall lifespan in a variety of ways. Past research has shown that cold environments can help slow aging, and these centuries-old sharks are most certainly benefiting from their chilly surroundings. --- Online news {\it Scientists find incredible shark that may be over 500 years old and still kicking\/}, 12.16.2017. (http://bgr.com/2017/12/14/oldest-shark-greenland-512-years-old/). } % 关键字-{中文}{英文} \Keyword{% 北航，学位论文，博士，硕士，中文，\LaTeX{}模板，\BUAAThesis{} }{% News, BGR, Shark } % 图标目录 \Listfigtab{on} % 启用: on[默认]; 关闭: off; % 缩写定义 按tabular环境或其他列表环境编写 \Abbreviations{ \centering \begin{tabular}{cl} $E$ & 能量 \\ $m$ & 质量 \\ $c$ & 光速 \\ $P$ & 概率 \\ $T$ & 时间 \\ $v$ & 速度 \\ \end{tabular} } \begin{document} %%================================================================= % 标题级别 %-------------------- % \chapter{第一章} % \section{1.1 小节} % \subsection{1.1.1 条} % \subsubsection{1.1.1.1} % \paragraph{1.1.1.1.1} % \subparagraph{1.1.1.1.1.1} %-------------------- %%================================================================= % 绪论 \input{tex/chap_intro} % 说明 \input{tex/chap_instruction} % 示例 \input{tex/chap_sample} % 总结 \input{tex/chap_summary} % 参考文献 % 2015版国标GBT7714-2015 % 2005版国标GBT7714-2005 \Bib{bst/GBT7714-2015}{ref} % 附录 \input{tex/chap_appendix} % 攻读学位期间成果 \input{tex/chap_achievement} % 致谢 \input{tex/chap_acknowledge} % 作者简介 \input{tex/chap_biography} \vspace{5cm} This is \BUAAThesis{}, Happy TeXing! --- from WeiQM. \end{document} content/publication/2019-mc-guire/cite.bib @article{2019-McGuire, author = {McGuire, and Shingledecker, and Willis, and Lee, and Martin-Drumel, Marie-Aline and Blake, and Brogan, and Burkhardt, and Caselli, Paola and Chuang, Ko-Ju and others}, doi = {10.3847/1538-4357/ab3b01}, journal = {The Astrophysical Journal}, pages = {201}, title = {Searches for Interstellar HCCSH and H$_2$CCS}, volume = {883}, year = {2019} } generated/tables/bel.abs-T.tex % latex table generated in R 3.2.3 by xtable 1.8-2 package % Fri Dec 7 21:35:31 2018 \begin{tabular}{rrrrrr} \lsptoprule & Menschen & Tiere & Konkreta & Abstrakta & Summe \\ \midrule mit dër & 478 & 21 & 105 & 196 & 800 \\ ohne dër & 1333 & 78 & 402 & 1106 & 2919 \\ Summe & 1811 & 99 & 507 & 1302 & 3719 \\ \lspbottomrule \end{tabular} kbroman/Teaching_UWStatGen2019 \documentclass[12pt]{article} \usepackage{times} \usepackage{amsmath} \usepackage{hyperref} % revise margins \setlength{\headheight}{0.0in} \setlength{\topmargin}{-0.45in} \setlength{\headsep}{0.0in} \setlength{\textheight}{9.55in} \setlength{\footskip}{0.35in} \setlength{\oddsidemargin}{-0.25in} \setlength{\evensidemargin}{-0.25in} \setlength{\textwidth}{7.0in} \setlength{\parskip}{6pt} \setlength{\parindent}{0pt} \hypersetup{pdfpagemode=UseNone} % don't show bookmarks on initial view \hypersetup{colorlinks, urlcolor={blue}} \begin{document} \thispagestyle{empty} Stat 877: Statistical methods for molecular biology (Spring, 2019)\\ \textbf{Homework \#3}: QTL mapping \textbf{Due 14 March 2019} \bigskip We will consider a set of simulated data from a backcross with 300 individuals, with a single quantitative phenotype. A selective genotyping strategy was used: only the top 46 and bottom 46 individuals, by phenotype, were genotyped. \begin{enumerate} \item Grab the comma-delimited data file at {\footnotesize \tt \verb| | \href{http://www.biostat.wisc.edu/~kbroman/teaching/uwstatgen/hw3.csv}{http://www.biostat.wisc.edu/{\textasciitilde}kbroman/teaching/uwstatgen/hw3.csv} } and place it in your R working directory. Within R, you'll need to install R/qtl via {\footnotesize \verb|install.packages("qtl")|} Then load R/qtl via {\footnotesize \verb|library(qtl)|} Then import the data file via {\footnotesize \verb|hw <- read.cross("csv", file="hw3.csv")|} \item Use each of standard interval mapping (by the EM algorithm) and Haley-Knott regression to map QTL in this cross. Also, use a permutation test to establish significance of identified QTL, and calculate 1.5-LOD support intervals for the locations of inferred QTL. Do two versions of the permutation test: the usual kind plus a stratified permutation test (with individuals stratified by the amount of genotyping, and with permutations performed within these two strata). \bigskip {\footnotesize To perform the stratified permutation test, do something like this: \vspace{-12pt} {\footnotesize \begin{verbatim} nt <- ntyped(hw) strat <- as.numeric(nt > mean(unique(nt))) operm <- scanone(hw, method="hk", n.perm=1000, perm.strat=strat) \end{verbatim} }} \item How do the results change if you omit the individuals that were not genotyped? \bigskip {\footnotesize To drop the non-genotyped, individuals, use code like \vspace{-12pt} {\footnotesize \begin{verbatim} hw_sub <- subset(hw, ind=(ntyped(hw) > 0)) \end{verbatim} }} \item What do you conclude, regarding the behavior of standard interval mapping vs Haley-Knott regression in the presence of selective genotyping, and regarding the use of an unstratified vs stratified permutation test? \end{enumerate} \end{document} \hypertarget{class_image_controller_base}{}\section{Image\+Controller\+Base クラス} \label{class_image_controller_base}\index{Image\+Controller\+Base@{Image\+Controller\+Base}} Image\+Controller\+Base の継承関係図\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=5.000000cm]{class_image_controller_base} \end{center} \end{figure} \subsection*{公開変数類} \begin{DoxyCompactItemize} \item \hypertarget{class_image_controller_base_a48518f64bd926bab64265066a1aa6145}{}{\bfseries \$http\+Status} = 200\label{class_image_controller_base_a48518f64bd926bab64265066a1aa6145} \item \hypertarget{class_image_controller_base_a46898a279a531bb5e6c8b0fe0e15d1cd}{}{\bfseries \$output\+Type} = \char`\"{}\char`\"{}\label{class_image_controller_base_a46898a279a531bb5e6c8b0fe0e15d1cd} \end{DoxyCompactItemize} \subsection*{限定公開メンバ関数} \begin{DoxyCompactItemize} \item \hyperlink{class_image_controller_base_a0eaea96edabf75b7ac4da68173ab4c56}{\+\_\+get\+Image} (\$arg\+File\+Path, \$arg\+Width=N\+U\+L\+L, \$arg\+Height=N\+U\+L\+L, \$arg\+Proportional=N\+U\+L\+L, \$arg\+Memcache\+D\+S\+N=N\+U\+L\+L) \end{DoxyCompactItemize} \subsection*{その他の継承メンバ} \subsection{関数詳解} \hypertarget{class_image_controller_base_a0eaea96edabf75b7ac4da68173ab4c56}{}\index{Image\+Controller\+Base@{Image\+Controller\+Base}!\+\_\+get\+Image@{\+\_\+get\+Image}} \index{\+\_\+get\+Image@{\+\_\+get\+Image}!Image\+Controller\+Base@{Image\+Controller\+Base}} \subsubsection[{\+\_\+get\+Image}]{\setlength{\rightskip}{0pt plus 5cm}Image\+Controller\+Base\+::\+\_\+get\+Image ( \begin{DoxyParamCaption} \item[{}]{\$arg\+File\+Path, } \item[{}]{\$arg\+Width = {\ttfamily NULL}, } \item[{}]{\$arg\+Height = {\ttfamily NULL}, } \item[{}]{\$arg\+Proportional = {\ttfamily NULL}, } \item[{}]{\$arg\+Memcache\+D\+S\+N = {\ttfamily NULL}} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [protected]}}\label{class_image_controller_base_a0eaea96edabf75b7ac4da68173ab4c56} \begin{DoxyParams}[1]{引数} unknown & {\em \$arg\+File\+Path} & \\ \hline string & {\em \$arg\+Width} & \\ \hline string & {\em \$arg\+Height} & \\ \hline string & {\em \$arg\+Proportional} & true\+:サイズ上の縦横比を維持しつつ、縮小する false\+:指定された比率にリサイズ(アスペクト比は変わらない) \\ \hline string & {\em \$arg\+Memcache\+D\+S\+N} & \\ \hline \end{DoxyParams} \begin{DoxyReturn}{戻り値} boolean \end{DoxyReturn} このクラス詳解は次のファイルから抽出されました\+:\begin{DoxyCompactItemize} \item Framework\+Package/class/\+M\+V\+C/Image\+Controller\+Base.\+class.\+php\end{DoxyCompactItemize} \hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result}{}\doxysection{Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result Class Reference} \label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result}\index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsection*{Properties} \begin{DoxyCompactItemize} \item string \mbox{\hyperlink{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_afd01c3ff3735c2210acae7b674804b43}{Mo\+Callback\+Url}}\hspace{0.3cm}{\ttfamily \mbox{[}get, set\mbox{]}} \begin{DoxyCompactList}\small\item\em The current or updated inbound message U\+RI \end{DoxyCompactList}\item string \mbox{\hyperlink{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_abbb6e8ef2de94b23a2a072fab457a530}{Dr\+Callbackurl}}\hspace{0.3cm}{\ttfamily \mbox{[}get, set\mbox{]}} \begin{DoxyCompactList}\small\item\em The current or updated delivery receipt U\+RI \end{DoxyCompactList}\item decimal \mbox{\hyperlink{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_aab43bb25be92ddc511a54cd15119a3bf}{Max\+Outbound\+Request}}\hspace{0.3cm}{\ttfamily \mbox{[}get, set\mbox{]}} \begin{DoxyCompactList}\small\item\em The maximum number of outbound messages per second. \end{DoxyCompactList}\item decimal \mbox{\hyperlink{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a476c23e14708d0948f33cb859f51fb8f}{Max\+Inbound\+Request}}\hspace{0.3cm}{\ttfamily \mbox{[}get, set\mbox{]}} \begin{DoxyCompactList}\small\item\em The maximum number of inbound messages per second. \end{DoxyCompactList}\item decimal \mbox{\hyperlink{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a02f9b0509ec037271d2e8d29c0276564}{Max\+Calls\+Per\+Second}}\hspace{0.3cm}{\ttfamily \mbox{[}get, set\mbox{]}} \begin{DoxyCompactList}\small\item\em The maximum number of A\+PI calls per second. \end{DoxyCompactList}\end{DoxyCompactItemize} \doxysubsection{Property Documentation} \mbox{\Hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_abbb6e8ef2de94b23a2a072fab457a530}\label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_abbb6e8ef2de94b23a2a072fab457a530}} \index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}!DrCallbackurl@{DrCallbackurl}} \index{DrCallbackurl@{DrCallbackurl}!Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsubsection{\texorpdfstring{DrCallbackurl}{DrCallbackurl}} {\footnotesize\ttfamily string Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result.\+Dr\+Callbackurl\hspace{0.3cm}{\ttfamily [get]}, {\ttfamily [set]}} The current or updated delivery receipt U\+RI \mbox{\Hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a02f9b0509ec037271d2e8d29c0276564}\label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a02f9b0509ec037271d2e8d29c0276564}} \index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}!MaxCallsPerSecond@{MaxCallsPerSecond}} \index{MaxCallsPerSecond@{MaxCallsPerSecond}!Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsubsection{\texorpdfstring{MaxCallsPerSecond}{MaxCallsPerSecond}} {\footnotesize\ttfamily decimal Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result.\+Max\+Calls\+Per\+Second\hspace{0.3cm}{\ttfamily [get]}, {\ttfamily [set]}} The maximum number of A\+PI calls per second. \mbox{\Hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a476c23e14708d0948f33cb859f51fb8f}\label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_a476c23e14708d0948f33cb859f51fb8f}} \index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}!MaxInboundRequest@{MaxInboundRequest}} \index{MaxInboundRequest@{MaxInboundRequest}!Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsubsection{\texorpdfstring{MaxInboundRequest}{MaxInboundRequest}} {\footnotesize\ttfamily decimal Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result.\+Max\+Inbound\+Request\hspace{0.3cm}{\ttfamily [get]}, {\ttfamily [set]}} The maximum number of inbound messages per second. \mbox{\Hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_aab43bb25be92ddc511a54cd15119a3bf}\label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_aab43bb25be92ddc511a54cd15119a3bf}} \index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}!MaxOutboundRequest@{MaxOutboundRequest}} \index{MaxOutboundRequest@{MaxOutboundRequest}!Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsubsection{\texorpdfstring{MaxOutboundRequest}{MaxOutboundRequest}} {\footnotesize\ttfamily decimal Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result.\+Max\+Outbound\+Request\hspace{0.3cm}{\ttfamily [get]}, {\ttfamily [set]}} The maximum number of outbound messages per second. \mbox{\Hypertarget{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_afd01c3ff3735c2210acae7b674804b43}\label{class_nexmo_1_1_api_1_1_accounts_1_1_account_settings_result_afd01c3ff3735c2210acae7b674804b43}} \index{Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}!MoCallbackUrl@{MoCallbackUrl}} \index{MoCallbackUrl@{MoCallbackUrl}!Nexmo.Api.Accounts.AccountSettingsResult@{Nexmo.Api.Accounts.AccountSettingsResult}} \doxysubsubsection{\texorpdfstring{MoCallbackUrl}{MoCallbackUrl}} {\footnotesize\ttfamily string Nexmo.\+Api.\+Accounts.\+Account\+Settings\+Result.\+Mo\+Callback\+Url\hspace{0.3cm}{\ttfamily [get]}, {\ttfamily [set]}} The current or updated inbound message U\+RI The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item C\+:/\+Users/slore/\+Documents/projects/slorello89/nexmo-\/dotnet/\+Nexmo.\+Api/\+Accounts/Account\+Settings\+Result.\+cs\end{DoxyCompactItemize} 1-10 %Articles @article{Bouchaud_2004, title={Fluctuations and response in financial markets: the subtle nature of “random” price changes}, volume={4}, ISSN={1469-7696}, url={http://dx.doi.org/10.1080/14697680400000022}, DOI={10.1080/14697680400000022}, number={2}, journal={Quantitative Finance}, publisher={Informa UK Limited}, author={ and and and }, year={2004}, month={Apr}, pages={176–190} } @article{Wang_2016_cross, title={Cross-response in correlated financial markets: individual stocks}, volume={89}, ISSN={1434-6036}, url={http://dx.doi.org/10.1140/epjb/e2016-60818-y}, DOI={10.1140/epjb/e2016-60818-y}, number={4}, journal={The European Physical Journal B}, publisher={Springer Science and Business Media LLC}, author={ and and }, year={2016}, month={Apr} } @article{Wang_2016_avg, author={ and {\"a}fer and }, title={Average cross-responses in correlated financial markets}, journal={The European Physical Journal B}, year={2016}, month={Sep}, day={26}, volume={89}, number={9}, pages={207}, issn={1434-6036}, doi={10.1140/epjb/e2016-70137-0}, url={https://doi.org/10.1140/epjb/e2016-70137-0} } @article{Wang_2018_a, author = { and }{\ss} and }, day = {30}, doi = {10.1140/epjb/e2018-80665-0}, issn = {1434-6036}, journal = {The European Physical Journal B}, month = {Aug}, number = {8}, pages = {191}, title = {{Statistical properties of market collective responses}}, url = {https://doi.org/10.1140/epjb/e2018-80665-0}, volume = {91}, year = {2018} } @article{Wang_2018_b, author = { and }{\ss} and }, title={Grasping asymmetric information in price impacts}, journal={The European Physical Journal B}, year={2018}, month={Nov}, day={01}, volume={91}, number={11}, pages={266}, issn={1434-6036}, doi={10.1140/epjb/e2018-80599-5}, url={https://doi.org/10.1140/epjb/e2018-80599-5} } @article{Wang_2018_copulas, doi = {10.1088/1742-5468/aab01c}, url = {https://doi.org/10.1088%2F1742-5468%2Faab01c}, year = 2018, month = {mar}, publisher = {{IOP} Publishing}, volume = {2018}, number = {3}, pages = {033407}, author = { and }, title = {Local fluctuations of the signed traded volumes and the dependencies of demands: a copula analysis}, journal = {Journal of Statistical Mechanics: Theory and Experiment}, } @article{trades_classification, author = { and }, title = {Inferring Trade Direction from Intraday Data}, journal = {The Journal of Finance}, volume = {46}, number = {2}, pages = {733-746}, doi = {10.1111/j.1540-6261.1991.tb02683.x}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1540-6261.1991.tb02683.x}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1540-6261.1991.tb02683.x}, year = {1991} } @article{empirical_facts, author = { and and and }, year = {2009}, month = {09}, pages = {}, title = {Econophysics: Empirical facts and agent-based models}, journal = {arXiv.org, Quantitative Finance Papers} } @article{large_prices_changes, author = { and and and and }, title = {What really causes large price changes?}, journal = {Quantitative Finance}, volume = {4}, number = {4}, pages = {383-397}, year = {2004}, publisher = {Routledge}, doi = {10.1080/14697680400008627}, URL = {https://doi.org/10.1080/14697680400008627}, eprint = {https://doi.org/10.1080/14697680400008627} } @article{subtle_nature, author = {}, title = {The subtle nature of financial random walks}, journal = {Chaos: An Interdisciplinary Journal of Nonlinear Science}, volume = {15}, number = {2}, pages = {026104}, year = {2005}, doi = {10.1063/1.1889265}, URL = {https://doi.org/10.1063/1.1889265}, eprint = {https://doi.org/10.1063/1.1889265} } @article{empirical_properties, author = {}, title = {Empirical properties of asset returns: stylized facts and statistical issues}, journal = {Quantitative Finance}, year = {2001}, volume = {1}, pages = {223--236} } @article{fluctions_market_friction, author = {}, title = {Fluctuations and market friction in financial trading}, journal = {International Journal of Modern Physics C}, volume = {13}, number = {03}, pages = {419-425}, year = {2002}, doi = {10.1142/S012918310200322X}, URL = {https://doi.org/10.1142/S012918310200322X}, eprint = {https://doi.org/10.1142/S012918310200322X} } @article{quant_stock_price_response, title = {Quantifying stock-price response to demand fluctuations}, author = { and and and }, journal = {Phys. Rev. E}, volume = {66}, issue = {2}, pages = {027104}, numpages = {4}, year = {2002}, month = {Aug}, publisher = {American Physical Society}, doi = {10.1103/PhysRevE.66.027104}, url = {https://link.aps.org/doi/10.1103/PhysRevE.66.027104} } @article{theory_market_impact, author = {}, year = {2008}, month = {04}, pages = {}, title = {A Theory for Market Impact: How Order Flow Affects Stock Price}, journal = {arXiv.org, Quantitative Finance Papers} } @article{order_flow_persistent, author={ and and and }, title={{Why is equity order flow so persistent?}}, journal={Journal of Economic Dynamics and Control}, year=2015, volume={51}, number={C}, pages={218-239}, month={}, doi={10.1016/j.jedc.2014.10.00}, url={https://ideas.repec.org/a/eee/dyncon/v51y2015icp218-239.html} } @article{dissecting_cross, title={Dissecting cross-impact on stock markets: an empirical analysis}, volume={2017}, ISSN={1742-5468}, url={http://dx.doi.org/10.1088/1742-5468/aa53f7}, DOI={10.1088/1742-5468/aa53f7}, number={2}, journal={Journal of Statistical Mechanics: Theory and Experiment}, publisher={IOP Publishing}, author={ and and and }, year={2017}, month={Feb}, pages={023406} } @article{spread_changes_affect, title={How spread changes affect the order book: comparing the price responses of order deletions and placements to trades}, author={ and }, journal={The European Physical Journal B}, year={2018}, volume={92}, pages={1-11} } @article{sampling_returns, author = { and }, title = {Sampling Returns for Realized Variance Calculations: Tick Time or Transaction Time?}, journal = {Econometric Reviews}, volume = {27}, number = {1-3}, pages = {230-253}, year = {2008}, publisher = {Taylor & Francis}, doi = {10.1080/07474930701873341}, URL = {https://doi.org/10.1080/07474930701873341}, eprint = {https://doi.org/10.1080/07474930701873341} } @article{asynchrony_effects_corr, title = {Compensating asynchrony effects in the calculation of financial correlations}, journal = {Physica A: Statistical Mechanics and its Applications}, volume = {389}, number = {4}, pages = {767 - 779}, year = {2010}, issn = {0378-4371}, doi = {https://doi.org/10.1016/j.physa.2009.10.033}, url = {http://www.sciencedirect.com/science/article/pii/S0378437109009030}, author = { and and }, } @article{tick_size_impact, title = {Impact of the tick-size on financial returns and correlations}, journal = {Physica A: Statistical Mechanics and its Applications}, volume = {389}, number = {21}, pages = {4828 - 4843}, year = {2010}, issn = {0378-4371}, doi = {https://doi.org/10.1016/j.physa.2010.06.037}, url = {http://www.sciencedirect.com/science/article/pii/S0378437110005662}, author = { and and }, } @article{causes_epps_effect, author = { and and }, title = {Statistical causes for the Epps effect in microstructure noise}, journal = {International Journal of Theoretical and Applied Finance}, volume = {14}, number = {08}, pages = {1231-1246}, year = {2011}, doi = {10.1142/S0219024911006838}, URL = {https://doi.org/10.1142/S0219024911006838}, eprint = {https://doi.org/10.1142/S0219024911006838}, } @article{account_spread, author = { and and }, year = {1997}, month = {01}, pages = {50-60}, title = {Accounting Information and Bid-Ask Spread}, volume = {11}, journal = {Accounting Horizons} } @article{effects_spread, ISSN = {00221082, 15406261}, URL = {http://www.jstor.org/stable/2328600}, author = { and }, journal = {The Journal of Finance}, number = {2}, pages = {479--486}, publisher = {[American Finance Association, Wiley]}, title = {The Effects of Beta, Bid-Ask Spread, Residual Risk, and Size on Stock Returns}, volume = {44}, year = {1989} } @article{limit_ord_spread, title = {Limit orders and the bid–ask spread}, journal = {Journal of Financial Economics}, volume = {53}, number = {2}, pages = {255 - 287}, year = {1999}, issn = {0304-405X}, doi = {https://doi.org/10.1016/S0304-405X(99)00022-7}, url = {http://www.sciencedirect.com/science/article/pii/S0304405X99000227}, author = { and [] and [Van Ness]}, keywords = {Limit order, Bid–ask spread, Specialists}, } @article{components_spread, title = {Estimating the components of the bid/ask spread}, journal = {Journal of Financial Economics}, volume = {21}, number = {1}, pages = {123 - 142}, year = {1988}, issn = {0304-405X}, doi = {https://doi.org/10.1016/0304-405X(88)90034-7}, url = {http://www.sciencedirect.com/science/article/pii/0304405X88900347}, author = { and }, } @article{components_spread_tokyo, title = {The components of the bid–ask spread in a limit-order market: evidence from the Tokyo Stock Exchange}, journal = {Journal of Empirical Finance}, volume = {9}, number = {4}, pages = {399 - 430}, year = {2002}, issn = {0927-5398}, doi = {https://doi.org/10.1016/S0927-5398(02)00003-8}, url = {http://www.sciencedirect.com/science/article/pii/S0927539802000038}, author = { and and and }, keywords = {Bid–ask spread, Adverse selection cost, Order-processing cost, Trade size}, } @article {EMH_lillo, author = { and }, title = {The Long Memory of the Efficient Market}, journal = {Studies in Nonlinear Dynamics \& Econometrics}, year = {2004}, publisher = {De Gruyter}, address = {Berlin, Boston}, volume = {8}, number = {3}, url = {https://www.degruyter.com/view/journals/snde/8/3/article-snde.2004.8.3.1226.xml.xml}, } @article{master_curve, author = { and and }, year = {2003}, month = {02}, pages = {129-30}, title = {Master Curve for Price-Impact Function}, volume = {421}, journal = {Nature}, doi = {10.1038/421129a} } @article{pow_law_dist, author = { and and and }, year = {2003}, month = {06}, pages = {267-70}, title = {A Theory of Power-Law Distributions in Financial Market Fluctuations}, volume = {423}, journal = {Nature}, doi = {10.1038/nature01624} } @article{ori_pow_law, author = {Vasiliki Plerou and and and Parameswaran Gopikrishnan}, title = {On the origin of power-law fluctuations in stock prices}, journal = {Quantitative Finance}, volume = {4}, number = {1}, pages = {11-15}, year = {2004}, publisher = {Routledge}, doi = {10.1088/1469-7688/4/1/C02}, URL = {https://doi.org/10.1088/1469-7688/4/1/C02}, eprint = {https://doi.org/10.1088/1469-7688/4/1/C02} } @article{rand_mat, title = {Random matrix approach to cross correlations in financial data}, author = { and and and Lu\'{\i}, and and }, journal = {Phys. Rev. E}, volume = {65}, issue = {6}, pages = {066126}, numpages = {18}, year = {2002}, month = {Jun}, publisher = {American Physical Society}, doi = {10.1103/PhysRevE.65.066126}, url = {https://link.aps.org/doi/10.1103/PhysRevE.65.066126} } @article{non_stationarity, doi = {10.1209/0295-5075/103/58003}, url = {https://doi.org/10.1209%2F0295-5075%2F103%2F58003}, year = 2013, month = {sep}, publisher = {{IOP} Publishing}, volume = {103}, number = {5}, pages = {58003}, author = { and and and }, title = {Non-stationarity in financial time series: Generic features and tail behavior}, journal = {{EPL} (Europhysics Letters)}, } @article{stat_prop, title={Statistical properties of stock order books: empirical results and models}, volume={2}, ISSN={1469-7696}, url={http://dx.doi.org/10.1088/1469-7688/2/4/301}, DOI={10.1088/1469-7688/2/4/301}, number={4}, journal={Quantitative Finance}, publisher={Informa UK Limited}, author={ and and }, year={2002}, month={Aug}, pages={251–256} } @article{stat_theory, Title = {Statistical theory of the continuous double auction}, Author = { and and and }, Journal = {Quantitative Finance}, Year = {2003}, Number = {6}, Pages = {481--514}, Volume = {3}, Doi = {10.1088/1469-7688/3/6/307}, Eprint = {http://dx.doi.org/10.1088/1469-7688/3/6/307}, Opturl = {http://dx.doi.org/10.1088/1469-7688/3/6/307} } @article{reg_and_irreg, Title = {Regularities and irregularities in order flow data}, Author = { and and }, Journal = {The European Physical Journal B}, Year = {2017}, Number = {218}, Volume = {90}, Doi = {10.1140/epjb/e2017-80087-6}, url = {https://doi.org/10.1140/epjb/e2017-80087-6} } @article{em_stylized_facts, Title = {Emergence of Stylized Facts During the Opening of Stock Markets}, Author = { and and }, Journal = {The European Physical Journal B}, Year = {2018}, url = {http://dx.doi.org/10.2139/ssrn.3303249} } @article{micro_eff, title = {Microstructure effects, bid–ask spreads and volatility in the spot foreign exchange market pre and post-EMU}, journal = {Global Finance Journal}, volume = {17}, number = {1}, pages = {23 - 49}, year = {2006}, issn = {1044-0283}, doi = {https://doi.org/10.1016/j.gfj.2006.06.004}, url = {http://www.sciencedirect.com/science/article/pii/S1044028306000251}, author = { and Owain {} and }, } @article{eur_int_curr, author={ and and }, title={The euro as an international currency: explaining puzzling first evidence from the foreign exchange markets}, journal={Journal of International Money and Finance}, year={2002}, volume={21}, number={3}, pages={351-383}, month={June}, url={https://ideas.repec.org/a/eee/jimfin/v21y2002i3p351-383.html} } @article{eur_change_forex, author={ and and }, title={{How has the euro changed the foreign exchange market?}}, journal={Economic Policy}, year=2002, volume={17}, number={34}, pages={149-192}, month={April}, url={https://ideas.repec.org/a/bla/ecpoli/v17y2002i34p149-192.html} } @article{intraday_forex, author = { and }, year = {2009}, month = {01}, pages = {35-45}, title = {Intraday spot foreign exchange market. Analysis of efficiency, liquidity and volatility}, volume = {6}, journal = {Investment Management and Financial Innovations} } @article{patterns_forex, author = { and and }, year = {2008}, month = {01}, title = {Patterns in high-frequency FX data: Discovery of 12 empirical scaling laws}, volume = {11}, journal = {arXiv.org, Quantitative Finance Papers}, doi = {10.1080/14697688.2010.481632} } @article{electronic_forex, title = {The electronic trading systems and bid-ask spreads in the foreign exchange market}, journal = {Journal of International Financial Markets, Institutions and Money}, volume = {20}, number = {4}, pages = {323 - 345}, year = {2010}, issn = {1042-4431}, doi = {https://doi.org/10.1016/j.intfin.2010.04.002}, url = {http://www.sciencedirect.com/science/article/pii/S1042443110000247}, author = { and }, } @article{info_forex, author = { and and and }, title = {Information Flows in Foreign Exchange Markets: Dissecting Customer Currency Trades}, journal = {The Journal of Finance}, volume = {71}, number = {2}, pages = {601-634}, doi = {10.1111/jofi.12378}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/jofi.12378}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/jofi.12378}, year = {2016} } @article{forex_inefficiency, title = {Intraday return inefficiency and long memory in the volatilities of forex markets and the role of trading volume}, journal = {Physica A: Statistical Mechanics and its Applications}, volume = {506}, pages = {433 - 450}, year = {2018}, issn = {0378-4371}, doi = {https://doi.org/10.1016/j.physa.2018.04.016}, url = {http://www.sciencedirect.com/science/article/pii/S0378437118304448}, author = { and and and }, } @article{teach_spread, author = {}, year = {2018}, month = {10}, pages = {55-62}, title = {Teaching The Bid-Ask Spread And Triangular Arbitrage For The Foreign Exchange Market}, volume = {11}, journal = {American Journal of Business Education (AJBE)}, doi = {10.19030/ajbe.v11i4.10212} } @article{spread_competition, title={FX Spreads and Dealer Competition Across the 24-Hour Trading Day}, author={ and }, journal={Capital Markets: Market Microstructure}, year={1999} } @article{political_forex, author = { and and and and }, title = {Economic and political effects on currency clustering dynamics}, journal = {Quantitative Finance}, volume = {19}, number = {5}, pages = {705-716}, year = {2019}, publisher = {Routledge}, doi = {10.1080/14697688.2018.1532101}, URL = {https://doi.org/10.1080/14697688.2018.1532101}, eprint = {https://doi.org/10.1080/14697688.2018.1532101} } @article{forex_microstructure, title={Microstructure of Foreign Exchange Markets}, author={ and }, journal={Available at SSRN 3345289}, year={2019}, doi = {http://dx.doi.org/10.2139/ssrn.3345289} } @article{forex_volatility, author = { and }, title = {The Global Transmission of Volatility in the Foreign Exchange Market}, journal = {The Review of Economics and Statistics}, volume = {85}, number = {3}, pages = {670-679}, year = {2003}, doi = {10.1162/003465303322369803}, URL = {https://doi.org/10.1162/003465303322369803}, eprint = {https://doi.org/10.1162/003465303322369803}, } @article{forex_liquidity, author = { and and }, title = {Liquidity in the Foreign Exchange Market: Measurement, Commonality, and Risk Premiums}, journal = {The Journal of Finance}, volume = {68}, number = {5}, pages = {1805-1841}, doi = {10.1111/jofi.12053}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/jofi.12053}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/jofi.12053}, year = {2013} } @article{forex_algorithmic, author = { and and and }, title = {Rise of the Machines: Algorithmic Trading in the Foreign Exchange Market}, journal = {The Journal of Finance}, volume = {69}, number = {5}, pages = {2045-2084}, doi = {10.1111/jofi.12186}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/jofi.12186}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/jofi.12186}, year = {2014} } @article{spread_futures, author = {}, title = {The determinants of bid-ask spreads in the foreign exchange futures market: A microstructure analysis}, journal = {Journal of Futures Markets}, volume = {19}, number = {3}, pages = {307-324}, doi = {10.1002/(SICI)1096-9934(199905)19:3<307::AID-FUT4>3.0.CO;2-5}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291096-9934%28199905%2919%3A3%3C307%3A%3AAID-FUT4%3E3.0.CO%3B2-5}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1002/%28SICI%291096-9934%28199905%2919%3A3%3C307%3A%3AAID-FUT4%3E3.0.CO%3B2-5}, year = {1999} } @article{order_flow_forex, title = {Order flow, bid-ask spread and trading density in foreign exchange markets}, journal = {Journal of Banking \& Finance}, volume = {36}, number = {2}, pages = {597-612}, year = {2012}, issn = {0378-4266}, doi = {https://doi.org/10.1016/j.jbankfin.2011.09.006}, url = {http://www.sciencedirect.com/science/article/pii/S0378426611002603}, author = { and and } } @article{local_forex, title = {Local information in foreign exchange markets}, journal = {Journal of International Money and Finance}, volume = {27}, number = {8}, pages = {1383 - 1406}, year = {2008}, issn = {0261-5606}, doi = {https://doi.org/10.1016/j.jimonfin.2007.10.003}, url = {http://www.sciencedirect.com/science/article/pii/S0261560607001179}, author = { and }, keywords = {Local information hypothesis, Price impact, Limit orders, Informed traders}, } @article{response_funct_fx, title = {Informed trade in spot foreign exchange markets: an empirical investigation}, journal = {Journal of International Economics}, volume = {61}, number = {2}, pages = {307-329}, year = {2003}, issn = {0022-1996}, doi = {https://doi.org/10.1016/S0022-1996(03)00003-5}, url = {https://www.sciencedirect.com/science/article/pii/S0022199603000035}, author = {}, } @article{physicists_contribution, title = {Econophysics: Can physicists contribute to the science of economics?}, journal = {Physica A: Statistical Mechanics and its Applications}, volume = {269}, number = {1}, pages = {156-169}, year = {1999}, issn = {0378-4371}, doi = {https://doi.org/10.1016/S0378-4371(99)00185-5}, url = {https://www.sciencedirect.com/science/article/pii/S0378437199001855}, author = { and and and and and }, } @article{my_paper_response_financial, title = {Price response functions and spread impact in correlated financial markets}, author = { and and }, year = {2021}, month = {04}, day = {08}, journal = {The European Physical Journal B}, volume = {94}, number = {4}, doi = {10.1140/epjb/s10051-021-00077-z}, url = {https://doi.org/10.1140/epjb/s10051-021-00077-z}, issn = {1434-6036}, } @article{prop_order_book, title = {More statistical properties of order books and price impact}, journal = {Physica A: Statistical Mechanics and its Applications}, volume = {324}, number = {1}, pages = {133 - 140}, year = {2003}, note = {Proceedings of the International Econophysics Conference}, issn = {0378-4371}, doi = {https://doi.org/10.1016/S0378-4371(02)01896-4}, url = {http://www.sciencedirect.com/science/article/pii/S0378437102018964}, } @article{data_01, title = {Big data analytics for financial Market volatility forecast based on support vector machine}, journal = {International Journal of Information Management}, volume = {50}, pages = {452-462}, year = {2020}, issn = {0268-4012}, doi = {https://doi.org/10.1016/j.ijinfomgt.2019.05.027}, url = {https://www.sciencedirect.com/science/article/pii/S0268401218313604}, author = { and and and and and and }, } @article{data_02, title = {FinTechs and the Market for Financial Analysis}, volume = {56}, DOI = {10.1017/S0022109020000721}, number = {6}, journal = {Journal of Financial and Quantitative Analysis}, publisher = {Cambridge University Press}, author = { and }, year={2021}, pages={1877–1907} } @article{data_03, title = {Correlation Structure and Evolution of World Stock Markets: Evidence from Pearson and Partial Correlation-Based Networks}, volume = {51}, DOI = {10.1007/s10614-016-9627-7}, journal = {Computational Economics}, author = { and and }, year={2018}, pages={607–635} } @article{data_04, title = {Financial markets under the global pandemic of COVID-19}, journal = {Finance Research Letters}, volume = {36}, pages = {101528}, year = {2020}, issn = {1544-6123}, doi = {https://doi.org/10.1016/j.frl.2020.101528}, url = {https://www.sciencedirect.com/science/article/pii/S1544612320304050}, author = { and and }, } @article{data_05, title = {Bid-ask spread determination in the FX swap market: Competition, collusion or a convention?}, journal = {Journal of International Financial Markets, Institutions and Money}, volume = {54}, pages = {78-97}, year = {2018}, issn = {1042-4431}, doi = {https://doi.org/10.1016/j.intfin.2017.08.001}, url = {https://www.sciencedirect.com/science/article/pii/S1042443117303797}, author = {}, } @article{data_06, title = {A multivariate stochastic volatility model with applications in the foreign exchange market}, journal = {Review of Derivatives Research}, volume = {21}, pages = {1-43}, year = {2018}, issn = {1573-7144}, doi = {10.1007/s11147-017-9132-8}, url = {https://doi.org/10.1007/s11147-017-9132-8}, author = { and }, } %============================================================================== %Books @book{latexcompanion, author = { and and }, title = {The \LaTeX\ Companion}, year = {1993}, publisher = {Addison-Wesley}, address = {Reading, Massachusetts} } @inbook{market_making, author = {}, editor = {Fr{\'e}d{\' and and and }, title = {``Market Making'' in an Order Book Model and Its Impact on the Spread}, bookTitle = {Econophysics of Order-driven Markets: Proceedings of Econophys-Kolkata V}, year = {2011}, publisher = {Springer Milan}, address = {Milano}, pages = {49--64}, isbn = {978-88-470-1766-5}, doi = {10.1007/978-88-470-1766-5_4}, url = {https://doi.org/10.1007/978-88-470-1766-5_4} } @inbook{intro_market_micro, author = {}, title = {Introduction to market microstructure and heterogeneity of investors}, bookTitle = {Computational Social Science and Complex Systems}, year = {2019}, pages = {73--89}, doi = {10.3254/190006}, } @inbook{forex_structure, author = { and and }, year = {2012}, month = {08}, title = {Foreign Exchange Market Structure, Players and Evolution}, isbn = {978-0-470-76883-9}, journal = {Handbook of Exchange Rates}, doi = {10.2139/ssrn.1935858} } @Inbook{forex_market_micro, author={}, editor={}, title={Market Microstructure, Foreign Exchange}, bookTitle={Complex Systems in Finance and Econometrics}, year={2011}, publisher={Springer New York}, address={New York, NY}, pages={580--614}, abstract={Glossary}, isbn={978-1-4419-7701-4}, doi={10.1007/978-1-4419-7701-4_33}, url={https://doi.org/10.1007/978-1-4419-7701-4_33} } @incollection{book_forex, title = {Chapter 5 - Stock Markets, Derivatives Markets, and Foreign Exchange Markets}, editor = {}, booktitle = {Strategies of Banks and Other Financial Institutions}, publisher = {Academic Press}, address = {San Diego}, pages = {125-164}, year = {2014}, isbn = {978-0-12-416997-5}, doi = {https://doi.org/10.1016/B978-0-12-416997-5.00005-1}, url = {https://www.sciencedirect.com/science/article/pii/B9780124169975000051}, author = {}, } @incollection{book_forex_2, title = {Chapter 1 - The Foreign Exchange Market}, editor = { and }, author = { and }, booktitle = {International Money and Finance (Eighth Edition)}, publisher = {Academic Press}, edition = {Eighth Edition}, pages = {3-24}, year = {2013}, isbn = {978-0-12-385247-2}, doi = {https://doi.org/10.1016/B978-0-12-385247-2.00001-9}, url = {https://www.sciencedirect.com/science/article/pii/B9780123852472000019}, } @incollection{book_forex_3, title = {Chapter 7 - Foreign Exchange Market and Interest Rates}, editor = { and }, booktitle = {Multi-Asset Risk Modeling}, publisher = {Academic Press}, address = {San Diego}, pages = {217-246}, year = {2014}, isbn = {978-0-12-401690-3}, doi = {https://doi.org/10.1016/B978-0-12-401690-3.00007-X}, url = {https://www.sciencedirect.com/science/article/pii/B978012401690300007X}, author = { and }, } @incollection{book_forex_4, title = {Chapter 11 - Microstructure of Currency Markets}, editor = { and and and and and and and }, booktitle = {Handbook of Safeguarding Global Financial Stability}, publisher = {Academic Press}, address = {San Diego}, pages = {107-119}, year = {2013}, isbn = {978-0-12-397875-2}, doi = {https://doi.org/10.1016/B978-0-12-397875-2.00012-X}, url = {https://www.sciencedirect.com/science/article/pii/B978012397875200012X}, author = {} } %============================================================================== %Doesn't fit in any other type @misc{r_walks_liquidity, title={Random walks, liquidity molasses and critical response in financial markets}, author={ and and }, year={2004}, eprint={cond-mat/0406224}, archivePrefix={arXiv}, primaryClass={cond-mat.other} } @misc{Wang_2017, title={Trading strategies for stock pairs regarding to the cross-impact cost}, author={}, year={2017}, eprint={1701.03098}, archivePrefix={arXiv}, primaryClass={q-fin.TR} } @incollection{market_digest, title = {CHAPTER 2 - How Markets Slowly Digest Changes in Supply and Demand}, editor = { and }, booktitle = {Handbook of Financial Markets: Dynamics and Evolution}, publisher = {North-Holland}, address = {San Diego}, pages = {57-160}, year = {2009}, series = {Handbooks in Finance}, issn = {15684997}, doi = {https://doi.org/10.1016/B978-012374258-2.50006-3}, url = {https://www.sciencedirect.com/science/article/pii/B9780123742582500063}, author = { and and }, } @incollection{complex_markets, title = {Econophysics and the Complexity of Financial Markets}, editor = {}, booktitle = {Philosophy of Complex Systems}, publisher = {North-Holland}, address = {Amsterdam}, pages = {531-565}, year = {2011}, volume = {10}, series = {Handbook of the Philosophy of Science}, issn = {18789846}, doi = {https://doi.org/10.1016/B978-0-444-52076-0.50019-5}, url = {https://www.sciencedirect.com/science/article/pii/B9780444520760500195}, author = {}, } @misc{predictive_pow, title={The Predictive Power of Zero Intelligence in Financial Markets}, author={ and and }, year={2003}, eprint={cond-mat/0309233}, archivePrefix={arXiv}, primaryClass={cond-mat.stat-mech} } @techreport{curr_speculation, title = {The Returns to Currency Speculation in Emerging Markets}, author = { and and }, institution = {National Bureau of Economic Research}, type = {Working Paper}, series = {Working Paper Series}, number = {12916}, year = {2007}, month = {February}, doi = {10.3386/w12916}, URL = {http://www.nber.org/papers/w12916}, } @misc{code, title={First release - Price response functions and spread impact in foreign exchange markets code}, author={}, year={2021}, version= {v1.0.0}, doi={10.5281/zenodo.4569918}, url={https://doi.org/10.5281/zenodo.4569918} } @misc{my_paper_response_financial_arxiv, title={Price response functions and spread impact in correlated financial markets}, author={ and and }, year={2020}, eprint={2010.15105}, archivePrefix={arXiv}, primaryClass={q-fin.ST} } @misc{bis, title={Triennial Central Bank Survey. Foreign exchange turnover in April 2013: preliminary global results}, institution={Monetary and Economic Department, Bank for International Settlements}, year={2013}, month={09}, url={https://www.bis.org/publ/rpfx13fx.pdf} } @manual{xetra, title = {T7 Release 8.0. Enhanced Order Book Interface}, organization = {Deutsche Börse Group}, year = {2020}, month = {03}, version={v8.0.4} } @misc{noauthor_differential_2021, title = {Differential privacy}, copyright = {Creative Commons Attribution-ShareAlike License}, url = {https://en.wikipedia.org/w/index.php?title=Differential_privacy&oldid=1061496127}, abstract = {Differential privacy (DP) is a system for publicly sharing information about a dataset by describing the patterns of groups within the dataset while withholding information about individuals in the dataset. The idea behind differential privacy is that if the effect of making an arbitrary single substitution in the database is small enough, the query result cannot be used to infer much about any single individual, and therefore provides privacy. Another way to describe differential privacy is as a constraint on the algorithms used to publish aggregate information about a statistical database which limits the disclosure of private information of records whose information is in the database. For example, differentially private algorithms are used by some government agencies to publish demographic information or other statistical aggregates while ensuring confidentiality of survey responses, and by companies to collect information about user behavior while controlling what is visible even to internal analysts. Roughly, an algorithm is differentially private if an observer seeing its output cannot tell if a particular individual's information was used in the computation. Differential privacy is often discussed in the context of identifying individuals whose information may be in a database. Although it does not directly refer to identification and reidentification attacks, differentially private algorithms probably resist such attacks.Differential privacy was developed by cryptographers and thus is often associated with cryptography, and draws much of its language from cryptography.}, language = {en}, urldate = {2022-02-02}, journal = {Wikipedia}, month = dec, year = {2021}, note = {Page Version ID: 1061496127}, file = {Snapshot:C\:\\Users\\Cescollino\\Zotero\\storage\\UVZGTSCT\\Differential_privacy.html:text/html}, } @article{mcmahan_communication-efficient_2017, title = {Communication-{Efficient} {Learning} of {Deep} {Networks} from {Decentralized} {Data}}, url = {http://arxiv.org/abs/1602.05629}, abstract = {Modern mobile devices have access to a wealth of data suitable for learning models, which in turn can greatly improve the user experience on the device. For example, language models can improve speech recognition and text entry, and image models can automatically select good photos. However, this rich data is often privacy sensitive, large in quantity, or both, which may preclude logging to the data center and training there using conventional approaches. We advocate an alternative that leaves the training data distributed on the mobile devices, and learns a shared model by aggregating locally-computed updates. We term this decentralized approach Federated Learning. We present a practical method for the federated learning of deep networks based on iterative model averaging, and conduct an extensive empirical evaluation, considering five different model architectures and four datasets. These experiments demonstrate the approach is robust to the unbalanced and non-IID data distributions that are a defining characteristic of this setting. Communication costs are the principal constraint, and we show a reduction in required communication rounds by 10-100x as compared to synchronized stochastic gradient descent.}, urldate = {2022-02-02}, journal = {arXiv:1602.05629 [cs]}, author = {McMahan, and Moore, Ramage, Hampson, Arcas, }, month = feb, year = {2017}, note = {arXiv: 1602.05629}, keywords = {Computer Science - Machine Learning}, file = {arXiv Fulltext PDF:C\:\\Users\\Cescollino\\Zotero\\storage\\PVMZ2F5C\\McMahan et al. - 2017 - Communication-Efficient Learning of Deep Networks .pdf:application/pdf;arXiv.org Snapshot:C\:\\Users\\Cescollino\\Zotero\\storage\\GTQ5Z5EF\\1602.html:text/html}, } @misc{jadhav_federated-learning_2022, title = {Federated-{Learning} ({PyTorch})}, copyright = {MIT}, url = {https://github.com/AshwinRJ/Federated-Learning-PyTorch}, abstract = {Implementation of Communication-Efficient Learning of Deep Networks from Decentralized Data}, urldate = {2022-02-02}, author = {.}, month = feb, year = {2022}, note = {original-date: 2018-11-16T23:51:14Z}, keywords = {deep-learning, distributed-computing, federated-learning, python, pytorch}, } storna_glowna.tex \thispagestyle{empty} \makeatletter \begin{titlepage} \sffamily\bfseries \center{ \fontsize{18}{18}\selectfont{ WYŻSZA SZKOŁA INFORMATYKI I ZARZĄDZANIA ,,COPERNICUS'' WE WROCŁAWIU } \rule[10pt]{\textwidth}{1pt} \raisebox{12pt}[0px]{ \fontsize{16}{16}\selectfont{WYDZIAŁ INFORMATYKI} } } \flushleft\fontsize{14}{14}\selectfont{ \parbox{110pt}{\textmd{Kierunek studiów:}} Informatyka \parbox{110pt}{\textmd{Poziom studiów:}} Studia pierwszego stopnia-inżynierskie \parbox{110pt}{\textmd{Specjalność:}} Systemy i sieci komputerowe } \vspace*{40pt} \center{ \fontsize{14}{14}\selectfont{PRACA DYPLOMOWA INŻYNIERSKA} \vspace*{30pt} \fontsize{14}{14}\selectfont\@author \vspace*{15pt} \fontsize{20}{20}\selectfont\@title \vspace*{20pt} \fontsize{14}{14}\selectfont\@engTitle } \vspace*{80pt} \flushright{ \fontsize{14}{14}\selectfont{Ocena pracy:} \makebox[220pt][r]{\fontsize{10}{10}\selectfont\dotfill} \fontsize{10}{10}\selectfont\textmd{(ocena pracy dyplomowej, data, podpis promotora) } \vspace*{50pt} \makebox[220pt][r]{\fontsize{10}{10}\selectfont\dotfill} \raisebox{4pt}{\fontsize{10}{10}\selectfont\textmd{(pieczątka uczelni)}} } \flushleft\fontsize{14}{14}\selectfont{ Promotor: \bigskip\@promoter } \vspace*{20pt} \center{ \rule[3pt]{\textwidth}{1pt} \fontsize{16}{16}\selectfont{WROCŁAW \@year} } \end{titlepage} \makeatother \clearpage %\newpage %\thispagestyle{empty} %\mbox{} %\newpage \section{Scope Definition} Due to limited time frame of this thesis, the scope shall be reduced to focus on the most important features. Naturally, these limitations contradict some of the requirements for an optimal \textit{decentralised streaming application}. \begin{itemize} \item Any user must be able to transmit video to $n$ other users. \item One producer and $n$ consumers are clustered in a channel. Any channel must only support one concurrent stream (one–to–many). \item Users need not be able to discover all other users in their room and video streams should be broadcast to all active nodes in a cluster. \item Methods to ensure message authenticity, user identity and media encryption shall be explored, but not implemented. \end{itemize} @techreport{Zheng2020, abstract = {Adversarial training is an effective defense method to protect classification models against adversarial attacks. However, one limitation of this approach is that it can require orders of magnitude additional training time due to high cost of generating strong adversarial examples during training. In this paper, we first show that there is high transferability between models from neighboring epochs in the same training process, i.e., adversarial examples from one epoch continue to be adversarial in subsequent epochs. Leveraging this property, we propose a novel method, Ad-versarial Training with Transferable Adversarial Examples (ATTA), that can enhance the robustness of trained models and greatly improve the training efficiency by accumulating adversarial perturbations through epochs. Compared to state-of-the-art adversarial training methods, ATTA enhances adversarial accuracy by up to 7.2{\%} on CIFAR10 and requires 12 ∼ 14× less training time on MNIST and CIFAR10 datasets with comparable model robustness.}, author = { and and }, file = {:Users/ziqizhang/Documents/Mendeley/Zheng et al/Zheng et al. - 2020 - Efficient Adversarial Training with Transferable Adversarial Examples.pdf:pdf}, pages = {1181--1190}, title = {{Efficient Adversarial Training with Transferable Adversarial Examples}}, year = {2020} } @article{Zheng2020a, abstract = {Deep Neural Networks (DNNs) are known to be vulnerable to adversarial attacks. Currently, there is no clear insight into how slight perturbations cause such a large difference in classification results and how we can design a more robust model architecture. In this work, we propose a novel interpretability method, InterpretGAN, to generate explanations for features used for classification in latent variables. Interpreting the classification process of adversarial examples exposes how adversarial perturbations influence features layer by layer as well as which features are modified by perturbations. Moreover, we design the first diagnostic method to quantify the vulnerability contributed by each layer, which can be used to identify vulnerable parts of model architectures. The diagnostic results show that the layers introducing more information loss tend to be more vulnerable than other layers. Based on the findings, our evaluation results on MNIST and CIFAR10 datasets suggest that average pooling layers, with lower information loss, are more robust than max pooling layers for the network architectures studied in this paper.}, archivePrefix = {arXiv}, arxivId = {2007.08716}, author = { and and }, eprint = {2007.08716}, file = {:Users/ziqizhang/Documents/Mendeley/Zheng et al/Zheng et al. - 2020 - Understanding and Diagnosing Vulnerability under Adversarial Attacks.pdf:pdf}, month = {jul}, title = {{Understanding and Diagnosing Vulnerability under Adversarial Attacks}}, url = {http://arxiv.org/abs/2007.08716}, year = {2020} } lsst-pst/pstn-037 \begin{abstract} This is the abstract. \end{abstract} Report/topo.tex \subsection{Method Selection} Before establishing models for prediction, we have to choose what algorithms to use in analyses. Besides common considerations, what we especially take into account is the global topology of the attribute space: because of the attributes WDF2 and WDF5, the space is in fact homeomorphic to $\mathbb R^{p-2}\times \mathbb S^1\times\mathbb S^1$ instead of $\mathbb R^p$. In this case, a hyperplane (or, more generally, an \emph{open} hypersurface) may not be able to seperate it into two \emph{disconnected} subspace. As the result, we exclude LDA, QDA and several other well-developed classification methods because an open hypersurface is always needed for these. Instead, we tend to apply methods that do not heavily depend on global topology (random forest, $k$-NN) or that classify with a \emph{closed} hypersurface (kernel SVM).\begin{titlepage} % AUTH Logo \centering\begin{minipage}{0.3\textwidth} \centering\includegraphics[height=3cm]{university.png} \end{minipage}% \begin{minipage}{0.7\textwidth} \begin{flushleft} \large Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης\\ Πολυτεχνική Σχολή\\ Τμήμα Ηλεκτρολόγων Μηχανικών \&\\Μηχανικών Υπολογιστών\\ Τομέας Ηλεκτρονικής και Υπολογιστών\\ Ομάδα Ευφυών Συστημάτων και Τεχνολογίας Λογισμικού (ISSEL)\\[5cm] \end{flushleft} \end{minipage} \\[1.7cm] \begin{center} \Large Διπλωματική Εργασία \\[0.8cm] \rule{450pt}{4pt} \\[0.4cm] \makeatletter{\fontsize{20.26pt}{1em}\selectfont{}\@title{}}\makeatother \rule{350pt}{4pt} \\[4cm] \noindent\begin{minipage}{0.5\textwidth} \begin{flushleft} \large \emph{Εκπόνηση:} \\ Φλώρος-Μαλιβίτσης Ορέστης\\ΑΕΜ:7796\\ \end{flushleft} \end{minipage}\hfill% \begin{minipage}{0.5\textwidth} \begin{flushright} \large \emph{Επίβλεψη:} \\ Ανδρέας Λ. Συμεωνίδης --\\ Αναπληρωτής καθηγητής\\ Εμμανουήλ Τσαρδούλιας --\\ Μεταδιδάκτορας \end{flushright} \end{minipage} \\[1cm] \vfill \large Θεσσαλονίκη, 4 Ιουλίου 2019 \end{center} \end{titlepage} % vim:ts=4:sw=4:expandtab:fo-=tc:tw=120 examples/sources/test9.tex \documentclass{article} \usepackage{tikz} \begin{document} \input diag \end{document} \section{Theorie} \label{sec:Theorie} Der Franck-Hertz-Versuch belegt die Aussage von Niels Bohrs Atommodell, dass Atome diskrete Energieniveaus besitzen. Dies bedeutet, dass Atome -- in diesem Fall Quecksilber -- durch diskrete Anregungsenergien in einen höherenergetischen Zustand versetzt werden können und bei der Rückkehr in den Anfangszustand einen Lichtquanten mit einer Energie emittieren, die der Energiedifferenz zwischen Anfangszustand und angeregtem Zustand, also \begin{equation} \label{eqn:photonemission} \symup{h} \nu = E_1 - E_0 \end{equation} entspricht. Hierbei entspricht $\symup{h}$ dem Planck'schen Wirkungsquantum und $\nu$ der Frequenz des emittierten Lichtquants. Die Anregung der Atome kann durch zwei unterschiedliche Methoden realisiert werden. Zum Einen durch Wechselwirkung elektromagnetischer Strahlung mit den Atomen. Zum Anderen mit den hier betrachteten Stoßprozessen von Elektronen mit den Atomen, den sogenannten \textbf{Elektronenstoßexperimenten}. Um dieses Phänomen näher zu behandeln wird zunächst der prinzipielle Aufbau des Franck-Hertz-Versuchs erläutert. \subsection{Prinzipieller Aufbau des Franck-Hertz-Versuchs} \begin{figure} \centering \includegraphics[width=0.7\textwidth]{Bilder/aufbau.png} \caption{Prinzipieller Aufbau des Frack-Hertz-Versuchs \cite{Anleitung}.} \label{fig:franckhertztheory} \end{figure} Der prinzipielle Aufbau des Franck-Hertz-Versuchs ist in Abbildung \ref{fig:franckhertztheory} dargestellt. In einem evakuierten Gefäß befinden sich ein Glühdraht, eine Beschleunigungselektrode, eine Auffängerelektrode und die Quecksilberatome. Die Dichte der Quecksilberatome ist gemäß der Dampfdruck-Kurve temperaturabhängig. Der Glühdraht, bestehend aus einem Material mit niedriger Austrittsarbeit, wird durch eine Gleichspannung erhitzt und es treten gemäß dem glühelektrischen Effekt Elektronen aus dem Draht aus. Da zwischen dem Glühdraht und der Beschleunigungselektrode eine positive Spannung anliegt, erhalten die Elektronen -- vorrausgesetzt sie besitzen unmittelbar nach dem Austritt aus dem Draht keine kinetische Energie -- bis zur Beschleunigungselektrode die kinetische Energie \begin{equation} E_{\mathrm{kin}} = \frac{\symup{m}_0 v_{\mathrm{vor}}^2}{2} = \symup{e}_0 U_{\mathrm{B}} \mathrm{,} \end{equation} wobei $\symup{m}_0$ der Elektronenmasse und $\symup{e}_0$ der Elementarladung entspricht. Am Ende des Gefäßes ist eine Auffängerelektrode angebracht, die ein Gegenfeld mit der Spannung $U_{\mathrm{A}}$ erzeugt, sodass nur Elektronen die Auffängerdiode erreichen, die die Ungleichung \begin{equation} \frac{\symup{m}_0}{2} v_{\mathrm{z}}^2 \geq \symup{e}_0 U_{\mathrm{A}} \end{equation} erfüllen. Das $v_{\mathrm{z}}$ stellt die z-Komponente des Geschwindigkeitsvektors dar, angenommen die z-Achse liegt in Richtung der Auffängerelektrode. Treffen die Elektronen auf Quecksilberatome können zwei Arten von Stößen auftreten: die Elastischen und die Unelastischen. Bei den elastischen Stößen reicht die Energie der Elektronen nicht aus um das Atom anzuregen. Aufgrund der großen Massendifferenz der beiden Stoßpartner kann die Energieabgabe der Elektronen vernachlässigt werden. Allerdings erfahren die Elektronen beim elastischen Stoß eine Richtungsänderung. Bei den unelastischen Stößen hingegen besitzen die Elektronen eine hinreichend große Energie um das Atom in einen höherenergetischen Zustand zu versetzen und übertragen diesen dafür nötigen Energiebetrag $E_1-E_0$ auf das Atom. Das angeregte Atom relaxiert in den Anfangszustand und emittiert dabei ein Photon gemäß Formel \eqref{eqn:photonemission}. \begin{figure} \centering \includegraphics[width=0.7\textwidth]{Bilder/ideale_hertzkurve.png} \caption{Charakteristische Gestalt der idealen Franck-Hertz-Kurve \cite{Anleitung}.} \label{fig:franckhertzkurve} \end{figure} Durch diese Stoßprozesse ergibt sich ein theoretischer Verlauf des an der Auffängerelektrode gemessenen Auffängerstroms $I_{\mathrm{A}}$ in Abhängigkeit von der Beschleunigungsspannung $U_{\mathrm{B}}$ wie in Abbildung \ref{fig:franckhertzkurve} zu sehen ist. Dieser Graph wird als \textbf{Franck-Hertz-Kurve} bezeichnet. Es wird idealisierend von einer monoenergetischen Energieverteilung der Elektronen ausgegangen. Dann werden die Elektronen beschleunigt, bis sie eine hinreichend hohe Energie besitzen, um das Gegenfeld zu durchqueren. Die Anzahl der Elektronen steigt mit steigender Beschleunigungsspannung bis ein instantaner Abfall des Auffängerstroms erfolgt. Die Elektronen haben an diesem Punkt genau die Energie, um die Quecksilberatome anzuregen. Dieser Vorgang wiederholt sich periodisch. Außerdem lässt sich an der Franck-Hertz-Kurve das Anregungspotential als äquidistanter Abstand der Maxima zu \begin{equation} \label{eqn:abstand} U_1 = \frac{E_1-E_0}{\symup{e}_0} \end{equation} bestimmen. Im Folgenden werden Einflüsse diskutiert, die zur Folge haben, dass die Franck-Hertz-Kurve nicht wie in Abbildung \ref{fig:franckhertzkurve} bestimmt werden kann. \subsection{Einflüsse auf die Gestalt der Franck-Hertz-Kurve} \subsubsection{Einfluss des Kontaktpotentials} \begin{figure} \centering \includegraphics[width=\textwidth]{Bilder/fermi_niveau.png} \caption{Darstellung der Potentialverhältnisse zwischen Glühkathode und Beschleunigungselektrode \cite{Anleitung}.} \label{fig:potentialverh} \end{figure} In der Regel besteht der Glühdraht aus einem Material mit niedriger Austrittsarbeit $\Phi_{\mathrm{G}}$, einer niedrigeren als bei der Beschleunigungselektrode $\Phi_{\mathrm{B}}$. Dadurch verändert sich die effektive Beschleunigungsspannung $U_{\mathrm{B,eff}}$ wie in Abbildung \ref{fig:potentialverh} illustriert. Es ergibt sich das effektive Beschleunigungspotential \begin{equation} U_{\mathrm{B,eff}} = U_{\mathrm{B}} - \frac{\Phi_{\mathrm{B}}-\Phi_{\mathrm{G}}}{\symup{e}_0} \mathrm{.} \end{equation} Hierbei wird der Subtrahend \begin{equation} K:=\frac{\Phi_{\mathrm{B}}-\Phi_{\mathrm{G}}}{\symup{e}_0} \end{equation} als \textbf{Kontaktpotential} bezeichnet. Die Franck-Hertz-Kurve wird durch diesen Einfluss um das Kontaktpotential $K$ verschoben. \subsubsection{Einfluss des Energiespektrums der Elektronen} Ein weiterer Einfluss auf die Franck-Hertz-Kurve ist die nicht gegebene monoenergetische Energieverteilung der Elektronen. Die Elektronen besitzen nach der Fermi-Dirac-Statistik ein statistisch verteiltes Energiespektrum im Draht und es ergibt sich somit nach Austritt der Elektronen aus dem Draht eine Geschwindigkeitsverteilung. Dies hat zur Folge, dass die Maxima der Franck-Hertz-Kurve nicht mehr einer bestimmten Beschleunigungsspannung zuzuordnen sind, sondern sich aufgrund der statistischen Verteilung über ein Intervall erstrecken. \subsubsection{Einfluss des Dampfdruckes} Voraussetzung für den Franck-Hertz-Versuch sind Stöße der Elektronen mit Atomen. Daher muss die mittlere freie Weglänge $\bar{w}$ klein gegen den Abstand $a$ zwischen Kathode und Beschleunigungselektrode sein. Die mittlere freie Weglänge ergibt sich über Sättigungsdampfdruck $p_{\mathrm{sät}}$ gemäß der kinetischen Gastheorie zu \begin{equation} \label{eqn:freiewegl} \bar{w} = \frac{0,0029}{p_{\mathrm{sät}}} \mathrm{.} \end{equation} Der Sättigungsdampfdruck ist abhängig von der Temperatur und berechnet sich wie folgt: \begin{equation} \label{eqn:p} p_\mathrm{sät}=5,5\cdot10^{7}\exp(-6876/T) \mathrm{.} \end{equation} %\begin{figure} % \centering % \includegraphics[width=0.7\textwidth]{Bilder/dampfdruckkurve.png} % \caption{Dampfdruckkurve des Quecksilbers \cite{Anleitung}.} % \label{fig:dampfi} %\end{figure} 0 \documentclass[a4paper]{article} \usepackage[paperwidth=180mm,paperheight=285mm,left=3cm,top=4cm,right=3cm,bottom=2cm,head=2.0cm,includefoot]{geometry} \usepackage{makeidx} \usepackage[spanish]{babel} \usepackage[utf8]{inputenc} \usepackage{amsmath} \usepackage{url} \usepackage{fancyvrb} \usepackage{babelbib} \usepackage{graphicx} \usepackage{lscape} \usepackage{fancyhdr} \graphicspath{ {img/} } %%%%%%%%%% Text with box \usepackage{tikz} \usetikzlibrary{shadows} \newcommand{\raisedtext}[1]{ \vspace{1mm} \begin{tikzpicture} [baseline=(X.base)]\node [drop shadow,fill=white,draw,very thin] (X) {#1}; \end{tikzpicture} \vspace{1mm} } %%%%%%%%%%%%%%%%%%% \newtheorem{definicion}{Definición} \urldef{\mails}\path|{,| \title{Módulo de resúmenes automáticos basado en TextRank con integración a Gensim} \author{ \and } \lhead{\includegraphics[scale=0.06]{logo_fiuba.pdf}} \chead{Módulo de Resúmenes Automáticos Basado\linebreak en TextRank con Integración a Gensim} \rhead{} \lfoot{Barrios -- López} \rfoot{\thepage} \cfoot{Mayo, 2015} \begin{document} \thispagestyle{empty} % T\'itulo del documento. \begin{center} \includegraphics{logo-fiuba.png}\\ \vspace{1cm} \textsc{\LARGE Universidad de Buenos Aires}\\[0.3cm] \textsc{\LARGE Facultad de Ingeniería}\\[1.2cm] \textsc{\LARGE Trabajo Profesional}\\[1.2cm] \textsc{\LARGE Módulo de resúmenes automáticos basado en TextRank con integración a Gensim}\\[0.3cm] \end{center} \vspace{20 mm} \begin{flushright} {\large -- 91954\\ -- 92278\\[0.1cm] \vspace{2cm} Mayo, 2015} \end{flushright} \pagestyle{fancy} \newpage \footnotesize \newpage \setcounter{page}{1} \tableofcontents \newpage \section{Motivación} Los resúmenes automatizados son muy utilizados en tareas relacionadas con el procesamiento de lenguaje natural y de aprendizaje automático. Su uso en motores de búsqueda, por ejemplo, mejora la eficiencia de indexación de textos y a su vez asiste en la presentación de resultados de manera efectiva. El incremento en la cantidad de información disponible en Internet ha intensificado su utilización en los últimos años y, en consecuencia, se ha dedicado enorme esfuerzo para mejorar los algoritmos existentes. El desarrollo de este Trabajo Profesional está motivado por el número de aplicaciones del tema en tareas de actualidad. \section{Objetivos} El presente trabajo consta de tres objetivos principales: \begin{itemize} \item Desarrollar el módulo para generar resúmenes automáticos usando un algoritmo conocido. \item Analizar, diseñar e implementar modificaciones para intentar mejorar el rendimiento del algoritmo seleccionado. \item Integrar las implementaciones a una herramienta de código abierto de procesamiento del lenguaje natural. \end{itemize} \section{Introducción teórica} \subsection{Resúmenes: generación y clasificación} Un resumen es una reducción a términos breves y precisos de lo esencial de una fuente de información. Su objetivo es el de extraer contenido intentando sintetizar sus conceptos más importantes, y su uso es altamente benéfico en tareas de aprendizaje debido a que: \begin{itemize} \item Facilitan la selección de información, \item Acortan tiempos de lectura, \item Simplifican búsquedas en textos, \item Optimizan la creación de índices. \end{itemize} La investigación indica, además, que contribuyen en tareas automatizadas: su utilización para presentar resultados de motores de búsquedas atrajo el interés de académicos desde principios de la década del 2000, convirtiéndose hoy en día en una funcionalidad básica de los principales buscadores de Internet. Sin embargo, si bien esta tarea puede no resultar costosa para un ser humano, durante muchos años fue considerada como difícil de automatizar. Este tema es uno de los que investiga el campo de procesamiento de lenguaje natural, dedicado a facilitar la interacción entre las computadoras y los seres humanos. \subsubsection{Enfoque abstractivo y extractivo} Existen dos métodos para generar resúmenes automatizados: abstractivos y extractivos. Los primeros se construyen regenerando el contenido extraído del texto original; esto es, se reformulan las frases por medio de técnicas de generación de lenguaje natural: fusión, combinación o supresión de términos. De esta manera, se obtienen pasajes que en principio no pertenecían al texto de origen, similar al modo en que lo harían las personas. Por el contrario, los resúmenes extractivos se crean a partir de la selección de un conjunto de palabras u oraciones consideradas sobresalientes en el texto original. Las oraciones se obtienen literalmente, se unen libremente, y se presentan con el objetivo de crear un resumen del texto. Las técnicas abstractivas son un área de creciente interés. Sin embargo, debido a la complejidad y las restricciones técnicas aparejadas, la investigación se ha volcado hacia el enfoque extractivo. \subsubsection{Aprendizaje supervisado y no supervisado} En el campo del aprendizaje automático, a los algoritmos que requieren de un conjunto de entrenamiento con datos etiquetados se denominan supervisados. Analizando estos ejemplos de entrenamiento, el sistema infiere funciones que pueden ser usadas para aplicarlas a nueva información. Por el contrario, a los algoritmos que no son entrenados se los denomina no supervisados. Este grupo utiliza técnicas para encontrar patrones o grupos ocultos en los datos a analizar. \subsubsection{Fuente de los datos} Los algoritmos de generación de resúmenes también se categorizan de acuerdo a la cantidad de documentos que se toman como fuente, determinando sumarización de un documento y multi-documento. Para este último caso se proporcionan textos con una temática afín. El resumen resultante permite a los lectores familiarizarse rápidamente con los tópicos principales de una colección de documentos. \subsubsection{Propósito del resumen} Los resúmenes de documentos pueden crearse de manera genérica o bien basados en consultas. Aquéllos resúmenes basados en consultas favorecen temas o aspectos específicos de un texto, de acuerdo a dicha consulta; teniendo en cuenta sus palabras clave o los tópicos a los que hace referencia. Los resúmenes genéricos, por otro lado, proporcionan los temas más relevantes tratados en el texto por sí solo. \subsection{Extracción de palabras claves} La extracción de palabras claves consiste en identificar el conjunto de términos que mejor describen a un documento de manera automatizada. Esto sirve como introducción a sus conceptos más relevantes, pues uno de los primeros pasos para modelar el conocimiento de una comunidad es tomar vocabulario del dominio. Es por ello que se han estudiado métodos para extraer este léxico técnico en base a documentos del área o comunidad en estudio. Dichas palabras se pueden utilizar para construir índices de colecciones, clasificar textos, extraer terminología de dominio específica, o incluso pueden servir como un breve resumen. El enfoque más sencillo para esta tarea ha empleado un criterio basado en la frecuencia de las palabras. Sin embargo, se han investigado otras debido a los malos resultados de esta técnica. Las principales se basan en métodos supervisados, donde el sistema es entrenado para reconocer palabras claves tomando en cuenta cuestiones sintácticas y/o léxicas. En la sección ‘TextRank para palabras claves’ se analiza y desarrolla la aplicación de TextRank como método no supervisado para esta tarea. Los resultados obtenidos se equiparan los de las técnicas principales en el área. \section{Trabajo previo} El campo de la generación automática de resúmenes ha atraido interés desde finales de la década de 1950 hasta la actualidad \cite{luhn}. Los métodos tradicionales tienen en cuenta la frecuencia de palabras o frases introductorias para identificar las oraciones más sobresalientes del texto. También se han desarrollado modelos estadísticos basados en corpus de entrenamiento para combinar varias heurísticas: palabras clave, posición de las oraciones, longitud de las oraciones, frecuencia de palabras y palabras contenidas en los títulos \cite{hovy}. Otros enfoques más modernos se basan en la representación del texto en forma de grafo: las oraciones importantes y los conceptos son las entidades altamente conectadas y, por esto, forman parte del resumen \cite{barzilay}. Con esta idea se usan métodos del campo de Recuperación de Información para identificar oraciones similares y determinar las más importantes, que formarán al resumen final \cite{salton}. El enfoque propuesto, tanto por Mihalcea y Tarau en TextRank \cite{mihalcea-tarau} como por Erkan y Radev en LexRank \cite{erkan}, consiste en utilizar el prestigio de las unidades léxicas (oraciones o palabras) dentro del grafo. \section{Herramientas y fundamentos} \subsection{PageRank} PageRank es un algoritmo creado en 1996 por y , fundadores de Google, como parte de una investigación sobre motores de búsqueda. Su objetivo es asignar un valor numérico a cada nodo en una red, con el propósito de medir la importancia relativa de cada elemento del conjunto. Para esto analiza los vínculos que componen la red. La idea subyacente es que los nodos más relevantes de la red serán los que tengan la mayor cantidad y calidad de conexiones con otros nodos. En sus orígenes, PageRank fue pensado para aplicarse sobre la red de Internet, pero puede aplicarse a cualquier otra red para estimar la importancia de los nodos. En términos matemáticos, PageRank brinda una manera iterativa de calcular el autovector principal de la matriz de adyacencia del grafo. Esto da como resultado la distribución de probabilidad de acceder a los nodos de la red, navegando de manera aleatoria a través de los vínculos o aristas. \subsection{TextRank} TextRank es un algoritmo no supervisado basado en grafos para realizar resúmenes automáticos extractivos u obtener palabras claves de un texto. Fue presentado en 2004 por y en \cite{mihalcea-tarau}. El algoritmo aplica una variación de PageRank \cite{pageetal98} sobre un grafo especialmente diseñado para la tarea. De esta manera permite explotar la estructura del texto, identificando los conceptos principales, sin necesidad de datos previos de entrenamiento. Debido a que se basa en PageRank, se sirve de la noción del “prestigio” o “recomendación” entre los elementos del grafo. Por este motivo, TextRank puede ser aplicado a cualquier texto, incluso en distintos idiomas, generando un resumen basado sólo en las propiedades intrínsecas del texto. El algoritmo modela el texto en base a un grafo, y luego, busca crear relaciones significativas (aristas) entre las entidades léxicas (vértices). Dependiendo de la aplicación que se desee dar al algoritmo, las entidades pueden ser palabras, frases, oraciones, párrafos, entre otros. De manera similar, también debe definirse el tipo de relación que se usa para unir los vértices: semántica, contextual, de superposición, y demás. Los pasos principales que se llevan a cabo son los siguientes: \begin{enumerate} \item Identificar las unidades del texto y agregarlas al grafo como vértices. \item Identificar relaciones que conectan a estas unidades, y agregarlas al grafo como aristas entre los vértices. Las aristas pueden ser dirigidas o no, y ponderadas o no. \item Aplicar PageRank para asignarle un puntaje a cada vértice. \item Ordenar los vértices de acuerdo al puntaje y utilizarlo para armar el resumen de acuerdo a algún criterio. \end{enumerate} \subsection{Gensim} Gensim es una biblioteca de código libre, escrita en Python, para el modelaje de tópicos y la indexación de documentos que está diseñada para trabajar con conjuntos de textos de gran tamaño \cite{rehurek_lrec}. Su uso se ha expandido tanto en lo comercial como lo académico, apuntando especialmente al área de procesamiento del lenguaje natural y a la búsqueda y recuperación de la información. Gensim utiliza un paradigma muy poderoso en el área de procesamiento de lenguaje natural, denominado VSM (modelo de espacio de vectores), en el cual se representan los documentos de un corpus como vectores de un espacio multidimensional. Esta representación explota la idea de que los textos en el área del modelado de tópicos se pueden expresar según un número de conceptos, lo que aumenta la eficiencia y también ayuda a eliminar ruido. La biblioteca está diseñada para proveer independencia del tamaño del corpus, es decir, para poder analizar textos que sean más grandes que la memoria RAM del sistema. El proyecto, además apunta a exponer una interfaz mínima e intuitiva utilizando terminología de procesamiento de lenguaje. Gensim proporciona implementaciones de algoritmos como TF-IDF, análisis semántico latente, proyecciones aleatorias y alocación de Dirichlet latente. \subsection{Evaluación de resúmenes} La herramienta por defecto para evaluar los sistemas generadores de resúmenes es \mbox{ROUGE} (Recall-Oriented Understudy for Gisting Evaluation) \cite{Lin2004a}. Este método fue desarrollado en la universidad de Southern California para automatizar la comparación entre resúmenes generados por el sistema bajo evaluación contra otros (ideales) escritos por seres humanos, tomados como referencia. Históricamente, los puntajes asignados en tareas de competencias como DUC fueron manuales; juzgando coherencia, capitalización, cohesión, consistencia y contenido \cite{duc2002}. Todo este trabajo es caro y difícil de coordinar, demandando de al menos 3.000 horas hombre. El surgimiento de ROUGE atiende la necesidad de automatizar ese proceso, reemplazando un método llamado BLEU, cuyos resultados no siempre se correlacionaban con los asignados por los jurados. El mecanismo de funcionamiento de los métodos del paquete ROUGE consiste en calcular la sensibilidad (recall) de unidades léxicas entre los resúmenes de sistema y los de referencia. Se generan así diferentes métricas: \begin{itemize} \item ROUGE-N: cuando se tienen en cuenta n-gramas (n palabras). Este método favorece a documentos con palabras compartidas por más candidatos. \item ROUGE-L: cuando se tiene en cuenta la subsecuencia máxima común (LCS, longest common subsequence). Este método propone que un conjunto de documentos va a ser más parecido mientras más larga sea la subsecuencia común máxima. \item ROUGE-W: le asigna pesos a las subsecuencias según su largo. \item ROUGE-S: es similar a ROUGE-2, teniendo en cuenta bigramas pero permitiendo una cantidad arbitraria de espacios entre palabras. \item ROUGE-SU: extiende la idea de ROUGE-S, pero además le otorga valor a las palabras que haya en común, sin importar que no verifiquen un orden. \end{itemize} ROUGE asigna un puntaje entre cero y uno a un esquema de evaluación conformado por una serie de documentos generados y sus referencias. Por diseño, se espera que buenos resúmenes tengan puntajes más altos; pero inevitablemente se requiere de trabajo humano para proveer las referencias. Debido a que los resultados arrojados por las métricas se correlacionan con los asignados manualmente, las evaluaciones en las competencias de generación de resúmenes automáticos se llevan a cabo usando exclusivamente ROUGE \cite{mihalcea-tarau}. \section{Implementación} \subsection{Preprocesamiento} El preprocesamiento del texto es una parte esencial en cualquier sistema que trate con datos del lenguaje natural. En esta fase se identifican las unidades léxicas que serán analizadas por la aplicación en las etapas subsiguientes. La calidad de este proceso tendrá un gran impacto en los resultados obtenidos. Es por ello que se deben seleccionar unidades léxicas significativas. Es decir, aquellas que contengan las características lingüísticas relevantes para el caso en estudio. A su vez, se deben filtrar las que no aportan información útil. En este proceso también, se unifica la codificación que pudiera existir en los distintos textos. Aquí se detallan las etapas principales. \subsubsection{Separación del texto} Esta etapa, también conocida como \textit{tokenización} consiste en separar el texto en oraciones o palabras. Separar palabras suele resultar más sencillo, dado que los espacios en blanco son una buena aproximación como límite de separación. Separar por oraciones suele ser caso más complejo. Definir los límites es una cuestión realmente problemática en el área del lenguaje natural. Generalmente se utilizan los símbolos .!? como separadores, pero se deben tratar con cuidado excepciones como \textit{Ph.D.}, \textit{Dr.}, \textit{Lic.}, \textit{\$19.99}, \textit{p.m.} o \textit{RR.HH}. El enfoque más utilizado para resolver esta cuestión se basa en la aplicación de expresiones regulares, aunque existen otras alternativas que implementan algoritmos entrenados con datos específicos del área sobre la cual se esté trabajando. \subsubsection{Filtrado} Las palabras más frecuentes muchas veces no tienen tanto significado por sí solas. En muchas tareas de procesamiento de lenguaje natural estas palabras aportan “ruido”, y por lo tanto se las debe quitar. La forma de filtrar palabras consiste en crear una lista de “palabras prohibidas” (\textit{stopwords}) a ser eliminadas del texto. \subsubsection{Stemming} El proceso de \textit{stemming} o \textit{lemmatización} consiste en quitar la desinencia de las diferentes términos, para llevarlos a su raíz. Esta etapa se lleva a cabo con el objetivo de que el algoritmo detecte cuando se están tratando los mismos conceptos, dado que se utilizan las mismas palabras. Por ejemplo, las palabras “canta”, “cantaba”, “cantando”, pueden ser asociadas al verbo “cantar”. El stemming es un proceso complejo dado que puede haber muchos casos excepcionales, y se necesitan reglas muy diferentes para cada idioma. El stemmer más usado en inglés es el Porter Stemmer. Para cada idioma existe una adaptación. Sin embargo, todos estos algoritmos tienen cierta tasa de error. \subsection{TextRank para extracción de oraciones} El problema de la extracción de oraciones apunta a identificar las secuencias más representativas del texto. Para este caso, las unidades léxicas tenidas en cuenta para aplicar el algoritmo serán oraciones completas \cite{introductionir}. Inicialmente, se construye un grafo en base al texto, agregando a cada oración como vértice. Para crear las aristas con su peso correspondiente, se debe definir una función de similitud entre dos oraciones dadas. Esta función será la que dicte cuánto una oración “recomienda” a otra por poseer contenidos similares y abordar los mismos conceptos. La función utilizada formalmente por el algoritmo original se define de la siguiente manera: \begin{definicion} Sean $S_i$, $S_j$ dos oraciones representadas por un conjunto de $n$ palabras que en $S_i$ aparecen como $S_i = w_{1}^{i}, w_{2}^{i},..., w_{n}^{i}$. La función de similitud para $S_i$, $S_j$ se define como: \begin{equation} Similitud(S_{i},S_{j}) = \frac{ | \{ w_{k} | w_{k} \in S_{i} \& w_{k} \in S_{j} \} | } { log(|S_{i}|) + log(|S_{j}|) } \end{equation} \end{definicion} El resultado de este proceso es el texto representado como un grafo altamente denso, al cuál se le aplicará PageRank para seleccionar los vértices más relevantes. Finalmente, a las oraciones más importantes se las presenta de acuerdo al orden de aparición en el texto original. \subsection{TextRank para palabras claves} La otra aplicación que se le puede dar a TextRank es la extracción de palabras clave (\textit{keywords}) de un texto. Es un problema similar a la extracción de oraciones, dado que el objetivo es identificar un conjunto de unidades que sean representativas de dicho texto. Aquí, las unidades serán palabras, sueltas o combinadas. Al igual que en el caso de la extracción de oraciones, luego de la etapa de preprocesamiento, las palabras no filtradas se agregan al grafo. Se pueden agregar filtros sintácticos adicionales para limitar el análisis sólo a, por ejemplo, sustantivos y verbos, o cualquier otra combinación. En lugar de una función de similitud, aquí se define una función de coocurrencia: dos vértices estarán conectados si las palabras que representan ocurren en el texto dentro de una ventana de n palabras, que suele tomar un valor entre dos y diez. Los vínculos de coocurrencia demuestran una relación entre los elementos sintácticos que sirve como indicador de la cohesión del texto. Es por ello que son útiles para indicar la relevancia de las palabras principales. Con el grafo no ponderado y no dirigido construido, el valor de cada vértice se establece en uno, y se aplica el PageRank. Una vez asignados los puntajes, se toman los k vértices más relevantes, donde k suele ser un tercio de la cantidad de vértices totales. Finalmente, las palabras preseleccionadas se marcan en el texto. En caso que estas palabras formen secuencias adyacentes, se las une en un “concepto clave”. Por ejemplo, si se preseleccionan las palabras “Puerto” y “Rico”, y en el texto figuran juntas, se las combina en “Puerto Rico”. \subsection{Métodos de evaluación} Considerando que parte del alcance planteado incluye la propuesta de mejoras al método, se trabajó con el paquete ROUGE, desarrollado en lenguaje Perl. Se usó un proyecto de Python que funciona como wrapper para adaptar los textos de entrada (pyROUGE) y ejecutar automáticamente la evaluación. Se implementó, además, una capa de abstracción adicional para acotar la cantidad de opciones que se manejan, y para automatizar la corrida de múltiples textos de una base de datos. Al ser un paquete de evaluación muy completo y versátil, ROUGE provee una gran cantidad de parámetros. Se configuraron sus opciones como se usan en las conferencias de DUC, calculando ROUGE-1, ROUGE-2 y ROUGE-SU4 en un intervalo de confianza del 95\% y aplicando un método de stemming \cite{duc2007}. Una de las restricciones que posee este método de evaluación es que requiere de resúmenes hechos por seres humanos para tomar como referencia, lo que implica un esfuerzo humano considerable si se quiere probar el rendimiento del método con varios textos. Se utilizó el corpus de DUC edición 2002, por ser en el que se desarrollaron las pruebas del paper original de TextRank. Todos los resultados obtenidos se compararon contra un sistema de referencia (\textit{baseline}) que genera resúmenes que consisten de las primeras 100 palabras de cada texto \cite{baselines}. Se configuró un ambiente de integración continua para permitir la ejecución de evaluaciones automatizadas y que además verifica la calidad del código programado. \section{Mejoras} En esta sección se analizan las propuestas de mejora que arrojaron mejores resultados. Los métodos listados a continuación consisten en redefinir la similaridad entre las oraciones. El listado total de ensayos realizados y sus puntajes se puede consultar en el Anexo I. \subsection{Propuestas} \subsubsection{Subcadena común} Dadas dos frases, el problema de la subcadena común más larga consiste en identificar la secuencia de caracteres de mayor extensión presente en ambas. Por ejemplo, entre “la cocina verde” y “una cocina vale más si es verde”, la secuencia más larga es “a cocina v” \cite{gusfield}. La propuesta consiste en modificar la función de distancia de la implementación original y reemplazarla por el largo de la subcadena común más larga. En el ejemplo anterior, la similitud sería de 10, dado que es el largo de la secuencia “a cocina v”. \subsubsection{Similitud coseno} La similitud coseno es una medida del parecido entre dos vectores en un espacio que posee un producto interior, y resulta de evaluar el valor del coseno del ángulo comprendido entre ellos. Para poder hacer uso de esta propiedad se utiliza el modelo vectorial, modelando a los documentos como vectores. Este modelo algebraico es utilizado para representar documentos en lenguaje natural de una manera formal, y tiene la característica de que favorece la dirección a la cuál apuntan los documentos independientemente de su longitud. Consecuentemente, textos que hablan de los mismos temas en distinta cantidad de palabras pueden tener una gran similitud con esta métrica \cite{singhal}. La propuesta se basa en aplicar este modelo para tratar a cada oración del texto como un vector n-dimensional (siendo n la cantidad de palabras distintas presentes en el documento), y luego compararlas utilizando la similitud coseno. Las componentes de cada vector estarán compuestas por el resultado de aplicar la función “frecuencia de término -- frecuencia inversa de documento” (TF--IDF de sus siglas en inglés) a cada palabra de la oración representada. Tomando como ejemplo el documento “Esta bien. Todo bien.”, las oraciones quedarían modeladas como vectores de la siguiente forma: \begin{center} \begin{Verbatim}[xleftmargin=3em] v1=[ TFIDF("Esta") , TFIDF("bien") , 0 ] v2=[ 0 , TFIDF("bien") , TFIDF("Todo") ] \end{Verbatim} \end{center} Dado que la imagen de la función TF--IDF está contenida en el intervalo [0,1], todos los vectores quedan conformados por entradas no negativas, haciendo que ninguna similitud sea menor a cero. \subsubsection{BM25} BM25 / Okapi BM25 es una función de ranking utilizada como estado de arte para tareas de Recuperación de Información. Esta función es una variación del modelo TF--IDF usando un modelo probabilístico \cite{robertson}. \begin{definicion} Dadas dos oraciones R, S, BM25 se define como: \begin{equation} BM25(D,Q) = \sum_{i=1}^{n} IDF(q_i) \cdot \frac{f(q_i, D) \cdot (k_1 + 1)}{f(q_i, D) + k_1 \cdot (1 - b + b \cdot \frac{|D|}{avgDL})} \end{equation} donde $k$ y $b$ son parámetros establecidos a $1,2$ y $0,75$, respectivamente. $avgDL$ es el largo promedio de las oraciones en el texto. \end{definicion} Esta función define que el valor de aquellas palabras que aparecen en más de la mitad de los textos resulte negativo. Dado que este comportamiento puede traer problemas en las fases posteriores del algoritmo, se aplica la siguiente fórmula correctiva: \begin{equation} IDF(q_i) = \begin{cases} log(N - n(q_i) + 0.5) - log(n(q_i) + 0.5) & \text{si } n(q_i) > N/2\\ \varepsilon \cdot avgIDF & \text{si } n(q_i) \leq N/2\\ \end{cases} \end{equation} donde $\varepsilon$ ronda entre 0,5 y 0,30 y avgIDF es el idf promedio para todos los términos. También se ensayaron otras alternativas como establecer $\varepsilon = 0$ y usar modificaciones más simples a la fórmula clásica de IDF. También se experimentó con una variante de este método, conocida como BM25+, que repara deficiencias de BM25 relacionadas a la frecuencia de término y a la penalización a documentos largos frente a documentos cortos irrelevantes \cite{lv}. \subsection{Ejemplo} En la Figura \ref{fig:text} se muestra un documento de ejemplo del conjunto de datos de la conferencia DUC del año 2002. El resumen generado por el algoritmo de TextRank usando la función BM25 como similitud superó al original por un 30\% en la métrica de ROUGE-1 y por un 54\% en ROUGE-2. Se muestran ambos resúmenes en las figuras \ref{fig:bm25} y \ref{fig:textrank}. \begin{figure}[h!] \caption{Documento de ejemplo del conjunto de datos de DUC 2002.} \label{fig:text} \raisedtext{{\scriptsize \parbox{\linewidth}{ \begin{enumerate} \itemsep0em \item Oscar Trophy Has No Copyright \item By \item Associated Press Writer \item LOS ANGELES (AP) \item The Oscar statuette, one of the most recognizable images in the entertainment world, has no copyright protection, a federal judge has ruled. \item The small Academy Award statue is part of the public domain, U.S. District Court Judge Laughlin Waters said in a ruling released Thursday. \item The decision was a setback for the Academy of Motion Picture Arts and Sciences, which had sued a Chicago-based manufacturer of an employee-incentive trophy similar to the Oscar. \item Academy President said the ruling ``comes as a shock to me''. \item The academy said it would appeal. \item The academy claimed that the Star Award, the trophy look-alike made by Creative House Promotions, violated copyright laws, diluted the academy's trademark and represented unfair competition. \item The Star Award depicted a naked, muscular male much like the Oscar, just two inches shorter and holding a star instead of a sword. \item It had a gold finish similar to the Oscar, and stood on a circular gold cap mounted on a cylindrical base. \item Although Waters acknowledged that the Creative House trophy is ``very similar'' to the Oscar, he rejected all of the academy's legal claims, saying that the statuette became part of the public domain prior to Jan. 1, 1978, the effective date of the Copyright Act of 1976. \item The film industry's top award was distributed from the first ceremony in 1929 until about 1941 without the ``c'' mark indicating a copyright. \item The judge ruled, too, that the Oscar not only was an honorary award but also was used to promote the film industry, and therefore did not have a ``limited purpose,'' as required by the copyright act. \item The academy noted that the ruling applied only to the statuette and that the academy's trademark and service marks for the Oscar and the Academy Award names and symbols are unaffected. \item ``We are surprised that the court would base its decision on events which occurred half a century ago, overlooking the meticulous fashion in which the Academy has protected the integrity of its institutional symbol for decades,'' Malden said in a statement. \end{enumerate} }}} \end{figure} \begin{figure}[h!] \caption{Resumen generado por TextRank usando similitud BM25.} \label{fig:bm25} \raisedtext{{\scriptsize \parbox{\linewidth}{ The small Academy Award statue is part of the public domain, U.S. District Court Judge Laughlin Waters said in a ruling released Thursday. The academy claimed that the Star Award, the trophy look-alike made by Creative House Promotions, violated copyright laws, diluted the academy's trademark and represented unfair competition. Although Waters acknowledged that the Creative House trophy is ``very similar'' to the Oscar, he rejected all of the academy's legal claims, saying that the statuette became part of the public domain prior to Jan. 1, 1978, the effective date of the Copyright Act of 1976. }}} \end{figure} \begin{figure}[h!] \caption{Resumen generado por la versión original de TextRank.} \label{fig:textrank} \raisedtext{{\scriptsize \parbox{\linewidth}{ Although Waters acknowledged that the Creative House trophy is ``very similar'' to the Oscar, he rejected all of the academy's legal claims, saying that the statuette became part of the public domain prior to Jan. 1, 1978, the effective date of the Copyright Act of 1976. The judge ruled, too, that the Oscar not only was an honorary award but also was used to promote the film industry, and therefore did not have a ``limited purpose,'' as required by the copyright act. The academy noted that the ruling applied only to the statuette and that the academy's trademark and service marks for the Oscar and the Academy Award names and symbols are unaffected. }}} \end{figure} \subsection{Resultados} Los mejores resultados se obtuvieron usando BM25 y BM25+. El incremento más alto se logró al reemplazar los valores negativos por la constante $\varepsilon$ = 0,25 en la ecuación 3, dando una mejora total de 2,92\% para BM25 y 2,60\% para BM25+. En el Cuadro 1 se detallan las alternativas ensayadas. \begin{table} \caption{Resultados de las propuestas.} \begin{center} \begin{tabular}{l*{5}{c}r} \hline \rule{0pt}{12pt} Método & ROUGE-1 & ROUGE-2 & ROUGE-SU4 & Mejora \\[2pt] \hline\rule{0pt}{12pt}\mbox{}\par\nobreak BM25 ($\varepsilon$ = 0,25) & 0,4042 & 0,1831 & 0,2018 & 2,92\% \\ BM25+ ($\varepsilon$ = 0,25) & 0,404 & 0,1818 & 0,2008 & 2,60\% \\ Similitud coseno & 0,4108 & 0,177 & 0,1984 & 2,54\% \\ BM25+ (IDF = log($N$/$N_i$)) & 0,4022 & 0,1805 & 0,1997 & 2,05\% \\ BM25 (IDF = log($N$/$N_i$)) & 0,4012 & 0,1808 & 0,1998 & 1,97\% \\ Subcadena común & 0,402 & 0,1783 & 0,1971 & 1,40\% \\ BM25+ ($\varepsilon$ = 0) & 0,3992 & 0,1803 & 0,1976 & 1,36\% \\ BM25 ($\varepsilon$ = 0) & 0,3991 & 0,1778 & 0,1966 & 0,89\% \\ \textbf{TextRank} & \textbf{0,3983} & \textbf{0,1762} & \textbf{0,1948} & \textbf{--}\\ BM25 & 0,3916 & 0,1725 & 0,1906 & -1,57\% \\ BM25+ & 0,3903 & 0,1711 & 0,1894 & -2,07\% \\ Referencia DUC & 0,39 & 0,1689 & 0,186 & -2,84\% \\ [2pt] \hline \end{tabular} \end{center} \end{table} Los tiempos de ejecución también se superaron. Se pudieron procesar los 567 documentos de la base de datos de DUC2002 utilizando 84\% del tiempo requerido por la versión original. El resultado de la métrica de similitud coseno también fue satisfactoria, presentando una mejora de 2,54\% por sobre el método original. A su vez, los métodos de secuencias de máxima longitud también mostraron una mejora considerable: cerca del 1,40\% por sobre TextRank. En la Figura 1 se comparan las distintas técnicas. \begin{figure}[h!] \centering \includegraphics[width=1\textwidth]{rouge-scores.pdf} \caption{Comparación de métricas.} \end{figure} \section{Integración a Gensim} La integración del módulo a Gensim\footnote{Pull request: \url{https://github.com/piskvorky/gensim/pull/324}} se desarrolló a partir de la última versión a la fecha, la 0.11.1-1. Se agregaron a la biblioteca las dos funcionalidades descriptas en el desarrollo del presente trabajo: la extracción de palabras claves y la generación automática de resúmenes extractivos. En la integración se incluyen tres casos de uso: \begin{itemize} \item Generación de resúmenes automáticos dado un texto. En este caso la función requiere una cadena de texto con el documento a resumir, y devuelve el texto resumido. \item Selección de documentos más importantes dado un corpus. Este caso es similar al anterior, con la diferencia de que se recibe un listado de documentos (que pueden ser oraciones de un texto) y se devuelve un listado de aquéllos más importantes. \item Obtención de palabras claves de un texto. En este caso se devuelve un listado de palabras clave dada una cadena de texto. \end{itemize} \section{Conclusiones} En este trabajo se analizaron variantes al algoritmo de TextRank. A partir del mismo se propusieron e implementaron optimizaciones cuyos resultados fueron significativos: se obtuvo una mejoría del 2,92\% por sobre el método original. Este número es notable si se tiene en cuenta que TextRank por sí solo performa 2,84\% por sobre el estándar de comparación. Las evaluaciones fueron hechas con los mismos conjuntos de datos y métricas utilizados en competencias internacionales, validando así las mediciones. Finalmente, la adaptación a la biblioteca de lenguaje natural Gensim también resultó sencilla gracias a las facilidades que provee su interfaz de programación y su gran cantidad de métodos. Queda para próximos trabajos explorar alternativas para la extracción de palabras claves, y el ensayo de distintos métodos algebraicos, haciendo uso del modelo del espacio vectorial. \newpage \section{Anexo I: modificaciones} Además de las mejoras descriptas en la sección “Mejoras”, también se ensayaron otras ideas cuyos resultados no fueron tan destacables como los mencionados. Como se explicó anteriormente, las modificaciones apuntan a cambiar el criterio de similaridad entre oraciones. Algunas de las que se tuvieron en cuenta tienen un concepto similar a la de subcadena común, que otorga puntajes de acuerdo a la cantidad de unidades léxicas en orden que estén presentes en las dos oraciones. Estas métricas utilizan las técnicas usadas por el paquete ROUGE: los métodos de subsecuencia de palabras común, subsecuencia de palabras común con peso, bigramas comunes y SU4 son idénticos a los usados por los métodos ROUGE-L, ROUGE-W, ROUGE-2 y ROUGE-SU4 respectivamente. Se implementaron también métricas de distancias de Levenshtein \cite{levenshtein}, un método que pondera con un valor más alto a las palabras clave del texto, y una lógica de envejecimiento que asigna puntaje adicional a aquéllas palabras que se mencionaron más recientemente. Finalmente, se experimentó también reemplazando el algoritmo de PageRank (que utiliza como solución el autovector asociado al valor asociado más grande de la matriz de similaridad) por el de HITS (que utiliza el vector singular asociado al valor singular más grande de la matriz) \cite{kleinberg}; y también se hicieron pruebas aproximando la solución al reducir el rango de la matriz con DVS. \subsection{Resultados} \begin{tabular}{l*{5}{c}r} \hline \rule{0pt}{12pt} Método & ROUGE-1 & ROUGE-2 & ROUGE-SU4 & Mejora \\[2pt] \hline\rule{0pt}{12pt}\mbox{}\par\nobreak BM25 ($\varepsilon$ = 0,25) & 0,4042 & 0,1831 & 0,2018 & 2,92\% \\ BM25+ ($\varepsilon$ = 0,25) & 0,404 & 0,1818 & 0,2008 & 2,60\% \\ Similitud coseno & 0,4108 & 0,177 & 0,1984 & 2,54\% \\ BM25+ (IDF = log($N$/$N_i$)) & 0,4022 & 0,1805 & 0,1997 & 2,05\% \\ BM25 (IDF = log($N$/$N_i$)) & 0,4012 & 0,1808 & 0,1998 & 1,97\% \\ Subcadena común & 0,402 & 0,1783 & 0,1971 & 1,40\% \\ Subsec. palabras común & 0,4032 & 0,1773 & 0,1968 & 1,38\% \\ Subsec. palabras común con peso & 0,4032 & 0,1773 & 0,1968 & 1,38\% \\ BM25+ ($\varepsilon$ = 0) & 0,3992 & 0,1803 & 0,1976 & 1,36\% \\ BM25 ($\varepsilon$ = 0) & 0,3991 & 0,1778 & 0,1966 & 0,89\% \\ Distancia temporal & 0,3992 & 0,1762 & 0,195 & 0,48\% \\ Levenshtein & 0,3961 & 0,1781 & 0,1955 & 0,38\% \\ Subsecuencia común & 0,396 & 0,1775 & 0,1949 & 0,22\% \\ \textbf{TextRank} & \textbf{0,3983} & \textbf{0,1762} & \textbf{0,1948} & \textbf{--} \\ SU4 & 0,3955 & 0,1732 & 0,1924 & -0,73\% \\ Prioridad keywords & 0,3952 & 0,1721 & 0,1915 & -1,03\% \\ Peso de conceptos & 0,3953 & 0,1713 & 0,1914 & -1,14\% \\ Distancia temporal invertida & 0,3913 & 0,1725 & 0,1912 & -1,53\% \\ BM25 & 0,3916 & 0,1725 & 0,1906 & -1,57\% \\ HITS & 0,3914 & 0,1699 & 0,1898 & -2,03\% \\ BM25+ & 0,3903 & 0,1711 & 0,1894 & -2,07\% \\ HITS + DVS & 0,3912 & 0,1686 & 0,189 & -2,33\% \\ Referencia DUC & 0,39 & 0,1689 & 0,186 & -2,84\% \\ Bigramas comunes & 0,393 & 0,1633 & 0,1842 & -3,42\% \\ Weighted LCSubstring & 0,3755 & 0,1509 & 0,1729 & -8,79\% \\ Levenshtein normalizado & 0,348 & 0,1374 & 0,1565 & -16,28\% \\ \end{tabular} \newpage \bibliography{informe}{} \bibliographystyle{babunsrt} \end{document} %%% Local Variables: %%% mode: latex %%% TeX-master: "DCI_manual_2017-2018_rev.pdf" %%% End: \documentclass[a4paper, 12pt]{article} % \usepackage[showframe]{geometry} \usepackage[utf8]{inputenc} \usepackage{fontspec} \usepackage{fancyhdr} \renewcommand\familydefault{\sfdefault} \usepackage{anyfontsize} \usepackage{mathtools} \usepackage{pdfpages} \usepackage{titlesec} \usepackage{amssymb} \usepackage{xcolor} \definecolor{dred}{HTML}{95043E} \usepackage{enumitem} % \usepackage{sectsty} % \allsectionsfont{\raggedleft} \usepackage{tikz} \usetikzlibrary{automata, positioning, arrows} \title{Space Invaders} \date{\today} \pagestyle{fancy} \fancyhf{} \rhead{Microprocesadores} \renewcommand{\headrulewidth}{0pt} \fancyfoot[RE,RO]{Pag. \thepage} \renewcommand{\contentsname}{Índice} \titleformat{\section} {\normalfont\Large}{\thesection}{3em}{}[\vskip-10pt{\makebox[\linewidth][l]{\rule{\textwidth}{1.5pt}}}] \setcounter{secnumdepth}{0} \begin{document} \begin{titlepage} % Background image of the title page \tikz[remember picture, overlay] \node[opacity=0.3,inner sep=0pt] at (current page.center){\includegraphics[width=\paperwidth,height=\paperheight]{background1}}; \vspace*{0.5cm} \noindent\rule{19cm}{0.3em} \vspace*{1.5cm} \noindent{\fontsize{50}{60}\selectfont Space Invaders}\\ \vspace*{1cm} \noindent{\fontsize{15}{15}\selectfont Diseño de circuitos integrados}\\ Grupo 21\\ \vspace*{2cm} \noindent \\ \noindent \\ \noindent \\ \thispagestyle{empty} \end{titlepage} \newpage \tableofcontents \thispagestyle{empty} \newpage \section{Introducción} En este documento se detallará el proceso realizado para crear el programa necesario para poder jugar al famoso juego \textit{Space Invaders} en una FPGA. \section{Control del juego} \begin{figure}[ht] \centering \begin{tikzpicture}[->, % makes the edges directed >=stealth, % makes the arrow heads bold node distance=3.5cm, % specifies the minimum distance between two nodes. Change if necessary. every state/.style={thick, fill=gray!30}, % sets the properties for each ’state’ node initial text=$ $, % sets the text that appears on the start arrow] ] \node[state, accepting] (1) {Inicio}; \node[state, above of=1] (2) {Test}; \node[state, below=1cm and 1cm of 1] (3) {Jugando}; \node[state, left=2cm and 3cm of 3] (4) {Ganado}; \node[state, right=2cm and 3cm of 3] (5) {Perdido}; \draw[->,line width=1pt] (1) edge[right, bend right] node{Test} (2) (2) edge[left, bend right] node{$\overline{\text{Test}}$} (1) (1) edge[right] node{Inicio} (3) (3) edge[below] node{num\_inv = 0} (4) (3) edge[below] node{y\_inv = 14} (5) (4) edge[left, bend left] node{Inicio} (1) (5) edge[right, bend right] node{Inicio} (1) (3) edge[loop below] node{$num\_inv > 0 \wedge y\_inv < 14$}(3); (5) edge[loop right] (5); \end{tikzpicture} \caption{Maquina de estados del juego} \label{fig:statemachine} \end{figure} \begin{itemize} \item \textcolor{dred}{Test}: En este estado se muestra un patron de ajedrez de 20$\times$15. Se entra en el estado cuando se pulsa en boton \textbf{Test} y se sale cuando se suelta el botón \end{itemize} \section{Implementación} \section{Recursos utilizados} \end{document} newmanne/logparser \documentclass[openany]{book} \usepackage{hyperlatex} \usepackage{hyperref} \usepackage{graphicx} \usepackage{float} \texonly{ % put borders around figures \floatstyle{boxed} \restylefloat{figure} % set dimensions of page/text \parindent=0in \parskip=10pt \oddsidemargin=-0.3in \evensidemargin=-0.3in \textwidth=7in \textheight=8.0in % redefine certain commands \renewcommand\code[1]{\path{#1}} \renewcommand\xlink[2]{\href{#2}{#1}} } \htmlonly{ \newcommand\code[1]{{\tt #1}} \newcommand\url[1]{\xlink{#1}{#1}} } \newcommand\bh{{\bf h}} \newcommand\bi{{\bf i}} \newcommand\bo{{\bf o}} \newcommand\bt{{\bf t}} \newcommand\bm{{\bf m}} \newcommand\be{{\bf e}} \newcommand\bg{{\bf f}} \newcommand\bb{{\bf b}} \newcommand\bl{{\bf l}} \newcommand\bp{{\bf p}} \newcommand\br{{\bf r}} \newcommand\bw{{\bf w}} \setcounter{htmldepth}{2} \setcounter{htmlautomenu}{3} \htmltitle{Chord} \htmladdress{} \title{Chord: A Versatile Platform for Program Analysis} \author{} \date{\today} \begin{document} \maketitle \input{whatis_chord} \input{getting_started} \input{example} \input{architecture} \input{properties} \input{predefined_analyses} \input{analyses} \input{dynamic_analysis} \input{datalog_analysis} \input{computing_scope} \input{program_representation} \input{acknowledgments} \end{document} %% bare_adv.tex %% V1.4b %% 2015/08/26 %% by %% See: %% http://www.michaelshell.org/ %% for current contact information. %% %% This is a skeleton file demonstrating the advanced use of IEEEtran.cls %% (requires IEEEtran.cls version 1.8b or later) with an IEEE Computer %% Society journal paper. %% %% Support sites: %% http://www.michaelshell.org/tex/ieeetran/ %% http://www.ctan.org/pkg/ieeetran %% and %% http://www.ieee.org/ %%************************************************************************* %% Legal Notice: %% This code is offered as-is without any warranty either expressed or %% implied; without even the implied warranty of MERCHANTABILITY or %% FITNESS FOR A PARTICULAR PURPOSE! %% User assumes all risk. %% In no event shall the IEEE or any contributor to this code be liable for %% any damages or losses, including, but not limited to, incidental, %% consequential, or any other damages, resulting from the use or misuse %% of any information contained here. %% %% All comments are the opinions of their respective authors and are not %% necessarily endorsed by the IEEE. %% %% This work is distributed under the LaTeX Project Public License (LPPL) %% ( http://www.latex-project.org/ ) version 1.3, and may be freely used, %% distributed and modified. A copy of the LPPL, version 1.3, is included %% in the base LaTeX documentation of all distributions of LaTeX released %% 2003/12/01 or later. %% Retain all contribution notices and credits. %% ** Modified files should be clearly indicated as such, including ** %% ** renaming them and changing author support contact information. ** %%************************************************************************* % *** Authors should verify (and, if needed, correct) their LaTeX system *** % *** with the testflow diagnostic prior to trusting their LaTeX platform *** % *** with production work. The IEEE's font choices and paper sizes can *** % *** trigger bugs that do not appear when using other class files. *** *** % The testflow support page is at: % http://www.michaelshell.org/tex/testflow/ % IEEEtran V1.7 and later provides for these CLASSINPUT macros to allow the % user to reprogram some IEEEtran.cls defaults if needed. These settings % override the internal defaults of IEEEtran.cls regardless of which class % options are used. Do not use these unless you have good reason to do so as % they can result in nonIEEE compliant documents. User beware. ;) % %\newcommand{\CLASSINPUTbaselinestretch}{1.0} % baselinestretch %\newcommand{\CLASSINPUTinnersidemargin}{1in} % inner side margin %\newcommand{\CLASSINPUToutersidemargin}{1in} % outer side margin %\newcommand{\CLASSINPUTtoptextmargin}{1in} % top text margin %\newcommand{\CLASSINPUTbottomtextmargin}{1in}% bottom text margin % \documentclass[10pt,journal,compsoc]{IEEEtran} \usepackage{amsmath}% \DeclareMathOperator*{\argmin}{arg\,min} % thin space, limits underneath in displays \usepackage{xeCJK}% 调用 xeCJK 宏包 \setCJKmainfont{SimSun}% 设置 CJK 主字体为 SimSun （宋体） \newtheorem{prop}{Proposition} % If IEEEtran.cls has not been installed into the LaTeX system files, % manually specify the path to it like: % \documentclass[10pt,journal,compsoc]{../sty/IEEEtran} % For Computer Society journals, IEEEtran defaults to the use of % Palatino/Palladio as is done in IEEE Computer Society journals. % To go back to Times Roman, you can use this code: %\renewcommand{\rmdefault}{ptm}\selectfont % Some very useful LaTeX packages include: % (uncomment the ones you want to load) % *** MISC UTILITY PACKAGES *** % %\usepackage{ifpdf} % 's ifpdf.sty is very useful if you need conditional % compilation based on whether the output is pdf or dvi. % usage: % \ifpdf % % pdf code % \else % % dvi code % \fi % The latest version of ifpdf.sty can be obtained from: % http://www.ctan.org/pkg/ifpdf % Also, note that IEEEtran.cls V1.7 and later provides a builtin % \ifCLASSINFOpdf conditional that works the same way. % When switching from latex to pdflatex and vice-versa, the compiler may % have to be run twice to clear warning/error messages. % *** CITATION PACKAGES *** % \ifCLASSOPTIONcompsoc % The IEEE Computer Society needs nocompress option % requires cite.sty v4.0 or later (November 2003) \usepackage[nocompress]{cite} \else % normal IEEE \usepackage{cite} \fi % cite.sty was written by % V1.6 and later of IEEEtran pre-defines the format of the cite.sty package % \cite{} output to follow that of the IEEE. Loading the cite package will % result in citation numbers being automatically sorted and properly % "compressed/ranged". e.g., [1], [9], [2], [7], [5], [6] without using % cite.sty will become [1], [2], [5]--[7], [9] using cite.sty. cite.sty's % \cite will automatically add leading space, if needed. Use cite.sty's % noadjust option (cite.sty V3.8 and later) if you want to turn this off % such as if a citation ever needs to be enclosed in parenthesis. % cite.sty is already installed on most LaTeX systems. Be sure and use % version 5.0 (2009-03-20) and later if using hyperref.sty. % The latest version can be obtained at: % http://www.ctan.org/pkg/cite % The documentation is contained in the cite.sty file itself. % % Note that some packages require special options to format as the Computer % Society requires. In particular, Computer Society papers do not use % compressed citation ranges as is done in typical IEEE papers % (e.g., [1]-[4]). Instead, they list every citation separately in order % (e.g., [1], [2], [3], [4]). To get the latter we need to load the cite % package with the nocompress option which is supported by cite.sty v4.0 % and later. % *** GRAPHICS RELATED PACKAGES *** % \ifCLASSINFOpdf % \usepackage[pdftex]{graphicx} % declare the path(s) where your graphic files are % \graphicspath{{../pdf/}{../jpeg/}} % and their extensions so you won't have to specify these with % every instance of \includegraphics % \DeclareGraphicsExtensions{.pdf,.jpeg,.png} \else % or other class option (dvipsone, dvipdf, if not using dvips). graphicx % will default to the driver specified in the system graphics.cfg if no % driver is specified. % \usepackage[dvips]{graphicx} % declare the path(s) where your graphic files are % \graphicspath{{../eps/}} % and their extensions so you won't have to specify these with % every instance of \includegraphics % \DeclareGraphicsExtensions{.eps} \fi % graphicx was written by and . It is % required if you want graphics, photos, etc. graphicx.sty is already % installed on most LaTeX systems. The latest version and documentation % can be obtained at: % http://www.ctan.org/pkg/graphicx % Another good source of documentation is "Using Imported Graphics in % LaTeX2e" by Keith Reckdahl which can be found at: % http://www.ctan.org/pkg/epslatex % % latex, and pdflatex in dvi mode, support graphics in encapsulated % postscript (.eps) format. pdflatex in pdf mode supports graphics % in .pdf, .jpeg, .png and .mps (metapost) formats. Users should ensure % that all non-photo figures use a vector format (.eps, .pdf, .mps) and % not a bitmapped formats (.jpeg, .png). The IEEE frowns on bitmapped formats % which can result in "jaggedy"/blurry rendering of lines and letters as % well as large increases in file sizes. % % You can find documentation about the pdfTeX application at: % http://www.tug.org/applications/pdftex % *** MATH PACKAGES *** % %\usepackage{amsmath} % A popular package from the American Mathematical Society that provides % many useful and powerful commands for dealing with mathematics. % % Note that the amsmath package sets \interdisplaylinepenalty to 10000 % thus preventing page breaks from occurring within multiline equations. Use: %\interdisplaylinepenalty=2500 % after loading amsmath to restore such page breaks as IEEEtran.cls normally % does. amsmath.sty is already installed on most LaTeX systems. The latest % version and documentation can be obtained at: % http://www.ctan.org/pkg/amsmath % *** SPECIALIZED LIST PACKAGES *** %\usepackage{acronym} % acronym.sty was written by . This package provides tools for % managing documents with large numbers of acronyms. (You don't *have* to % use this package - unless you have a lot of acronyms, you may feel that % such package management of them is bit of an overkill.) % Do note that the acronym environment (which lists acronyms) will have a % problem when used under IEEEtran.cls because acronym.sty relies on the % description list environment - which IEEEtran.cls has customized for % producing IEEE style lists. A workaround is to declared the longest % label width via the IEEEtran.cls \IEEEiedlistdecl global control: % % \renewcommand{\IEEEiedlistdecl}{\IEEEsetlabelwidth{SONET}} % \begin{acronym} % % \end{acronym} % \renewcommand{\IEEEiedlistdecl}{\relax}% remember to reset \IEEEiedlistdecl % % instead of using the acronym environment's optional argument. % The latest version and documentation can be obtained at: % http://www.ctan.org/pkg/acronym %\usepackage{algorithmic} % algorithmic.sty was written by and . % This package provides an algorithmic environment fo describing algorithms. % You can use the algorithmic environment in-text or within a figure % environment to provide for a floating algorithm. Do NOT use the algorithm % floating environment provided by algorithm.sty (by the same authors) or % algorithm2e.sty (by ) as the IEEE does not use dedicated % algorithm float types and packages that provide these will not provide % correct IEEE style captions. The latest version and documentation of % algorithmic.sty can be obtained at: % http://www.ctan.org/pkg/algorithms % Also of interest may be the (relatively newer and more customizable) % algorithmicx.sty package by : % http://www.ctan.org/pkg/algorithmicx % *** ALIGNMENT PACKAGES *** % %\usepackage{array} % and 's array.sty patches and improves % the standard LaTeX2e array and tabular environments to provide better % appearance and additional user controls. As the default LaTeX2e table % generation code is lacking to the point of almost being broken with % respect to the quality of the end results, all users are strongly % advised to use an enhanced (at the very least that provided by array.sty) % set of table tools. array.sty is already installed on most systems. The % latest version and documentation can be obtained at: % http://www.ctan.org/pkg/array %\usepackage{mdwmath} %\usepackage{mdwtab} % Also highly recommended is 's extremely powerful MDW tools, % especially mdwmath.sty and mdwtab.sty which are used to format equations % and tables, respectively. The MDWtools set is already installed on most % LaTeX systems. The lastest version and documentation is available at: % http://www.ctan.org/pkg/mdwtools % IEEEtran contains the IEEEeqnarray family of commands that can be used to % generate multiline equations as well as matrices, tables, etc., of high % quality. %\usepackage{eqparbox} % Also of notable interest is 's eqparbox package for creating % (automatically sized) equal width boxes - aka "natural width parboxes". % Available at: % http://www.ctan.org/pkg/eqparbox % *** SUBFIGURE PACKAGES *** %\ifCLASSOPTIONcompsoc % \usepackage[caption=false,font=footnotesize,labelfont=sf,textfont=sf]{subfig} %\else % \usepackage[caption=false,font=footnotesize]{subfig} %\fi % subfig.sty, written by , is the modern replacement % for subfigure.sty, the latter of which is no longer maintained and is % incompatible with some LaTeX packages including fixltx2e. However, % subfig.sty requires and automatically loads Axel Sommerfeldt's caption.sty % which will override IEEEtran.cls' handling of captions and this will result % in non-IEEE style figure/table captions. To prevent this problem, be sure % and invoke subfig.sty's "caption=false" package option (available since % subfig.sty version 1.3, 2005/06/28) as this is will preserve IEEEtran.cls % handling of captions. % Note that the Computer Society format requires a sans serif font rather % than the serif font used in traditional IEEE formatting and thus the need % to invoke different subfig.sty package options depending on whether % compsoc mode has been enabled. % % The latest version and documentation of subfig.sty can be obtained at: % http://www.ctan.org/pkg/subfig % *** FLOAT PACKAGES *** % %\usepackage{fixltx2e} % fixltx2e, the successor to the earlier fix2col.sty, was written by % and . This package corrects a few problems % in the LaTeX2e kernel, the most notable of which is that in current % LaTeX2e releases, the ordering of single and double column floats is not % guaranteed to be preserved. Thus, an unpatched LaTeX2e can allow a % single column figure to be placed prior to an earlier double column % figure. % Be aware that LaTeX2e kernels dated 2015 and later have fixltx2e.sty's % corrections already built into the system in which case a warning will % be issued if an attempt is made to load fixltx2e.sty as it is no longer % needed. % The latest version and documentation can be found at: % http://www.ctan.org/pkg/fixltx2e %\usepackage{stfloats} % stfloats.sty was written by . This package gives LaTeX2e % the ability to do double column floats at the bottom of the page as well % as the top. (e.g., "\begin{figure*}[!b]" is not normally possible in % LaTeX2e). It also provides a command: %\fnbelowfloat % to enable the placement of footnotes below bottom floats (the standard % LaTeX2e kernel puts them above bottom floats). This is an invasive package % which rewrites many portions of the LaTeX2e float routines. It may not work % with other packages that modify the LaTeX2e float routines. The latest % version and documentation can be obtained at: % http://www.ctan.org/pkg/stfloats % Do not use the stfloats baselinefloat ability as the IEEE does not allow % \baselineskip to stretch. Authors submitting work to the IEEE should note % that the IEEE rarely uses double column equations and that authors should try % to avoid such use. Do not be tempted to use the cuted.sty or midfloat.sty % packages (also by ) as the IEEE does not format its papers in % such ways. % Do not attempt to use stfloats with fixltx2e as they are incompatible. % Instead, use Morten Hogholm'a dblfloatfix which combines the features % of both fixltx2e and stfloats: % % \usepackage{dblfloatfix} % The latest version can be found at: % http://www.ctan.org/pkg/dblfloatfix %\ifCLASSOPTIONcaptionsoff % \usepackage[nomarkers]{endfloat} % \let\MYoriglatexcaption\caption % \renewcommand{\caption}[2][\relax]{\MYoriglatexcaption[#2]{#2}} %\fi % endfloat.sty was written by , and % . This package may be useful when used in conjunction with % IEEEtran.cls' captionsoff option. Some IEEE journals/societies require that % submissions have lists of figures/tables at the end of the paper and that % figures/tables without any captions are placed on a page by themselves at % the end of the document. If needed, the draftcls IEEEtran class option or % \CLASSINPUTbaselinestretch interface can be used to increase the line % spacing as well. Be sure and use the nomarkers option of endfloat to % prevent endfloat from "marking" where the figures would have been placed % in the text. The two hack lines of code above are a slight modification of % that suggested by in the endfloat docs (section 8.4.1) to ensure that % the full captions always appear in the list of figures/tables - even if % the user used the short optional argument of \caption[]{}. % IEEE papers do not typically make use of \caption[]'s optional argument, % so this should not be an issue. A similar trick can be used to disable % captions of packages such as subfig.sty that lack options to turn off % the subcaptions: % For subfig.sty: % \let\MYorigsubfloat\subfloat % \renewcommand{\subfloat}[2][\relax]{\MYorigsubfloat[]{#2}} % However, the above trick will not work if both optional arguments of % the \subfloat command are used. Furthermore, there needs to be a % description of each subfigure *somewhere* and endfloat does not add % subfigure captions to its list of figures. Thus, the best approach is to % avoid the use of subfigure captions (many IEEE journals avoid them anyway) % and instead reference/explain all the subfigures within the main caption. % The latest version of endfloat.sty and its documentation can obtained at: % http://www.ctan.org/pkg/endfloat % % The IEEEtran \ifCLASSOPTIONcaptionsoff conditional can also be used % later in the document, say, to conditionally put the References on a % page by themselves. % *** PDF, URL AND HYPERLINK PACKAGES *** % %\usepackage{url} % url.sty was written by . It provides better support for % handling and breaking URLs. url.sty is already installed on most LaTeX % systems. The latest version and documentation can be obtained at: % http://www.ctan.org/pkg/url % Basically, \url{my_url_here}. % NOTE: PDF thumbnail features are not required in IEEE papers % and their use requires extra complexity and work. %\ifCLASSINFOpdf % \usepackage[pdftex]{thumbpdf} %\else % \usepackage[dvips]{thumbpdf} %\fi % thumbpdf.sty and its companion Perl utility were written by . % It allows the user a way to produce PDF documents that contain fancy % thumbnail images of each of the pages (which tools like acrobat reader can % utilize). This is possible even when using dvi->ps->pdf workflow if the % correct thumbpdf driver options are used. thumbpdf.sty incorporates the % file containing the PDF thumbnail information (filename.tpm is used with % dvips, filename.tpt is used with pdftex, where filename is the base name of % your tex document) into the final ps or pdf output document. An external % utility, the thumbpdf *Perl script* is needed to make these .tpm or .tpt % thumbnail files from a .ps or .pdf version of the document (which obviously % does not yet contain pdf thumbnails). Thus, one does a: % % thumbpdf filename.pdf % % to make a filename.tpt, and: % % thumbpdf --mode dvips filename.ps % % to make a filename.tpm which will then be loaded into the document by % thumbpdf.sty the NEXT time the document is compiled (by pdflatex or % latex->dvips->ps2pdf). Users must be careful to regenerate the .tpt and/or % .tpm files if the main document changes and then to recompile the % document to incorporate the revised thumbnails to ensure that thumbnails % match the actual pages. It is easy to forget to do this! % % Unix systems come with a Perl interpreter. However, MS Windows users % will usually have to install a Perl interpreter so that the thumbpdf % script can be run. The Ghostscript PS/PDF interpreter is also required. % See the thumbpdf docs for details. The latest version and documentation % can be obtained at. % http://www.ctan.org/pkg/thumbpdf % NOTE: PDF hyperlink and bookmark features are not required in IEEE % papers and their use requires extra complexity and work. % *** IF USING HYPERREF BE SURE AND CHANGE THE EXAMPLE PDF *** % *** TITLE/SUBJECT/AUTHOR/KEYWORDS INFO BELOW!! *** \newcommand\MYhyperrefoptions{bookmarks=true,bookmarksnumbered=true, pdfpagemode={UseOutlines},plainpages=false,pdfpagelabels=true, colorlinks=true,linkcolor={black},citecolor={black},urlcolor={black}, pdftitle={Bare Demo of IEEEtran.cls for Computer Society Journals},%},%.dvi->.ps->.pdf workflow if the respective packages/scripts are % loaded/invoked with the correct driver options (dvips, etc.). % As most IEEE papers use URLs sparingly (mainly in the references), this % may not be as big an issue as with other publications. % % That said, created his breakurl.sty package which % permits hyperref to easily break URLs even in dvi mode. % Note that breakurl, unlike most other packages, must be loaded % AFTER hyperref. The latest version of breakurl and its documentation can % be obtained at: % http://www.ctan.org/pkg/breakurl % breakurl.sty is not for use under pdflatex pdf mode. % % The advanced features offer by hyperref.sty are not required for IEEE % submission, so users should weigh these features against the added % complexity of use. % The package options above demonstrate how to enable PDF bookmarks % (a type of table of contents viewable in Acrobat Reader) as well as % PDF document information (title, subject, author and keywords) that is % viewable in Acrobat reader's Document_Properties menu. PDF document % information is also used extensively to automate the cataloging of PDF % documents. The above set of options ensures that hyperlinks will not be % colored in the text and thus will not be visible in the printed page, % but will be active on "mouse over". USING COLORS OR OTHER HIGHLIGHTING % OF HYPERLINKS CAN RESULT IN DOCUMENT REJECTION BY THE IEEE, especially if % these appear on the "printed" page. IF IN DOUBT, ASK THE RELEVANT % SUBMISSION EDITOR. You may need to add the option hypertexnames=false if % you used duplicate equation numbers, etc., but this should not be needed % in normal IEEE work. % The latest version of hyperref and its documentation can be obtained at: % http://www.ctan.org/pkg/hyperref % *** Do not adjust lengths that control margins, column widths, etc. *** % *** Do not use packages that alter fonts (such as pslatex). *** % There should be no need to do such things with IEEEtran.cls V1.6 and later. % (Unless specifically asked to do so by the journal or conference you plan % to submit to, of course. ) % correct bad hyphenation here \hyphenation{op-tical net-works semi-conduc-tor} \begin{document} % % paper title % Titles are generally capitalized except for words such as a, an, and, as, % at, but, by, for, in, nor, of, on, or, the, to and up, which are usually % not capitalized unless they are the first or last word of the title. % Linebreaks \\ can be used within to get better formatting as desired. % Do not put math or special symbols in the title. \title{SemiBoost:适用于半监督学习的Boosting算法} % % % author names and IEEE memberships % note positions of commas and nonbreaking spaces ( ~ ) LaTeX will not break % a structure at a ~ so this keeps an author's name from being broken across % two lines. % use \thanks{} to gain access to the first footnote area % a separate \thanks must be used for each paragraph as LaTeX2e's \thanks % was not built to handle multiple paragraphs % % %\IEEEcompsocitemizethanks is a special \thanks that produces the bulleted % lists the Computer Society journals use for "first footnote" author % affiliations. Use \IEEEcompsocthanksitem which works much like \item % for each affiliation group. When not in compsoc mode, % \IEEEcompsocitemizethanks becomes like \thanks and % \IEEEcompsocthanksitem becomes a line break with idention. This % facilitates dual compilation, although admittedly the differences in the % desired content of \author between the different types of papers makes a % one-size-fits-all approach a daunting prospect. For instance, compsoc % journal papers have the author affiliations above the "Manuscript % received ..." text while in non-compsoc journals this is reversed. Sigh. \author{Pavan~Kumar~Mallapragada,~\IEEEmembership{Student~Member, ~IEEE,} Rong~Jin, ~\IEEEmembership{Member,~IEEE,} Anil~K.~Jain,~\IEEEmembership{Fellow,~IEEE,} and~Yi~Liu,~\IEEEmembership{Student~Member,~IEEE}% \IEEEcompsocitemizethanks{\IEEEcompsocthanksitem The authors are with the Department of Computer Science and Engineering, Michigan State University, 3115, Engineering Building, East Lansing, MI 48823.\protect\\ % note need leading \protect in front of \\ to get a newline within \thanks as % \\ is fragile and will error, could use \hfil\break instead. E-mail: \{pavanm, , , . }% <-this % stops a space \thanks{Manuscript received April 19, 2005; revised August 26, 2015.} \thanks{Manuscript received 28 Oct. 2007; revised 3 June 2008; accepted 8 Sept. 2008; published online 16 Sept. 2008.} \thanks{Recommended for acceptance by .} \thanks{For information on obtaining reprints of this article, please send e-mail to , and reference IEEECS Log Number TPAMI-2007-10-0729.} \thanks{Digital Object Identifier no. 10.1109/TPAMI.2008.235.} } % note the % following the last \IEEEmembership and also \thanks - % these prevent an unwanted space from occurring between the last author name % and the end of the author line. i.e., if you had this: % % \author{....lastname \thanks{...} \thanks{...} } % ^------------^------------^----Do not want these spaces! % % a space would be appended to the last name and could cause every name on that % line to be shifted left slightly. This is one of those "LaTeX things". For % instance, "\textbf{A} \textbf{B}" will typeset as "A B" not "AB". To get % "AB" then you have to do: "\textbf{A}\textbf{B}" % \thanks is no different in this regard, so shield the last } of each \thanks % that ends a line with a % and do not let a space in before the next \thanks. % Spaces after \IEEEmembership other than the last one are OK (and needed) as % you are supposed to have spaces between the names. For what it is worth, % this is a minor point as most people would not even notice if the said evil % space somehow managed to creep in. % The paper headers \markboth{Journal of \LaTeX\ Class Files,~Vol.~14, No.~8, August~2015}% {Shell \MakeLowercase{\textit{et al.}}: Bare Advanced Demo of IEEEtran.cls for IEEE Computer Society Journals} % The only time the second header will appear is for the odd numbered pages % after the title page when using the twoside option. % % *** Note that you probably will NOT want to include the author's *** % *** name in the headers of peer review papers. *** % You can use \ifCLASSOPTIONpeerreview for conditional compilation here if % you desire. % The publisher's ID mark at the bottom of the page is less important with % Computer Society journal papers as those publications place the marks % outside of the main text columns and, therefore, unlike regular IEEE % journals, the available text space is not reduced by their presence. % If you want to put a publisher's ID mark on the page you can do it like % this: %\IEEEpubid{0000--0000/00\$00.00~\copyright~2015 IEEE} % or like this to get the Computer Society new two part style. %\IEEEpubid{\makebox[\columnwidth]{\hfill 0000--0000/00/\$00.00~\copyright~2015 IEEE}% %\hspace{\columnsep}\makebox[\columnwidth]{Published by the IEEE Computer Society\hfill}} % Remember, if you use this you must call \IEEEpubidadjcol in the second % column for its text to clear the IEEEpubid mark (Computer Society journal % papers don't need this extra clearance.) % use for special paper notices %\IEEEspecialpapernotice{(Invited Paper)} % for Computer Society papers, we must declare the abstract and index terms % PRIOR to the title within the \IEEEtitleabstractindextext IEEEtran % command as these need to go into the title area created by \maketitle. % As a general rule, do not put math, special symbols or citations % in the abstract or keywords. \IEEEtitleabstractindextext{% \begin{abstract} 半监督学习已经在模式识别和机器学习方面引起了巨大的关注。在这之前的大多数研究关注于设计特殊的算法，用于有效地利用已标记数据与未标记数据的结合。我们的目标是，通过使用可用的未标记样本，提高所有监督学习算法的分类准确度。区分于现有流行的方法，我们把这个难题称作半监督提升问题。基于一个监督学习算法，我们设计了一个偏半监督学习算法，而且利用未标记数据提升了它的性能。当我们需要训练一个监督学习算法，且只有非常有限的已标记数据和未标记数据的时候，这个问题就变得非常重要。我们提出了一个适用半监督学习的boosting框架，称之为SemiBoost。我们所推荐的半监督学习方法的主要优点是：1）在有大量未标记数据的情况下，所有的监督学习算法都可以得到性能上的提升，2）通过迭代的boosting算法可以实现高效率计算，3）在训练分类模型的时候利用流形假设和聚类假设。在有大量未标记数据的情况下，我们对16个不同的数据集和文字分类进行了实证研究，结果表明，几种常用的监督学习算法的性能都得到了提升。我们也证实了，SemiBoost算法可以与当代最先进的半监督学习算法相媲美。 \end{abstract} % Note that keywords are not normally used for peerreview papers. \begin{IEEEkeywords} 机器学习, 半监督学习, 半监督提升, 流形假设, 聚类假设, boosting. \end{IEEEkeywords}} % make the title area \maketitle % To allow for easy dual compilation without having to reenter the % abstract/keywords data, the \IEEEtitleabstractindextext text will % not be used in maketitle, but will appear (i.e., to be "transported") % here as \IEEEdisplaynontitleabstractindextext when compsoc mode % is not selected if conference mode is selected - because compsoc % conference papers position the abstract like regular (non-compsoc) % papers do! \IEEEdisplaynontitleabstractindextext % \IEEEdisplaynontitleabstractindextext has no effect when using % compsoc under a non-conference mode. % For peer review papers, you can put extra information on the cover % page as needed: % \ifCLASSOPTIONpeerreview % \begin{center} \bfseries EDICS Category: 3-BBND \end{center} % \fi % % For peerreview papers, this IEEEtran command inserts a page break and % creates the second title. It will be ignored for other modes. \IEEEpeerreviewmaketitle \ifCLASSOPTIONcompsoc \IEEEraisesectionheading{\section{Introduction}\label{sec:introduction}} \else \section{Introduction} \label{sec:introduction} \fi % Computer Society journal (but not conference!) papers do something unusual % with the very first section heading (almost always called "Introduction"). % They place it ABOVE the main text! IEEEtran.cls does not automatically do % this for you, but you can achieve this effect with the provided % \IEEEraisesectionheading{} command. Note the need to keep any \label that % is to refer to the section immediately after \section in the above as % \IEEEraisesectionheading puts \section within a raised box. % The very first letter is a 2 line initial drop letter followed % by the rest of the first word in caps (small caps for compsoc). % % form to use if the first word consists of a single letter: % \IEEEPARstart{A}{demo} file is .... % % form to use if you need the single drop letter followed by % normal text (unknown if ever used by the IEEE): % \IEEEPARstart{A}{}demo file is .... % % Some journals put the first two words in caps: % \IEEEPARstart{T}{his demo} file is .... % % Here we have the typical use of a "T" for an initial drop letter % and "HIS" in caps to complete the first word. 半监督学习在模式识别和机器学习上引起了广泛关注。虽然半监督学习分类是一个相对新颖的领域，但是，早在几十年前，使用未标记的样本来进行预测的想法就被提出来了。对半监督学习的研究，最开始被认为是Scudders在1965年进行的关于“自学习”的工作，早些年Robbins和Monro在顺序学习上的研究也可以被看作跟半监督学习有关的研究。半监督学习，尤其是半监督分类的主要思想是，同时利用已标记和未标记数据来训练一个分类模型。现在，每天都有大量的数据正在因文章、文件、图像、电子邮件等媒体而产生，这些数据的大多数都是没有被分类，或者说是没有被标记的。因此，通过监督学习方法来使个人新闻过滤、电子垃圾邮件过滤、文本分类和图像分类等软件实现自动化，是非常困难的。现实中通常只有一小部分的已标记数据是可用的， 比如说，基于某个用户已标记为“喜欢”的文章，或者说该用户标记为“垃圾”的邮件，但是另外一边有更大量的数据没有被标记，这造成的结果就是，人们急需能够利用少量的已标记数据，同时又能结合大量的未标记数据和有助于构建有效的分类系统的算法。\\ 根据算法潜在的基本假设，现存的半监督学习分类算法可以分为两大类。一种是基于流形假设的算法，这种算法假设数据来自一个高维空间，但是同时也可以用一个低维空间来表示。通常地，数据的基本几何图形，是通过将数据表示为图表、将样本表示为节点、样本之间的成对相似性作为边权重来捕获的。一些基于图的算法，比如说标签传播算法, 马可洛夫随机游走,图割算法, 频谱图传感器,还有低密度分离等算法，经常在文献中被推荐，就是基于此流形假设的。\\ 还有一些被推荐用于半监督学习的自然归纳的算法。这些算法通常是基于聚类假设。聚类假设表明具有高度相似性的样本一定属于同一类标签。换一种说法就是，类之间的决策边界必须穿过低密度区域。这个假设允许未标记数据来正则化决策边界，这将反作用于影响分类模型的选择，许多像TSVM、半监督SVM这样成功的半监督学习算法都是采取了这个方法。这些算法基于决策边界假定模型，结果就变成一个归纳的分类器。\\ 流形正则化也是一种归纳方法，但是其基于流形假设。流行正则化试图在数据集上构建一个最大化边界的分类器，同时最小化相应的相似矩阵的不连续性，这是通过往一个基于SVM的目标函数里添加一个基于图的正则项来实现的。一个与此相关且名为“LIAM”的算法，使用将先验度量信息编码在图拉普拉斯算子的方法来正则化SVM决策边界，同时这个算法有一个快速的优化算法。\\ 大多数的半监督学习方法特意地设计能够有效利用已标记和未标记数据的算法。然而，情况通常是这样的：使用者的应用中已经有一个非常适合的监督学习算法，他所需要的是通过利用可行的未标记数据来提升算法的性能。从这点出发，更实际的方法是设计一种利用未标记样本的技术，而不考虑底层的学习算法。这样的方法能够适应于分类器的基于任务选择，同时能够有效地利用未标记数据。区别于我们所研究的标准的半监督学习问题，我们把这种，通过使用未标记数据来提高任何一种监督学习算法的性能的难题，称作半监督提升。\\ 为了实现半监督提升，我们推出了一种boosting框架，我们将其称作“SemiBoost”，用于通过未标记数据实现某中监督学习算法的提升。跟大多数的boosting算法相似，SemiBoost迭代地提高分类的准确度。在每一次迭代中，选择一部分的未标记样本，使用特定的监督学习算法来训练出一个新的分类模型。在最后，每次迭代训练所得的模型线性组合在一起，形成一个最终的分类模型。SemiBoost过程的概览呈现在图1中（图1为原文的Fig 1.）设计SemiBoost的主要难点在于：1)如何对未标记的例子进行抽样，用于每次迭代并训练出新的分类模型；2）所被选择的未标记例子应该被分配成什么类标签。值得注意的是，不像监督boosting算法那样，我们选择难以分类的已标记示例，SemiBoost需要在每次迭代过程中选择未标记的示例。\\ 一个解决上述问题的方法是，利用聚类设想和大边距标准。我们可以通过选择具有最高分类置信度的未标记示例来提高分类界限，并为它们分配由当前分类器预测的类标签。这些被分配的标签被称作伪标签。已标记的数据和被选中的被标上伪标签的数据被用于下一次迭代，以便产生第二个分类器。这个方法被类似于自学习、ASSEMBLE和半监督MarginBoost等采用。然而，这个策略的一个问题是，这些带有被预测的类标签的示例的引入可能仅仅起到加大分类边距的作用，并没有给分类器提供任何新奇的信息。因为被选中的未标记示例是可以被置信分类的，他们通常是远离决策边界的。这样的结果是，使用被选中的未标记示例进行训练得出的分类器，和原来使用已标记示例训练出来的分类器很大可能上会共享决策边界。这是因为，通过调整决策边界，有较高分类置信度的示例会获得更高的置信度，这意味着我们可能需要额外的指引和最大化边距准则来提高基分类器的性能。\\ 为了解决上面的这个问题，我们推荐在每一次迭代的过程中，使用成对相似性测量来指引未标记示例的选择，同时也给他们分配类标签。对于每一个没有标记的示例$x_i$，我们分别计算将$x_i$分配为正类和负类的置信度。这两个置信度的计算基于强化后的分类器的预测结果和不同示例之间的相似度。我们接着在每次迭代中，选择拥有最高分类置信度的示例，再带上已标记的示例训练出一个新的分类模型。现有的分类模型将会线性组合在一起形成一个新的分类模型，用于更好地预测结果。需要注意的是，我们推荐的这个方法跟利用流形预测的基于图的半监督学习方法密切相关。接下来这一结将会讨论现存的半监督学习方法和他们跟SemiBoost之间的关系。 \section{RELATED WORK} 表1（表1为原文pdf中的TABLE 1）简要地展示了目前存在地半监督学习方法和他们底层的假设。 第一列展示的是算法对数据的假设；第二列是这些半监督学习方法的名字和引用；紧接着第三列是简要的方法描述；第四列说明算法是自然归纳（I）还是转导的（T）。 一个归纳算法可以被用作预测在训练过程中不曾出现的样本的标签（无论是已标记的还是未标记的）；所谓转导算法限于用来预测那些在训练过程中被使用的未标记的样本的标签。\\ 基于图的方法将已标记和未标记示示例表示为一个连接图，连接图中，每一个示例都用一个节点来表示，如果两个节点之间有非常大的相似性的话，这对节点会通过一条边连接起来。 属于这类方法中比较出名的有，基于谐波函数的方法、光谱图传感器方法、基于高斯处理的方法、流形正则化还有标签传播方法。 使用监督类标签和图结构，最小化未标记示例的不一致性，可以找到这些示例的最佳类标签。\\ 二次准则是一个很受欢迎的方法，它定义了标签$y=\{y_i\}^n_{i=1}$在样本$\{x_i\}^n_{i=1}$之上的不一致性和成对相似性$S_{i,j}$\\ \begin{center} \[F(y)=\sum_{i=1}^n\sum^n_{j=1}S_{i,j}(y_i-y_j)^2=y^TLy,\] \end{center} 公式中，L是组合图拉普拉斯算子。 给定一个半监督设定，在上面的一致性测量中，仅假设已知一些标签，剩下的标签都是位置的。 我们的任务是给未知的标签赋值，使得整体的不一致性最小化。 在[6]中的方法认为这种情况下应有$y_i\in\{\pm1\}$时，于是制定了一个离散最优化问题，使用最小割方法解决。 然而，最小割易于使解决方案退化，因此，目标函数是使用一种混合整数编程方法来最小化的，这种编程方法需要大量的计算资源，经常使很多人望而却步。 在好几种方法中，对目标函数在$y_i\in[0,1]$时的持续松弛都有被考虑到，这是通过马可洛夫随机场、高斯随机场和谐波函数来解决的。\\ 我们推出的这个框架利用了半监督学习中的成对相似性，从某种意义上说，这是于基于图的方法是密切相关的。 我们所提出的这个方法中使用的不一致性度量，遵循了类似的定义，除了使用指数代价而不是平方代价的函数来违反标签。 跟大多数的基于图方法不同，我们通过从未标记和已标记示例中学习，创造了一个特别的分类器模型。 这对于半监督提升来说尤为重要，因其目标是使用大量的未标记数据实现对监督学习算法的提升。\\ 基于聚类假设的方法利用未标记数据来正则化决策边界。 特别情况下，穿过低密度数据区域地决策边界，往往被认为是被未标记示例稠密包围的。 这些方法都特别地扩展了SVM或者是相关得最大化边距分类器，而且，对于决策树等非基于边距的分类器来说不易于扩展。 这类方法有：转导支持向量机（TSVM）、半监督支持向量机（S3VM）还有具有零类噪声模型的高斯过程等。 另一方面，我们推荐的算法是一个能使基分类器很好地适应特定任务的普通方法。\\ 最后，我们强调，这个方法跟用于半监督学习的集成方法家族是密切相关的。 在类似于AdaBoost等算法大获成功后，集成方法在监督分类领域中受到热烈的欢迎。 集成算法的半监督对应物是基于聚类假设的，主要的示例包括ASSEMBLE[16]，还有半监督MarginBoost(SSMB)[17]。 这两个算法的工作原理都是，给未标记样本分配伪标签，然后将这些新标记的数据作为训练下一个新的监督分类器时所需的样本。 SSMB和ASSEMBLE都是基于边距boosting算法，他们通过下面这个公式来最小化成本函数： \begin{center} \[J(H)=C(y_iH(x_i))+C(\vert H(x_i)\vert),\] \end{center} 公式中，H是构建中的集成分类器，C是单调递减的成本函数。 $y_iH(x_i)$这一项对应了已标记样本的边距定义。 边距的定义包含对未标记样本来说不可用的真实标签$y_i$。 ASSEMBLE中定义了像$\vert H(x_i)\vert$这样的伪边距，在SSMB中则为$H(x_i)^2$，同时考虑到$y_i\in\{\pm1\}$这一事实，这样在目标函数中可以省去$y_i$这一项。 然而，这个算法依赖于每次迭代时现有的集成分类器所预测出来的伪标签。 相反地，我们推荐的算法，结合了相似度信息和分类器预测结果，获得更多可靠的伪标签，这跟现有的方法非常地不同。 另外一边，SSMB需要基学习器本身就是一个半监督算法，因此，SSMB用于解决与我们所推荐的算法相反的boosting半监督算法问题。\\ 在本质上，SemiBoost算法结合了基于图和集成方法的优点，产生了一个在半监督学习上更加普遍而且更加强大的方法。% \section{SEMI-SUPERVISED BOOSTING} 我们首先来正式描述一下半监督提升，然后展示SemiBoost算法% \subsection{Semi-Supervised Improvement} 使用$D=\{x_1, x_2, ... , x_n\}$来表示整个数据集，包括已标记的和未标记的实例。 假设首先有$n_l$个实例已被标记，在$y_l=(y_1^l,y_2^l,...,y_{nl}^l)$中，每一个类标签$y_i^l$都是+1或者-1。 我们使用$y_u=(y_1^u,y_2^u,...,y_{n_u}^u)$来表示估算出的未标记实例的类标签，此处$n_u=n-n_l$。整个数据集的标签表示为$y=[y_l;y_u]$。 对称相似矩阵表示为$S=[S_{i,j}]_{n\times n}$，$S_{i,j}\geq0$代表着$x_i$和$x_j$之间的相似性。 而且，$S^{ll}$表示对应$n_l$个已标记实例的相似矩阵的$n_l\times n_l$子矩阵，同时$S^{lu}$表示已标记和未标记实例的$n_l\times n_u$子矩阵。 $S^{uu}$和$S^{ul}$的子矩阵可以相应地定义。 用$A$表示为给定的监督学习算法。 半监督提升的目标是，把$A$当做一个黑盒，使用未标记实例和成对相似性$S$来迭代地提高$A$的性能。 在图2（对应原文pdf Fig.2）中给出了一个用于半监督提升的SemiBoost算法的简要概述。\\ 将半监督提升问题和现存的半监督分类方法区分开来是很重要的。 就像在第二节中讨论的那样，任何集成算法的新分类器都必须依赖于伪标签。 另一方面，基于图的算法使用样本之间的成对相似性，给未标记的实例分配标签，达到样本在相似度上的一致性。 在半监督提升问题上，我们致力于构建一个能够像基于图方法那样利用未标记样本的集成分类器。 \subsection{SemiBoost} 为了提高给定的学习算法$A$，我们遵循boosting的核心思想，迭代地运行算法$A$。 在每一次迭代中，算法$A$都学习到一个新的分类模型，而且在不同的迭代中的分类模型都将会被线性组合在一起，从而形成最终的分类模型。 \subsubsection{Objective Function} 必须根据这两个主要的准则给未标记实例分配标签：1)在未标记样本之中有高相似度的点必须分配到同样的标签；2)未标记样本如果跟一个已标记样本有高度相似性的话，那么未标记样本被分配该已标记样本的标签。 我们的目标函数是$F(y, S)$包含了两个项：一个度量已标记和未标记实例$F_l(y,S)$之间的不一致性；另一个$F_u(y_u, S)$度量未标记实例之间的不一致性。\\ 受到谐波函数方法的启发，我们定义$F_u(y, S)$作为类标签$y$和相似度测量$S$之间的不一致性，也就是说： \begin{center} \begin{equation} F_u(y_u, S)=\sum^{n_u}_{i,j=1}S^{uu}_{i,j}exp(y^u_i-y^u_j) \end{equation} \end{center} 许多使用相似性或者核矩阵的目标函数需要正半定核，这样才能保证目标函数的凸性。（比如说SVM） 然而，由于$exp(x)$是一个凸函数，而且我们假设$S^{uu}_{i,j}$是非负的，不管相似矩阵的正定性如何，对于每个$i,j$来说，$F_u(y_u,S)$都是凸函数。 这就出现了相似矩阵之间互不对称，而且不用改变目标函数的凸性。 不对称相似矩阵在使用有向图进行分类问题建模的时候大有用处，而且在与文本分类有关的应用中有更好的表现。\\ 尽管如此，我们的方法还是可以用于一般的相似矩阵，我们假设提供的相似矩阵是对称的。 上面的公式可以扩展成以下形式： \[F_u(y_u,S)=\frac{1}{2}\sum S^{uu}_{j,i}exp(y_j^u-y_i^u)+\frac 12\sum S^{uu}_{i,j}exp(y^u_i-y^u_j)\] 同时根据$S$的对称性，我们可以得到 \begin{center} \begin{equation} F_u(y_u,S)=\sum^{n_u}_{i,j=1}S^{uu}_{j,i}cosh(y_i^u-y_j^u)\end{equation} \end{center} 此处$cosh(y_i-y_j)=(exp(-y_i+y_j)+exp(y_i-y_j))/2$是双曲线余弦函数。$cosh(x)$是一个凸函数，它的最小值在$x=0$时取得。 通过使用$cosh(.)$函数重写该函数，我们可以发现在基于图拉普拉斯方法的二次惩罚因子和现在方法中的指数惩罚因子之间的联系。 使用$cosh(.)$惩罚因子不仅有助于基于boosting算法的求导，同事也加大了分类边距。 用于boosting算法的指数成本的效用是众所周知的[27]。\\ 已标记和未标记实例$F_l(y,S)$之间的不连续性定义为： \begin{center} \begin{equation} F_l(y,S)=\sum^{n_l}_{i=1}\sum^{n_u}_{j=1}S^{uu}_{i,j}exp(-2y^l_iy^u_j). \end{equation} \end{center} 结合公式(1)和(3)可以得到以下的目标函数： \begin{center} \begin{equation} F(y,S)=F_l(y,S^{lu})+CF_u(y_u,S^{uu}). \end{equation} \end{center} 引入常量C用于权衡已标记和未标记数据的重要性。 给定在(4)中的目标函数的最优类标签$y_u$，可以通过最小化$F$来取得\\ 使用$\hat{y}_i^l,i=1,...,n_l$来表示学习算法在训练数据中的已标记实例上预测得到的标签。注意，在公式(4)中，没有一项对应了预测得到的标签和真实标签之间的不一致性，也就是$F_{ll}=\sum\nolimits^{n_l}_{i=1}exp(y_i^l,\hat{y}^l_i)$。把这一项添加到公式中，当没有未标记样本时，可以使算法专攻于AdaBoost。 然而在实践中，可用的已标记数据数量有限，$F_{ll}$这一项通常要比$F_l$和$F_u$不那么重要，因此在公式(4)中也就被去除了。\\ 选择一个更小的样本子集来训练分类器可能没那么有效。所以，当前的方法是，将已标记数据的预测包含在约束的形式中，这也就能在每次迭代中，有效利用所有的已标记数据来为训练分类器。 这个问题现在可以正式地表示为如下： \begin{center} \begin{equation} \begin{split} \min \quad &F(y,S) \\ s.t. \quad &\hat{y}^l_i=y^l_i,i-1,\ldots,n_l \end{split} \end{equation} \end{center} 这是一个凸优化问题，也就是说该问题能够使用数字方法有效解决。 但是，我们的目标是，通过未标记数据和相似矩阵$S$，提升给定的学习算法$A$，所以我们推出的boosting算法能够有效地最小化目标函数$F$。 下面是推导出该boosting算法的过程： \begin{itemize} \item 未标记实例$y_i^u$的标签由在相对应的数据样本上集成预测出的标签来代替 \item 使用边界最优化方法寻找集成分类器，最小化目标函数 \item 边界进一步被简化，以获得抽样方案和其他需要的参数 \end{itemize} 上面提到的目标函数和几个基于图的、流形正则化和集成方法有很密切的相关性。在本文更长篇的版本中的附录F和G，我们讨论了SemiBoost和被广泛使用过的几种半监督算法之间的关系。 \subsection{Algorithm} 我们使用边界最优化方法派生出了这个boosting算法，另外，在[29]中介绍了使用传统的函数梯度方法来推导boosting算法 这种方法也可以被视为能通过线性函数逼近原有的目标函数的一种松弛。 然而，这样一种方法包含了参数“步长”的规范。 在我们的推导中，步长是自动确定的，所以这就解决了确定步长的问题。 SemiBoost算法在图3（原文pdf中的Fig.3）中有简要的总结。\\ 用$h^{(t)}(x):\chi \rightarrow \{-1, +1\}$来表示才从算法$A$在第$t$迭代中中学习到的二分类模型。 用$H(x):\chi\rightarrow IR$来表示经过$T$次迭代得到的组合分类模型。 它是将$T$个分类模型线性组合在一起而得来的： \begin{center} \[ H(x)=\sum^T_{t=1}\alpha_th^{(t)}(x), \] \end{center} 公式中$\alpha_t$是组合权重。在第$(T+1)$次迭代时，我们的目标是找到能有效最小化目标函数$F$的一个新的分类器$h(x)$和组合权重$\alpha$。\\ 现在问题变成了下面的最优化问题: \begin{center} \begin{equation} \begin{split} \argmin_{h(x),\alpha}\quad &\sum^{n_l}_{i=1}\sum^{n_u}_{j=1}S^{lu}_{i,j}exp(-2y^l_i(H_j+\alpha h_j))\\ &+C\sum^{n_u}_{i,j=1}S^{uu}_{i,j}exp(H_i-H_j)exp(\alpha(h_i-h_j)) \\ \end{split} \end{equation} \begin{equation} s.t.\quad h(x_i) = y^l_i,i=1,\ldots,n_l, \end{equation} \end{center} 公式中,$H_i\equiv H(x_i)$且$h_i\equiv h(x_i)$。\\ 这个表达式包含了变量$\alpha$和$h_i$的乘积，使其变得非线性，也因此变得很难优化。 然而，这些限制能够通过包含所有的已标记样本到每个分类器的训练集中来实现。 为了简化计算，我们构建了目标函数的上限，如命题1所述： \begin{prop} 最小化公式（7）等价于最小化公式： \begin{equation} \overline{F}_1=\sum^{n_u}_{i=1}exp(-2\alpha h_i)p_i+exp(2\alpha h_i)q_i \end{equation} 公式中： \begin{align} p_i &= \sum^{n_l}S^{ul}_{i,j}e^{-2H_i}\delta(y_j,1)+\frac{C}{2}\sum^{n_u}_{j=1}S^{uu}_{i,j}e^{H_j-H_i},\\ q_i &= \sum^{n_l}_{j=1}S^{yl}_{i,j}e^{2H_i}\delta(y_j,-1)+\frac{C}{2}\sum^{n_u}_{j=1}S^{uu}_{i,j}e^{H_i-H_j}_{i,j}, \end{align} 且当$x=y$的时候有$\delta(x,y)=1$，当$x\neq y$时有$\delta(x,y)=0$ \end{prop} 证明速写：将$F(y,S)$替换为$H_i\leftarrow H_i +\alpha h_i$，重组各项，我们就能够得到想要的答案\\ 数量$p_i$和$q_i$可以分别被译作将$x_i$归类为正类或负类的置信度。\\ 在(8)中的表达式很难优化，因为权重$\alpha$和分类器$h(x)$是耦合在一起的。我们使用接下来定义的上限来简化这个问题： \begin{prop} 最小化公式(8)等价与最小化此公式： \begin{equation*} \overline{F}_1\leq\sum^{n_u}_{i=1}(p_i+q_i)(e^{2\alpha}+e^{-2\alpha}-1)-\sum^{n_u}_{i=1}2\alpha h_i(p_i-q_i) \end{equation*} \end{prop} 证明：见[28] 我们把上面这个等式的上限定义为$\overline{F}_2$ \begin{prop} 为了最小化$\overline{F}_2$，实例$x_i$的最优的类标签$z_i$是$z_i=sign(p_i-q_i)$，同时其权重为$\vert p_i-q_i\vert$ 使$\overline{F}_1$最小的最优$\alpha$为： \begin{equation} \alpha = \frac{1}{4}ln\frac{\sum\nolimits^{n_u}_{i=1}p_i\delta(h_i,1)+\sum\nolimits^{n_u}_{i=1}q_i\delta(h_i,-1)}{\sum\nolimits^{n_u}_{i=1}p_i\delta(h_i,-1)+\sum\nolimits^{n_u}_{i=1}q_i\delta(h_i,1)} \end{equation} \end{prop} 证明速写：在(11)中的表达式可以通过区分$\overline{F}_2$和$\alpha$和设置为0获得。 观察可得，上面的函数在$h_i(p_i-q_i)$是线性的，如果我们选择$h_i=sign(p_i-q_i)$，那么函数就能被最小化，最大值为$\vert p_i-q_i\vert$\\ 命题1-3证明了在SemiBoost的求导中的松弛是可行的。在每次松弛中，“接触点”维持在目标函数和上限之间。结果是，这个过程保证了以下几点：1）目标函数在迭代过程中单调递减；2）最终的方案收敛到一个极小值。在[30]中可见更多细节。命题3建立了一个boosting算法的关键因素。有了这些，SemiBoost算法可以以图3（原文pdf的Fig. 3）的形式展现出来。\\ 用$\epsilon_i$来表示分类器得出的权重的误差，则有： \begin{center} \begin{equation*} \epsilon_t=\frac{\sum^{n_u}_{i=1}p_i\delta(h_i,-1)+\sum\nolimits^{n_u}_{i=1}q_i\delta(h_i,1)}{\sum\nolimits_i(p_i+q_i)} \end{equation*} \end{center} 在AdaBoost[29]中,$\alpha$可以表示为： \begin{center} \begin{equation} \alpha_t=\frac14ln(\frac{1-\epsilon_t}{\epsilon_t}), \end{equation} \end{center} 这跟AdaBoost的权重因子非常地相似，它们仅仅不同在于是否存在常熟因子$\frac{1}{2}$。 同时，如果AdaBoost遇到了基分类器的错误率要比随机错误率高的情况（也就是说$\epsilon_{t+1}\geq\frac{1}{2}$），它会返回当前的分类器$H_t$。 这跟SemiBoost遇到的当$\alpha\leq0$就停止的情况时有直接对应关系的。 从公式（11）（或者直接从公式（12））观察，我们可以看到这种情况仅仅会在分母大于分子的时候出现，也就是说，$\epsilon_{t+1}\geq\frac{1}{2}$是等价于$\alpha\leq0$。 然而，在没有大量的分类器被训练过出来的情况下，这种条件很难满足，通常有一个参数用来定义要使用的分类器的个数。 以往的经验告诉我们，固定数量（比如说20）的分类器能够给AdaBoost带来更好的性能[27]\\ 在SemiBoost中使用的样本计划跟AdaBoost的有很大的不同。AdaBoost知道数据的标签，而且也因此能够基于上一次迭代地结果，增加或减少分配到每个样本上的权重。 在SemiBoost里面，对于未标记数据，我们没有掌握真实的标签，评估分类的难度都变得非常具有挑战性。 然而，命题2给出的答案是，选择最具有置信度未标记数据样本对降低目标函数来说是最优解。 直观上看，使用具有高置信度的标签无疑是一个很好的选择，因为他们在成对相似性信息和分类上是一致的。 $p_i$和$q_i$的值在这几种情况下会变得很大：1）$x_i$不能置信地被分类出来，也就是说$\vert H_i \vert$很小，而且它的某个临近点已经被标记，这对应着公式（9）和（10）的第一项；2）实例$x_i$和一些已经被自信分类的未标记实例高度相似，也就是说，对于实例$x_j$有巨大的$s_{i,j}$和$\vert H_j \vert$。 这对应着公式（9）和（10）中的第二项。 这意味着相似度信息在样本选择的导向中扮演着一个极其重要的角色。与之相反，之前的方法像ASSEMBLE和SSMB的，样本只是被选择来仅仅提高$\vert H_i \vert$的值。\\ 类似于大多数的boosting算法，我们可以看到所推荐的半监督boosting算法指数级别地减少了原来的目标函数。 结果可以用下面的理论来概括：\\ \textbf{Theorem 1.} 定义$\alpha_1, \ldots, \alpha_t$代表SemiBoost算法计算出来的组合权重。然后，在第$(t+1)$次迭代时的目标函数，也就是 $\overline{F}_{t+1}$的界限如下： \begin{center} \begin{equation*} F_{t+1}\leq\kappa_S exp(-\sum^t_{i=1}\gamma_i), \end{equation*} \end{center} 其中： \begin{center} \begin{equation*} \kappa_S=[\sum^{n_u}_{i=1}(\sum^{n_l}_{j=1}S^{ul}_{i,j}+C\sum^{n_u}_{j=1}S^{uu}_{i,j})] \end{equation*} \end{center} 同时$\gamma_i=log(cosh(\alpha_i))$.\\ \textbf{Proof}见[28]\\ 上述定理表明，尽管在上面的命题中做出了松弛，目标函数还是遵循了指数级衰减。\\ \textbf{Corollary 1.}就错误$\epsilon_t$而言，目标函数在第$(t+1)$次的表示方法为： $F_{t+1} \leq \kappa_S \prod\nolimits^t_{i=1}(\frac{1-\epsilon_i}{\epsilon_i})^{\frac{1}{4}}$\\ \textbf{Proof Sketch.}可以通过替换定理1中的$\alpha_i$为公式（12）来验证。 \\ 在上面的求导过程中，我们限制了目标函数，这样一来分类器在已标记样本上的预测必定会与我们提供的真实标签相匹配。 然而，如果真实的标签中噪点太多，半监督分类的结果可能不会达到最理想的状态。 如果已标记数据的被预测出来的标签跟真实标签不符，那么可以通过增加一项惩罚项来推导出另外一个相似于SemiBoost的算法。 在这篇论文中，我们假设给定的标签是正确的。 这是合理假设，因为有非常少的已标记样本，确保他们的正确性并不会遇到太多的困难。 \subsection{Implementation} \subsubsection{Sampling} 就跟其他的boosting算法一样，取样组成了SemiBoost的最重要的部分。取样的准则通常考虑到下面的因素：1）有多少样本是必须要从未标记样本中选取出来进行训练的；2）必须进行的抽样分布是什么\\ 像AdaBoost这样的监督boost算法有可用的真实标签，这使得算法很容易地去选择或者不选择哪些样本。 另一方面，在SemiBoost的迭代过程中分配的标签都是伪标签，而且很有可能是错误的。 这就表明在SemiBoost算法工作中，我们应该从少数中选择最具有置信度的数据点。 但是选择少量的样本可能会使得收敛的过程变得缓慢，同时选择过多的样本可能会包含一些无信息甚至很差劲的样本到训练集中去。当前的选择标准是：经验法则；选择样本中最有可能在实践中表现良好的10\%。 从命题3中可得，要减少$\overline{F}_1$，选择使$\vert p_i-q_i \vert$的值增大的样本是较好的。 这种选择给分类器提供了高度可靠的被分配伪标签的样本。取样可以通过下面的分布来完成： \begin{center} \begin{equation*} P_s(x_i)=\frac{\vert p_i - q_i \vert}{\sum\nolimits^{n_u}_{i=1}\vert p_i - q_i \vert} \end{equation*} \end{center} 上面的$P_s(x_i)$是数据点$x_i$为来自转导集的样本的可能性。 \subsubsection{Stopping Criterion} 根据优化过程，SemiBoost会在$\alpha\leq0$的时候停止，也就意味着分类器的增加会提升目标函数的值而非降低。 然而，$\alpha$的值在开始的时候减小得特别快，到最后，减小的速率会急剧下降，需要非常多次数的迭代才能将其变为负数。 我们现在在集成中使用由经验得出的固定的分类器个数，定义为$T$，我们把$T$的值设置为20 \subsubsection{Similarity Matrix} 受到径向基在基于图方法中的成功的激励，我们决定使用径向基相似性。对于两个样本$x_i$和$x_j$来说，他们之间的相似性$S_{i,j}$可以通过$S_i,j=exp(-\Vert x_i - x_j\Vert^2_2/\delta^2)$来计算，此处$\delta$是控制径向基扩散程度的缩放参数。 众所周知，$\delta$的选择对于算法的整体性能有非常重要的影响[18]。我们把对于相似性值的缩放参数定在第十个百分位到第一百个百分位，以10步长，$\mu$设置为相似矩阵$S$的平均值。实验证明，在$\sigma$的选择范围内，转导和归纳方法的性能都是稳定的。 这个属性是我们一直想要得到的，毕竟考虑到选择一个正确的缩放参数是一件困难的事情。 \section{RESULTS AND DISCUSISON} SemiBoost专注的事情是使用未标记数据来提高任何给定的监督分类器，因此我们主要的目标是，根据基分类器的归纳性能的提升，对SemiBoost进行评估。\\ 一个使用了SemiBoost和“ring-norm”数据集的监督学习器的性能提升在图4（原文pdf中Fig. 4）中所展示。 整个数据集有两个类，每一类有500个样本。 图中每个类有10个已标记样本。 实线代表着决策边界，黑暗和明亮的区域代表着两个类区域。 SemiBoost在每次迭代中的性能都在每幅图下的括号中。 图4a，4b，4c展示了SemiBoost在前三次迭代中获得的分类器，图4d展示了经过12次迭代的最终分类器。 \subsection{Data Sets} 我们使用了16个不同的数据集来评估SemiBoost：4个[9]中提供的基准数据集，10个UCI数据集，还有2个用于脸部种族识别和文本分类的数据集。 由于SemiBoost适用于二分类问题，所以我们从基准数据集中选择了二分类数据及。UCI数据集中的多类数据也通过选择最多的两类的方法被转化成二类数据集。数据集通过去除有特征损失、对分类特征进行二进制加密、使用PCA降维以保持95\%的方差等方法进一步处理。 所选数据集名称、选中的类中的样品数量$n$、还有数据集的维度$d$都总结在表2(Table 2)的第一列上。 除了这些，我们还在文本识别问题上评估了我们的方法，结果将会在下一节展示。\\ 半监督学习算法的转导性能已经被研究得很彻底了[9, Chapter 21]，然而，半监督学习并不仅限于转导学习，同时，无样本扩展已经吸引了非常多人的注意。 事实上，归纳学习非常的重要，因为在训练过程中仅仅只有一部分的未标记样本是可见的。 在这些情况下，学习的真正效用在于给新的测试样本分类的能力。 在这个目标的驱动下，我们对SemiBoost和当前最先进的三种归纳监督学习算法进行了比较。他们分别是转导SVM，归纳版的低密度分离（LDS）还有来自流形正则化方法的拉普拉斯SVM。 LDS不是一个归纳算法因为它包含了基于图的维度下降步骤。 我们使用LDS在转导集上预测的标签，在原有的数据上进行归纳分类器的训练。 \subsection{Experimental Setup} 实验的设置旨在研究，通过使用未标记数据、SemiBoost算法的表现和其他三种当前最先进的版监督学习算法后，监督学习器的性能提升。\\ 我们把分类准确性作为评估的度量。 每个实验的百分比精度的平均偏差和标准偏差都是经过20次的反复试验得来的。 20次重复的试验分别都是用不同的训练集和测试集的子集。 为了测量归纳性能，我们把数据随机分割成两半。 我们把这些乘作训练集和测试集。 训练集有10个已标记点，剩下的都是未标记点。 使用在测试集上的预测结果，评估SemiBoost在训练集上学习到的集成的分类器的性能。\\ SemiBoost在未标记数据中取样，在每次迭代中给他们分配标签，并且构建一个分类器$h_t(x)$。 这样构建的分类器的个数取决于迭代次数$T$。 $T$设置为10，我们在公式（11）中计算得到的权重$\alpha_t$为负时停止boosting。 我们把目标函数里$C$的值设置为已标记样本数目和未标记样本数目的比例$C=n_l/n_u$。\\ 第一个实验研究了三种不同的基分类器$(h_t(x))$在使用了SemiBoost之后的性能提升：DS, J48决策树算法(J48), 还有关于顺序最小优化算法的SVM。 我们使用软件WEKA[33]来实现这三个分类器。 所有的算法都是使用默认的参数运行（比如对SVM算法来说，默认的C和线性的核） 我们选择决策树（DS和J48）还有SVM作为基分类器，是因为他们在监督学习文献中，可用于各种学习任务、表现为最为出色的三种基本分类器。 \subsection{Results} \subsubsection{Choice of Base Classifier} 表2（Table 2）将监督和三种基准半监督算法与SemiBoost算法进行了比较。 DS，J48和SVM这三列给出了基分类器在归纳集上的表现。 SB-X一列给出了SemiBoost在基分类器X上的表现。 表2中的最后三列对应基准半监督学习算法TSVM，LDS，和LapSVM的归纳表现 需要注意的是，我们的目的不是构造对某个分类问题最适用的分类器，而是说明SemiBoost可以在所有的分类问题上对监督分类器在性能上进行可能的提升。 结果表明SemiBoost几乎在所有的数据集上，都显著地提升了三种基分类器的性能。 使用一个独立样本配对的测试集，我们发现在SemiBoost的影响下，Decision Stump在12个数据集上的表现有了显著提升。对于J48而言，分类器在13个数据集上有了更好的效果，但是在房屋数据集上有了严重的退化。SVM则在7个数据集上有了显著提升，但是在其他三个数据中上有严重退化。 在SVM性能退化的情况下，基准算法跟监督分类器相比起来表现得很差，这表明：未标记数据在这些场合并没有什么帮助。 通过SemiBoost得到的集成分类器相对来说更加地稳定，因为它的分类精度跟基分类器相比起来有更低的标准误差。\\ \subsubsection{Performance Comparison of SemiBoost with Benchmark Algorithms} 我们比较了SemiBoost和另外三种不同的算法：TSVM，LapSVM，还有ILDS。SemiBoost所展示出来的性能能够与基准算法相媲美。 SemiBoost几乎在所有的数据及上比ILDS表现得更好，而且在四个数据集上使DS获得了更加巨大的提升，在2个数据集上使SVM发挥更好。 与TSVM相比，SemiBoost在十个数据集上使SVM更好，在8个数据集上使DS更好。 同时，TSVM在三个数据集上在一段合理的时间内（20小时）难以收敛。 SemiBoost和LapSVM的表现相差无几；SB-DS的性能在两个数据集上要优于LapSVM，在一个数据集上要差于LapSVM。 类似地，SB-SVM和LapSVM显著地在8个情况中的3个要优于彼此。 在一些数据集上，有基分类器要比SemiBoost表现更好。但是，在这些情况中，这个基分类器的性能都要比所有的半监督算法都要好。（比如说SVM在COIL2、vehicle、sat、还有房屋数据集上要优于所有的算法） 这意味着，未标记数据不总是能够提升基分类器的性能的，或者更通俗地说，未标记数据不总能在学习过程中起到帮助作用。 当一个基分类器优于半监督学习算法时，我们可以发现，与其他的算法比较起来，SemiBoost更加接近基准线。\\ \subsubsection{Performance with Unlabeled Data Increment} 图5展示了SemiBoost在UCI数据集上的表现。每一个数据集都被分成两等分，一份用作训练，另外一份用作归纳测试。 训练集中的十个样本被标记。 SVM在训练集上使用默认的参数进行训练，在测试集上的表现使用点线表示。 训练集中未标记的实例以10\%的比例增加到已标记实例中。 实线代表着SemiBoost算法在未标记数据增加的情况下的性能。 虚线代表着SVM在所有被添加进去的样本被标记后的性能。 同时可以观察到，SemiBoost的性能随着所添加的未标记数据的增多而提升，无论何时，这样的提升都是有可能存在的。 \subsubsection{Sensitivity to Parameter $\sigma$} 图6展示了SemiBoost—SVM在参数$\sigma$取不同的值时的不同性能。Sigma被选作相似性分布的第$\rho$个百分位。$\rho$在第十个到第一百个百分位之间。选择合适的$\sigma$的值是构造图最困难的部分之一，而且已经有好几种启发式方法被提出，用于决定$\sigma$。在上面所展示的大多数数据集上，SemiBoost在缩放参数方面是相对稳定的一个。 然而，基于经验所得，一般推荐$\sigma$的选择在成对距离的第十到第二十个百分位之间。 \subsubsection{Margin and Confidence} 在这个实验中，我们凭经验证明了SemiBoost倾向于最大化平均边距。 对于未标记数据，一个流行的关于边距的定义是$\vert H(x_i) \vert$[16], [17]。 平均边距是$\vert H(x_i) \vert$在被用作训练的未标记数据上的经验平均值。 图7a，7b，和7c相对应地展示了DS，J48，SVM这三个基分类器在optdigits数据集上的平均边距值。 平均边距值随着迭代次数的增多而增大，无论选择的是哪个基分类器。 然而，需要注意的是，即使在每一次迭代中测试错误不断减少，最小的边距可能不会一直增加。 当训练集包含了一小部分的能够被完美地分类的已标记样本时，边距就会很大程度上取决于未标记数据。 考虑到在未标记数据上的边距，在第一次迭代中分类器在所有未分类的数据上的边距为$\alpha_1$，而在第二次迭代中，最小的边距是 $\min_{x_i}\vert H^{(2)}(x_i)\vert = \vert\alpha_1-\alpha_2\vert\leq\alpha_1=\min_{x_i}\vert H^{(1)}(x_i)\vert$。 事实上，在所有的迭代中，最小边距的值可能被牺牲用于获得性能上的提升，也就是说，与相似矩阵达成一致。 最近有资料表明，最大化最小边距不一定能使得分类器表现得更好。 有人认为，在boosting的情况下，贪婪地最大化平均边距的方法要比最大化最小边距的方法更可取。 图8是关于$H(x_i)$的值在每次迭代之后的分布。 直方图中明亮和黑暗的条状分别代表类2和4，还有optdigits数据集。 值得注意的是，随着迭代过程的进行，这些类变得越来越分开。 \subsubsection{Convergence} 根据定理1，SemiBoost指数级收敛。接下来的节证明了SemiBoost在一个样例数据集上的收敛。 为了说明收敛，我们选择了optdigits数据集中最稠密了两个类，分别是数字2和数字4。 图9a演示了在迭代过程中，目标函数在不断有新的分类器加入的情况下的变更。 目标函数的变更是遵循指数下降的。 图9b展示了$\alpha$在迭代过程中的值。 最初，$\alpha$的值急速下降，接着在大概20次迭代之后，$\alpha$的值跟最初的分类器比起来就微不足道了。 这表明了，尽管SemiBoost仍然需要更多迭代来达到收敛，但是新加进去的分类器在促进方面不会有重大的影响。 图9c展示了SemiBoost以Decision Stump作为基分类器的精确度。 \subsubsection{Comparison with AdaBoost} 为了评估未标记数据在提高一个基分类器的性能的贡献，我们基于相同的基分类器（或者说弱学习器）、使用像在Section4.2中相似的实验步骤。 表3展示了3个基分类器：Decision Stump、J48和SVM（第一列）在第一行的六个数据集上的性能。 对每个分类器来说，头两行展示了分类器的归纳能力和它的使用10个已标记样本进行训练的加强版本（使用AdaBoost）。 第三行展示了在未标记数据增加到已标记样本集中时SemiBoost的性能。 第四和第五行，被标记为$large$和$AB-large$，展示了分类器的性能和它的在给用于SemiBoost的未标记数据分配标签之后的数据上训练过后得到的加强版本。\\ 从表3中，我们可以看到SemiBoost强化过后的分类器（SB-DS，SB-J48，和SB-SVM）都要显著地比仅使用已标记数据来强化的（AdaBoost）和没有使用boosting算法的分类器要好。 当然，当所有的未标记数据都被分配标签后，分类器和他们的boosted版本在性能上都要显著地强于SemiBoost。 跟基分类器相比较，AB-small在几个数据集上的归纳能力的下降，可能缘由过少训练样本导致的过拟合。 SemiBoost中，不断增加的未标记数据就像一个正则化机制，避免了过拟合，也因此能够实现一个能力有所提高的分类器。 \section{PERFORMANCE ON TEXT CATEGORIZATION} 我们使用备受欢迎的20-newsgroups数据集来解决文本分类问题，进一步评估SemiBoost算法。 我们用DS，J48和SVM作为基分类器进行了评估，解决使用了10个最受欢迎的数据集创建的二分类问题。 注意，这次实验设置跟一些其他的关于半监督学习、采用一对多方法分类的研究有所不同。 跟一对多的方法相比，一对一的评估方法有以下的优点： \begin{itemize} \item 在二分类任务中，监督分类器的最好性能之间有很大的差距。这使我们能够展示：当SVM对某个问题不是最好的分类器是，那么使用未标记数据来提升的SVM可能不是最好的半监督算法。 \item 半监督学习算法依赖于特定的关于数据集、类的结构的假设。在一对多方法中，这些假设很可能都被违反了。举个例子，许多半监督算法假设两个类之间有巨大的团簇间隙。通过把许多类聚集为负类，我们希望能够看到负类中存在着巨大的团簇间隙。流形假设的违反也可以用相似的思路来解释。 \item 当我们采用一对多的分类方法是，数据中有高度的不平衡性。虽然应用先验知识结合这种不平衡性到半监督学习中有助于提高性能，但是我们的假设是：除了相关性信息和一些训练实例，我们对数据一无所知。 \item 一对一已经是非常受欢迎的用来创建多类分类器的方法了。在一对一方法下，使用基于DAG的架构，测试时间可以得到非常显著的减少。 \end{itemize} 我们生成了对于这10个类可能产生的45个二分类问题。优于空间有限，我们仅仅把由5个类组成的10个二分类问题的结果包括进来。总结在表4中。 这些结果跟45个二分类问题的结果非常详细。 表4的第一列展示了10个二分类问题。 每个分类任务都包含了大约有2000份文档的数据集。 我们使用备受欢迎的tf-idf特征，也就是在2000份文档中总共出现10次以上的词语。 稍后，tf-idf特征在每份文档中被规范化。 每个数据集的维度展示在表4的列2上。 我们跟从Section4.2中同样的归纳评估过程。 我们使用每个类中的两个已标记样本来训练分类器。 我们使用线性核（特征矢量的点乘）作为相似性，这在文本分类任务中很常见。 Decision Stump， J48, SVM，还有他们的Boosted版本，TSVM，ILDS，LapSVM，这些不同算法的性能都展示在表4中。 所有SVM的参数值$C$都设置为1，保证性能的比较是公平的。 性能的平均误差和标准误差都是在20次重复试验过后才记录的。\\ 表4说明，在使用DS和J48的情况下，SemiBoost显著（95\%的置信度水平，使用独立样本配对t-test测量所得）提高了在所有类对上的性能。 SVM在5个类对上的性能得到了极大提升。 而且我们也注意到，对于所有的类对，SemiBoosted版本的DS比SemiBoosted版本的SVM表现得更好。 通过比较基于SVM的方法，SB-SVM在7个类对上明显优于LapSVM，在10个类对上明显优于ILDS。 TSVM在5个类对上优于SB-SVM。总的来说，SB-SVM跟TSVM不相上下，但是要比LapSVM和ILDS好很多。SB-DS在5个类对上优于TSVM，同时在所有类对上优于LapSVM和ILDS。 ILDS的差劲的性能可能要归咎于ILDS使用的核函数。 ILDS使用一个基于图距离的核函数，这可能不适用于基于文本分类的任务。 从实验中，我们可以看到，SemiBoosting Decision Stumps是一个可替代基于SVM的半监督学习的可行方法。 \section{CONCLUSIONS AND FUTURE WORK} 我们通过一个boosting框架，推出了一个用于半监督学习的算法。SemiBoost的用处在于，在有未标记样本的时候，它能提高任何给定的基分类器性能。 总的来说，在UCI数据集和文本分类数据集上的结果说明了这个方法的可行性。 SemiBoost的性能能与当前最先进的半监督学习算法相提并论。 我们所观察到SemiBoost的稳定性，表明它可以在实践中起到十分大的用处。 SemiBoost，就像几乎所有其他的半监督学习分类算法一样，当前是一个二分类算法。我们正在通过重定义一致性的测量来探索多分类的扩展。 我们正在致力于获得相关的理论性成果，在相似矩阵揭示潜在的数据结构时，能够保证SemiBoost的性能。 \section{ACKNOWLEGEMENTS} 作者在此感谢所有匿名审核者给出的宝贵的意见。本实验由US Office of Naval Research grant no.N000140710225 和 US National Science Foundation grant no. IIS-0643494提供部分支持 \section{REFERENCES} 翻不动了。 \hfill 杨健威 \hfill January 14, 2019 \end{document} --- --- @article{biggs_97, author = { and }, journal = {Appl. Opt.}, keywords = {Algorithms; Atmospheric turbulence; Image restoration; Linear filtering; Phase retrieval; Space telescopes}, number = {8}, pages = {1766--1775}, publisher = {OSA}, title = {Acceleration of iterative image restoration algorithms}, volume = {36}, month = {Mar}, year = {1997}, url = {http://ao.osa.org/abstract.cfm?URI=ao-36-8-1766}, doi = {10.1364/AO.36.001766}, abstract = {A new technique for the acceleration of iterative image restorationalgorithms is proposed. The method is based on the principles of vectorextrapolation and does not require the minimization of a cost function. Thealgorithm is derived and its performance illustrated withRichardson--Lucy (R--L) and maximum entropy (ME) deconvolutionalgorithms and the Gerchberg--Saxton magnitude and phase retrievalalgorithms. Considerable reduction in restoration times is achieved withlittle image distortion or computational overhead per iteration. The speedupachieved is shown to increase with the number of iterations performed and iseasily adapted to suit different algorithms. An example R--L restorationachieves an average speedup of 40 times after 250 iterations and an ME method20 times after only 50 iterations. An expression for estimating theacceleration factor is derived and confirmed experimentally. Comparisons withother acceleration techniques in the literature reveal significantimprovements in speed and stability.}, } t2s/doc/programming-guide/rewriting-math-equations-into-basic-ures.tex \chapter{Rewriting math equations into basic UREs} Based on the work of Evans~\cite{broadcastElim,broadcastElimQR}, we can translate math equations into UREs, following the simple steps below. Going through an example, we would find that this process is pretty much mechanical and intuitive, and only middle-school level of math knowledge is needed. So do not be scared by the math symbols. In fact, you want to work out the UREs through the simple math process, since the correctness can be proved and that would save you time in debugging. We strongly recommend a beginner tries one example so as to quickly master the skills. These skills are generally applicable to real world problems. Use auto-regressive filter~\cite{autoRegressiveModel} for example: \begin{equation} X_t=c+\sum\limits^I_{i=1}\varphi_i X_{t-i}+\varepsilon_t \end{equation} where $c$ is a constant, $\varphi_1, ... \varphi_I$ are the parameters, $\varepsilon _t$ is white noise, and $X_t$ is the output at the current time $t$ and is dependent on the previous $I$ outputs $ X_{t-1}, ..., X_{t-I}$. Table~\ref{tab:deriving-ures-for-auto-regressive-filter} shows a complete process how UREs are derived for the problem, and the steps are explained in more detail below: \begin{table*}[!ht] \begin{tabular}{|l|ll|ll|} \hline \multicolumn{1}{|c|}{\textbf{Step}} & \multicolumn{2}{|c|}{\textbf{Equations}} & \multicolumn{2}{|c|}{\textbf{Initial values}} \\\hline\hline \multirow{1}{*}{1: Iterative form} & $X_t=c+\sum\limits^I_{i=1}\varphi_i X_{t-i}+\varepsilon_t$ & $t=1, ..., T$ & $X_{t-i}=0$ &if $t-i \le 0$ \\\hline \multirow{2}{*}{2: Recursive form} & $X_t=X_t+\varphi_i X_{t-i}$ & $t=1,...,T$ & $X_t=c+\varepsilon_t$ & \\ & & $i=1,..,I$ & $X_{t-i}=0$ &if $t-i \le 0$ \\\hline \multirow{2}{*}{3: DSA} & $X_t^i=X_t^{i-1}+\varphi_i X_{t-i}^I$ & $t=1,...,T$ & $X_t^{0}=c+\varepsilon_t$ & \\ & & $i=1,..,I$& $X_{t-i}^I=0$ &if $t-i \le 0$\\\hline \multirow{2}{*}{4: Full index form} & $X_t^i=X_t^{i-1}+\varphi_i^0 X_{t-i}^I$ & $t=1,...,T$ & $X_t^{0}=c+\varepsilon_t^0$ & \\ & &$i=1,..,I$ & $X_{t-i}^I=0$ &if $t-i \le 0$\\\hline \multirow{4}{*}{5: UREs} & \multicolumn{2}{|l|}{$\Phi_t^i=(t-1=0) ? \varphi_i^0 : \Phi_{t-1}^i $} & & \\ & \multicolumn{2}{|l|}{$\chi_t^i=(t-1=0) ? 0 : (i-1=0 ? X_{t-1}^{i-(1-I)} : \chi_{t-1}^{i-1})$}&&\\ & $X_t^i=(i-1=0 ? c+\varepsilon_t^0 : X_t^{i-1})+\Phi_t^i\chi_t^i$ &$t=1,...,T$ & &\\ & & $i=1,..,I$& & \\\hline \end{tabular} \caption{Deriving UREs for auto-regressive filter. Here $c? a : b$ is an expression returning $a$ when condition $c$ is true, and $b$ otherwise.} \label{tab:deriving-ures-for-auto-regressive-filter} \end{table*} \section{Step 1: Iterative form} First, write down the math equation(s) of the original problem. Usually, in an equation, a domain is iterated (like $i=1,...,I$ and $t=1,...,T$), and some variable (like $X$) is computed. A variable might have some initial values (like $X_{t-i}=0$ for $t-i \le 0$). \section{Step 2: Recursive form} Translating the iterative form to a recursive form is straightforward: according to the iterative form, initialize a variable (e.g. $X_t=c+\varepsilon_t$), and update the variable every iteration with a new value based on its previous value (in the form of $X_t=X_t+...$). \section{Step 3: DSA (Dynamic Single Assignment)} When updating a variable every iteration, save it to a distinct memory location. Every reference to a variable thus exposes the iteration in which the variable is defined. For example, $X_t=X_t + ... X_{t-i}$ is changed into $X_t^i=X_t^{i-1} + ... X_{t-i}^I$. After this renaming, it is clear that the 3 references to $X$ are referring to the $X$ values defined in the current iteration $\begin{psmallmatrix}i\\t\end{psmallmatrix}$ and previous iterations $\begin{psmallmatrix}i-1\\t\end{psmallmatrix}$ and $\begin{psmallmatrix}I\\t-i\end{psmallmatrix}$, respectively. Consequently, dataflow/dependences between these iterations are made explicit. For another example, the initialization $X_t=c+\varepsilon_t$ is changed into $X_t^0=c+\varepsilon_t$ by adding one 0 for the missing index $i$. $X_t^0$ refers to the $X$ value defined in iteration $\begin{psmallmatrix}0\\t\end{psmallmatrix}$, which is outside the domain, because index $i$ starts from 1 with a step of 1. In general, if index $i$ starts from $s$ with a step $h$, the initial value should be $X_t^{s-h}$. \section{Step 4: Full index form} Now variables are referenced with full indices, but constants are not: $\varphi_i$ and $\varepsilon_t$ are inputs never modified when computing $X_t$ (The other constant $c$ is just a number and we do not care). Similar to the handling of initialization in step 3, we give these constants full indices by adding 0's for the missing index $i$: change $\varphi_i$ and $\varepsilon_t$ into $\varphi_i^0$ and $\varepsilon_t^0$. They refer to the $\varphi$ and $\varepsilon$ values defined in iteration $\begin{psmallmatrix}0\\i\end{psmallmatrix}$ and $\begin{psmallmatrix}0\\t\end{psmallmatrix}$, respectively, which are outside the domain. In general, if iteration index $i$ starts from $s$ with a step $h$, we should rename $\varphi_i$ and $\varepsilon_t$ into $\varphi_i^{s-h}$ and $\varepsilon_t^{s-h}$. After being full indexed, variables and constants will be processed in the same way. At this point, the equations we get are AREs (Affine Recurrence Equations), that is, the current iteration $\begin{psmallmatrix}i\\t\end{psmallmatrix}$ reads a value defined in a previous iteration with a distance $d$ that is in the form of $d=A\begin{psmallmatrix}i\\t\end{psmallmatrix}+d_0$, where $A$ is a matrix and $d_0$ a constant vector. In other words, there is a read-after-write affine dependence between the two iterations. The dependence distance can be calculated as the current iteration - the previous iteration = the write's index - the read's index: Remember that every write/read is fully indexed with the iteration that defines its value; thus the write is indexed with the current iteration, the read is indexed with the previous iteration. See Table~\ref{tab:deps-full-index-form-auto-regressive-filter} for all the dependences: \begin{table*}[!ht] \begin{tabular}{|c|c|c|l|} \hline \multicolumn{1}{|c|}{\textbf{Dependence No.}} & \multicolumn{1}{|c|}{\textbf{Write}} & \multicolumn{1}{|c|}{\textbf{Read}} & \multicolumn{1}{|c|}{\textbf{Dependence distance}}\\\hline\hline \multirow{1}{*}{1} & $X_t^i$ & $X_t^{i-1}$ & $\begin{pmatrix} i\\t\end{pmatrix} - \begin{pmatrix} i-1 \\ t\end{pmatrix} = \begin{pmatrix} 1 \\ 0\end{pmatrix}$ \\\hline \multirow{1}{*}{2} & $X_t^i$ & $\varphi_i^{0}$ & $\begin{pmatrix} i\\t\end{pmatrix} - \begin{pmatrix} 0 \\ i \end{pmatrix} = \begin{pmatrix} i \\ t-i\end{pmatrix} = \begin{pmatrix} 1 & 0 \\ -1 & 1\end{pmatrix} \begin{pmatrix} i \\ t\end{pmatrix}$ \\\hline \multirow{1}{*}{3} & $X_t^i$ & $X_{t-i}^{I}$ & $\begin{pmatrix} i\\t\end{pmatrix} - \begin{pmatrix} I \\ t-i \end{pmatrix} = \begin{pmatrix} i-I \\ i\end{pmatrix} = \begin{pmatrix} 1 & 0 \\ 1 & 0\end{pmatrix} \begin{pmatrix} i \\ t\end{pmatrix} + \begin{pmatrix} -I \\ 0 \end{pmatrix}$ \\\hline \end{tabular} \caption{Dependences of the full index form in step 4 of Table~\ref{tab:deriving-ures-for-auto-regressive-filter}}. \label{tab:deps-full-index-form-auto-regressive-filter} \end{table*} \section{Step 5: UREs} We translate AREs into UREs by converting a broadcast into a pipeline. After that, every dependence has a constant distance uniformly in the entire domain. There are two ways to convert a broadcast into a pipeline, either graphically or mathematically. \subsection{First way to translate AREs into UREs: drawing a dataflow graph} \label{sec:are-to-ures-with-dfg} We can draw the dataflow and intuitively figure out how to change a broadcast into a pipeline, as exemplified in Fig.~\ref{fig:broadcast-to-pipeline-arf}. According to the 2nd dependence in Table~\ref{tab:deps-full-index-form-auto-regressive-filter}, originally, a datum $\varphi_i^0$ is broadcast to iterations $X_t^i$ for all $t$, as shown in the left of the figure. Equivalently, the same datum can be loaded in an iteration at a boundary of the domain, and from that iteration, propagated in a pipeline fashion to all the other iterations, as shown in the right of the figure. As we can see, $\varphi_i^0$ is loaded at a bounary iteration $(i, 1)$, and then is propagated to iteration $(i, 2)$, and from there to iteration $(i, 3)$, etc. \begin{figure}[!ht] \centering \includegraphics[width=\textwidth]{./img/broadcast-to-pipeline-arf.png} \caption{For the 2nd dependence of auto-regressive filter in Tabel~\ref{tab:deps-full-index-form-auto-regressive-filter}, the broadcasts due to this dependence are changed to pipelines. Here we assume $T=3$, and every point is an iteration annotated with its indices $\begin{psmallmatrix} i\\ t\end{psmallmatrix}$.} \label{fig:broadcast-to-pipeline-arf} \end{figure} \subsubsection{Expressing a pipeline} \label{sec:express-pipeline} Now we can modify the full index form \begin{equation} X_t^i=...\varphi_i^0... \end{equation} into \begin{equation} X_t^i=...\Phi_t^i... \end{equation} where $\Phi$ values are propagated along the $t$ dimension until out of the domain: \begin{equation} \Phi_t^i=(t-1=0) ? \varphi_i^0 : \Phi_{t-1}^i \end{equation} As you can see, the keys to change a broadcast into a pipeline are: (1) the direction of the pipeline, along which data would be propagated from one iteration to the next, and (2) the boundary conditions when the pipeline is fed with some initial values. In the same way, we can convert a broadcast due to the third dependence into a pipeline (Could you do it?). After that, we get the UREs shown in step 5 of Table~\ref{tab:deriving-ures-for-auto-regressive-filter}. \subsection{Second way to translate AREs into UREs: Calculating a propagation direction} \label{sec:are-to-ures-with-math} We can generalize the example in Fig.~\ref{fig:broadcast-to-pipeline-arf}, and directly find out a propagation direction vector in simple math: \begin{quotation} \noindent If a read-after-write dependence is affine, i.e. the distance $d$ is in the form of $Az+d_0$, where $A$ is a matrix, $z$ is the current iteration, and $d_0$ is a constant vector, then the dependence incurs broadcasts of values, and such a broadcast can be changed into a pipeline by propagating a value along a direction $r$ that is a solution of $(E-A)r=\mathbf{0}$, where $E$ is the identity matrix. \end{quotation} \begin{figure}[!ht] \centering \includegraphics[width=\textwidth]{./img/broadcast-to-pipeline.png} \caption{Changing a broadcast into a pipeline} \label{fig:broadcast-to-pipeline} \end{figure} The left part of Fig.~\ref{fig:broadcast-to-pipeline} shows that due to an affine dependence, some data are broadcast to multiple consumers, including an iteration $b$ at the boundary of the domain, and two other iterations $c_1$ and $c_2$, etc. Remember that data are fully indexed, i.e. they can be considered having been defined in an iteration $y$, even though that iteration is actually out of the domain. Let the dependence distance be $d=Az+d_0$. Because both iteration $c1$ and $c_2$ get data from the same iteration $y$, using the dependence distance, we have \begin{equation} y = c_2 - (Ac_2+d_0) = c_1 - (Ac_1+d_0) \end{equation} So \begin{equation} (E-A)c_2 = (E-A)c_1 \end{equation} Let $r=c_2-c1$, we have \begin{equation} (E-A)r = 0. \end{equation} Therefore, we can imagine that a datum that is defined outside the domain is first sent to a boundary iteration $b$, and then following a propagation direction $r$ to the next iteration $b+r$, and so on, and eventually the data reach iteration $c_1$, and then $c_2$, etc. This constructs a pipeline as shown in the right part of Fig.~\ref{fig:broadcast-to-pipeline}. For example, the second dependence for the auto-regressive filter is $\begin{psmallmatrix} 1 & 0 \\ -1 & 1\end{psmallmatrix} \begin{psmallmatrix} i \\ t\end{psmallmatrix}$ (Table~\ref{tab:deps-full-index-form-auto-regressive-filter}), and thus $A=\begin{psmallmatrix} 1 & 0 \\ -1 & 1\end{psmallmatrix}$. Solve the equation $(E-A)r=\begin{psmallmatrix} 0 & 0 \\ 1 & 0\end{psmallmatrix}r=\mathbf{0}$. We get a solution $r=\begin{psmallmatrix} 0 &\\ *\end{psmallmatrix}$, where the second element (the $t$ dimension) can be arbitrary. Take a non-zero solution $r=\begin{psmallmatrix} 0 &\\ 1\end{psmallmatrix}$, and we see exactly from the right part of Fig.~\ref{fig:broadcast-to-pipeline-arf} that a pipeline can be constructed along the $t$ dimension for a datum. With this propagation direction, we convert a broadcast due to the second dependence into a pipeline in the same way as shown in Section~\ref{sec:express-pipeline}. Following the same approach, for the third dependence, we can find a propagation direction $r=\begin{psmallmatrix} 1 &\\ 1\end{psmallmatrix}$, and convert the broadcasts due to this dependence into pipelines as well (Could you do it?). After that, we get the UREs shown in step 5 of Table~\ref{tab:deriving-ures-for-auto-regressive-filter}. \subsection{Testing correctness of UREs in standard C} \label{sec:test-correctness-ures} We can express the UREs in Table~\ref{tab:deriving-ures-for-auto-regressive-filter} in standard C, adding necessary helping code for testing, and see if the UREs can produce correct results. See Listing~\ref{lst:test-correctness-ures-auto-regressive-filter-in-c}. \lstinputlisting[language=c, caption={Testing the correctness of the UREs in Table~\ref{tab:deriving-ures-for-auto-regressive-filter} in standard C code.}, label={lst:test-correctness-ures-auto-regressive-filter-in-c}]{code/auto-regressive-filter-testing-ures-in-c.c} Now test it: \begin{verbatim} $gcc auto-regressive-filter-testing-ures.c $./a.out Success! \end{verbatim} \subsection{Expressing the UREs in T2S} \label{sec:express-ures-in-t2s} It is straightforward to express the UREs in T2S. See Listing~\ref{lst:test-correctness-ures-auto-regressive-filter-in-t2s}. The code is similar to the C code before. The major difference is that the T2S code builds a symbolic dataflow graph where inputs are symbolic parameters and computation is symbolic expressions (Line 13-27), then instantiates the graph to compute with concrete inputs (Line 29-37), and after that compares with a golden baseline for correctness. \lstinputlisting[language=c, caption={Testing the correctness of the UREs in Table~\ref{tab:deriving-ures-for-auto-regressive-filter} in T2S.}, label={lst:test-correctness-ures-auto-regressive-filter-in-t2s}]{code/auto-regressive-filter-testing-ures-in-t2s.cpp} Now test it: \begin{verbatim} $ export T2S_PATH=path_to_your_t2s_installation $ export PATH=$T2S_PATH/Halide/bin:$T2S_PATH/install/gcc-7.5.0/bin:$PATH $ export LD_LIBRARY_PATH=$T2S_PATH/Halide/bin:$LD_LIBRARY_PATH $ g++ -I$T2S_PATH/Halide/include -L$T2S_PATH/Halide/bin -lHalide -std=c++11 \ auto-regressive-filter-testing-ures-in-t2s.cpp $ ./a.out Success! \end{verbatim} 0 % Capítulo de libro que explica counterfactual explanations: https://christophm.github.io/interpretable-ml-book/counterfactual.html. Incluye referencias a algunas técnicas interesantes para obtener explicaciones contrafácticas de forma agnóstica al modelo de machine learning. % -> Uso de metaheurísticas multiobjetivo (minimizar cambios en variables, nºvariables cambiadas,etc): @inproceedings{dandl2020multi, title = {Multi-objective counterfactual explanations}, author = { and }, booktitle = {International Conference on Parallel Problem Solving from Nature}, pages = {448--469}, year = {2020}, organization = {Springer} } % -> Basado en resolver problemas de satisfacibilidad @inproceedings{karimi2020model, title = {Model-agnostic counterfactual explanations for consequential decisions}, author = { }, booktitle = {International Conference on Artificial Intelligence and Statistics}, pages = {895--905}, year = {2020}, organization = {PMLR} } % -> Basado en determinantal point processes @inproceedings{mothilal2020explaining, title = {Explaining machine learning classifiers through diverse counterfactual explanations}, author = { and and }, booktitle = {Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency}, pages = {607--617}, year = {2020} } % Estas técnicas podrían ser aplicados sobre distintos **clasificadores** interpretables (uno por cada ola) para un estudio de variables cuyo cambio invierte el resultado final de mortalidad del paciente a uno de supervivencia. Sobre esta idea puede tomarse un enfoque más pragmático, e.g. estudiando variables en control del médico y/o paciente (estado de vacunación, medicamentos administrados...), o centrándose en obtener un algoritmo de búsqueda de explicaciones contrafácticas maximizando la usabilidad (pocos hiperparámetros, ponderación de variables a cambiar con métodos basados en preguntas en vez de asignación manual de pesos, etc.) % En el libro se mencionan otras técnicas de interpretabilidad agnósticas al modelo que podría ser interesante explorar. Por ejemplo uso de Partial dependence plot para el estudio del impacto de las variables que se sospecha que cambian significativamente el modelo de razonamiento en las nuevas olas, como el estado de vacunación. Puede profundizarse más la idea y estudiar como empiezan o dejan de ser relevantes para un clasificador algunas variables en las distintas olas. % #################### % Puede también centrarse el foco en las técnicas de obtención y aplicación de explicaciones contrafácticas que sean especiales de las redes Bayesianas: % Este artículo sienta unas bases de notación para el problema. Además nos indica que no es posible realizar consultas intercausales a una red Bayesiana si no utilizamos una estructura simétrica mundo hipotético - mundo real, con un puente de disturbance variables. No tengo claro ahora mismo como se conseguirían estas variables. @article{balke2011probabilistic, title = {Probabilistic evaluation of counterfactual queries}, author = { and }, year = {2011} } % Bien es cierto que este artículo se refiere a clasificadores Bayesianos, sin embargo, vale la pena acordarse de este trabajo si se acaban utilizando en el enfoque del TFM adicionalmente/en vez de redes Bayesianas en general. @inproceedings{albini2020relation, title = {Relation-Based Counterfactual Explanations for Bayesian Network Classifiers.}, author = { and and and }, booktitle = {IJCAI}, pages = {451--457}, year = {2020} } % Creo que la idea de este artículo podría trasladarse a uno de los objetivos del TFM: análisis forense: ¿Qué podría haber hecho para evitar la muerte? @article{constantinou2016value, title = {Value of information analysis for interventional and counterfactual Bayesian networks in forensic medical sciences}, author = { and and }, journal = {Artificial Intelligence in Medicine}, volume = {66}, pages = {41--52}, year = {2016}, publisher = {Elsevier} } % Restricciones de factibilidad a los ejemplos contrafactuales, aplicado además sobre Redes Bayesianas @article{mahajan2019preserving, title = {Preserving causal constraints in counterfactual explanations for machine learning classifiers}, author = { and and }, journal = {arXiv preprint arXiv:1912.03277}, year = {2019} } % Se utiliza este tipo de razonamiento aplicado sobre redes Bayesianas. Cabe notar que el artículo advierte de la necesidad de que la red modele relaciones causales; esto podría llegar a ser un problema, como creo que se comentó en la última reunión. @article{richens2019counterfactual, title = {Counterfactual diagnosis}, author = { and and }, journal = {arXiv preprint arXiv:1910.06772}, year = {2019} } % Uso de términos médicos en el desarrollo de Redes Bayesianas, no es inmediatamente relevante en la decisión de la dirección del TFM, pero podría ser útil según el enfoque que se decida. @article{kyrimi2020medical, title = {Medical idioms for clinical Bayesian network development}, author = { Neves, and McLachlan, Scott and and and }, journal = {Journal of Biomedical Informatics}, volume = {108}, pages = {103495}, year = {2020}, publisher = {Elsevier} } % Teoría de unificación de contrafactuales y modelos gráficos causales. Parece muy potente, pero plantea muchos desafíos en términos de complejidad e implementación. @article{richardson2013single, title = {Single world intervention graphs (SWIGs): A unification of the counterfactual and graphical approaches to causality}, author = {Richardson, and Robins, }, journal = {Center for the Statistics and the Social Sciences, University of Washington Series. Working Paper}, volume = {128}, number = {30}, pages = {2013}, year = {2013}, publisher = {Citeseer} } % Revisa dos aproximaciones a este razonamiento en Redes Bayesianas, a priori no lo he entendido muy bien. @article{rips2010two, title = {Two causal theories of counterfactual conditionals}, author = {}, journal = {Cognitive science}, volume = {34}, number = {2}, pages = {175--221}, year = {2010}, publisher = {Wiley Online Library} } % Tampoco he entendido demasiado bien este trabajo, parece una propuesta de un tipo de grafo para explicar un proceso de inferencia. Queda pendiente una lectura más a fondo @article{jaimini2022causalkg, title = {CausalKG: Causal Knowledge Graph Explainability using interventional and counterfactual reasoning}, author = { and }, journal = {arXiv preprint arXiv:2201.03647}, year = {2022} } % Uso de contrafactuales con árboles de decisión. Esta incompleto? @inproceedings{sokol2019desiderata, title = {Desiderata for interpretability: explaining decision tree predictions with counterfactuals}, author = { and }, booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence}, volume = {33}, number = {01}, pages = {10035--10036}, year = {2019} } % Uso de contrafactuales en Redes Bayesianas, tiene mejor pinta que otros enfoques más teóricos. Eso sí, parece más bien enfocado a explicar que a encontrar alternativas factibles @inproceedings{koopman2021persuasive, title = {Persuasive contrastive explanations for Bayesian networks}, author = { and }, booktitle = {European Conference on Symbolic and Quantitative Approaches with Uncertainty}, pages = {229--242}, year = {2021}, organization = {Springer} } % Framework de explicación con contrafactuales. Documento muy pesado (es una tesis doctoral!), pero incluye usos con Redes Bayesianas. Quizás más apropiado para el SOTA que para otra cosa @phdthesis{wijaya2021multilayer, title = {Multilayer Counterfactual Explanations for Machine Learning Classifiers}, author = {}, year = {2021}, school = {Imperial College London} }Bibtex/A Fast and Accurate Dependency Parser using Neural Networks.bib @InProceedings{chen2014afast, author = { and }, title = {A Fast and Accurate Dependency Parser using Neural Networks}, booktitle = {Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)}, year = {2014}, publisher = {Association for Computational Linguistics}, pages = {740--750}, location = {Doha, Qatar}, doi = {10.3115/v1/D14-1082}, url = {http://www.aclweb.org/anthology/D14-1082} }% -*-latex-*- \title{Orbiter Technical Notes: Inertia calculations for composite vessels} \author{} \date{September 28, 2017} \documentclass[a4paper]{article} \usepackage[dvips]{graphicx} \usepackage{amsmath} \usepackage{amsfonts} \usepackage{cite} \usepackage{setspace} \usepackage{times} \newcommand\mat[1]{\mathsf{#1}} \begin{document} \maketitle \section{Introduction} This document describes the method Orbiter uses to define the inertial behaviour for composite vessels from the parameters of its components. \section{Definitions} To describe the dynamics of a rigid body we use Euler's rotation equations that describe the rotation of the body in the rotating system: \begin{equation} \mat{I} \dot{\vec{\omega}} + \vec{\omega} \times (\mat{I}\vec{\omega}) = \vec{M} \end{equation} where $\vec{\omega}$ are the components of angular velocity around the three main vessel axes, $\vec{M}$ are the applied torques, and $\mat{I}$ is the inertia matrix. In the principal body frame, the inertia matrix is diagonal, where diagonal elements $I_n\;(n=1,2,3)$ are the principal moments of inertia (PMI). We then get \begin{equation} \begin{split} I_1 \dot\omega_1 + (I_3 - I_2) \omega_2 \omega_3 &= M_1 \\ I_2 \dot\omega_2 + (I_1 - I_3) \omega_3 \omega_1 &= M_2 \\ I_3 \dot\omega_3 + (I_2 - I_1) \omega_1 \omega_2 &= M_3 \end{split} \end{equation} Note that Orbiter assumes vessels to be defined in the principal frame but does not enforce it, that is, only the diagonal elements of $\mat{I}$ are considered. For most vessel layouts, an intrinsic symmetry is present, and the vessel frame naturally aligns closely with the principal frame, so that neglecting the off-diagonal elements of $\mat{I}$ incurs a small error. For composite structures this may not be true, so a more general framework may be introduced in the future. \section{Composite structures} In spaceflight scenarios, rigid structures may commonly be formed by connecting individual components (launch stack composed from individual stages, spacecraft docked at a space station, Apollo CSM+LM assembly, etc.) These rigid assemblies act as a single body under rotation, governed by its own composite PMI. While the composite PMI may be pre-calculated for commonly encountered arrangements, not all combinations are predictable. In Orbiter, any vessels with suitable docking ports can be connected to form arbitrary new ``super-vessels'' of two or more components. It is therefore necessary to find a method to compute the composite PMI from the inertia parameters of the individual components and the geometry of the assembly. Orbiter uses the following method: \begin{itemize} \item Represent a vessel by a small number of discrete point masses arranged so that they generate the same PMI as the original vessel. \item Transform the point masses of all individual vessels into the common barycentric frame of the super-vessel. \item Compute the PMI of the super-vessel by summing over the point masses of all individual components. \end{itemize} \subsection{Point mass representation of a vessel} In general, the PMI components of an object occupying a volume $V$ with mass density distribution $\rho(\vec{r}),\; \vec{r}\in V$ are given by \begin{equation} \begin{split} I_1 &= \int_V \rho(\vec{r}) (y^2 + z^2) d\vec{r} \\ I_2 &= \int_V \rho(\vec{r}) (z^2 + x^2) d\vec{r} \\ I_3 &= \int_V \rho(\vec{r}) (x^2 + y^2) d\vec{r} \end{split} \end{equation} with $\vec{r} = \lbrace x,y,z \rbrace$. For an object composed of $n$ individual point masses $m_i$ at locations $\vec{r}_i$ this simplifies to \begin{equation} \begin{split} I_1 &= \sum_{i=1}^n m_i (y_i^2 + z_i^2) \\ I_2 &= \sum_{i=1}^n m_i (z_i^2 + x_i^2) \\ I_3 &= \sum_{i=1}^n m_i (x_i^2 + y_i^2) \end{split} \end{equation} We can now represent a vessel with arbitrary PMI values by 6 point masses that produce the same PMI values: \begin{equation} \begin{split} \hat{\vec{r}}_{1,2} &= \lbrace \pm \hat x, 0, 0 \rbrace \\ \hat{\vec{r}}_{3,4} &= \lbrace 0, \pm \hat y, 0 \rbrace \\ \hat{\vec{r}}_{5,6} &= \lbrace 0, 0, \pm \hat z \rbrace \end{split} \end{equation} and $m_i = \hat m = m/6,\; i=1..6$, given vessel mass $m$. To compute the axis offsets $\hat x, \hat y, \hat z$ of the point masses we note that \begin{equation} \begin{split} I_1 &= \hat m (2 \hat y^2 + 2 \hat z^2) \\ I_2 &= \hat m (2 \hat z^2 + 2 \hat x^2) \\ I_3 &= \hat m (2 \hat x^2 + 2 \hat y^2) \end{split} \end{equation} and thus \begin{equation}\label{eq:smp_ofs} \begin{split} \hat x &= \frac{1}{2} \sqrt{\left| -\frac{I_1}{\hat m} + \frac{I_2}{\hat m} + \frac{I_3}{\hat m} \right|} \\ \hat y &= \frac{1}{2} \sqrt{\left| \frac{I_1}{\hat m} - \frac{I_2}{\hat m} + \frac{I_3}{\hat m} \right|} \\ \hat z &= \frac{1}{2} \sqrt{\left| \frac{I_1}{\hat m} + \frac{I_2}{\hat m} - \frac{I_3}{\hat m} \right|}. \end{split} \end{equation} It is left to the reader to confirm that this point sample arrangement generates the correct PMI values. \subsection{Super-vessel definition} A super-vessel is a composite structure arranged by rigid assembly of individual vessels where the geometry is defined by the position and orientation of the participating docking ports of the component vessels. The origin of the super-vessel frame is given by the barycentre of the assembly. It should be noted that the origin can shift continuously relative to the vessel assembly if the masses of the component vessels change (e.g. as a result of fuel consumption). The orientation of the super-vessel frame is currently arbitrarily set to the orientation of the first vessel in the assembly. No attempt is made to rotate to the principal frame. This may need some thought. With this arrangement, we can define transformations for each component vessel $j$ into the supervessel frame: \begin{equation} \vec{r}_j^S = \mat{R}_j \vec{r}_j + \vec{r}_j^0, \end{equation} where $\mat{R}_j$ is a rotation matrix and $\vec{r}_j^0$ is the position of the vessel CoG in the super-vessel frame. The transformation of the point-mass representations of all participating vessels into the super-vessel frame is given by \begin{equation} \hat{\vec{r}}_{ji}^S = \mat{R}_j \hat{\vec{r}}_{ji} + \vec{r}_j^0, \; j=1..N, \; i=1..6, \end{equation} where $N$ is the number of participating vessels. We can now assemble the components $I_n^S$ of the super-vessel PMI by summing over the point samples of all vessels: \begin{equation} \begin{split} I_1^S & = \sum_{j=1}^N \sum_{i=1}^6 \hat m_j ([\hat y_{ji}^S]^2 + [\hat z_{ji}^S]^2)\\ I_2^S & = \sum_{j=1}^N \sum_{i=1}^6 \hat m_j ([\hat z_{ji}^S]^2 + [\hat x_{ji}^S]^2)\\ I_3^S & = \sum_{j=1}^N \sum_{i=1}^6 \hat m_j ([\hat x_{ji}^S]^2 + [\hat y_{ji}^S]^2) \end{split} \end{equation} \section{Mass-normalised PMI} It should be noted that Orbiter uses \emph{mass-normalised} PMI values $\tilde I_n = I_n/m$, which has the advantage that PMI values can remain unchanged even if the vessel mass changes. All Orbiter API calls by convention return and expect $\tilde I_n$ instead of $I_n$. Applying this convention to the super-vessel definition yields \begin{equation} \tilde{I}_n^S = \frac{I_n^S}{\sum_{j=1}^N m_j} \end{equation} Likewise, for the computation of the sample offsets in Eq.~\ref{eq:smp_ofs} we can write \begin{equation} \hat x = \frac{1}{2} \sqrt{\left| -\frac{\tilde{I}_1 m}{\hat m} + \frac{\tilde{I}_2 m}{\hat m} + \frac{\tilde{I}_3 m}{\hat m} \right|} = \sqrt{\frac{3}{2} \left| -\tilde{I}_1 + \tilde{I}_2 + \tilde{I}_3 \right|} \end{equation} with corresponding expressions for $\hat y$ and $\hat z$. \end{document} 0 %\section{Impossibility of SNOW properties with two clients with restricted communication }\label{two-client} \section{Two Client Open Question} \label{sec:2c2s} %{ \color{blue} This section closes the open question of whether SNOW properties can be implemented in the MWSR setting. This subsumes the two-clients setting explicitly left open in~\cite{SNOW2016}. We first prove that SNOW remains impossible in a MWSR and 2-server system if C2C communication is disallowed.  Next, we present an algorithm that implements SNOW properties in an MWSR setting with at least two servers. Hence we resolve a more general version of the open question raised in~\cite{SNOW2016}: the feasibility of SNOW in this setting depends on whether C2C communication is allowed. %} % the presence of client to client communication, it is possible to have all SNOW properties with two clients and at least two servers. %Fig~\ref{fig:architecture2a}. \subsection{No SNOW Without C2C Messages} \label{subsec:no_snow_no_c2c} In this section, we use the same system model as in Section~\ref{sec:formal_proof}: two servers $s_x$ and $s_y$ with two clients, a reader $r_1$ that issues only \rots{} and a writer $w$ that issues only \wots{}. A \wot{} $W$ writes $(x_1, y_1)$ to $s_x$ and $s_y$, and a \rot{} $R$ reads both servers. We assume that there is a bi-directional communication channel between any pair of client and server and any pair of servers. There is no communication channel between clients. We assume that each transaction can be identified by a unique number, e.g., transaction identifier. %We denote the automata for servers $s_x$ and $s_y$ by $s_x$ and $s_y$, respectively, and the automata for the clients $r$ and $w$ by $r$ and $w$, respectively. %We denote by $S_1$ the subsystem consisting of $s_x$, $s_y$ and $w$, and by $S_1$ the automata composed of $s_x$, $s_y$ and $w$, i.e., $s_x \times s_y \times w$. %Also, we denote by $\mathcal{A}$, which can also be interpreted as the algorithm, the automaton representing the entire system consisting of $S_1$ and $r$ (i.e., $s_x \times s_y$ $ \times w$ $ \times r$). % We assume that between any client $c$, %$c \in \{r, w \}$ and any server $s$, $s \in \{s_x, s_y\}$ there are channel automata $Channel_{c, s}$ and $Channel_{s, c}$. %Again, with abuse of notation we denote by $\mathcal{S}$ the automata composed of the automatons $s_x$, $s_y$ and $w$. Also, by $\mathcal{T}$ we denote the composition of automaton $\mathcal{S}$ and $r$, i.e., represents the entire system. %We used the notation $\writeop{o}{v}$ to mean a write operation with value $v$ on object $o$; and similarly, by $\readop{o}$ we mean a read operation on $o$. %We denote by $S_2$ the system, and also the composed automaton, consisting of $s_x$, $s_y$, $r$ and $w$, by composing the automatons $S_1$ and $r$, i.e., the automaton $S_1 \times r$. %We assume that clients are non-faulty. %Here, we consider only executions of some algorithm ${\mathcal A}$ that satisfied SNOW properties. %For contradiction, we also assume that any execution of $\mathcal{A}$ respects the SNOW properties. % Also, we assume that in an execution $\epsilon$ of $\mathcal{A}$ we can identify each transaction with a unique identifier. %\sloppy Consider an execution of ${\mathcal A}$ with two transactions in it as: a \wot{} $W \equiv $ $WRITE($$ (o_1, x_1), (o_2, y_1) )$ invoked by $w$, where $x_1 \neq x_0$ and $y_1 \neq y_0$; and a \rot{} $R \equiv READ$$(o_1, o_2)$, invoked by $r_1$. %Let us denote by %$op_x^{r_1}$ and $op_y^{r_1}$ the read operations $\readop{o_1}$ and $ \readop{o_2} $, respectively. %Due to the wait-free requirement for the \wots{}, $W$ completes, and we denote by $\INV{W}$ the invocation action and $\RESP{W}$ the response action in $\alpha$. % %\paragraph{Notations and Definitions:} %We introduce the following notations and definitions in the context of an execution $\epsilon$, of $\mathcal{A}$, with transactions $R_1$ and $W$ in it. % %\begin{notation} %We introduce the following notations for actions relevant to $R$ in $\epsilon$, % where $ j \in \{1, 2\}$: %\begin{enumerate} %\item $\INV{R}$ and $\RESP{R}$: invocation and response actions for $R$, at $r$; %\item $\INV{W}$ and $\RESP{W}$: invocation and response actions for $W$, at $w$; %\item $\send{m_j^{r}}{r, s_j}$: an output action at $r$, which sends a message $m_j^{r}$ from reader $r$ to server $s_j$, requesting the value for $o_j$; %\item $\recv{m_j^{r}}{r, s_j}$: an input action at $s_j$, that receives the message $m_j^{r}$, sent from $r$; %\item $\send{v_j}{ s_j, r}$: an output action at $s_j$, that sends value $v_j$, for $o_j$, to $r$. %\item $\recv{v_j}{ s_j, r}$: an input action at $r$, to receive a message $v_j$ from $s_j$ at $r$. % %\end{notation} % %\begin{definition} %\item \emph{Non-blocking fragments $\frag{i}{\epsilon}$, $i \in \{1, 2\}$.} %Suppose there is a fragment of execution in $\epsilon$ where the first action is $recv(m_i^r)_{r, s_i}$ and the last action is %$send(v_i)_{s_i, r}$ , both of which occur at $s_i$. Moreover, suppose there is % no other input action at $s_i$ in this fragment. Then we call this execution fragment % a \emph{non-blocking response fragment} for $op_i^r$ at $s_i$. We use the notation $\frag{i}{\epsilon}$ to denote this fragment of %execution of $\epsilon$. %Similarly, if such an execution fragment occurs with $m_y^r$, $y$ and $s_y$ then we denote this by $\frag{1,y}{\epsilon}$ and refer to it as a non-blocking response fragment for $op_2^r$ at $s_y$. %\end{definition} %\begin{notation} %\item We use the notations $R(\epsilon)$ and $W(\epsilon)$ to denote the transactions $R$ and $W$, in $\epsilon$. When the underlying execution is clear from the context we simply use $R$ and $W$. %\end{notation} %\begin{notation} % \item If the non-blocking fragment $\frag{1,x}{\epsilon}$ appears in $\epsilon$ such that $recv(m^{r_1}_2)_{r, s_y}$, at $s_y$, does not occur before % $\frag{1,x}{\epsilon}$ completes and $\epsilon$ is of the form % $\finiteprefix{\ell-1}{\ell} \circ \frag{1,x}{\epsilon} \circ \frage{S}{\epsilon}{}$, where %$\frage{S}{\epsilon}{}$ is any continuation of the execution, then we denote $\finiteprefix{\ell-1}{\ell} $ by $\prefix{\epsilon}$. %\item \sloppy If $\epsilon$ is of the form % $\finiteprefix{\ell-1}{\ell} \circ \frag{1,x}{\epsilon} \circ \kappa \circ \frag{1,y}{\epsilon} , a_p, \sigma_p, \ldots$ or % $\finiteprefix{\ell-1}{\ell} \circ \frag{1,y}{\epsilon} \circ \kappa \circ \frag{1,x}{\epsilon} \circ \frage{S}{\epsilon}{}$, where % $\ell$ is a positive integer, % $\kappa$ is a segment of $\epsilon$, possibly even of length zero and $\frage{S}{\epsilon}{}$ is any suffix part of the execution % then we denote $\finiteprefix{\ell-1}{\ell} $ by $\prefix{\epsilon}$. Clearly, we can write $\epsilon$ as $\prefix{\epsilon} \circ \frag{1,x}{\epsilon}\circ \kappa \circ \frag{1,y}{\epsilon} \circ \suffix{\epsilon}$ or $\prefix{\epsilon} \circ \frag{1,y}{\epsilon}\circ \kappa \circ \frag{1,x}{\epsilon} \circ \suffix{\epsilon}$. %\end{notation} %\end{enumerate} %However, unlike in the case of \rots{}, for \wots{} the SNOW properties do not impose strict conditions on the occurrence of external events at the clients or the servers. %However, the W property guarantees termination of any \wot{}. Therefore, all we can claim is $\RESP{W}$ eventually appears after $\INV{W}$. Our strategy is still proof by contradiction: We assume there exists some algorithm $\mathcal{A}$ that satisfies all SNOW properties, and then we show the existence of a sequence of executions of $\mathcal{A}$, finally leading to an execution contradicting the S property. \remove{ %If $\epsilon$ is an execution of $\mathcal{A}$, with \rot{} $R$. % We denote by $\frag{1,x}{\epsilon}$ and $\frag{1,y}{\epsilon}$ the execution fragments of $\epsilon$ between the actions $recv(x)_{ s_x, r}$ and $send(m_x^r)_{r, s_x}$ at $s_x$; and $recv(x)_{ s_x, r}$ and $send(m_y^r)_{r, s_y}$ at $s_y$, respectively, and in $\frag{1,x}{\epsilon}$ and $\frag{1,y}{\epsilon}$ no other internal actions occur at $s_x$ and $s_y$, respectively. First, we show the existence of an execution $\alpha$ of $\mathcal{A}$ where $R_1$ is invoked after $W$ completes, where the send actions $send(m_x^{r_1})_{r_1, s_x}$ and $send(m_y^{r_1})_{r_1, s_y}$ at the $r_1$ occur consecutively in $P(\alpha)$, which is a prefix of $\alpha$. Then we show that $\alpha$ can be written in the form $\prefix{\alpha} \circ \frag{1,x}{\alpha}$ (Fig.~\ref{fig:execution1} $(a)$, Lemma~\ref{lem:exec_alpha}). %\delete{where the receive and response occurs at $s_x$ without any input action at $s_x$ between these two actions (Fig.~\ref{fig:execution1} $(a)$). } We then prove the existence of another execution $\beta$, which can be written in the form $\prefix{\beta} \circ \frag{1,x}{\beta} \circ \frag{1,y}{\beta}$ by extending $\alpha$ with an execution fragment $\frag{1,y}{\beta}$, such that $\frag{1,x}{\beta} \stackrel{s_x}{\sim} \frag{1,x}{\alpha}$ (Fig.~\ref{fig:execution1} $(b)$; Lemma~\ref{lem:exec_beta}). Note that in any arbitrary extension of $\beta$, $R_1$ eventually returns $(x_1, y_1)$. % Next, we show the existence of an execution $\gamma$ of the form $\prefix{\gamma} \circ \frag{1,x}{\gamma} \circ \frag{1,y}{\gamma}$, where the send actions $send(m_x^{r_1})_{r_1, s_x}$ and $send(m_y^{r_1})_{r_1, s_y}$ at $r_1$ occur before $W$ is invoked (Fig.~\ref{fig:execution1} $(c)$; $\gamma$), but $\frag{1,x}{\gamma}$ and $\frag{1,y}{\gamma}$ occur after $\RESP{W}$ as in $\beta$. % % Based on $\gamma$, we show the existence of an execution $\delta$ of the form $\prefix{\eta} \circ \frag{1,x}{\eta} \circ \frag{1,y}{\eta}\circ \suffix{\eta}$, where $R_1$ responds with $(x_1, y_1)$. % %of $\mathcal{A}$ such that, $R$ is invoked before $W$ is invoked and $R$ completes after $W$. % Additionally, in $\beta$ the servers %$s_x$ and $s_y$ receive the object value requests from $r$ after $W$ completes and $R$ responds with $(x_1, y_1)$ %From $\gamma$, we show the existence of an execution $\eta$ of $\mathcal{A}$ where $F_1$ occurs before $F_2$, and eventually $R$ responds with $(x_1, y_1)$. % Finally, starting with $\eta$, we create a sequence of executions $\delta (\equiv \eta)$, $\delta^{(1)}$, $\cdots$ $\delta^{(f)}$, of $\mathcal{A}$, where in each of them $R_1$ responds with $(x_1, y_1)$ (Fig.~\ref{fig:execution2} $(e)$ and $(g)$; Lemma~\ref{thm:two-snow}). Additionally, for any $\delta^{(i)}$, the fragments $\frag{1,x}{\delta^{(i)}}$ and $\frag{1,y}{\delta^{(i)}}$ appear before $\delta^{(i-1)}$. % i.e., $\prefix{\delta^{(i)}}$ is a prefix of $\prefix{\delta^{(i-1)}}$. Based on $\delta^{(f)}$, we prove the existence of an execution $\phi$ (Fig.~\ref{fig:execution2} $(h)$), where $R_1$ returns $(x_1, y_1)$ even before $W$ begins, which violates the $S$ property. } %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%555 The following theorem states when client-to-client communication is not allowed it is impossible to have the SNOW properties even with two clients. % We start with an execution $\eta$ and create a sequence of executions of $\mathcal{A}$, where each one is of the form $\prefix{\cdot} \circ \frag{1,x}{\cdot}\circ \frag{1,y}{\cdot}\circ \suffix{\cdot}$, with progressively shorter $\prefix{\cdot}$ until we have a final execution that contradicts the $S$ property. \begin{theorem}\label{thm:two-snow} The SNOW properties cannot be implemented in a system with two clients and two servers, where the clients are not allowed to communicate with each other. \end{theorem} \remove{ \begin{figure}[!ht] \centering \includegraphics[width=0.50\textwidth]{figures/architecture3_1.png} \vspace{-2.5em} \caption{ \small{The architecture of a typical web service with clients, servers, and % The external requests and responses at the clients to and from the end-users are represented as invocation and responses for the transaction. the communication channels, between every pair of processes, inside a datacenter is modeled as a collection of I/O automata. Note that, unlike the architecture in Fig.~\ref{fig:architecture2a}, in this setup there are communication channels between every pair of clients.}} \label{fig:architecture2c} \end{figure} } BeardOfDoom/adatbizt-adatvedelem-szabalyozasai1-10 \section{Jogalkotás az Európai Közösségben} Az Európai Unió joga több forrásból merít, ezeket fogom alább ismertetni. \cite{EU-jog} Elsődlegesen a szerződésekből, mint a Római Szerződések, vagy a Lisszaboni Szerződés. "A szerződés az uniós tagállamok között létrejött, kötelező erejű megállapodás, amely meghatározza az uniós célkitűzéseket, az uniós intézményekre vonatkozó szabályokat, a döntéshozatal módját és az EU és a tagállamai közötti viszonyt. A szerződések módosítására az EU hatékonyságának és átláthatóságának javítása, az új tagállamok csatlakozására való felkészülés, valamint új együttműködési területek (egységes valuta) bevezetése érdekében kerül sor." \cite{EU-szerzodesek} A szerződésekkel már egyenrangúvá vált az Európai Unió Alapjogi Chartája, amely "Belefoglalja az EU jogába az uniós polgárok, illetve az EU területén tartózkodó személyek számos személyes, állampolgári, politikai, gazdasági és társadalmi jogát." "Azáltal, hogy világosabbá teszi az alapjogokat és felhívja rájuk a figyelmet, a charta jogbiztonságot teremt az EU-ban." \cite{EU-charta} A szerződéseket követik az Unió által kötött nemzetközi megállapodások. Ezek a nemzetközi közjog szerinti egyezmények, és a szerződő felek számára jogokat és kötelezettségeket hoznak létre, amelyeket az EU egészében alkalmazni kell. \cite{EU-nemzetkozi} Az egyel lentebbi szinten a másodlagos jog van. Ezen joganyagok akkor tekinthetők érvényesnek, ha összhangban vannak a hierarchiában felette szereplő jogszabályokkal. "Másodlagos vagy származtatott jogforrásnak az Európai Unió intézményei által alkotott joganyagot nevezzük. E jogforrások az alapszerződéseken alapulnak, kizárólag az alapszerződésekben meghatározott szervek által és csak az ott meghatározott eljárás keretei között, megfelelő felhatalmazás alapján kerülhetnek kibocsátásra." "A másodlagos jogforrások közül a rendelet, az irányelv és a határozat kötelező erejű, az ajánlás és vélemény pedig nem bír kötelező erővel." \cite{EU-masodlagos} A jogalkotásban részt vesz az Európai Parlament, az Európai Unió Tanácsa, az Európai Bizottság, az Európai Gazdasági és Szociális Bizottság, és a Régiók Európai Bizottsága.ComputationalMethods/disedit.tex \chapter{Computational Methods Used} \thispagestyle{plain} \vspace{-.5cm} \input{ComputationalMethods/text.tex} \input{preamble} \input{macros} \begin{document} \mainmatter \setcounter{chapter}{8} \input{conclusions/main} \bibliography{bibliography/bibliography} \end{document} @Manual{orange, AUTHOR = "US Department of Defence", TITLE = "Trusted Computer System Evaluation Criteria", MONTH = "December", YEAR = "1995" } @Manual{redbook, AUTHOR = "US Department of Defence", TITLE = "Trusted Network Interpretation", MONTH = "April", YEAR = "1985", NOTE = "CSC-STD-002-85 \newline \texttt{http://www.radium.ncsc.mil/tpep/library/rainbow/NCSG-TG-021.html}" } @Manual{rainbow, AUTHOR = "US Department of Defence", TITLE = "The Rainbow Series", NOTE = "\newline \texttt{http://www.radium.ncsc.mil/tpep/library/rainbow/}" } @Manual{itsec, AUTHOR = "Commission of the European Communities", TITLE = "Information Technology Security Evaluation Criteria", MONTH = "June", YEAR = "1991" } @Manual{ctcpec, AUTHOR = "Canadian System Security Centre, Communications Security Establishment, Government of Canada", TITLE = "Canadian Trusted Computer Product Evaluation Criteria", MONTH = "January", YEAR = "1993" } @Manual{federal, AUTHOR = "United States Government", TITLE = "Federal Criteria for Information Technology Security, Version 1.0", MONTH = "December", YEAR = "1992" } @Manual{tpepfaq, AUTHOR = "Trusted Product Evaluation Program (TPEP)", TITLE = "The Computer Security Evaluation Criteria Frequently Asked Questions (V2.1)", NOTE = "\newline\texttt{http://www.radium.ncsc.mil/tpep/process/faq.html}" } @Manual{jt01, AUTHOR = "Joint Task 1 (JT01)", TITLE = "Relating Functionality Class and Security Sub-profile Specification", } @Manual{harmony, AUTHOR = "Cooperation on Security of Information Systems Joint Task 01", TITLE = "Foundations for the Harmonization of Information Technology Security Standards", MONTH = "April", YEAR = "1993" } @Manual{cc, AUTHOR = "NSA, NIST", TITLE = "Common Criteria for Information Technology Security Evaluation", MONTH = "January", YEAR = "1996" } @BOOK{unix, AUTHOR = " and ", TITLE = "Practical UNIX and Internet Security", PUBLISHER = "O'Reilly and Associates, Inc", YEAR = "1996" } @ARTICLE{nt, AUTHOR = "", TITLE = "The Handy Security Toolkit", JOURNAL = "Windows NT Magazine", MONTH = "July", YEAR = "1997" } @Manual{cesg, AUTHOR = "Communications-Electronics Security Group", TITLE = "UK Systems Security Confidence Levels, CESG, Memorandum No. 3", MONTH = "January", YEAR = "1989" } @Manual{dtiec, AUTHOR = "Department of Trade and Industry, United Kingdom", TITLE = "DTI Commercial Computer Security Centre Evaluation Levels Manual, V22", MONTH = "February", YEAR = "1989" } @Manual{german, AUTHOR = "German Information Technology Security Agency", TITLE = "Criteria for the Evaluation of Trustworthiness of Information Technology (IT) Systems", MONTH = "January", YEAR = "1989" } @Manual{french, AUTHOR = "Service Central de la S curit des Syst mes d'Information, France", TITLE = "Catalogue de Crit res Destin s valuer le Degr de Confiance des Syst mes d'Information", MONTH = "July", YEAR = "1989" } @Manual{dsd, AUTHOR = "Department of Defence, Commonwealth of Australia", TITLE = "Defence Signals Directorate", NOTE = "\newline \texttt{http://www.dsd.gov.au/}" } @article{liu_exploiting_2019, abstract = {The conductor-like polarizable continuum model (C-PCM) with switching/Gaussian smooth discretization is a widely used implicit solvation model in quantum chemistry. We have previously implemented C-PCM solvation for Hartree-Fock (HF) and density functional theory (DFT) on graphical processing units (GPUs), enabling the quantum mechanical treatment of large solvated biomolecules. Here, we first propose a GPU-based algorithm for the PCM conjugate gradient linear solver that greatly improves the performance for very large molecules. The overhead for PCM-related evaluations now consumes less than 15% of the total runtime for DFT calculations on large molecules. Second, we demonstrate that our algorithms tailored for ground state C-PCM are transferable to excited state properties. Using a single GPU, our method evaluates the analytic gradient of the linear response PCM time-dependent density functional theory energy up to 80× faster than a conventional central processing unit (CPU)-based implementation. In addition, our C-PCM algorithms are transferable to other methods that require electrostatic potential (ESP) evaluations. For example, we achieve speed-ups of up to 130× for restricted ESP-based atomic charge evaluations, when compared to CPU-based codes. We also summarize and compare the different PCM cavity discretization schemes used in some popular quantum chemistry packages as a reference for both users and developers.}, author = {, . and Kulik, . and Martínez, .}, copyright = {© 2018 Wiley Periodicals, Inc.}, doi = {10.1002/qua.25760}, file = {Accepted Version:/Users/asteeves/Zotero/storage/SFC9MPLD/Liu et al. - 2019 - Exploiting graphical processing units to enable qu.pdf:application/pdf;Snapshot:/Users/asteeves/Zotero/storage/LPFDW69N/qua.html:text/html}, issn = {1097-461X}, journal = {International Journal of Quantum Chemistry}, keywords = {excited state, graphical processing unit, nonequilibrium solvation, polarizable continuum}, language = {en}, note = {_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/qua.25760}, number = {1}, pages = {e25760}, title = {Exploiting graphical processing units to enable quantum chemistry calculation of large solvated molecules with conductor-like polarizable continuum models}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/qua.25760}, urldate = {2021-06-20}, volume = {119}, year = {2019} } UMBC-DREAM-Lab/UMBC-DREAM-Lab.github.io1-10 --- --- @inproceedings{eren2021DocEng21, abstract = {The unprecedented outbreak of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2), or COVID-19, continues to be a significant worldwide problem. As a result, a surge of new COVID-19 related research has followed suit. The growing number of publications requires document organization methods to identify relevant information. In this paper, we expand upon our previous work with clustering the CORD-19 dataset by applying multi-dimensional analysis methods. Tensor factorization is a powerful unsupervised learning method capable of discovering hidden patterns in a document corpus. We show that a higher-order representation of the corpus allows for the simultaneous grouping of similar articles, relevant journals, authors with similar research interests, and topic keywords. These groupings are identified within and among the latent components extracted via tensor decomposition. We further demonstrate the application of this method with a publicly available interactive visualization of the dataset.}, author = {Eren, and Solovyev, Nick and Hamer, Chris and McDonald, Renee and Alexandrov, Boian and }, booktitle = {Proceedings of the ACM Symposium on Document Engineering 2021}, doi = {10.1145/3469096.3474927}, pages = {1--4}, title = {COVID-19 Multidimensional Kaggle Literature Organization}, pdf = {https://arxiv.org/pdf/2107.08190.pdf}, url = {https://doi.org/10.1145/3469096.3474927}, year = {2021} } @article{eren2021RFoT, title={Random Forest of Tensors (RFoT)}, author={Eren, and Nicholas, Charles and McDonald, Renee and Hamer, Chris}, abstract={Machine learning has become an invaluable tool in the fight against malware. Traditional supervised and unsupervised methods are not designed to capture the multi-dimensional details that are often present in cyber data. In contrast, tensor factorization is a powerful unsupervised data analysis method for extracting the latent patterns that are hidden in a multi-dimensional corpus. In this poster we explore the application of tensors to classification, and we describe a hybrid model that leverages the strength of multi-dimensional analysis combined with clustering. We introduce a novel semi-supervised ensemble classifier named Random Forest of Tensors (RFoT) that is based on generating a forest of tensors in parallel, which share the same first dimension, and randomly selecting the remainder of the dimensions and entries of each tensor from the features set.}, year={2021}, note={Presented at the 12th Annual Malware Technical Exchange Meeting, Online, 2021}, url = {https://www.maksimeren.com/publication/eren_rfot_mtem2021/}, pdf= {https://www.maksimeren.com/poster/Random_Forest_of_Tensors_RFoT_MTEM.pdf}, journal={Presented at the 12th Annual Malware Technical Exchange Meeting, Online.} } @article{boutsikas2021evading, title={Evading Malware Classifiers via Monte Carlo Mutant Feature Discovery}, author={ Eren, Varga, Raff, Matuszek, Cynthia and }, abstract = {The use of Machine Learning has become a significant part of malware detection efforts due to the influx of new malware, an ever changing threat landscape, and the ability of Machine Learning methods to discover meaningful distinctions between malicious and benign software. Antivirus vendors have also begun to widely utilize malware classifiers based on dynamic and static malware analysis features. Therefore, a malware author might make evasive binary modifications against Machine Learning models as part of the malware development life cycle to execute an attack successfully. This makes the studying of possible classifier evasion strategies an essential part of cyber defense against malice. To this extent, we stage a grey box setup to analyze a scenario where the malware author does not know the target classifier algorithm, and does not have access to decisions made by the classifier, but knows the features used in training. In this experiment, a malicious actor trains a surrogate model using the EMBER-2018 dataset to discover binary mutations that cause an instance to be misclassified via a Monte Carlo tree search. Then, mutated malware is sent to the victim model that takes the place of an antivirus API to test whether it can evade detection.}, url = {https://arxiv.org/abs/2106.07860}, pdf={https://arxiv.org/pdf/2106.07860.pdf}, journal={Presented at the 12th Annual Malware Technical Exchange Meeting, Online.}, year={2021} } @string{aps = {American Physical Society,}} @article{raff_ngram_2016, abstract = {Malware classification using machine learning algorithms is a difficult task, in part due to the absence of strong natural features in raw executable binary files. Byte n-grams previously have been used as features, but little work has been done to explain their performance or to understand what concepts are actually being learned. In contrast to other work using n-gram features, in this work we use orders of magnitude more data, and we perform feature selection during model building using Elastic-Net regularized Logistic Regression. We compute a regularization path and analyze novel {\{}$\backslash$em multi-byte identifiers{\}}. Through this process, we discover significant previously unreported issues with byte n-gram features that cause their benefits and practicality to be overestimated. Three primary issues emerged from our work. First, we discovered a flaw in how previous corpora were created that leads to an over-estimation of classification accuracy. Second, we discovered that most of the information contained in n-grams stem from string features that could be obtained in simpler ways. Finally, we demonstrate that n-gram features promote overfitting, even with linear models and extreme regularization.}, author = {, Sylvester, , }, doi = {10.1007/s11416-016-0283-1}, issn = {2263-8733}, journal = {Journal of Computer Virology and Hacking Techniques}, keywords = {byte n-grams,elastic-net,malware classification,multi-byte identifier}, month = {sep}, title = {{An investigation of byte n-gram features for malware classification}}, url = {http://link.springer.com/10.1007/s11416-016-0283-1}, pdf={https://www.edwardraff.com/publications/investigation_byte_ngrams.pdf}, year = {2016} } @inproceedings{raff2017peheader, abstract = {Many efforts have been made to use various forms of domain knowledge in malware detection. Currently there exist two common approaches to malware detection without domain knowledge, namely byte n-grams and strings. In this work we explore the feasibility of applying neural networks to malware detection and feature learning. We do this by restricting ourselves to a minimal amount of domain knowledge in order to extract a portion of the Portable Executable (PE) header. By doing this we show that neural networks can learn from raw bytes without explicit feature construction, and perform even better than a domain knowledge approach that parses the PE header into explicit features.}, address = {New York, NY, USA}, author = { and and }, booktitle = {Proceedings of the 10th ACM Workshop on Artificial Intelligence and Security}, doi = {10.1145/3128572.3140442}, isbn = {978-1-4503-5202-4}, keywords = {cyber security,deep learning,malware detection}, pages = {121--132}, publisher = {ACM}, series = {AISec '17}, title = {{Learning the PE Header, Malware Detection with Minimal Domain Knowledge}}, url = {http://doi.acm.org/10.1145/3128572.3140442}, pdf={https://arxiv.org/pdf/1709.01471}, year = {2017} } @inproceedings{Zak2017, abstract = {Recent work has shown that byte n-grams learn mostly low entropy features, such as function imports and strings, which has brought into question whether byte n-grams can learn information corresponding to higher entropy levels, such as binary code. We investigate that hypothesis in this work by performing byte n-gram analysis on only specific sub-sections of the binary file, and compare to results ob- tained by n-gram analysis on assembly code generated from disassembled binaries. We do this by leveraging the change in model performance and ensembles to glean insights about the data. In doing so we discover that byte n-grams can learn from the code regions, but do not necessarily learn any new information. We also discover that assembly n-grams may not be as effective as previously thought and that disam- biguating instructions by their binary opcode, an approach not previously used for malware detection, is critical for model generalization.}, author = { and and }, booktitle = {2017 12th International Conference on Malicious and Unwanted Software (MALWARE)}, doi = {10.1109/MALWARE.2017.8323963}, isbn = {978-1-5386-1436-5}, month = {oct}, pages = {109--118}, publisher = {IEEE}, title = {{What can N-grams learn for malware detection?}}, url = {http://ieeexplore.ieee.org/document/8323963/}, pdf={https://www.edwardraff.com/publications/what_can_ngrams_learn.pdf}, year = {2017} } @inproceedings{raff_lzjd_2017, abstract = {The Normalized Compression Distance (NCD) has been used in a number of domains to compare objects with varying feature types. This flexibility comes from the use of general purpose compression algorithms as the means of computing distances between byte sequences. Such flexibility makes NCD particularly attractive for cases where the right features to use are not obvious, such as malware classification. However, NCD can be computationally demanding, thereby restricting the scale at which it can be applied. We introduce an alternative metric also inspired by compression, the Lempel-Ziv Jaccard Distance (LZJD). We show that this new distance has desirable theoretical properties, as well as comparable or superior performance for malware classification, while being easy to implement and orders of magnitude faster in practice.}, address = {New York, New York, USA}, author = { and }, booktitle = {Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD '17}, doi = {10.1145/3097983.3098111}, isbn = {9781450348874}, pages = {1007--1015}, publisher = {ACM Press}, title = {{An Alternative to NCD for Large Sequences, Lempel-Ziv Jaccard Distance}}, url = {http://dl.acm.org/citation.cfm?doid=3097983.3098111}, pdf={https://www.edwardraff.com/publications/alternative-ncd-lzjd.pdf}, year = {2017}, selected={true} } @article{JMLR:v18:16-131, abstract = {Java Statistical Analysis Tool (JSAT) is a Machine Learning library written in pure Java. It works to fill a void in the Java ecosystem for a general purpose library that is relatively high performance and flexible, which is not adequately fulfilled by Weka (Hall et al., 2009) and Java-ML (Abeel et al., 2009). Almost all of the algorithms are independently implemented using an Object- Oriented framework. JSAT is made available under the GNU GPL license here: github.com/EdwardRaff/JSAT.}, author = {}, journal = {Journal of Machine Learning Research}, number = {23}, pages = {1--5}, title = {{JSAT: Java Statistical Analysis Tool, a Library for Machine Learning}}, url = {http://jmlr.org/papers/v18/16-131.html}, pdf = {http://www.jmlr.org/papers/volume18/16-131/16-131.pdf}, html = {http://jmlr.org/papers/v18/16-131.html}, volume = {18}, year = {2017} } @inproceedings{raff_shwel, abstract = {There are currently few methods that can be applied to malware classification problems which don't require domain knowledge to apply. In this work, we develop our new SHWeL feature vector representation, by extending the recently proposed Lempel-Ziv Jaccard Distance. These SHWeL vectors improve upon LZJD's accuracy, outperform byte n-grams, and allow us to build efficient algorithms for both training (a weakness of byte n-grams) and inference (a weakness of LZJD). Furthermore, our new SHWeL method also allows us to directly tackle the class imbalance problem, which is common for malware-related tasks. Compared to existing methods like SMOTE, SHWeL provides significantly improved accuracy while reducing algorithmic complexity to O(N). Because our approach is developed without the use of domain knowledge, it can be easily re-applied to any new domain where there is a need to classify byte sequences.}, address = {New York, NY, USA}, author = { and }, booktitle = {Proceedings of the 10th ACM Workshop on Artificial Intelligence and Security}, doi = {10.1145/3128572.3140446}, isbn = {978-1-4503-5202-4}, keywords = {cyber security,lzjd,malware classification,shwel}, pages = {111--120}, publisher = {ACM}, series = {AISec '17}, title = {{Malware Classification and Class Imbalance via Stochastic Hashed LZJD}}, url = {http://doi.acm.org/10.1145/3128572.3140446}, pdf={https://www.edwardraff.com/publications/raff_shwel.pdf}, year = {2017} } @inproceedings{MalConv, abstract = {In this work we introduce malware detection from raw byte sequences as a fruitful research area to the larger machine learning community. Building a neural network for such a problem presents a number of interesting challenges that have not occurred in tasks such as image processing or NLP. In particular, we note that detection from raw bytes presents a sequence problem with over two million time steps and a problem where batch normalization appear to hinder the learning process. We present our initial work in building a solution to tackle this problem, which has linear complexity dependence on the sequence length, and allows for interpretable sub-regions of the binary to be identified. In doing so we will discuss the many challenges in building a neural network to process data at this scale, and the methods we used to work around them.}, archivePrefix = {arXiv}, arxivId = {1710.09435}, author = { and Sylvester, Jared and and }, booktitle = {AAAI Workshop on Artificial Intelligence for Cyber Security}, eprint = {1710.09435}, month = {oct}, title = {{Malware Detection by Eating a Whole EXE}}, url = {http://arxiv.org/abs/1710.09435}, pdf = {http://arxiv.org/pdf/1710.09435}, year = {2018}, selected={true} } @inproceedings{Fleshman2018, abstract = {As machine-learning (ML) based systems for malware detection become more prevalent, it becomes necessary to quantify the benefits compared to the more traditional anti-virus (AV) systems widely used today. It is not practical to build an agreed upon test set to benchmark malware detection systems on pure classification performance. Instead we tackle the problem by creating a new testing methodology, where we evaluate the change in performance on a set of known benign {\&} malicious files as adversarial modifications are performed. The change in performance combined with the evasion techniques then quantifies a system's robustness against that approach. Through these experiments we are able to show in a quantifiable way how purely ML based systems can be more robust than AV products at detecting malware that attempts evasion through modification, but may be slower to adapt in the face of significantly novel attacks.}, archivePrefix = {arXiv}, arxivId = {1806.04773}, author = { McLean, }, booktitle = {2018 13th International Conference on Malicious and Unwanted Software (MALWARE)}, doi = {10.1109/MALWARE.2018.8659360}, eprint = {1806.04773}, isbn = {978-1-7281-0155-2}, month = {oct}, pages = {1--10}, publisher = {IEEE}, title = {{Static Malware Detection {\&} Subterfuge: Quantifying the Robustness of Machine Learning and Current Anti-Virus}}, url = {https://ieeexplore.ieee.org/document/8659360/}, pdf={http://arxiv.org/pdf/1806.04773}, year = {2018} } @inproceedings{scws_18, address = {New York, NY, USA}, author = { and and }, booktitle = {Proceedings of the 27th ACM International Conference on Information and Knowledge Management}, doi = {10.1145/3269206.3271690}, isbn = {978-1-4503-6014-2}, keywords = {consistent weighted sampling,jaccard similarity,min-hashing}, pages = {1203--1212}, publisher = {ACM}, series = {CIKM '18}, title = {{Engineering a Simplified 0-Bit Consistent Weighted Sampling}}, url = {http://doi.acm.org/10.1145/3269206.3271690}, pdf={http://arxiv.org/pdf/1804.00069}, year = {2018} } @inproceedings{NEURIPS2019_c429429b, author = {}, abstract = {What makes a paper independently reproducible? Debates on reproducibility center around intuition or assumptions but lack empirical results. Our field focuses on releasing code, which is important, but is not sufficient for determining reproducibility. We take the first step toward a quantifiable answer by manually attempting to implement 255 papers published from 1984 until 2017, recording features of each paper, and performing statistical analysis of the results. For each paper, we did not look at the authors code, if released, in order to prevent bias toward discrepancies between code and paper.}, booktitle = {Advances in Neural Information Processing Systems}, pages = {}, publisher = {Curran Associates, Inc.}, title = {A Step Toward Quantifying Independently Reproducible Machine Learning Research}, url = {https://proceedings.neurips.cc/paper/2019/file/c429429bf1f2af051f2021dc92a8ebea-Paper.pdf}, pdf = {https://proceedings.neurips.cc/paper/2019/file/c429429bf1f2af051f2021dc92a8ebea-Paper.pdf}, volume = {32}, year = {2019} } @inproceedings{grad_dsaa, author = { and }, booktitle = {The 5th IEEE International Conference on Data Science and Advanced Analytics (DSAA)}, title = {{Gradient Reversal Against Discrimination : A Fair Neural Network Learning Approach}}, year = {2018}, pdf={https://www.edwardraff.com/publications/grad_dsaa.pdf}, } @article{raff_lzjd_digest, abstract = {Recent work has proposed the Lempel-Ziv Jaccard Distance (LZJD) as a method to measure the similarity between binary byte sequences for malware classification. We propose and test LZJD's effectiveness as a similarity digest hash for digital forensics. To do so we develop a high performance Java implementation with the same command-line arguments as sdhash, making it easy to integrate into existing work-flows. Our testing shows that LZJD is effective for this task, and significantly outperforms sdhash and ssdeep in its ability to match related file fragments and is faster at comparison time.}, archivePrefix = {arXiv}, arxivId = {1708.03346}, author = { Nicholas, .}, doi = {10.1016/j.diin.2017.12.004}, eprint = {1708.03346}, issn = {17422876}, journal = {Digital Investigation}, month = {feb}, title = {{Lempel-Ziv Jaccard Distance, an effective alternative to ssdeep and sdhash}}, url = {https://doi.org/10.1016/j.diin.2017.12.004}, pdf={https://arxiv.org/pdf/1708.03346}, year = {2018} } @inproceedings{hashgram_2018, abstract = {N-grams have long been used as features for classification problems, and their distribution often allows selection of the top-k occurring n-grams as a reliable first-pass to feature selection. However, this top-k selection can be a performance bottleneck, especially when dealing with massive item sets and corpora. In this work we introduce Hash-Grams, an approach to perform top-k feature mining for classification problems. We show that the Hash-Gram approach can be up to three orders of magnitude faster than exact top-k selection algorithms. Using a malware corpus of over 2 TB in size, we show how Hash-Grams retain comparable classification accuracy, while dramatically reducing computational requirements.}, address = {Halifax, NS, Canada}, author = { and }, booktitle = {Proceedings of the ACM Symposium on Document Engineering 2018}, doi = {10.1145/3209280.3229085}, publisher = {ACM}, title = {{Hash-Grams: Faster N-Gram Features for Classification and Malware Detection}}, url = {http://doi.acm.org/10.1145/3209280.3229085}, pdf = {https://www.edwardraff.com/publications/Hash_Grams_On_Many_Cores_and_Skewed_Distributions.pdf}, year = {2018} } @inproceedings{pylzjd-proc-scipy-2019, author = { and and }, booktitle = {Proceedings of the 18th Python in Science Conference}, doi = {10.25080/Majora-7ddc1dd1-00e}, editor = {Calloway, Chris and Lippa, David and Niederhut, Dillon and Shupe, David}, pages = {97--102}, title = {{PyLZJD: An Easy to Use Tool for Machine Learning}}, url = {http://conference.scipy.org/proceedings/scipy2019/pylzjd.html}, html = {http://conference.scipy.org/proceedings/scipy2019/pylzjd.html}, pdf = {http://conference.scipy.org/proceedings/scipy2019/pdfs/pylzjd.pdf}, year = {2019} } @inproceedings{Kilograms_2019, abstract = {N-grams have been a common tool for information retrieval and machine learning applications for decades. In nearly all previous works, only a few values ofn are tested, with n {\textgreater} 6 being exceed- ingly rare. Larger values ofn are not tested due to computational burden or the fear of overfitting. In this work, we present a method to find the top-k most frequent n-grams that is 60× faster for small n, and can tackle large n ≥ 1024. Despite the unprecedented size ofn considered, we show how these features still have predictive ability for malware classification tasks. More important, large n- grams provide benefits in producing features that are interpretable by malware analysis, and can be used to create general purpose signatures compatible with industry standard tools like Yara. Fur- thermore, the counts of common n-grams in a file may be added as features to publicly available human-engineered features that rival efficacy of professionally-developed features when used to train gradient-boosted decision tree models on the EMBER dataset.}, author = { Finlayson, . and }, booktitle = {Proceedings of KDD 2019 Workshop on Learning and Mining for Cybersecurity (LEMINCS'19)}, title = {{KiloGrams: Very Large N-Grams for Malware Classification}}, url = {https://arxiv.org/abs/1908.00200}, pdf = {https://arxiv.org/pdf/1908.00200}, year = {2019} } @inproceedings{Raff2019d, abstract = {Defenses against adversarial examples, when using the ImageNet dataset, are historically easy to defeat. The common understanding is that a combination of simple image transformations and other various defenses are insufficient to provide the necessary protection when the obfuscated gradient is taken into account. In this paper, we explore the idea of stochastically combining a large number of individually weak defenses into a single barrage of randomized transformations to build a strong defense against adversarial attacks. We show that, even after accounting for obfuscated gradients, the Barrage of Random Transforms (BaRT) is a resilient defense against even the most difficult attacks, such as PGD. BaRT achieves up to a 24x improvement in accuracy compared to previous work, and has even extended effectiveness out to a previously untested maximum adversarial perturbation of ϵ=32.}, author = {. and . and . and .}, booktitle = {Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition}, doi = {10.1109/CVPR.2019.00669}, isbn = {9781728132938}, issn = {10636919}, keywords = {Big Data,Categorization,Deep Learning,Large Scale Methods,Recognition: Detection,Retrieval}, title = {{Barrage of random transforms for adversarially robust defense}}, pdf={https://openaccess.thecvf.com/content_CVPR_2019/papers/Raff_Barrage_of_Random_Transforms_for_Adversarially_Robust_Defense_CVPR_2019_paper.pdf}, volume = {2019-June}, year = {2019} } @article{Fleshman2018a, abstract = {Adversarial attacks against Neural Networks are a problem of considerable importance, for which effective defenses are not yet readily available. We make progress toward this problem by showing that non-negative weight constraints can be used to improve resistance in specific scenarios. In particular, we show that they can provide an effective defense for binary classification problems with asymmetric cost, such as malware or spam detection. We also show how non-negativity can be leveraged to reduce an attacker's ability to perform targeted misclassification attacks in other domains such as image processing.}, archivePrefix = {arXiv}, arxivId = {1806.06108}, author = { }, eprint = {1806.06108}, journal = {AAAI-2019 Workshop on Artificial Intelligence for Cyber Security}, title = {{Non-Negative Networks Against Adversarial Attacks}}, url = {http://arxiv.org/abs/1806.06108}, pdf = {http://arxiv.org/pdf/1806.06108}, year = {2019} } @inproceedings{Nguyen2019_filename_malicious, author = { }, booktitle = {2019 IEEE International Conference on Big Data (Big Data)}, doi = {10.1109/BigData47090.2019.9006132}, isbn = {978-1-7281-0858-2}, month = {dec}, pages = {1322--1331}, publisher = {IEEE}, title = {{Would a File by Any Other Name Seem as Malicious?}}, url = {https://ieeexplore.ieee.org/document/9006132/}, pdf={https://arxiv.org/pdf/1910.04753.pdf}, year = {2019} } @inproceedings{Zhang2020a, abstract = {Artificial intelligence (AI)-based decision-making systems are employed nowadays in an ever growing number of online as well as offline services–some of great importance. Depending on sophisticated learning algorithms and available data, these systems are increasingly becoming automated and data-driven. However, these systems can impact individuals and communities with ethical or legal consequences. Numerous approaches have therefore been proposed to develop decision making systems that are discrimination-conscious by-design. However, these methods assume the underlying data distribution is stationary without drift, which is counterfactual in many real world applications. In addition, their focus has been largely on minimizing discrimination while maximizing prediction performance without necessary flexibility in customizing the tradeoff according to different applications. To this end, we propose a learning algorithm for fair classification that also adapts to evolving data streams and further allows for a flexible control on the degree of accuracy and fairness. The positive results on a set of discriminated and non-stationary data streams demonstrate the effectiveness and flexibility of this approach.}, author = { and and and and and and }, booktitle = {2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI)}, doi = {10.1109/ICTAI50040.2020.00069}, isbn = {978-1-7281-9228-4}, keywords = {ai fairness,flexible fairness,online classification}, month = {nov}, pages = {399--406}, publisher = {IEEE}, title = {{Flexible and Adaptive Fairness-aware Learning in Non-stationary Data Streams}}, url = {https://ieeexplore.ieee.org/document/9288346/}, pdf={https://mdsoar.org/bitstream/handle/11603/20072/ICTAI20.pdf?sequence=1&isAllowed=y}, year = {2020} } @inproceedings{Pillai2020, abstract = {Ordering the selection of training data using active learning can lead to improvements in learning efficiently from smaller corpora. We present an exploration of active learning approaches applied to three grounded language problems of varying complexity in order to analyze what methods are suitable for improving data efficiency in learning. We present a method for analyzing the complexity of data in this joint problem space, and report on how characteristics of the underlying task, along with design decisions such as feature selection and classification model, drive the results. We observe that representativeness, along with diversity, is crucial in selecting data samples.}, archivePrefix = {arXiv}, arxivId = {2011.08021}, author = { and }, booktitle = {2020 IEEE International Conference on Big Data (Big Data)}, eprint = {2011.08021}, title = {{Sampling Approach Matters: Active Learning for Robotic Language Acquisition}}, url = {http://arxiv.org/abs/2011.08021}, pdf = {http://arxiv.org/pdf/2011.08021}, year = {2020} } @inproceedings{Eren2020, abstract = {The world has faced the devastating outbreak of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2), or COVID-19, in 2020. Research in the subject matter was fast-tracked to such a point that scientists were struggling to keep up with new findings. With this increase in the scientific literature, there arose a need for organizing those documents. We describe an approach to organize and visualize the scientific literature on or related to COVID-19 using machine learning techniques so that papers on similar topics are grouped together. By doing so, the navigation of topics and related papers is simplified. We implemented this approach using the widely recognized CORD-19 dataset to present a publicly available proof of concept.}, author = {Eren, and Solovyev, Nick and Raff, Edward and Nicholas, Charles and Johnson, Ben}, booktitle = {Proceedings of the ACM Symposium on Document Engineering 2020}, doi = {10.1145/3395027.3419591}, keywords = {acm reference format,charles nicholas,clustering,covid-19,dimensionality reduction,document visualization,edward raff,maksim ekin eren,nick solovyev}, pages = {1--4}, title = {{COVID-19 Kaggle Literature Organization}}, pdf={https://arxiv.org/pdf/2008.13542}, url = {https://dl.acm.org/doi/10.1145/3395027.3419591}, year = {2020} } @inproceedings{Raff2020a, abstract = {Malware classification is a difficult problem, to which machine learning methods have been applied for decades. Yet progress has often been slow, in part due to a number of unique difficulties with the task that occur through all stages of the developing a machine learning system: data collection, labeling, feature creation and selection, model selection, and evaluation. In this survey we will review a number of the current methods and challenges related to malware classification, including data collection, feature extraction, and model construction, and evaluation. Our discussion will include thoughts on the constraints that must be considered for machine learning based solutions in this domain, and yet to be tackled problems for which machine learning could also provide a solution. This survey aims to be useful both to cybersecurity practitioners who wish to learn more about how machine learning can be applied to the malware problem, and to give data scientists the necessary background into the challenges in this uniquely complicated space.}, archivePrefix = {arXiv}, arxivId = {2006.09271}, author = { and }, booktitle = {NeurIPS 2020 Workshop: ML Retrospectives, Surveys {\&} Meta-Analyses (ML-RSA)}, eprint = {2006.09271}, keywords = {68t01,68t99,ams subject classifications,cyber security,machine learning,malware detection}, title = {{A Survey of Machine Learning Methods and Challenges for Windows Malware Classification}}, url = {http://arxiv.org/abs/2006.09271}, pdf = {http://arxiv.org/pdf/2006.09271}, year = {2020} } @inproceedings{Ordun2020a, abstract = {With the increased attention on thermal imagery for Covid-19 screening, the public sector may believe there are new opportunities to exploit thermal as a modality for computer vision and AI. Thermal physiology research has been ongoing since the late nineties. This research lies at the intersections of medicine, psychology, machine learning, optics, and affective computing. We will review the known factors of thermal vs. RGB imaging for facial emotion recognition. But we also propose that thermal imagery may provide a semi-anonymous modality for computer vision, over RGB, which has been plagued by misuse in facial recognition. However, the transition to adopting thermal imagery as a source for any human-centered AI task is not easy and relies on the availability of high fidelity data sources across multiple demographics and thorough validation. This paper takes the reader on a short review of machine learning in thermal FER and the limitations of collecting and developing thermal FER data for AI training. Our motivation is to provide an introductory overview into recent advances for thermal FER and stimulate conversation about the limitations in current datasets.}, archivePrefix = {arXiv}, arxivId = {2009.10589}, author = { }, booktitle = {AAAI FSS-20: Artificial Intelligence in Government and Public Sector}, eprint = {2009.10589}, title = {{The Use of AI for Thermal Emotion Recognition: A Review of Problems and Limitations in Standard Design and Data}}, url = {http://arxiv.org/abs/2009.10589}, pdf = {http://arxiv.org/pdf/2009.10589}, year = {2020} } @inproceedings{Rahnama2020, abstract = {Deep neural networks (DNNs) are vulnerable to subtle adversarial perturbations applied to the input. These adversarial perturbations, though imperceptible, can easily mislead the DNN. In this work, we take a control theoretic approach to the problem of robustness in DNNs. We treat each individual layer of the DNN as a nonlinear dynamical system and use Lyapunov theory to prove stability and robustness locally. We then proceed to prove stability and robustness globally for the entire DNN. We develop empirically tight bounds on the response of the output layer, or any hidden layer, to adversarial perturbations added to the input, or the input of hidden layers. Recent works have proposed spectral norm regularization as a solution for improving robustness against l2 adversarial attacks. Our results give new insights into how spectral norm regularization can mitigate the adversarial effects. Finally, we evaluate the power of our approach on a variety of data sets and network architectures and against some of the well-known adversarial attacks.}, archivePrefix = {arXiv}, arxivId = {1911.04636}, author = { . and }, booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, eprint = {1911.04636}, pages = {8178--8187}, title = {{Robust Design of Deep Neural Networks against Adversarial Attacks based on Lyapunov Theory}}, url = {http://arxiv.org/abs/1911.04636}, pdf = {http://arxiv.org/pdf/1911.04636}, year = {2020} } @inproceedings{Raff2020autoyara, abstract = {Yara rules are a ubiquitous tool among cybersecurity practitioners and analysts. Developing high-quality Yara rules to detect a malware family of interest can be labor- and time-intensive, even for expert users. Few tools exist and relatively little work has been done on how to automate the generation of Yara rules for specific families. In this paper, we leverage large n-grams ({\$}n \backslashgeq 8{\$}) combined with a new biclustering algorithm to construct simple Yara rules more effectively than currently available software. Our method, AutoYara, is fast, allowing for deployment on low-resource equipment for teams that deploy to remote networks. Our results demonstrate that AutoYara can help reduce analyst workload by producing rules with useful true-positive rates while maintaining low false-positive rates, sometimes matching or even outperforming human analysts. In addition, real-world testing by malware analysts indicates AutoYara could reduce analyst time spent constructing Yara rules by 44-86{\%}, allowing them to spend their time on the more advanced malware that current tools can't handle. Code will be made available at https://github.com/NeuromorphicComputationResearchProgram .}, archivePrefix = {arXiv}, arxivId = {2009.03779}, author = { and and Munoz, and and . and and and }, booktitle = {13th ACM Workshop on Artificial Intelligence and Security (AISec'20)}, doi = {10.1145/3411508.3421372}, eprint = {2009.03779}, title = {{Automatic Yara Rule Generation Using Biclustering}}, url = {http://arxiv.org/abs/2009.03779}, pdf = {http://arxiv.org/pdf/2009.03779}, year = {2020} } @inproceedings{Raff2020, abstract = {Prior work inspired by compression algorithms has described how the Burrows Wheeler Transform can be used to create a distance measure for bioinformatics problems. We describe issues with this approach that were not widely known, and introduce our new Burrows Wheeler Markov Distance (BWMD) as an alternative. The BWMD avoids the shortcomings of earlier efforts, and allows us to tackle problems in variable length DNA sequence clustering. BWMD is also more adaptable to other domains, which we demonstrate on malware classification tasks. Unlike other compression-based distance metrics known to us, BWMD works by embedding sequences into a fixed-length feature vector. This allows us to provide significantly improved clustering performance on larger malware corpora, a weakness of prior methods.}, archivePrefix = {arXiv}, arxivId = {1912.13046}, author = { Nicholas, }, booktitle = {The Thirty-Fourth AAAI Conference on Artificial Intelligence}, doi = {10.1609/aaai.v34i04.5994}, eprint = {1912.13046}, pages = {5444--5453}, title = {{A New Burrows Wheeler Transform Markov Distance}}, url = {http://arxiv.org/abs/1912.13046}, url = {http://arxiv.org/pdf/1912.13046.pdf}, year = {2020} } @inproceedings{Nolet2020, abstract = {The Uniform Manifold Approximation and Projection (UMAP) algorithm has become widely popular for its ease of use, quality of results, and support for exploratory, unsupervised, supervised, and semi-supervised learning. While many algorithms can be ported to a GPU in a simple and direct fashion, such efforts have resulted in inefficent and inaccurate versions of UMAP. We show a number of techniques that can be used to make a faster and more faithful GPU version of UMAP, and obtain speedups of up to 100x in practice. Many of these design choices/lessons are general purpose and may inform the conversion of other graph and manifold learning algorithms to use GPUs. Our implementation has been made publicly available as part of the open source RAPIDS cuML library(https://github.com/rapidsai/cuml).}, archivePrefix = {arXiv}, arxivId = {2008.00325}, author = {Nolet, . and Lafargue, Raff, Oates, , }, booktitle = {The Thirty-Fifth AAAI Conference on Artificial Intelligence}, eprint = {2008.00325}, title = {{Bringing UMAP Closer to the Speed of Light with GPU Acceleration}}, url = {http://arxiv.org/abs/2008.00325}, year = {2021}, pdf={https://arxiv.org/pdf/2008.00325.pdf} } @inproceedings{Raff2020c, abstract = {There has been increasing concern within the machine learning community that we are in a reproducibility crisis. As many have begun to work on this problem, all work we are aware of treat the issue of reproducibility as an intrinsic binary property: a paper is or is not reproducible. Instead, we consider modeling the reproducibility of a paper as a survival analysis problem. We argue that this perspective represents a more accurate model of the underlying meta-science question of reproducible research, and we show how a survival analysis allows us to draw new insights that better explain prior longitudinal data. The data and code can be found at https://github.com/EdwardRaff/Research-Reproducibility-Survival-Analysis}, archivePrefix = {arXiv}, arxivId = {2012.09932}, author = {}, booktitle = {The Thirty-Fifth AAAI Conference on Artificial Intelligence}, eprint = {2012.09932}, title = {{Research Reproducibility as a Survival Analysis}}, url = {http://arxiv.org/abs/2012.09932}, pdf = {http://arxiv.org/pdf/2012.09932}, year = {2021} } @inproceedings{Raff2020b, abstract = {Recent works within machine learning have been tackling inputs of ever-increasing size, with cybersecurity presenting sequence classification problems of particularly extreme lengths. In the case of Windows executable malware detection, inputs may exceed {\$}100{\$} MB, which corresponds to a time series with {\$}T=100,000,000{\$} steps. To date, the closest approach to handling such a task is MalConv, a convolutional neural network capable of processing up to {\$}T=2,000,000{\$} steps. The {\$}\backslashmathcal{\{}O{\}}(T){\$} memory of CNNs has prevented further application of CNNs to malware. In this work, we develop a new approach to temporal max pooling that makes the required memory invariant to the sequence length {\$}T{\$}. This makes MalConv {\$}116\backslashtimes{\$} more memory efficient, and up to {\$}25.8\backslashtimes{\$} faster to train on its original dataset, while removing the input length restrictions to MalConv. We re-invest these gains into improving the MalConv architecture by developing a new Global Channel Gating design, giving us an attention mechanism capable of learning feature interactions across 100 million time steps in an efficient manner, a capability lacked by the original MalConv CNN. Our implementation can be found at https://github.com/NeuromorphicComputationResearchProgram/MalConv2}, archivePrefix = {arXiv}, arxivId = {2012.09390}, author = { and . and and }, booktitle = {The Thirty-Fifth AAAI Conference on Artificial Intelligence}, eprint = {2012.09390}, title = {{Classifying Sequences of Extreme Length with Constant Memory Applied to Malware Detection}}, url = {http://arxiv.org/abs/2012.09390}, pdf = {http://arxiv.org/pdf/2012.09390}, year = {2021} } @inproceedings{Ordun2020, abstract = {This paper illustrates five different techniques to assess the distinctiveness of topics, key terms and features, speed of information dissemination, and network behaviors for Covid19 tweets. First, we use pattern matching and second, topic modeling through Latent Dirichlet Allocation (LDA) to generate twenty different topics that discuss case spread, healthcare workers, and personal protective equipment (PPE). One topic specific to U.S. cases would start to uptick immediately after live White House Coronavirus Task Force briefings, implying that many Twitter users are paying attention to government announcements. We contribute machine learning methods not previously reported in the Covid19 Twitter literature. This includes our third method, Uniform Manifold Approximation and Projection (UMAP), that identifies unique clustering-behavior of distinct topics to improve our understanding of important themes in the corpus and help assess the quality of generated topics. Fourth, we calculated retweeting times to understand how fast information about Covid19 propagates on Twitter. Our analysis indicates that the median retweeting time of Covid19 for a sample corpus in March 2020 was 2.87 hours, approximately 50 minutes faster than repostings from Chinese social media about H7N9 in March 2013. Lastly, we sought to understand retweet cascades, by visualizing the connections of users over time from fast to slow retweeting. As the time to retweet increases, the density of connections also increase where in our sample, we found distinct users dominating the attention of Covid19 retweeters. One of the simplest highlights of this analysis is that early-stage descriptive methods like regular expressions can successfully identify high-level themes which were consistently verified as important through every subsequent analysis.}, archivePrefix = {arXiv}, arxivId = {2005.03082}, author = { }, booktitle = {epiDAMIK 2020: 3rd epiDAMIK ACM SIGKDD International Workshop on Epidemiology meets Data Mining and Knowledge Discovery}, eprint = {2005.03082}, title = {{Exploratory Analysis of Covid-19 Tweets using Topic Modeling, UMAP, and DiGraphs}}, url = {http://arxiv.org/abs/2005.03082}, pdf = {http://arxiv.org/pdf/2005.03082}, year = {2020} } @inproceedings{DBLP:conf/dsaa/ShahapureN20, author = { and }, editor = { and and and and and }, title = {Cluster Quality Analysis Using Silhouette Score}, booktitle = {7th {IEEE} International Conference on Data Science and Advanced Analytics, {DSAA} 2020, Sydney, Australia, October 6-9, 2020}, pages = {747--748}, publisher = {{IEEE}}, year = {2020}, url = {https://doi.org/10.1109/DSAA49011.2020.00096}, doi = {10.1109/DSAA49011.2020.00096}, timestamp = {Tue, 24 Nov 2020 12:27:16 +0100}, biburl = {https://dblp.org/rec/conf/dsaa/ShahapureN20.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/corr/abs-2005-02911, author = { and }, title = {A Quantum Algorithm To Locate Unknown Hashes For Known N-Grams Within {A} Large Malware Corpus}, journal = {CoRR}, volume = {abs/2005.02911}, year = {2020}, url = {https://arxiv.org/abs/2005.02911}, archivePrefix = {arXiv}, eprint = {2005.02911}, timestamp = {Sun, 10 May 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/corr/abs-2005-02911.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/doceng/Nicholas17, author = {}, editor = { and }, title = {Document Engineering Issues in Malware Analysis}, booktitle = {Proceedings of the 2017 {ACM} Symposium on Document Engineering, DocEng 2017, Valletta, Malta, September 4-7, 2017}, pages = {3}, publisher = {{ACM}}, year = {2017}, url = {https://doi.org/10.1145/3103010.3103027}, doi = {10.1145/3103010.3103027}, timestamp = {Tue, 06 Nov 2018 16:57:31 +0100}, biburl = {https://dblp.org/rec/conf/doceng/Nicholas17.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/doceng/NicholasB16, author = { and }, editor = { and }, title = {Document Engineering Issues in Malware Analysis}, booktitle = {Proceedings of the 2016 {ACM} Symposium on Document Engineering, DocEng 2016, Vienna, Austria, September 13 - 16, 2016}, pages = {3}, publisher = {{ACM}}, year = {2016}, url = {https://doi.org/10.1145/2960811.2967174}, doi = {10.1145/2960811.2967174}, timestamp = {Tue, 06 Nov 2018 16:57:33 +0100}, biburl = {https://dblp.org/rec/conf/doceng/NicholasB16.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/doceng/NicholasB15, author = { and }, editor = { and {\`{e}}s}, title = {Document Engineering Issues in Document Analysis}, booktitle = {Proceedings of the 2015 {ACM} Symposium on Document Engineering, DocEng 2015, Lausanne, Switzerland, September 8-11, 2015}, pages = {229--230}, publisher = {{ACM}}, year = {2015}, url = {https://doi.org/10.1145/2682571.2801033}, doi = {10.1145/2682571.2801033}, timestamp = {Tue, 06 Nov 2018 16:57:32 +0100}, biburl = {https://dblp.org/rec/conf/doceng/NicholasB15.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/sigweb/NicholasM13, author = { and }, title = {Document engineering education: workshop report}, journal = {{SIGWEB} Newsl.}, volume = {2013}, number = {Winter}, pages = {1:1--1:5}, year = {2013}, url = {https://doi.org/10.1145/2430733.2430734}, doi = {10.1145/2430733.2430734}, timestamp = {Thu, 16 Jul 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/sigweb/NicholasM13.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/vizsec/LeschkeN13, author = { and }, editor = { and Kwan{-} and and }, title = {Change-link 2.0: a digital forensic tool for visualizing changes to shadow volume data}, booktitle = {10th Workshop on Visualization for Cyber Security, VizSec 2013, Atlanta, GA, USA, October 14, 2013}, pages = {17--24}, publisher = {{ACM}}, year = {2013}, url = {https://doi.org/10.1145/2517957.2517960}, doi = {10.1145/2517957.2517960}, timestamp = {Wed, 24 Feb 2021 16:44:20 +0100}, biburl = {https://dblp.org/rec/conf/vizsec/LeschkeN13.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/naacl/McNameeMN09, author = { and and }, title = {Translation Corpus Source and Size in Bilingual Retrieval}, booktitle = {Human Language Technologies: Conference of the North American Chapter of the Association of Computational Linguistics, Proceedings, May 31 - June 5, 2009, Boulder, Colorado, USA, Short Papers}, pages = {25--28}, publisher = {The Association for Computational Linguistics}, year = {2009}, url = {https://www.aclweb.org/anthology/N09-2007/}, timestamp = {Wed, 18 Sep 2019 12:15:54 +0200}, biburl = {https://dblp.org/rec/conf/naacl/McNameeMN09.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/sigir/McNameeNM09, author = { and and }, editor = { and and and and }, title = {Addressing morphological variation in alphabetic languages}, booktitle = {Proceedings of the 32nd Annual International {ACM} {SIGIR} Conference on Research and Development in Information Retrieval, {SIGIR} 2009, Boston, MA, USA, July 19-23, 2009}, pages = {75--82}, publisher = {{ACM}}, year = {2009}, url = {https://doi.org/10.1145/1571941.1571957}, doi = {10.1145/1571941.1571957}, timestamp = {Wed, 14 Nov 2018 10:58:10 +0100}, biburl = {https://dblp.org/rec/conf/sigir/McNameeNM09.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/cikm/WilsonN08, author = { and }, editor = { and and }, title = {Topological analysis of an online social network for older adults}, booktitle = {Proceeding of the 2008 {ACM} Workshop on Search in Social Media, {SSM} 2008, Napa Valley, California, USA, October 30, 2008}, pages = {51--58}, publisher = {{ACM}}, year = {2008}, url = {https://doi.org/10.1145/1458583.1458596}, doi = {10.1145/1458583.1458596}, timestamp = {Tue, 06 Nov 2018 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/cikm/WilsonN08.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/sigir/McNameeNM08, author = { and and }, editor = {Sung{-} and and and Tat{-} and Mun{-}}, title = {Don't have a stemmer?: be un+concern+ed}, booktitle = {Proceedings of the 31st Annual International {ACM} {SIGIR} Conference on Research and Development in Information Retrieval, {SIGIR} 2008, Singapore, July 20-24, 2008}, pages = {813--814}, publisher = {{ACM}}, year = {2008}, url = {https://doi.org/10.1145/1390334.1390518}, doi = {10.1145/1390334.1390518}, timestamp = {Tue, 06 Nov 2018 11:07:25 +0100}, biburl = {https://dblp.org/rec/conf/sigir/McNameeNM08.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/eor/VolkovichKN07, author = { and and }, title = {Building initial partitions through sampling techniques}, journal = {Eur. J. Oper. Res.}, volume = {183}, number = {3}, pages = {1097--1105}, year = {2007}, url = {https://doi.org/10.1016/j.ejor.2005.12.045}, doi = {10.1016/j.ejor.2005.12.045}, timestamp = {Fri, 21 Feb 2020 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/eor/VolkovichKN07.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @incollection{DBLP:books/daglib/p/VolkovichKN06, author = { and and }, editor = { and and }, title = {Sampling Methods for Building Initial Partitions}, booktitle = {Grouping Multidimensional Data - Recent Advances in Clustering}, pages = {161--185}, publisher = {Springer}, year = {2006}, url = {https://doi.org/10.1007/3-540-28349-8\_6}, doi = {10.1007/3-540-28349-8\_6}, timestamp = {Sun, 25 Oct 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/books/daglib/p/VolkovichKN06.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @book{DBLP:books/daglib/0015184, editor = { and and }, title = {Grouping Multidimensional Data - Recent Advances in Clustering}, publisher = {Springer}, year = {2006}, url = {https://doi.org/10.1007/3-540-28349-8}, doi = {10.1007/3-540-28349-8}, isbn = {978-3-540-28348-5}, timestamp = {Tue, 16 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/books/daglib/0015184.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/ir/KoganTN05, author = { and and }, title = {Data Driven Similarity Measures for \emph{k}-Means Like Clustering Algorithms}, journal = {Inf. Retr.}, volume = {8}, number = {2}, pages = {331--349}, year = {2005}, url = {https://doi.org/10.1007/s10791-005-5666-8}, doi = {10.1007/s10791-005-5666-8}, timestamp = {Sat, 27 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/ir/KoganTN05.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/dke/BhatOSN04, author = { and and and }, title = {Finding aliases on the web using latent semantic analysis}, journal = {Data Knowl. Eng.}, volume = {49}, number = {2}, pages = {129--143}, year = {2004}, url = {https://doi.org/10.1016/j.datak.2003.10.006}, doi = {10.1016/j.datak.2003.10.006}, timestamp = {Sat, 20 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/dke/BhatOSN04.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/cse/KoganNV03, author = { and and }, title = {Text mining with information-theoretic clustering}, journal = {Comput. Sci. Eng.}, volume = {5}, number = {6}, pages = {52--59}, year = {2003}, url = {https://doi.org/10.1109/MCISE.2003.1238704}, doi = {10.1109/MCISE.2003.1238704}, timestamp = {Thu, 12 Mar 2020 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/cse/KoganNV03.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/trec/KallurkarSCNJJRSBO03, author = { and and and and and and and and and }, editor = { and }, title = {{UMBC} at {TREC} 12}, booktitle = {Proceedings of The Twelfth Text REtrieval Conference, {TREC} 2003, Gaithersburg, Maryland, USA, November 18-21, 2003}, series = {{NIST} Special Publication}, volume = {500-255}, pages = {699--706}, publisher = {National Institute of Standards and Technology {(NIST)}}, year = {2003}, url = {http://trec.nist.gov/pubs/trec12/papers/umbc.web.novelty.pdf}, timestamp = {Thu, 12 Mar 2020 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/trec/KallurkarSCNJJRSBO03.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/expert/CostFJPNSCKPZT02, author = { and and and and and and and and and and }, title = {ITtalks: {A} Case Study in the Semantic Web and {DAML+OIL}}, journal = {{IEEE} Intell. Syst.}, volume = {17}, number = {1}, pages = {40--47}, year = {2002}, url = {https://doi.org/10.1109/5254.988447}, doi = {10.1109/5254.988447}, timestamp = {Mon, 26 Oct 2020 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/expert/CostFJPNSCKPZT02.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/ir/SoboroffN02, author = { and }, title = {Related, but not Relevant: Content-Based Collaborative Filtering in {TREC-8}}, journal = {Inf. Retr.}, volume = {5}, number = {2-3}, pages = {189--208}, year = {2002}, url = {https://doi.org/10.1023/A:1015797928606}, doi = {10.1023/A:1015797928606}, timestamp = {Sat, 27 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/ir/SoboroffN02.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/cia/CostKMNS02, author = { and and and and }, editor = { and and }, title = {Integrating Distributed Information Sources with {CARROT} {II}}, booktitle = {Cooperative Information Agents VI, 6th International Workshop, {CIA} 2002, Madrid, Spain, September 18-20, 2002, Proceedings}, series = {Lecture Notes in Computer Science}, volume = {2446}, pages = {194--201}, publisher = {Springer}, year = {2002}, url = {https://doi.org/10.1007/3-540-45741-0\_17}, doi = {10.1007/3-540-45741-0\_17}, timestamp = {Tue, 14 May 2019 10:00:54 +0200}, biburl = {https://dblp.org/rec/conf/cia/CostKMNS02.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/trec/CostKMNS02, author = { and and and and }, editor = { and }, title = {{CARROTT} 11 and the {TREC} 11 Web Track}, booktitle = {Proceedings of The Eleventh Text REtrieval Conference, {TREC} 2002, Gaithersburg, Maryland, USA, November 19-22, 2002}, series = {{NIST} Special Publication}, volume = {500-251}, publisher = {National Institute of Standards and Technology {(NIST)}}, year = {2002}, url = {http://trec.nist.gov/pubs/trec11/papers/umbc.kallurkar.pdf}, timestamp = {Thu, 12 Mar 2020 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/trec/CostKMNS02.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/wrac/KagalPCTZFJPCN02, author = { and and and and and and and and and }, editor = { and and }, title = {Agents Making Sense of the Semantic Web}, booktitle = {Innovative Concepts for Agent-Based Systems, First International Workshop on Radical Agent Concepts, {WRAC} 2002, McLean, VA, USA, January 16-18, 2002, Revised Papers}, series = {Lecture Notes in Computer Science}, volume = {2564}, pages = {417--433}, publisher = {Springer}, year = {2002}, url = {https://doi.org/10.1007/978-3-540-45173-0\_32}, doi = {10.1007/978-3-540-45173-0\_32}, timestamp = {Sun, 25 Oct 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/conf/wrac/KagalPCTZFJPCN02.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{10.1145/383952.383961, author = { and }, title = {Ranking Retrieval Systems without Relevance Judgments}, year = {2001}, isbn = {1581133316}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/383952.383961}, doi = {10.1145/383952.383961}, abstract = {The most prevalent experimental methodology for comparing the effectiveness of information retrieval systems requires a test collection, composed of a set of documents, a set of query topics, and a set of relevance judgments indicating which documents are relevant to which topics. It is well known that relevance judgments are not infallible, but recent retrospective investigation into results from the Text REtrieval Conference (TREC) has shown that differences in human judgments of relevance do not affect the relative measured performance of retrieval systems. Based on this result, we propose and describe the initial results of a new evaluation methodology which replaces human relevance judgments with a randomly selected mapping of documents to topics which we refer to aspseudo-relevance judgments.Rankings of systems with our methodology correlate positively with official TREC rankings, although the performance of the top systems is not predicted well. The correlations are stable over a variety of pool depths and sampling techniques. With improvements, such a methodology could be useful in evaluating systems such as World-Wide Web search engines, where the set of documents changes too often to make traditional collection construction techniques practical.}, booktitle = {Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval}, pages = {66–73}, numpages = {8}, location = {New Orleans, Louisiana, USA}, series = {SIGIR '01}, selected={true} } @inproceedings{DBLP:conf/esaw/PerichKCTZFJPCN01, author = { and and and and and and and and and }, editor = { and and }, title = {{ITTALKS:} An Application of Agents in the Semantic Web}, booktitle = {Engineering Societies in the Agents World II, Second International Workshop, {ESAW} 2001, Prague, Czech Republic, July 7, 2001, Revised Papers}, series = {Lecture Notes in Computer Science}, volume = {2203}, pages = {175--194}, publisher = {Springer}, year = {2001}, url = {https://doi.org/10.1007/3-540-45584-1\_12}, doi = {10.1007/3-540-45584-1\_12}, timestamp = {Sun, 25 Oct 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/conf/esaw/PerichKCTZFJPCN01.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/semweb/CostFJPNCKPZT01, author = { and and and and and and and and and }, editor = { and and J{\'{e}}r{\^{o}} and }, title = {{ITTALKS:} {A} Case Study in the Semantic Web and {DAML}}, booktitle = {Proceedings of SWWS'01, The first Semantic Web Working Symposium, Stanford University, California, USA, July 30 - August 1, 2001}, pages = {477--494}, year = {2001}, url = {http://www.semanticweb.org/SWWS/program/full/paper41.pdf}, timestamp = {Fri, 19 Sep 2003 08:41:21 +0200}, biburl = {https://dblp.org/rec/conf/semweb/CostFJPNCKPZT01.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/sigir/SoboroffNC01, author = { and and }, editor = { and and and }, title = {Ranking Retrieval Systems without Relevance Judgments}, booktitle = {{SIGIR} 2001: Proceedings of the 24th Annual International {ACM} {SIGIR} Conference on Research and Development in Information Retrieval, September 9-13, 2001, New Orleans, Louisiana, {USA}}, pages = {66--73}, publisher = {{ACM}}, year = {2001}, url = {https://doi.org/10.1145/383952.383961}, doi = {10.1145/383952.383961}, timestamp = {Tue, 06 Nov 2018 11:07:24 +0100}, biburl = {https://dblp.org/rec/conf/sigir/SoboroffNC01.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/vissym/AtkisonPNEAM01, author = { and and and and and }, editor = { and and }, title = {Case Study: Visualization and Information Retrieval Techniques for Network Intrusion Detection}, booktitle = {3rd Joint Eurographics - {IEEE} {TCVG} Symposium on Visualization, VisSym 2001, Ascona, Switzerland, May 28-30, 2001}, pages = {283--290}, publisher = {Eurographics Association}, year = {2001}, url = {https://doi.org/10.1007/978-3-7091-6215-6\_30}, doi = {10.1007/978-3-7091-6215-6\_30}, timestamp = {Mon, 12 Oct 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/conf/vissym/AtkisonPNEAM01.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/jodi/MillarSLN00, author = { and and and }, title = {Performance and Scalability of a Large-Scale N-gram Based Information Retrieval System}, journal = {J. Digit. Inf.}, volume = {1}, number = {5}, year = {2000}, url = {http://journals.tdl.org/jodi/article/view/22}, timestamp = {Tue, 25 Jan 2011 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/jodi/MillarSLN00.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/sigir/SoboroffN00, author = { and }, editor = { and and and Mun{-}}, title = {Collaborative filtering and the generalized vector space model}, booktitle = {{SIGIR} 2000: Proceedings of the 23rd Annual International {ACM} {SIGIR} Conference on Research and Development in Information Retrieval, July 24-28, 2000, Athens, Greece}, pages = {351--353}, publisher = {{ACM}}, year = {2000}, url = {https://doi.org/10.1145/345508.345646}, doi = {10.1145/345508.345646}, timestamp = {Tue, 06 Nov 2018 11:07:24 +0100}, biburl = {https://dblp.org/rec/conf/sigir/SoboroffN00.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/jodl/ShawKSENZMR99, author = { and and and and and and and }, title = {Interactive Volumetric Information Visualization for Document Corpus Management}, journal = {Int. J. Digit. Libr.}, volume = {2}, number = {2-3}, pages = {144--156}, year = {1999}, url = {https://doi.org/10.1007/s007990050043}, doi = {10.1007/s007990050043}, timestamp = {Fri, 11 Sep 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/jodl/ShawKSENZMR99.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/sigir/SoboroffNP99, author = { and and }, title = {Workshop on Recommender Systems: Algorithms and Evaluation}, journal = {{SIGIR} Forum}, volume = {33}, number = {1}, pages = {36--43}, year = {1999}, url = {https://doi.org/10.1145/331403.331413}, doi = {10.1145/331403.331413}, timestamp = {Tue, 06 Nov 2018 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/sigir/SoboroffNP99.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/pdpta/MillerSLNC99, author = { and and and and }, editor = {}, title = {Techniques for Gigabyte-Scale N-gram Based Information Retrieval on Personal Computers}, booktitle = {Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications, {PDPTA} 1999, June 28 - Junlly 1, 1999, Las Vegas, Nevada, {USA}}, pages = {1410--1416}, publisher = {{CSREA} Press}, year = {1999}, timestamp = {Tue, 20 Apr 2004 14:47:51 +0200}, biburl = {https://dblp.org/rec/conf/pdpta/MillerSLNC99.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/ep/NicholasD98, author = { and }, editor = { and and }, title = {Spotting Topics with the Singular Value Decomposition}, booktitle = {Principles of Digital Document Processing, 4th International Workshop, PODDP'98, Saint Malo, France, March 29-30, 1998, Proceedings}, series = {Lecture Notes in Computer Science}, volume = {1481}, pages = {82--91}, publisher = {Springer}, year = {1998}, url = {https://doi.org/10.1007/3-540-49654-8\_7}, doi = {10.1007/3-540-49654-8\_7}, timestamp = {Tue, 14 May 2019 10:00:48 +0200}, biburl = {https://dblp.org/rec/conf/ep/NicholasD98.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/atal/CostSLFMN97, author = { and and and and and }, editor = { and and }, title = {{TKQML:} {A} Scripting Tool for Building Agents}, booktitle = {Intelligent Agents IV, Agent Theories, Architectures, and Languages, 4th International Workshop, {ATAL} '97, Providence, Rhode Island, USA, July 24-26, 1997, Proceedings}, series = {Lecture Notes in Computer Science}, volume = {1365}, pages = {339--343}, publisher = {Springer}, year = {1997}, url = {https://doi.org/10.1007/BFb0026770}, doi = {10.1007/BFb0026770}, timestamp = {Tue, 29 Dec 2020 18:27:45 +0100}, biburl = {https://dblp.org/rec/conf/atal/CostSLFMN97.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/cikm/SoboroffNKE97, author = { and and and }, title = {Visualizing Document Authorship Using n-grams and Latent Semantic Indexing}, booktitle = {Proceedings of the Workshop on New Paradigms in Information Visualization and Manipulation {(NIPV} '97), in conjuction with {CIKM} '97, November 10-14, 1997, Las Vegas, NV, {USA}}, pages = {43--48}, publisher = {{ACM}}, year = {1997}, url = {https://doi.org/10.1145/275519.275529}, doi = {10.1145/275519.275529}, timestamp = {Wed, 14 Nov 2018 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/cikm/SoboroffNKE97.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/tcltk/CostSLFMN97, author = { and and and and and . Nicholas}, editor = { and }, title = {Agent Development Support for Tcl}, booktitle = {Proceedings of the Fifth Annual Tcl/Tk Workshop 1997, Boston, Massachusetts, USA, July 14-17, 1997}, publisher = {{USENIX} Association}, year = {1997}, url = {https://www.usenix.org/legacy/publications/library/proceedings/tcl97/poster\_cost.html}, timestamp = {Tue, 02 Feb 2021 08:04:40 +0100}, biburl = {https://dblp.org/rec/conf/tcltk/CostSLFMN97.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/jasis/PearceN96, author = { and }, title = {{TELLTALE:} Experiments in a Dynamic Hypertext Environment for Degraded and Multilingual Data}, journal = {J. Am. Soc. Inf. Sci.}, volume = {47}, number = {4}, pages = {263--275}, year = {1996}, url = {https://doi.org/10.1002/(SICI)1097-4571(199604)47:4\&\#60;263::AID-ASI2\&\#62;3.0.CO;2-V}, doi = {10.1002/(SICI)1097-4571(199604)47:4\&\#60;263::AID-ASI2\&\#62;3.0.CO;2-V}, timestamp = {Wed, 13 May 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/jasis/PearceN96.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/cn/RoweN95, author = { and }, title = {Reliability of {WWW} Name Servers}, journal = {Comput. Networks {ISDN} Syst.}, volume = {27}, number = {6}, pages = {773--780}, year = {1995}, url = {https://doi.org/10.1016/0169-7552(95)00045-9}, doi = {10.1016/0169-7552(95)00045-9}, timestamp = {Thu, 14 May 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/cn/RoweN95.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/epodd/NicholasR93, author = { and }, title = {Canto: a Hypertext Data Model}, journal = {Electron. Publ.}, volume = {6}, number = {2}, pages = {93--113}, year = {1993}, timestamp = {Tue, 01 Sep 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/epodd/NicholasR93.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/ijcis/NicholasY93, author = { and }, title = {Information and Knowledge Management: Guest Editors' Introduction}, journal = {Int. J. Cooperative Inf. Syst.}, volume = {2}, number = {2}, pages = {105--106}, year = {1993}, url = {https://doi.org/10.1142/S0218215793000216}, doi = {10.1142/S0218215793000216}, timestamp = {Thu, 18 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/ijcis/NicholasY93.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/ijcis/MayfieldN93, author = { and }, title = {Snitch: Augmenting Hypertext Documents with a Semantic Net}, journal = {Int. J. Cooperative Inf. Syst.}, volume = {2}, number = {3}, pages = {335--354}, year = {1993}, url = {https://doi.org/10.1142/S0218215793000150}, doi = {10.1142/S0218215793000150}, timestamp = {Thu, 18 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/ijcis/MayfieldN93.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/cikm/PearceN93, author = { and }, editor = { and and }, title = {Generating a Dynamic Hypertext Environment with n-gram Analysis}, booktitle = {{CIKM} 93, Proceedings of the Second International Conference on Information and Knowledge Management, Washington, DC, USA, November 1-5, 1993}, pages = {148--153}, publisher = {{ACM}}, year = {1993}, url = {https://doi.org/10.1145/170088.170121}, doi = {10.1145/170088.170121}, timestamp = {Tue, 06 Nov 2018 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/cikm/PearceN93.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/epodd/NicholasW92, author = { and }, title = {On the Interchangeability of {SGML} and {ODA}}, journal = {Electron. Publ.}, volume = {5}, number = {3}, pages = {105--130}, year = {1992}, timestamp = {Tue, 01 Sep 2020 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/epodd/NicholasW92.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/tse/MamrakKNS89, author = { and and and }, title = {Chameleon: {A} System for Solving the Data-Translation Problem}, journal = {{IEEE} Trans. Software Eng.}, volume = {15}, number = {9}, pages = {1090--1108}, year = {1989}, url = {https://doi.org/10.1109/32.31367}, doi = {10.1109/32.31367}, timestamp = {Wed, 17 May 2017 01:00:00 +0200}, biburl = {https://dblp.org/rec/journals/tse/MamrakKNS89.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @article{DBLP:journals/cacm/MamrakKNS87, author = { and and and }, title = {A Software Architecture for Supporting the Exchange of Electronic Manuscripts}, journal = {Commun. {ACM}}, volume = {30}, number = {5}, pages = {408--414}, year = {1987}, url = {https://doi.org/10.1145/22899.22905}, doi = {10.1145/22899.22905}, timestamp = {Tue, 06 Nov 2018 00:00:00 +0100}, biburl = {https://dblp.org/rec/journals/cacm/MamrakKNS87.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{DBLP:conf/icdcs/MamrakMGJN82, author = { and and and and }, title = {Guest Layer in Distributed Processing Support on Local Operating Systems}, booktitle = {Proceedings of the 3rd International Conference on Distributed Computing Systems, Miami/Ft. Lauderdale, Florida, USA, October 18-22, 1982}, pages = {854--859}, publisher = {{IEEE} Computer Society}, year = {1982}, timestamp = {Fri, 19 Mar 2021 00:00:00 +0100}, biburl = {https://dblp.org/rec/conf/icdcs/MamrakMGJN82.bib}, bibsource = {dblp computer science bibliography, https://dblp.org} } @inproceedings{Bouthillier2021, abstract = {Strong empirical evidence that one machine-learning algorithm A outperforms another one B ideally calls for multiple trials optimizing the learning pipeline over sources of variation such as data sampling, data augmentation, parameter initialization, and hyperparameters choices. This is prohibitively expensive, and corners are cut to reach conclusions. We model the whole benchmarking process, revealing that variance due to data sampling, parameter initialization and hyperparameter choice impact markedly the results. We analyze the predominant comparison methods used today in the light of this variance. We show a counter-intuitive result that adding more sources of variation to an imperfect estimator approaches better the ideal estimator at a 51 times reduction in compute cost. Building on these results, we study the error rate of detecting improvements, on five different deep-learning tasks/architectures. This study leads us to propose recommendations for performance comparisons.}, archivePrefix = {arXiv}, arxivId = {2103.03098}, author = {Bouthillier, , , Justin and and and Madan, Kanika and and and Serdyuk, Dmitriy and and Pal, Chris and Varoquaux, Ga{\"{e}}l and }, booktitle = {Machine Learning and Systems (MLSys)}, eprint = {2103.03098}, title = {{Accounting for Variance in Machine Learning Benchmarks}}, url = {http://arxiv.org/abs/2103.03098}, pdf = {http://arxiv.org/pdf/2103.03098}, year = {2021} } @inproceedings{Raff2021, abstract = {K-Means++ and its distributed variant K-Means‖ have become de facto tools for selecting the initialseeds of K-means. While alternatives have been developed, the effectiveness, ease of implementation, and theoretical grounding of the K-means++ and ‖ methods have made them difficult to “best” from a holistic perspective. By considering the limited opportunities within seed selection to perform pruning, we develop specialized triangle inequalitypruning strategies and a dynamic priority queue to show the first acceleration of K-Means++ and K-Means‖ that is faster in run-time while being algorithmicly equivalent. For both algorithms we are able to reduce distance computations by over500×. For K-means++ this results in up to a 17×speedup in run-time and a 551× speedup for K-means‖. We achieve this with simple, but carefully chosen, modifications to known techniques which makes it easy to integrate our approach into existing implementations of these algorithms}, author = {}, booktitle = {30th International Joint Conference on Artificial Intelligence (IJCAI-21)}, title = {{Exact Acceleration of K-Means ++ and K-Means}}, url = {https://arxiv.org/abs/2105.02936}, pdf = {https://arxiv.org/pdf/2105.02936}, year = {2021} } @inproceedings{Ordun2021, abstract = {Thermal images reveal medically important physiological information about human stress, signs of inflammation, and emotional mood that cannot be seen on visible images. Providing a method to generate thermal faces from visible images would be highly valuable for the telemedicine community in order to show this medical information. To the best of our knowledge, there are limited works on visible-to-thermal (VT) face translation, and many current works go the opposite direction to generate visible faces from thermal surveillance images (TV) for law enforcement applications. As a result, we introduce favtGAN, a VT GAN which uses the pix2pix image translation model with an auxiliary sensor label prediction network for generating thermal faces from visible images. Since most TV methods are trained on only one data source drawn from one thermal sensor, we combine datasets from faces and cityscapes. These combined data are captured from similar sensors in order to bootstrap the training and transfer learning task, especially valuable because visible-thermal face datasets are limited. Experiments on these combined datasets show that favtGAN demonstrates an increase in SSIM and PSNR scores of generated thermal faces, compared to training on a single face dataset alone.}, archivePrefix = {arXiv}, arxivId = {2106.08091}, author = { and and }, booktitle = {ICIP}, eprint = {2106.08091}, title = {{Generating Thermal Human Faces for Physiological Assessment Using Thermal Sensor Auxiliary Labels}}, url = {http://arxiv.org/abs/2106.08091}, year = {2021} } @inproceedings{Nguyen2021, abstract = {The detection of malware is a critical task for the protection of computing environments. This task often requires extremely low false positive rates (FPR) of 0.01{\%} or even lower, for which modern machine learning has no readily available tools. We introduce the first broad investigation of the use of uncertainty for malware detection across multiple datasets, models, and feature types. We show how ensembling and Bayesian treatments of machine learning methods for static malware detection allow for improved identification of model errors, uncovering of new malware families, and predictive performance under extreme false positive constraints. In particular, we improve the true positive rate (TPR) at an actual realized FPR of 1e-5 from an expected 0.69 for previous methods to 0.80 on the best performing model class on the Sophos industry scale dataset. We additionally demonstrate how previous works have used an evaluation protocol that can lead to misleading results.}, archivePrefix = {arXiv}, arxivId = {2108.04081}, author = {. and and and }, booktitle = {IJCAI-21 1st International Workshop on Adaptive Cyber Defense}, eprint = {2108.04081}, file = {:Users/eman7613/Downloads/2108.04081.pdf:pdf}, title = {{Leveraging Uncertainty for Improved Static Malware Detection Under Extreme False Positive Constraints}}, url = {http://arxiv.org/abs/2108.04081}, year = {2021} } \graphicspath{{Pics/}} \newpage \section{Orthocenter--Circumcircle--NinePoint Circle} Notes everyone need to memorize by heart top to bottom: \begin{enumerate} [itemsep=0pt] \item \href{http://yufeizhao.com/olympiad/imo2008/zhao-circles.pdf}{Circles - Yufei Zhao} \item \href{http://yufeizhao.com/olympiad/cyclic_quad.pdf}{Big Picture - Yufei Zhao} \item \href{http://yufeizhao.com/olympiad/power_of_a_point.pdf}{POP - Yufei Zhao} \item \href{http://yufeizhao.com/olympiad/three_geometry_lemmas.pdf}{3 Lemmas - Yufei Zhao} \end{enumerate} \den{The usual notations}{ Unless stated otherwise, we assume that $\triangle ABC$ is an arbitrary triangle, with circumcenter $O$, orthocenter $H$, $\omega$ is the circumcircle. Usually, $DEF$ is the orthic triangle in this chapter. $MNP$ is the median triangle. } \begin{minipage}{.65\textwidth} \lem{Collinearity with antipode and center}{ Let $A'$ be the antipode of $ A $ in $ \odot ABC $. Let $BDEC$ be a cyclic quadrilateral with $ D\in AB $ and $ E\in AC $. Let $ P $ be the center of $ BDEC $. Also, let $ X=BE\cap CD $. Then $ A', P, X $ are collinear. } \solu{ Using ``The Big Picture" property to show that if $ Q=\odot ADE \cap \odot ABC $, then $ P, X, Q $ collinear and $ PQ\perp AQ $. Which implies that $ P, A', Q $ are collinear. } \end{minipage}\hfill% \begin{minipage}{.33\textwidth} \figdf{}{ChinaTST2018T1P3_lemm}{} \end{minipage} \begin{minipage}{.65\linewidth} \lem{Weird point Y}{ Let $Y$ be a point on $AC$ such that $\triangle CBY \sim CAB$. Let $E$ bet the foot of $B$ on $AC$, let $N$ be the midpoint of $AB$. Then $NE, CO, BE$ are concurrent. } \begin{solution} Draw te circles, $BNKO$ and $BKYC$. \end{solution} \end{minipage}\hfill% \begin{minipage}{.33\linewidth} \figdf{}{CHKMO_2014_p4_lem_1}{} \end{minipage} \newpage \subsection{Problems} \begin{minipage}{.55\textwidth} \prob{https://artofproblemsolving.com/community/c6h1441717p8209926}{Balkan MO 2017 P3}{S}{ Let $t_B$ and $t_C$ be the tangents $\omega$ at $B$ and $C$, they meet at $L$. The straight lines passing through $B, C$ and parallel to $AC, AB$ intersects $t_C, t_B$ at points $D, E$ respectively. $T = AC \cap \odot BDC, S = AB\cap \odot CBE$. Prove that $ST$, $AL$, and $BC$ are concurrent. } \begin{solution} We have $\triangle ABT \sim \triangle ACB \sim ASC$, which leads to $BT\parallel CS$, and $AL$ becomes the median of $\triangle ABT$. \end{solution} \vspace{1em} \prob{https://artofproblemsolving.com/community/c4512_2014_usamo}{USAMO 2014 P5}{E}{ Let $P$ be the second intersection of $\odot AHC$ with the internal bisector of $\angle BAC$. Let $X$ be the circumcenter of triangle $APB$ and $Y$ the orthocenter of triangle $APC$. Prove that the length of segment $XY$ is equal to the circumradius of triangle $ABC$. } \solu{ No length conditions given, yet we need to prove that two lengths are equal. \emph{Parallelogram} !!! Just need to prove that $ Y\in \odot ABC \text{ and } YD\perp AB $ } \vspace{1em} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ $P$ is a point. $(K)$ is the circle with diameter $AP$. $(K)$ cuts $CA, AB$ again at $ E, F $. $ PH $ cuts $ (K) $ again at $ G $. Tangent line at $E, F $ of $ (K) $ intersect at $ T $. $ M $ is midpoint of $ BC $. $L$ is the point on $ MG $ such that $ AL \parallel MT. $ Prove that $LA \perp LH $. } \begin{solution}[Phantom Point] Take $ L' = MG \cap AZYH $, then use spiral similarity to show that $ AL'\parallel MT $. \end{solution} \end{minipage}\hfill% \begin{minipage}{.4\textwidth} \figdf{.7}{Balkan_MO_2017_P3}{} \figdf{.99}{USAMO_2014_P5}{} \figdf{.9}{SATST2015proposed_by_bura/derakynay1134-1}{} \end{minipage} \newpage \begin{minipage}{.5\textwidth} \prob{}{Bewarish 1}{E}{ Let $ DEF $ be the orthic triangle, and let $ EF\cap BC = P $. Let the tangent at $ A $ to $ \odot ABC $ meet $ BC $ at $ Q $. Let $ T $ be the reflection of $ Q $ over $ P $. Let $ K $ be the orthogonal projection of $ H $ on $ AM $. Prove that $ \angle OKT = 90 $. } \solu{ Spiral similarity from $O$ to get rid of $Q$ and $T$. Then spiral similarity again from $P$ to get a trivial circle. } \end{minipage}\hfill% \begin{minipage}{.48\textwidth} \figdf{.99}{Bewarish_1}{} \end{minipage} \begin{minipage}{.6\linewidth} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ $P$ lies on $(O)$. The line passes through $P$ and parallel to $BC$ cuts $CA$ at $E$. $K$ is circumcenter of triangle $PCE$ and $L$ is nine point center of triangle $PBC$. Prove that the line passes through $L$ and parallel to $PK,$ always passes through a fixed point when $P$ moves. } \begin{solution}[Construction] Notice that if we reflect $ P $ over $ L $ to get $ P' $, then $ OP=AH $ and $ OP\perp BC $ where $ O $ is the circumcenter of $ \odot ABC $. Which trivially implies that the line throught $ L $ passes throught the midpoint of $ P'D $ where $ D $ is the reflection of $ H $ over $ BC $. \end{solution} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ Altitude $AH$, $H$ lies on $BC$. $P$ is a point that lies on bisector $\angle BAC$ and $P$ is inside triangle $ABC$. Circle diameter $AP$ cuts $(O)$ again at $G$. $L$ is projection of $P$ on $AH$. Assume that $GL$ bisects $HP$. Prove that $P$ is incenter of $ABC$. } \begin{solution}[Angle Chase] Since $ \angle APL = \angle ABD = \angle AGD $, $ G, L, M $ are collinear. Let $ E\in BC $ and $ PE\perp BC $. Then $ E $ also lies on $ DG $. Again we have, $ \triangle DPE\sim \triangle DGP $. Which implies $ DP=DB=DC $. \end{solution} \end{minipage}\hfill% \begin{minipage}{.35\linewidth} \figdf{}{SATST2015proposed_by_bura/derakynay1134-2}{} \figdf{}{SATST2015proposed_by_bura/derakynay1134-4}{} \end{minipage} \newpage \begin{minipage}{.6\linewidth} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ $M$ lies on small arc $\overline{BC}$ . $P$ lies on $AM$. Circle diameter $MP$ cuts $(O)$ again at $N$. $MO$ cuts circle diameter $MP$ again at $Q$. $AN$ cuts circle diameter $MP$ again at $R$. Prove that $\angle PRA = \angle PQA$. } \begin{solution}[Angle Chase] Let $ MO \cap \odot ABC = D $. Becase $ NP\perp MN $, we have $ N, P, D $ collinear, and $ APQD $ cyclic. So, $ \triangle APQ\sim \triangle ANM \sim \triangle APR $. \end{solution} \vspace{1em} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ Let $ABC$ be right triangle with hypotenuse $BC$, bisector $BE$, $E$ lies on $CA$. Assume that circumcircle of triangle $BCE$ cuts segment $AB$ again at $F$. $K$ is projection of $A$ on $BC$. $L$ lies on segment $AB$ such that $BL = BK$. Prove that $\dfrac{AL}{AF} = \sqrt{\dfrac{BK}{BC}}$. } \vspace{1em} \prob{https://artofproblemsolving.com/community/c74453h1225408_some_geometric_problems} {Saudi Arab 2015}{E}{ $AD$ is diameter of $(O)$. $M, N$ lie on $BC$ such that $OM \parallel AB$, $ON \parallel AC$. $DM, DN$ cut $(O)$ again at $P, Q$. Prove that $BC = DP = DQ$. } \begin{solution} We will prove that $MB = MD$. Since $OD = OA$, we have $EM = MD$. And since $\triangle EBD$ is a right triangle, $MB = ME = MD$. And so the arcs $PB$ and $DC$ are equal. \end{solution} \end{minipage}\hfill% \begin{minipage}{.35\linewidth} \figdf{}{SATST2015proposed_by_bura/derakynay1134-5}{} \figdf{}{SATST2015proposed_by_bura/derakynay1134-6}{} \figdf{}{SATST2015proposed_by_bura/derakynay1134-7}{} \end{minipage} \newpage \prob{http://igo-official.ir/wp-content/uploads/2017/09/4th_IGO_Problems-and-Solutions.pdf} {IGO 2017 A3}{E}{ Let $O$ be the circumcenter of $\triangle ABC$. Line $CO$ intersects the altitude through $A$ at point $K$. Let $P, M$ be the midpoints of $AK, AC$ respectively. If $PO$ intersects $BC$ at $Y$ , and the circumcircle of $\triangle BCM$ meets $AB$ at $X$, prove that $BXOY$ is cyclic } \begin{minipage}{.5\linewidth} \solu{ We want to prove that $\angle POX = \angle AMX$. But we also notice by Reim's theorem that $\angle BOY = \angle BMC$, which leads to $\angle XPO = \angle XAM$. \\ Now, we want to show that \[\frac{XA}{AP} = \frac{XM}{OM}\] Which is simple length chase. } \end{minipage}\hfill% \begin{minipage}{.49\linewidth} \figdf{}{IGO_2017_Advanced_P3}{} \end{minipage} \begin{minipage}{.5\linewidth} \prob{} {}{EM}{ Let $EF\parallel BC$ be two points on the circumcircle. Let $D$ be the center of $HE$, and let $K$ be the point on $AB$ for which $OK\parallel AF$. Prove that $DK\perp DC$. } \begin{solution} The main part of the solution is to get rid of $OK$ in some way. We take $T$ to be the point such that $T\in OK$ and $\angle OTC = 90$. And let $S = CT\cap \odot ABC$. Then we have, $ES\perp AB$, and if we can show that $D$ lies on $\odot PTC$, we are done. But then that is straightforward as $DT\parallel FS\parallel CH$ and $DP\parallel EQ$. \end{solution} \end{minipage}\hfill% \begin{minipage}{.48\linewidth} \figdf{}{bdmo_forum_geo_1}{} \end{minipage} \newpage \prob{https://artofproblemsolving.com/community/c6h1615884p10101617}{Turkey TST 2018 P4}{E}{ In a non-isosceles acute triangle $ABC$, $D$ is the midpoint of $BC$. The points $E$ and $F$ lie on $AC$ and $AB$, respectively, and the circumcircles of $CDE$ and $AEF$ intersect in $P$ on $AD$. The angle bisector from $P$ in triangle $EFP$ intersects $EF$ in $Q$. Prove that the tangent line to the circumcircle of $AQP$ at $A$ is perpendicular to $BC$. } \begin{minipage}{.5\linewidth} \begin{solution}[angle chase] Note that $AQ$ is the angle bisector of $\angle BAC$. Using this fact, we can easily prove that $\angle HAQ = \angle APQ$. \end{solution} \solu{[inversion] Inverting around $A$ with radius $\sqrt{AP\cdot AD}$ sends $EF$ to the circumcircle of $ABC$ and $P$ to $D$. Since $AQ$ bisects $\angle BAC$, we have $DQ'\perp BC$. } \end{minipage}\hfill% \begin{minipage}{.47\linewidth} \figdf{}{Turkey_TST_2018_p4}{} \end{minipage} \prob{https://artofproblemsolving.com/community/q1h1970138p13654481}{USA Winter TST 2020 P2}{E}{ Two circles $\Gamma_1$ and $\Gamma_2$ have common external tangents $\ell_1$ and $\ell_2$ meeting at $T$. Suppose $\ell_1$ touches $\Gamma_1$ at $A$ and $\ell_2$ touches $\Gamma_2$ at $B$. A circle $\Omega$ through $A$ and $B$ intersects $\Gamma_1$ again at $C$ and $\Gamma_2$ again at $D$, such that quadrilateral $ABCD$ is convex. Suppose lines $AC$ and $BD$ meet at point $X$, while lines $AD$ and $BC$ meet at point $Y$. Show that $T$, $X$, $Y$ are collinear. } \begin{solution}[Radical Axis] It is easy to see that $ X $ lies on the radical axis of $ \Gamma_1 $ and $ \Gamma_2 $. Let $ B' = l_1\cap \Gamma_2 $ and $ A' = l_2\cap \Gamma_1 $. Let $ C'=A'X\cap \Gamma_1 $ and $ D'=B'X\cap \Gamma_2 $. Let $ A'C\cap AC'=Z $.\\ We have $ AD'CB' $ and $ A'DC'B $ cyclic. Also $ T, D, C' $ and $ T, D', C $ are collinear. Which implies $ A'D'CB $ and $ ADC'B' $ are cyclic too.\\ Applying pascal on $ AAC'CA'A' $, we have $ T, X, Z $ are collinear.\\ Now, it is easy to see that $ Z, Y, T $ lie on the radical axis of $ A'D'CB $ and $ ADC'B' $. So we have $ T, X, Y, Z $ collinear. \end{solution} \figdf{.8}{USA_Winter_TST_2020_P2}{} \vspace{1em} \begin{solution}[mOvInG pOiNtS, by shawnee03] Fix $\Gamma_1$ and $\Gamma_2$ (and hence $\ell, T, A, B$) and animate $X$ linearly on $\ell$. Then \begin{itemize} \item $C$ moves projectively on $\Gamma_1$ (it is the image of the perspectivity through $A$ from $\ell$ to $\Gamma_1$) and thus has degree $2$, and similarly for $D$. \item $\overline{AD}$ has degree at most $0+2=2$, and similarly for $\overline{BC}$. \item $Y=\overline{AD}\cap\overline{BC}$ has degree at most $2+2=4.$ \item The collinearity of $T,X,Y$ has degree at most $0+1+4=5.$ \end{itemize} Thus it suffices to verify the problem for six different choices of $X$. We choose: \begin{itemize} \item $\ell\cap \ell_1$: here $Y$ approaches $A$ as $X$ approaches $\ell\cap \ell_1$. \item $\ell\cap\ell_2$: here $Y$ approaches $B$ as $X$ approaches $\ell\cap \ell_2$. \item $\ell\cap \overline{AB}$: here $Y$ approaches $\ell\cap \overline{AB}$ as $X$ approaches $\ell\cap \overline{AB}$. \item the point at infinity along $\ell$: here $Y=T$. \item the two intersections of $\Gamma_1$ and $\Gamma_2$: here $Y=X$. \end{itemize} (The final two cases may be chosen because we know that there exists a choice of $A,B,C,D$ for which $ABCD$ is convex; this forces $\Gamma_1$ and $\Gamma_2$ to intersect.) \end{solution} \gene{https://artofproblemsolving.com/community/c6h1970138p13654501} {USA Winter TST 2020 P2}{ Let $ABCD$ be a cyclic quadrilateral, $X=AC\cap BD$, and $Y=AB\cap CD$. Let $T$ be a point on line $XY$, $\Gamma_1$ be the circle through $A$ and $C$ tangent to $TA$, and $\Gamma_2$ be the circle through $B$ and $D$ tangent to $TD$. Then $\Gamma_1$ and $\Gamma_2$ are viewed at equal angles from $T$. } \begin{minipage}{.5\linewidth} \begin{solution}[Length Chase, by a1267ab] If the radiuses of $ \Gamma_1 $ and $ \Gamma_2 $ are $ r_1, r_2 $, then we have to show, \[\frac{TA}{r_1}=\frac{TD}{r_2}\] We have, \[r_1= \frac{AB}{2\sin\angle TAB},\ r_2= \frac{CD}{2\sin\angle TDC}\] To get the sine ratios, we compare the areas of $ \triangle TAB $ and $ \triangle TDC $. We have, \[\begin{aligned} \frac{TA\cdot AB\ \sin\angle TAB}{TD\cdot CD\ \sin\angle TDC} &= \frac{[TAB]}{[TCD]}\\ &= \frac{[XAB]}{[XCD]} = \frac{AB^2}{CD^2}\\[.7em] \implies \frac{r_1}{TA}&=\frac{r_2}{TD} \end{aligned}\] \end{solution} \end{minipage}\hfill% \begin{minipage}{.5\linewidth} \figdf{.9}{USA_Winter_TST_2020_P2_generalization}{} \end{minipage} \begin{minipage}{.5\linewidth} \prob{https://artofproblemsolving.com/community/c5h1629606p10226149} {USAJMO 2018 P3}{E}{ Let $ABCD$ be a quadrilateral inscribed in circle $\omega$ with $\overline{AC} \perp \overline{BD}$. Let $E$ and $F$ be the reflections of $D$ over lines $BA$ and $BC$, respectively, and let $P$ be the intersection of lines $BD$ and $EF$. Suppose that the circumcircle of $\triangle EPD$ meets $\omega$ at $D$ and $Q$, and the circumcircle of $\triangle FPD$ meets $\omega$ at $D$ and $R$. Show that $EQ = FR$. } \end{minipage}\hfill% \begin{minipage}{.45\linewidth} \figdf{1}{USAJMO_2018_p3}{} \end{minipage} \prob{https://artofproblemsolving.com/community/c6h1289432p6815893} {IRAN 3rd Round 2016 P1}{E}{ Let $ABC$ be an arbitrary triangle, $P$ is the intersection point of the altitude from $C$ and the tangent line from $A$ to the circumcircle. The bisector of angle $A$ intersects $BC$ at $D$. $PD$ intersects $AB$ at $K$, if $H$ is the orthocenter then prove : $HK\perp AD$ } \solu{Finding a set of Collinear points.} \prob{} {}{E}{ Let $\triangle ABC$ be a triangle. $F, G$ be arbitrary points on $AB, AC$. Take $D, E$ midpoint of $BF, CG$. Show that the nine-point centers of $\triangle ABC,\ \triangle ADE,\ \triangle AFG$ are collinear. }\label{eriq_lemma_3} \prob{http://igo-official.ir/wp-content/uploads/2017/09/4th_IGO_Problems-and-Solutions.pdf} {IGO 2017 A4}{}{ Three circles $W_1 , W_2$ and $W_3$ touches a line $l$ at $A ,B ,C$ respectively ($B$ lies between $A$ and $C$). $W_2$ touches $W_1$ and $W_3$. Let $l_2$ be the other common external tangent of $W_1$ and $W_3$. $l_2$ cuts $W_2$ at $X ,Y$. Perpendicular to $l$ at $B$ intersects $W_2$ again at $K$. Prove that $KX$ and $KY$ are tangent to the circle with diameter $AC$. } \figdf{}{IGO_2017_A4}{} \begin{solution} Using the names of the vertices in the diagram, we let $UV$ be a segment parallel to $O_1, O_2$. Step by step we prove that \begin{enumerate} \item $O$ is the center of $SHB$, then $O, O_1, O_2$ are collinear. \item $ZM$ bisects $\angle XZY$. \item $BUZVM$ is cyclic. \item $ZU^2 = ZH.ZB$. \end{enumerate} \end{solution} \prob{https://artofproblemsolving.com/community/c6h1513396p8996816} {IGO 2017 A2}{E}{ We have six pairwise non-intersecting circles that the radius of each is at least one (no circle lies in the interior of any other circle). Prove that the radius of any circle intersecting all the six circles, is at least one. } \begin{solution} We first expand the circles so that they touch each other in a ring like shape. Then we take the largest diameter of the convex hexagon with the centers. We show that any circle that intersects those two cirlces must have radius at least $1$. \end{solution} \prob{https://artofproblemsolving.com/community/c6h1709992p11022223} {IGO 2017 A4}{M}{ Quadrilateral $ABCD$ is circumscribed around a circle. Diagonals $AC,BD$ are not perpendicular to each other. The angle bisectors of angles between these diagonals, intersect the segments $AB,BC,CD$ and $DA$ at points $K,L,M$ and $N$. Given that $KLMN$ is cyclic, prove that so is $ABCD$. } \begin{minipage}{.5\linewidth} \begin{solution} If we let $K', L', M', N'$ be the points where the incenter touches the sides, then we wish to prove that $K= K'$ and so on. To prove this, we first prove that $KL, MN, AC$ are concurrent. \\ Then we prove that $K'L'$ and $M'N'$ also passes through the same point. This lets us use the lemmas of complete cyclic quadrilaterals. \end{solution} \end{minipage}\hfill% \begin{minipage}{.5\linewidth} \figdf{.9}{IGO_2018_A4}{} \end{minipage} \prob{https://artofproblemsolving.com/community/c6h1632766p10256362} {ARO 2018 P10.2}{E}{ Let $\triangle ABC$ be an acute-angled triangle with $ABBC$. Let $\Omega $ be the circumcircle of $ ABC$. The tangents to $ \Omega $ at the points $A$ and $C$ meet at $P$, and $BP$ and $AC$ intersect at $S$. Let $AD$ be the altitude of the triangle $ABP$ and $\omega$ the circumcircle of the triangle $CSD$. Suppose $ \omega$ and $ \Omega $ intersect at $K\not= C$. Prove that $ \angle CKM=90^\circ $. } \prob{https://artofproblemsolving.com/community/c6h79788p456609} {APMO 1999 P3}{E}{ Let $\Gamma_1$ and $\Gamma_2$ be two circles intersecting at $P$ and $Q$. The common tangent, closer to $P$, of $\Gamma_1$ and $\Gamma_2$ touches $\Gamma_1$ at $A$ and $\Gamma_2$ at $B$. The tangent of $\Gamma_1$ at $P$ meets $\Gamma_2$ at $C$, which is different from $P$, and the extension of $AP$ meets $BC$ at $R$. Prove that the circumcircle of triangle $PQR$ is tangent to $BP$ and $BR$. } \prob{} {Simurgh 2019 P2}{E}{ Let $ ABC $ be an isosceles triangle, $ AB=AC $. Suppoe that $ Q $ is a point such that $ AQ=AB,\ AQ||BC $. Let $ P $ be the foot of perpendicular line from $ Q $ to $ BC $. Prove that the circle with diameter $ PQ $ is tangent to the circumcircle of $ ABC $. } \prob{http://emc.mnm.hr/wp-content/uploads/2018/12/EMC_2018_Seniors_ENG_Solutions-2.pdf} {European Mathematics Cup 2018 P2}{E}{ Let $ABC$ be a triangle with $|AB|<|AC|$. Let $k$ be the circumcircle of $\triangle ABC$ and let $O$ be the center of $k$. Point $M$ is the midpoint of the arc $\widehat{BC}$ of $k$ not containing $A$. Let $D$ be the second intersection of the perpendicular line from $M$ to $AB$ with $k$ and $E$ be the second intersection of the perpendicular line from $M$ to $AC$ with $k$. Points $X$ and $Y$ are the intersections of $CD$ and $BE$ with $OM$ respectively. Denote by $k_{b}$ and $k_{c}$ circumcircles of triangles $BDX$ and $CEY$ respectively. Let $G$ and $H$ be the second intersections of $k_{b}$ and $k_{c}$ with $AB$ and $AC$ respectively. Denote by $k_{a}$ the circumcircle of triangle $AGH$. Prove that $O$ is the circumcenter of $\triangle O_{a}O_{b}O_{c},$ where $O_{a}, O_{b}, O_{c}$ are the centers of $k_{a}, k_{b}, k_{c}$ respectively. } \prob{https://artofproblemsolving.com/community/c6h1789909p11836144}{RMM 2019 P2}{E}{ Let $ABCD$ be an isosceles trapezoid with $AB\parallel CD$. Let $E$ be the midpoint of $AC$. Denote by $\omega$ and $\Omega$ the circumcircles of the triangles $ABE$ and $CDE$, respectively. Let $P$ be the crossing point of the tangent to $\omega$ at $A$ with the tangent to $\Omega$ at $D$. Prove that $PE$ is tangent to $\Omega$. } \prob{https://artofproblemsolving.com/community/c6h1709995p11022258}{IGO 2018 A5}{E}{ $ABCD$ is a cyclic quadrilateral. A circle passing through $A,B$ is tangent to segment $CD$ at point $E$. Another circle passing through $C,D$ is tangent to $AB$ at point $F$. Point $G$ is the intersection point of $AE,DF$, and point $H$ is the intersection point of $BE,CF$. Prove that the incenters of triangles $AGF,BHF,CHE,DGE$ lie on a circle. } \begin{solution}[juckter] The cases where two opposite sides of $ABCD$ are parallel are easily dealt with. Let $X = AB \cap CD$. Then $XE^2 = XA \cdot XB = XC \cdot XD = XF^2$, so $XE = XF$. Reflect $E$ through $X$ onto $E'$, and notice that $XE^2 = XC \cdot XD$ implies $(C, D; E, E') = -1$. Because $\angle EFE' = 90^{\circ}$ (which follows from $XE = XF = XE'$) it follows that $FE$ bisects $\angle CFD$ and analogously $EF$ bisects $\angle AEB$. It then follows easily that $G$ and $H$ are symmetric about $EF$.\\ \figdf{.95}{IGO2018A5}{\autoref{problem:IGO 2018 A5} IGO 2018 A5} Now let $I_1, I_2, I_3$ and $I_4$ be the incenters of $AGF, DGE, CHE, BHF$ respectively. Then $I_1I_2$ and $I_3I_4$ are the external bisectors of angles $EGF$ and $EHF$ respectively, and by symmetry about $EF$ these lines intersect at a (possibly ideal) point $X \in EF$.\\ Finally, we may angle chase to find that $E, I_1, I_2, F$ and $E, I_3, I_4, F$ are quadruples of concyclic points. If $I_1I_2$ is parallel to $I_3I_4$ then we may easily conclude by symmetry about the perpendicular bisector of $EF$. Otherwise by Power of a Point from $X$ we have $XI_1 \cdot XI_2 = XE \cdot XF = XI_3 \cdot XI_4$, so $I_1, I_2, I_3, I_4$ are concyclic, as desired. \end{solution} \prob{https://artofproblemsolving.com/community/c6h418983p2365045} {ISL 2011 G8}{EM}{ Let $ABC$ be an acute triangle with circumcircle $\Gamma$. Let $\ell$ be a tangent line to $\Gamma$, and let $\ell_a, \ell_b$ and $\ell_c$ be the lines obtained by reflecting $\ell$ in the lines $BC$, $CA$ and $AB$, respectively. Show that the circumcircle of the triangle determined by the lines $\ell_a, \ell_b$ and $\ell_c$ is tangent to the circle $\Gamma$. } \figdf{.8}{ISL_2011_G8}{} \solu{ The main problem here are the reflected lines. We need to somehow know more about them. So we come up with some ways to construct the three lines without drawing the tangent $l$, which leads us to the reflection of $D$ over the three sides idea. And after doing some angle chasing to find out the angles of the triangle, we begin to see relationships between the reflection points and the vertices of the triangle. } \prob{https://artofproblemsolving.com/community/c6h1818716p12141505}{ELMO 2019 P3}{EM}{ Let $ABC$ be a triangle such that $\angle CAB > \angle ABC$, and let $I$ be its incentre. Let $D$ be the point on segment $BC$ such that $\angle CAD = \angle ABC$. Let $\omega$ be the circle tangent to $AC$ at $A$ and passing through $I$. Let $X$ be the second point of intersection of $\omega$ and the circumcircle of $ABC$. Prove that the angle bisectors of $\angle DAB$ and $\angle CXB$ intersect at a point on line $BC$. } \begin{solution}[Angle Chase] Suppose the bisector of $ \angle BAD $ meet $ BC $ at $ G' $. Then we have, \begin{align*} \angle BG'A &= \frac{\angle A-\angle B}{2}\\ \therefore \angle CG'A &= \angle B + \angle BG'A\\ &=\frac{A+B}{2}\\[1em] \implies CG'&=CA\\ \therefore \angle G'ID &= \angle B \end{align*} Now, let $ M $ be the midpoint of the minor arc $ BC $. Let $ G=XM\cap BC $. So we have \[\triangle MGI \sim \triangle MIX \implies \angle MIG = \angle MXI\] Let $ XI\cap \odot ABC = N \neq X $. Since $ AC $ is tangent to $ \odot AXI $, $ NC\parallel AM $. Which means $$ \angle MXI = \angle B = \angle MIG $$ Which completes our proof by implying that $ G'\equiv G $. \figdf{.4}{elmo_2019_P3}{} \end{solution} \prob{https://artofproblemsolving.com/community/c6h596927p3542092}{ISL 2014 G5}{M}{ Convex quadrilateral $ABCD$ has $\angle ABC = \angle CDA = 90^{\circ}$. Point $H$ is the foot of the perpendicular from $A$ to $BD$. Points $S$ and $T$ lie on sides $AB$ and $AD$, respectively, such that $H$ lies inside triangle $SCT$ and \[ \angle CHS - \angle CSB = 90^{\circ}, \quad \angle THC - \angle DTC = 90^{\circ}.\] Prove that line $BD$ is tangent to the circumcircle of triangle $TSH$. } \solu{ First construct using nice circles, then prove the center is on $ AH $ using angle bisector theorem. \figdf{.6}{ISL2014G5}{Construction} \figdf{.3}{ISL2014G5_lemma1}{Lemma} } \prob{https://artofproblemsolving.com/community/c6h1113194p5083564}{ISL 2014 G7}{M}{ Let $ABC$ be a triangle with circumcircle $\Omega$ and incentre $I$. Let the line passing through $I$ and perpendicular to $CI$ intersect the segment $BC$ and the arc $BC$ (not containing $A$) of $\Omega$ at points $U$ and $V$ , respectively. Let the line passing through $U$ and parallel to $AI$ intersect $AV$ at $X$, and let the line passing through $V$ and parallel to $AI$ intersect $AB$ at $Y$ . Let $W$ and $Z$ be the midpoints of $AX$ and $BC$, respectively. Prove that if the points $I, X,$ and $Y$ are collinear, then the points $I, W ,$ and $Z$ are also collinear. } \solu{Draw a nice diagram, and use the parallel property to find circles. \figdf{.7}{ISL2014G7}{ISL 2014 G7}} \prob{https://artofproblemsolving.com/community/c6h1112748p5079655}{ISL 2015 G6}{E}{ Let $ABC$ be an acute triangle with $AB > AC$. Let $\Gamma $ be its cirumcircle, $H$ its orthocenter, and $F$ the foot of the altitude from $A$. Let $M$ be the midpoint of $BC$. Let $Q$ be the point on $\Gamma$ such that $\angle HQA = 90^{\circ}$ and let $K$ be the point on $\Gamma$ such that $\angle HKQ = 90^{\circ}$. Assume that the points $A$, $B$, $C$, $K$ and $Q$ are all different and lie on $\Gamma$ in this order.\\ Prove that the circumcircles of triangles $KQH$ and $FKM$ are tangent to each other. } \solu{ Draw the tangent line, and find angles. \figdf{.7}{ISL2015G6}{ISL 2015 G6} } \begin{minipage}{.5\linewidth} \prob{https://artofproblemsolving.com/community/c6h1268908p6622796}{ISL 2015 G5}{M}{ Let $ABC$ be a triangle with $CA \neq CB$. Let $D$, $F$, and $G$ be the midpoints of the sides $AB$, $AC$, and $BC$ respectively. A circle $\Gamma$ passing through $C$ and tangent to $AB$ at $D$ meets the segments $AF$ and $BG$ at $H$ and $I$, respectively. The points $H'$ and $I'$ are symmetric to $H$ and $I$ about $F$ and $G$, respectively. The line $H'I'$ meets $CD$ and $FG$ at $Q$ and $M$, respectively. The line $CM$ meets $\Gamma$ again at $P$. Prove that $CQ = QP$. } \end{minipage}\hfill% \begin{minipage}{.46\linewidth} \figdf{}{ISL2015G5}{ISL 2015 G5} \end{minipage} \prob{https://artofproblemsolving.com/community/c6h418635p2361976}{ISL 2010 G5}{E}{ Let $ABCDE$ be a convex pentagon such that $BC \parallel AE,$ $AB = BC + AE,$ and $\angle ABC = \angle CDE.$ Let $M$ be the midpoint of $CE,$ and let $O$ be the circumcenter of triangle $BCD.$ Given that $\angle DMO = 90^{\circ},$ prove that $2 \angle BDA = \angle CDE.$ } \solu{First try to construct the point. Do this the long way, then find a easier way that includes $ B, C $, not $ B, A $ to do that. Then try to translate what $ 90 $ degree condition into angles, and take midpoints, since we have midpoints involved. \figdf{.7}{ISL2010G5}{ISL 2010 G5} } \prob{https://artofproblemsolving.com/community/c6h1918849p13155415}{IGO 2019 A5}{MH}{ Let points $A, B$ and $C$ lie on the parabola $\Delta$ such that the point $H$, orthocenter of triangle $ABC$, coincides with the focus of parabola $\Delta$. Prove that by changing the position of points $A, B$ and $C$ on $\Delta$ so that the orthocenter remain at $H$, inradius of triangle $ABC$ remains unchanged. } \solu{ \figdf{.7}{IGO2019G5}{IGO 2019 A5} I think the idea for inversion should have been pretty natural after finding that the incircle is fixed. } \prob{https://artofproblemsolving.com/community/c6h1140464p5353018}{Iran 3rd Round 2015 P5}{M}{ Let $ABC$ be a triangle with orthocenter $H$ and circumcenter $O$. Let $R$ be the radius of circumcircle of $\triangle ABC$. Let $A',B',C'$ be the points on $\overrightarrow{AH},\overrightarrow{BH},\overrightarrow{CH}$ respectively such that $AH.AA'=R^2,BH.BB'=R^2,CH.CC'=R^2$. Prove that $O$ is incenter of $\triangle A'B'C'$. } \solu{The condition easily leads to a nice construction of the points. It should be trivial to figure that the construction is really important. Also, noticing a similarity among the triangles is really important.} \newpage \subsection{The line parallel to BC} \den{Important Points}{ Let the line parallel to $ BC $ through $ O $ meet $ AB, AC $ at $ D, E $. Let $ K $ be the midpoint of $ AH $, $ M $ be the midpoint of $ BC $. $ F $ be the feet of $ A $-altitude on $ BC $ and let $ H' $ be the reflection of $ H $ on $ F $. Let $ O' $ be the circumcenter of $ KBC $. } \lem{}{ $ \angle DKC = \angle EKB = 90^{\circ} $ }\label{lemma:linethroughO_lemma1} \fig{1}{linethroughO_1}{} \lem{}{$ CD, BE, OH', AM, KO' $ are concurrent. (by \autoref{isogonality_lemma}) }\label{lemma:linethroughO_lemma2} \fig{1}{linethroughO_2}{} \prob{https://artofproblemsolving.com/community/c6h1412598_circumcircles_intersect_on_ao}{InfinityDots MO Problem 3}{EM}{ Let $\triangle ABC$ be an acute triangle with circumcenter $O$ and orthocenter $H$. The line through $O$ parallel to $BC$ intersect $AB$ at $D$ and $AC$ at $E$. $X$ is the midpoint of $AH$. Prove that the circumcircles of $\triangle BDX$ and $\triangle CEX$ intersect again at a point on line $AO$. } \solu{Just using \autoref{lemma:linethroughO_lemma1} to get another pair of circle where we can apply radical axis arguments.} \solu{Noticing that the resulting point is the isogonal conjugate of a well defined point,} \lem{}{ Let $ P, Q $ be on $ AB, AC $ resp. such that $ PQ\parallel BC $. And let $ A' $ be such that $ A'\in \odot ABC, AA'\parallel BC $. Let $ CP\cap BQ = X $, and let the perpendicular bisector of $ BC $ meet $ PQ $ at $ Y $. Prove that $ A', X, Y $ are collinear. }\label{lemma:concurrency_in_prallel_lines_with_BC} \solu{No angles... Do Lengths...} \fig{1}{concurrency_in_prallel_lines_with_BC}{} \prob{https://artofproblemsolving.com/community/c6h1634961p10278557}{ARO 2018 P11.4}{M}{ $ P \in AB,\ Q \in AC,\ PQ\parallel BC,\ BQ\cap CP = X $. $ A' $ is the reflection of $ A $ wrt $ BC $. $ A'X\cap \odot APQ = Y $. Prove that $ \odot BYC $ is tangent to $ \odot APQ $. } \solu{Of co it can be solved using angle chase, \autoref{lemma:concurrency_in_prallel_lines_with_BC} makes it almost trivial.} \prob{https://artofproblemsolving.com/community/c6h520197}{buratinogigle}{EM}{ Let $(O)$ be a circle and $E,F$ are two points inside $(O)$. $(K),(L)$ are two circles passing though $E,F$ and tangent internally to $(O)$ at $A,D$, respectively. $AE,AF$ cut $(O)$ again at $B,C$, respectively. $BF$ cuts $CE$ at $G$. Prove that reflection of $A$ though $EF$ lies on line $DG$.\\ Rephrasing the problem as such: In the setup of \autoref{lemma:concurrency_in_prallel_lines_with_BC}, let $ A'X\cap \odot ABC = Z $, then $ \odot PQZ $ is tangent to $ \odot ABC $. } \solu{Simple angle chase. } \solu{Another solution to this is by taking $ D $ as a phantom point.} \solu{Another solution is with cross ratios} \newpage \subsection{Simson Line and Stuffs} \lem{Simson Line Parallel}{ Let $P$ be a point on the circumcircle, let $P'$ be the reflection of $P$ on $BC$ and let $PP'\cap\Omega = Q$, and let $l_p$ be the Simson line of $P$. Prove that $l_p\parallel AD\parallel HP'$. } \lem{Simson Line Angle}{ Given triangle $ABC$ and its circumcircle $(O)$. Let $E, F$ be two arbitrary points on $(O)$. Then the angle between the Simson lines of two points $E$ and $F$ is half the measure of the arc $EF$. } \begin{minipage}{.5\linewidth} \figdf{.7}{SimsonLineLemma1}{\autoref{lemma:Simson Line Parallel}} \end{minipage}\hfill% \begin{minipage}{.5\linewidth} \figdf{.8}{SimsonLineLemma2}{\autoref{lemma:Simson Line Angle}} \end{minipage} \newpage\subsection{Euler Line} \theo{}{Perspectivity Line with Orthic triangle is perpendicular to Euler line}{Let $ DEF $ be the orthic triangle. Then $ BC\cap EF, CA\cap FD, AB\cap ED $ are collinear, and the line is perpendicular to the Euler line. In fact this line is the radical axis of the Circumcircle and the NinePoint circle} \lem{}{$ DEF $ is orthic triangle of $ ABC $, $ XYZ $ is the orthic triangle of $ DEF $. Prove that the perspective point of $ ABC $ and $ XYZ $ lies on the Euler line of $ ABC $ } \begin{prooof} Thinking the stuff wrt to the incircle and using cross ratio. \end{prooof} \begin{minipage}{.5\linewidth} \lem{Perpendicular on Euler Line}{ Let $E, F$ be the feet of altitudes from $B, C$, and let $M, N$ be the midpoints of $AC, AB$. $D = MN\cap AB$. Prove that $AF$ is perpendicular to the euler line. } \begin{prooof} We know that $MNEF$ lie on a circle. So $D$ is the radical center of $\odot MNEF, \odot AMN, \odot AEF$. \end{prooof} \end{minipage}\hfill% \begin{minipage}{.45\linewidth} \figdf{.9}{perp_to_euler_line}{} \end{minipage} \begin{minipage}{.54\linewidth} \prob{https://artofproblemsolving.com/community/c6h624551p3740031} {CHKMO 2014 P4}{EM}{ Let $\triangle ABC$ be a scalene triangle, and let $D$ and $E$ be points on sides $AB$ and $AC$ respectively such that the circumcircles of triangles $\triangle ACD$ and $\triangle ABE$ are tangent to $BC$. Let $F$ be the intersection point of $BC$ and $DE$. Prove that $AF$ is perpendicular to the Euler line of $\triangle ABC$. } \begin{solution} Let $B', C'$ be the points on $AC, AB$ such that $BB' = BA$, $CA = CC'$. By simple angle chasing, we can show that $BE, CD$ are tangent to $\odot BB'CC'$.\\ Now by Pascal's theorem on $BBB'CCC'$, we can show that $D, E, F$ are collinear where $F = BC\cap B'C'$. Then by \autoref{lemma:Perpendicular on Euler Line}, we have that $AF\perp HO$. \end{solution} \lem{Extension of CHKMO}{ $AS$, $BB'$, $CC'$ are concurrent. } \end{minipage}\hfill% \begin{minipage}{.45\linewidth} \figdf{.9}{CHKMO_2014_p4}{} \end{minipage} \newpage \subsection{Assorted Diagrams} \figdf{.5}{random_circle_pic_1}{$ H $ lies on the line, circles vary} \datedsubsection{09/2014—06/2018} {% 西南林业大学，昆明} {% \textbf{学士} 计算机科学与技术专业} {% % I am a \highlight{culture hero} and \highlight{trickster figure} who created humanity % from clay, and who defies the gods by stealing fire and giving it to humanity as % civilization. I am known for my \highlight{intelligence} and for being a champion of % humankind. I am also seen as the author of the human arts and sciences generally. This % is why I can use `I' in an CV; you, however, should not~(at least that's what they % say).} 主修课程数据结构、操作系统、组成原理、网络等，这几门主要课程均是90分以上。大三时作为学 校交换生派到泰国学习，英文教学环境。 } %\datedsubsection{Jan. 1234 -- Aug. 5432} % {% % School of Demigods, Greece} % {% % \textbf{M.Sc.}~in Trickery\begin{footnotesize} % ~(summa cum laude) % \end{footnotesize}} % {% % Thesis title: ``Stealing fire from Zeus to give it back to humanity''.} \setcounter{section}{4} \section{Parallel Algorithm to Create Muscle and Fiber Meshes}\label{sec:parallel_algorithm} % The previously presented algorithm to create 3D and 1D meshes is not parallelized. %This restricts the resources that can be employed during the execution of the algorithm to those accessible by one hardware core. Thus, the size of the handled meshes is limited by the available memory of the computer. An algorithm that can be used with distributed memory parallelization could, in contrast, benefit from more total memory that is accessible at different compute nodes. Furthermore, the tracing of the streamlines could be performed in parallel which has the potential to reduce runtimes. In the following, we present an extended algorithm based on the one presented in \cref{sec:ser_alg_meshes} that can be run in parallel on multiple cores. The extended algorithm employs a partitioning of the 3D volume. Every process only stores data corresponding to its own partition. This allows to run the algorithm on a distributed memory system, where data transfer between the processes occurs by sending messages using the Message Passing Interface (MPI). It is possible to create meshes with larger sizes than could fit into a single nodes' memory. This enables us to run the algorithm for meshes with very high resolution that can be used for simulations in the field of High Performance Computing. These meshes are partitioned into subdomains for every compute core and can be read from and written to disk concurrently. \subsection{Overview of the Parallel Algorithm to Create Muscle and Fiber Meshes} The steps of the algorithm and its input and output are given in \cref{alg:parallel_algorithm_1}. Input and output are the same as for the \cref{alg:serial_algorithm_1} presented in \cref{sec:ser_alg_meshes}. The input is a triangulated tubular surface of the muscle that can be obtained as described in \cref{sec:preprocessing_of_the_muscle_geometry}. A second input, the variable called \emph{boundary\_points}, is used only during recursive calls and is not set at the beginning. The output consists of the 3D mesh of the muscle volume $\Omega_M$ and embedded 1D fiber meshes $\Omega_{F,i}$. During execution of the algorithm, the 3D mesh of the muscle is recreated iteratively with increasing resolution and increasing number of subdomains. The algorithm is formulated recursively. At the finest resolution when the recursion terminates, the fiber meshes are finally generated together in all subdomains. At first, a single process executes all the steps of \cref{alg:parallel_algorithm_1} from lines \ref{line:3.2} to \ref{line:3.11a}. This corresponds to recursion level $\ell=0$. Then, in line \ref{line:3.12}, the procedure is called again and in the first recursion executed by eight processes with eight subdomains. On the $\ell$th recursion level, the number of involved processes and subdomains is $8^\ell$. After a specified maximum recursion depth $\ell_\text{max}$ is reached, all involved processes execute the first branch of the \code{if} statement in line \ref{line:3.8} and generate the final 3D and 1D output meshes in line \ref{line:3.9}. The steps in \cref{alg:parallel_algorithm_1} are executed concurrently by the involved processes at the respective levels. Some of the steps only operate on the locally stored data and, thus, are independent of other processes. Other steps involve communication between processes. Whether an instruction effects only the own domain of the process or involves global communication is denoted in parentheses at the beginning of the lines in \cref{alg:parallel_algorithm_1}. \begin{algorithm} \begin{algorithmic}[1]% \Procedure{Create\_3D\_meshes\_parallel}{} \Require Triangulated tubular surface \Require boundary\_points: $4\times 4$ points per slice \Ensure Structured 3D volume mesh \Ensure 1D fiber meshes \Statex \State (own domain) Create\_3D\_mesh(boundary\_points) \label{line:3.2} \State (own domain) Fix and smooth 2D meshes \label{line:3.3} \State (global)$\hskip2.4em$ Solve Laplace problem \label{line:3.4} \State (global)$\hskip2.4em$ Communicate ghost elements to neighboring subdomains \label{line:3.5} \State (own domain) Trace streamlines for new subdomain boundaries \label{line:3.6} \State (global)$\hskip2.4em$ new\_boundary\_points $\leftarrow$ Construct new subdomains \label{line:3.7} \Statex \If{recursion ends} \label{line:3.8} \State (own domain) Trace streamlines for fiber meshes \label{line:3.9} \Else \label{line:3.10} \State (global)$\hskip2.4em$ communicate boundary points \label{line:3.11a} \State (global)$\hskip2.4em$ Create\_3D\_meshes\_parallel(new\_boundary\_points) \label{line:3.11} \EndIf \label{line:3.12} \EndProcedure \end{algorithmic}% \caption{Parallel algorithm to create muscle and fiber meshes}% \label{alg:parallel_algorithm_1}% \end{algorithm}% \subsection{Overview of the Subdomain Refinement} The goal during the recursive calls is to determine smooth boundaries for the new subdomains. Each process splits its own subdomain into eight subdomains and then proceeds to the next recursion level. The subdomain boundaries are determined by tracing streamlines in a divergence-free vector field through the entire muscle volume, similar to the approach in \cref{alg:serial_algorithm_2}. The divergence-free vector field is computed from the solution of a Laplace problem, which is solved in parallel on the entire mesh of the muscle in every recursion. The mesh width of this mesh gets halved in every recursion, subsequently leading to an increasingly fine mesh. The subdomain boundaries are always aligned to streamlines in the mesh that was created last. On each recursion level, the existing subdomain boundaries and the new boundaries for the subdomains on the next recursion level are all created anew and, thus, change slightly as the mesh refines. As the subdomain boundaries in the interior of the volume refine, so do the outer boundaries given by the surface of the muscle. The given triangulation of the surface is sampled again on each recursion level yielding increasingly fine representations. At the final recursion level $l_\text{max}$, the muscle is partitioned into $8^{l_\text{max}}$ subdomains and a respective fine 3D mesh in the muscle volume exists. Then, the algorithm traces the specified amount of streamlines through the whole mesh to produce 1D meshes for the muscle fibers. By construction of the subdomains, the streamlines enter and leave the subdomains through their top and bottom bounding planes. This allows parallel execution of the final streamline tracing step. The reason that the algorithm constructs the partitioning iteratively and not once at the beginning using an initial mesh lies in the requirements for parallel streamline tracing. Each subdomain should be able to trace streamlines in longitudinal ($z$) direction of the muscle without communication to their neighbors in $x$ and $y$ directions. To ensure this property, the partitioning involves a small overlap of neighboring subdomains, i.e., a ghost layer. This ghost layer can consist of a lower number of elements if the mesh is iteratively refined than if the partitioning was created directly on a coarser mesh. % -- \subsection{Data Structure of Boundary Points}\label{sec:data_structure_of_boundary_points} In the following, \cref{alg:parallel_algorithm_1} is illustrated in more detail. The execution starts with one process and the only input is the tubular muscle surface. It is given either as triangulation or in parametric form as NURBS surface. The first step is to construct a quadrilateral mesh of this surface. This is done using the procedure explained in \cref{sec:slicing_of_the_geometry}, which creates horizontal slices of the muscle and places equidistant points on the \say{rings} of the boundaries of these slices. As explained earlier, the points are arranged such that the resulting quadrangulation of the surface has good quality. A result for the biceps muscle is visualized in \cref{fig:serial_alg_0}. Initially, the parallel algorithm stores the points on these rings in the variable \code{boundary\_}\\\code{points}. If the procedure in \cref{alg:parallel_algorithm_1} is called recursively, the contents of this variable is passed as an argument from the previous recursion. The set of points in \code{boundary\_points} defines the boundaries of the subdomain of the process where it is stored. The points on each ring in the $x$-$y$-plane are organized such that they enclose a grid of $n_\text{el,x} \times n_\text{el,x}$ elements, where the number $n_\text{el,x}$ of elements per coordinate direction can be specified as parameter. In the following, an example with $n_\text{el,x}=4$ is considered. The grid is shown in \cref{fig:boundary_grid_1}, the $4\,n_\text{el,x}$ boundary points on the ring are visualized by red color. Note that \cref{fig:boundary_grid_1} depicts the ring as a square whereas in reality it has the potentially more irregular shape of the subdomain. A number $(n_\text{el,z}+1)$ of these rings are stacked in $z$ direction to approximate the enclosing surface of the subdomain, where the number $n_\text{el,z}$ of elements in $z$ direction is again given by a parameter. Thus, the variable \code{boundary\_points} contains a total of $4\,n_\text{el,x}\,(n_\text{el,z}+1)$ points. Typical parameter values are $n_\text{el,x}=4$ and $n_\text{el,z}=50$. \begin{figure}% \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=3cm]{images/parallel_fiber_estimation/boundary_grid.pdf}% \caption{Grid of $4 \times 4$ boundary points, which occurs at the beginning of the procedure of \cref{alg:parallel_algorithm_1}.}% \label{fig:boundary_grid_1}% \end{subfigure} \quad \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=3cm]{images/parallel_fiber_estimation/boundary_grid_2.pdf}% \caption{Grid with $4 \times 8$ boundary points, which occurs after a refinement step at beginning of the procedure of \cref{alg:parallel_algorithm_1}.}% \label{fig:boundary_grid_2}% \end{subfigure} \caption{Logical subdomain boundaries (red) and interior grid (gray) before and after the refinement at the beginning of \cref{alg:parallel_algorithm_1}.}% \label{fig:boundary_grid}% \end{figure}% % refine border points As the goal on every recursion level is to construct a mesh with half the mesh width of the mesh on the previous level, the given boundary points are refined to twice the amount by inserting new points at the centers between neighboring points. The refinement happens in all three coordinate directions. For the example with $n_\text{el,x}=4$, the resulting grid with the $4\times 8$ refined boundary points is shown in \cref{fig:boundary_grid_2}. In $z$ direction, we get $(2\,n_\text{el,z}+1)$ slices with points. % logical structure of new subdomains The task in the recursive procedure is now to determine boundaries for eight subdomains. This is achieved by subdividing the given 2D slices into four 2D subdomains each. Additionally, the 3D volume is split at its vertical center. Thus, the upper and lower parts contain four subdomains each. \Cref{fig:subdomain} visualizes this scheme for the eight subdomains on recursion level $l=1$. The boundary points of the first and the eighth subdomain are shown. The boundary points have already been refined such that every slice in \cref{fig:subdomain} consists of $4 \times 8$ points and corresponds to the grid in \cref{fig:boundary_grid_2} \begin{figure}% \centering% \includegraphics[height=8cm]{images/parallel_fiber_estimation/subdomains_2.png}% \caption{Parallel 3D mesh generation: Partitioning of the muscle volume into eight subdomains during the first call to the procedure in \cref{alg:parallel_algorithm_1}. The first (red) and the eighth subdomain (green) are shown.}% \label{fig:subdomain}% \end{figure}% % ^v^v^ %For this reason, the given $4 \times 4$ boundary points are refined to twice the amount of boundary points by inserting new points at the centers between neighboring points. The resulting grid is shown in \cref{fig:boundary_grid_2}. Now, it would be possible to subdivide the grid to obtain four instances of the needed grid in \cref{fig:boundary_grid_1}. %However, this would result in constant straight connection lines between the initial boundary points. In all further recursive calls, the additional points would all be placed on these lines and thereby not properly refine the subdomain boundaries. Instead, a different approach is desired where the subdomain's boundaries in the volume follow the directions of streamlines and fibers. Thus, the approach is to define the subdomain boundaries in the interior of the global domain by traced streamlines and sample the outer boundaries from the surface triangulation with the desired mesh width. The required steps of this approach are discussed next. % -- \subsection{Generation and Smoothing of the 3D Mesh} % create 3D mesh After the \code{boundary\_points} variables has been set, the next step of \cref{alg:parallel_algorithm_1} is to construct a 3D mesh in the domain. In line \ref{line:3.2} of \cref{alg:parallel_algorithm_1}, the harmonic map algorithm \cref{alg:serial_algorithm_1} described in \cref{sec:ser_alg_meshes} is called. Its input consists of the boundary points that define the 2D slices of the volume. This means that \cref{alg:serial_algorithm_1} does not need to construct the slice boundary rings from the surface triangulation, instead, the formulation of \cref{alg:serial_algorithm_1} can directly start with line \ref{alg:1.2} to triangulate the slices and then compute the harmonic map. For the harmonic map computation, the second triangulation method is used with a circular reference domain quadrangulated by the second scheme. The result is a set of quadrangulated 2D slices that forms a 3D mesh by vertically connecting the elements of neighbor slices. % smoothing Next, line \ref{line:3.3} of \cref{alg:parallel_algorithm_1} improves the mesh quality of the 2D muscle slices $S_M$ from which the 3D mesh is formed. This action consists of two steps. The first step is to ensure that no self-intersecting or degenerate quadrilaterals exist in the slice. The second step applies Laplacian smoothing to improve the mesh quality of the slice. Theoretically, the first step should not be necessary, as the chosen quadrangulation algorithm always produces valid elements. However, in practice, small or irregularly shaped, concave domains occur and together with rounding and numerical errors in the Laplace problem computations occasionally lead to invalid meshes with self intersecting elements, especially for higher recursion depths in \cref{alg:parallel_algorithm_1}. Executing the first step therefore increases the robustness of the implementation. \begin{figure}% \centering% \def\svgwidth{0.7\textwidth}% \input{images/parallel_fiber_estimation/quads.pdf_tex}% \caption{Decomposition of quadrilateral elements into triangles as substep of the validity check of muscle slice quadrangulations. A quadrilateral element (left) and the four triangles (right) that can be constructed from its four points. These triangles are needed for the check in \cref{alg:parallel_algorithm_1} whether the quadrilateral element is valid.}% \label{fig:quads_tris}% \end{figure}% \begin{figure}% \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=3cm]{images/parallel_fiber_estimation/triangle_score_3.pdf}% \caption{Convex quadrilateral with score $s=4$ and the contained triangles, which are all oriented counterclockwise.}% \label{fig:triangle_score_3}% \end{subfigure} \quad \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=3cm]{images/parallel_fiber_estimation/triangle_score_4.pdf}% \caption{Concave quadrilateral with score $s=3$ and the contained triangles. Only the red triangle is oriented clockwise.}% \label{fig:triangle_score_4}% \end{subfigure} \caption{Check for valid elements in the muscle slice quadrangulations that occurs in \cref{alg:parallel_algorithm_1}: Illustration of the score of valid concave and convex quadrilaterals.}% \label{fig:triangle_score}% \end{figure}% The algorithm performs this step by repeatedly iterating over all interior mesh points in every slice $S_M$ and fixing invalid elements. To find invalid elements, for every quadrilateral the four triangles that can be formed from the points of the quadrilateral are considered, as shown in \cref{fig:quads_tris}. For every triangle with points $\bfp^0,\bfp^1$ and $\bfp^2$, the orientation of the triangle is determined. The orientation is counterclockwise if the oriented triangle area $A_{012}$ is positive. The oriented triangle area is the determinant of the $3 \times 3$ matrix that contains the row vectors $(p^i_x,p^i_y,1)$ for the triangle points $\bfp^i=(p^i_x,p^i_y)^\top$ and can be computed by the following formula \cite{sedgewick2011algorithms}: % \begin{align*} A_{012} = (p^1_x-p^0_x)\,(p^2_y-p^0_y) - (p^2_x-p^0_x)\,(p^1_y-p^0_y). \end{align*} If the orientation is counterclockwise, a score value of the triangle is set to one, if it is clockwise, the score is set to zero. The score values of the four triangles are added up to yield a score $s$ for the quadrilateral. Only if this score is $s \geq 3$, the quadrilateral is valid. \Cref{fig:triangle_score} illustrates the cases of valid quadrilateral elements. In a valid, convex element, all four triangles lie inside the element and, thus, the score is $s=4$. If only one triangle is located outside, the quadrilateral is also valid and concave. In this case the score has the value $s=3$. At the current mesh point in the loop over all points that are not at the boundary of the mesh, the four adjacent quadrilaterals are considered. If any of them is invalid, the algorithm tries to improve the situation by deflecting the point by a random, small vector. A maximum of 200 random deflections from the original position with exponentially increasing deflection vector sizes are tried. After each modification of the point, the scores of the four adjacent quadrilateral elements are evaluated. If the sum of the four element scores increases, the point is kept and the iteration over all interior mesh points starts anew. Note that this does not necessarily mean that the invalid element was fixed, only its score was improved. If it was not fixed, it will be considered again in the next iteration. For example, a convex element that initially is oriented clockwise instead of counterclockwise has a score of $s=0$. In the first iteration, one point is deflected such that the quadrilateral intersects itself but has a higher score $s\geq 0$. At least one more iteration is needed until the quadrilateral is oriented correctly. When all elements in the slice $S_M$ are valid, this step is complete. \subsection{Laplacian Smoothing} The second step is the smoothing step that improves the mesh quality of the 2D slices. \num{20} iterations of Laplacian smoothing \cite{field1988laplacianSmoothingAndDelaunayTriangulations} are executed. Laplacian smoothing in our case subsequently visits all interior points of the mesh and sets the location of a point to the center of gravity of its four direct neighbors. %The order in which the points of the mesh are traversed is changed after every iteration: In the first iteration, the points are traversed starting at the bottom left. In the second iteration, the traversal starts at the top right and moves in opposite direction compared to the first iteration. The third and fourth iterations begin at the bottom right and top left nodes of the mesh. After these four iterations the scheme is repeated. The reason for this change of the traversal is to \Cref{fig:laplace_smoothing} shows the effect of Laplacian smoothing for a slice in a subdomain on the first recursion level. It can be seen how the smoothing equalizes the element side lengths and angles. However, this smoothing step can invalidate a mesh by introducing overlapping quadrilaterals. An example for this case is given in \cref{fig:laplace_smoothing_0}. The initial mesh in \cref{fig:world_mesh_0} is concave and occurs during recursion level $l=2$. \Cref{fig:world_mesh_improved_0} shows the result of the smoothing, which contains one invalid element. The smoothing operation placed the fourth point of the element that also contains the three boundary points at the concavity outside the mesh. As a remedy, the smoothing method checks the validity of the adjacent elements before a point is moved. If the move results in an invalid element, the action is not carried out and the traversal continues with the next point instead. \Cref{fig:world_mesh_improved_1} shows the resulting mesh if this check is enabled. The mesh has slightly different boundaries because the check influenced the behavior already on lower recursion levels. % smoothing fix \begin{figure}% \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/world_mesh.png} \caption{Initial 2D mesh of a subdomain at the boundary of the biceps muscle.}% \label{fig:world_mesh}% \end{subfigure} \quad \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/world_mesh_improved.png} \caption{The mesh of (\subref{fig:world_mesh}) after 20 iterations of Laplacian smoothing.}% \label{fig:world_mesh_improved}% \end{subfigure} \caption{Quality improvement of 2D muscle slice quadrangulation: Effect of Laplacian smoothing of a 2D grid which occurs in line \ref{line:3.3} of \cref{alg:parallel_algorithm_1}.}% \label{fig:laplace_smoothing}% \end{figure}% \begin{figure}% \centering% \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/loop_025_p3728344_world_mesh.png} \caption{Initial 2D mesh.}% \label{fig:world_mesh_0}% \end{subfigure} \quad \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/loop_025_p3728344_world_mesh_improved_3.png} \caption{The mesh of (\subref{fig:world_mesh_0}) after 20 iterations of Laplacian smoothing, yielding an invalid quadrangulation.}% \label{fig:world_mesh_improved_0}% \end{subfigure} \quad \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/loop_025_p30402_world_mesh_improved.png} \caption{The mesh of (\subref{fig:world_mesh_0}) after 20 iterations of Laplacian smoothing with the validity check, yielding a valid quadrangulation.}% \label{fig:world_mesh_improved_1}% \end{subfigure} \caption{Quality improvement of 2D muscle slice quadrangulation: Effects of Laplacian smoothing on concave domains.}% \label{fig:laplace_smoothing_0}% \end{figure}% \subsection{Solution of the Laplace Problem}\label{sec:solution_of_the_laplace_problem} After the 3D mesh has been created and smoothened, the next steps are to solve the Laplace problem in the muscle domain, to trace streamlines in the gradient of the solution vector field and finally to construct the eight subdomains for the recursive calls by subdividing the own domain along the streamlines. % refine 3D mesh Prior to the solution of the Laplace problem, the 3D mesh gets refined further by increasing the number of elements per coordinate direction by a specified factor $r\in \N$. The rationale is to increase the number of degrees of freedom and, thus, the resolution to get a smaller numerical error in the subsequent Laplace computation. This refinement is in addition to the refinement of the initial boundary points by a factor of two described in \cref{sec:data_structure_of_boundary_points}. The mesh with $2\,n_\text{el,x} \times 2\,n_\text{el,x} \times 2\,n_\text{el,z}$ elements gets refined to $2\,r\,n_\text{el,x} \times 2\,r\,n_\text{el,x} \times 2\,r\,n_\text{el,z}$ elements. The new points are found by interpolating in the existing mesh. For example, the 3D mesh of \cref{fig:boundary_grid} with $2\cdot 4 \times 2\cdot 4 \times 2\cdot 50$ elements gets refined with the factor $r=2$ to $16 \times 16 \times 200$ elements. \Cref{fig:02_boundary_points} shows the refined boundary points in this example in a view in negative $z$ direction towards the bottom of the muscle. The red points are the boundary points of the $4 \times 8$ grid, the additional white points are added in between by the refinement with $r=2$. Because this refinement is carried out by interpolating between the initial points, the new points are located on straight lines between the initial points. This can especially be seen at the lower left of the figure (indicated by arrows and lines) where always five neighboring points lie on a straight line. \begin{figure}% \centering% \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[height=9cm]{images/parallel_fiber_estimation/02_boundary_points_2.png} \caption{Initial (arrows) and refined boundary points (red) and points after additional refinement by a factor of $r=2$ (white).}% \label{fig:02_boundary_points}% \end{subfigure} \quad \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[height=9cm]{images/parallel_fiber_estimation/03_seed_points.png} \caption{Seed points for the streamlines (blue).}% \label{fig:03_seed_points}% \end{subfigure} \\ \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[height=8cm]{images/parallel_fiber_estimation/05_corner_streamlines_corner.png} \caption{The four boundary streamlines (red) and the layer of ghost elements (green) at the bottom and right of the subdomain.}% \label{fig:05_corner_streamlines}% \end{subfigure} \quad \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[height=8cm]{images/parallel_fiber_estimation/07_filled_corner.png} \caption{New boundary points on the outer (light green) and interior boundary (dark green).}% \label{fig:07_filled}% \end{subfigure} \caption{Parallel generation of 3D meshes: refined boundaries, streamlines and subdomain refinement in the first subdomain for recursion level $l=1$.}% \label{fig:03_boundary_points_and_seed_points}% \end{figure}% % laplace problem Next, in line \ref{line:3.4} of \cref{alg:parallel_algorithm_1} the Laplace problem gets solved. The same step also occurs in \cref{alg:serial_algorithm_2} and is explained in \cref{sec:generation_of_fiber_meshes}. The equation is formulated globally and the discretization uses the existing partitioning. Dirichlet boundary conditions of $p(\bfx) = 0$ and $p(\bfx) = 1$ are prescribed at the bottom and top of the domain, as shown by the spheres in \cref{fig:dirichlet_bc_1}. Alternatively, Neumann boundary conditions can be used. A parallel GMRES solver is employed to obtain the solution in a few iterations. E.g., for the biceps muscle a linear system at $l=0$ has 4131 degrees of freedom and 26 iterations are needed to obtain a residual norm below \num{1e-4}. After the solution $p(\bfx)$ is obtained, the gradient field $\nabla p(\bfx)$ is computed. The solution and the gradient directions are visualized in \cref{fig:laplace_1}. \begin{figure} \centering \begin{subfigure}[t]{0.23\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/dirichlet_bc_1.png} \caption{Location of Dirichlet boundary condition nodes at the bottom and top.}% \label{fig:dirichlet_bc_1}% \end{subfigure} \, \begin{subfigure}[t]{0.24\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/laplace_1.png} \caption{Solution $p$ of the Laplace problem and direction of the gradient $\nabla p$.}% \label{fig:laplace_1}% \end{subfigure} \qquad \begin{subfigure}[t]{0.19\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/boundary_points_1.png} \caption{Traced streamlines that split the domain into eight subdomains.}% \label{fig:boundary_points_1}% \end{subfigure} \, \begin{subfigure}[t]{0.24\textwidth}% \centering% \includegraphics[height=7.2cm]{images/parallel_fiber_estimation/slices_2.png} \caption{Rings of the slices $S_M$ and traced streamlines in the interior.}% \label{fig:slices_2}% \end{subfigure} \caption{Parallel 3D mesh generation: Process of subdividing the muscle volume into eight subdomains using the solution of a Laplace problem, which is an important step in the procedure of \cref{alg:parallel_algorithm_1}.} \label{fig:determining_subdomains}% \end{figure} \subsection{Communication of the Ghost Layer} % ghost communication Subsequently, the gradient field $\nabla p(\bfx)$ is used to trace streamlines to determine new boundaries of the subdomain. This involves tracing streamlines that start exactly on the boundary. These streamlines potentially switch between the subdomain owned by the current process and the subdomains of neighboring processes. Streamline tracing requires the gradient field values of the elements where the streamline passes through. To avoid repeated communications in these cases, a ghost layer of a specified number $n_\text{ghost\_layer\_width}$ of elements is added to the subdomains at all parts of the boundary that touch a neighboring subdomain directly or diagonally adjacent in $x$ and $y$ direction. The ghost layer is constructed and the node positions and values of $p$ and $\nabla p$ associated with the ghost elements are communicated between the neighboring processes after the solution of the Laplace problem. This occurs in line \ref{line:3.5} of the algorithm. \Cref{fig:05_corner_streamlines} shows $n_\text{ghost\_layer\_width}=1$ layer of ghost elements on a subdomain at recursion level $l=1$. \subsection{Selection of Seed Points for the Streamlines}\label{sec:selection_of_seed_points} % seed points-1 Next, the seed points from which the streamlines start are determined on the subdomain. All seed points are selected from the set of nodes in the structured mesh of a horizontal 2D slice. \Cref{fig:seed_points_to_send_1} visualizes the structured mesh in light gray in the first call to the procedure for recursion level $l=0$ where the domain is not yet partitioned. The selected seed points are shown by the yellow and red points. As can be seen, the seed points consist of the nodes of the 2D mesh at the horizontal and vertical centers in this view which form the \emph{plus sign} shape given by the yellow points. In addition, the four red points near the corners of the structured mesh are selected. %\Cref{fig:seed_points_to_send_1} also visualizes the boundary of the slice together with a method of splitting it into eight sectors by choosing the splitting points such that they are the closest to the given outer seed points. % seed points interior for l=0 The seed points of the \emph{plus sign} yield the streamlines that subdivide the domain into four parts in $x$ and $y$ direction. With the additional split in $z$ direction, the inner boundaries of the eight subdomain are obtained. The resulting boundaries are given in \cref{fig:fixed_1}. The streamlines of these seed points are also depicted in \cref{fig:boundary_points_1}. The interior boundary points for the eight subdomains that partition the muscle volume at level $l=1$ are shown by different colors. The full subdomain boundaries include also the outer surface of the muscle, which is given by the rings of the muscle slices. \Cref{fig:slices_2} shows the streamlines in black and the circumferential rings of the muscle slices in blue that were extracted during the call to \cref{alg:serial_algorithm_1} in line \ref{line:3.2}. \begin{figure} \centering \begin{subfigure}[t]{0.30\textwidth}% \centering% %\includegraphics[height=4cm]{images/parallel_fiber_estimation/seed_points_to_send_1a.png} \includegraphics[height=5cm]{images/parallel_fiber_estimation/fixed_0.pdf} \caption{The seed points of the streamlines used to determine the subdomain boundaries. }% \label{fig:seed_points_to_send_1}% \end{subfigure} \, \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[height=5cm]{images/parallel_fiber_estimation/fixed_1.png} \caption{Streamlines and lines on the muscle surface that define the new subdomain boundaries.}% \label{fig:fixed_1}% \end{subfigure} \, \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[height=5cm]{images/parallel_fiber_estimation/final_interior_1.png} \caption{All streamlines and lines on the muscle surface that are created if the algorithm is run with one process and $l_\text{max}=0$.}% \label{fig:final_interior_1}% \end{subfigure} \caption{Parallel mesh generation: Seed points and streamlines that occur during the first call to the procedure in \cref{alg:serial_algorithm_2}, in a view from the top of the muscle.} \label{fig:seed_points}% \end{figure} % seed points interior for l>0 At higher recursion levels $l>0$, the boundaries for the new subdomains consist of those at the outer boundary of the muscle defined by the surface representation and those in the interior of the muscle. Similar to the previously considered case at recursion level $l=0$, for $l\geq 1$ the boundaries in the interior of the muscle have to be sampled by a set of streamlines. In addition to the streamlines associated with the plus sign shaped seed points, new streamlines at the boundaries of the current subdomain have to be obtained. \Cref{fig:03_seed_points} shows in blue all seed points that are selected in a subdomain on recursion level $l=1$ in order to create subdomain boundaries for level $l=2$. As can be seen, in addition to the plus sign shape and the four outer seed points two lines of points in approximate $x$ and $y$ directions are selected at the lower and right edges of the image. These are seed points for the new boundaries in the interior of the muscle. Note that the current recursion level $l=1$ also has boundaries at these locations. However, for level $l=2$ these boundaries are recreated by the new streamlines. Tracing of these streamlines potentially uses the ghost layer. The resulting streamlines and the ghost layer for $n_\text{ghost\_layer\_width}=1$ are shown in \cref{fig:05_corner_streamlines,fig:07_filled}. \subsection{Determination of Subdomain Boundaries on the Outer Muscle Surface} % seed points-3 outer ring Next, the boundary points on the outer surface of the muscle are determined for the new partitioning. They are obtained by sampling the circumferential rings of the muscle surface with the resolution required in the current recursion level. In our implementation, this can be done either by sampling the original surface triangulation of the muscle or by directly evaluating the parametric form of the NURBS surface that approximates the muscle surface. At recursion level $l=0$, the entire muscle surface is touched by the new subdomains. Thus, when traversing the circumference of the muscle four new subdomains are encountered. In consequence, every circumferential ring needs to be split into four quarter parts for the four adjacent subdomains. For each of these new subdomains, the quarter part corresponds to two neighboring sides of the subdomain boundary in \cref{fig:boundary_grid_1}. \Cref{fig:seed_points_to_send_1} also visualizes the two neighboring sides per new subdomain as dark and light portions of the outer boundary. To obtain these sides, a splitting point is needed that further splits every quarter part of the circumferential ring into the two sides for the new subdomain. In summary, the ring needs to be split into eight parts that fit to the inner subdomain boundaries. The eight split points are determined by the eight outer streamline points. In \Cref{fig:seed_points_to_send_1}, the four outer yellow points of the plus sign and the four red points are considered. For each split location, the nearest point on the circumferential ring is determined. The employed algorithm for calculating the coordinates of the point on a ring that has the shortest distance to a given, second point is described in \cref{sec:slicing_of_the_geometry}. After the two sections of the circumferential rings have been determined for all new subdomains, the sections are equidistantly sampled in circumferential direction with $n_\text{el,x}$ elements each to create the outer boundary points for the subdomains. Also in longitudinal direction of the muscle, i.e., the $z$ axis, points are sampled on each streamline and on the outer boundary surface to yield the required number of $n_\text{el,z}$ points per subdomain. The resulting boundary points obtained during recursion level $l=0$ are shown in \cref{fig:fixed_1}. This method is also similarly required on higher recursion levels $l>0$. Then, however, two cases have to be considered separately. The first case involves splitting the muscle surface boundary on one process into two parts, analogously to the described method at $l=0$. The second case involves two neighboring processes that have to agree on the split point of their common part of the outer surface boundary. In the example at recursion level $l=1$ in \cref{fig:05_corner_streamlines}, the red streamlines are used to split the boundary sides at the outer boundary of the global domain. The first case occurs for the upper left red streamline, which is used to bisect the shown white part of the muscle surface. The second case occurs at the lower left and upper right borders between the shown subdomain and the neighboring subdomains of two other processes. For the case at the upper right, \cref{fig:seed_points_case_2} visualizes the following method: First, the point on the outer surface that is closest to the point of the red streamline is determined on both subdomains, visualized by the yellow stars. These points are communicated between the two processes. Each process computes the center point of these two points (orange star) and then finds the closest point to this center point on the boundary. This is done for all rings of the muscle in $z$ direction. In result, both processes have the same line on the surface in longitudinal direction of the muscle that is then used as one edge of the new subdomains. \begin{figure} \centering \def\svgwidth{0.7\textwidth} \input{images/parallel_fiber_estimation/fixed_1.pdf_tex} \caption{Parallel mesh generation: Case of partitioning the outer boundary surface that occurs for recursion level $l \geq 1$ in \cref{alg:serial_algorithm_2}. Shown are the meshes on two subdomains of processes $p_0$ and $p_1$ and the location of the streamline at the corner (red points). The orange star is the newly determined border point between the subdomain boundaries.} \label{fig:seed_points_case_2}% \end{figure} Thus, the new subdomain boundaries on the outer boundary can be found using the red extra streamlines shown in \cref{fig:05_corner_streamlines}. For simplicity, the algorithm always computes the four streamlines in every corner of the mesh although all of them are only required for $l=0$. In the shown example for $l=1$, the streamline in the lower right corner is not needed for the sampling of the new boundary points. A summary of the streamlines that are used for the new subdomain boundaries in this example is given in \cref{fig:07_filled}. The sampled boundary points at the muscle surface are shown in light green color. These two sides of the own domain will be split into four sides for the new subdomains. In this example with $n_\text{el,x}=4$, the surface therefore gets sampled at $4\times 4=16$ lines. A comparison with the white lines in \cref{fig:05_corner_streamlines} shows that the newly sampled points are different from the initially sampled points. While in \cref{fig:05_corner_streamlines} always five neighboring boundary points are located on a straight line, the points in \cref{fig:07_filled} follow the curved outer boundary better and, thus, refine the boundary representation. \subsection{Parallel Algorithm for Streamline Tracing}\label{sec:parallel_streamline_tracing} % streamline tracing method In line \ref{line:3.6} of \cref{alg:parallel_algorithm_1}, streamlines have to be traced through the gradient field $\nabla p(\bfx)$ of the Laplace solution for all the seed points given in \cref{sec:selection_of_seed_points}. In the following, more details on the parallel method of streamline tracing is given. This step is similar to the analog step in \cref{alg:serial_algorithm_2}. The same method of explicit Euler integration is used. The seed points are located at the horizontal plane at the center in vertical direction of the muscle. From there, streamlines are traced in both directions towards the ends of the muscle following the positive and negative gradient directions. The tracing algorithm uses the efficient scheme of selecting the subsequently traversed elements described in \cref{sec:algorithm_for_streamline_tracing}. The implementation is adjusted in a way to also take into account the layers of ghost elements. % streamline tracing in parallel-1 Since the streamlines traverse the entire muscle from the center to the bottom and top, multiple processes are involved in the computation of every streamline. To describe the scheme, all processes are numbered in $z$ direction from bottom to top by an index $i_z \in \{0, 1, \dots, n_z-1\}$ where $n_z = 2^l$ is the number of processes in $z$ direction on the current recursion level $l$. % streamline tracing in parallel-2 The initial seed points are determined on the processes at the vertical center with index $\lfloor n_z/2\rfloor$. They are communicated to the processes below with index $\lfloor n_z/2\rfloor-1$. These two groups of processes begin with tracing the streamlines through their subdomains starting from the same seed points, the upper processes in upward and the lower processes in downward direction. Then, the end points of the traced streamlines are communicated to the next processes, which continue the tracing. The procedure repeats with further processes until the streamlines reach the bottom and top ends of the overall muscle domain. The time complexity of this approach is $\O(n_z) = \O(\sqrt[3]{n_\text{proc}})$ with the number $n_\text{proc}$ of processes. % streamline sampling After the streamlines have been traced, they are sampled at equidistant positions with a distance according to the required distance between the boundary points of the subdomains. \subsection{Recursion End: Generation of the Resulting Meshes} % summary/überleitung In result, one pass of \cref{alg:parallel_algorithm_1} from lines \ref{line:3.2} to \ref{line:3.7} creates boundaries for eight new subdomains. Line \ref{line:3.8} checks whether the maximum recursion $l_\text{max}$ is reached and the recursion ends. If the recursion ends, the final 3D mesh and 1D fiber meshes are constructed in line \ref{line:3.9}. In this case, the prepared boundary points are not needed for a further subdivision of the domain but are used to construct the final meshes instead. Every resulting fiber mesh is generated by one streamline. In line \ref{line:3.9}, additional streamlines are traced starting at the remaining grid points of the 2D slice at the vertical center of the muscle that were not selected as seed points earlier. The parallel method described in \cref{sec:parallel_streamline_tracing} is used. As an example, \cref{fig:seed_points_level2} shows all seed points of streamlines for $l=2$ at the beginning of line \ref{line:3.9} in a run of \cref{alg:parallel_algorithm_1} with $n_\text{el,x}=4$, $l_\text{max}=2$ and 64 processes. The shown points are located at the top of the lowest 16 subdomains in the muscle, i.e., the subdomains of processes 0 to 15. The corresponding streamlines get traced in line \ref{line:3.6} to be used for the new subdomain boundary. However, since the recursion ends at $l=2$ the missing seed points of the subdomain grids are subsequently filled in and the remaining streamlines are traced. From the full grid of $31\times 31=\num{961}$ streamlines, \num{449} or $~\SI{47}{\percent}$ have already been traced at this point. \begin{figure} \centering \includegraphics[width=0.5\textwidth]{images/parallel_fiber_estimation/seed_points_level2.png} \caption{Parallel mesh generation: Seed points of the streamlines that are traced on processes 0 to 15 in the procedure of \cref{alg:parallel_algorithm_1} at recursion level $l=2$.} \label{fig:seed_points_level2}% \end{figure} %If more fiber meshes are desired, the 2D elements of this slice can be sampled equidistantly in the parameter space used for the harmonic map computation to obtain more seed points from which additional streamlines are traced. In summary, the 2D quadrilateral mesh at the center slice of the muscle defines the location of the resulting muscle fibers. Because the construction of this 2D mesh ensured a good mesh quality with similar element sizes, the distance between the resulting fibers is similar and a spatially homogeneous set of muscle fibers is generated. To obtain the final 1D fiber meshes $\Omega_{F,i}$, the streamlines are sampled at equidistant $z$ intervals, specified by a parameter $\Delta z$. Because the streamlines are directed mainly along the $z$ axis, the constant $z$ interval for the sampling approximately corresponds to the resulting 1D mesh width, i.e., the distance between the points of a fiber. An advantage of this method is that the points of all fibers lie in the same $x$-$y$ planes. Thus, the total set of points can also be interpreted as a structured 3D mesh of the muscle volume $\Omega_M$. This 3D mesh is aligned with the fiber meshes and planes through the $x$ and $y$ axes. These properties are advantageous for data mapping between the 3D mesh and the 1D fiber meshes and for the numerical solution of models with anisotropic advection processes in the 3D mesh that is oriented according to the direction of the fibers. % file output At the end, the data is written collectively by all processes into a single file. This is done using the parallel file I/O functionality of MPI. This can be done because the absolute position in the file of every point can be calculated from the index of the point in the structured mesh. % Examples \Cref{fig:final_interior_1} gives an example of the resulting streamlines if the recursion ends already after one pass of the procedure at $l_\text{max}=0$. For the example with $l_\text{max}=1$, the selected seed points and the parts of the resulting streamlines in the considered subdomain are shown in \cref{fig:08_final}. Here, the dark blue streamlines on the boundary were traced as part of the refinement actions in line \ref{line:3.6}. Because the recursion ends for $l=1$, these streamlines are now reused for the final fiber meshes instead of further parallel partitioning. Additionally, the light blue streamlines in the interior were traced to obtain a full grid of fibers for the output of the algorithm. \subsection{Continuation on the Next Recursion Level} If line \ref{line:3.8} of \cref{alg:parallel_algorithm_1} does not detect the recursion end because the maximum recursion levels is not yet reached, the \code{else} branch in line \ref{line:3.10} is chosen. Execution continues with the eight times higher number of processes $8^{\ell+1}$. The processes that executed the previous parts of the algorithm send their determined boundaries of the new subdomains to seven other respective processes in line \cref{line:3.11a}. Only the first subdomain remains on the same process. Every process stores the boundary points for its new subdomain in the variable \code{boundary\_points}. In line \cref{line:3.11}, the procedure is called recursively and the next recursion level $(l+1)$ begins. \Cref{fig:06_subdomain} shows the \code{boundary\_points} of the first new subdomain on level $l=2$ for the example on recursion level $l=1$. It consists of the outer boundary (dark yellow lines) and the interior boundary (brown streamlines) and is nearly geometrically similar to the subdomain on level $l=1$. \begin{figure}% \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[height=10cm]{images/parallel_fiber_estimation/08_final.png} \caption{Seed points (yellow), traced interior streamlines (light blue) and boundary points (dark blue), generated if $l_\text{max}=1$.}% \label{fig:08_final}% \end{subfigure} \quad \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[height=10cm]{images/parallel_fiber_estimation/06_subdomain.png} \caption{Boundary points of the first subdomain on level $l=2$ (dark yellow and brown) embedded in the boundary (white) of level $l=1$, generated if $l_\text{max} > 1$.}% \label{fig:06_subdomain}% \end{subfigure} \caption{Generation of 3D and 1D meshes in subdomains: Resulting streamlines after the pass of \cref{alg:parallel_algorithm_1} for recursion level $l=1$.}% \label{fig:improved}% \end{figure}% \subsection{Repair of Incomplete Streamlines}\label{sec:repair_of_incomplete_streamlines} Practical tests have shown that, for irregular muscle geometries, occasionally some streamlines generated in lines \ref{line:3.6} and \ref{line:3.9} of \cref{alg:parallel_algorithm_1} can be incomplete. This means that it was not possible to obtain a streamline that runs through the entire subdomain or the entire muscle domain from top to bottom, instead points are missing for some ranges of $z$ values. This can happen if the streamlines leave the subdomains (because the ghost layer width was chosen too small) or due to numerical errors in irregularly shaped elements mainly on high recursion levels where the system matrix is badly conditioned. To obtain meaningful results even in these cases, three different repair mechanisms are introduced that interpolate the missing data from valid streamlines. \Cref{fig:fix_invalid} visualizes the cases by examples in a setting of four subdomains with grids of $5 \times 5$ fibers each. The repair mechanisms $\#1$ to $\#3a$ only apply to boundary points. They are executed in line \ref{line:3.6} of the algorithm after the local portions of the streamlines have been traced and before the end points of the streamlines are sent to the neighbor processes below and above that continue the streamline tracing. Mechanismn $\#3b$ and $\#3c$ repair invalid streamlines in the final result and are executed during line \ref{line:3.9} of \cref{alg:parallel_algorithm_1}. Mechanism \#1 checks all streamlines at subdomain boundaries in the interior, which are shared between neighboring processes. If a streamline is incomplete on one process but complete on the neighbor process, the data of the complete side are transferred such that both processes have the same valid points for this streamline. In the example in \cref{fig:fix_invalid}, the valid streamline data are sent from the top left to the top right subdomain. Mechanism \#2 checks streamlines at the outer corners of the subdomains. Incomplete streamlines at these locations are recreated from the given boundary points. Because the set of boundary points is twice as coarse as the required number of sample points at these streamlines, every second point gets interpolated from the top and bottom neighbor points. Mechanism \#3a is concerned with streamlines at interior subdomain boundaries that could not be fixed by mechanism \#1 because the streamlines are incomplete on both sharing processes. In this case, the streamlines are interpolated from the two complete neighboring streamlines that are located next along the boundary as shown in the example in \cref{fig:fix_invalid}. Instead of the factors $\frac13$ and $\frac23$, the actual relation of distances between the seed points of the streamlines is used. The same interpolation is executed independently on both involved processes. Because the valid streamlines have the same data on both subdomains, the resulting fixed streamlines will also be identical. Mechanisms \#3b and \#3c follow the same approach. They are applied to the interior fibers of the final result and can repair any number of incomplete fibers that are located between complete fibers. This case rarely occurs, a cause can be errors in the numerical solution of the Laplace problem. In example \#3b in \cref{fig:fix_invalid}, the two invalid streamlines are interpolated from their left and right valid neighbors. In example \#3c, no valid right neighbor exists. Instead, the streamlines are interpolated by using valid positions from the upper and lower neighbors. \begin{figure}% \centering% \def\svgwidth{0.8\textwidth} \input{images/parallel_fiber_estimation/fix_invalid.pdf_tex} \caption{Repair of streamlines used during partitioning and fiber approximation in the 3D and 1D mesh generation: Examples of the four repair mechanisms for estimating incomplete streamlines during the parallel algorithm. Invalid streamlines are indicated by red circles, valid streamlines by black circles. The brown arrows show the direction of data transfer.}% \label{fig:fix_invalid}% \end{figure}% \subsection{Post-processing and Output of the Generated Streamlines}\label{sec:postprocessing_of_the_generated_streamlines} After repairing invalid streamlines, the final result of the algorithm is a grid with $(2\,n_\text{el,x}\,n_x+1) \times (2\,n_\text{el,x}\,n_x+1)$ fibers in the $x$-$y$ plane and a configurable number of points in $z$ direction, where $n_x = 2^{\ell_\text{max}}$ is the number of subdomains per coordinate direction on the last recursion level. If a higher number of fibers is desired than is naturally generated by the parallel algorithm, additional fibers can be created by interpolation in the existing grid of fibers, which is parallel partitioned. The implementation of the presented algorithm in \opendihu{} includes this post-processing functionality as part of the mesh generation program. Alternatively, the step can be applied separately on any binary output file that contains a grid of fibers. The action of increasing the number of fibers proceeds as follows. The initial grid contains the fibers that were created from the streamlines, called \emph{key fibers}. A specified number $m$ of additional fibers is placed between the key fibers in both $x$ and $y$ coordinate directions. The additional fibers together with the key fibers form a grid of fibers in the muscle cross-sections with an $m$ times finer mesh width. In the grid of key fibers, every portion bounded by $2 \times 2$ key fibers contains $(2+m)^2 - 4$ additional fine fibers. The total number of fibers depending on $n_x$ and $m$, therefore, is $N=(2\,n_\text{el,x}\,n_x\,(1+m)+1)^2$. Due to construction, this number is always odd. This is a desired property because it yields an even number of elements per coordinate direction and this allows to construct a mesh with quadratic ansatz functions. The new fibers are computed by barycentric interpolation. The location of every new point $\bfp$ is calculated from the nearest points $\bfp_0, \bfp_1, \bfp_2$ and $\bfp_3$ of key fibers in the $x$-$y$ plane, numbered according to \cref{fig:quads_tris}, by% \begin{align}\label{eq:mesh_barycentric_interpolation} \bfp = (1-\alpha_x)\,(1-\alpha_y)\,\bfp_0 + \alpha_x\,(1-\alpha_y)\,\bfp_1 + (1-\alpha_x)\,\alpha_y\,\bfp_2 + \alpha_x\,\alpha_y\,\bfp_3. \end{align} Here, the factors $\alpha_x,\alpha_y \in [0,1]$ are chosen in a way to create the fine grid of fibers: % \begin{align*} \alpha_x = i / (m+1), \quad \alpha_y = j / (m+1)\quad \text{ for }i,j = 0, \dots,m, \quad (i,j) \neq (0,0). \end{align*} As a result, we can generate a 3D mesh where the number of points in $x$ and $y$ directions can be adjusted by the parameter $m$. An advantage of this algorithm is that each process only has to keep the data of its own subdomain in memory at any time. This allows parallel processing of very large meshes. For small-enough meshes that do not fall under this restriction, the utility script \code{resample_bin_fibers.py} can be used to create meshes of any resolution from any other mesh using the barycentric interpolation in \cref{eq:mesh_barycentric_interpolation}. An example is given in \cref{fig:left_biceps_brachii_33x33fibers_refined}, where a mesh of $33\times 33$ fibers is refined by interpolation to a mesh with $71\times 71$ fibers. % seed points 33x33, 71x71 from scaling script \begin{figure}[H] \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/left_biceps_brachii_33x33fibers_bin_csv.pdf}% \caption{Mesh points in a $33\times 33$ grid at the center cross-section of the biceps muscle.}% \label{fig:left_biceps_brachii_33x33fibers_bin_csv}% \end{subfigure} \quad \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/left_biceps_brachii_71x71fibers_bin_csv.png} % also png \caption{Refined mesh points in a $71\times 71$ grid that were obtained from (a) by barycentric interpolation.}% \label{fig:left_biceps_brachii_71x71fibers_bin_csv}% \end{subfigure} \caption{Refinement of existing meshes to obtain derived meshes with any number of nodes.}% \label{fig:left_biceps_brachii_33x33fibers_refined}% \end{figure}% The resulting points are stored in a binary file format. The contents of this output file can either be interpreted as grid points of a 3D mesh or as points of individual 1D fibers. This is an advantage in a multi-scale simulation where both a 3D muscle mesh and multiple embedded 1D fiber meshes occur: First, all mesh information of both $\Omega_M$ and $\Omega_{F,i}$ can be given by a single file. And second, the 3D mesh is aligned with the 1D fibers and all 3D mesh points are also 1D mesh points. The spacing in $z$ direction between points on a fiber is typically chosen as $\Delta z = \SI{0.01}{\cm}$. This value was found to ensure a low error in the model for propagation of electric stimuli along the muscle. The value leads to 1481 points per fiber on the belly of the biceps muscle. Every point coordinate is stored in the output file as double precision value with eight bytes. The file contains a header of 72 bytes with descriptive information such as the number of fibers, some parameter values and a time stamp. The total file size therefore can be calculated by $72+N\cdot 1481\cdot 3\cdot 8$ bytes. Often, the spatial resolution of the 3D mesh does not need to be as high as those of the fibers. The relation of the 3D and 1D mesh widths as well as the number of 1D meshes should be chosen such that the numerical error of the simulation in both domains is balanced. In case the 3D mesh should be coarser than the output of the algorithm, we can use only a subset of the points contained in the output file. Then, a stride in $x$, $y$, and $z$ direction is specified in the settings for the simulation. The corresponding coarse grid of points is extracted and used to construct the 3D mesh that is then used for the simulation. %The 1D fiber meshes use all given points in the file. % When a 3D mesh with a smaller spatial resolution in $x$ and $y$ direction than the number of fibers is used in a simulation together with 1D fiber meshes from the same file, some of the fibers at the outer layer can be located outside of the 3D mesh. A way to avoid this is to not use the outer layer of fibers. A remaining issue concerns the mesh quality on the outer boundary. In general, the 3D mesh created by \cref{alg:parallel_algorithm_1} has good quality because the interior points result from smooth streamlines that were traced through a divergence free vector field. The points at the boundary, however, are either sampled from a triangulation of a tubular surface of the muscle or computed from the NURBS formulation. This surface is derived from imaging data, as described in \cref{sec:preprocessing_of_the_muscle_geometry}. If the triangulation is used, the quality of the boundary points of the created mesh depends on the quality of the muscle surface and its triangulation. In a case where this quality is poor, only the outer layer of elements of the created 3D mesh is affected. \Cref{fig:poor_boundary_33x33} shows an example for this effect in a grid of $9 \times 9$ fibers. It can be seen that only the fibers at the bottom of the image have an irregularity at their center. Such an irregularity potentially occurs at every $z$ coordinates where a new subdomain begins. The cause is that, at these locations, the points on the rings are slightly shifted relative to each other. A remedy in such a case is to discard the outer layer of fibers and construct the mesh only from points of the inner streamlines. Accordingly, our implementation of the presented algorithm \cref{alg:parallel_algorithm_1} always creates two different output files. The first output file contains all fibers, the second contains all except the outer layer of fibers. The second file contains only $N=(2\,n_\text{el,x}\,n_x\,(1+m)-1)^2$ instead of $N=(2\,n_\text{el,x}\,n_x\,(1+m)+1)^2$ fibers. \begin{figure}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/poor_boundary_33x33.png}% \caption{Evaluation of the parallel mesh generation algorithm, \cref{alg:parallel_algorithm_1}: Resulting fibers and points on the fibers created with the parallel algorithm, $9\times 9$ fibers with $1481$ nodes each. Irregularities in the outer surface can be seen in the center at the bottom of the image.}% \label{fig:poor_boundary_33x33}% \end{figure}% \section{Results and Discussion} The following section presents results of the parallel algorithm for mesh generation, \cref{alg:parallel_algorithm_1}. In addition, the effect of various parameters is investigated. Two types of parameters can be distinguished. Parameters of the first type influences the number of nodes in the resulting mesh. These parameters have to be set such that the desired mesh resolution is achieved. Often, multiple, different parameter combinations are possible to achieve a given mesh resolution. Parameters of the second type have no effect on the mesh resolution but on the quality of the mesh. Usually, the parameter combination that gives the highest mesh quality should be chosen. In the following, \cref{sec:mesh_generation_resulting_meshes} shows results of the algorithm. Then, \cref{sec:mesh_generation_mesh_size_parameters} outlines how parameters of the first type affect the mesh resolution. A specific parameter, the recursion width, is discussed in \cref{sec:mesh_generation_recursion_width}. Subsequently, \cref{sec:mesh_generation_mesh_quality_parameters} evaluates and discusses parameters of the second type, which affect the mesh quality. \subsection{Resulting Meshes}\label{sec:mesh_generation_resulting_meshes} At first, results of the whole workflow described in \cref{sec:overview_and_notation_of_required_meshes} to \cref{sec:parallel_algorithm} are presented. The input for the mesh generation algorithm is a geometry representation, which is typically extracted from biomedical imaging. The output of the parallel algorithm, \cref{alg:parallel_algorithm_1}, comprises a 3D mesh with hexahedral elements as well as multiple, embedded 1D fiber meshes. \Cref{fig:muscle_meshes} visualizes some results for the biceps and triceps muscles. The parameter values $n_\text{el,x}=4$, $n_\text{el,z}=50$ and $m=0$ are chosen. If the recursion level is set to $l_\text{max}=0$, the algorithm generates meshes with the smallest possible number of fibers, which is a grid of $7 \times 7$ fibers. \Cref{fig:muscle_mesh_0} shows a grid of $7\times 7$ fibers and the corresponding 3D mesh that was sampled from the fiber data using every 50th point in $z$ direction of the fiber meshes. It can be seen that the generated fibers traverse all nodes of the generated 3D mesh and, thus, the 3D mesh is aligned with the fiber direction. \Cref{fig:muscle_mesh_1} shows a similar result with $9 \times 9$ fibers. Here, the colors correspond to the solution of an electrophysiology simulation. Blue regions indicate that the fiber membranes have an electric potential equal to their resting potential, which indicates no activation. Orange and red colors correspond to activated regions. It can be seen that the activation is present at the same locations on both the fibers and the 3D mesh. In the simulation, this requires data mapping from the fiber meshes to the 3D mesh. Because all nodes of the 3D mesh are located on the fibers, this data transfer becomes trivial. \Cref{fig:muscle_mesh_2,fig:muscle_mesh_3} present grids with $13 \times 13$ and $67 \times 67$ fibers of the biceps muscle, respectively. Results with larger numbers of fibers are not shown here because in such visualizations the fibers become less distinguishable. \Cref{fig:muscle_mesh_4,fig:muscle_mesh_5} show fibers for the triceps geometry. \begin{figure}% \centering% \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/muscle_mesh.png} \caption{Grid of $7 \times 7$ fibers (red) and the aligned 3D mesh with $7 \times 7 \times 30$ nodes (yellow).}% \label{fig:muscle_mesh_0}% \end{subfigure} \quad \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/muscle_mesh_1b.png} \caption{Grid of $9 \times 9$ fibers and 3D mesh with the solution of an electrophysiology simulation.}% \label{fig:muscle_mesh_1}% \end{subfigure} \quad \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[height=7cm]{images/parallel_fiber_estimation/muscle_mesh_2b.png} \caption{Grid of $13 \times 13$ fibers.}% \label{fig:muscle_mesh_2}% \end{subfigure} \\ \begin{subfigure}[t]{0.25\textwidth}% \centering% \includegraphics[height=5cm]{images/parallel_fiber_estimation/muscle_mesh_3.png} \caption{Grid $67 \times 67$ muscle fibers for the biceps geometry}% \label{fig:muscle_mesh_3}% \end{subfigure} \, \begin{subfigure}[t]{0.15\textwidth}% \centering% \includegraphics[height=5cm]{images/parallel_fiber_estimation/triceps_25x25.png} \caption{$25 \times 25$ fibers of triceps}% \label{fig:muscle_mesh_5}% \end{subfigure} \hfill \begin{subfigure}[t]{0.55\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/triceps_inside_67x67b.png} \caption{Grid of $67 \times 67$ fibers for the triceps geometry as seen from within the muscle. The total number of points is \num{8982489}.}% \label{fig:muscle_mesh_4}% \end{subfigure} \caption{Evaluation of the parallel mesh generation algorithm, \cref{alg:parallel_algorithm_1}: 1D fiber meshes and corresponding 3D meshes. The biceps geometry is used in (a)-(d), the triceps geometry in (e) and (f). The fibers have 1481 nodes each in the biceps muscle and 2001 nodes each in the triceps muscle.}% \label{fig:muscle_meshes}% \end{figure}% \subsection{Effect of Mesh Size Parameters}\label{sec:mesh_generation_mesh_size_parameters} Next, the type of parameters that affect the resulting mesh resolution is discussed. The choices of the maximum recursion level $\ell_\text{max}$, the number $n_\text{el,x}$ of elements in $x$ direction of the subdomains and the fine grid parameter $m$ determine the resulting number $N$ of fibers and, thus, the file size of the binary output file. The formulas for these numbers were given in \cref{sec:postprocessing_of_the_generated_streamlines}. \Cref{tab:file_sizes} lists exemplary numbers of fibers and file sizes for $n_\text{el,x}=4$ and different values of $\ell_\text{max}$ and $m$. The number $n_\text{proc}$ of required processes to reach the maximum recursion level is also listed, it depends on $\ell_\text{max}$ by $n_\text{proc}=8^{\ell_\text{max}}$. Two different numbers of fibers and corresponding file sizes are listed for every parameter combination. The two variants correspond to the two files that include respectively omit the fibers at the boundary. The table shows that meshes with different sizes can be constructed by appropriate choices of parameters. A realistic biceps muscle contains about \num{200000} to \num{400000} muscle fibers \cite{MacDougall1984}. The table shows that constructing a mesh in this range yields a file with a size of $\approx$\SI{10}{\gibi\byte}.\footnote{ In this work, file sizes are given using multiples of bytes (B) and the prefixes defined in the ISO/IEC International System of Quantities \cite{ISOmebi}. The prefixes are: 1 kibibyte (\SI{1}{\kibi\byte})=$2^{10}$ bytes, 1 mebibyte (\SI{1}{\mebi\byte})=$2^{20}$ bytes, 1 gibibyte (\SI{1}{\gibi\byte})=$2^{30}$ bytes} A mesh that contains \SI{1}{\percent} of the realistic number of fibers can be stored in a file with size of $\approx$\SI{100}{\mebi\byte}. The binary files to store the generated meshes are small compared to ASCII-based file formats as each point coordinate is represented by only eight bytes. For comparison, the ASCII-based \emph{exnode} format defined within the OpenCMISS framework uses 24 characters, i.e., 24 bytes to store one point coordinate. Additionally, a larger memory overhead for the description of the data is needed such that \emph{exnode} files are more than three times larger than the binary files used in \opendihu{}. The binary file format uses no compression that could further reduce the file size. The reason is that no extra effort should be needed when writing programs that parse these files. Thus, they can easily be handled by codes in different programming languages. For example, within \opendihu{} the file format is understood by various Python scripts and C++ programs. \begin{table} \centering% \begin{tabular}{|l|l|l| r@{\,=\,}l r|rr|} \hline max. &fine grid &\# proc. & \multicolumn{2}{c}{\begin{tabular}{lrr} \# fibers && \end{tabular}}&file size \\ level $\ell_\text{max}$ & $m$ & $n_\text{proc}$ & \multicolumn{2}{c}{\begin{tabular}{lll} &&\end{tabular}}&\\ \hline 0 & 0 & 1 & $9\times 9$ & \num{81} & \SI{2.7}{\mebi\byte} \\ & & & $7\times 7$ & \num{49} & \SI{1.7}{\mebi\byte}\\ 0/1 & 1/0 & 1/8 & $17\times 17$ & \num{289} & \SI{9.8}{\mebi\byte}\\ & & & $15\times 15$ & \num{225} & \SI{7.6}{\mebi\byte}\\ 0/1/2 & 3/1/0 & 1/8/64 & $33\times 33$ & \num{1089} & \SI{36.9}{\mebi\byte}\\ & & & $31\times 31$ & \num{961} & \SI{32.6}{\mebi\byte}\\\hline 0 & 7 & 1 & $65\times 65$ & \num{4225} & \SI{143.2}{\mebi\byte}\\ & & & $63\times 63$ & \num{3969} & \SI{134.5}{\mebi\byte}\\ 2 & 7 & 64 & $257 \times 257$ & \num{66049} & \SI{2.2}{\gibi\byte}\\ & & & $255 \times 255$ & \num{65025} & \SI{2.2}{\gibi\byte}\\ 2 & 15 & 64 & $513 \times 513$ & \num{263169} & \SI{8.7}{\gibi\byte}\\ & & & $511 \times 511$ & \num{261121} & \SI{8.6}{\gibi\byte}\\ \hline \end{tabular} \caption{Parallel 1D and 3D mesh generation: Different parameter choices of $l_\text{max}$ and $m$ and the resulting number $n_\text{proc}$ of processes, number of fibers and file size. Some results can be achieved with different parameter combinations, e.g., both $\ell_\text{max}=0, m=1$ and $\ell_\text{max}=1, m=0$ result in $17\times 17$ fibers. These combinations are separated by slashes.}% \label{tab:file_sizes}% \end{table} \subsection{Effect of the Recursion Width}\label{sec:mesh_generation_recursion_width} Some numbers of fibers can be achieved with multiple, different parametrizations that use different recursion widths. \Cref{tab:file_sizes} contains such alternatives for $\ell_\text{max}$ and $m$ separated by slashes in the second and third row. For example, the three combinations $(\ell_\text{max} = 0, m=3)$, $(\ell_\text{max} = 1, m=1)$, and $(\ell_\text{max} = 2, m=0)$ all lead to a grid of $31 \times 31$ fibers (without boundary layer). However, the spatial location of the fibers in the muscle is not identical for these alternatives, because the intermediate mesh used for the streamline tracing of the fibers is differently resolved. In the case with recursion depth $\ell_\text{max} = 0$ and fine grid interpolation parameter $m=3$, numerous of the resulting fibers are interpolated from a coarse grid whereas in the case with $\ell_\text{max} = 2$ and $m=0$ all fibers are key fibers and are obtained by streamlines tracing through a fine mesh. \Cref{fig:different_recursion_levels} shows parts of the resulting meshes at the longitudinal center of the muscle for these two cases. In \cref{fig:31x31_l0_center}, the mesh obtained with $l_\text{max}=0$ consists of a grid of traced key fibers and an interpolated finer grid of fibers. The key fiber grid is given by the corners of the gray checkerboard pattern in the image. It can be seen that the mesh consists of patches with $4 \times 5$ or $5 \times 5$ fibers that each have equal element lengths and angles. In comparison, the mesh in \cref{fig:31x31_l2_center} that was obtained with $l_\text{max}=2$ consists only of key fibers. Here, the change in shape going from one element to its neighbors occurs more smoothly than in \cref{fig:31x31_l0_center}. This qualitatively implies a higher mesh quality. To quantify this effect, we introduce a measure for mesh quality and compare the scores of the three alternatives in the present example. We consider all angles that occur in an element in the $x$-$y$ plane. The mean value of all angles is obviously $\pi/2$. The variance of all angles can be used as the measure for mesh quality. If the variance is low, this indicates similar elements and, thus, good mesh quality. For the present example, the variance was computed for the mesh with $31 \times 31$ fibers and \num{1481} nodes per fiber and, thus, $\num{1332000}$ 3D elements in total. \Cref{fig:2mesh_quality} plots the variance for the three cases given in the third row of \cref{tab:file_sizes}, i.e., parameter combinations $(l_\text{max}=0,m=3)$, $(l_\text{max}=1,m=1)$ and $(l_\text{max}=2,m=0)$. The 3D mesh corresponding to the lowest bar contains the 2D mesh shown in \cref{fig:31x31_l0_center} and the 3D mesh corresponding to the upper-most bar contains \cref{fig:31x31_l2_center}. It can be seen that the quality of meshes on higher recursion levels with less interpolation increases as expected. This emphasizes the benefit of the parallel algorithm that uses finer meshes compared to the mesh used during serial execution of the algorithm. \begin{figure}% \centering% \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[width=0.9\textwidth,trim=0 0 8mm 1cm, clip]{images/parallel_fiber_estimation/31x31fibers_l0_m3_2In_dirichlet.pdf} \caption{Result for parameters $l_\text{max}=0, m=3$, i.e., with three interpolated fibers between every two traced fibers.}% \label{fig:31x31_l0_center}% \end{subfigure} \quad \begin{subfigure}[t]{0.45\textwidth}% \centering% \includegraphics[width=0.9\textwidth,trim=0 0 8mm 1cm, clip]{images/parallel_fiber_estimation/31x31fibers_l2_m0_2In_dirichlet.pdf} \caption{Result for parameters $l_\text{max}=2, m=0$, i.e., without interpolation.}% \label{fig:31x31_l2_center}% \end{subfigure} \caption{Comparison of generated meshes of the biceps with different maximum recursion levels $l_\text{max}$. A lower left portion of the full mesh with $31 \times 31$ fibers is shown.}% \label{fig:different_recursion_levels}% \end{figure}% \begin{figure}% \centering% \includegraphics[height=6cm]{images/parallel_fiber_estimation/mesh_quality_recursion_level.pdf}% \caption{Variance of the element angles for meshes with the same number of $31 \times 31$ fibers, but created by different recursion levels $\ell_\text{max}$. The parameters correspond to the third row of \cref{tab:file_sizes}. A lower variance means better mesh quality.}% \label{fig:2mesh_quality}% \end{figure}% \subsection{Effect of Mesh Quality Parameters}\label{sec:mesh_generation_mesh_quality_parameters} In addition to the parameters that affect the resulting mesh sizes, $n_\text{el,x}$, $l_\text{max}$ and $m$, further options exist to tune the behavior of \cref{alg:parallel_algorithm_1} and in result lead to meshes with different quality. These options are described in the following. The surface that is the input to \cref{sec:parallel_algorithm} can be represented either as triangulation or in parametric form as NURBS surface. The triangulation can either be the result of the image segmentation step or it can be obtained by triangulating a NURBS surface. Thus, if the approximation of the geometry by a smooth spline surface is desired it is possible to choose between both options. One difference is the resulting runtime. To sample a point on the surface using the NURBS representation, the nonlinear equation has to be inverted using a Newton scheme for each point. This is slower than using the triangulation where rings on $x$-$y$ planes are extracted initially and then equidistantly sampled, as explained in \cref{sec:slicing_of_the_geometry} and \cref{sec:data_structure_of_boundary_points}. The Laplace problem $Δp=0$ that is solved in every recursion depends on the discretization and mesh resolution on every subdomain. In addition to the number $n_\text{el,x}$ of elements in $x$ and $y$ directions, the mesh resolution also follows from the number $n_\text{el,z}$ of elements in $z$ direction. The number of elements in this intermediate mesh is also influenced by the factor $r\in \N$ of the refinement described in \cref{sec:solution_of_the_laplace_problem}. While $r=1$ corresponds to no refinement, for $r>1$ the number of elements is increased by the factor $r^3$. Note that the output meshes of the algorithm depend on $n_\text{el,x}$ but not on $n_\text{el,z}$ nor $r$ as they are generated later after the process of streamline tracing. Furthermore, the finite element discretization of the Laplace problem can either use linear or quadratic ansatz functions, leading to the respective linear or quadratic elements in the mesh. The type of boundary conditions for the Laplace problem in \cref{eq:fiberest_laplace} can be selected among the Neumann boundary conditions given by \cref{eq:fiberest_neumann} or the Dirichlet boundary conditions given by \cref{eq:fiberest_dirichlet}. After the Laplace problem is solved, the gradient direction $\nabla p(\bfx)$ at a point $\bfx$ in the domain needed for streamline tracing can be determined by two different methods. Either the gradient vector field is precomputed using finite differences and the nodal values of the solution field $p$ and then evaluated at $\bfx$. Or the gradient value is directly interpolated at $\bfx$ in the 3D element using a linear combination of the solution values and derivatives of the ansatz functions of the element. During parallel streamline tracing, the width $n_\text{ghost\_layer\_width}$ of the ghost layer is important. If it is too small, streamlines leave the domain of the process and have to be repaired, i.e., approximated by neighboring streamlines as described in \cref{sec:repair_of_incomplete_streamlines}. We found that a value of $n_\text{ghost\_layer\_width}=2\,r$ is enough and minimizes the number of invalid streamlines leaving the domain. With some parameter combinations, invalid streamlines still occur occasionally. Those result from badly conditioned elements and gradient values with high numerical errors that cannot be fixed by a larger ghost layer. By including the factor $r$ in the ghost layer width, the actual sizes of the ghost layer is always the same independent of the chosen refinement. To investigate the effect of all options, a parameter study is conducted in the following. We fix the values of $n_\text{el,x}=4$, $n_\text{el,z}=50$, $m=1$, $l_\text{max}=1$, and $n_\text{ghost\_layer\_width}=2\,r$ and vary all other parameters. The resulting meshes consist of $33 \times 33$ fibers and $31 \times 31$ fibers if the boundary layer is omitted, as described in \cref{sec:postprocessing_of_the_generated_streamlines}. We compare the mesh quality of the 3D meshes that result from the $31 \times 31$ fibers. As before, the variance of the element angles is used to rate the quality of each resulting mesh. To identify a parameter combination in the study, a scenario name is composed of one character each for the various options, as explained in the following. The linear or quadratic formulation of the Laplace problem is indicated by the characters \say{$\ell$} or \say{q}. Neumann and Dirichlet boundary conditions are indicated by \say{N} and \say{D}. The refinement level $r$ is specified by the respective integer value. Finally, \say{g} or \say{s} indicates whether the precomputed gradient field (\say{g}) is used in streamline tracing or the solution values (\say{s}) and derivatives of the ansatz functions. For example, the scenario considered in \cref{fig:different_recursion_levels,fig:2mesh_quality} can be specified as \say{$\ell$D2s}, as it uses the linear mesh with Dirichlet boundary conditions, a refinement factor $r=2$ and the solution values to compute the gradient. The following study was performed for the biceps geometry in two variants, firstly using the approximated NURBS surface directly and secondly using a triangulation obtained from the NURBS surface. These two variants are indicated by \say{splines} for the NURBS surface and \say{stl} for the STL file containing the triangulation. %Additionally, also the variance of relative element lengths in the $x$-$y$ planes was computed, using the calculation explained in \cref{sec:mesh_generation_0_results_and_discussion}. Most scenarios yielded a value of \num{2.2e-2}. Scenarios with significantly different values were discarded, as their results contained incomplete streamlines. \Cref{fig:3mesh_quality} presents the resulting variances of the element angles. The scenarios are sorted according to their mesh quality score, i.e., the variance of their element angles. This means the results are ordered by improving mesh quality from bottom to top. \begin{figure}% \centering% \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh_quality_options_0.pdf}% \caption{Scenario using the surface triangulation.}% \label{fig:mesh_quality_options_0}% \end{subfigure} \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh_quality_options_1.pdf}% \caption{Scenario using the spline surface.}% \label{fig:mesh_quality_options_1}% \end{subfigure} \caption{Parallel mesh generation algorithm: Comparison of the mesh quality that results from different options in the mesh generation algorithm. A lower variance means better mesh quality.}% \label{fig:3mesh_quality}% \end{figure}% Two separate plots for the \say{stl} and \say{splines} scenarios are shown in \cref{fig:mesh_quality_options_0} and \cref{fig:mesh_quality_options_1}. The two resulting meshes of the best options, \say{qN1s\_stl} and \say{qN1s\_splines} are visualized left and right in \cref{fig:stl_splines_results}. In the top plane of the muscle belly, it can be seen that the orientation of the mesh is slightly different. This explains the large difference of the angle variance values between the two scenarios in \cref{fig:3mesh_quality}, which are higher for \say{splines} than for \say{stl}. The scores of parameter combinations should therefore only be compared among the same surface representation. A statement regarding which of the two options is better is not reasonable from this data set. \begin{figure}% \centering% \includegraphics[width=0.8\textwidth]{images/parallel_fiber_estimation/stl_splines_results.png}% \caption{Parallel mesh generation algorithm: Results of the scenarios using a surface triangulation (stl, left) and a NURBS surface (splines, right).}% \label{fig:stl_splines_results}% \end{figure}% The results in \cref{fig:3mesh_quality} show that almost all values are close together, which indicates similar good mesh qualities for different parameter combinations. Nevertheless, the ranking reveals some differences between the options. A comparison of the rankings in the columns for \say{stl} or \say{splines} shows that some parameter choices consistently resulted in better meshes. A better result was achieved if Neumann boundary conditions were used (\say{N}) compared to Dirichlet boundary conditions (\say{D}). Similarly, quadratic ansatz functions (\say{q}) performed better than linear ansatz functions (\say{$\ell$}). This is reasonable as quadratic ansatz functions yield a higher spatial consistency in the finite element formulation. A higher refinement factor $r$ of the internal mesh was beneficial for the cases with linear ansatz functions. For quadratic ansatz functions, the effect of $r$ is less clear. The study shows that no refinement ($r=1$) is often better in this case. The variant without the precomputed gradient field (\say{s}) performed generally better than the variant with gradient field computation (\say{g}). However, further studies with different recursion depths showed that the effect of some options also depended on the scenario. For a higher recursion depth, Dirichlet boundary conditions turned out to be more robust in the sense that fewer incomplete streamlines occurred. %Larger refinement factors led to smaller mesh widths and in consequence to a smaller ghost layer for the streamline tracing, which always has a thickness of one element. Thus, more streamlines left their subdomain and became invalid. Therefore, e.g., a smaller value of $r$ yielded better results for $\ell_\text{max}=2$. In summary, the quadratic formulation (\say{q}) and the streamline tracing using solution values (\say{s}) could be shown to be better options than their alternatives. For a maximum recursion level $\ell_\text{max}=1$, the best parameter combination among the tested combination was \say{qN1s}. In a separate study for $\ell_\text{max}=2$, the combination \say{qD2s} was found to be as robust as \say{qN1s}. % \subsection{Post-processing of the Meshes} To further improve the mesh quality on every cross-section, we apply two more post-processing steps, one local and one global transformation. As can be seen in the cross-section of \cref{fig:stl_splines_results}, some rows of elements in the generated mesh have zigzag lines, and not all elements are equally sized. Furthermore, there are almost degenerate elements with small interior angles. Thus, we apply a first, local transformation on the mesh. This operation consists of Laplacian smoothing and randomly deflecting points, where interior element angles are smaller than \SI{20}{\degree}. If such deflections result in invalid self-intersecting elements, the self-intersection is resolved, which potentially again introduces small interior angles. The total transformation consists of 25 iterations of alternatingly applying the smoothing step and the improvement step of small interior angles. An exemplary resulting mesh after this transformation is shown in \cref{fig:mesh47_a}. It can be seen that all lines in the mesh are smooth and almost straight. At the right center of the shown mesh, the effects of the deflection step, which improves small interior angles, can be seen. However, at the left, top and bottom boundary of the mesh, degenerate elements with small interior angles remain. This is especially true for the four mesh points on the boundary that correspond to the corners of the quadrangulation. At these points, elements are present that have two sides that are part of the mesh boundary, forming an interior angle of almost \SI{180}{\degree}. Three points of these elements are located in an almost straight line. As a consequence, the Laplacian smoothing step moves the fourth point of these elements close to this straight line, which adds another large interior angle and degenerates the element. This effect also occurs for interior elements of the mesh that are close to these points on the boundary. \Cref{fig:mesh47_b} shows the detail of the left boundary of the mesh in \cref{fig:mesh47_a}, where this effect can be seen. The area of the elements decreases towards the boundary and the elements get more degenerate in this direction. As a remedy, we perform the second, global transformation step to counteract this tendency. In this step, most of the points in every cross-section of the mesh are translated by a fixed mapping, such that the small elements at the boundary are transformed into elements of better quality. % plot of function f(r) \begin{figure} \centering% \includegraphics[width=0.5\textwidth]{images/parallel_fiber_estimation/extend_mesh_plot.pdf}% \caption{Transformation of a mesh to improve the mesh quality. The depicted function $f$ transforms the elements in radial direction.}% \label{fig:extend_mesh_plot}% \end{figure}% Each mesh point on a cross-section of the mesh is represented by polar coordinates $(r,\varphi)$. The radius $r$ is transformed according to the function $r_\text{new} = y(r_\text{old})$ that is depicted in the lower plot of \cref{fig:extend_mesh_plot}. This piecewise defined function $f\in \CC^1([0,1]\to [0,1])$ is linear for $x\in [0,s]$ and a polynomial of degree 3 for $x \in [s,1]$. It passes through the yellow point in \cref{fig:extend_mesh_plot}, which in $x$ direction is at the center of the interval $[s,1]$ and in $y$ direction is at fraction $\alpha$ of the shown yellow line. The parameter $\alpha$ controls the extent, to which the mesh is transformed in radial direction. The parameter $s$ specifies the size of a region around the center of the mesh where no transformation occurs. We obtain good results by choosing $s=0.05$ and $\alpha=0.7$. The resulting function takes the form $f(r) = 1.330\,r^3 - 1.463\,r^2 + 1.136\,r - 0.003$. The first derivative $f'(x)$ of this function is shown in the upper plot of \cref{fig:extend_mesh_plot}. It quantifies the amount of extension or compression of the mesh elements in radial direction. The right side of the plot corresponds to the outer boundary of the mesh. There, the elements are extended, since $f'(r) > 0$. To compensate this extension, the elements have to be compressed towards the interior of the mesh where $f'(r) < 0$. The range of $r \in [0,s]$ corresponds to the region in the interior of the mesh that is not transformed. The points of the mesh are transformed in radial direction by adjusting their coordinate $r$ and not transformed in circumferential direction, i.e., the angle $\varphi$ remains constant. However, the application of the function $f$ on $r$ is additionally modulated by a piecewise sine function depending on $\varphi$. The transformation $f$ is only fully applied at the four radii corresponding to the described special points on the boundary, around which the degenerated elements occur. In between these lines, the transformation is reduced and some points of the mesh are not transformed at all. \Cref{fig:mesh47_b} shows the mesh of \cref{fig:mesh47_a} after this transformation has been applied. \Cref{fig:mesh47b_} shows the extract of the mesh from the left boundary that corresponds to the same extract of the original mesh in \cref{fig:mesh47a_}. It can be seen that the quality of the elements is improved close to the boundary. The area of the rectangles is now approximately equal and no small interior angles occur. In \cref{fig:mesh_47ab}, the previous and the transformed mesh are overlaid to show the regions that remain unmodified. The unmodified parts form a \say{cross} shape that touches the boundary at the regions where the mesh quality is also good in the original mesh. % original and transformed mesh \begin{figure} \centering% \begin{tabular}{cc} \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh47_a.png}% \caption{Original mesh at the upper-most cross-section of the muscle.}% \label{fig:mesh47_a}% \end{subfigure} & \begin{subfigure}[t]{0.30\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh47_b.png} \caption{Transformed mesh with better mesh quality.}% \label{fig:mesh47_b}% \end{subfigure} \\ \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh47a_.png} \caption{Extract at the left boundary of the original mesh in (a), rotated clockwise by \SI{90}{\degree}.}% \label{fig:mesh47a_}% \end{subfigure} & \begin{subfigure}[t]{0.48\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh47b_.png} \caption{Analog extract to (c) of the transformed mesh (b).}% \label{fig:mesh47b_}% \end{subfigure} \end{tabular} \begin{subfigure}[t]{0.7\textwidth}% \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/mesh_47ab.png} \caption{Overlay of the original mesh (a) and the transformed mesh (b), rotated clockwise by \SI{90}{\degree}. The unmodified regions can be seen.}% \label{fig:mesh_47ab}% \end{subfigure} \caption{Transformation of a biceps mesh with $47\times 47$ points to improve the mesh quality.}% \label{fig:mesh47}% \end{figure}% % \subsection{Usage of the Generated Meshes in Simulations} In some biomechanical simulations, also a body fat and skin layer on top of the muscle is considered. In this case, an appropriate mesh is required that is attached to the muscle mesh and seamlessly matches the elements of the muscle mesh. The construction of such a mesh is discussed later in \cref{sec:construction_and_partitioning_of_the_mesh} together with the parallel partitioning. A visualization of such a parallel setting is given in \cref{fig:partitioning_biceps}. Tendons connect the muscle belly of the biceps brachii muscle to the humerus and ulna bones at the top left and bottom, respectively. The muscle mesh consists of fibers that are colored according to a parallel partitioning. In a parallel partitioning, every process contributes calculations only on its associated spatial subdomain to the overall computation. On the right-hand side of the muscle, a layer of adipose tissue is attached to the muscle belly. This layer is needed, if electromyography on the skin surface is simulated. % fat layer mesh and partitioning \begin{figure} \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/partitioning_biceps.png}% \caption{Summary visualization of the simulation setup in this work (I): Biceps muscle with tendons and bones, parallel partitioned fibers and a fat layer mesh.}% \label{fig:partitioning_biceps}% \end{figure}% \Cref{fig:muscle_meshes_raytrace} shows another use case of various meshes in a multi-scale simulation model. The muscle is cut open for visualization purposes. The figure depicts numerous muscle fibers in the upper part of the muscle belly. The fibers are colored according to simulation results of the electric membrane potential, which is responsible for the activation of the muscle. The lower part of the muscle shows elements of the 3D mesh. The coloring corresponds to the electric potential, that is measured during intramuscular electromyography in the interior of the muscle or during surface electrophysiology on the outside of the muscle. % 3D and 1D meshes \begin{figure} \centering% \includegraphics[width=\textwidth]{images/parallel_fiber_estimation/muscle_meshes_raytrace.png}% \caption{Summary visualization of the simulation setup in this work (II): Muscle fibers in the upper part and 3D mesh elements in the lower part of the muscle belly.}% \label{fig:muscle_meshes_raytrace}% \end{figure} \clearpage %19G resulting_meshes %The created meshes have been written to "resulting_meshes". Now you can copy these files to the examples/electrophysiology/input directory %real 1047m29.662s %user 3970m33.404s %sys 4m6.717s \label{sec:repro_tendon_meshes} \begin{reproduce} The parallel algorithm is implemented in the example \code{parallel_fiber_estimation}. Numerous parameters can be set on the command line. After compilation, run the program as follows to get a description of all available options. \begin{lstlisting}[columns=fullflexible,breaklines=true,postbreak=\mbox{\textcolor{gray}{$\hookrightarrow$}\space}] cd $\$$OPENDIHU_HOME/examples/fiber_tracing/parallel_fiber_estimation/build_release ./generate ../settings_generate.py --help \end{lstlisting} Running the program without options and \code{--help} uses sensible default values. A given surface triangulation of the biceps muscle gets used by default. To compute the examples shown in this section, use and adjust the following script that runs the \code{generate} program and computes the mesh quality: \begin{lstlisting}[columns=fullflexible,breaklines=true,postbreak=\mbox{\textcolor{gray}{$\hookrightarrow$}\space}] cd $\$$OPENDIHU_HOME/examples/fiber_tracing/parallel_fiber_estimation/build_release ../run.sh \end{lstlisting} Computation of mesh and file sizes as shown in \cref{tab:file_sizes} can be done using the \emph{compute\_sizes.py} script. While the previously given commands are good for exploring the algorithms, generation of the meshes used for the simulation involves some more steps. Dedicated scripts exist that perform all steps and call the algorithms with the proper parameters. Starting from the STL file extracted from cmgui, as explained in \cref{sec:surf_extr}, the next steps are: \begin{itemize}[leftmargin=1cm] \item[(i)] Scale the points from millimeters (used in the Visible Human dataset) to centimeters (used in the simulation), \item[(ii)] remove the interior triangles, \item[(iii)] translate the mesh such that the bounding box begins at $z=0$, this is needed for the programs used in the next steps, \item[(iv)] create the spline surface representation as explained in \cref{sec:nurbs}, \item[(v)] compile and run the \opendihu{} programs to create the binary files of the 3D mesh and the 1D fibers meshes, the algorithm in \cref{sec:parallel_algorithm} is used, \item[(vi)] adjust the indexing and undo the translation in (iii), \item[(vii)] refine the created meshes of key fibers by different numbers $m$ of fine grid fibers, in total 10 different mesh sizes are created for differently refined simulations, \item[(viii)] create meshes for the fat layer $\Omega_B$ on top of the muscle surface, also in 10 different resolutions. \end{itemize} Two scripts are given for the biceps brachii and triceps brachii muscles. They perform all listed steps and also create intermediate output files that can be used to understand the process. Some steps are automatically skipped if the resulting output file already exists from a previous run. This is especially helpful for the removal of the interior triangles from the initial file which takes nearly a full day. A third scripts creates three meshes $\Omega_{T,i}$ for the tendons of the biceps muscle, as visualized in \cref{fig:tendon_meshes}. At the bottom, a single tendon mesh is created whereas at the top, two separate tendons exist. The script involves numerous rotation and cropping operations of the initial surface, before the algorithm of \cref{sec:ser_alg_meshes} is executed. The three output files of the tendon meshes have the same file format as the muscle meshes. The files have the extension \emph{.bin} for \say{binary}. The script \code{examine_bin_fibers.py} can be used to debug the created binary files. The three scripts can be executed as follows: \begin{lstlisting}[columns=fullflexible,breaklines=true,postbreak=\mbox{\textcolor{gray}{$\hookrightarrow$}\space}] cd $\$$OPENDIHU_HOME/examples/electrophysiology/meshes ./process_meshes_biceps.sh ./process_meshes_triceps.sh ./process_meshes_tendons.sh \end{lstlisting} The output can be found in the subdirectory \code{processed_meshes}. For the total output about \SI{68}{\gibi\byte} of drive space is required, however, the resulting meshes have a size of only \SI{19}{\gibi\byte}. A total runtime of more than a day is to be expected. \end{reproduce} \section{Conclusion and Future Work}\label{sec:meshes_summary_and_conclusion} This chapter presented algorithms for creating muscle meshes that are needed for multi-scale simulations of the musculoskeletal system. For the biceps muscle, 3D meshes for tendons on both ends and the muscle were created. Additionally, 1D fibers meshes were generated that are embedded in the mesh of the muscle. The 3D mesh and the 1D meshes resulting from the parallel algorithm are aligned with each other. This facilitates data mapping between the meshes and reduces numerical errors. All generated meshes are structured, which allows an efficient parallelization. First, an overview of available meshing software and known algorithms in the literature was given. Very little software tools were capable of generating structured meshes and none fitted our special needs. Therefore, own algorithms were developed to generate meshes starting from medical imaging data. A workflow was presented to generate a smooth surface triangulation from imaging data. Our base data was the male dataset from the Visible Human Project. Two alternatives within this workflow were presented, where the first alternative executed automatic image segmentation based on morphological operations and the second alternative used semi-automatic segmentation tools from the Physiome project. Then, smooth NURBS surfaces were fitted to the extracted boundaries of the muscle volumes. Next, a novel algorithm to create structured meshes from a triangulated muscle surface was presented. The algorithm used harmonic maps on 2D slices in combination with regular grids in a parameter space to achieve good mesh quality. A method of computing streamlines in a divergence-free vector field to estimate muscle fibers, which is established in the literature, was used. It allowed embedding 1D meshes for muscle fibers in the created 3D meshes of the muscle. Numerical experiments tested and evaluated different choices of triangulation and quadrangulation schemes for the 2D cross-section and reference domains in our algorithm. Next, a parallelized algorithm was introduced that was based on our first, serial algorithm. The algorithm used distributed memory parallelism and provided the same features as the serial algorithm, having the same formats for input and output. The difference was that it constructed a fine, partitioned mesh for streamline tracing that was distributed over all employed processes. Thus, it was possible to create finer meshes using more compute nodes. Differently resolved meshes of the biceps and triceps muscle volumes and muscle fibers were created using this algorithm. The superiority of the parallel algorithm using a higher number of processes compared to the serial execution was explained and demonstrated in a numerical experiment. Several options to fine-tune the algorithm were evaluated. Post-processing methods were described that improved the mesh quality of the resulting meshes. The presented algorithms and their implementation in \opendihu{} are the basis for further computations within this work. They are used to generating structured hexahedral meshes with good mesh quality. These meshes are required for efficient, parallel finite element simulations of various aspects of the neuromuscular system. The presented algorithms are specialized for fusiform muscles and require the muscle geometry to be oriented along one coordinate axis (the $z$ axis) in order to generate a structured mesh that comprises planar slices that are normal to that direction. The algorithms can also be applied to any tubular surface geometry of more complex muscles and will construct the corresponding structured 3D mesh. The generated 1D fiber meshes, however, are only valid for muscles, where the approach of streamline tracing through the solution of the Laplacian potential flow problem with boundary conditions at the bottom and top ends of the muscle can be applied. In literature, this approach has been successfully used for various muscles with more complex fiber architectures, such as the tibialis anterior, gluteus maximus and deltoid muscles \cite{Choi2013}. However, the locations where boundary conditions were prescribed was not always at the bottom and top ends of the muscle. If in future work muscles with more complex layouts should be simulated, the approach could be as follows. Depending on the complexity of the outer geometry, first the presented algorithms (either \cref{alg:serial_algorithm_1,alg:serial_algorithm_2} or \cref{alg:parallel_algorithm_1}) can be used to create a structured 3D mesh. Then, a potential flow simulation can be manually setup in \opendihu{} using the 3D mesh and boundary conditions defined at proper locations. Seed points have to be defined and the streamline tracer of \opendihu{} can be used to create fiber meshes. In consequence, the resulting 1D fibers will not be aligned with the 3D mesh. Algorithmically, this poses no problem to the simulations in \opendihu{} as the data mapping functionality can handle arbitrarily positioned meshes. However, the parallel partitioning gets more involved as the combined domain of 1D and 3D meshes has to be partitioned equally for both mesh types. % \begin{tiny}(Cee04)\end{tiny} Une inégalité vient de \begin{displaymath} \rg(P Q)\leq \min(\rg(P), \rg(Q)) \end{displaymath} Pour l'autre, on note $r=\rg(A)$ et on considère les colonnes \begin{displaymath} C_{i_1},\cdots, C_{i_r} \end{displaymath} formant une base de l'espace des colonnes de $A$. La matrice du produit scalaire canonique dans cet espace est inversible. Elle est extraite de $\trans A\, A$. 0 \newcommand{\sshare}[1]{\langle #1 \rangle} \newcommand{\bsshare}[1]{\Big\langle #1 \Big\rangle} \newcommand{\BList}{\mathsf{BitsList}} \newcommand{\SList}{\mathsf{SquaresList}} \newcommand{\MList}{\mathsf{MultList}} \newcommand{\Zk}{\Z_{\langle k \rangle}} \newcommand{\Zgam}{\Z_{\langle \gamma \rangle}} \newcommand{\Qk}[1]{\Q_{\langle k, #1 \rangle}} \newcommand{\Random}{\mathsf{Random}} \newcommand{\Open}{{\mathsf{Open}}} \newcommand{\Mult}{\mathsf{Mult}} \newcommand{\abort}{\mathsf{abort}} \newcommand{\asn}{\leftarrow} \mathchardef\mhyphen="2D \newcommand{\F}{{\mathbb{F}}} \newcommand{\N}{{\mathbb{N}}} \newcommand{\Q}{{\mathbb{Q}}} \newcommand{\Z}{{\mathbb{Z}}} \newcommand{\true}{{\mathsf{true}}} \newcommand{\false}{{\mathsf{false}}} \newcommand{\floor}[1]{{\lfloor #1 \rfloor}} \newcommand{\ceil}[1]{{\lceil #1 \rceil}} \newcommand{\angles}[1]{{\langle #1 \rangle}} \newcommand{\err}{\mathsf{err}} \newcommand{\modp}{\bmod \; p} \newcommand{\modn}{\bmod \; 2^{64}} \newcommand{\note}[2]{{\textcolor{blue}{\sf (#1's Note:) {\sl{#2}}}}} \newcommand{\todo}[1]{{\textcolor{red}{\sf (TO DO:) {\sl{#1}}}}} \newcommand{\notetoreader}[1]{{\bf \em Note}: #1} \usepackage[small,nohug,heads=vee]{diagrams} \diagramstyle[labelstyle=\scriptstyle] \titleformat*{\paragraph}{\normalfont\normalsize\bfseries\color{blue}} \lstset{ % language=Python, % choose the language of the code basicstyle=\ttfamily\small, % the size of the fonts that are used for the code numbers=none, % where to put the line-numbers numberstyle=\tiny, % the size of the fonts that are used for the line-numbers stepnumber=1, % the step between two line-numbers. If it is 1 each line will be numbered numbersep=5pt, % how far the line-numbers are from the code %backgroundcolor=none, % choose the background color. You must add \usepackage{color} showspaces=false, % show spaces adding particular underscores showstringspaces=false, % underline spaces within strings showtabs=false, % show tabs within strings adding particular underscores frame=none, % adds a frame around the code tabsize=2, % sets default tabsize to 2 spaces captionpos=b, % sets the caption-position to bottom breaklines=true, % sets automatic line breaking breakatwhitespace=false, % sets if automatic breaks should only happen at whitespace escapeinside={\%*}{*)} % if you want to add a comment within your code } \lstnewenvironment{mylisting}{% \vspace{0.5em} }{% \vspace{0.8em} } \newenvironment{class}[1] { {\noindent\colorbox{black!10}{\texttt{#1}}}\hspace{0pt} \begin{quote} } { \end{quote} } \newcommand{\displaytt}[1]{\vspace{1em}\centerline{\texttt{#1}}\vspace{1em}} \newcommand{\func}[1]{\paragraph{\texttt{#1}}\hspace{0pt} \vspace{0.5em}\\} \newcommand{\id}{\mathfrak{i}} \newcommand{\Red}[1]{{\color{red}{#1}}} \newcommand{\Blue}[1]{{\color{blue}{#1}}} \newcommand{\grp}[1]{{\langle #1 \rangle}} \mathchardef\mhyphen="2D \newcommand{\Sqr}{\mathsf{Sqr}} \newcommand{\Ran}{\mathsf{Ran}} \newcommand{\SolvedBits}{\mathsf{Solved{\mhyphen}Bits}} \newcommand{\Symm}{\mathsf{Symm}} \newcommand{\Pre}{\mathsf{Pre}} \newcommand{\BitLT}{\mathsf{Bit{\mhyphen}LT}} \newcommand{\BitAdd}{\mathsf{Bit{\mhyphen}Add}} \newcommand{\Xor}{\mathsf{Xor}} \newcommand{\ModTm}{\mathsf{Mod2m}} \newcommand{\Trunc}{\mathsf{Trunc}} \newcommand{\TruncPr}{\mathsf{TruncPr}} \newcommand{\Bits}{\mathsf{Bits}} \newcommand{\Sgn}{\mathsf{Sgn}} \newcommand{\LTZ}{\mathsf{LTZ}} \newcommand{\GTZ}{\mathsf{GTZ}} \newcommand{\EQZ}{\mathsf{EQZ}} \newcommand{\LT}{\mathsf{LT}} \newcommand{\GE}{\mathsf{GE}} \newcommand{\EQ}{\mathsf{EQ}} \newcommand{\Scale}{\mathsf{Scale}} \newcommand{\FPAdd}{\mathsf{FP{\mhyphen}Add}} \newcommand{\FPMult}{\mathsf{FP{\mhyphen}Mult}} \newcommand{\ow}{\overline{w}} \newcommand{\ox}{\overline{x}} \newcommand{\oy}{\overline{y}} \newcommand{\oz}{\overline{z}} \newcommand{\open}{\text{open}} \newcommand{\startopen}{\mathtt{startopen}} \newcommand{\stopopen}{\mathtt{stopopen}} \newcommand{\triple}{\mathtt{triple}} \newcommand{\squ}{\mathtt{square}} \newcommand{\bit}{\mathtt{bit}} \newcommand{\inp}{\mathtt{input}} \newcommand{\inv}{\mathtt{inv}} \mathchardef\mhyphen="2D \newcommand{\ind}{\stackrel{c}{\approx}} \newcommand{\C}{\mathbb{C}} \newcommand{\R}{\mathbb{R}} \newcommand{\gc}{\mathfrak{c}} \newcommand{\ct}{\mathfrak{c}} \newcommand{\xt}{\mathfrak{d}} \newcommand{\sk}{\mathfrak{s}} \newcommand{\pk}{\mathfrak{pk}} \newcommand{\epk}{\mathfrak{epk}} \newcommand{\lev}{\ell} \newcommand{\enc}{\mathfrak{enc}} \newcommand{\zero}{\mathfrak{zero}} \newcommand{\va}{\mathbf{a}} \newcommand{\vb}{\mathbf{b}} \newcommand{\vc}{\mathbf{c}} \newcommand{\vct}{\underline{\mathbf{c}}} \newcommand{\vd}{\mathbf{d}} \newcommand{\ve}{\mathbf{e}} \newcommand{\vf}{\mathbf{f}} \newcommand{\vk}{\mathbf{k}} \newcommand{\vm}{\mathbf{m}} \newcommand{\vr}{\mathbf{r}} \newcommand{\vs}{\mathbf{s}} \newcommand{\vt}{\mathbf{t}} \newcommand{\vv}{\mathbf{v}} \newcommand{\vx}{\mathbf{x}} \newcommand{\vy}{\mathbf{y}} \newcommand{\vz}{\mathbf{z}} \newcommand{\norm}[1]{{\Vert #1 \Vert}} \newcommand{\can}{\mathsf{can}} \newcommand{\DDsecp}{\mathsf{DD\_sec}} \newcommand{\Soundsecp}{\mathsf{Sound\_sec}} \newcommand{\ZKsecp}{\mathsf{ZK\_sec}} \newcommand{\Sacsecp}{\mathsf{sacrifice\_sec}} \newcommand{\Macsecp}{\mathsf{macs\_sec}} \newcommand{\Sample}{\mathsf{Sample}} \newcommand{\comm}{\mathsf{comm}} \newcommand{\FRand}{\mathcal{F}_{\mathsf{Rand}}} \newcommand{\KeyGen}{\mathsf{KeyGen}} \newcommand{\Enc}{\mathsf{Enc}} \newcommand{\Dec}{\mathsf{Dec}} \newcommand{\DistDec}{\mathsf{DistDec}} \newcommand{\SwitchModulus}{\mathsf{SwitchModulus}} \newcommand{\SwitchKey}{\mathsf{SwitchKey}} \newcommand{\Seed}{\mathsf{Seed}} \newcommand{\PRF}{\mathsf{PRF}} \newcommand{\cnt}{\mathsf{cnt}} \newcommand{\cond}{\mathsf{cond}} \newcommand{\Rand}{\mathsf{Rand}} \newcommand{\MACCheck}{\mathsf{MACCheck}} \newcommand{\EncCommit}{\mathsf{EncCommit}} \newcommand{\Reshare}{\mathsf{Reshare}} \newcommand{\DataGen}{\mathsf{DataGen}} \newcommand{\DataCheck}{\mathsf{DataCheck}} \newcommand{\MACKeyGen}{\mathsf{MACKeyGen}} \newcommand{\clean}{\mathsf{clean}} \newcommand{\scale}{\mathsf{scale}} \newcommand{\dec}{\mathsf{dec}} \newcommand{\dishonest}{\mathsf{dishonest}} \newcommand{\KS}{\mathsf{KS}} \newcommand{\Var}{\mathsf{Var}} \newcommand{\chall}{\mathsf{chall}} \newcommand{\RLWE}{\mathsf{RLWE}} \newcommand{\dN}{\mathcal{DG}} \newcommand{\HWT}{\mathcal{HWT}} \newcommand{\ZO}{\mathcal{ZO}} \newcommand{\RC}{\mathcal{RC}} \newcommand{\calU}{\mathcal{U}} \newcommand{\cH}{\mathcal{H}} \newcommand{\Comm}{\mathsf{Commit}} \newcommand{\Challenge}{\mathsf{Chal}} \newcommand{\Resp}{\mathsf{Response}} \newcommand{\Verify}{\mathsf{Verify}} \newcommand{\Deliver}{\mathsf{Deliver}} \newcommand{\Abort}{\mathsf{Abort}} \newcommand{\Skip}{\mathsf{Skip}} \newcommand{\Cheat}{\mathsf{Cheat}} \newcommand{\Proceed}{\mathsf{Proceed}} \newcommand{\Pass}{\mathsf{Pass}} \newcommand{\Continue}{\mathsf{Continue}} \newcommand{\CheatSwitch}{\mathsf{CheatSwitch}} \newcommand{\Success}{\mathsf{Success}} \newcommand{\NoSuccess}{\mathsf{NoSuccess}} \newcommand{\Start}{\mathsf{Start}} \newcommand{\Command}{\mathsf{Command}} \newcommand{\DataType}{\mathsf{DataType}} \newcommand{\InputPrep}{\mathsf{InputPrep}} \newcommand{\Triples}{\mathsf{Triples}} \newcommand{\Squares}{\mathsf{Squares}} \newcommand{\Angle}{\mathsf{Angle}} \newcommand{\SEncCommit}{\mathsf{SEncCommit}} \newcommand{\SReshare}{\mathsf{SReshare}} \newcommand{\BreakDown}{\mathsf{BreakDown}} \newcommand{\Decision}{\mathsf{Decision}} \newcommand{\resp}{\mathsf{resp}} \newcommand{\Func}{\mathcal{F}} \newcommand{\Sim}{\mathcal{S}} \newcommand{\Env}{\mathcal{Z}} \newcommand{\Adv}{\mathcal{A}} \newcommand{\FKEYGEN}{\ensuremath{\mathcal{F}_\textsc{KeyGen}}} \newcommand{\PIKEYGEN}{\ensuremath{\Pi_\textsc{KeyGen}}} \newcommand{\SKEYGEN}{\ensuremath{\mathcal{S}_\textsc{KeyGen}}} \newcommand{\FCOMM}{\ensuremath{\mathcal{F}_\textsc{Commit}}} \newcommand{\PICOMM}{\ensuremath{\Pi_\textsc{Commit}}} \newcommand{\SCOMM}{\ensuremath{\mathcal{S}_\textsc{Commit}}} \newcommand{\FENCCOMMIT}{\ensuremath{\mathcal{F}_\textsc{EncCommit}}} \newcommand{\PIENCCOMMIT}{\ensuremath{\Pi_\textsc{EncCommit}}} \newcommand{\SENCCOMMIT}{\ensuremath{\mathcal{S}_\textsc{EncCommit}}} \newcommand{\FCOMMOPEN}{\ensuremath{\mathcal{F}_\textsc{Comm}}} \newcommand{\FMACCHECK}{\ensuremath{\mathcal{F}_\textsc{MACCheck}}} \newcommand{\SMACCHECK}{\ensuremath{\mathcal{S}_\textsc{MACCheck}}} \newcommand{\FPREP}{\ensuremath{\mathcal{F}_\textsc{Prep}}} \newcommand{\FMACKEYGEN}{\ensuremath{\mathcal{F}_\textsc{MACKeyGen}}} \newcommand{\SMACKEYGEN}{\ensuremath{\mathcal{S}_\textsc{MACKeyGen}}} \newcommand{\STUPLEGEN}{\ensuremath{\mathcal{S}_\textsc{TupleGen}}} \newcommand{\STUPLECHECK}{\ensuremath{\mathcal{S}_\textsc{TupleCheck}}} \newcommand{\SPREP}{\ensuremath{\mathcal{S}_\textsc{Prep}}} \newcommand{\FSHE}{\ensuremath{\mathcal{F}_\textsc{SHE}}} \newcommand{\PISHE}{\ensuremath{\Pi_\textsc{SHE}}} \newcommand{\SSHE}{\ensuremath{\mathcal{S}_\textsc{SHE}}} \newcommand{\Online}{\ensuremath{\Pi_\textsc{Online}}} \newcommand{\FONLINE}{\ensuremath{\mathcal{F}_\textsc{Online}}} \newcommand{\SONLINE}{\ensuremath{\mathcal{S}_\textsc{Online}}} \newcommand{\Prep}{\ensuremath{\Pi_\textsc{Prep}}} \newcommand{\Proto}[1]{\Pi_\mathsf{#1}} \newcommand{\Simu}[1]{\mathcal{S}_{\mathsf{#1}}} \newcommand{\Pgpok}{\Proto{gZKPoK}} \newcommand{\Sgpok}{\Simu{gZKPoK}^S} \newcommand{\gZKPoK}{\mathsf{ZKPoK}} \newcommand{\flag}{\mathsf{flag}} \newcommand{\Pred}{\mathsf{P}} \newcommand{\Diag}{\mathsf{Diag}} \newcommand{\ui}{^{(i)}} \newcommand{\uti}{\ui{}^{\intercal}} \newcommand{\transpose}[1]{#1^{\intercal}} \def\tick{\tikz\fill[scale=0.4](0,.35) -- (.25,0) -- (1,.7) -- (.25,.15) -- cycle;} \newcommand{\figlab}[1]{\label{fig:#1}} \newcommand{\figref}[1]{Figure~\ref{fig:#1}} \newcommand{\tablab}[1]{\label{tab:#1}} \newcommand{\tabref}[1]{Table~\ref{tab:#1}} % Environments \newenvironment{boxfig}[2]{% {#1}{#2} = {Caption}{label} \vspace{0.3cm} \begin{figure}[htbp!] \renewcommand{\figurename}{{\bf Figure~}} \newcommand{\FigCaption}{#1} \newcommand{\FigLabel}{#2} \vspace{-0.60cm} \begin{center} \begin{small} \begin{tabular}{@{}|@{~~}l@{~~}|@{}} \hline \rule[-1.5ex]{0pt}{1ex}\begin{minipage}[b]{.96\linewidth} \vspace{1ex} \smallskip \begin{center}\FigCaption\end{center} }{% \end{minipage}\\ \hline \end{tabular} \end{small} \vspace{-0.2cm} \caption{\FigCaption} \figlab{\FigLabel} \end{center} \end{figure} } %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Better Box environment % Usage: \begin{Boxfig}[placement]{Caption}{Label}{Title} \newenvironment{Boxfig}[4][]{ \begin{figure}[thb!] \renewcommand{\figurename}{{\bf Figure~}} \newcommand{\FigCaption}{#2} \newcommand{\FigLabel}{#3} \vspace{-0.25cm} \begin{center} \begin{small} \begin{tabular}{@{}|@{~~}l@{~~}|@{}} \hline \rule[-1.5ex]{0pt}{1ex}\begin{minipage}[b]{.96\linewidth} \vspace{1ex} \smallskip \begin{center} \textbf{#4} \end{center} } {% \end{minipage}\\ \hline \end{tabular} \end{small} \vspace{-0.25cm} \caption{\FigCaption} \label{fig:\FigLabel} \end{center} \end{figure} } \noindent where $ \mathbf{x}_{j} $ is the $ j^{\mathrm{th}} $ column of a matrix $ \mathbf{X} $, $ \bar{x}_{j} $ is the mean of $ \mathbf{x}_{j} $, and $ s_{j} $ is the standard deviation of $ \mathbf{x}_{j} $. pretextual/resumo.tex1-10 % ---------------------------------------------------------------------------- % % Resumo \chapter*{Resumo} \noindent \citacaoautor. \textbf{\titulotrabalho}. \ano. \paginas~f. \tipotrabalho~(\tipoprograma) - Instituto de Matemática e Estatística, Universidade de São Paulo, São Paulo, \ano. \\ Elemento obrigatório, constituído de uma sequência de frases concisas e objetivas, em forma de texto. Deve apresentar os objetivos, métodos empregados, resultados e conclusões. O resumo deve ser redigido em parágrafo único, conter no máximo 500 palavras e ser seguido dos termos representativos do conteúdo do trabalho (palavras-chave). Texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto. Texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto texto. \\ \noindent \textbf{Palavras-chave:} palavra-chave1, palavra-chave2, palavra-chave3. @inproceedings{ercan2015automatic, author = { Stokkink, }, booktitle = {2015 IEEE/ACM 12th Working Conference on Mining Software Repositories}, organization = {IEEE}, pages = {442--445}, title = {Automatic assessments of code explanations: Predicting answering times on stack overflow}, year = {2015} } kachark/FormFlight \hypertarget{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn}{}\section{D\+O\+T\+\_\+assignment.\+dynamics.\+L\+T\+I\+Dyn Class Reference} \label{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn}\index{DOT\_assignment.dynamics.LTIDyn@{DOT\_assignment.dynamics.LTIDyn}} Linear Dynamics. \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item def \mbox{\hyperlink{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_ae9fd78b382de1c3b12fe9b66ef493135}{\+\_\+\+\_\+init\+\_\+\+\_\+}} (self, A, B) \item def \mbox{\hyperlink{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_a0dc3f0dfdbdc223af24ca048858ec02e}{rhs}} (self, t, x, u) \end{DoxyCompactItemize} \subsection*{Public Attributes} \begin{DoxyCompactItemize} \item \mbox{\Hypertarget{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_ae114e0e285737ce44939e97a940c8b07}\label{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_ae114e0e285737ce44939e97a940c8b07}} {\bfseries A} \item \mbox{\Hypertarget{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_a153320385ebcf731bcbc66fa456d0d09}\label{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_a153320385ebcf731bcbc66fa456d0d09}} {\bfseries B} \end{DoxyCompactItemize} \subsection{Detailed Description} Linear Dynamics. \begin{DoxyVerb}Class representing linear time-invariant dynamics \end{DoxyVerb} \subsection{Constructor \& Destructor Documentation} \mbox{\Hypertarget{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_ae9fd78b382de1c3b12fe9b66ef493135}\label{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_ae9fd78b382de1c3b12fe9b66ef493135}} \index{DOT\_assignment.dynamics.LTIDyn@{DOT\_assignment.dynamics.LTIDyn}!\_\_init\_\_@{\_\_init\_\_}} \index{\_\_init\_\_@{\_\_init\_\_}!DOT\_assignment.dynamics.LTIDyn@{DOT\_assignment.dynamics.LTIDyn}} \subsubsection{\texorpdfstring{\_\_init\_\_()}{\_\_init\_\_()}} {\footnotesize\ttfamily def D\+O\+T\+\_\+assignment.\+dynamics.\+L\+T\+I\+Dyn.\+\_\+\+\_\+init\+\_\+\+\_\+ (\begin{DoxyParamCaption}\item[{}]{self, }\item[{}]{A, }\item[{}]{B }\end{DoxyParamCaption})} \begin{DoxyVerb}LTIDyn constructor Input: - A: state matrix - B: control input matrix\end{DoxyVerb} \subsection{Member Function Documentation} \mbox{\Hypertarget{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_a0dc3f0dfdbdc223af24ca048858ec02e}\label{class_d_o_t__assignment_1_1dynamics_1_1_l_t_i_dyn_a0dc3f0dfdbdc223af24ca048858ec02e}} \index{DOT\_assignment.dynamics.LTIDyn@{DOT\_assignment.dynamics.LTIDyn}!rhs@{rhs}} \index{rhs@{rhs}!DOT\_assignment.dynamics.LTIDyn@{DOT\_assignment.dynamics.LTIDyn}} \subsubsection{\texorpdfstring{rhs()}{rhs()}} {\footnotesize\ttfamily def D\+O\+T\+\_\+assignment.\+dynamics.\+L\+T\+I\+Dyn.\+rhs (\begin{DoxyParamCaption}\item[{}]{self, }\item[{}]{t, }\item[{}]{x, }\item[{}]{u }\end{DoxyParamCaption})} \begin{DoxyVerb}Computes the right-hand side of the LTI system Input: - t: time - x: state vector - u: control input vector Output: - \dot{x} = Ax + Bu\end{DoxyVerb} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item D\+O\+T\+\_\+assignment/dynamics.\+py\end{DoxyCompactItemize} cridemichel/mywebsite @article{Piazza2013-2, abstract = {We investigate numerically pseudo-first-order irreversible bimolecular reactions of the type A + B → B between hard spheres undergoing event-driven Brownian dynamics. We study the encounter rate and the survival probability of A particles as functions of the packing fraction φ in the trapping (a single particle diffusing among static non-overlapping traps) and target (many traps diffusing in the presence of a single static target particle) settings, as well as in the case of diffusing traps and particles (full mobility). We show that, since inertial effects are accounted for in our simulation protocol, the standard Smoluchowski theory of coagulation of non-interacting colloids is recovered only at times greater than a characteristic time Δt, marking the transition from the under-damped to the over-damped regime. We show that the survival probability S(t) decays exponentially during this first stage, with a rate 1/τ0 ∝ φ. Furthermore, we work out a simple analytical expression that is able to capture to an excellent extent the numerical results for t < Δt at low and intermediate densities. Moreover, we demonstrate that the time constant of the asymptotic exponential decay of S(t) for diffusing traps and particles is , where kS = 4π(DA + DB)Rρ is the Smoluchowski rate. Detailed analyses of the effective decay exponent β = d [log(-logS(t))]/d (logt) and of the steady-state encounter rate reveal that the full mobility and trapping problem are characterized by very similar kinetics, rather different from the target problem. Our results do not allow one to ascertain whether the prediction S(t) ∝ exp(-at3/2) (a = const.) as t → ∞ for the trapping problem in 3D is indeed recovered. In fact, at high density, S(t) is dominated by short encounter times, which makes it exceedingly hard to record the events corresponding to the exploration of large, trap-free regions. As a consequence, at high densities the steady-state rate simply tends to 1/τ0. Finally, we work out an analytical formula for the rate that shows a remarkable agreement with the numerics up φ = 0.4. © 2013 IOP Publishing Ltd.}, affiliation = {Centre de Biophysique Moléculaire, CNRS-UPR 4301, Université d'Orléans, F-45071 Orléans Cedex, France; Ecole Polytechnique Fédérale de Lausanne, Institute of Theoretical Physics, BSP, CH-1015 Lausanne, Switzerland; Laboratoire de Physique de Solides, UMR 8502, Université Paris Sud, F-91405 Orsay Cedex, France; Dipartimento di Fisica, 'Sapienza' Università di Roma, Piazzale Aldo Moro 2, I-00185 Rome, Italy}, art_number = {245101}, author = {. and . and .}, document_type = {Article}, doi = {10.1088/0953-8984/25/24/245101}, journal = {Journal of Physics Condensed Matter}, note = {cited By 2}, number = {24}, source = {Scopus}, title = {Irreversible bimolecular reactions with inertia: From the trapping to the target setting at finite densities}, url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84878527635&doi=10.1088%2f0953-8984%2f25%2f24%2f245101&partnerID=40&md5=91ab9d77e41c1a584a2ff731d46760d2}, volume = {25}, year = {2013} } @article{DBLP:journals/tog/KangUWS03, author = { and and and }, title = {High dynamic range video}, journal = {ACM Trans. Graph.}, volume = {22}, number = {3}, year = {2003}, pages = {319-325}, ee = {http://doi.acm.org/10.1145/882262.882270}, bibsource = {DBLP, http://dblp.uni-trier.de} } tables/continent_sector_n_studies.tex \begin{tabular}{ll p{1.5cm}} \toprule & & Gridcell sums \\ \midrule South America & D\&A & 3674 \mbox{(2400-5760)} \\ & Other & 1061 \mbox{(605-1995)} \\ North America & D\&A & 21745 \mbox{(14364-31884)} \\ & Other & 8868 \mbox{(5002-15196)} \\ Africa & D\&A & 5323 \mbox{(3391-8104)} \\ & Other & 2251 \mbox{(1220-4105)} \\ Europe & D\&A & 13991 \mbox{(9105-21466)} \\ & Other & 3762 \mbox{(2089-7232)} \\ Asia & D\&A & 20885 \mbox{(14705-29783)} \\ & Other & 6764 \mbox{(3999-11548)} \\ Oceania & D\&A & 5482 \mbox{(3579-8202)} \\ & Other & 1922 \mbox{(1047-3441)} \\ Global & D\&A & 57366 \mbox{(38371-85227)} \\ & Other & 20419 \mbox{(11697-35705)} \\ Without location & D\&A & 0 \mbox{(0-0)} \\ & Other & 23954 \mbox{(13897-42921)} \\ \bottomrule \end{tabular} 10-100 \frac{4 \sqrt{14} \sqrt{\pi} \nu}{315 c^{3}} \left(54 i S_{\lambda} \nu - 18 i S_{\lambda} + 138 S_{n} \nu - 46 S_{n} + 111 i \Sigma_{\lambda} \delta \nu - 18 i \Sigma_{\lambda} \delta + 177 \Sigma_{n} \delta \nu - 46 \Sigma_{n} \delta\right) r^{3}{\left (0 \right )} v^{13}{\left (0 \right )}Robiq/Thesis % Chapter 8 % \chapter{Conclusion} % Main chapter title % \label{Chapter8} % For referencing the chapter elsewhere, use \ref{Chapter8} % REDO WHEN UPDATING This thesis has given an introduction to the current solutions available for file transfers, and reviewed the advantages and pitfalls of these solutions. Afterwards, the basic technology and principles of SendIt were explained. Then the general design and implementations were explained. After which it went through the specifics of each mode. Following, was a review of the experiments done and their contributions to the system, and an evaluation of SendIt and comparable systems. Finally, this chapter will summarize the work done and explore directions for future research. % \section{Summary} % This thesis has introduced a system and a prototype for improving the current situation of file transfers, focusing on e-mail attachments specifically. It did so by finding technology that fit well with the requirements, and by developing an alternative solution. In an effort to return to the original idea behind the Internet as a decentralized system, this thesis suggests a serverless system and shows a prototype of an implementation of such a system. A serverless implementation also follows the logical idea of file transfers as a transaction only between two end-nodes. There is no reason to involve a third party, unless absolutely necessary. In addition, it also reduces the cost of running services, since central servers do not need to be hosted or managed. The disadvantages of a serverless implementation is that there is no central management where one can access information easily, nor a central entity who can co-ordinate communication. % \paragraph{} % Since P2P technology allows us to avoid including a third party, this becomes the natural choice to use for the actual transfer of the files. While the usual way of creating P2P connections is by getting information from a server, or through DHT, SendIt suggests leaving it up to the users, since it will make the system more unpredictable, and as such harder to attack, while also allowing for easier use. Since this creates additional burden on the user, the Assisted Connection Setup mode was also developed. Endpoints send connection setup information to a server over secure WebSockets, which then relays it to the correct endpoint. While this does use a server, the server acts solely as a forwarding agent. Users can either use the default server, or host their own. An example of the default server will be available with the code and installation files for SendIt. This is mainly to cater to the inexperienced users who care more about usability than maximizing security. The opposite is true for the completely serverless version. An added benefit, and another reason for choosing P2P communication, was that it minimizes the attack surface of the application. Restricting the time of which the file, or metadata about the file, is available online greatly reduces the risk of an attacker gaining access to it. The contrast from the proposed system compared to regular solutions is enormous considering the fact that files are usually left on servers even after the transfer has completed, which leaves the information accessible for infinitely longer than with a direct solution. As such, the suggested system is a clear improvement, and has distinct advantages when it comes to reducing the attack surface of file transfers. The drawback of P2P is also the inherent advantage: It is synchronous communication. This means that it is up to developers to find ways to facilitate asynchronous transfers by other means, if such functionality is wanted. % \paragraph{} % When choosing how to implement these features, WebRTC came forward as the logical option, because of its inherent support for many of the desired features. It offers serverless connection setup, creates P2P connections, and also incorporates security functionality. It offers authentication of peers, based on the Offer and Answer, and automatically encrypts the communications channel. It is also easy to deploy because it is a web technology, which means it can be used in an internet browser, making it easy to deploy and install. There are two drawbacks of using WebRTC. The first is that it offers no consistent way to reach or address endpoints, which means the connection setup has to be repeated every time a connection is to be made. The second is that there is no way to consistently identify or authenticate users. For the first issue, there are no easy remedies, the second can be solved by adding an additional identification and authentication scheme. % \paragraph{} % For such schemes, there are many options available. The most used scheme is public-key cryptography, and for good reason. As outlined in \Cref{sec:pkc}, it is a system for encryption and decryption that ensures confidentiality and integrity, while also being easy to manage. By associating each identity with a public key, the system guarantees that only the entity with the corresponding private key can access the information. This protects user privacy, as well as, making sure that it is not possible for attackers to see or manipulate the contents of the data. The downside of public-key cryptography is that it is hard to understand and manage correctly by people who are less technically capable. They have a hard time understanding the concept behind how it works, the difference between the two keys, and how to utilize it in practice. In the proposed system, this is taken care of programmatically and the users do not have to worry about such issues. It is important to use the correct key for the corresponding endpoint, in order for the encryption and decryption scheme to be useful. This is why identity management and trust is necessary. To be able to use public-key cryptography, users need to exchange keys and, only then, can data be encrypted. The exchange of keys is usually done through a trusted third party, in the form of a server. Since SendIt does not have a central server to rely on, the first connection, and with it the exchange of keys, is regarded as a trusted interaction. What this means is that the system assumes the information is correct and not being supplied by a malicious user. Afterwards, the system can guarantee all the benefits of public-key cryptography, but it relies on this first setup being correct. In order to reduce the risk of an attacker exploiting this, a trust system is suggested. The web of trust model allows users to re-evaluate and nuance their trust in endpoints. This gives a collective view of who is trustworthy and who is not, and can also help detect attackers trying to abuse the first trust. While this initial trust is detrimental to the security of the system, changing it would affect usability severely. As such, the system is built upon this first trust, with the web of trust model as a continuous evaluation to detect attackers, and to minimize the effect of potential attackers. It is possible to extend or change this functionality, if one so pleases, when using SendIt as a platform for further development. % \paragraph{} % These features create the basis for a reasonably secure system. In addition, the system is made to be as user friendly as possible. Installation is extremely easy, all that is required is downloading the installer and executing it. There are no options or dialogs during the installation. There is also no requirement to sign up, to do any key management, or any other aspects that may take focus away from the main goal: transferring files. The user only has to care about placing the files in the correct folder and initiating the transfer to the correct destination. In a similar fashion, when receiving files, all one has to do is decide whether to accept or decline such an offer. The functionality is clear and easy to use. The flow of the program is easy to follow, and requires little interaction from the user. Once the transfer is complete, the connection is automatically torn down and the user can start a new transfer. All of these features are there to make sure that people with low technical understanding can still get the benefit of using the system. The difficult, technical parts are automatically set up and taken care of for the user. Advanced users are also kept in mind, as they are allowed access to details about the system and can make changes in the settings window. % \paragraph{} % Comparing SendIt to existing solutions clearly illustrates the advantage of direct communication and the increase this yields in regards to protecting users security and privacy. It gives the user better control over where their data is, and also reduces the chance of an attacker gaining access to the data. Additionally, it requires no setup on the user side, and can be used directly after installation has finished. While other solutions have encryption and authentication, the systems used cannot be inspected by the user, and they can easily be tricked by false E2E encryption. This is not the case for SendIt as it can be easily inspected and managed. One can also easily switch identities, if desired, without having to create a new account or go through additional steps. This is in stark contrast to the comparative solutions. % \paragraph{} % The contributions of this thesis can be summed up as a user-friendly system that allows for direct, secure file transfers. SendIt can also serve as a platform to expand for a wide range of functionality and services. SendIt, as a platform, would likely be used for it's identity management and authentication functionality, not the file transfer functionality. While this is one potential use case, another can be adding more functionality to the file transfer aspect. This would allow it to become more of a complete solution that could function as a stand alone system, instead of a complimentary one. It also contributes to exploring new implementations of e-mail attachments. The serverless functionality also opens for a new way of software developing, and contributes and encourages a break with the current server-based development. Finally, it shows that secure solutions do not necessarily have to be hard to use and incentivizes development of more user-friendly security applications. % \paragraph{} % \textbf{In conclusion:} This thesis introduced a system that improves the current implementation of e-mail attachments. It does so by directly transferring files, lowering the attack surface significantly. It also guarantees end-to-end encryption and endpoint authentication (\emph{based on the key exchange done as part of the first connection}), which is not commonly used for e-mail attachments, making it resistant to both active and passive attacks. It achieves this while keeping the system easy to use, by taking care of key management and authentication on behalf of the user, without the need for any setup or central corroboration. SendIt outperforms e-mail attachments in regards to security, as well as usability, and cost efficiency. The drawbacks of SendIt are the limitation of only supporting synchronous transfers, as well as, it's inherent trust in the first connection. % \section{Future work} % There are still many venues and ways of improvement for systems such as SendIt, and e-mail attachments in general. By separating future work into two sections, it is easier to get a clear view of which improvements are necessary for SendIt, and what changes are needed in regards to e-mail attachments in general. SendIt has the core features and design developed, but since it has been a one person job, the depth and complexity has been limited. With more time or more manpower, many of the issues currently existing can be fixed, and a complete alternative to the current e-mail attachment functionality can be finalized. As such, the future research for SendIt focuses more on implementations and testing of the system. For e-mail attachments in general, a more holistic approach is taken in regards to future work. Things like large scale evaluation and system design become more important and need to be prioritized more than when designing a simpler system. % \subsection{SendIt} % The following topics can be future research for the development for SendIt: \begin{itemize} \item \textbf{Developing and testing extendability of the connection setup for the Serverless mode.} Adding to the lifetime of the connection setup would greatly increase the usability and reliability of this mode. Adding some kind of connection-manager or middleware was considered as part of this research, but could not be done in time. % \item \textbf{Improving reliability of connections and connection setup.} In conjunction with the point above, increasing the success rate of the connection setup would make the system more reliable and easier to use. Making sure connections are stable and do not disconnect will help the usability and stability of the system. % \item \textbf{Exploring more ways to evaluate trust.} Testing, implementing, and comparing different systems in practice, as well as finding an algorithm for evaluation. This work would add great value to the trust management of SendIt. % \item \textbf{Improve the initial trust.} Finding a way to increase the trust in the first connection, or some way of authentication that does not rely on a central service, would be a big improvement. In addition, allowing for sharing keys across applications may also be a way to bridge this initial trust issue. Such problems are left up to future research. % \item \textbf{Adding support for pausing transfers.} Having a way to pause transfers should be easy to implement, and will benefit the system greatly. Especially for big transfers, this will allow users to take breaks temporarily, if needed. % \item \textbf{Checking overhead of the file transfer and optimizing the protocol.} Optimizing the speed of the file transfer, as well as, minimizing the overhead of the protocol, will ensure that the communication is as effective as possible. This would allow for shorter transfer-times, less data transferred, and higher efficiency. % %\item Finalize protocol and ACS server. Completing this work will add extra trust to the system and help make the key exchange go smoothly. While this is not critical to the system in any way, it would make for an increase in security and improve the trust in an identity. \end{itemize} % \subsection{E-mail attachments} % The following topics can be future research for e-mail attachments and how to improve the current system: \begin{itemize} \item \textbf{Finding a better way to asynchronously transfer files.} Any improvement to the way e-mail attachments currently work would be a welcome contribution, but particularly an improvement to user privacy is necessary. Avoiding server-storage would be optimal, but finding a way to ensure that the file does not reside on the server after the message has been transferred would be an acceptable compromise. % \item \textbf{Exploring other systems and alternatives to SendIt.} Evaluating their usability and functionality. By doing this, one can clearly indicate which solution is optimal, and recommend a new standard which can be adopted by e-mail clients. % \item \textbf{Creating alternative solutions to complement SendIt.} As mentioned in the introduction (\Cref{Chapter1}), replacing the current email attachment system with SendIt alone is not recommended. Developing a complementary system to SendIt, which in combination can be implemented to completely replace the current system, would go a long way towards improving the current situation. \end{itemize} % \section{Final remarks} % As a final note, I would like to add that the source code for both SendIt and the ACS server will be made available \href{https://github.com/Robiq}{on github}\footnote{\href{https://github.com/Robiq}{https://github.com/Robiq}} once it has been cleared for release. I hope it will be developed further, and that it can succeed as an open source application that provides improved security and privacy for everyone.\documentclass[12pt]{article} \usepackage[utf8]{inputenc} \usepackage[english,russian]{babel} \usepackage{hyperref} \usepackage{microtype} \hypersetup{colorlinks=true,urlcolor=blue} \setlength{\parindent}{0pt} \usepackage[absolute]{textpos} \TPGrid{16}{16} \usepackage[top=2\TPVertModule, bottom=2\TPVertModule, left=3\TPHorizModule, right=2\TPHorizModule]{geometry} \usepackage{fancyhdr} \pagestyle{fancy} \renewcommand{\headrulewidth}{0pt} \fancyhf{} \rhead{} \rfoot{\small Page~\thepage} \begin{document} \section*{Ар} \href{mailto:}{} / \href{https://t.me/artemiy312}{telegram.me/artemiy312} \newline \href{https://stackoverflow.com/users/6800156}{StackOverflow} / \href{https://github.com/artemiy312}{GitHub} / \href{https://gitlab.com/artemiy312}{GitLab} \medskip\textbf{Бэкэнд разработчик}. Работаю с Python и JavaScript. \newline Иногда пишу на Go и Rust. Катаюсь на лыжах, занимаюсь живописью, делаю игры. Россия, Санкт-Петербург. \subsection*{Кратко} \begin{itemize} \item 3+ года в сложных проектах с Python/JavaScript \item Опыт с PostgreSQL, Redis, RabbitMQ, Docker, Swarm, Heroku, Unix \item Практикую TDD, CI/CD, короткие итерации, рефакторинг \end{itemize} \subsection*{Проекты} \textbf{2019}. \href{https://app.fama.family/}{Fama.family}. EdTech платформа о финансах. Разрабатывал REST API, админку и SPA. Делал синтез речи и плеер для текстовых уроков, интеграцию уроков с чат-ботом для ввода данных, индексацию источников с финансовыми статьями и полнотекстовый поиск. Falcon, Django, Vue.js, PostgreSQL, Redis, Apache Solr, Centrifugo, Docker. \medskip \textbf{2018}. \href{https://github.com/fidals/cryptotrader}{Cryptotrader}. Торговый-бот на криптовалютных биржах. Разрабатывал клиентский API для четырех бирж на Python с asyncio, торговый алгоритм арбитража, отчеты о сделках в Telegram, веб-панель состояния бота. Aiohttp, Pytest, TimescaleDB, Redis, Docker, Drone CI, Sentry. \medskip \textbf{2017}. \href{https://start.ru/}{Start.ru}. Видео-стриминговая платформа. Разрабатывал сервисы конфигурации стриминга, транскодинга медиафайлов и админку. Делал полнотекстовый поиск, репликацию данных из MongoDB в ElasticSearch. Aiohttp, Flask, Pytest, RabbitMQ, Docker, Datadog, GitLab CI. \medskip \textbf{2016}. \href{https://github.com/fidals/shopelectro}{Shopelectro.ru} \href{https://github.com/fidals/stroyprombeton}{Stroyprombeton.ru}. E-commerce проекты. Разрабатывал фротэнд и бэкэнд на общей \href{https://github.com/fidals/refarm-site}{платформе для магазинов}. Делал импорт базы товаров из 1С, генерацию Excel прайс-листа, полнотекстовый поиск, табличный редактор товаров, генерацию текстов по фильтрам. Django, Celery, jQuery, Selenium, PostgreSQL, Redis, RabbitMQ, Docker, Drone CI. \subsection*{Карьера} Команда \href{https://github.com/fidals}{Фидалс}. Июнь 2016 - текущий момент. Разрабатывали Shopelectro, Stroyprombeton, Start.ru и Cryptotrader. \end{document} \hypertarget{test_2CMakeLists_8txt}{}\doxysection{test/\+C\+Make\+Lists.txt File Reference} \label{test_2CMakeLists_8txt}\index{test/CMakeLists.txt@{test/CMakeLists.txt}} \doxysubsection*{Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{test_2CMakeLists_8txt_a8b5d9d230fe35c22ce1d6e90014082a9}{set}} (G\+T\+E\+S\+T\+\_\+\+S\+H\+U\+F\+F\+LE 1) \mbox{\hyperlink{app_2CMakeLists_8txt_ab3d3ab91616d93113cb589d83b42a377}{add\+\_\+executable}}(cpp-\/test main.\+cpp camtest.\+cpp cascadetest.\+cpp rectplottest.\+cpp \$ \end{DoxyCompactItemize} \doxysubsection{Function Documentation} \mbox{\Hypertarget{test_2CMakeLists_8txt_a8b5d9d230fe35c22ce1d6e90014082a9}\label{test_2CMakeLists_8txt_a8b5d9d230fe35c22ce1d6e90014082a9}} \index{CMakeLists.txt@{CMakeLists.txt}!set@{set}} \index{set@{set}!CMakeLists.txt@{CMakeLists.txt}} \doxysubsubsection{\texorpdfstring{set()}{set()}} {\footnotesize\ttfamily set (\begin{DoxyParamCaption}\item[{G\+T\+E\+S\+T\+\_\+\+S\+H\+U\+F\+F\+LE}]{1 }\end{DoxyParamCaption})} Definition at line 1 of file C\+Make\+Lists.\+txt. \begin{DoxyCode}{0} \DoxyCodeLine{9 \{CMAKE\_SOURCE\_DIR\}/app/cam.cpp } \end{DoxyCode} 10-100 %% %% pytorch-neural-doodle/docs/content/chapters/implementation.tex %% %% Created by <> on 21/08/2018. %% Updated by <> on 07/10/2018. %% \section{Implementation} \label{section:implementation} \textbf{by } \\ As a starting point, Champandard examines user behavior while authoring style transfers. This is made possible through a social media bot \cite{deepforger2015}, making artistic style transfer algorithms available to a general public. It can be observed, that in the cases where the algorithm does not meet the users expectation, it is often due to a lack of semantic segmentation of style and content images. As an example, when transfering an artists style to a photographic portrait, one would expect the algorithm to transfer the color and texture of skintones the artist chose to the skin areas of the portrait. While this expectation may at times be met, the patch based style loss constructed above does not generally enforce such a behavior. The usage of semantic segmentation is especially prominent in more specialized approaches to style transfer \cite{yang2017semantic}, suggesting additional merit to the presented intuition. Note, that convolutional neural networks such as VGG do implicitly learn semantic segmentation of images \cite{thoma2016survey}, but this segmentation is not put to use in the above style loss construction because nearest neighbour patches are selected only with respect to similarity in texture. A segment of sky in the background of a style painting may absolutely be selected as nearest neighbour patch for a skintone area in the content picture, if local texture happens to be similar. This lack of semantic segmentation causes glitches and subverts the intention of the user but the disregard for the semantic segmentation extracted by the convolutional neural network is in some respects a byproduct of the design -- the style loss intentionally uses low layer responses to capture local texture information and to not capture higher level (content) responses such as semantic information. This leaves the user with a very limited number of control levers. Choosing the parameter \(\alpha\) which weights style loss against content loss presents a spectrum between a faithful reproduction of the content image and an unstructured reproduction of style texture with no regard to the content. This is insufficient in practice, especially for abstract styles and in those -- particularly interesting -- cases where style and content image are very dissimilar in perspective or subject. The core idea of Champandard lies in incorporating segmentation from a semantic map both indirectly as part of the nearest neighbour patch computation and directly into the style loss term. \subsection{Semantic maps} \textbf{by } \\ A semantic map is an image dispaying a semantic segmentation of another image. It has the same aspect ratio as the segmented image and encodes semantically similar sections as areas with the same color. This information can be provided by a user as an artistic control lever of the process. In fact, as \cite{doodles2016} points out, it is possible to go one step further, encoding not only the semantic affilliation of pixels in a given area with each other, but also the relationship of different semantic areas. This is done, by choosing similar colors for the segmentation of semantically similar (but not equal) areas. When working with an image of a landscape, one may choose to segment different variations of woodland appearance with different shades of one color, say green. The process of authoring semantic maps is technically easy, as will be described in the next section, but it can be artistically challenging. The latter can also be seen as opportunity: additional leeway in the artistic shaping of images presents the possibility of more appealing results. This fact has been noted by \cite{doodles2016} and is clearly corroborated through our experiments. \clearpage \subsection{Style Loss} \textbf{by } \\ The semantic map of the style image is processed by the VGG network and activations for specific layers are gathered. In the case of a VGG 19 network, we choose the layers \texttt{conv3\_1} and \texttt{conv4\_1}. Denote the response tensor with \(m^s\). This tensor has two spatial dimensions and one channel dimension. The vector \(m^s\) is subsequently scaled by a factor \(\gamma\) and concatenated along the channel dimension with the response vector \(y^s\) of the style image \[x^s = y \,||\,\gamma m \] The analogous preparation steps are applied to the content image \[x^c = y^c \,||\,\gamma m^c \] For the reasons outlined in the introduction, a patch-based style loss approach is prefered over a Gram-based approach. We therefore compute \(k\times k\) patches \(\Psi (x^s)\) and \(\Psi (x^c)\) along the spatial dimensions. Our implementation specifically uses \(3\times 3\) patches. Given \(x^s\) and \(x^c\), the patch-based approach of \cite{mrf2016} can be applied without the need for significant changes. Nearest neighbour patches are computed with respect to normalized cross-correlation and we have the finished patch-based style loss \[\mathcal{L}_{s,p}(x^s,x^c) = \sum_i \|\Psi_i(x^c)-\Psi_{\text{NN}(i)}(x^s)\|_2^2\] The parameter \(\gamma\) can be seen as a user control point, specifying how heavily the segmentation should be weighted against local pixel conformance during nearest neighbour computation. This specific way of incorporating segmentation maps has two principal upsides: \begin{enumerate} \item The segmentation maps can easily be authored, as they do not need to conform to many formal specifications. They can have an arbitrary number of channels and can be drawn with a reduced resolution as compared to the original style- and content image. These are both conveniences in practical use. % embedding segmentation \item The algorithm being used is essentially the same as the patch based approach of \cite{mrf2016} because the datastructure being worked on does only change along the channel dimension and the operations being applied remain the same. In the common library implementation of respective convolutions, one does not need to change anything. The only extra implementation is the weighted concatenation of segmentation maps. \end{enumerate} Manual work can be saved by using a lower resolution for the segmentations and computational effort as well as memory can be saved by increasing the stride between successive patches. Applied moderately, neither of these simplifications have a large impact on practical results. \subsection{Content Loss} \textbf{by } \\ Content loss can be added without additional modifications to the original approach of Gatys et al. \cite{gatys2015neural}. With regard to semantic maps, it merits mentioning that much like the original implementation of \cite{doodles2016}, our present implementation can be run in a mode that requires a content segmentation but does not require a content image. In this mode, texture is merely semantically transfered from the style image to the content image. Content loss ist omitted, the total loss being minimized is exactly the style loss. Figure \ref{fig::nocontent} shows results for this mode. This is what the title of the present work refers to as a \emph{Neural Doodle}. \begin{figure} \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/renoir.png} \end{subfigure}%\hspace{5mm} ~ \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/out_result.png} \end{subfigure}%\hspace{5mm} ~ \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/out2_result.png} \end{subfigure} \\ \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/renoir_style.png} \end{subfigure}%\hspace{5mm} ~ \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/renoir_out_style.png} \end{subfigure}%\hspace{5mm} ~ \begin{subfigure}[t]{0.32\textwidth} \centering \includegraphics[width=4.5cm]{renoir/renoir_out_style_2.png} \end{subfigure} \caption[]{Reproduction of the paper results generated by our implementation. Original style image (Renoir) with respective segmentation in the left column. Two different segmentation maps (lower row) and respective transfer results (upper row). No content image was used and the content loss omitted.} \label{fig::nocontent} \end{figure}docs/Zmat_appendix.tex \chapter{Specifying your geometry with a Z-Matrix} A Z-Matrix is a convenient way to specify the geometry of a molecule or crystal in terms of bond lengths, bond angles, and dihedral angles. There are several styles of Z-matrix used in various programs, \calcprog\ uses a format similar to that used in Gaussian. This is intended to be a brief introduction to how a Z-matrix works. If you already know how to use a Z-matrix, here's all you need to know about the implementation in \calcprog: \begin{itemize} \item The first atom is put at the origin. \item The second atom is put along the Z axis. \item The third atom is in the XZ plane. \item Dihedrals are evaluated using the right hand rule. \end{itemize} The easiest way to explain a Z-matrix is to show one and then explain it, so that's what we'll do. Before we start, however, we need to briefly define a dihedral angle. A dihedral is specified by 4 atoms, we'll call them A, B, C, and D. The dihedral A--B--C--D is the angle between the plane defined by A--B--C and the plane defined by B--C--D. Here's an alternative explanation: the dihedral A--B--C--D is the angle between the lines C--D and B--A if you are looking down the line C--B. There's one more piece of information we need to fully understand the dihedral: there is a handedness associate with them. If you think about it, looking down the line C--B there are two different angles between lines C--D and B--A: $\theta$ and 360-$\theta$. The dihedrals in \calcprog\ are defined using the right hand rule: Take your right hand and point the thumb down the line C--B, now align your fingers with the line C--D, curling your fingers shows the direction in which the dihedral angle is measured. % This is all about a million times easier to understand using a picture, here's a picture demonstrating both views of dihedrals and their handedness. % \begin{center} \epsfig{file=dihedral.eps,width=5.0in} \end{center} % With that definition under our belt, here's the Geometry specification for a square pyramidal (CH$_3$)BiI$_4$ fragment, where the CH$_3$ group is along the Z axis and the Bi and four I's lie in the XY plane: \shrinkspacing \begin{verbatim} Geometry Z Matrix 9 1 Bi 2 C 1 2.1 3 I 1 2.7 2 90.0 4 I 1 2.7 2 90.0 3 90.0 5 I 1 2.7 2 90.0 3 180.0 6 I 1 2.7 2 90.0 3 270.0 7 H 2 1.1 1 109.5 3 0.0 8 H 2 1.1 1 109.5 3 120.0 9 H 2 1.1 1 109.5 3 240.0 \end{verbatim} \resumespacing Let's look at the first few entries in more detail. \begin{enumerate} \item The first atom is a Bi and it's placed at the origin. Cartesian: (0 0 0). \item Atom two is a C. It's placed on the Z axis, 2.1 \AA\ away from atom 1. Cartesian: (0 0 2.1). \item Atom three is an I. It's placed 2.7 \AA\ away from atom 1 and the angle between atoms 3--1--2 in the XZ plane is 90.0 degrees. Cartesian: (2.7 0 0). \item Atom four is an I. It's placed 2.7 \AA\ away from atom 1, the angle 4--1--2 is 90 degrees. This angle puts us in the XY plane. At this point we know that atom 4 lies on a circle in the XY plane with radius 2.7 \AA. The dihedral 4--1--2--3 (90 degrees) tells us where on the circle we are. This dihedral is particularly easy to see: if we look down the bond 2--1 (which is looking down the Z axis), the angle between the bond 3--1 and the bond 4--1 is 90 degrees. So atom 4 lies on the Y axis. Taking the right-handedness of dihedrals into account, we know that atom 4 lies on the negative Y axis. Cartesian (0 -2.7 0). \item Atom five is an I. It's 2.7 \AA\ away from atom 1, making an angle of 90 degrees with 2 and a dihedral of 180 with 3. This puts us on the negative X axis. Cartesian (-2.7 0 0). \end{enumerate} If you find the handedness of dihedrals confusing, just play around with a couple of molecules defined using Z matrices, you'll get the hang of it fairly quickly. 1-10 \begin{frame}{Анализ свойств меры Хартли} \noindent Экспериментатор одновременно подбрасывает монету (М) и кидает игральную кость (К). Какое количество информации содержится в эксперименте (Э)?\\ \vspace{1.5em} \color[rgb]{0,0.5,0.0}\noindent\textbf{Аддитивность:} \color{black}$i(\mbox{Э})=i(M)+i(K)=>i(\mbox{12 исходов})=i(\mbox{2 исхода})+i(\mbox{6 исходов}):\ \log_x12=\log_x2+\log_x6$ \color[rgb]{0,0.5,0.0}\noindent\textbf{Неотрицательность:} \color{black}Функция $log_xN$ неотрицательно при любом $x>1$ и $N\geq1$ \color[rgb]{0,0.5,0.0}\noindent\textbf{Монотонность:} \color{black}С увеличением $p(M)$ или $p(K)$ функция $i(\mbox{Э})$ монотонно возрастает. \color[rgb]{0,0.5,0.0}\noindent\textbf{Принцип неопределённости:} \color{black}При наличии всегда только одного исхода (монета и кость с магнитом) количество информации равно нулю: $\log_x1+log_x1=0$ \end{frame}bellicapelli0/FDSAssignement10 @misc{sentiment_vader, title={Simple Sentiment Analysis for NLP Beginners and Everyone Else using VADER and TextBlob}, url={https://medium.com/swlh/simple-sentiment-analysis-for-nlp-beginners-and-everyone-else-using-vader-and-textblob-728da3dbe33d}, journal={Medium}, author={}, year=2020, } @article{roder_2015, title={Exploring the Space of Topic Coherence Measures}, DOI={10.1145/2684822.2685324}, journal={WSDM}, author={ and Both, Andreas and }, year=2015, pages={399–408} } @misc{sentiment_vader, title={Simple Sentiment Analysis for NLP Beginners and Everyone Else using VADER and TextBlob}, url={https://365datascience.com/pca-k-means/}, journal={Medium}, author={}, year={2020}, } @misc{degrave_2016, title={A Naive Bayes Tweet Classifier}, url={https://www.kaggle.com/degravek/a-naive-bayes-tweet-classifier}, journal={Kaggle}, author={}, year={2016}, } @article{roder_2015, title={Exploring the Space of Topic Coherence Measures}, author={ and Both, Andreas and }, year={2015}, DOI={10.1145/2684822.2685324}, pages={399–408}, series={WSDM '15}, } @article{go_2009, title={Twitter Sentiment Classification using Distant Supervision}, author={, }, year=2009, month=01, volume=150, journal={Processing} }@InProceedings{xua15, supplementary = {Supplementary:xua15-supp.pdf}, title = {CUR Algorithm for Partially Observed Matrices}, author = { and and }, pages = {1412-1421}, abstract = {CUR matrix decomposition computes the low rank approximation of a given matrix by using the actual rows and columns of the matrix. It has been a very useful tool for handling large matrices. One limitation with the existing algorithms for CUR matrix decomposition is that they cannot deal with entries in a {\it partially observed} matrix, while incomplete matrices are found in many real world applications. In this work, we alleviate this limitation by developing a CUR decomposition algorithm for partially observed matrices. In particular, the proposed algorithm computes the low rank approximation of the target matrix based on (i) the randomly sampled rows and columns, and (ii) a subset of observed entries that are randomly sampled from the matrix. Our analysis shows the relative error bound, measured by spectral norm, for the proposed algorithm when the target matrix is of full rank. We also show that only $O(n r\ln r)$ observed entries are needed by the proposed algorithm to perfectly recover a rank $r$ matrix of size $n\times n$, which improves the sample complexity of the existing algorithms for matrix completion. Empirical studies on both synthetic and real-world datasets verify our theoretical claims and demonstrate the effectiveness of the proposed algorithm.}, } sjgknight/starter-hugo-research-group @article{kearneyProspectiveScienceTeachers2006, author = {}, citation = {https://scholar.google.com/citations?view\textsubscriptop=view\textsubscriptcitation&hl=en&user=kR2bHbkAAAAJ&pagesize=100&citation\textsubscriptfor\textsubscriptview=kR2bHbkAAAAJ:LkGwnXOMwfcC}, doi = {10.14742/ajet.1300}, journal = {Australasian Journal of Educational Technology}, number = {2}, title = {Prospective Science Teachers as E-Learning Designers}, type = {Journal Article}, volume = {22}, year = {2006} } \documentclass[../psets.tex]{subfiles} \pagestyle{main} \renewcommand{\leftmark}{Problem Set 3} \begin{document} \begin{enumerate}[label={\Roman*)}] \item \marginnote{2/4:}Ammonia undergoes a facile inversion ("umbrella flip") as shown below. The activation barrier for inversion is low ($\Delta G^\ddagger\sim\SI{5}{kcal\per\mole}$), and the transition state for this motion is planar \ce{NH3}. Note that the relevant valence shell IP's are $\ce{N}_{2s}=\SI{-26.0}{eV}$, $\ce{N}_{2p}=\SI{-13.4}{eV}$, and $\ce{H}_{1s}=\SI{-13.6}{eV}$. \begin{center} \schemestart \chemfig{\charge{90=\:}{N}(-[:-30]H)(>:[:-150]H)(<[:-110]H)}\arrow{<=>} \chemleft{[} \chemfig{\charge{180=\:}{N}(-H)(>:[:150]H)(<[:-150]H)} \chemright{]^{\ddagger}}\arrow{<=>} \chemfig{\charge{-90=\:}{N}(-[:30]H)(>:[:150]H)(<[:110]H)} \schemestop \end{center} \begin{enumerate}[label={\alph*)}] \item Construct an MO diagram for \emph{planar} \ce{NH3}. \begin{proof}[Answer] Point group: $D_{3h}$\par Basis functions: all three \ce{H} orbitals, \ce{N_{$2s$}}, \ce{N_{$2p_x$}}, \ce{N_{$2p_y$}}, and \ce{N_{$2p_z$}}.\par Apply operations, generate reducible representations, and reduce to irreducible representations: \begin{align*} \Gamma_{\ce{H}} &= (3,0,1,3,0,1) = A_1'+E'\\ \Gamma_{\ce{N_{$2s$}}} &= A_1'\\ \Gamma_{\ce{N_{$2p_x$}}} &= E'\\ \Gamma_{\ce{N_{$2p_y$}}} &= E'\\ \Gamma_{\ce{N_{$2p_z$}}} &= A_2'' \end{align*} Combine central and peripheral orbitals by their symmetry: \begin{figure}[H] \centering \begin{tikzpicture}[ yscale=0.3, every node/.prefix style={black} ] \footnotesize \draw [ultra thick,gry] (2,-26.0) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$s(A_1')$} ++(0.5,0); \draw [ultra thick,gry] (2,-13.4) -- node{\Large$\upharpoonleft$} node[below=2mm]{$p_x(E')$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=7mm]{$p_y(E')$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$p_z(A_2'')$} ++(0.5,0) ; \draw [ultra thick,gry] (-3.7,-13.6) -- node{\Large$\upharpoonleft$} node[below=2mm]{$A_1'$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$E'$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$E'$} ++(0.5,0) ; \draw [ultra thick] (-0.55,-28) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$2a_1'$} ++(1.1,0) (-0.55,-23) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.5,0) node[below=2mm,xshift=0.05cm]{$1e'$} ++(0.1,0) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.5,0) (-0.55,-13.4) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$1a_2''$} ++(1.1,0) (-0.55,-10) -- node[below]{$3a_1'$} ++(1.1,0) (-0.55,-4) -- ++(0.5,0) node[below,xshift=0.05cm]{$2e'$} ++(0.1,0) -- ++(0.5,0) ; \draw [grx,densely dashed] (-3.2,-13.6) -- (-0.55,-28) (-3.2,-13.6) -- (-0.55,-10) (-2,-13.6) -- (-0.55,-23) (-2,-13.6) -- (-0.55,-4) (2,-13.4) -- (0.55,-4) (2,-13.4) -- (0.55,-23) (2,-13.4) -- (0.55,-13.4) (2,-26.0) -- (0.55,-10) (2,-26.0) -- (0.55,-28) ; \small \node [label={[yshift=1mm]left:\footnotesize$3\e[-]$}] at (-2.85,-31) {$3\times\ce{H}$}; \node at (0,-31) {\ce{NH3}}; \node [label={[yshift=1mm]right:\footnotesize$5\e[-]$}] at (2.85,-31) {\ce{N}}; \end{tikzpicture} \caption{Planar ${\ce{NH3}}^\ddagger$ orbital diagram.} \label{fig:orbitalDiagram-NH3-planar} \end{figure} \end{proof} \item Label the MOs with the appropriate Mulliken symbols ($a_{1g}$, $e_g$, etc.) and add electrons to show the proper orbital occupancies. \begin{proof}[Answer] See Figure \ref{fig:orbitalDiagram-NH3-planar}. \end{proof} \item Compare your MO diagram with that for pyramidal \ce{NH3} (Figure 5.30 in your text), and comment qualitatively on why this process is a low-energy one. \begin{proof}[Answer] It appears that the only change between the two MO diagrams is that the two $3a_1$ electrons in the pyramidal \ce{NH3} diagram must be excited to the $1a_2''$ orbital in the planar \ce{NH3} diagram. Since $1a_2''$ is higher in energy than $3a_1$, there will be an increase in energy, but since it is only marginally higher, the increase will be very small. \end{proof} \item What vibrational mode is responsible for the inversion? \begin{proof}[Answer] If any vibrational mode is responsible for the inversion, it certainly won't be a stretching mode since these have no effect on molecular geometry about the central atom. On the other hand, a bending mode could well achieve such a transition. Thus, we will find the bending modes in both pyramidal and planar \ce{NH3} and compare.\par\medskip For pyramidal \ce{NH3}, we can determine that $\Gamma_{x,y,z}=(3,0,1)$. We can also figure out that the number of atoms unmoved after applying each symmetry operation is $(4,1,2)$. Thus, $\Gamma_{3N}=(12,0,2)$. We can decompose this by inspection to $\Gamma_{3N}=3A_1+A_2+4E$. Since $\Gamma_\text{trans}=A_1+E$ and $\Gamma_\text{rot}=A_2+E$, we have by subtraction that $\Gamma_\text{vibs}=2A_1+2E$.\par We can determine that $\Gamma_\nu=(3,0,1)$ by counting how many $\overrightarrow{\ce{N-H}}$ vectors stay the same under each symmetry operation. We can decompose this by inspection to $\Gamma_\nu=A_1+E$. Thus, we have by subtraction that $\Gamma_\delta=A_1+E$.\par\smallskip For planar \ce{NH3}, we can determine that $\Gamma_{x,y,z}=(3,0,-1,1,-2,1)$. We can also figure out that the number of atoms unmoved after applying each symmetry operation is $(4,1,2,4,1,2)$. Thus, $\Gamma_{3N}=(12,0,-2,4,-2,2)$. We can decompose this by repeated applications of the reduction formula to $\Gamma_{3N}=A_1'+A_2'+3E'+2A_2''+E''$. Since $\Gamma_\text{trans}=E'+A_2''$ and $\Gamma_\text{rot}=A_2'+E''$, we have by subtraction that $\Gamma_\text{vibs}=A_1'+2E'+A_2''$.\par We can determine that $\Gamma_\nu=(3,0,1,3,0,1)$ by counting how many $\overrightarrow{\ce{N-H}}$ vectors stay the same under each symmetry operation. We can decompose this by inspection to $\Gamma_\nu=A_1'+E'$. Thus, we have by subtraction that $\Gamma_\delta=E'+A_2''$.\par\medskip Since the $E$ pyramidal bending modes transform into the analogous $E'$ planar bending modes, but the $A_1$ pyramidal bending mode has no planar analogue, it is the $A_1$ bending mode in pyramidal \ce{NH3} that causes the inversion. \end{proof} \end{enumerate} \newpage \item ${\color{white}x}$ \begin{enumerate}[label={\alph*)}] \item Use group theory to construct an MO diagram for octahedral \ce{SF6}. Consider only $\sigma$-bonding between \ce{S} and the \ce{F}'s and use only the sulfur $3s$ and $3p$ valence orbitals (i.e., ignore the $3d$-orbital involvement). For fluorine, just use a "$\sigma$-type" orbital to determine the $6\times\ce{F}$ group orbitals. \begin{proof}[Answer] Point group: $O_h$\par Basis functions: all six \ce{F} orbitals, \ce{S_{$3s$}}, \ce{S_{$3p_x$}}, \ce{S_{$3p_y$}}, and \ce{S_{$3p_z$}}.\par Apply operations, generate reducible representations, and reduce to irreducible representations: \begin{align*} \Gamma_{\ce{F}} &= (6,0,0,2,2,0,0,0,4,2) = A_{1g}+E_g+T_{1u}\\ \Gamma_{\ce{S_{$3s$}}} &= A_{1g}\\ \Gamma_{\ce{S_{$3p_x$}}} &= T_{1u}\\ \Gamma_{\ce{S_{$3p_y$}}} &= T_{1u}\\ \Gamma_{\ce{S_{$3p_z$}}} &= T_{1u} \end{align*} Combine central and peripheral orbitals by their symmetry: \begin{figure}[h!] \centering \begin{tikzpicture}[ yscale=0.15, every node/.prefix style={black} ] \footnotesize \draw [ultra thick,gry] (2,-22.71) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$s(A_{1g})$} ++(0.5,0); \draw [ultra thick,gry] (2,-11.62) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$p_x(T_{1u})$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=7mm]{$p_y(T_{1u})$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$p_z(T_{1u})$} ++(0.5,0) ; \draw [ultra thick,gry] (-5.5,-40.17) -- node{\Large$\upharpoonleft$} node[below=2mm]{$E_g$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$E_g$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$A_{1g}$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$T_{1u}$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$T_{1u}$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$} node[below=2mm]{$T_{1u}$} ++(0.5,0) ; \draw [ultra thick] (-0.85,-60) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} node[below=2mm]{$a_{1g}$} ++(1.7,0) (-0.85,-52) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.5,0) ++(0.1,0) -- node[below=2mm]{$t_{1u}$} node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.5,0) ++(0.1,0) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.5,0) (-0.85,-40.17) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.8,0) node[below=2mm,xshift=0.05cm]{$e_g$} ++(0.1,0) -- node{\Large$\upharpoonleft$\hspace{-1mm}$\downharpoonright$} ++(0.8,0) (-0.85,-6) -- ++(0.5,0) ++(0.1,0) -- node[below]{$t_{1u}$} ++(0.5,0) ++(0.1,0) -- ++(0.5,0) (-0.85,-1) -- node[below]{$a_{1g}$} ++(1.7,0) ; \draw [grx,densely dashed] (-3.8,-40.17) -- (-0.85,-60) (-3.8,-40.17) -- (-0.85,-1) (-2,-40.17) -- (-0.85,-52) (-2,-40.17) -- (-0.85,-6) (-2,-40.17) -- (-0.85,-40.17) (2,-11.62) -- (0.85,-6) (2,-11.62) -- (0.85,-52) (2,-22.71) -- (0.85,-1) (2,-22.71) -- (0.85,-60) ; \small \node [label={[yshift=1mm]left:\footnotesize$6\e[-]$}] at (-3.75,-67) {$6\times\ce{F}$}; \node at (0,-67) {\ce{SF6}}; \node [label={[yshift=1mm]right:\footnotesize$6\e[-]$}] at (2.85,-67) {\ce{S}}; \end{tikzpicture} \caption{\ce{SF6} orbital diagram.} \label{fig:orbitalDiagram-SF6} \end{figure} \end{proof} \item Label the MO's with the appropriate Mulliken symbols and show the orbital occupancies (i.e., fill in the MO levels with the proper number of electrons). \begin{proof}[Answer] See Figure \ref{fig:orbitalDiagram-SF6}. \end{proof} \item Based on the MO diagram, comment on the number of bonding electrons in \ce{SF6} and the bond-order of each \ce{S-F} bond. \begin{proof}[Answer] There are 8 bonding electrons (the two in the $1a_{1g}$ orbital, and the six in the degenerate $1t_{1u}$ orbitals; the four in the degenerate $1e_g$ orbitals are nonbonding and all anti-bonding orbitals are unfilled). Since the bond order is one half the number of bonding electrons divided by the number of bonds, we have $\text{B.O.}=\frac{2}{3}$. \end{proof} \end{enumerate} \end{enumerate} % Maziotti is a good QMech professor. % Chem libre texts for sketching MOs. % Show coefficients. % Bond order correct. % Show valence shell IPs in diagram. \end{document}@book{Abdullah-Faridan-et-al-1983, Address = {Jakarta}, Author = { and and Usman, Umar and .}, Iso_code = {smr}, Olac_field = {semantics; typology; general_linguistics; syntax}, Publisher = {Pusat Pembinaan dan Pengembangan Bahasa, Departemen Pendidikan dan Kebudayan}, Title = {{M}orfologi dan {S}intaksis {B}ahasa {S}imeulue}, Wals_code = {sim}, Year = {1983} } @incollection{Adelaar-1995a, Address = {Berlin / New York}, Author = {}, Booktitle = {{C}omparative {A}ustronesian {D}ictionary 1}, Editor = {.}, Iso_code = {min}, Olac_field = {general_linguistics; typology; phonetics; phonology}, Pages = {433-442}, Publisher = {Mouton de Gruyter}, Title = {{M}inangkabau}, Wals_code = {min}, Year = {1995} } @incollection{Aikhenvald-1998, Address = {Berlin and New York}, Author = {Aikhenvald, .}, Booktitle = {{H}andbook of {A}mazonian {L}anguages 4}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {gae; bwi}, Olac_field = {typology; morphology; syntax; semantics; general_linguistics}, Pages = {225-439}, Publisher = {Mouton de Gruyter}, Title = {{W}arekena}, Wals_code = {bnw; wrk}, Year = {1998} } @book{Alexander-1968, Address = {Rutland}, Author = {.}, Iso_code = {haw}, Olac_field = {semantics; syntax; typology; general_linguistics}, Publisher = {}, Title = {{H}awaiian {G}rammar}, Wals_code = {haw}, Year = {1968} } @book{Alieva-and-Bui-Khanh-The-1999, Address = {St.Petersburg}, Author = {Alieva, . and }, Publisher = {Orientalia}, Title = {{T}he {C}ham {L}anguage. {S}poken {I}dioms of the {E}astern {D}ialect}, Wals_code = {cme}, Year = {1999} } @incollection{Allan-1976b, Address = {East Lansing}, Author = {.}, Booktitle = {{T}he {N}on-{S}emitic {L}anguages of {E}thiopia}, Editor = {.}, Iso_code = {mdx}, Olac_field = {morphology; semantics; phonetics; phonology; syntax; general_linguistics; typology}, Pages = {377-392}, Publisher = {African Studies Center, Michigan State University}, Title = {{D}izi}, Wals_code = {diz}, Year = {1976} } @incollection{Andrade-1933, Address = {New York}, Author = {.}, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 3}, Editor = {}, Iso_code = {qui}, Olac_field = {morphology; general_linguistics; syntax; typology; semantics}, Pages = {151-292}, Publisher = {Columbia University Press}, Series = {Bureau of American Ethnology, Bulletin}, Title = {{Q}uileute}, Volume = {40.3}, Wals_code = {qui}, Year = {1933} } @misc{Applegate-1972, Author = {Applegate, .}, Iso_code = {inz}, Olac_field = {morphology; syntax; typology; general_linguistics}, School = {University of California at Berkeley}, Title = {{I}neseño {C}humash {G}rammar}, Wals_code = {cin}, Year = {1972} } @book{Armbruster-1960, Address = {Cambridge}, Author = {Armbruster, }, Iso_code = {kzh}, Olac_field = {general_linguistics; syntax; semantics; morphology; phonetics; typology; phonology}, Publisher = {The University Press}, Title = {{D}ongolese {N}ubian: {A} {G}rammar}, Wals_code = {nbd}, Year = {1960} } @book{Arnott-1970, Address = {Oxford}, Author = {.}, Iso_code = {fuv}, Olac_field = {semantics; general_linguistics; typology; morphology; syntax}, Publisher = {Clarendon Press}, Title = {{T}he {N}ominal and {V}erbal {S}ystems of {F}ula}, Wals_code = {fma}, Year = {1970} } @book{Aronson-1982, Address = {Columbus, Ohio}, Author = {.}, Iso_code = {kat}, Olac_field = {phonology; general_linguistics; phonetics; typology}, Publisher = {Slavica}, Title = {{G}eorgian - {A} reading {G}rammar}, Wals_code = {geo}, Year = {1982} } @article{Baird-2001, Author = {}, Iso_code = {xxk}, Journal = {Linguistika, Wahana Pengembang Cakrawala Linguistik}, Olac_field = {typology; syntax; general_linguistics; semantics}, Pages = {1-22}, Title = {{N}umerals and {C}lassifiers in {K}eo}, Volume = {14}, Wals_code = {keo}, Year = {2001} } @book{Barber-1977, Address = {Aberdeen}, Author = {}, Iso_code = {ban}, Olac_field = {phonetics; syntax; general_linguistics; phonology; typology}, Publisher = {University of Aberdeen}, Series = {Occasional Publications}, Title = {{A} {G}rammar of the {B}alinese {L}anguage}, Volume = {3}, Wals_code = {bal}, Year = {1977} } @incollection{Barnes-1990, Address = {Austin}, Author = {}, Booktitle = {{A}mazonian {L}inguistics, {S}tudies in {L}owland {S}outh {A}merican {L}anguages}, Editor = {.}, Iso_code = {tue}, Olac_field = {general_linguistics; semantics; typology; syntax}, Pages = {273-292}, Publisher = {University of Texas Press}, Title = {{C}lassifiers in {T}uyuca}, Wals_code = {tuy}, Year = {1990} } @book{Barr-1965, Address = {Nairobi}, Author = {.}, Iso_code = {lgg}, Olac_field = {morphology; semantics; syntax; general_linguistics; typology}, Publisher = {East African Literature Bureau}, Title = {{A} {C}ourse in {L}ugbara}, Wals_code = {lug}, Year = {1965} } @incollection{Baumgartner-1998, Address = {Canberra}, Author = {Baumgartner, .}, Booktitle = {{P}apers in {S}outheast {A}sian {L}inguistics. {V}olume 15: {F}urther {C}hamic {S}tudies}, Editor = {}, Pages = {1-20}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series A}, Title = {{A} grammar sketch of {W}estern ({C}ambodian) {C}ham}, Volume = {89}, Wals_code = {chw}, Year = {1998} } @book{Beaton-1968, Address = {Khartoum}, Author = {.}, Iso_code = {fvr}, Olac_field = {phonetics; phonology; general_linguistics; semantics; syntax; morphology; typology}, Publisher = {Sudan Research Unit, University of Khartoum}, Series = {Linguistic Monograph Series}, Title = {{A} {G}rammar of the {F}ur {L}anguage}, Volume = {1}, Wals_code = {fur}, Year = {1968} } @incollection{Beckwith-1992, Address = {Szeged}, Author = {Beckwith, .}, Booktitle = {{T}he {S}tructure of {H}ungarian}, Editor = {Kenese. and .}, Iso_code = {hun}, Olac_field = {general_linguistics; semantics; typology; syntax}, Pages = {197-206}, Publisher = {JATE}, Series = {Approaches to Hungarian}, Title = {{C}lassifiers in {H}ungarian}, Volume = {4}, Wals_code = {hun}, Year = {1992} } @incollection{Beeler-1976, Address = {The Hague}, Author = {.}, Booktitle = {{H}okan {S}tudies}, Editor = { and }, Iso_code = {boi}, Olac_field = {morphology; semantics; general_linguistics; typology; syntax}, Pages = {251-270}, Publisher = {Mouton}, Title = {{B}arbareno {C}humash grammar: a farrago}, Wals_code = {cba}, Year = {1976} } @book{Benjamin-1976, Address = {Honolulu}, Author = {}, Booktitle = {{A}ustroasiatic {S}tudies 1}, Editor = { al.}, Iso_code = {tea}, Olac_field = {morphology; semantics; syntax; typology; general_linguistics}, Pages = {129-187}, Publisher = {University of Hawaii Press}, Series = {Oceanic Linguistics Special Publication}, Title = {{A}n {O}utline of {T}emiar {G}rammar}, Wals_code = {tmr}, Year = {1976} } @book{Berry-and-Berry-1999, Address = {Canberra}, Author = { and }, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{A} {D}escription of {A}bun: {A} {W}est {P}apuan {L}anguage of {I}rian {J}aya}, Volume = {115}, Wals_code = {abu}, Year = {1999} } @book{Besnier-2000, Address = {London / New York}, Author = {}, Iso_code = {tvl}, Olac_field = {semantics; syntax; typology; general_linguistics}, Publisher = {Routledge}, Series = {Descriptive Grammars}, Title = {{T}uvaluan, a {P}olynesian {L}anguage of the {C}entral {P}acific}, Wals_code = {tvl}, Year = {2000} } @incollection{Bickel-2003, Address = {London}, Author = {}, Booktitle = {{T}he {S}ino-{T}ibetan {L}anguages}, Editor = { and LaPolla, .}, Iso_code = {byw}, Olac_field = {general_linguistics; semantics; syntax; typology; morphology}, Pages = {546-570}, Publisher = {Curzon Press}, Title = {{B}elhare}, Wals_code = {bel}, Year = {2003} } @book{Biligiri-1965, Address = {Poona}, Author = {.}, Iso_code = {khr}, Olac_field = {phonology; typology; semantics; general_linguistics; syntax; phonetics}, Publisher = {Deccan College Postgraduate and Research Institute}, Series = {Building Centenary and Silver Jubilee Series}, Title = {{K}haria: {P}honology, {G}rammar and {V}ocabulary}, Volume = {3}, Wals_code = {khr}, Year = {1965} } @book{Birk-1976, Address = {Canberra}, Author = {Birk, .}, Iso_code = {mpb}, Olac_field = {morphology; general_linguistics; semantics; typology; syntax}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{T}he {M}alakmalak {L}anguage, {D}aly {R}iver ({W}estern {A}rnhem {L}and)}, Volume = {45}, Wals_code = {mlk}, Year = {1976} } @article{Bleek-1928, Author = {Bleek, .}, Besttxt = {ptxt2\africa\bleek_bushman1928-1929_o.txt}, Cfn = {africa\bleek_bushman1928-1929_o.pdf}, Delivered = {africa\bleek_bushman1928-1929_o.pdf}, Fn = {africa\bleek_bushman1928-1929_o.pdf, africa\bleek_bushman1928-1929v2_o.pdf}, Hhtype = {grammar_sketch}, Inlg = {English [eng]}, Journal = {Zeitschrift für Eingeborenensprachen}, Keywords = {;saf;rsa;lng;grd;w.231;}, Lgcode = {/Xam [xam]}, Macro_area = {Africa}, Pages = {81-98}, Src = {eballiso2009, hh}, Title = {{B}ushman grammar: a grammatical sketch of the language of the /{X}am-ka-!k'e}, Volume = {XIX}, Wals_code = {xam}, Year = {1928-1929} } @incollection{Boas-1911b, Address = {Washington, D. C.}, Author = {}, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 1}, Editor = {}, Iso_code = {wac; chh}, Olac_field = {semantics; syntax; morphology; typology; general_linguistics}, Pages = {559-678}, Publisher = {Government Printing Office}, Series = {Smithsonian Institution Bureau of American Ethnology Bulletin}, Title = {{C}hinook}, Volume = {40}, Wals_code = {cku}, Year = {1911} } @incollection{Boas-1911c, Address = {Washington, D. C.}, Author = {}, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 1}, Editor = {}, Iso_code = {tsi}, Olac_field = {typology; phonetics; phonology; general_linguistics; morphology; syntax; semantics}, Pages = {283-422}, Publisher = {Government Printing Office}, Series = {Smithsonian Institution Bureau of American Ethnology Bulletin}, Title = {{T}simshian}, Volume = {40}, Wals_code = {tsi}, Year = {1911} } @book{Bodomo-1997, Address = {Stanford}, Author = {}, Publisher = {Center for the Study of Language and Information (CSLI) at Stanford University}, Series = {Stanford Monographs in African Languages}, Title = {{T}he {S}tructure of {D}agaare}, Wals_code = {dga}, Year = {1997} } @book{Bohm-1985, Address = {Wien}, Author = {}, Iso_code = {xuu; naq}, Olac_field = {typology; general_linguistics; semantics; syntax}, Publisher = {Institut für Afrikanistik und Ägyptologie der Universität Wien}, Series = {Beiträge zur Afrikanistik}, Title = {{K}hoe - {K}owap: {E}inführung in die {S}prache der {H}ottentotten, {N}ama - {D}ialekt}, Volume = {25}, Wals_code = {kho}, Year = {1985} } @incollection{Borgman-1990, Address = {Berlin}, Author = {.}, Booktitle = {{H}andbook of {A}mazonian {L}anguages 2}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {xsu}, Olac_field = {morphology; syntax; typology; semantics; phonology; general_linguistics; phonetics}, Pages = {15-248}, Publisher = {Mouton de Gruyter}, Title = {{S}anuma}, Wals_code = {snm}, Year = {1990} } @misc{Bowden-1997a, Author = {}, Iso_code = {kmo; mky}, Olac_field = {phonetics; morphology; semantics; syntax; phonology; typology; general_linguistics}, School = {University of Melbourne}, Title = {{T}aba ({M}akian {D}alam): {D}escription of an {A}ustronesian language from {E}astern {I}ndonesia}, Wals_code = {tab}, Year = {1997} } @misc{Braine-1970, Author = {field}, Iso_code = {caq}, Olac_field = {syntax; typology; morphology; general_linguistics; semantics}, School = {University of California at Berkeley}, Title = {{N}icobarese {G}rammar ({C}ar {D}ialect)}, Wals_code = {nca}, Year = {1970} } @book{Bray-1909, Address = {Calcutta}, Author = {}, Publisher = {Superintendent Government Printing}, Title = {{T}he {B}rahui {L}anguage 1: {I}ntroduction and {G}rammar}, Wals_code = {brh}, Year = {1909} } @book{Bright-1957, Address = {Berkeley}, Author = {}, Iso_code = {kyh}, Olac_field = {typology; phonetics; semantics; morphology; general_linguistics; syntax; phonology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{T}he {K}arok {L}anguage}, Volume = {13}, Wals_code = {krk}, Year = {1957} } @book{Bril-2002, Address = {Paris}, Author = {}, Iso_code = {nee}, Olac_field = {general_linguistics; syntax; semantics; typology}, Publisher = {Peeters}, Title = {{L}e nêlêmwa ({N}ouvele-{C}alédonie): analyse syntaxique et sémantique}, Wals_code = {nel}, Year = {2002} } @book{Broadbent-1964, Address = {Berkeley}, Author = {.}, Iso_code = {skd}, Olac_field = {typology; general_linguistics; syntax; morphology; phonetics; phonology; semantics}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{T}he {S}outhern {S}ierra {M}iwok {L}anguage}, Volume = {38}, Wals_code = {mss}, Year = {1964} } @book{Bruce-1984, Address = {Canberra}, Author = {}, Iso_code = {amp}, Olac_field = {semantics; phonetics; morphology; phonology; typology; syntax; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{T}he {A}lamblak {L}anguage of {P}apua {N}ew {G}uinea ({E}ast {S}epik)}, Volume = {81}, Wals_code = {ala}, Year = {1984} } @book{Brustad-2000, Address = {Georgetown}, Author = {.}, Iso_code = {arz}, Olac_field = {general_linguistics; typology; morphology}, Publisher = {University Press}, Title = {{S}yntax of spoken {A}rabic. {A} {C}omparative {S}tudy of {M}oroccan, {E}gyptian, {S}yrian, and {K}uwaiti {D}ialects}, Wals_code = {aeg; ako}, Year = {2000} } @unpublished{Burenhult-2000, Author = {}, Iso_code = {jhi}, Note = {Paper presented at the Fifth International Symposium on Language and Linguistics, Pan-Asiatic Linguistics, Ho Chi Minh City, Vietnam, 17 November 2000}, Olac_field = {typology; syntax; semantics; general_linguistics}, Title = {{U}nitizer and {N}ominalizer: the /n/ affix in {J}ahai}, Type = {paper}, Wals_code = {jah}, Year = {2000} } @book{Burling-1961, Address = {Pune}, Author = {}, Publisher = {Deccan College Postgraduate and Research Institute}, Series = {Deccan College Monograph Series}, Title = {{A} {G}aro {G}rammar}, Volume = {25}, Wals_code = {gar}, Year = {1961} } @book{CIDCA-1985, Address = {Managua}, Author = {CIDCA}, Iso_code = {miq}, Olac_field = {general_linguistics; typology; syntax}, Publisher = {Centro de Investigaciones y Documentación de la Costa Atlántica}, Title = {{M}iskitu {B}ila {A}isanka: {G}ramatica {M}iskita}, Wals_code = {mis}, Year = {1985} } @book{Camaj-1984, Address = {Wiesbaden}, Author = {}, Iso_code = {als; aln}, Olac_field = {syntax; semantics; typology; general_linguistics}, Publisher = {Harrassowitz}, Title = {{A}lbanian grammar}, Wals_code = {alb}, Year = {1984} } @book{Campbell-1985, Address = {Berlin}, Author = {}, Iso_code = {ppl}, Olac_field = {general_linguistics; syntax; phonetics; phonology; morphology; semantics; typology}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{T}he {P}ipil {L}anguage of {E}l {S}alvador}, Volume = {1}, Wals_code = {pip}, Year = {1985} } @book{Canonici-1995, Address = {Durban}, Author = {.}, Iso_code = {zul}, Olac_field = {morphology; typology; syntax; general_linguistics; semantics}, Publisher = {Department of Zulu Language and Literature, University of Natal}, Title = {{Z}ulu grammatical structure}, Wals_code = {zul}, Year = {1995} } @book{Capell-1971, Address = {Canberra}, Author = {}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{A}rosi {G}rammar}, Volume = {20}, Wals_code = {aro}, Year = {1971} } @book{Capell-and-Hinch-1970, Address = {The Hague}, Author = { and .}, Iso_code = {mph}, Olac_field = {phonology; semantics; phonetics; general_linguistics; syntax; morphology; typology}, Publisher = {Mouton de Gruyter}, Series = {Janua Linguarum, Series Practica}, Title = {{M}aung {G}rammar}, Volume = {98}, Wals_code = {mau}, Year = {1970} } @book{Carlin-1993, Address = {Köln}, Author = {}, Iso_code = {teu}, Olac_field = {general_linguistics; typology; syntax; semantics}, Publisher = {Institut für Afrikanistik, Universität zu Köln}, Series = {Afrikanistische Monographien}, Title = {{T}he {S}o {L}anguage}, Volume = {2}, Wals_code = {so}, Year = {1993} } @book{Carlson-1994, Address = {Berlin}, Author = {}, Iso_code = {spp}, Olac_field = {phonetics; semantics; syntax; general_linguistics; morphology; typology; phonology}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {S}upyire}, Volume = {14}, Wals_code = {sup}, Year = {1994} } @book{Caughley-1982, Address = {Canberra}, Author = {Caughley, }, Iso_code = {cdm}, Olac_field = {phonetics; syntax; general_linguistics; semantics; phonology; typology}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{T}he {S}yntax and {M}orphology of the {V}erb in {C}hepang}, Volume = {84}, Wals_code = {cpn}, Year = {1982} } @article{Cauty-1974a, Author = {.}, Iso_code = {pbh}, Journal = {Revista Colombiana de Antropologia}, Olac_field = {phonology; typology; phonetics; general_linguistics}, Pages = {251-254}, Title = {{L}os sistemas fonologicos y silabicos de la lengua {P}anare}, Volume = {17}, Wals_code = {pnr}, Year = {1974} } @book{Cech-and-Heinschink-1996, Address = {München}, Author = { and Heinschink, .}, Olac_field = {typology; general_linguistics; syntax; semantics}, Publisher = {Lincom Europa}, Series = {Languages of the World / Materials}, Title = {{S}epečides-{R}omani}, Volume = {106}, Wals_code = {rse}, Year = {1996} } @book{Chadwick-1975, Address = {Canberra}, Author = {}, Iso_code = {jig}, Olac_field = {syntax; general_linguistics; phonology; semantics; phonetics; morphology; typology}, Publisher = {Australian Institute of Aboriginal Studies}, Title = {{A} {D}escriptive {S}tudy of the {D}jingili {L}anguage}, Wals_code = {dji}, Year = {1975} } @incollection{Chafe-1996, Address = {Washington}, Author = {.}, Booktitle = {{H}andbook of {N}orthamerican {I}ndians. {V}olume 17: {L}anguages}, Editor = {}, Iso_code = {see}, Olac_field = {semantics; general_linguistics; phonology; phonetics; syntax; typology}, Pages = {551-579}, Publisher = {Smithsonian Institute}, Title = {{S}ketch of {S}eneca, an {I}roquoian {L}anguage}, Wals_code = {snc}, Year = {1996} } @book{Chaffanjon-1889, Address = {Paris}, Author = {.}, Iso_code = {bae}, Olac_field = {typology; syntax; general_linguistics}, Publisher = {Librairie Hachette Et.}, Title = {{L}'{O}rénoque et le {C}aura}, Wals_code = {bae}, Year = {1889} } @incollection{Chapin-1978, Address = {Austin}, Author = {.}, Booktitle = {{S}yntactic {T}ypology}, Editor = {Lehmann, .}, Iso_code = {rap}, Olac_field = {general_linguistics; typology; morphology; semantics; syntax}, Pages = {139-168}, Publisher = {University of Texas Press}, Title = {{E}aster {I}sland: a characteristic {VSO} language}, Wals_code = {rap}, Year = {1978} } @incollection{Chapman-and-Derbyshire-1991, Address = {Berlin}, Author = { and Derbyshire, .}, Booktitle = {{H}andbook of {A}mazonian {L}anguages 3}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {pad}, Olac_field = {phonology; general_linguistics; syntax; semantics; typology; morphology; phonetics}, Pages = {161-352}, Publisher = {Mouton de Gruyter}, Title = {{P}aumari}, Wals_code = {pau}, Year = {1991} } @book{Christensen-1930, Address = {Copenhagen}, Author = {}, Iso_code = {glk}, Olac_field = {general_linguistics; semantics; typology; syntax}, Publisher = {Andr. Fred. Høst \& søn.}, Title = {{C}ontributions à la {D}ialectologie {I}ranienne}, Wals_code = {gil}, Year = {1930} } @book{Cole-1982, Address = {Amsterdam}, Author = {}, Iso_code = {qvi}, Olac_field = {syntax; morphology; typology; phonology; general_linguistics; phonetics; semantics}, Publisher = {North-Holland}, Series = {Lingua Descriptive Studies}, Title = {{I}mbabura {Q}uechua}, Volume = {5}, Wals_code = {qim}, Year = {1982} } @book{Crofts-1973, Address = {Brasilia}, Author = {}, Iso_code = {myu}, Olac_field = {general_linguistics; typology; semantics; syntax}, Publisher = {Summer Institute of Linguistics}, Series = {Série Lingüística}, Title = {{G}ramática mundurukú}, Volume = {2}, Wals_code = {muu}, Year = {1973} } @incollection{Crowley-1983, Address = {Amsterdam}, Author = {}, Booktitle = {{H}andbook of {A}ustralian {L}anguages 3}, Editor = {Dixon, . and .}, Iso_code = {urf}, Olac_field = {typology; semantics; syntax; phonology; phonetics; general_linguistics; morphology}, Pages = {306-428}, Publisher = {John Benjamins}, Title = {{U}radhi}, Wals_code = {uhi}, Year = {1983} } @book{Crowley-1998a, Address = {Honolulu}, Author = {}, Publisher = {University of Hawaii Press}, Title = {{A}n {E}rromangan ({S}ye) {G}rammar}, Wals_code = {err}, Year = {1998} } @book{Crowley-1998c, Address = {München}, Author = {}, Iso_code = {uur}, Olac_field = {general_linguistics; syntax; semantics; typology}, Publisher = {Lincom Europa}, Series = {Languages of the World / Materials}, Title = {{U}ra}, Volume = {240}, Wals_code = {ura}, Year = {1998} } @book{Cyffer-1998, Address = {Köln}, Author = {}, Iso_code = {knc}, Olac_field = {typology; general_linguistics; semantics; syntax; morphology}, Publisher = {Köppe}, Title = {{A} {S}ketch of {K}anuri}, Wals_code = {knr}, Year = {1998} } @incollection{Datooga-2000, Address = {Köln}, Author = {}, Booktitle = {{P}roceedings of the 2nd world congress of {A}frican linguistics, {L}eipzig 1997}, Editor = { and Gensler, }, Hhtype = {specific_feature (computerized assignment from "verb")}, Inlg = {English [eng]}, Keywords = {;nea;eaf;lng;lxl;grm;v.141;}, Lgcode = {Datooga [tcc] (autotranslated from Maho's coding system)}, Macro_area = {Africa}, Pages = {603-616}, Publisher = {Rüdiger Köppe Verlag}, Src = {eballiso2009, weball}, Subject_headings = {nea, eaf, lng, lxl, grm, v.141}, Title = {{V}erb classes in {N}ilotic: evidence from {D}atooga (southern {N}ilotic)}, Wals_code = {dat}, Year = {2000} } @book{Davies-1989b, Address = {London}, Author = {}, Iso_code = {kpw}, Olac_field = {typology; semantics; general_linguistics; syntax}, Publisher = {Routledge}, Series = {Croom Helm Descriptive Grammars}, Title = {{K}obon}, Wals_code = {kob}, Year = {1989} } @book{Dench-1995, Address = {Canberra}, Author = {}, Iso_code = {vma}, Olac_field = {syntax; general_linguistics; semantics; morphology; typology; phonology; phonetics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{M}artuthunira: {A} {L}anguage of the {P}ilbara {R}egion of {W}estern {A}ustralia}, Volume = {125}, Wals_code = {mrt}, Year = {1995} } @book{Dench-1998, Address = {München}, Author = {}, Iso_code = {yia}, Olac_field = {syntax; general_linguistics; typology; semantics; morphology}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{Y}ingkarta}, Volume = {137}, Wals_code = {yng}, Year = {1998} } @book{Deni-1991, Address = {Porto Velho}, Author = {}, Lgcode = {Dení [dny]}, Macro_area = {South America}, Publisher = {New Tribes Mission (ms.)}, Src = {fabreall2009ann}, Title = {{G}ender agreement in {D}eni}, Wals_code = {den}, Year = {1991} } @book{Derbyshire-1985, Address = {Dallas}, Author = {Derbyshire, .}, Iso_code = {hix}, Olac_field = {syntax; phonetics; general_linguistics; semantics; phonology; morphology; typology}, Publisher = {Summer Institute of Linguistics.}, Title = {{H}ixkaryana and {L}inguistic {T}ypology}, Wals_code = {hix}, Year = {1985} } @incollection{Derbyshire-1986, Address = {Berlin}, Author = {Derbyshire, .}, Booktitle = {{H}andbook of {A}mazonian languages 1}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {wau; jaa; dny; pab; plu; ter}, Olac_field = {general_linguistics; syntax; typology}, Pages = {469-566}, Publisher = {Mouton de Gruyter}, Title = {{C}omparative survey of morphology and syntax in {B}razilian {A}rawakan}, Wals_code = {prc; trn}, Year = {1986} } @incollection{Derbyshire-and-Payne-1990, Address = {Austin}, Author = {Derbyshire, . and Payne, .}, Booktitle = {{A}mazonian {L}inguistics, {S}tudies in {L}owland {S}outh {A}merican {L}anguages}, Editor = {Payne, .}, Iso_code = {jaa; pib; tuo; myp; amr; arl; snn; qvi; xsu; cul; boa; ore; dny; oca; apu; yad; kaq; pad; wau; pbh; tob; ter; pab; cbt; myu; auc; hix; plu; pid}, Olac_field = {typology; syntax; general_linguistics; semantics}, Pages = {243-271}, Publisher = {University of Texas Press}, Title = {{N}oun {C}lassification {S}ystems of {A}mazonian {L}anguages}, Wals_code = {wur}, Year = {1990} } @book{Dickens-1992, Address = {Windhoek}, Author = {}, Iso_code = {ktz}, Olac_field = {typology; morphology; semantics; syntax; general_linguistics}, Publisher = {Nyae Development Foundation}, Title = {{J}u|'hoan {G}rammar}, Wals_code = {juh}, Year = {1992} } @phdthesis{Dienst-2006, Author = {}, Hhtype = {grammar}, Inlg = {English [eng]}, Lgcode = {Culina [cul]}, Macro_area = {South America}, Pages = {740}, School = {LaTrobe University}, Src = {hh}, Title = {{A} reference grammar of {K}ulina}, Wals_code = {cul}, Year = {2006} } @incollection{Diller-1994, Address = {Amsterdam}, Author = {}, Booktitle = {{S}emantic and lexical universals}, Editor = { and }, Iso_code = {tha}, Olac_field = {syntax; semantics; typology; general_linguistics}, Pages = {149-170}, Publisher = {Benjamins}, Title = {{T}hai}, Wals_code = {tha}, Year = {1994} } @book{Dixon-1910, Address = {Berkeley}, Author = {Dixon, .}, Iso_code = {cid}, Note = {Reprinted by Kraus Reprint Corporation, New York, 1964}, Olac_field = {semantics; phonetics; phonology; general_linguistics; morphology; syntax; typology}, Pages = {293-380}, Publisher = {Berkeley University Press}, Series = {University of California Publications in American Archaeology and Ethnology}, Title = {{T}he {C}himariko {I}ndians and {L}anguage}, Volume = {5.5}, Wals_code = {chi}, Year = {1910} } @book{Dixon-1972, Address = {Cambridge}, Author = {Dixon, .}, Iso_code = {dbl}, Olac_field = {general_linguistics; phonetics; morphology; phonology; syntax; typology; semantics}, Publisher = {Cambridge University Press}, Series = {Cambridge Studies in Linguistics}, Title = {{T}he {D}yirbal {L}anguage of {N}orth {Q}ueensland}, Volume = {9}, Wals_code = {dyi}, Year = {1972} } @article{Djawanai-2002, Author = {Djawanai, Stephanus}, Iso_code = {nxg}, Journal = {Linguistika, Wahana Pengembang Cakrawala Linguistik}, Olac_field = {general_linguistics; semantics; syntax; typology}, Pages = {116-129}, Title = {{T}he {C}lassifier {S}ystem of {N}gadha {L}anguage: {A}n {E}thnolinguistic {P}erspective}, Volume = {16}, Wals_code = {ngd}, Year = {2002} } @misc{Dol-1999, Author = {}, Iso_code = {ayz}, Olac_field = {general_linguistics; typology; semantics; phonetics; phonology; morphology; syntax}, School = {University of Leiden}, Title = {{A} {G}rammar of {M}aybrat: {A} {L}anguage of the {B}ird's {H}ead, {I}rian {J}aya, {I}ndonesia}, Wals_code = {may}, Year = {1999} } @incollection{Donaldson-1999, Address = {New South Wales, Australia}, Author = {}, Booktitle = {{M}acquarie {A}boriginal {W}ords: {A} {D}ictionary of {W}ords from {A}ustralian {A}boriginal and {T}orres {S}trait {I}slander {L}anguages}, Editor = { and }, Iso_code = {wyb}, Olac_field = {general_linguistics; typology; syntax}, Pages = {23-40}, Publisher = {The Macquarie Library Pty Ltd.}, Title = {{N}giyampaa}, Wals_code = {ngi}, Year = {1999} } @article{Donohue-1996, Author = {}, Doi = {10.2307/416102}, Iso_code = {bdl}, Journal = {Language}, Number = {4}, Olac_field = {general_linguistics; syntax; typology}, Pages = {782-793}, Title = {{B}ajau: a symmetrical {A}ustronesian language}, Volume = {72}, Wals_code = {bjs}, Year = {1996} } @book{Donohue-1999a, Address = {Berlin / New York}, Author = {}, Iso_code = {khc; bhq}, Olac_field = {syntax; morphology; phonology; typology; general_linguistics; phonetics; semantics}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {T}ukang {B}esi}, Volume = {20}, Wals_code = {tuk}, Year = {1999} } @book{Dougherty-1983, Address = {Berkeley}, Author = {Dougherty, .}, Iso_code = {fut}, Olac_field = {general_linguistics; semantics; typology; phonology; syntax; phonetics; morphology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{W}est {F}utuna-{A}niwa: {A}n {I}ntroduction to a {P}olynesian {O}utlier {L}anguage}, Volume = {102}, Wals_code = {fut}, Year = {1983} } @book{Dugast-1971, Address = {Paris}, Author = {}, Iso_code = {baz}, Olac_field = {typology; semantics; general_linguistics; syntax; morphology}, Publisher = {Edition Klincksieck}, Series = {Langues et Littératures de l'Afrique Noire}, Title = {{G}rammaire du tunen}, Volume = {8}, Wals_code = {tnn}, Year = {1971} } @misc{Dunn-1999, Author = {}, Iso_code = {ckt}, Olac_field = {phonology; general_linguistics; syntax; semantics; morphology; typology; phonetics}, School = {Australian National University}, Title = {{A} {G}rammar of {C}hukchi}, Wals_code = {chk}, Year = {1999} } @book{Dutton-1996, Address = {München}, Author = {Dutton, .}, Iso_code = {kbk}, Olac_field = {typology; phonetics; syntax; general_linguistics; phonology; morphology; semantics}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{K}oiari}, Volume = {10}, Wals_code = {koi}, Year = {1996} } @unpublished{Eaton-2008, Author = {}, Iso_code = {sad}, Olac_field = {typology; morphology; syntax; semantics; general_linguistics}, Title = {{A} {S}andawe grammar}, Type = {manuscript}, Wals_code = {sdw}, Year = {2008} } @book{Ehrman-1972b, Address = {Washington}, Author = {Ehrman, .}, Iso_code = {khm}, Olac_field = {general_linguistics; syntax; semantics; typology; phonetics; phonology}, Publisher = {Foreign Service Institute}, Title = {{C}ontemporary {C}ambodian: {G}rammatical {S}ketch}, Wals_code = {khm}, Year = {1972} } @book{Emeneau-1944, Address = {Berkeley / Los Angeles}, Author = {Emeneau, .}, Iso_code = {kfe}, Olac_field = {general_linguistics; syntax; phonetics; typology; phonology}, Publisher = {University of California Press}, Title = {{K}ota texts 1}, Wals_code = {kot}, Year = {1944} } @book{Estrada-Fernandez-1996, Address = {München}, Author = {.}, Iso_code = {pia}, Olac_field = {general_linguistics; morphology; syntax; typology; semantics}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{P}ima {B}ajo}, Volume = {71}, Wals_code = {pba}, Year = {1996} } @book{Everett-1991, Address = {Campinas, Brazil}, Author = {.}, Iso_code = {myp}, Olac_field = {morphology; typology; general_linguistics}, Publisher = {Editora de Unicamp}, Title = {{A} lingua piraha e a teoria da sintaxe: descricao, perspectivas e teoria}, Wals_code = {prh}, Year = {1991} } @book{Everett-and-Kern-1997, Address = {London}, Author = {. and }, Iso_code = {pav}, Olac_field = {typology; morphology; phonology; general_linguistics; semantics; syntax; phonetics}, Publisher = {Routledge}, Series = {Descriptive Grammar Series}, Title = {{W}ari: the {P}acaas {N}ovos {L}anguage of {W}estern {B}razil}, Wals_code = {war}, Year = {1997} } @book{Ezard-1997, Address = {Canberra}, Author = {}, Iso_code = {tbo}, Olac_field = {syntax; semantics; morphology; typology; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar of {T}awala, an {A}ustronesian {L}anguage of the {M}ilne {B}ay {A}rea, {P}apua {N}ew {G}uinea}, Volume = {137}, Wals_code = {taw}, Year = {1997} } @misc{Facundes-2000, Author = {Facundes, }, Iso_code = {apu}, Olac_field = {semantics; phonetics; general_linguistics; typology; phonology; syntax; morphology}, School = {State University of New York, Buffalo}, Title = {{T}he {L}anguage of the {A}purinã {P}eople of {B}razil ({M}aipure/{A}rawak)}, Wals_code = {ara}, Year = {2000} } @book{Fairbanks-1958, Address = {New York}, Author = {.}, Iso_code = {hye}, Olac_field = {phonology; typology; phonetics; general_linguistics}, Publisher = {American Council of Learned Societies}, Title = {{S}poken {E}ast {A}rmenian}, Wals_code = {arm}, Year = {1958} } @incollection{Faller-2001, Address = {Amherst}, Author = {}, Booktitle = {{P}roceedings of {SULA}. {T}he {S}emantics of {U}nder-{R}epresented {L}anguages of the {A}mericas}, Editor = {. and }, Iso_code = {quz}, Olac_field = {typology; syntax; semantics; general_linguistics}, Pages = {38-47}, Publisher = {The Graduate Linguistics Students' Association, The University of Massachusetts}, Title = {{T}he problem of {Q}uechua -nka: distributivity vs. group forming}, Wals_code = {qcu}, Year = {2001} } @incollection{Faltz-1995, Address = {Dordrecht}, Author = {.}, Booktitle = {{Q}uantification in {N}atural {L}anguages}, Editor = {. and . and .}, Iso_code = {lkt}, Olac_field = {typology; semantics; general_linguistics; syntax}, Pages = {271-319}, Publisher = {Kluwer Academic Publishers}, Title = {{T}owards a {T}ypology of {N}atural {L}ogic}, Wals_code = {lkt}, Year = {1995} } @book{Farr-1999, Address = {Canberra}, Author = {}, Iso_code = {kpr}, Olac_field = {typology; syntax; semantics; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{T}he {I}nterface {B}etween {S}yntax and {D}iscourse in {K}orafe, a {P}apuan {L}anguage of {P}apua {N}ew {G}uinea}, Volume = {148}, Wals_code = {krf}, Year = {1999} } @book{Feldman-1986, Address = {Canberra}, Author = {}, Iso_code = {kmn}, Olac_field = {semantics; syntax; typology; phonology; general_linguistics; phonetics; morphology}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{A} {G}rammar of {A}wtuw}, Volume = {94}, Wals_code = {awt}, Year = {1986} } @book{Fennell-and-Gelsen-1980, Address = {The Hague, Berlin}, Author = {Fennell, . and }, Iso_code = {lav}, Olac_field = {phonetics; phonology; general_linguistics; syntax; typology; semantics}, Publisher = {Mouton}, Title = {{A} {G}rammar of {M}odern {L}atvian. {V}olumes 1-3}, Wals_code = {lat}, Year = {1980} } @misc{Fernandez-1967, Author = {}, Iso_code = {bfw}, Olac_field = {typology; morphology; semantics; general_linguistics; syntax}, School = {University of North Carolina at Chapel Hill}, Title = {{A} {G}rammatical {S}ketch of {R}emo: {A} {M}unda {L}anguage}, Wals_code = {rem}, Year = {1967} } @book{Fleisch-2001, Address = {Köln}, Author = {}, Iso_code = {lch}, Olac_field = {semantics; general_linguistics; syntax; typology}, Publisher = {Rüdiger Köppe}, Title = {{L}ucazi {G}rammar. {A} {M}orphosemantic {A}nalysis}, Wals_code = {lch; luc}, Year = {2001} } @book{Foley-1991, Address = {Stanford}, Author = {.}, Iso_code = {yee}, Olac_field = {phonetics; typology; general_linguistics; morphology; phonology; semantics; syntax}, Publisher = {Stanford University Press}, Title = {{T}he {Y}imas {L}anguage of {P}apua {N}ew {G}uinea}, Wals_code = {yim}, Year = {1991} } @book{Foreman-1974, Address = {Ukarumpa, Papua New Guinea}, Author = {}, Iso_code = {yss}, Olac_field = {general_linguistics; syntax; typology; morphology; semantics}, Publisher = {Summer Institute of Linguistics}, Series = {Language Data, Asian-Pacific Series}, Title = {{G}rammar of {Y}essan-{M}ayo}, Volume = {4}, Wals_code = {yes}, Year = {1974} } @incollection{Frachtenberg-1922a, Address = {Washington}, Author = {Frachtenberg, }, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 2}, Editor = {}, Iso_code = {csz}, Olac_field = {typology; phonetics; general_linguistics; phonology; morphology; syntax; semantics}, Pages = {297-429}, Publisher = {Government Printing Office}, Series = {Smithsonian Institution Bureau of American Ethnology Bulletin}, Title = {{C}oos}, Volume = {40}, Wals_code = {coo}, Year = {1922} } @incollection{Frachtenberg-1922b, Address = {Washington}, Author = {Frachtenberg, }, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 2}, Editor = {Boas, Franz}, Iso_code = {sis}, Olac_field = {syntax; typology; morphology; semantics; general_linguistics}, Pages = {431-629}, Publisher = {Government Printing Office}, Series = {Smithsonian Institution Bureau of American Ethnology Bulletin}, Title = {{S}iuslawan ({L}ower {U}mpqua)}, Volume = {40}, Wals_code = {siu}, Year = {1922} } @book{Frajzyngier-1993, Address = {Berlin}, Author = {}, Iso_code = {sur}, Olac_field = {typology; syntax; morphology; semantics; general_linguistics}, Publisher = {Dietrich Reimer Verlag}, Title = {{A} {G}rammar of {M}upun}, Wals_code = {mup}, Year = {1993} } @book{Frank-1990, Address = {Arlington}, Author = {}, Iso_code = {arh}, Olac_field = {phonetics; semantics; syntax; phonology; typology; general_linguistics; morphology}, Publisher = {Summer Institute of Linguistics and The University of Texas at Arlington}, Series = {Studies in the Language of Colombia}, Title = {{I}ka {S}yntax}, Volume = {1}, Wals_code = {ika}, Year = {1990} } @book{Franklin-1971, Address = {Canberra}, Author = {Franklin, }, Iso_code = {kjs; kew}, Olac_field = {semantics; phonetics; morphology; syntax; phonology; typology; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar of {K}ewa, {N}ew {G}uinea}, Volume = {16}, Wals_code = {kew}, Year = {1971} } @book{Gair-1970, Address = {The Hague}, Author = {.}, Iso_code = {sin}, Olac_field = {semantics; typology; syntax; general_linguistics}, Publisher = {Mouton}, Title = {{C}olloquial {S}inhalese {C}lause {S}tructures}, Wals_code = {snh}, Year = {1970} } @book{Gani-et-al-1986, Address = {Jakarta}, Author = { and , }, Iso_code = {kge}, Olac_field = {semantics; general_linguistics; typology; syntax}, Publisher = {Pusat Pembinaan dan Pengembangan Bahasa}, Title = {{M}orfologi dan {S}intaksis {B}ahasa {K}ayu {A}gung}, Wals_code = {kag}, Year = {1986} } @book{Garantjang-et-al-1989, Address = {Jakarta}, Author = { and , . and }, Iso_code = {npy}, Olac_field = {general_linguistics; typology; semantics; syntax}, Publisher = {Pusat Pembinaan dan Pengembangan Bahasa, Departemen Pendidikan dan Kebudayan}, Title = {{S}truktur {B}ahasa {N}apu}, Wals_code = {npu}, Year = {1989} } @misc{Genetti-1990, Author = {}, Iso_code = {new}, Olac_field = {syntax; semantics; typology; general_linguistics}, School = {University of Oregon}, Title = {{A} {D}escriptive and {H}istorical {A}ccount of the {D}olakha {N}ewari {D}ialect}, Wals_code = {new}, Year = {1990} } @book{Geurtjens-1921, Address = {Weltevreden and 's Gravenhage}, Author = {}, Iso_code = {kei}, Olac_field = {general_linguistics; typology; phonetics; syntax; phonology}, Publisher = {Albrecht and }, Series = {Verhandelingen van het Bataviaansch Genootschap voor Kunsten en Wetenschappen}, Title = {{S}praakleer der {K}eieesche taal}, Volume = {63.2}, Wals_code = {kei}, Year = {1921} } @book{Glinert-1989, Address = {New York}, Author = {}, Iso_code = {heb}, Olac_field = {semantics; general_linguistics; phonetics; syntax; morphology; phonology; typology}, Publisher = {Cambridge University Press}, Title = {{T}he {G}rammar of {M}odern {H}ebrew}, Wals_code = {heb}, Year = {1989} } @incollection{Goddard-1911, Address = {Washington, D. C.}, Author = {Goddard, }, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages}, Editor = {}, Iso_code = {hup}, Olac_field = {syntax; general_linguistics; semantics; typology; morphology}, Pages = {85-158}, Publisher = {Government Printing Office}, Series = {Bureau of American Ethnology Bulletin}, Title = {{A}thapascan ({H}upa)}, Volume = {40.1}, Wals_code = {hpd}, Year = {1911} } @book{Gordon-1986, Address = {Berkeley}, Author = {}, Iso_code = {mrc}, Olac_field = {phonology; syntax; typology; general_linguistics; morphology; semantics; phonetics}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{M}aricopa {M}orphology and {S}yntax}, Volume = {108}, Wals_code = {mar}, Year = {1986} } @book{Goswami-1970, Address = {Gauhati, Assam}, Author = {}, Iso_code = {asm}, Olac_field = {morphology; typology; general_linguistics}, Publisher = {Department of Historical Antiquarian Studies}, Title = {{A} {S}tudy of {K}amrupi, a dialect of {A}ssamese}, Wals_code = {ass}, Year = {1970} } @unpublished{Green-and-Green-1972, Author = { and }, Iso_code = {plu}, Olac_field = {syntax; morphology; semantics; general_linguistics; typology}, School = {Summer Institute of Linguistics, Brazil}, Title = {{S}urface {S}tructure of {P}alikur {G}rammar}, Type = {manuscript}, Wals_code = {plk}, Year = {1972} } @incollection{Greppin-2001, Address = {New York / Dublin}, Author = {}, Booktitle = {{F}acts {A}bout the {W}orld's {L}anguages, {A}n {E}ncyclopedia of the {W}orld's {L}anguages: {P}ast and {P}resent}, Editor = { and }, Iso_code = {hye}, Olac_field = {typology; general_linguistics; morphology}, Pages = {39-42}, Publisher = {HW Wilson}, Title = {{A}rmenian}, Wals_code = {arz}, Year = {2001} } @book{Groves-et-al-1985, Address = {Canberra}, Author = {. and . and }, Iso_code = {gil}, Olac_field = {morphology; semantics; syntax; phonology; typology; phonetics; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series D}, Title = {{K}iribatese: {A}n {O}utline {D}escription}, Volume = {64}, Wals_code = {krb}, Year = {1985} } @book{Gruzdeva-1998, Address = {München and Newcastle}, Author = {}, Iso_code = {niv}, Olac_field = {phonology; general_linguistics; typology; phonetics; syntax; morphology; semantics}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{N}ivkh}, Volume = {111}, Wals_code = {niv; nvs}, Year = {1998} } @article{Guduf-1966, Author = {.}, Besttxt = {ptxt2\africa\scheytt_yaghwatadaxa1966-1967_o.txt}, Cfn = {africa\scheytt_yaghwatadaxa1966-1967_o.pdf}, Delivered = {africa\scheytt_yaghwatadaxa1966-1967_o.pdf}, Fn = {africa\scheytt_yaghwatadaxa1966-1967.pdf, africa\scheytt_yaghwatadaxa1966-1967_o.pdf}, Hhtype = {text}, Inlg = {German [deu]}, Journal = {Afrika und Übersee}, Keywords = {;waf;nga;ltr;x.534d;}, Lgcode = {Guduf-Gava [gdf]}, Macro_area = {Africa}, Pages = {4-34}, Src = {eballiso2009, hh}, Title = {{P}roben der {S}prache der {Y}aghwatadaxa in {G}avva ({N}ordostnigerien)}, Volume = {L}, Wals_code = {gdf}, Year = {1966/1967} } @book{Guma-1971, Address = {Pietermaritzburg}, Author = {Guma, }, Iso_code = {sot}, Olac_field = {typology; syntax; semantics; general_linguistics}, Publisher = {Shuter and Shooter}, Title = {{A}n {O}utline {S}tructure of {S}outhern {S}otho}, Wals_code = {ses}, Year = {1971} } @book{Haas-1940, Address = {New York}, Author = {Haas, .}, Iso_code = {tun}, Olac_field = {phonology; general_linguistics; morphology; semantics; phonetics; typology; syntax}, Publisher = {}, Title = {{T}unica}, Wals_code = {tun}, Year = {1940} } @book{Haiman-1980, Address = {Amsterdam}, Author = {}, Iso_code = {ygr}, Olac_field = {syntax; typology; semantics; phonetics; morphology; phonology; general_linguistics}, Publisher = {}, Series = {Studies in Language Companion Series}, Title = {{H}ua: {A} {P}apuan {L}anguage of the {E}astern {H}ighlands of {N}ew {G}uinea}, Volume = {5}, Wals_code = {hua}, Year = {1980} } @misc{Hale-1959, Author = {.}, Iso_code = {ood}, Olac_field = {general_linguistics; typology; phonetics; phonology}, School = {Indiana University, Bloomington}, Title = {{A} {P}apago {G}rammar}, Wals_code = {ood}, Year = {1959} } @book{Hamel-1994, Address = {Canberra}, Author = {Hamel, .}, Iso_code = {los}, Olac_field = {syntax; typology; phonology; semantics; phonetics; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar and {L}exicon of {L}oniu, {P}apua {N}ew {G}uinea}, Volume = {103}, Wals_code = {lon}, Year = {1994} } @misc{Handel-and-Nichols-2000, Author = { and .}, HowPublished = {online}, Iso_code = {inh}, Note = {Downloaded from http://ingush.berkeley.edu:7012.}, Olac_field = {syntax; general_linguistics; typology}, Title = {{I}ngush {G}rammar}, Url = {http://ingush.berkeley.edu:7012/}, Wals_code = {ing}, Year = {2000} } @book{Hardman-1983, Address = {Peru}, Author = {.}, Iso_code = {jqr}, Olac_field = {semantics; general_linguistics; syntax; typology}, Publisher = {Instituto des estudios peruanos}, Title = {{J}aqaru: {C}ompendio de estructura fonologica y morfologica}, Wals_code = {jaq}, Year = {1983} } @book{Harms-1994, Address = {Dallas / Arlington}, Author = {Harms, }, Publisher = {Summer Institute of Linguistics and University of Texas}, Series = {Studies in the Languages of Colombia}, Title = {{E}pena {P}edee {S}yntax}, Volume = {4}, Wals_code = {epe}, Year = {1994} } @book{Harrell-et-al-1965, Address = {Washington}, Author = { al.}, Publisher = {Georgetown University Press}, Title = {{A} {B}asic {C}ourse in {M}oroccan {A}rabic}, Wals_code = {amr}, Year = {1965} } @incollection{Harris-1988, Address = {London}, Author = {}, Booktitle = {{T}he {R}omance {L}anguages}, Editor = { and }, Iso_code = {fra}, Olac_field = {phonology; morphology; typology; phonetics; syntax; general_linguistics; semantics}, Pages = {209-245}, Publisher = {Croom Helm}, Title = {{F}rench}, Wals_code = {fre}, Year = {1988} } @book{Harrison-1976, Address = {Honolulu}, Author = {.}, Iso_code = {mkj}, Olac_field = {general_linguistics; syntax; semantics; typology}, Publisher = {University Press of Hawaii}, Title = {{M}okilese reference grammar}, Wals_code = {mok}, Year = {1976} } @incollection{Hartzler-1994, Address = {Amsterdam}, Author = {}, Booktitle = {{T}ypological {S}tudies in {N}egation}, Editor = { and , René}, Iso_code = {set}, Olac_field = {syntax; typology; general_linguistics}, Pages = {51-64}, Publisher = {}, Series = {Typological Studies in Language}, Title = {{S}entani}, Volume = {29}, Wals_code = {snt}, Year = {1994} } @misc{Harvey-1992, Author = {}, School = {University of Sydney}, Title = {{T}he {G}aagudju {P}eople and {T}heir {L}anguage}, Wals_code = {gaa}, Year = {1992} } @book{Haspelmath-1993, Address = {Berlin}, Author = {}, Iso_code = {lez}, Olac_field = {phonetics; general_linguistics; semantics; typology; morphology; phonology; syntax}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {L}ezgian}, Volume = {9}, Wals_code = {lez}, Year = {1993} } @incollection{Hausenberg-1998, Address = {London}, Author = {}, Booktitle = {{T}he {U}ralic {L}anguages}, Editor = {}, Iso_code = {kpv}, Olac_field = {semantics; typology; syntax; general_linguistics}, Pages = {305-326}, Publisher = {Routledge}, Title = {{K}omi}, Wals_code = {kzy}, Year = {1998} } @book{Hayward-1984, Address = {Hamburg}, Author = {}, Publisher = {}, Title = {{T}he {A}rbore {L}anguage: {A} {F}irst {I}nvestigation (including a vocabulary)}, Wals_code = {abo}, Year = {1984} } @inbook{Healey-1965b, Address = {Canberra}, Author = {.}, Booktitle = {{P}apers in {N}ew {G}uinea {L}inguistics 3}, Iso_code = {tlf}, Olac_field = {morphology; typology; syntax; general_linguistics}, Pages = {1-26}, Publisher = {Australian National University}, Series = {Linguistic Circle of Canberra Publications, Pacific Linguistics, Series A}, Title = {{T}elefol clause structure}, Volume = {5}, Wals_code = {tlf}, Year = {1965} } @book{Heath-1980b, Address = {Canberra}, Author = {}, Iso_code = {wnd}, Olac_field = {semantics; syntax; morphology; typology; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{B}asic {M}aterials in {W}arndarang: {G}rammar, {T}exts, and {D}ictionary}, Volume = {72}, Wals_code = {wrn}, Year = {1980} } @book{Heath-1984, Address = {Atlantic Highlands N. J. / Canberra}, Author = {}, Iso_code = {nuy}, Olac_field = {syntax; general_linguistics; typology; semantics; morphology; phonology; phonetics}, Publisher = {Humanities Press / Australian Institute of Aboriginal Studies}, Title = {{A} {F}unctional {G}rammar of {N}unggubuyu}, Wals_code = {nug}, Year = {1984} } @book{Heath-1999b, Address = {Berlin}, Author = {}, Iso_code = {khq}, Olac_field = {phonology; typology; semantics; general_linguistics; morphology; phonetics; syntax}, Publisher = {Mouton de Gruyter}, Title = {{A} {G}rammar of {K}oyra {C}hiini: the {S}onghay of {T}imbuktu}, Wals_code = {kch}, Year = {1999} } @misc{Helberg-Chavez-1984, Author = {, }, Iso_code = {amr}, Olac_field = {syntax; typology; general_linguistics}, School = {Universität Tübingen}, Title = {{S}kizze einer {G}rammatik des {A}marakueri}, Wals_code = {amk}, Year = {1984} } @incollection{Hercus-1986b, Address = {Canberra}, Author = {Hercus, .}, Booktitle = {{V}ictorian {L}anguages, {A} {L}ate {S}urvey}, Editor = {.}, Note = {revised edition}, Olac_field = {general_linguistics; semantics; typology; syntax; morphology}, Pages = {3-71}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{O}utline of the {W}embawemba language}, Volume = {77}, Wals_code = {wem}, Year = {1986} } @book{Hewitt-1989d, Address = {London}, Author = {}, Iso_code = {abk}, Olac_field = {typology; syntax; general_linguistics}, Publisher = {Routledge}, Title = {{A}bkhaz}, Wals_code = {abk}, Year = {1989} } @book{Hoffmann-1963, Address = {London}, Author = {}, Iso_code = {mrt}, Olac_field = {phonetics; general_linguistics; typology; phonology; semantics; morphology; syntax}, Publisher = {Oxford University Press for International African Institute}, Title = {{A} {G}rammar of the {M}argi {L}anguage}, Wals_code = {mrg}, Year = {1963} } @incollection{Hoijer-1946b, Address = {New York}, Author = {}, Booktitle = {{L}inguistic {S}tructures of {N}ative {A}merica}, Editor = {}, Iso_code = {tqw}, Note = {Reprinted 1971 by Johnson Reprint Corp., New York}, Olac_field = {typology; phonology; phonetics; syntax; general_linguistics; semantics}, Pages = {289-311}, Publisher = {Viking Fund Inc.}, Title = {{T}onkawa}, Wals_code = {ton}, Year = {1946} } @phdthesis{Hokkien-2011, Author = {}, Besttxt = {ptxt2\eurasia\chen_huian2011.txt}, Cfn = {eurasia\chen_huian2011.pdf}, Delivered = {eurasia\chen_huian2011.pdf}, Fn = {eurasia\chen_huian2011_o.pdf, eurasia\chen_huian2011.pdf}, Hhtype = {grammar}, Inlg = {English [eng]}, Lgcode = {Hui'an = Southern Min = Min Nan [nan]}, Macro_area = {Eurasia}, Pages = {xv+439}, School = {University of Hong Kong}, Src = {hh}, Title = {{T}he {S}outhern {M}in dialect of {H}ui'an: morphosyntax and grammaticalization}, Wals_code = {hok}, Year = {2011} } @book{Holmer-1996, Address = {Lund, Sweden}, Author = {.}, Iso_code = {trv}, Olac_field = {semantics; typology; general_linguistics; syntax}, Publisher = {Lund University Press}, Series = {Travaux de l'Institut de Linguistique de Lund}, Title = {{A} {P}arametric {G}rammar of {S}eediq}, Volume = {30}, Wals_code = {see}, Year = {1996} } @book{Holt-1999a, Address = {München}, Author = {}, Iso_code = {jic}, Olac_field = {syntax; semantics; phonetics; general_linguistics; typology; phonology}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{T}ol ({J}icaque)}, Volume = {170}, Wals_code = {tol}, Year = {1999} } @book{Householder-and-Lofti-1965, Address = {Bloomington}, Author = {. and .}, Publisher = {Indiana University Press}, Series = {Indiana University Publications, Ural and Altaic Series}, Title = {{B}asic {C}ourse in {A}zerbaijani}, Volume = {45}, Wals_code = {aze}, Year = {1965} } @book{Hualde-and-Ortiz-de-Urbina-2003, Address = {Berlin}, Author = {Hualde, and , eds.}, Iso_code = {eus}, Olac_field = {general_linguistics; semantics; morphology; syntax; typology}, Publisher = {Mouton de Gruyter}, Title = {{A} {G}rammar of {B}asque}, Wals_code = {bsq}, Year = {2003} } @incollection{Hughes-2000, Address = {Canberra}, Author = {}, Booktitle = {{S}pices from the {E}ast. {P}apers in {L}anguages of {E}astern {I}ndonesia}, Editor = {.}, Pages = {131-180}, Publisher = {Pacific Linguistics, Australian National University}, Title = {{T}he {M}orphology of {D}obel, {A}ru, with {S}pecial {R}eference to {R}eduplication}, Wals_code = {dob}, Year = {2000} } @book{Hurd-and-Hurd-1966, Address = {Port Moresby, Papua New Guinea}, Author = { and }, Iso_code = {nas}, Olac_field = {general_linguistics; phonetics; syntax; morphology; phonology; typology; semantics}, Publisher = {Summer Institute of Linguistics and Department of Information and Extension Services}, Title = {{N}asioi {L}anguage {C}ourse}, Wals_code = {nas}, Year = {1966} } @book{Huttar-and-Huttar-1994, Address = {London}, Author = {. and }, Iso_code = {djk}, Olac_field = {morphology; syntax; semantics; phonetics; typology; phonology; general_linguistics}, Publisher = {Routledge}, Series = {Descriptive Grammar Series}, Title = {{N}dyuka}, Wals_code = {ndy}, Year = {1994} } @phdthesis{Hyow-2018, Author = {}, Besttxt = {ptxt2\eurasia\zakaria_hyow2018.txt}, Cfn = {eurasia\zakaria_hyow2018.pdf}, Fn = {eurasia\zakaria_hyow2018_o.pdf, eurasia\zakaria_hyow2018.pdf}, Hhtype = {grammar}, Inlg = {English [eng]}, Lgcode = {Hyow = Asho Chin in Bangladesh = Asho Chin [csh]}, Macro_area = {Eurasia}, Pages = {882}, School = {Nanyang Technological University}, Src = {hh}, Title = {{A} grammar of {H}yow}, Wals_code = {hyo}, Year = {2018} } @book{Hyslop-2001, Address = {Canberra}, Author = {}, Publisher = {Australian National University}, Series = {Pacific Linguistics}, Title = {{T}he {L}olovoli dialect of the {N}orth-{E}ast {A}mbae language, {V}anuatu}, Volume = {515}, Wals_code = {aml}, Year = {2001} } @phdthesis{Idaan-2005, Address = {Utrecht}, Author = {}, Besttxt = {ptxt2\papua\goudswaard_begak2005_s.txt}, Cfn = {papua\goudswaard_begak2005_s.pdf}, Class_loc = {P381.S23}, Delivered = {papua\goudswaard_begak2005_s.pdf}, Document_type = {B}, Filenames = {dbj_goudswaard2005.pdf}, Fn = {papua\goudswaard_begak-idaan2005_o.pdf, papua\goudswaard_begak2005_s.pdf, papua/goudswaard_begak2005_s.pdf}, Hhtype = {grammar}, ISBN = {9789076864730}, Inlg = {English [eng]}, Lgcode = {Ida'an [dbj]}, Macro_area = {Papunesia}, Mpi_eva_library_shelf = {PL 5246 GOU 2005}, Mpifn = {begak_goudswaard2005_s.pdf}, Oclc = {783333598}, Pages = {xix+520}, Publisher = {LOT}, School = {Vrije Universiteit Amsterdam}, Series = {LOT Dissertation Series}, Src = {hh, ldh, mpieva, phoible}, Subject_headings = {Sabah –Languages –Phonology, Sabah –Languages –Phonology}, Title = {{T}he {B}egak ({I}da'an) {L}anguage of {S}abah}, Url = {http://www.lotpublications.nl/publish/issues/Goudswaard/index.html}, Volume = {107}, Wals_code = {beg}, Year = {2005} } @incollection{Ikekeonwu-1999, Address = {Cambridge}, Author = {Ikekeonwu, .}, Booktitle = {{H}andbook of the {I}nternational {P}honetic {A}ssociation}, Iso_code = {ibo}, Olac_field = {typology; phonetics; phonology; general_linguistics}, Pages = {108-110}, Publisher = {Cambridge University Press}, Title = {{I}gbo}, Wals_code = {igb}, Year = {1999} } @book{Ikoro-1996, Address = {Leiden}, Author = {}, Iso_code = {ogo}, Olac_field = {syntax; semantics; general_linguistics; phonetics; morphology; typology; phonology}, Publisher = {Research School CNWS, Leiden University}, Title = {{T}he {K}ana {L}anguage}, Wals_code = {kan}, Year = {1996} } @book{Innes-1971, Address = {London}, Author = {}, Iso_code = {men}, Note = {2nd imprint}, Olac_field = {general_linguistics; syntax; semantics; typology}, Publisher = {School Of Oriental and African Studies}, Title = {{A} {P}ractical {I}ntroduction to {M}ende}, Wals_code = {mde}, Year = {1971} } @book{Irwin-1974, Address = {Canberra}, Author = {}, Iso_code = {sll}, Olac_field = {semantics; general_linguistics; syntax; morphology; typology}, Publisher = {Linguistic Circle of Canberra}, Title = {{S}alt-{Y}ui {G}rammar}, Wals_code = {syu}, Year = {1974} } @incollection{Isaev-2004, Address = {New York}, Author = {.}, Booktitle = {{T}he {N}orth {E}ast {C}aucasian {L}anguages}, Iso_code = {dar}, Olac_field = {syntax; typology; semantics; general_linguistics}, Publisher = {Caravan Books}, Series = {The Indigenous Languages of the Caucasus}, Title = {{D}argwa}, Volume = {3}, Wals_code = {drg}, Year = {2004} } @book{Jama-2004, Address = {Oxford}, Author = {.}, Besttxt = {ptxt2\south_america\dixon_jarawara2004.txt}, Cfn = {south_america\dixon_jarawara2004_o.pdf}, Class_loc = {PM6257}, Document_type = {B}, Fn = {south_america\dixon_jarawara2004.pdf, south_america\dixon_jarawara2004_o.pdf}, Hhtype = {grammar}, ISBN = {9780199270675}, Inlg = {English [eng]}, Keywords = {Amazonia, Arawá, Grammar, South America}, Lgcode = {Jaruára [jaa]}, Macro_area = {South America}, Mpi_eva_library_shelf = {PM 6257 DIX 2004}, Oclc = {252668841}, Pages = {xx+636}, Phoible_inventoryid = {1847}, Phoible_languagecode = {jaa}, Phoible_languagename = {Jarawara}, Phoible_languagevariantcode = {jaa_jrw}, Publisher = {Oxford University Press}, Src = {fabreall2009ann, hh, mpieva, phoible, seifart}, Title = {{T}he {J}arawara {L}anguage of {S}outhern {A}mazonia}, Wals_code = {jmm}, Year = {2004} } @book{Jensen-et-al-1977a, Address = {Honolulu}, Author = {Jensen, and Defeg), Raphael}, Iso_code = {yap}, Olac_field = {syntax; phonology; general_linguistics; typology; phonetics; morphology; semantics}, Publisher = {University of Hawaii Press}, Series = {PALI Language Texts}, Title = {{Y}apese {R}eference {G}rammar}, Volume = {Micronesia}, Wals_code = {yap}, Year = {1977} } @incollection{Jessen-1999, Address = {Berlin}, Author = {}, Booktitle = {{W}ord {P}rosodic {S}ystems in the {L}anguages of {E}urope}, Editor = {van , H.}, Iso_code = {deu}, Olac_field = {typology; phonology; phonetics; general_linguistics}, Pages = {515-544}, Publisher = {Mouton de Gruyter}, Title = {{G}erman}, Wals_code = {ger}, Year = {1999} } @book{Jones-1998b, Address = {München}, Author = {Jones, }, Publisher = {Lincom Europa}, Title = {{T}he {B}oko / {B}usa {L}anguage {C}luster}, Wals_code = {bok}, Year = {1998} } @incollection{Karhunen-1994, Address = {Jakarta}, Author = {}, Booktitle = {no booktitle}, Editor = {, }, Iso_code = {pdo}, Olac_field = {syntax; typology; general_linguistics; semantics}, Pages = {17-47}, Publisher = {NUSA}, Series = {Studies in Sulawesi Linguistics}, Title = {{T}he noun phrase in {P}adoe}, Volume = {3}, Wals_code = {pad}, Year = {1994} } @book{Kayser-1993, Address = {Yarralumla, Australia}, Author = {Kayser, MSC}, Iso_code = {nau}, Note = {Edited by }, Olac_field = {semantics; morphology; typology; general_linguistics; syntax}, Publisher = {Embassy of the Federal Republic of Germany}, Title = {{N}auru grammar}, Wals_code = {nau}, Year = {1993} } @incollection{Kazenin-1993, Address = {Strasbourg}, Author = {Kazenin, .}, Booktitle = {{T}he noun phrase in the {A}ndalal dialect of {A}var as spoken at {S}ogratl}, Editor = {.}, Iso_code = {ava}, Olac_field = {typology; syntax; general_linguistics}, Pages = {70-77}, Publisher = {European Social Fund}, Series = {EUROTYP Working Papers}, Title = {{A}ction nominal constructions in the {S}ogratl sialect of {A}var}, Volume = {7/18}, Wals_code = {ava}, Year = {1993} } @incollection{Keen-1983, Address = {Amsterdam}, Author = {}, Booktitle = {{H}andbook of {A}ustralian {L}anguages 3}, Editor = {. and Blake, .}, Iso_code = {gcd}, Olac_field = {general_linguistics; syntax; semantics; typology; morphology}, Pages = {191-304}, Publisher = {John Benjamins}, Title = {{Y}ukulta}, Wals_code = {yuk}, Year = {1983} } @book{Ketapang-1994, Address = {Singapore}, Author = {}, Besttxt = {ptxt2\papua\mintz_malay-indonesian1994_o.txt}, Fn = {papua\mintz_malay-indonesian1994.pdf, papua\mintz_malay-indonesian1994_o.pdf}, Hhtype = {grammar}, ISBN = {9789971003982}, Inlg = {English [eng]}, Lgcode = {Malay [zlm], Indonesian [ind]}, Macro_area = {Eurasia, Papunesia}, Oclc = {34909955}, Pages = {421}, Publisher = {EPB Publishers}, Src = {hh, wals}, Title = {{A} student's grammar of {M}alay \& {I}ndonesian}, Wals_code = {ktp}, Year = {1994} } @book{Key-1967, Address = {The Hague}, Author = {Key, .}, Iso_code = {cyb}, Olac_field = {semantics; phonetics; morphology; phonology; syntax; general_linguistics; typology}, Publisher = {Mouton}, Title = {{M}orphology of {C}ayuvava}, Wals_code = {cyv}, Year = {1967} } @book{Khan-1999, Address = {Leiden / Boston}, Author = {}, Iso_code = {aij}, Olac_field = {general_linguistics; phonetics; semantics; morphology; syntax; phonology; typology}, Publisher = {Brill}, Series = {Handbuch der Orientalistik. Erste Abteilung: Der Nahe und Mittlere Osten}, Title = {{A} {G}rammar of {N}eo-{A}ramaic: the dialect of the {J}ews of {A}rbel}, Volume = {47}, Wals_code = {naj}, Year = {1999} } @book{Khumi-1964, Address = {Rangoon}, Author = {}, Hhtype = {grammar_sketch}, Inlg = {English [eng]}, Lgcode = {Khumi Chin [cnk]}, Macro_area = {Eurasia}, Publisher = {Daw Aye Tin, Linn Press}, Src = {hh}, Title = {{A} handbook of the {C}hin language: {K}humi section}, Wals_code = {khu}, Year = {1964} } @book{Kibrik-2001, Address = {Moskva}, Author = {Anonymous}, Editor = {}, Publisher = {Im}, Title = {{B}agvalinskij jazyk: {G}rammatika, teksty, slovari}, Wals_code = {bgv}, Year = {2001} } @unpublished{Kinkade-2001, Author = {}, Title = {{T}he {A}real {Q}uestion: {N}orthwest {C}oast and {C}alifornia}, Type = {paper}, Wals_code = {eya; klp; klm; mll; tli}, Year = {2001} } @incollection{Klaiman-1990, Address = {Oxford}, Author = {.}, Booktitle = {{T}he {W}orld's {M}ajor {L}anguages}, Editor = {}, Iso_code = {ben}, Olac_field = {general_linguistics; phonetics; typology; phonology}, Pages = {490-513}, Publisher = {Oxford University Press}, Title = {{B}engali}, Wals_code = {ben}, Year = {1990} } @book{Klamer-1998, Address = {Berlin}, Author = {}, Iso_code = {xbr}, Olac_field = {syntax; typology; semantics; general_linguistics}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {K}ambera}, Volume = {18}, Wals_code = {kam}, Year = {1998} } @book{Kretschmar-1995, Address = {Bonn}, Author = {}, Iso_code = {loy}, Olac_field = {general_linguistics; typology; syntax}, Publisher = {VGH-Wissenschaftsverlag}, Series = {Erzählungen und Dialekt aus Südmustang}, Title = {{U}ntersuchung zur {G}rammatik des {S}üdmustang-{D}ialekts}, Wals_code = {brl}, Year = {1995} } @book{Krivoshein-de-Canese-1983, Address = {Asunción}, Author = {, Natalia}, Iso_code = {gug}, Olac_field = {general_linguistics; syntax; typology}, Publisher = {Coleccion Nemity}, Title = {{G}ramatica de la lengua {G}uarani}, Wals_code = {gua}, Year = {1983} } @book{Kroeber-1906, Address = {Berkeley}, Author = {.}, Iso_code = {was}, Olac_field = {general_linguistics; morphology; semantics; typology; phonology; syntax; phonetics}, Pages = {251-317}, Publisher = {University of California Press}, Series = {University of California Publications in American Archaeology and Ethnology}, Title = {{T}he {W}asho {L}anguage of {E}ast {C}entral {C}alifornia and {N}evada}, Volume = {4.5}, Wals_code = {was}, Year = {1907} } @book{Krueger-1961, Address = {Bloomington}, Author = {.}, Publisher = {Indiana University Press}, Series = {Indiana University Publications, Uralic and Altaic Series}, Title = {{C}huvash {M}anual: {I}ntroduction, {G}rammar, {R}eader, and {V}ocabulary}, Volume = {7}, Wals_code = {chv}, Year = {1961} } @book{Krueger-1977, Address = {Bloomington}, Author = {.}, Iso_code = {tyv}, Olac_field = {typology; morphology; syntax; general_linguistics; semantics}, Publisher = {Indiana University Press}, Title = {{T}uvan {M}anual}, Wals_code = {tuv}, Year = {1977} } @misc{Kruspe-1999, Author = {}, Iso_code = {sza}, Olac_field = {phonology; general_linguistics; phonetics; typology; morphology; syntax; semantics}, School = {University of Melbourne}, Title = {{A} {G}rammar of {S}emelai}, Wals_code = {sml}, Year = {1999} } @misc{Krute-1989, Author = {}, Iso_code = {pid}, Olac_field = {semantics; morphology; syntax; general_linguistics; typology}, School = {Columbia University}, Title = {{P}iaroa {N}ominal {M}orphosemantics}, Wals_code = {pia}, Year = {1989} } @book{Kuiper-1962, Address = {Amsterdam}, Author = {Kuiper, .}, Iso_code = {nll}, Olac_field = {typology; general_linguistics; syntax; morphology; semantics}, Publisher = {Noord-Hollandsche Uitgevers Maatschappij}, Title = {{N}ahali: a comparative study}, Wals_code = {nah}, Year = {1962} } @incollection{Laidig-and-Laidig-1995, Address = {Jakarta}, Author = {. and .}, Booktitle = {no booktitle}, Editor = {.}, Iso_code = {alo}, Olac_field = {typology; general_linguistics; syntax; semantics}, Pages = {19-42}, Publisher = {NUSA}, Series = {Descriptive Studies in Languages of Maluku}, Title = {{A} synopsis of {L}arike phonology and syntax}, Volume = {2}, Wals_code = {lrk}, Year = {1995} } @incollection{Lastra-de-Suarez-1984, Address = {Austin}, Author = {, Yolanda}, Booktitle = {{S}upplement to the {H}andbook of {M}iddle {A}merican {L}anguages: {L}inguistics}, Editor = {.}, Iso_code = {pei}, Olac_field = {syntax; morphology; general_linguistics; typology}, Pages = {20-42}, Publisher = {University of Texas Press}, Title = {{C}hichimeco {J}onaz}, Wals_code = {cjo}, Year = {1984} } @book{Lee-1975, Address = {Honolulu}, Author = {}, Iso_code = {kos}, Olac_field = {typology; syntax; semantics; general_linguistics; morphology}, Publisher = {The University Press of Hawaii}, Title = {{K}usaiean {R}eference {G}rammar}, Wals_code = {kos}, Year = {1975} } @book{Lee-1989, Address = {Oxford}, Author = {Lee, .}, Iso_code = {kor}, Olac_field = {general_linguistics; syntax; typology}, Publisher = {Oxford University Press}, Title = {{K}orean {G}rammar}, Wals_code = {kor}, Year = {1989} } @incollection{Leslau-1997b, Address = {Winona Lake}, Author = {}, Booktitle = {{P}honologies of {A}sia and {A}frica 2}, Editor = {.}, Iso_code = {amh}, Olac_field = {phonology; phonetics; general_linguistics; typology}, Pages = {399-429}, Publisher = {Eisenbrauns}, Title = {{A}mharic}, Wals_code = {amh}, Year = {1997} } @book{Lewis-1967, Address = {Oxford}, Author = {Lewis, .}, Iso_code = {tur}, Olac_field = {typology; syntax; morphology; semantics; general_linguistics}, Publisher = {Clarendon Press}, Title = {{T}urkish {G}rammar}, Wals_code = {tur}, Year = {1967} } @misc{Loos-1969, Author = {}, Iso_code = {kaq}, Olac_field = {phonology; typology; phonetics; general_linguistics}, School = {University of Oklahoma}, Series = {SIL of the University of Oklahoma publications}, Title = {{T}he {P}honology of {C}apanahua and its {G}rammatical {B}asis}, Volume = {20}, Wals_code = {cap}, Year = {1969} } @incollection{Lori-1989, Author = {}, Booktitle = {{É}tudes irano-aryennes offertes à {G}ilbert {L}azard}, Hhtype = {grammar_sketch}, Inlg = {French [fra]}, Lgcode = {Lori = Luri-Northern [lrc]}, Macro_area = {Eurasia}, Pages = {37-58}, Publisher = {Paris}, Series = {Studia Iranica: Cahier}, Src = {hh}, Title = {{L}e kurde lori}, Volume = {7}, Wals_code = {lur}, Year = {1989} } @book{Lorimer-1935, Address = {Cambridge, Massacgusetts / Oslo}, Author = {}, Iso_code = {bsk}, Olac_field = {morphology; semantics; syntax; general_linguistics; typology}, Publisher = {Harvard University Press / Aschehoug}, Series = {Instituttet for Sammenlignende Kulturforskning, Serie B}, Title = {{T}he {B}urushaski {L}anguage. {V}olume 1: {I}ntroduction and {G}rammar}, Volume = {29}, Wals_code = {bur}, Year = {1935} } @book{Love-2000, Address = {München}, Author = {.}, Iso_code = {unp}, Olac_field = {semantics; syntax; typology; general_linguistics}, Publisher = {Lincom Europa}, Title = {{T}he {G}rammatical {S}tructure of the {W}orora {L}anguage of {N}orth-{W}estern {A}ustralia}, Wals_code = {wor}, Year = {2000} } @mastersthesis{Loven-2001, Asjp_name = {Laven}, Author = {}, Besttxt = {ptxt2\eurasia\jacq_jruq2001.txt}, Cfn = {eurasia\jacq_jruq2001.pdf}, Class_loc = {PL4310}, Delivered = {eurasia\jacq_jruq2001.pdf}, Document_type = {CD}, Fn = {eurasia\jacq_jruq2001.pdf}, Hhtype = {grammar}, Inlg = {English [eng]}, Lgcode = {Laven [lbo]}, Macro_area = {Eurasia}, Mpi_eva_library_shelf = {ON ORDER}, Mpifn = {jruq_jacq2001.pdf}, Oclc = {805828193}, Pages = {xx+562}, Publisher = {MA Thesis, Australian National University}, School = {Australian National University}, Src = {asjp2010, hh, mpieva}, Subject_headings = {Laven language, Mon-Khmer languages-Laos, Laven language - Mon-Khmer languages-Laos}, Title = {{A} description of {J}ruq ({L}oven): a {M}on-{K}hmer language of the {L}ao {PDR}}, Wals_code = {lov}, Year = {2001} } @book{Lynch-1998, Address = {Honolulu}, Author = {}, Publisher = {University of Hawai'i Press}, Title = {{P}acific {L}anguages}, Wals_code = {cuu; poh; tng; uli}, Year = {1998} } @book{Macaulay-1996, Address = {Berkeley}, Author = {}, Iso_code = {mig}, Olac_field = {general_linguistics; phonology; semantics; syntax; phonetics; morphology; typology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{A} {G}rammar of {C}halcatongo {M}ixtec}, Volume = {127}, Wals_code = {mxc}, Year = {1996} } @book{Mahootian-1997, Address = {London}, Author = {}, Iso_code = {pes}, Olac_field = {morphology; typology; semantics; syntax; general_linguistics}, Publisher = {Routledge}, Series = {Descriptive Grammars}, Title = {{P}ersian}, Wals_code = {prs}, Year = {1997} } @book{Makaruku-et-al-1997, Address = {Ambon / Maluku}, Author = {Makaruku, et al.}, Iso_code = {alp}, Olac_field = {phonology; general_linguistics; phonetics; typology}, Publisher = {Summer Institute of Linguistics}, Title = {{K}amus {M}asyarakat. {I}ndonesia-{A}lune {A}lune-{I}ndonesia}, Wals_code = {aln}, Year = {1997} } @misc{Maku-2008, Author = {.}, Besttxt = {ptxt2\south_america\migliazza_maku2008_o.txt}, Fn = {south_america\migliazza_maku2008.pdf, south_america\migliazza_maku2008_o.pdf, south_america\migliazza_maku2008v2_o.pdf}, Hhtype = {grammar_sketch}, HowPublished = {Paper presented at the 4th Conference on Endangered Languages and Cultures of Native America, University of Utah}, Inlg = {English [eng]}, Lgcode = {Máku [xak]}, Macro_area = {South America}, Pages = {26}, Src = {hh}, Title = {{M}áku}, Wals_code = {mkw}, Year = {2008} } @article{Malecot-1963b, Author = {}, Iso_code = {lui}, Journal = {International Journal of American Linguistics}, Olac_field = {phonology; phonetics; general_linguistics; typology}, Pages = {89-95}, Title = {{L}uiseno, a structural analysis 1: {P}honology}, Volume = {29}, Wals_code = {lui}, Year = {1963} } @book{Manyambeang-et-al-1996, Address = {Jakarta}, Author = {. and . and Nasruddin}, Iso_code = {mak}, Olac_field = {general_linguistics; phonology; phonetics; typology}, Publisher = {Pusat Pembinaan dan Pengembangan Bahasa}, Title = {{T}ata {B}ahasa {M}akassar}, Wals_code = {mks}, Year = {1996} } @book{Marsack-1962, Address = {London}, Author = {.}, Iso_code = {smo}, Note = {first edition}, Olac_field = {morphology; semantics; syntax; general_linguistics; typology}, Publisher = {The English Universities Press Ltd}, Series = {Teach Yourself Books}, Title = {{S}amoan}, Wals_code = {sam}, Year = {1962} } @book{Martin-1991a, Address = {Rutland}, Author = {.}, Iso_code = {jpn}, Olac_field = {typology; semantics; syntax; general_linguistics}, Publisher = {Tuttle}, Series = {Tuttle language library}, Title = {{A} reference grammar of {J}apanese: a complete guide to the grammar and syntax of the japanese language}, Wals_code = {jpn}, Year = {1991} } @misc{Maslova-1999, Author = {}, Iso_code = {yux}, Note = {Published as Maslova 2003a}, Olac_field = {phonology; semantics; typology; morphology; syntax; general_linguistics; phonetics}, School = {University of Bielefeld}, Title = {{A} {G}rammar of {K}olyma {Y}ukaghir}, Wals_code = {yko}, Year = {1999} } @book{Mason-1918, Address = {Berkeley}, Author = {}, Iso_code = {sln}, Olac_field = {semantics; typology; morphology; syntax; general_linguistics}, Publisher = {University of California Press}, Title = {{T}he {L}anguage of the {S}alinan {I}ndians}, Wals_code = {sal}, Year = {1918} } @book{Matisoff-1973, Address = {Berkeley}, Author = {Matisoff, .}, Iso_code = {lhu}, Olac_field = {semantics; general_linguistics; phonology; syntax; phonetics; typology; morphology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{A} {G}rammar of {L}ahu}, Volume = {75}, Wals_code = {lah}, Year = {1973} } @book{Matteson-1965, Address = {Berkeley}, Author = {}, Iso_code = {pib}, Olac_field = {syntax; phonetics; general_linguistics; phonology; semantics; morphology; typology}, Publisher = {University of California Press}, Title = {{T}he {P}iro ({A}rawakan) {L}anguage}, Wals_code = {pir}, Year = {1965} } @incollection{Mbugu-1994, Address = {Amsterdam}, Author = {}, Besttxt = {ptxt2\africa\mous_maa-mbugu1994_o.txt}, Booktitle = {{M}ixed {L}anguages: 15 {C}ase {S}tudies in {L}anguage {I}ntertwining}, Cfn = {africa\mous_maa-mbugu1994_o.pdf}, Editor = { and }, Fn = {africa\mous_mbugu1994v2.pdf, africa\mous_mbugu1994.pdf, africa\mous_maa-mbugu1994_o.pdf, south_america\bakker-mous_15-mixed-languages1994_o.pdf, africa\mous_maa-mbugu1994.pdf}, Hhtype = {grammar_sketch}, Inlg = {English [eng]}, Keywords = {;eaf;tnz;lng;bnt;g.221;z.g.20a;}, Lgcode = {Mbugu [mhd]}, Macro_area = {Africa}, Pages = {175-200}, Publisher = {IFOTT}, Series = {Studies of Language and Language Use}, Src = {eballiso2009, hh, weball, zurich}, Subject_headings = {eaf, tnz, lng, bnt, g.221, z.g.20a}, Title = {{M}a'a or {M}bugu}, Volume = {13}, Wals_code = {mbg}, Year = {1994}, Zurichcode = {Maa} } @book{McGregor-1990, Address = {Amsterdam}, Author = {McGregor, William}, Publisher = {}, Title = {{A} {F}unctional {G}rammar of {G}ooniyandi}, Wals_code = {goo}, Year = {1990} } @book{McGregor-1993, Address = {München}, Author = {}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{G}unin / {K}wini}, Volume = {11}, Wals_code = {gnn}, Year = {1993} } @book{McGregor-1996, Address = {München}, Author = {McGregor, William}, Iso_code = {nyv}, Olac_field = {general_linguistics; typology; semantics; morphology; syntax}, Publisher = {Lincom Europa}, Series = {Languages of the World, Materials}, Title = {{N}yulnyul}, Volume = {88}, Wals_code = {nyu}, Year = {1996} } @book{Meri-2014, Address = {Kuala Lumpur}, Author = {Omar, }, Hhtype = {phonology;ethnographic;socling}, Inlg = {English [eng]}, Lgcode = {Mah Meri [mhe]}, Macro_area = {Eurasia}, Pages = {xxii+202}, Publisher = {University of Malaya Press}, Src = {hh}, Title = {{T}he {M}ah {M}eri language: an introduction}, Wals_code = {mhm}, Year = {2014} } @book{Merlan-1982, Address = {Amsterdam}, Author = {Merlan, .}, Iso_code = {mpc}, Olac_field = {phonetics; syntax; phonology; morphology; semantics; general_linguistics; typology}, Publisher = {North-Holland}, Series = {Lingua Descriptive Studies}, Title = {{M}angarayi}, Volume = {4}, Wals_code = {myi}, Year = {1982} } @book{Miller-1966, Address = {Berkeley / Los Angeles}, Author = {Miller, .}, Iso_code = {kjq}, Olac_field = {typology; general_linguistics; phonology; phonetics}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{A}coma {G}rammar and {T}exts}, Volume = {40}, Wals_code = {aco}, Year = {1966} } @book{Mintz-1994, Address = {Singapore}, Author = {Mintz, .}, Iso_code = {ind}, Olac_field = {general_linguistics; syntax; morphology; typology}, Publisher = {EPB publishers}, Title = {{A} {S}tudent's {G}rammar of {M}alay and {I}ndonesian}, Wals_code = {ind}, Year = {1994} } @book{Mixco-1997, Address = {München}, Author = {Mixco, .}, Iso_code = {mhq}, Olac_field = {general_linguistics; syntax; typology; semantics}, Publisher = {Lincom Europa}, Series = {Languages of the World/Materials}, Title = {{M}andan}, Volume = {159}, Wals_code = {mdn}, Year = {1997} } @incollection{Moag-2001, Address = {New York / Dublin}, Author = {.}, Booktitle = {{F}acts {A}bout the {W}orld's {L}anguages, {A}n {E}ncyclopedia of the {W}orld's {L}anguages: {P}ast and {P}resent}, Editor = { and }, Iso_code = {mal}, Olac_field = {typology; general_linguistics; morphology}, Pages = {459-462}, Publisher = {HW Wilson}, Title = {{M}alayalam}, Wals_code = {mym}, Year = {2001} } @misc{Moore-1984, Author = {}, School = {City University of New York}, Title = {{S}yntax of the {L}anguage of the {G}avião {I}ndians of {R}ondônia, {B}razil}, Wals_code = {gav}, Year = {1984} } @misc{Morgan-1991, Author = {}, Iso_code = {kut}, Olac_field = {morphology; typology; general_linguistics; phonology; phonetics}, School = {University of California, Berkeley}, Title = {{A} {D}escription of the {K}utenai {L}anguage}, Wals_code = {kut}, Year = {1991} } @article{Morgenstierne-1954, Author = {}, Iso_code = {wbk}, Journal = {Norsk Tidsskrift for Sprogvidenskap}, Olac_field = {general_linguistics; syntax; typology}, Pages = {146-219}, Title = {{T}he {W}aigali language}, Volume = {17}, Wals_code = {wgl}, Year = {1954} } @book{Morice-1932, Address = {Wien}, Author = {.}, Iso_code = {crx}, Olac_field = {general_linguistics; syntax; typology}, Publisher = {Anthropos}, Title = {{T}he {C}arrier {L}anguage}, Wals_code = {crq}, Year = {1932} } @book{Mosel-1980, Address = {Canberra}, Author = {}, Iso_code = {tpi}, Olac_field = {typology; semantics; general_linguistics; syntax}, Publisher = {The Australian National University}, Title = {{T}olai and {T}ok {P}isin}, Wals_code = {tla}, Year = {1980} } @book{Mottin-1978, Address = {Bangkok}, Author = {.}, Iso_code = {mww}, Olac_field = {general_linguistics; syntax; typology}, Publisher = {Don Bosco Press}, Title = {{E}léments de grammaire {H}mong {B}lanc}, Wals_code = {hmd}, Year = {1978} } @misc{Mous-1992, Author = {}, Iso_code = {irk}, Olac_field = {general_linguistics; typology; syntax}, School = {Rijksuniversiteit Leiden}, Title = {{A} {G}rammar of {I}raqw}, Wals_code = {irq}, Year = {1992} } @book{Moyse-Faurie-1983, Address = {Paris}, Author = {Moyse-Faurie, Claire}, Iso_code = {dhv}, Olac_field = {syntax; general_linguistics; typology; phonology; phonetics; semantics; morphology}, Publisher = {Société d'Études Linguistiques et Anthropologiques de France}, Title = {{L}e drehu, langue de {L}ifou ({Ì}les {L}oyauté)}, Wals_code = {dre}, Year = {1983} } @book{Muller-1954, Address = {Fosieux, Switzerland}, Author = {}, Iso_code = {kyx}, Olac_field = {syntax; typology; general_linguistics}, Publisher = {Anthropos}, Series = {Microbibliotheca Anthropos}, Title = {{G}rammar and {V}ocabulary of the {K}onua language}, Volume = {12}, Wals_code = {knu}, Year = {1954} } @article{Mushin-2005, Author = {}, Iso_code = {gbc}, Journal = {Australian Journal of Linguistics}, Olac_field = {morphology; syntax; general_linguistics; typology}, Pages = {253-273}, Title = {{W}ord order pragmatics and narrative functions in {G}arrwa}, Volume = {25}, Wals_code = {grr}, Year = {2005} } @book{Muthalib-et-al-1992, Address = {Jakarta}, Author = { .}, Iso_code = {mdr}, Olac_field = {syntax; typology; general_linguistics; semantics}, Publisher = {Pusat Pembinaan dan Pengembangan Bahasa}, Title = {{T}ata {B}ahasa {M}andar}, Wals_code = {mnr}, Year = {1992} } @book{Naess-2000, Address = {München}, Author = {}, Iso_code = {piv}, Olac_field = {syntax; general_linguistics; semantics; typology}, Publisher = {Lincom Europa}, Series = {Languages of the World, Materials}, Title = {{P}ileni}, Volume = {325}, Wals_code = {pil}, Year = {2000} } @book{Nagaraja-1999, Address = {Tokyo}, Author = {.}, Iso_code = {kfq}, Olac_field = {general_linguistics; typology; morphology; semantics; syntax}, Publisher = {Tokyo University of Foreign Studies}, Title = {{K}orku {L}anguage: {G}rammar, {T}exts and {V}ocabulary}, Wals_code = {kku}, Year = {1999} } @book{Nedjalkov-1997, Address = {London / New York}, Author = {}, Publisher = {Routledge}, Series = {Routledge Descriptive Grammars}, Title = {{E}venki}, Wals_code = {eve}, Year = {1997} } @book{Nehlil-1909, Address = {Paris}, Author = {.}, Iso_code = {thv}, Olac_field = {general_linguistics; typology; semantics; morphology; syntax}, Publisher = {Leroux}, Series = {Publications de l'école des lettres, Bulletin de correspondance africaine}, Title = {{É}tude sur le dialecte de {G}hat}, Volume = {38}, Wals_code = {tgh}, Year = {1909} } @book{Neukom-2001, Address = {München}, Author = {}, Iso_code = {sat}, Olac_field = {semantics; morphology; syntax; general_linguistics; typology}, Publisher = {Lincom Europa}, Title = {{S}antali}, Wals_code = {stl}, Year = {2001} } @incollection{Newman-1946, Address = {New York}, Author = {}, Booktitle = {{L}inguistic {S}tructures of {N}ative {A}merica}, Editor = {}, Iso_code = {yok}, Olac_field = {syntax; typology; general_linguistics; semantics}, Pages = {222-248}, Publisher = {Viking Fund}, Title = {{T}he {Y}awelmani dialect of {Y}okuts}, Wals_code = {ywl}, Year = {1946} } @incollection{Newman-1996, Address = {Washington}, Author = {}, Booktitle = {{H}andbook of {A}merican {I}ndians. {V}olume 17: {L}anguages}, Editor = {}, Iso_code = {zun}, Olac_field = {syntax; typology; semantics; general_linguistics}, Pages = {483-506}, Publisher = {Smithsonian Institute}, Title = {{S}ketch of the {Z}uni {L}anguage}, Wals_code = {zun}, Year = {1996} } @book{Newman-2000, Address = {New Haven}, Author = {}, Iso_code = {hau}, Olac_field = {typology; phonology; general_linguistics; phonetics; semantics; syntax; morphology}, Publisher = {Yale University Press}, Title = {{T}he {H}ausa {L}anguage: {A}n {E}ncyclopedic {R}eference {G}rammar}, Wals_code = {hau}, Year = {2000} } @book{Ngarinyman-2014, Address = {Berlin}, Author = { and }, Besttxt = {ptxt2\australia\meakins-nordlinger_bilinarra2014.txt}, Cfn = {australia\meakins-nordlinger_bilinarra2014.pdf}, Delivered = {australia\meakins-nordlinger_bilinarra2014.pdf}, Fn = {australia\meakins-nordlinger_bilinarra2014_o.pdf, australia\meakins-nordlinger_bilinarra2014.pdf}, Hhtype = {grammar}, ISBN = {9781614512684}, Inlg = {English [eng]}, Lgcode = {Bilinarra = Ngarinman [nbj]}, Macro_area = {Australia}, Oclc = {870589917}, Pages = {xxxiii+525}, Publisher = {De Gruyter Mouton}, Series = {Pacific Linguistics}, Src = {hh}, Title = {{A} {G}rammar of {B}ilinarra: {A}n {A}ustralian {A}boriginal {L}anguage of the {N}orthern {T}erritory}, Volume = {640}, Wals_code = {ngy}, Year = {2014} } @incollection{Nichols-1994a, Address = {Delmar, New York}, Author = {.}, Booktitle = {{T}he {I}ndigenous {L}anguages of the {C}aucasus 4}, Editor = {}, Iso_code = {che}, Olac_field = {syntax; semantics; general_linguistics; morphology; typology}, Pages = {1-77}, Publisher = {Caravan Books}, Title = {{C}hechen}, Wals_code = {chc}, Year = {1994} } @book{Nicklas-1979, Address = {Durant, Oklahoma}, Author = {}, Iso_code = {cho}, Olac_field = {semantics; morphology; syntax; typology; general_linguistics}, Publisher = {Choctaw Bilingual Education Program, Southeastern Oklahoma State University}, Title = {{R}eference {G}rammar to the {C}hoctaw {L}anguage}, Wals_code = {cct}, Year = {1979} } @book{Noonan-1992, Address = {Berlin}, Author = {}, Iso_code = {laj}, Olac_field = {phonology; typology; morphology; phonetics; syntax; semantics; general_linguistics}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {L}ango}, Volume = {7}, Wals_code = {lan}, Year = {1992} } @incollection{Noonan-2003a, Address = {London}, Author = {}, Booktitle = {{T}he {S}ino-{T}ibetan {L}anguages}, Editor = { and .}, Pages = {336-352}, Publisher = {Routledge}, Title = {{N}ar-{P}hu}, Wals_code = {ath}, Year = {2003} } @incollection{Noonan-2003b, Address = {London}, Author = {}, Booktitle = {{T}he {S}ino-{T}ibetan {L}anguages}, Editor = { and LaPolla, .}, Iso_code = {chx}, Olac_field = {typology; general_linguistics; syntax; semantics}, Pages = {315-335}, Publisher = {Routledge}, Title = {{C}hantyal}, Wals_code = {chn}, Year = {2003} } @phdthesis{Ocaina-2004, Address = {Paris}, Author = {}, Document_type = {B}, Hhtype = {grammar (computerized assignment from "descriptive")}, Inlg = {French [fra]}, Lgcode = {Ocaina [oca]}, Macro_area = {South America}, Mpi_eva_library_shelf = {DESIDERAT}, Note = {Paris 7}, Publisher = {Univ.}, School = {University of Paris}, Src = {mpieva}, Title = {{L}a langue ocaina (sous-famille {U}itoto): première approche descriptive de la variété dialectale uvohsa}, Wals_code = {oca}, Year = {2004} } @book{Okell-1969, Address = {London}, Author = {}, Publisher = {Oxford University Press.}, Title = {{A} {R}eference {G}rammar of {C}olloquial {B}urmese (two volumes)}, Wals_code = {brm}, Year = {1969} } @book{Omar-1983, Address = {Kuala Lumpur}, Author = {}, Publisher = {Dewan Bahasa dan Pustaka}, Title = {{T}he {M}alay {P}eoples of {M}alaysia and {T}heir {L}anguages}, Wals_code = {bit; kbr; klt; lud; mel; nrm; kua; tim}, Year = {1983} } @book{Onya-1999, Address = {Pontianak}, Author = {}, Hhtype = {ethnographic;wordlist}, Inlg = {Indonesian [ind]}, Lgcode = {Semandang [sdm]}, Macro_area = {Papunesia}, Publisher = {Institute of Dayakology Research and Development}, Src = {hh}, Title = {{S}uku dan bahasa {D}ayak di {K}abupaten {K}etapang}, Wals_code = {smd}, Year = {1999} } @book{Osborne-1974, Address = {Canberra}, Author = {.}, Iso_code = {tiw}, Olac_field = {phonetics; typology; semantics; general_linguistics; phonology; syntax; morphology}, Publisher = {Australian Institute of Aboriginal Studies}, Series = {Australian Aboriginal Studies}, Title = {{T}he {T}iwi {L}anguage}, Volume = {55}, Wals_code = {tiw}, Year = {1974} } @inproceedings{Ostler-1994, Address = {Newark}, Author = {}, Booktitle = {{L}anguage in the {A}ndes, {L}atin {A}merican {S}tudies}, Editor = {. and . and .}, Iso_code = {chb}, Note = {Based on papers from a conference entitled 'International Conference on Language, Language Policy and Education in the Andes' held at the University of Delaware, Oct. 28-30, 1991}, Olac_field = {syntax; general_linguistics; semantics; typology}, Pages = {205-230}, Publisher = {University of Delaware}, Series = {Occasional Monographs in Latin American Studies}, Title = {{S}yntactic {T}ypology of {M}uisca - a {S}ketch}, Volume = {4}, Wals_code = {msc}, Year = {1994} } @book{Owens-1985, Address = {Hamburg}, Author = {}, Iso_code = {hae}, Olac_field = {typology; general_linguistics; phonology; phonetics; morphology; syntax; semantics}, Publisher = {}, Series = {Kuschitische Sprachstudien, Cushitic Language Studies}, Title = {{A} {G}rammar of {H}arar {O}romo ({N}ortheastern {E}thiopia)}, Volume = {4}, Wals_code = {orh}, Year = {1985} } @book{Ozanne-Rivierre-1998, Address = {Paris}, Author = {}, Iso_code = {yly}, Olac_field = {syntax; general_linguistics; typology}, Publisher = {Peeters}, Title = {{L}e nyelâyu de {B}alade ({N}ouvelle-{C}aledonie)}, Wals_code = {iaa; nyl}, Year = {1998} } @book{Pandharipande-1997, Address = {London}, Author = {Pandharipande, .}, Iso_code = {mar}, Olac_field = {typology; morphology; semantics; syntax; general_linguistics}, Publisher = {Routledge}, Series = {Descriptive Grammar Series}, Title = {{M}arathi}, Wals_code = {mhi}, Year = {1997} } @book{Parks-1976, Address = {New York}, Author = {Parks, .}, Iso_code = {paw}, Olac_field = {syntax; morphology; typology; general_linguistics; semantics}, Publisher = {Garland Publishing}, Title = {{A} {G}rammar of {P}awnee}, Wals_code = {pwn}, Year = {1976} } @incollection{Payne-and-Payne-1990, Address = {Berlin}, Author = {. and }, Booktitle = {{H}andbook of {A}mazonian {L}anguages 2}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {yad}, Olac_field = {general_linguistics; semantics; morphology; phonetics; syntax; phonology; typology}, Pages = {249-474}, Publisher = {Mouton de Gruyter}, Title = {{Y}agua}, Wals_code = {yag}, Year = {1990} } @incollection{Peeke-1994, Address = {Amsterdam}, Author = {}, Booktitle = {{T}ypological {S}tudies in {N}egation}, Editor = {Kahrel, Peter and Berg, }, Iso_code = {auc}, Olac_field = {syntax; typology; general_linguistics; morphology}, Pages = {267-290}, Publisher = {John Benjamins}, Title = {{W}aorani}, Wals_code = {wao}, Year = {1994} } @incollection{Peterson-2003, Address = {London}, Author = {.}, Booktitle = {{T}he {S}ino-{T}ibetan languages}, Editor = { and LaPolla, .}, Iso_code = {cnh}, Olac_field = {typology; morphology; syntax; general_linguistics}, Pages = {409-426}, Publisher = {Routledge}, Title = {{H}akha {L}ai}, Wals_code = {lai}, Year = {2003} } @book{Pilhofer-1933, Address = {Berlin / Hamburg}, Author = {}, Iso_code = {kmg}, Journal = {Zeitschrift für Eingeborenen-Sprachen, Beiheft}, Note = {Reprinted 1969 by Kraus Reprint, Nendeln, Liechtenstein}, Olac_field = {syntax; general_linguistics; typology; semantics}, Publisher = {D. Reimer (Ernst Vohsen)}, Title = {{G}rammatik der {K}ate-{S}prache in {N}euguinea}, Volume = {14}, Wals_code = {kat}, Year = {1933} } @book{Pitkin-1984, Address = {Berkeley}, Author = {}, Iso_code = {wit}, Olac_field = {syntax; general_linguistics; phonology; phonetics; semantics; typology; morphology}, Publisher = {University of California Press}, Title = {{W}intu {G}rammar}, Wals_code = {win}, Year = {1984} } @incollection{Popjes-and-Popjes-1986, Address = {Berlin}, Author = { and Popjes, Jo}, Booktitle = {{H}andbook of {A}mazonian {L}anguages 1}, Editor = {Derbyshire, . and Pullum, .}, Iso_code = {xra; ram}, Olac_field = {syntax; semantics; morphology; typology; phonetics; phonology; general_linguistics}, Pages = {128-199}, Publisher = {Mouton de Gruyter}, Title = {{C}anela-{K}rahô}, Wals_code = {ckr}, Year = {1986} } @book{Poppe-1960, Address = {Bloomington}, Author = {}, Publisher = {Indiana University}, Series = {Indiana University Publications, Uralic and Altaic Series}, Title = {{B}uriat {G}rammar}, Volume = {2}, Wals_code = {but}, Year = {1960} } @book{Poppe-1968, Address = {Bloomington}, Author = {}, Iso_code = {tat}, Note = {second revised edition}, Olac_field = {general_linguistics; typology; syntax; semantics; morphology}, Publisher = {Indiana University Press}, Title = {{T}atar {M}anual. {D}escriptive {G}rammar and {T}exts with {T}atar-{E}nglish-{G}lossary}, Wals_code = {tvo}, Year = {1968} } @book{Poppe-1970, Address = {Washington}, Author = {}, Iso_code = {khk}, Olac_field = {semantics; phonology; syntax; phonetics; morphology; typology; general_linguistics}, Publisher = {The Center for Applied Linguistics}, Series = {Handbook Series}, Title = {{M}ongolian {L}anguage}, Volume = {4}, Wals_code = {kha}, Year = {1970} } @book{Premsrirat-1987, Address = {Canberra}, Author = {}, Iso_code = {kjg}, Olac_field = {general_linguistics; typology; phonology; morphology; syntax; semantics; phonetics}, Pages = {1-143}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series A}, Title = {{A} {K}hmu {G}rammar}, Volume = {75}, Wals_code = {kmu}, Year = {1987} } @book{Prost-1964g, Address = {Dakar}, Author = {.}, Booktitle = {{C}ontribution à l'{É}tude des langues voltaiques}, Iso_code = {wib; tsp}, Olac_field = {general_linguistics; syntax; typology; morphology; semantics}, Pages = {251-342}, Publisher = {Institut Français d’Afrique Noire (IFAN)}, Series = {Memoires de l'Institut Français d'Afrique Noire}, Title = {{L}e toussian}, Volume = {70}, Wals_code = {tow}, Year = {1964} } @book{Purepecha-2000, Address = {München}, Author = {}, Besttxt = {ptxt2\north_america\chamoreau_purepecha2000v2_o.txt}, Cfn = {north_america\chamoreau_purepecha2000v2_o.pdf}, Fn = {north_america\chamoreau_purepecha2000v2_o.pdf, north_america\chamoreau_purepecha2000_o.pdf}, Hhtype = {grammar}, ISBN = {9783895869419}, Inlg = {French [fra]}, Lgcode = {Purepecha [tsz]}, Macro_area = {North America}, Oclc = {747256941}, Pages = {viii+336}, Publisher = {Lincom}, Series = {LINCOM Studies in Native American Linguistics}, Src = {hh}, Title = {{G}rammaire du {P}urépecha}, Volume = {34}, Wals_code = {pur}, Year = {2000} } @book{Quesada-2000, Address = {München}, Author = {}, Iso_code = {tfr}, Olac_field = {syntax; typology; general_linguistics; phonetics; semantics; phonology; morphology}, Publisher = {Lincom Europa}, Title = {{A} {G}rammar of {T}eribe}, Wals_code = {trb}, Year = {2000} } @book{Quirk-et-al-1985, Address = {London}, Author = { and Greenbaum, Sidney and Leech, Geoffrey and }, Iso_code = {eng}, Olac_field = {semantics; general_linguistics; typology; syntax}, Publisher = {Longman}, Title = {{A} {C}omprehensive {G}rammar of the {E}nglish {L}anguage}, Wals_code = {eng}, Year = {1985} } @book{Radin-1929, Address = {Berkeley}, Author = {}, Iso_code = {wao}, Olac_field = {syntax; general_linguistics; phonology; semantics; morphology; typology; phonetics}, Publisher = {University of California Press}, Series = {California University Publications in American Archaeology and Ethnology}, Title = {{A} {G}rammar of the {W}appo {L}anguage}, Volume = {27}, Wals_code = {wap}, Year = {1929} } @incollection{Rasoloson-and-Rubino-2004, Address = {London}, Author = { and }, Booktitle = {{T}he {A}ustronesian {L}anguages of {A}sia and {M}adagascar}, Editor = {. and .}, Iso_code = {plt}, Olac_field = {morphology; general_linguistics; typology}, Pages = {456-488}, Publisher = {Routledge Curzon}, Title = {{M}alagasy}, Wals_code = {mal}, Year = {2004} } @misc{Rau-1992, Author = {}, Iso_code = {tay}, Note = {Also published 1992 by Crane Publishing Co., Taipei}, Olac_field = {syntax; morphology; typology; semantics; general_linguistics}, School = {Cornell University}, Title = {{A} {G}rammar of {A}tayal}, Wals_code = {ata}, Year = {1992} } @book{Ray-1933, Address = {Port Moresby, Papua}, Author = {}, Iso_code = {kiw; kjd}, Olac_field = {morphology; semantics; general_linguistics; syntax; typology}, Publisher = {Ed, Government Printer}, Title = {{A} {G}rammar of the {K}iwai {L}anguage, {F}ly {D}elta, {P}apua. {W}ith a {K}iwai {V}ocabulary by {R}ev. {E}. {B}axter {R}iley}, Wals_code = {kiw}, Year = {1933} } @incollection{Reesink-1996, Address = {Jakarta}, Author = {Reesink, .}, Booktitle = {{S}tudies in {I}rian {L}anguages 1}, Editor = {.}, Iso_code = {had; ayz; kps}, Olac_field = {semantics; general_linguistics; typology; syntax}, Pages = {1-20}, Publisher = {Universitas Katolik Indonesia Atma Jaya}, Series = {NUSA - Linguistic Studies of Indonesian and Other Languages of Indonesia}, Title = {{M}orphosyntactic {F}eatures of the {B}ird's {H}ead {L}anguages}, Volume = {40}, Wals_code = {tht}, Year = {1996} } @book{Reesink-1999, Address = {Canberra}, Author = {Reesink, .}, Iso_code = {had}, Olac_field = {typology; general_linguistics; morphology; semantics; syntax}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar of {H}atam, {B}ird's {H}ead {P}eninsula, {I}rian {J}aya}, Volume = {146}, Wals_code = {hat}, Year = {1999} } @book{Refsing-1986, Address = {Aarhus}, Author = {}, Publisher = {Aarhus University Press}, Title = {{T}he {A}inu {L}anguage: {T}he {M}orphology and {S}yntax of the {S}hizunai {D}ialect}, Wals_code = {ain}, Year = {1986} } @book{Rennison-1997, Address = {London}, Author = {.}, Iso_code = {kfz}, Olac_field = {morphology; phonology; phonetics; semantics; typology; syntax; general_linguistics}, Publisher = {Routledge}, Series = {Descriptive Grammar Series}, Title = {{K}oromfe}, Wals_code = {kfe}, Year = {1997} } @book{Rischel-1995, Address = {Copenhagen}, Author = {}, Iso_code = {mra}, Olac_field = {typology; semantics; general_linguistics; syntax}, Publisher = {Museum Tusculanum Press}, Title = {{M}inor {M}labri: {A} {H}unter-{G}atherer {L}anguage of {N}orthern {I}ndochina}, Wals_code = {mlm}, Year = {1995} } @book{Roberts-1987, Address = {London}, Author = {.}, Iso_code = {aey}, Olac_field = {morphology; syntax; semantics; phonetics; typology; general_linguistics; phonology}, Publisher = {Croom Helm}, Series = {Croom Helm Descriptive Grammar Series}, Title = {{A}mele}, Wals_code = {ame}, Year = {1987} } @book{Robson-1992, Address = {Clayton}, Author = {}, Iso_code = {jav}, Olac_field = {typology; syntax; general_linguistics; semantics}, Publisher = {Centre of Southeast Asian Studies, Monash University}, Series = {Monash papers on Southeast Asia}, Title = {{J}avanese {G}rammar for {S}tudents}, Volume = {26}, Wals_code = {jav}, Year = {1992} } @inbook{Ross-1980, Address = {Canberra}, Author = {}, Booktitle = {{P}apers in {N}ew {G}uinea {L}inguistics 20}, Iso_code = {vam}, Olac_field = {semantics; morphology; phonetics; syntax; typology; general_linguistics; phonology}, Pages = {77-109}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series A}, Title = {{S}ome elements of {V}animo, a {N}ew {G}uinea tone language}, Volume = {56}, Wals_code = {dum}, Year = {1980} } @book{Rowlands-1959, Address = {London}, Author = {Rowlands, }, Iso_code = {mnk}, Olac_field = {typology; general_linguistics; semantics; syntax}, Publisher = {University of London}, Title = {{A} grammar of {G}ambian {M}andinka}, Wals_code = {mdk}, Year = {1959} } @book{Rubongoya-1999, Address = {Köln}, Author = {.}, Iso_code = {nyo}, Olac_field = {general_linguistics; typology; semantics; syntax}, Publisher = {Köppe}, Series = {East African Languages and Dialects}, Title = {{A} {M}odern {R}unyoro-{R}utooro {G}rammar}, Volume = {9}, Wals_code = {rru}, Year = {1999} } @book{Rumsey-1982, Address = {Canberra}, Author = {.}, Iso_code = {ung}, Olac_field = {morphology; syntax; semantics; general_linguistics; typology; phonology; phonetics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{A}n {I}ntra-{S}entence {G}rammar of {U}ngarinjin, {N}orth-{W}estern {A}ustralia}, Volume = {86}, Wals_code = {ung}, Year = {1982} } @misc{Sakel-2002a, Author = {}, Iso_code = {cas}, Olac_field = {syntax; semantics; typology; general_linguistics}, School = {University of Nijmegen}, Title = {{A} {G}rammar of {M}osetén}, Wals_code = {mos}, Year = {2002} } @incollection{Salminen-1998, Address = {London}, Author = {}, Booktitle = {{T}he {U}ralic {L}anguages}, Editor = {}, Iso_code = {yrk}, Olac_field = {semantics; phonology; typology; syntax; morphology; general_linguistics; phonetics}, Pages = {516-547}, Publisher = {Routledge}, Title = {{N}enets}, Wals_code = {nen}, Year = {1998} } @book{Samarin-1966, Address = {Berkeley}, Author = {.}, Iso_code = {gbp}, Olac_field = {semantics; typology; phonology; general_linguistics; syntax; phonetics; morphology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{T}he {G}beya {L}anguage: {G}rammar, {T}exts and {V}ocabularies}, Volume = {44}, Wals_code = {gbb}, Year = {1966} } @book{Sandahl-2000, Address = {Leuven}, Author = {.}, Iso_code = {hin}, Olac_field = {morphology; typology; general_linguistics}, Publisher = {Peeters}, Title = {{A} {H}indi {R}eference {G}rammar}, Wals_code = {hin}, Year = {2000} } @incollection{Sapir-1922b, Address = {Washington}, Author = {}, Booktitle = {{H}andbook of {A}merican {I}ndian {L}anguages 2}, Editor = {}, Iso_code = {tkm}, Olac_field = {general_linguistics; syntax; morphology; typology; semantics}, Pages = {1-296}, Publisher = {Government Printing Office}, Series = {Smithsonian Institution Bureau of American Ethnology Bulletin}, Title = {{T}he {T}akelma {L}anguage of {S}outhwestern {O}regon}, Volume = {40}, Wals_code = {tkl}, Year = {1922} } @article{Sapir-1930, Author = {}, Iso_code = {ute}, Journal = {Proceedings of the American Society of Arts and Sciences}, Note = {reprinted 1992, The Collected Works of Edward Sapir, ed. by : }, Olac_field = {phonetics; semantics; general_linguistics; phonology; syntax; morphology; typology}, Pages = {1-3}, Title = {{S}outhern {P}aiute: a {S}hoshonean {L}anguage}, Volume = {65}, Wals_code = {put}, Year = {1930} } @book{Scatton-1984, Address = {Columbus, Ohio}, Author = {.}, Iso_code = {bul}, Olac_field = {syntax; general_linguistics; phonetics; typology; semantics; morphology; phonology}, Publisher = {Slavica Publishers}, Title = {{A} {R}eference {G}rammar of {M}odern {B}ulgarian}, Wals_code = {bul}, Year = {1984} } @book{Schachter-and-Otanes-1983, Address = {Berkeley}, Author = { and Otanes, .}, Iso_code = {tgl}, Note = {reprinted edition}, Olac_field = {syntax; general_linguistics; typology}, Publisher = {University of California Press}, Title = {{T}agalog {R}eference {G}rammar}, Wals_code = {tag}, Year = {1983} } @book{Schadeberg-and-Elias-1979, Address = {Tervuren, Belgium}, Author = {Schadeberg, . and }, Iso_code = {ras}, Olac_field = {morphology; syntax; typology; general_linguistics}, Publisher = {Musee Royal de L'Afrique Centrale}, Title = {{A} {D}escription of the {O}rig {L}anguage ({S}outhern {K}ordofan), {B}ased on the {N}otes of {F}r. {C}arlo {M}uratori}, Wals_code = {ori}, Year = {1979} } @book{Schiffman-1983, Address = {Seattle}, Author = {.}, Iso_code = {kan}, Olac_field = {morphology; semantics; syntax; general_linguistics; typology}, Publisher = {University of Washington Press}, Title = {{A} {R}eference {G}rammar of {S}poken {K}annada}, Wals_code = {knd}, Year = {1983} } @misc{Schuh-1972, Author = {Schuh, .}, Iso_code = {ngi}, Olac_field = {syntax; morphology; semantics; general_linguistics; typology}, School = {University of California at Los Angeles}, Title = {{A}spects of {N}gizim {S}yntax}, Wals_code = {ngz}, Year = {1972} } @book{Schuh-1998, Address = {Berkeley}, Author = {Schuh, .}, Iso_code = {mkf}, Olac_field = {syntax; semantics; general_linguistics; typology; morphology}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{A} {G}rammar of {M}iya}, Volume = {130}, Wals_code = {miy}, Year = {1998} } @misc{Schultze-Berndt-2000, Author = {Schultze-Berndt, Eva}, Iso_code = {djd}, Olac_field = {syntax; semantics; typology; general_linguistics}, School = {Katholieke Universiteit Nijmegen}, Title = {{S}imple and {C}omplex {V}erbs in {J}aminjung: {A} {S}tudy of {E}vent {C}ategorization in an {A}ustralian {L}anguage}, Wals_code = {jam}, Year = {2000} } @book{Schutz-1985, Address = {Honolulu}, Author = {Schütz, .}, Publisher = {University of Hawaii Press}, Title = {{T}he {F}ijian {L}anguage}, Wals_code = {fij}, Year = {1985} } @book{Seiler-1985, Address = {Canberra}, Author = {}, Iso_code = {imn}, Olac_field = {typology; semantics; morphology; phonology; syntax; phonetics; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{I}monda, a {P}apuan {L}anguage}, Volume = {93}, Wals_code = {imo}, Year = {1985} } @book{Sem-1975, Address = {Leningrad}, Author = {.}, Iso_code = {gld}, Olac_field = {typology; general_linguistics; syntax}, Publisher = {Nauka}, Title = {{O}cherki dialektov nanajsogo jazyka: {B}ikinskij (ussurijskij) dialekt}, Wals_code = {nai}, Year = {1975} } @book{Senft-1952, Address = {Berlin}, Author = {}, Iso_code = {kij}, Olac_field = {general_linguistics; syntax; semantics; typology}, Publisher = {Mouton de Gruyter}, Title = {{K}ilivila. {T}he {L}anguage of the {T}robriand {I}slanders}, Wals_code = {klv}, Year = {1952} } @incollection{Shepherd-and-Shepherd-1973, Address = {Norman}, Author = { and }, Booktitle = {{C}lause, {S}entence, and {D}iscourse {P}atterns in {S}elected {L}anguages of {N}epal. {P}art 3: {T}exts}, Editor = {}, Iso_code = {mgp}, Olac_field = {semantics; syntax; typology; general_linguistics}, Pages = {301-434}, Publisher = {Summer Institute of Linguistics}, Title = {{M}agar texts}, Wals_code = {msy}, Year = {1973} } @book{Shipley-1964, Address = {Berkeley}, Author = {Shipley, .}, Iso_code = {nmu}, Olac_field = {semantics; general_linguistics; morphology; typology; syntax; phonology; phonetics}, Publisher = {University of California Press}, Title = {{M}aidu {G}rammar}, Wals_code = {mne}, Year = {1964} } @mastersthesis{Shuar-2014, Author = {}, Besttxt = {ptxt2\south_america\saad_shuar2014_s.txt}, Cfn = {south_america\saad_shuar2014_s.pdf}, Delivered = {south_america\saad_shuar2014_s.pdf}, Fn = {south_america\saad_shuar2014_s.pdf, south_america\saad_shuar2014_o.pdf}, Hhtype = {grammar_sketch}, Inlg = {English [eng]}, Lgcode = {Shuar [jiv]}, Macro_area = {South America}, Pages = {188}, School = {Radboud Universiteit Nijmegen}, Src = {hh}, Title = {{A} sketch grammar of {S}huar}, Wals_code = {jiv}, Year = {2014} } @misc{Silver-1966, Author = {}, Iso_code = {sht}, Olac_field = {typology; phonetics; phonology; general_linguistics; syntax}, School = {University of California, Berkeley}, Title = {{T}he {S}hasta {L}anguage}, Wals_code = {shs}, Year = {1966} } @incollection{Snapp-et-al-1982, Address = {Dallas}, Author = { and and }, Booktitle = {{U}to-{A}ztecan {G}rammatical {S}ketches}, Editor = {.}, Iso_code = {pao}, Olac_field = {general_linguistics; typology; semantics; morphology; syntax}, Pages = {1-92}, Publisher = {Summer Institute of Linguistics and University of Arlington}, Series = {Studies in Uto-Aztecan Grammar}, Title = {{N}orthern {P}aiute}, Volume = {3}, Wals_code = {pno}, Year = {1982} } @book{Solnit-1997, Address = {Honolulu}, Author = {.}, Iso_code = {eky}, Olac_field = {semantics; general_linguistics; typology; phonology; syntax; morphology; phonetics}, Publisher = {University of Hawaii Press}, Title = {{E}astern {K}ayah {L}i: {G}rammar, {T}exts, {G}lossary}, Wals_code = {kyl}, Year = {1997} } @book{Stappers-1973, Address = {Tervuren}, Author = {}, Iso_code = {zmq}, Olac_field = {semantics; typology; general_linguistics; syntax}, Publisher = {Musée royal de l'Afrique centrale}, Series = {Musee royal de l'Afrique centrale. Annales}, Title = {{E}squisse de la langue mituku}, Volume = {80}, Wals_code = {mit}, Year = {1973} } @unpublished{Stebbins-2002, Author = {}, Iso_code = {gcc; tuh}, Note = {Paper presented at the 5th International Conference on Oceanic Linguistics, Australian National University, Canberra, 14 January 2002.}, Olac_field = {general_linguistics; semantics; syntax; typology}, Title = {{N}oun {C}lass {M}arkers in {M}ali-{B}aining}, Type = {paper}, Wals_code = {mli; tll}, Year = {2002} } @incollection{Stilo-2004, Address = {Amsterdam}, Author = {}, Booktitle = {{C}oordinating {C}onstructions}, Editor = {.}, Iso_code = {vaf}, Olac_field = {typology; general_linguistics; syntax}, Pages = {269-330}, Publisher = {Benjamins}, Title = {{C}oordination in three {W}estern {I}ranian {L}anguages: {V}afsi, {P}ersian and {G}ilaki}, Wals_code = {vaf}, Year = {2004} } @book{Suarez-1983b, Address = {Cambridge}, Author = {.}, Iso_code = {jai; tzh; yua; jac; top; tsz; tzb}, Olac_field = {syntax; general_linguistics; typology; semantics}, Publisher = {Cambridge University Press}, Title = {{T}he {M}esoamerican {I}ndian {L}anguages}, Wals_code = {jak; tpa; tze; yct}, Year = {1983} } @book{Sulkala-and-Karjalainen-1992, Address = {London}, Author = {Sulkala, Helena and Karjalainen, Merja}, Iso_code = {fin}, Olac_field = {general_linguistics; typology; syntax; semantics; morphology; phonology; phonetics}, Publisher = {Routledge}, Series = {Descriptive Grammar Series}, Title = {{F}innish}, Wals_code = {fin}, Year = {1992} } @incollection{Swadesh-1946b, Address = {New York}, Author = {}, Booktitle = {{L}inguistic {S}tructures of {N}ative {A}merica}, Editor = {}, Iso_code = {ctm}, Note = {Johnson Reprint Corporation, New York}, Olac_field = {morphology; semantics; typology; syntax; general_linguistics}, Pages = {312-326}, Publisher = {Viking Fund Publications in Anthropology}, Title = {{C}hitimacha}, Wals_code = {ctm}, Year = {1946} } @article{Swanton-1929, Author = {Swanton, .}, Doi = {10.1086/463777}, Iso_code = {aqp}, Journal = {International Journal of American Linguistics}, Number = {2-4}, Olac_field = {typology; morphology; semantics; general_linguistics; syntax}, Pages = {121-149}, Title = {{A} sketch of the {A}takapa language}, Volume = {5}, Wals_code = {atk}, Year = {1929} } @book{Swanton-1991, Author = {.}, Iso_code = {ncz}, Note = {Edited and annotated by . Edited version of Smithsonian Institution Manuscript 4142}, Olac_field = {syntax; general_linguistics; typology; semantics}, Title = {{A} {G}rammatical {S}ketch of the {N}atchez {L}anguage 1918}, Wals_code = {hai}, Year = {1991} } @book{Taljaard-et-al-1991, Address = {Pretoria}, Author = {. and . and .}, Iso_code = {ssw}, Olac_field = {general_linguistics; typology; semantics; syntax}, Publisher = {}, Title = {{H}andbook of {S}iswati}, Wals_code = {swt}, Year = {1991} } @inbook{Tharp-1996, Address = {Ukarumpa, Papua New Guinea}, Author = {}, Booktitle = {{T}wo {N}on-{A}ustronesian {G}rammars from the {I}slands}, Editor = {.}, Iso_code = {sua}, Note = {also available online: http://www.sil.org/pacific/png/pubs/37102/Sulka_Grammar.pdf}, Olac_field = {general_linguistics; morphology; typology; phonetics; phonology; syntax; semantics}, Pages = {77-179}, Publisher = {Summer Institute of Linguistics}, Series = {Data Papers on Papua New Guinea Languages}, Title = {{S}ulka {G}rammar {E}ssentials}, Volume = {42}, Wals_code = {sul}, Year = {1996} } @book{Thiesen-1996, Address = {Lima}, Author = {}, Iso_code = {boa}, Olac_field = {syntax; general_linguistics; typology}, Publisher = {Summer Institute of Linguistics}, Title = {{G}ramatica del idioma {B}ora}, Wals_code = {bor}, Year = {1996} } @book{Thompson-1965, Address = {Seattle}, Author = {.}, Iso_code = {vie}, Olac_field = {typology; phonology; syntax; semantics; phonetics; general_linguistics; morphology}, Publisher = {University of Washington Press}, Title = {{A} {V}ietnamese {G}rammar}, Wals_code = {vie}, Year = {1965} } @incollection{Thompson-1976a, Address = {East Lansing}, Author = {}, Booktitle = {{T}he {N}on-{S}emitic {L}anguages of {E}thiopia}, Editor = {}, Iso_code = {nrb}, Olac_field = {typology; syntax; phonology; semantics; phonetics; general_linguistics; morphology}, Pages = {484-494}, Publisher = {African Studies Center, Michigan State University}, Title = {{N}era}, Wals_code = {nar}, Year = {1976} } @book{Thompson-and-Thompson-1992, Address = {Missoula, Montana}, Author = {. and }, Iso_code = {thp}, Olac_field = {general_linguistics; syntax; typology; semantics}, Publisher = {Linguistics Laboratory, University of Montana}, Series = {University of Montana Occasional Papers in Linguistics}, Title = {{T}he {T}hompson {L}anguage}, Volume = {8}, Wals_code = {tho}, Year = {1992} } @book{Topping-1973, Address = {Honolulu}, Author = {. (with the assistance of )}, Iso_code = {cha}, Note = {reprinted in 1980}, Olac_field = {typology; phonology; general_linguistics; semantics; phonetics; morphology; syntax}, Publisher = {University of Hawaii Press}, Title = {{C}hamorro {R}eference {G}rammar}, Wals_code = {cha}, Year = {1973} } @book{Traill-1994, Address = {Köln}, Author = {}, Iso_code = {nmn}, Olac_field = {general_linguistics; semantics; typology; phonetics; phonology; syntax}, Publisher = {Rüdiger Köppe Verlag}, Title = {{A} !{X}óõ {D}ictionary}, Wals_code = {xoo}, Year = {1994} } @book{Tucker-1994, Address = {Köln}, Author = {Tucker, .}, Iso_code = {luo}, Olac_field = {general_linguistics; semantics; typology; syntax}, Publisher = {Rüdiger Köppe Verlag}, Title = {{A} {G}rammar of {K}enya {L}uo ({D}holuo)}, Wals_code = {luo}, Year = {1994} } @book{Tucker-and-Mpaayei-1955, Address = {London}, Author = {. and }, Iso_code = {mas}, Olac_field = {syntax; semantics; morphology; general_linguistics; phonology; typology; phonetics}, Publisher = {Longmans, Green \& Co}, Series = {Publications of the African Institute, Leyden}, Title = {{A} {M}aasai {G}rammar with {V}ocabulary}, Volume = {2}, Wals_code = {maa}, Year = {1955} } @incollection{Tuggy-1979, Address = {Arlington}, Author = {}, Booktitle = {{M}odern {A}ztec {G}rammatical {S}ketches}, Editor = {.}, Iso_code = {nhg}, Olac_field = {syntax; semantics; phonetics; phonology; general_linguistics; morphology; typology}, Pages = {1-140}, Publisher = {Summer Institute of Linguistics}, Series = {Studies in Uto-Aztecan Grammar}, Title = {{T}etelcingo {N}ahuatl}, Volume = {2}, Wals_code = {nht}, Year = {1979} } @book{Tung-1964, Address = {Taipei}, Author = {}, Iso_code = {tsu}, Olac_field = {general_linguistics; phonology; phonetics; morphology; typology}, Publisher = {Institute of History and Philology, Academia Sinica}, Series = {Special Publications}, Title = {{A} {D}escriptive {S}tudy of the {T}sou {L}anguage, {F}ormosa}, Volume = {48}, Wals_code = {tso}, Year = {1964} } @article{Valenzuela-et-al-2001, Author = { . and .}, Doi = {10.1017/S0025100301002109}, Iso_code = {shp}, Journal = {Journal of the International Phonetic Association}, Olac_field = {phonetics; phonology; general_linguistics; typology}, Pages = {287-290}, Title = {{S}hipibo}, Volume = {31}, Wals_code = {shk}, Year = {2001} } @book{Velie-and-Velie-1981, Address = {Yarinacocha, Pucallpa, Perú}, Author = { and }, Iso_code = {ore}, Olac_field = {syntax; general_linguistics; typology}, Publisher = {Instituto Lingüístico de Verano}, Title = {{V}ocabulario {O}rejon}, Wals_code = {ore}, Year = {1981} } @book{Wade-1992, Address = {Oxford}, Author = {Wade, .}, Iso_code = {rus}, Note = {reprinted in 1995}, Olac_field = {phonology; semantics; phonetics; morphology; typology; general_linguistics; syntax}, Publisher = {Blackwell}, Title = {{A} {C}omprehensive {R}ussian {G}rammar}, Wals_code = {rus}, Year = {1992} } @book{Wagner-1934, Address = {New York, NY}, Author = {}, Iso_code = {yuc}, Note = {Extract from the Handbook of American Indian Languages 3, ed. by F. Boas}, Olac_field = {semantics; typology; morphology; syntax; general_linguistics}, Publisher = {Columbia University Press}, Title = {{Y}uchi}, Wals_code = {yuc}, Year = {1934} } @book{Walker-1982, Address = {Jakarta}, Author = {.}, Iso_code = {hvn}, Olac_field = {typology; general_linguistics; semantics; syntax}, Publisher = {Universitas Atma Jaya}, Series = {NUSA Linguistic Studies in Indonesian and Languages in Indonesia}, Title = {{G}rammar of {S}awu}, Volume = {13}, Wals_code = {saw}, Year = {1982} } @book{Waterhouse-1962, Address = {Bloomington}, Author = {}, Iso_code = {chd; clo}, Olac_field = {general_linguistics; syntax; morphology; typology}, Publisher = {Indiana University Press}, Series = {Indiana University Research Center in Anthropology, Folklore and Linguistics Memoir}, Title = {{T}he {G}rammatical {S}tructure of {O}axaca {C}hontal}, Volume = {19}, Wals_code = {chx}, Year = {1962} } @book{Waterhouse-1963, Address = {Norman}, Author = {Waterhouse, .}, Country = {Peru [PE]}, Hhtype = {comparative (computerized assignment from "languages")}, Lgcode = {Aguaruna [agr], Arabela [arl], Candoshi-Shapra [cbu], Murui Huitoto [huu], Iquito [iqu], Cashinahua [cbs], Machiguenga [mcb]}, Macro_area = {South America}, Pages = {vii+220}, Publisher = {Summer Institute of Linguistics of the University of Oklahoma}, Series = {Summer Institute of Linguistics Publications in Linguistics and Related Fields}, Sil_id = {10023}, Src = {sil16}, Subject = {Tagmemics [TAG]}, Title = {{S}tudies in {P}eruvian {I}ndian languages 1}, Url = {http://www.sil.org/acpub/repository/10023_front.pdf, http://www.sil.org/acpub/repository/10023.pdf}, Volume = {9}, Wals_code = {arb}, Year = {1963} } @misc{Watkins-1980, Author = {Watkins, .}, Iso_code = {kio}, Olac_field = {general_linguistics; syntax; typology}, School = {University of Kansas}, Title = {{A} grammar of {K}iowa}, Wals_code = {kio}, Year = {1980} } @book{Watters-2002, Address = {Cambridge}, Author = {Watters, .}, Iso_code = {kgj}, Olac_field = {semantics; morphology; typology; syntax; general_linguistics}, Publisher = {Cambridge University Press}, Title = {{A} {G}rammar of {K}ham}, Wals_code = {kmh}, Year = {2002} } @book{Weber-1989, Address = {Berkeley}, Author = {Weber, }, Iso_code = {qvh}, Olac_field = {morphology; typology; semantics; general_linguistics; syntax}, Publisher = {University of California Press}, Series = {University of California Publications in Linguistics}, Title = {{A} {G}rammar of {H}uallaga ({H}uánaco) {Q}uechua}, Volume = {112}, Wals_code = {qhu}, Year = {1989} } @book{Welmers-1973, Address = {Berkeley / Los Angeles}, Author = {Welmers, .}, Iso_code = {xpe; ibo; yor; swh; lug}, Olac_field = {phonology; semantics; typology; general_linguistics; phonetics; syntax}, Publisher = {University of California Press}, Title = {{A}frican {L}anguage {S}tructures}, Wals_code = {kpe; lda; swa; yor}, Year = {1973} } @book{West-1980, Address = {Bogotá}, Author = {}, Iso_code = {tuo}, Olac_field = {semantics; general_linguistics; morphology; typology; syntax}, Publisher = {Summer Institute of Linguistics}, Title = {{G}ramática {P}opular del {T}ucano}, Wals_code = {tuc}, Year = {1980} } @misc{Wheeler-1970, Author = {.}, Iso_code = {snn}, Olac_field = {syntax; general_linguistics; morphology; typology}, School = {University of California at Berkeley}, Title = {{G}rammar of the {S}iona {L}anguage, {C}olombia, {S}outh {A}merica}, Wals_code = {sin}, Year = {1970} } @incollection{Will-1989, Address = {Hamburg}, Author = {}, Booktitle = {{T}opics in {N}ilo-{S}aharan {L}inguistics}, Editor = {}, Iso_code = {mym}, Olac_field = {morphology; syntax; general_linguistics; semantics; typology}, Pages = {129-150}, Publisher = {Helmut Buske Verlag}, Title = {{S}ketch of {M}e'en syntax}, Wals_code = {mee}, Year = {1989} } @book{Willett-1991, Address = {Arlington}, Author = {.}, Iso_code = {stp}, Olac_field = {phonology; general_linguistics; typology; syntax; phonetics; semantics}, Publisher = {Summer Institute of Linguistics and University of Texas at Arlington.}, Title = {{A} {R}eference {G}rammar of {S}outheastern {T}epehuan}, Wals_code = {tps}, Year = {1991} } @book{Wilson-1974, Address = {Ukarumpa, Papua New Guinea}, Author = {}, Iso_code = {sue}, Olac_field = {semantics; syntax; general_linguistics; phonetics; phonology; morphology; typology}, Publisher = {Summer Institute of Linguistics}, Series = {Workpapers in Papua New Guinea Languages}, Title = {{S}uena {G}rammar}, Volume = {8}, Wals_code = {sue}, Year = {1974} } @book{Wilson-1980, Address = {Ukarumpa, Papua New Guinea}, Author = {Wilson, .}, Iso_code = {abt}, Olac_field = {morphology; syntax; semantics; general_linguistics; typology}, Publisher = {Summer Institute of Linguistics}, Series = {Workpapers in Papua New Guinea Languages}, Title = {{A}mbulas {G}rammar}, Volume = {26}, Wals_code = {amb}, Year = {1980} } @book{Wolfart-1973, Address = {Philadelphia}, Author = {Wolfart, }, Iso_code = {crk}, Olac_field = {morphology; phonetics; syntax; semantics; typology; phonology; general_linguistics}, Publisher = {American Philosophical Society}, Series = {Transactions of the American Philosophical Society, New Series}, Title = {{P}lains {C}ree: {A} {G}rammatical {S}tudy}, Volume = {63.5}, Wals_code = {cre}, Year = {1973} } @book{Woollams-1996, Address = {Canberra}, Author = {}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar of {K}aro {B}atak, {S}umatra}, Volume = {130}, Wals_code = {bkr}, Year = {1996} } @misc{Wu-2006, Author = {}, Iso_code = {ami}, Olac_field = {typology; general_linguistics; syntax}, School = {Buffalo State UNiversity of New York}, Title = {{V}erb classification, case marking, and grammatical relations in {A}mis}, Wals_code = {ami}, Year = {2006} } @mastersthesis{Yami-1990, Author = {}, Hhtype = {grammar_sketch}, Inlg = {English [eng]}, Lgcode = {Yami [tao]}, Macro_area = {Papunesia}, Pages = {ii+146}, School = {National Tsing Hua University}, Src = {hh}, Title = {{Y}ami {S}tructure: {A} {D}escriptive {S}tudy of the {Y}ami {L}anguage}, Wals_code = {ymi}, Year = {1990} } @book{Yar-Shater-1969, Address = {The Hague}, Author = {}, Iso_code = {tks}, Olac_field = {syntax; semantics; typology; general_linguistics}, Publisher = {Mouton}, Title = {{A} {G}rammar of {S}outhern {T}ati {D}ialects}, Wals_code = {tts}, Year = {1969} } @book{Yidiny-1977, Address = {Cambridge}, Aiatsis_callnumber = {B D621.81/G2 1 Book Open Stack}, Aiatsis_code = {Y117}, Aiatsis_reference_language = {Yidiny^}, Author = {.}, Besttxt = {ptxt2\australia\dixon_yidin1977v2.txt}, Cfn = {australia\dixon_yidin1977.pdf}, Class_loc = {PL7101.Y53}, Delivered = {australia\dixon_yidin1977.pdf}, Document_type = {B}, Fn = {australia\dixon_yidin1977v3.pdf, australia\dixon_yidin1977.pdf, australia\dixon_yidin1977v2.pdf, australia/dixon_yidin1977.pdf}, Hhtype = {grammar}, ISBN = {9780521214629}, Inlg = {English [eng]}, Iso_code = {yii}, Keywords = {Yidiny word domains}, Languageid = {580}, Lgcode = {Yidiny [yii]}, Lgfamily = {Pama-Nyungan, Yidinic}, Macro_area = {Australia}, Mpi_eva_library_shelf = {PL 7101 .Y53 DIX 1977}, Mpifn = {yidin_dixon1977.pdf}, Olac_field = {semantics}, Ozbib_id = {1526}, Ozbibnote = {[Review Bulletin of the School of Oriental Studies 41, Robins; Bulletin de la Société de Linguistique de Paris 73, Lazard; Lingua 46, Comrie; AUMLA 51, Tryon; Language in Society 8, Haviland; Language 55, Heath]}, Ozbibreftype = {6}, Pages = {xxiii+563}, Publisher = {Cambridge University Press}, Refdb_id = {http://wals.info/refdb/record/205}, Series = {Cambridge Studies in Linguistics}, Src = {autotyp, hh, mpieva, ozbib, phoible}, Subject_headings = {Grammar, Yidiny language, Yidiny language – Grammar, Grammar – Yidiny language – Yidiny language – Grammar}, Title = {{A} {G}rammar of {Y}idiɲ}, Volume = {19}, Wals_code = {yid}, Year = {1977} } @book{Yoshie-1996, Address = {Tokyo}, Author = {}, Iso_code = {mzn}, Olac_field = {general_linguistics; semantics; typology; syntax}, Publisher = {Institute for the Study of Languages and Cultures of Asia and Africa}, Title = {{S}ari {D}ialect}, Wals_code = {mzn}, Year = {1996} } @book{Young-and-Morgan-1980, Address = {Albuquerque}, Author = {. and }, Iso_code = {nav}, Olac_field = {phonetics; typology; general_linguistics; syntax; phonology; semantics; morphology}, Publisher = {University of New Mexico Press}, Title = {{T}he {N}avajo {L}anguage: {A} {G}rammar and {C}olloquial {D}ictionary}, Wals_code = {nav}, Year = {1980} } @incollection{Zaicz-1998, Address = {London}, Author = {}, Booktitle = {{T}he {U}ralic {L}anguages}, Editor = {}, Iso_code = {myv}, Olac_field = {general_linguistics; syntax; semantics; typology; morphology}, Pages = {184-218}, Publisher = {Routledge}, Title = {{M}ordva}, Wals_code = {moe}, Year = {1998} } @incollection{Zee-1999, Address = {Cambridge}, Author = {}, Booktitle = {{H}andbook of the {I}nternational {P}honetic {A}ssociation}, Iso_code = {yue}, Olac_field = {general_linguistics; typology; phonetics; phonology}, Pages = {58-60}, Publisher = {Cambridge University Press}, Title = {{C}hinese ({H}ong {K}ong {C}antonese)}, Wals_code = {cnt}, Year = {1999} } @book{Zeitoun-2007, Address = {Taipei}, Author = {}, Iso_code = {dru}, Olac_field = {morphology; syntax; typology; general_linguistics; semantics}, Publisher = {Academia Sinica}, Series = {Language and Linguistics Monograph}, Title = {{A} grammar of {M}antauran ({R}ukai)}, Volume = {A4-2}, Wals_code = {ruk}, Year = {2007} } @incollection{Zhu-2001, Address = {New York / Dublin}, Author = {}, Booktitle = {{F}acts {A}bout the {W}orld's {L}anguages, {A}n {E}ncyclopedia of the {W}orld's {L}anguages: {P}ast and {P}resent}, Editor = { and }, Iso_code = {cmn}, Olac_field = {typology; morphology; general_linguistics}, Pages = {146-150}, Publisher = {HW Wilson}, Title = {{M}andarin {C}hinese}, Wals_code = {mnd}, Year = {2001} } @book{Zuniga-2000, Address = {München}, Author = {}, Iso_code = {arn}, Olac_field = {syntax; typology; general_linguistics}, Publisher = {Lincom Europa}, Title = {{M}apudungun}, Wals_code = {map}, Year = {2000} } @mastersthesis{bourdeau_ergativity_2015, Author = {}, School = {Radboud University}, Title = {{E}rgativity in {Shawi} ({Chavahuita})}, Wals_code = {chy}, Year = {2015} } @book{de-Vries-1993, Address = {Canberra}, Author = {}, Iso_code = {wng; tyn}, Olac_field = {semantics; morphology; typology; syntax; general_linguistics}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series B}, Title = {{F}orms and {F}unctions in {K}ombai, an {A}wyu {L}anguage of {I}rian {J}aya}, Volume = {108}, Wals_code = {kmb}, Year = {1993} } @book{dryer_world_2013, Address = {Leipzig}, Author = { and }, Publisher = {Max Planck Institute for Evolutionary Anthropology}, Title = {{T}he world atlas of language structures online}, Wals_code = {keu; rcp}, Year = {2013} } @book{lichtenberk_grammar_1983, Address = {Honolulu}, Author = {}, Publisher = {University of Hawaii Press}, Title = {{A} {Grammar} of {Manam}}, Wals_code = {toq}, Year = {1983} } @book{mous_alagwa_2016, Address = {Paris}, Author = {}, Publisher = {Broché}, Title = {{A}lagwa: {A} {South} {Cushitic} language of {Tanzania}: {Grammar}, texts and lexicon}, Wals_code = {agw}, Year = {2016} } @book{pickering_apurina_2009, Address = {Anápolis}, Author = {}, Publisher = {Associação Internacional de Linguística - SIL Brasil}, Title = {{A}purinã {Grammar}}, Wals_code = {apu}, Year = {2009} } @book{van-Driem-1987, Address = {Berlin}, Author = {}, Iso_code = {lif}, Olac_field = {semantics; phonology; syntax; morphology; typology; phonetics; general_linguistics}, Publisher = {Mouton de Gruyter}, Series = {Mouton Grammar Library}, Title = {{A} {G}rammar of {L}imbu}, Volume = {4}, Wals_code = {lim}, Year = {1987} } @misc{van-Engelenhoven-1995, Address = {Ridderkerk}, Author = {, Aone}, Iso_code = {lti}, Olac_field = {phonology; phonetics; semantics; general_linguistics; typology; syntax}, Publisher = {Offsetdrukkerij Ridderprint}, School = {University of Leiden}, Title = {{A} description of the {L}eti {L}anguage (as spoken in {T}utukei)}, Wals_code = {let}, Year = {1995} } @book{van-Klinken-1999, Address = {Canberra}, Author = {, }, Iso_code = {tet}, Olac_field = {semantics; syntax; general_linguistics; typology}, Publisher = {Australian National University}, Series = {Pacific Linguistics, Series C}, Title = {{A} {G}rammar of the {F}ehan {D}ialect of {T}etun, an {A}ustronesian {L}anguage of {W}est {T}imor}, Volume = {155}, Wals_code = {ttn}, Year = {1999} } @misc{van-Staden-2000, Author = {, Miriam}, Iso_code = {tvo}, Olac_field = {semantics; syntax; general_linguistics; typology; morphology}, School = {Leiden University}, Title = {{T}idore: {A} {L}inguistic {D}escription of a {L}anguage of the {N}orth {M}oluccas}, Wals_code = {tid}, Year = {2000} } @book{van-den-Berg-1995, Address = {München}, Author = {, Helma}, Iso_code = {huz}, Olac_field = {general_linguistics; syntax; morphology; phonetics; phonology; typology; semantics}, Publisher = {Lincom Europa}, Series = {Lincom Studies in Caucasian Linguistics}, Title = {{A} {G}rammar of {H}unzib}, Volume = {1}, Wals_code = {hzb}, Year = {1995} } @book{van-der-Tuuk-1971, Address = {The Hague}, Author = {, .}, Iso_code = {bbc}, Note = {Reprint of 1864}, Olac_field = {phonology; typology; syntax; general_linguistics; phonetics; semantics}, Publisher = {}, Title = {{A} {G}rammar of {T}oba {B}atak}, Wals_code = {tob}, Year = {1971} } @misc{van-der-Voort-2000, Author = {, Hein}, Olac_field = {typology; semantics; syntax; general_linguistics}, School = {University of Leiden}, Title = {{A} {G}rammar of {K}waza}, Wals_code = {kwz}, Year = {2000} } @article{Ayers062019TheNeedFor, author = { and and }, doi = {10.1001/jama.2019.4432}, journal = {{JAMA}}, month = {jun}, number = {22}, pages = {2163}, publisher = {American Medical Association ({AMA})}, title = {The Need for Federal Regulation of Marijuana Marketing}, url = {https://doi.org/10.1001%2Fjama.2019.4432}, volume = {321}, year = {2019} } hertrste/mythos10-100 % Beamer theme by <> % December 2010 version 0.1 % % based on the LaTeX-Beamer package : % Copyright 2003 by <> % % This program can be redistributed and/or modified under the terms % of the GNU Public License, version 2. % \mode % \definecolor{beamer@btublack}{cmyk}{0.1,0,0,0.8} \definecolor{beamer@btured}{cmyk}{0,1,0.9,0} \definecolor{beamer@fak1Color}{cmyk}{0.10,1.00,0,0} % magenta \definecolor{beamer@fak2Color}{cmyk}{1,0.70,0,0} % dark blue \definecolor{beamer@fak3Color}{cmyk}{1,0,0.10,0} % light blue \definecolor{beamer@fak4Color}{cmyk}{0.45,0,1,0} % light green % \definecolor{beamer@highlightColor}{named}{black} % \setbeamercolor{structure}{fg=beamer@highlightColor} % \setbeamercolor{title}{parent=normal} % \setbeamercolor{section in toc}{parent=structure} \setbeamercolor{section in toc shaded}{parent=normal} % \setbeamercolor{section number projected}{parent=item projected} % \setbeamercolor{subsection in toc}{parent=section in toc} \setbeamercolor{subsection in toc shaded}{parent=normal} % \setbeamercolor{footline}{fg=beamer@highlightColor} % \setbeamercolor{item}{parent=normal} \setbeamercolor{alerted text}{fg=beamer@fak1Color} \setbeamercolor{example text}{fg=beamer@fak4Color} % \setbeamercolor{mini frame}{parent=section in toc} %\setbeamercolor{mini frame shaded}{parent=beamer@fak4Color} \setbeamercolor{mini frame shaded}{use=normal text,fg=normal text.fg!30!bg} % \mode TrickyTroll/ML-intro1-10 %% Generated by Sphinx. \def\sphinxdocclass{report} \documentclass[letterpaper,10pt,french]{sphinxmanual} \ifdefined\pdfpxdimen \let\sphinxpxdimen\pdfpxdimen\else\newdimen\sphinxpxdimen \fi \sphinxpxdimen=.75bp\relax \PassOptionsToPackage{warn}{textcomp} \usepackage[utf8]{inputenc} \ifdefined\DeclareUnicodeCharacter % support both utf8 and utf8x syntaxes \ifdefined\DeclareUnicodeCharacterAsOptional \def\sphinxDUC#1{\DeclareUnicodeCharacter{"#1}} \else \let\sphinxDUC\DeclareUnicodeCharacter \fi \sphinxDUC{00A0}{\nobreakspace} \sphinxDUC{2500}{\sphinxunichar{2500}} \sphinxDUC{2502}{\sphinxunichar{2502}} \sphinxDUC{2514}{\sphinxunichar{2514}} \sphinxDUC{251C}{\sphinxunichar{251C}} \sphinxDUC{2572}{\textbackslash} \fi \usepackage{cmap} \usepackage[T1]{fontenc} \usepackage{amsmath,amssymb,amstext} \usepackage{babel} \usepackage{times} \expandafter\ifx\csname T@LGR\endcsname\relax \else % LGR was declared as font encoding \substitutefont{LGR}{\rmdefault}{cmr} \substitutefont{LGR}{\sfdefault}{cmss} \substitutefont{LGR}{\ttdefault}{cmtt} \fi \expandafter\ifx\csname T@X2\endcsname\relax \expandafter\ifx\csname T@T2A\endcsname\relax \else % T2A was declared as font encoding \substitutefont{T2A}{\rmdefault}{cmr} \substitutefont{T2A}{\sfdefault}{cmss} \substitutefont{T2A}{\ttdefault}{cmtt} \fi \else % X2 was declared as font encoding \substitutefont{X2}{\rmdefault}{cmr} \substitutefont{X2}{\sfdefault}{cmss} \substitutefont{X2}{\ttdefault}{cmtt} \fi \usepackage[Sonny]{fncychap} \ChNameVar{\Large\normalfont\sffamily} \ChTitleVar{\Large\normalfont\sffamily} \usepackage[,numfigreset=1,mathnumfig]{sphinx} \fvset{fontsize=\small} \usepackage{geometry} % Include hyperref last. \usepackage{hyperref} % Fix anchor placement for figures with captions. \usepackage{hypcap}% it must be loaded after hyperref. % Set up styles of URL: it should be placed after hyperref. \urlstyle{same} \usepackage{sphinxmessages} \title{Introduction à la reconnaissance optique de caractère} \date{nov. 27, 2020} \release{} \author{, , , } \newcommand{\sphinxlogo}{\vbox{}} \renewcommand{\releasename}{} \makeindex \begin{document} \ifdefined\shorthandoff \ifnum\catcode`\=\string=\active\shorthandoff{=}\fi \ifnum\catcode`\"=\active\shorthandoff{"}\fi \fi \pagestyle{empty} \sphinxmaketitle \pagestyle{plain} \sphinxtableofcontents \pagestyle{normal} \phantomsection\label{\detokenize{intro::doc}} La documentation sur l’utilisation de ce site n’a pas encore été faite. :) \chapter{Rapport Final} \label{\detokenize{rapport_final:rapport-final}}\label{\detokenize{rapport_final::doc}} Cette section contient notre rapport final. \section{Introduction} \label{\detokenize{intro_finale:introduction}}\label{\detokenize{intro_finale::doc}} \subsection{Mise en contexte} \label{\detokenize{intro_finale:mise-en-contexte}} L’intelligence artificielle est au coeur de l’actualité depuis près d’une décennie. Elle est déjà entrain de changer le monde , et ce, dans plusieurs secteurs incluant la finance, la sécurité, la santé, la justice criminelle, les moyens de transport, la publicité, et plusieurs autres. Que ça soit des décisions sur l’investissement d’un portefeuille d’un individu ou de la détection de fraude en identifiant des anormalités, l’intelligence artificielle est de plus en plus présente dans le secteur de la finance. \sphinxcite{zbib:nytimes} Du côté de la sécurité, un excellent exemple serait \sphinxhref{https://en.wikipedia.org/wiki/Project\_Maven}{Project Maven} un projet d” intelligence artificielle du \sphinxhref{https://en.wikipedia.org/wiki/The\_Pentagon}{Pentagon} des États\sphinxhyphen{}Unis qui est capable de passer à travers plusieurs informations, vidéos et photos pour détecter des dangers potentiels. L’intelligence artificielle est très importante dans la santé avec des compagnies comme \sphinxhref{https://www.merantix.com/}{Merantix}, une compagnie allemande qui a permis de détecter des ganglions lymphatiques ainsi que des problèmes liés à ceux\sphinxhyphen{}ci tels que des lésions ou des cancers. L’étude de séquence d’ADN par l’intelligence artificielle permet de détecter des maladies génétiques et des cancers. Un des domaines le plus importants en ce moment serait, les moyens de transport avec plus de \$80 milliards investis dans des véhicules de conduite autonome entre 2014 et 2017. L’intelligence artificielle dans ce domaine aurait pour but de diminuer grandement l’erreur humaine dans les transports et réduire à presque zéro les accidents si la majorité des autos était intelligente. De plus, cela réduirait aussi grandement le trafic grâce à la communication entre les automobiles intelligentes. La compagnie \sphinxhref{https://www.tesla.com/}{Tesla} en est déjà très avancée pour ce qui est de leur auto intelligente. \sphinxcite{zbib:gouvqc} Comme on peut le voir, cette technologie a permis de multiples avancées dans des domaines où il se fait extrêmement difficile de modéliser la problématique selon une fonction mathématique particulière. L’analyse de langage en est un bon exemple. Le travail ne peut être modélisé par une seule fonction mathématique puisque les conditions souvent changeantes nécessiteraient une multitude de fonctions différentes pour chaque environnement qui n’est pas réaliste. La solution est plutôt « d’entraîner » un ordinateur à comprendre le monde qui l’entoure. Pour continuer avec l’exemple de l’analyse du langage, une solution serait de fournir à l’ordinateur une immense quantité d’exemples et de solutions afin qu’il développe la capacité de prédire la solution à de nouveaux exemples. \sphinxhref{https://github.com/openai/gpt-3}{GPT\sphinxhyphen{}3}, un nouveau modèle d’intelligence artificielle produit par \sphinxhref{https://openai.com}{OpenAI}, a permis à des développeurs de créer un programme lui\sphinxhyphen{}même capable de programmer à partir de demandes spécifiques faites par un utilisateur. \subsection{Le début de la découverte des inconvénients} \label{\detokenize{intro_finale:le-debut-de-la-decouverte-des-inconvenients}} Malgré les avancées incroyables que l’intelligence artificielle a déjà permis et continuera de permettre dans le futur, elle n’est pas sans ses inconvénients. \subsubsection{Le biais} \label{\detokenize{intro_finale:le-biais}} Au courant des dernières années, les systèmes intelligents sont de plus en plus reconnus coupables de discrimination envers certains groupes d’individus. Une étude réalisée par le \sphinxhref{https://www.nist.gov/}{NIST} à étudié le taux d’erreur de différents programmes de reconnaissance faciale en fonction des différences de sexe et d’ethnicité des individus sur les photos analysées. L’étude présente des taux d’erreur jusqu’à cent fois plus élevés pour des personnes d’origine asiatique ou africaine lorsque comparé à des personnes d’origine européenne \sphinxcite{zbib:nistbias}. Le taux d’erreur est aussi plus élevé chez les femmes que chez les hommes, et ce, peut importe l’origine. Un autre résultat important de cette étude est que le taux d’erreur associé à la reconnaissance de personnes asiatiques n’est pas présent dans des programmes réalisés dans des pays d’Asie. Cette observation permet de déduire l’un des plus grands problèmes liés à l’intelligence artificielle: le biais. Contrairement à une fonction mathématique qui transforme un chiffre de manière définie, les procédés menant à la reconnaissance faciale sont beaucoup plus flous et souvent très mal compris. Plusieurs considèrent les programmes entraînés comme des « boîtes noires ». Il est difficile de prédire ce qui sortira de la boîte lorsque l’on y insère quelque chose, et il est encore plus difficile de comprendre pourquoi le programme prend certaines décisions plus que d’autres. Cette imprévisibilité inquiète plusieurs. Elle rend la tâche de corriger le biais assez ardue. Elle fait aussi en sorte qu’il est difficile de prédire le comportement du programme dans des cas extrêmes sans avoir à lui faire passer des tests dans ces conditions. Le biais est donc un phénomène difficile à corriger, ce qui entraîne des questionnements en rapport aux bienfaits de l’utilisation de l’intelligence artificielle. Certaines régions du monde commencent à bannir l’utilisation de la reconnaissance faciale par les forces de l’ordre. C’est le cas de la ville de Portland, en Oregon \sphinxcite{zbib:cnnportland}. La ville a décidé de bannir l’utilisation de la technologie suite à des craintes en liées à son manque de précision, surtout lorsqu’utilisée sur des individus appartenant à une minorité visible. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{black_box}.png} \caption{L’analogie de la boîte noire.}\label{\detokenize{intro_finale:boite-noire}}\end{figure} \subsubsection{Une deuxième révolution industrielle} \label{\detokenize{intro_finale:une-deuxieme-revolution-industrielle}} Une autre inquiétude liée à l’intelligence artificielle est l’importante quantité d’emplois qui risque de disparaître puisqu’ils seront maintenant occupés par des ordinateurs. Ces inquiétudes sont justifiées. Plusieurs articles, dont \sphinxhref{https://www.cnbc.com/2019/01/14/the-oracle-of-ai-these-kinds-of-jobs-will-not-be-replaced-by-robots-.html}{celui\sphinxhyphen{}ci} publié par CNN ainsi que \sphinxhref{https://medium.com/@ChanPriya/15-jobs-that-will-never-be-replaced-by-ai-512bfbbed0d6}{cette publication} sur Medium tentent de rassurer la population en mentionnant des emplois qui ne pourraient apparemment jamais être remplacés par des ordinateurs. Ils mentionnent entre autres les emplois créatifs, accompagnés des emplois nécessitant beaucoup d’interactions humaines. Pourtant, le domaine de l’IA avance chaque année, et il existe maintenant une panoplie de programmes capable de \sphinxhref{https://openai.com/blog/musenet/}{composer de la musique}, \sphinxhref{https://www.nvidia.com/en-us/research/ai-playground/}{maîtriser les arts visuels} ainsi qu”\sphinxhref{https://www.youtube.com/watch?v=D5VN56jQMWM}{entretenir des conversations au téléphone}. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{duplex}.jpeg} \caption{Le PDG de Google présentant une démonstration de Google Duplex.}\label{\detokenize{intro_finale:duplex-presentation}}\end{figure} Il est dangereux d’extrapoler le progrès qui a été fait au courant des dernières années sur les décennies à venir. Certaines lois limitant le développement de l’IA, ou des limitations physiques au présent rythme d’augmentation de la puissance de calcul des ordinateurs pourraient survenir grandement ralentir le développement de la technologie. Si nous tentons tout de même de le faire, les inquiétudes vécues par plusieurs semblent raisonnables. \subsection{Comprendre la technologie pour démystifier les inquiétudes} \label{\detokenize{intro_finale:comprendre-la-technologie-pour-demystifier-les-inquietudes}} Bien que les précédentes inquiétudes face à l’intelligence artificielle soient totalement justifiées, elles ne sont pas sans solution. Si son développement est fait de manière éthique et s’il est bien encadré, nous pourrions en retirer plus d’avantages que d’inconvénient. Pour bien comprendre les inquiétudes, il faut d’abord comprendre les enjeux. C’est pourquoi nous tenterons de répondre à la question suivante. \sphinxstyleemphasis{Quel est le fonctionnent de l’intelligence artificielle et comment devrait\sphinxhyphen{}elle être utilisée afin de bénéficier l’être humain?} \section{Les librairies nécessaires} \label{\detokenize{explications_librairies:les-librairies-necessaires}}\label{\detokenize{explications_librairies::doc}} Afin de réaliser ce projet dans des temps raisonnables, nous utilisons des outils et des données réalisés par des organisations réputées comme Google, \sphinxhref{https://numpy.org/}{Numpy} et la \sphinxhref{https://matplotlib.org/}{Matplotlib} development team. \subsection{Tensorflow} \label{\detokenize{explications_librairies:tensorflow}} \DUrole{xref,myst}{Tensorflow} est une plateforme nous permettant d’accélérer le développement de notre application d’apprentissage machine. La librairie procure un \sphinxhref{https://en.wikipedia.org/wiki/API}{API} en Python donnant accès à de multiples fonctions utilisées pour obtenir des données et les utiliser pour entraîner notre programme. \subsubsection{Historique} \label{\detokenize{explications_librairies:historique}} Développée par Google et rendue publique en 2015 \sphinxcite{zbib:wikitf}, la librairie a depuis permis aux masses de développer toutes sortes d’applications bénéficiant de l’intelligence artificielle. TensorFlow est une version polie du système DistBelief. DistBelief est un produit du projet The Google Brain. Après avoir été utilisé pendant quelques années pour des produits Google ainsi que pour de la recherche, DistBelief est amélioré et rendu publique sous le nom TensorFlow \sphinxcite{zbib:tfpaper}. Aujourd’hui, la \sphinxhref{https://github.com/tensorflow/tensorflow}{page Github} de TensorFlow mentionne plus de 100 000 utilisateurs et 2 780 contributeurs. La librairie est utilisée par de multiples entreprises dont Google, Coca\sphinxhyphen{}Cola, airbnb, Twitter et Intel \sphinxcite{zbib:tfmain}. \subsubsection{Notre utilisation} \label{\detokenize{explications_librairies:notre-utilisation}} Nous utilisons TensorFlow afin de faciliter l’accès à nos données et afin de créer notre modèle. \paragraph{Accès aux données} \label{\detokenize{explications_librairies:acces-aux-donnees}} Pour entrainer notre modèle, nous utilisons la base de données \sphinxhref{http://yann.lecun.com/exdb/mnist/}{MNIST}. Bien qu’elle soit très complète, le format de cette base de donnée est assez complexe ({\hyperref[\detokenize{preprocessing::doc}]{\sphinxcrossref{\DUrole{doc,std,std-doc}{Voir la section sur le \sphinxstyleemphasis{preprocessing})}}}}. Heureusement, la libraire TensorFlow procure un \sphinxhref{https://www.tensorflow.org/api\_docs/python/tf/keras/datasets/mnist/load\_data}{interface simple} avec le langage de programmation que nous utilisons. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{p}{(}\PYG{p}{(}\PYG{n}{train\PYGZus{}data}\PYG{p}{,} \PYG{n}{train\PYGZus{}labels}\PYG{p}{)}\PYG{p}{,} \PYG{p}{(}\PYG{n}{test\PYGZus{}data}\PYG{p}{,} \PYG{n}{test\PYGZus{}labels}\PYG{p}{)}\PYG{p}{)} \PYG{o}{=} \PYG{n}{mnist}\PYG{o}{.}\PYG{n}{load\PYGZus{}data}\PYG{p}{(}\PYG{p}{)} \PYG{n}{num\PYGZus{}data} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{vstack}\PYG{p}{(}\PYG{p}{[}\PYG{n}{train\PYGZus{}data}\PYG{p}{,} \PYG{n}{test\PYGZus{}data}\PYG{p}{]}\PYG{p}{)} \PYG{n}{num\PYGZus{}labels} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{hstack}\PYG{p}{(}\PYG{p}{[}\PYG{n}{train\PYGZus{}labels}\PYG{p}{,} \PYG{n}{test\PYGZus{}labels}\PYG{p}{]}\PYG{p}{)} \end{sphinxVerbatim} La classe \sphinxcode{\sphinxupquote{mnist}} fournie par la librairie \sphinxcode{\sphinxupquote{Tensorflow}}nous permet de charger en mémoire toutes les données nécessaires en seulement trois lignes. \paragraph{Entraînement du modèle} \label{\detokenize{explications_librairies:entrainement-du-modele}} Le domaine de l’IA s’avère assez complexe. Programmer et entraîner un modèle nécessite des connaissances en mathématiques avancées \sphinxcite{zbib:goodfellow-et-al-2016}. \sphinxcode{\sphinxupquote{Tensorflow}}vise à accélérer le développement de l’intelligence artificielle ainsi que de rendre ce développement accessible aux masses. Afin de simplifier la réalisation de notre programme et afin de le rendre plus efficace, nous comptons donc utiliser les méthodes d’entraînement fournies par la librairie. Pour aider le lecteur à comprendre le fonctionnement du programme (et donc pour éviter le phénomène de la {[}boîte noire{]}(lien vers la section de la boîte noire), chaque fonction utilisée sera décortiquée et expliquée en détails. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{model} \PYG{o}{=} \PYG{n}{keras}\PYG{o}{.}\PYG{n}{Sequential}\PYG{p}{(}\PYG{p}{[} \PYG{n}{keras}\PYG{o}{.}\PYG{n}{layers}\PYG{o}{.}\PYG{n}{Flatten}\PYG{p}{(}\PYG{n}{input\PYGZus{}shape}\PYG{o}{=}\PYG{p}{(}\PYG{l+m+mi}{28}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{)}\PYG{p}{)}\PYG{p}{,} \PYG{n}{keras}\PYG{o}{.}\PYG{n}{layers}\PYG{o}{.}\PYG{n}{Dense}\PYG{p}{(}\PYG{l+m+mi}{128}\PYG{p}{,} \PYG{n}{activation}\PYG{o}{=}\PYG{l+s+s1}{\PYGZsq{}}\PYG{l+s+s1}{relu}\PYG{l+s+s1}{\PYGZsq{}}\PYG{p}{)}\PYG{p}{,} \PYG{n}{keras}\PYG{o}{.}\PYG{n}{layers}\PYG{o}{.}\PYG{n}{Dense}\PYG{p}{(}\PYG{l+m+mi}{10}\PYG{p}{)} \PYG{p}{]}\PYG{p}{)} \PYG{n}{model}\PYG{o}{.}\PYG{n}{compile}\PYG{p}{(}\PYG{n}{optimizer}\PYG{o}{=}\PYG{l+s+s1}{\PYGZsq{}}\PYG{l+s+s1}{adam}\PYG{l+s+s1}{\PYGZsq{}}\PYG{p}{,} \PYG{n}{loss}\PYG{o}{=}\PYG{n}{tf}\PYG{o}{.}\PYG{n}{keras}\PYG{o}{.}\PYG{n}{losses}\PYG{o}{.}\PYG{n}{SparseCategoricalCrossentropy}\PYG{p}{(}\PYG{n}{from\PYGZus{}logits}\PYG{o}{=}\PYG{k+kc}{True}\PYG{p}{)}\PYG{p}{,} \PYG{n}{metrics}\PYG{o}{=}\PYG{p}{[}\PYG{l+s+s1}{\PYGZsq{}}\PYG{l+s+s1}{accuracy}\PYG{l+s+s1}{\PYGZsq{}}\PYG{p}{]}\PYG{p}{)} \PYG{n}{model}\PYG{o}{.}\PYG{n}{fit}\PYG{p}{(}\PYG{n}{train\PYGZus{}images}\PYG{p}{,} \PYG{n}{train\PYGZus{}labels}\PYG{p}{,} \PYG{n}{epochs}\PYG{o}{=}\PYG{l+m+mi}{10}\PYG{p}{)} \end{sphinxVerbatim} Seulement quelques lignes sont nécessaires à l’entraînement de notre modèle. \subsection{Numpy} \label{\detokenize{explications_librairies:numpy}} \sphinxhref{https://numpy.org}{NumPy} est une librairie facilitant le calcul avec le langage de programmation que nous utilisons pour créer notre modèle. \subsubsection{Historique} \label{\detokenize{explications_librairies:id5}} La libraire à été créée en 2005 et était à l’époque basée sur les librairies \sphinxhref{https://docs.python.org/3/library/numeric.html}{numériques et les modules mathématiques} de Python \sphinxcite{zbib:numpy}. Numpy vise à rendre la réalisation de grands calculs numériques plus simples et optimisée. Plusieurs de ses capacités ont des fonctions homologue dans les logiciels \sphinxhref{https://www.mathworks.com}{Matlab} et \sphinxhref{https://maplesoft.com}{Maple}. \subsubsection{Notre utilisation} \label{\detokenize{explications_librairies:id7}} Comme sera possible de l’observer tout au long de ce rapport, les mathématiques constituent le pilier principal sur lequel repose le domaine de l’intelligence artificielle. De plus, la plupart des composantes des réseaux neuronaux peuvent être représentés comme des matrices. Numpy permet d’utiliser certaines propriétés des matrices afin de paralléliser les calculs menant à l’entraînement du modèle. De plus, comme nous le verrons dans la section suivante, Numpy peut être utilisé afin de représenter des images comme des matrices, et donc faciliter les opérations sur chacun des pixels. \subsection{Matplolib} \label{\detokenize{explications_librairies:matplolib}} \sphinxhref{https://matplotlib.org}{Matplotlib} est une librairie de visualisation implémentée dans le langage de programmation Python. \subsubsection{Historique} \label{\detokenize{explications_librairies:id8}} La libraire est un projet à code source ouvert financé par \sphinxhref{https://numfocus.org}{Numfocus}. Déployé depuis plus de 17 ans cite\sphinxcode{\sphinxupquote{wikimatplotlib}}, la librairie permet, tout comme \sphinxcode{\sphinxupquote{Numpy}}, au langage de programmation Python de se rapprocher de l’application Matlab, tout en restant libre de droit. \subsubsection{Notre utilisation} \label{\detokenize{explications_librairies:id9}} La librairie fourni un \sphinxstyleemphasis{API} simple à utiliser permettant de réaliser des graphiques directement dans nos notebooks. Nous utilisons aussi la fonctionnalité permettant de représenter des images comme suit: \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{plt}\PYG{o}{.}\PYG{n}{imshow}\PYG{p}{(}\PYG{n}{train\PYGZus{}data}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{)} \end{sphinxVerbatim} Ce qui permet d’obtenir la figure suivante: \sphinxincludegraphics{{plot-demo}.png} \sphinxcode{\sphinxupquote{train\_data{[}0{]}}} est un \sphinxcode{\sphinxupquote{array}} de pixels, représentés par leur \sphinxcode{\sphinxupquote{grayscale value}}. Nous discuterons plus de ces termes dans la section suivante. \section{Traitement antérieur à l’entrainement} \label{\detokenize{preprocessing:traitement-anterieur-a-lentrainement}}\label{\detokenize{preprocessing::doc}} Dans cette section, nous discuterons du traitement nécessaire afin d’utiliser des images pour entrainer un réseau neuronal. Nous discuterons aussi de l’importance de ce traitement, ainsi que de la raison pour laquelle il doit aussi être réalisé sur les images que nous voudrons par la suite reconnaître. \subsection{Pourquoi faire du \sphinxstyleemphasis{preprocessing}?} \label{\detokenize{preprocessing:pourquoi-faire-du-preprocessing}} Comme nous en avons discuté dans la section précédente, notre programme utilise des méthodes fournies par la librairie \sphinxcode{\sphinxupquote{Tensorflow}} afin de charger les données dans le bon format. Malheureusement, dans une majorité des cas, les données ne vous seront pas fournies sur un plateau d’argent. Les programmes d’apprentissage machine visent à faire du calcul statistique sur le jeu de données fourni \begin{sphinxadmonition}{note}{Note:} Dans le domaine de l’IA, un jeu de données représente l’ensemble des données traitées ainsi que leur étiquettes. \end{sphinxadmonition} \subsection{Mise en bouche sur l’apprentissage machine} \label{\detokenize{preprocessing:mise-en-bouche-sur-lapprentissage-machine}} L’entrainement d’un modèle se rapproche beaucoup des mathématiques, plus précisément de la statistique, comme en témoigne le \sphinxstyleemphasis{Deep Learning Book} \sphinxcite{zbib:goodfellow-et-al-2016}. L’apprentissage machine vise à ingérer des quantités massives de données provenant de sources différentes. Par la suite, à l’aide de calculs statistiques, le programme tente de faire une certaine classification du jeu données. Selon le besoin, le programme pourrait alors poser une étiquette sur des données non étiquetées similaires à celles retrouvées dans le jeu de donnée \DUrole{bibtex}{{[}mitclassification{]}}. Le modèle pourrait aussi être entrainer afin de reconnaître des anomalies, grouper des informations similaires par classes et bien d’autres \DUrole{bibtex}{{[}wikisupervised{]}}. Toutes ces informations seront discutées plus en détails au courant de la prochaine section. Ce qu’il est important de retenir, c’est qu’entrainer un modèle nécessite \sphinxstyleemphasis{\sphinxstylestrong{beaucoup}} de données. \begin{sphinxadmonition}{note}{Sur la signification de «beaucoup de données»} Il y a 60 000 exemples dans nos données d’entraînement. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{In} \PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n}{train\PYGZus{}data}\PYG{o}{.}\PYG{n}{shape} \PYG{n}{Out}\PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{p}{(}\PYG{l+m+mi}{60000}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{)} \end{sphinxVerbatim} \end{sphinxadmonition} \subsection{Traiter beaucoup de données} \label{\detokenize{preprocessing:traiter-beaucoup-de-donnees}} Pour reprendre l’exemple précédent, la \sphinxcode{\sphinxupquote{shape}} de l’objet \sphinxcode{\sphinxupquote{train\_data}}est une liste de 60 000 images représentées par des matrices carrées de dimension 28. Dans notre cas, si le programme passait tous les pixels un à un, il faudrait qu’il réalise séquentiellement \(28 \times 28 \times 60 000 = 47 040 000\) opérations. Ce serait par exemple le cas dans une \sphinxcode{\sphinxupquote{for loop}}. Bien que les ordinateurs modernes sont particulièrement rapides% \begin{footnote}[16]\sphinxAtStartFootnote Un ordinateur moderne possédant un processeur de 2GHz peut réaliser 2 000 000 000 opérations par seconde sur chacun de ses coeurs. % \end{footnote}, les modèles récents sont eux aussi entrainés avec des jeux de données de plus en plus massifs. Celui utilisé pour le service \sphinxhref{https://translate.google.ca}{Google Translate}, par exemple, compte des milliards d’exemples \DUrole{bibtex}{{[}googledatasize{]}}. Heureusement, il existe des méthodes permettant de paralléliser% \begin{footnote}[17]\sphinxAtStartFootnote Exécuter plusieurs opérations parallèlement plutôt que séquentiellement. % \end{footnote} les opérations statistiques réalisées sur notre modèle. \subsubsection{Le parallélisme} \label{\detokenize{preprocessing:le-parallelisme}} Pour réduire le coût monétaire et temporel de l’entrainement d’un modèle, la tâche peut être séparée sur plusieurs des coeurs% \begin{footnote}[18]\sphinxAtStartFootnote Un «coeur» est une unité du processeur pouvant faire du calcul indépendamment des autres coeurs. Un processeur d’ordinateur portable moderne possède de deux à quatre coeurs. Une carte graphique moderne en possède quelques milliers, quoique moins performants que ceux du processeur. % \end{footnote} de la machine. Pour se faire, nous profiterons des propriétés des matrices. \paragraph{Les propriétés des matrices} \label{\detokenize{preprocessing:les-proprietes-des-matrices}} Afin de trouver comment il serait possible de réaliser nos calculs en parallèle, analysons les propriétés des matrices. \subparagraph{La multiplication} \label{\detokenize{preprocessing:la-multiplication}} Assumons les matrices de dimensions compatibles% \begin{footnote}[19]\sphinxAtStartFootnote Pour réaliser une multiplication entre deux matrices, il faut que le nombre de colonnes de la première matrice soit égal au nombre de rangées de la deuxième. % \end{footnote} \(A\) et \(B\): \([A \times B]_{i,j} = \displaystyle\sum_{k=1}A_{i,k}B_{k,j}\) Assumons aussi que la matrice \(C\) est produite par l’opération \(A \times B\) et que les matrices \(A\) et \(B\) sont carrées% \begin{footnote}[20]\sphinxAtStartFootnote Une matrice carrée est une matrice qui contient autant de rangées que de colonnes. % \end{footnote}. Le calcul de \(C\) pourrait alors être implémenté de la manière suivante. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{c+c1}{\PYGZsh{} Initialisation de la matrice de départ.} \PYG{c+c1}{\PYGZsh{} Nous assumons que les matrices sont 3x3.} \PYG{n}{C} \PYG{o}{=} \PYG{p}{[}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{,} \PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{,} \PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{,}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{]} \PYG{c+c1}{\PYGZsh{} Pour chaque rangée de la matrice A.} \PYG{k}{for} \PYG{n}{i} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{A}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} Pour chaque valeur d\PYGZsq{}une rangée de la matrice B.} \PYG{k}{for} \PYG{n}{j} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{B}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} Pour chaque rangée de la matrice B} \PYG{k}{for} \PYG{n}{k} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{B}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} [AxB]\PYGZus{}\PYGZob{}i,j\PYGZcb{} += A\PYGZus{}\PYGZob{}i,k\PYGZcb{} * B\PYGZus{}\PYGZob{}k, j\PYGZcb{}} \PYG{n}{C}\PYG{p}{[}\PYG{n}{i}\PYG{p}{]}\PYG{p}{[}\PYG{n}{j}\PYG{p}{]} \PYG{o}{+}\PYG{o}{=} \PYG{n}{A}\PYG{p}{[}\PYG{n}{i}\PYG{p}{]}\PYG{p}{[}\PYG{n}{k}\PYG{p}{]} \PYG{o}{*} \PYG{n}{B}\PYG{p}{[}\PYG{n}{k}\PYG{p}{]}\PYG{p}{[}\PYG{n}{j}\PYG{p}{]} \end{sphinxVerbatim} Quoi qu’assez simple à implémenter, cette façon de calculer \(C\) est particulièrement inefficace. Alors que les matrices A et B augmentent en taille, le nombre d’opérations requises augmente…\sphinxstylestrong{au cube!} Si \(A\) passe d’une matrice \(2X2\) à une matrice \(3X3\), chaque \sphinxcode{\sphinxupquote{for loop}} doit être réalisée \(n\)% \begin{footnote}[21]\sphinxAtStartFootnote \(n\) représente le nombre de rangées et de colonnes d’une matrice carrée. % \end{footnote} fois de plus. Comme le programme contient 3 for loops imbriquées, si la première doit être faite \(n\) fois de plus, alors c’est de même pour la deuxième, puis la troisième. Le calcul est alors \(n \times n \times n = n^3\) fois plus complexe à réaliser \DUrole{bibtex}{{[}wikimatrixmulti{]}}. Heureusement, ce problème n’est pas sans issues. Reprenons l’équation de la multiplication de deux matrices. \([A \times B]_{i,j} = \displaystyle\sum_{k=1}A_{i,k}B_{k,j}\) Dans ce cas, chaque élément de \(C\) est produit par une sommation sur des multiplications d’éléments de \(A\) et \(B\). Il est aussi important de noter qu’aucun calcul pour un élément de \(C\) dépend d’un calcul pour un autre élément de \(C\)% \begin{footnote}[22]\sphinxAtStartFootnote Ils peuvent être calculés dans n’importe quel ordre et le résultat sera toujours le même. Un résultat pourrait aussi % \end{footnote}. Il serait donc possible de calculer plusieurs éléments de \(C\) en même temps! Bien que le calcul en parallèle ne réduit pas l’ordre de complexité, il permet tout de même de diviser le temps requis par le nombre de coeurs utilisés% \begin{footnote}[23]\sphinxAtStartFootnote Plus ou moins. Voir \sphinxhref{https://en.wikipedia.org/wiki/Parallel\_computing\#Amdahl\%27s\_law\_and\_Gustafson\%27s\_law}{1.1 Amdahl’s law and Gustafson’s law} % \end{footnote}. \subparagraph{L’addition} \label{\detokenize{preprocessing:laddition}} L’addition de deux matrices compatibles% \begin{footnote}[24]\sphinxAtStartFootnote Pour que deux matrices puissent être additionnées, elles doivent avoir le même nombre de rangées et de colonnes. % \end{footnote} se définit par l’addition de chacun des éléments homologues des deux matrices. Si nous reprenons les matrices carrées \(A\) et \(B\) utilisées plus haut, la somme de ces deux matrices serait: \([A+B]_{i,j} = A_{i,j} + B_{i,j}\) Supposons que la matrice \(C\) résulte de la somme de \(A\) et \(B\). Il est encore une fois possible d’affirmer que la valeur de \(C_{i,j}\) ne dépend pas de la valeur de \(C_{k,l}\). Il serait possible d’additionner chaque composante des deux matrices dans n’importe quel ordre en obtenant toujours le même résultat. Encore une fois, l’addition de deux matrices peut être parallélisé afin de réduire le temps de calcul% \begin{footnote}[25]\sphinxAtStartFootnote En assumant que le calcul est réalisé sur une machine possédant plusieurs coeurs. % \end{footnote}. \subparagraph{La multiplication par un scalaire} \label{\detokenize{preprocessing:la-multiplication-par-un-scalaire}} Bien que la multiplication par un scalaire s’avère facile à réaliser à la main pour de petites matrices, l’opération doit tout de même être réalisée sur chaque élément de la matrice. \( \lambda \begin{bmatrix} x_{11} & x_{12} & x_{13} & \dots & x_{1n} \\ x_{21} & x_{22} & x_{23} & \dots & x_{2n} \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ x_{d1} & x_{d2} & x_{d3} & \dots & x_{dn} \end{bmatrix} \) Revient à faire le calcul: \(\begin{bmatrix} \lambda x_{11} & \lambda x_{12} & \lambda x_{13} & \dots & \lambda x_{1n} \\ \lambda x_{21} & \lambda x_{22} & \lambda x_{23} & \dots & \lambda x_{2n} \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ \lambda x_{d1} & \lambda x_{d2} & \lambda x_{d3} & \dots & \lambda x_{dn} \end{bmatrix}\) Encore une fois, aucun résultat n’est dépendant d’un autre. Il serait donc possible d’effectuer plusieurs multiplications en même temps, puis grouper les résultats dans une matrice. \subparagraph{La sommation} \label{\detokenize{preprocessing:la-sommation}} Nous avons ici un calcul légèrement différent des autres. Dans le cas de la sommation des éléments d’une matrice de dimension \((1,n)\), le calcul s’avère commutatif en plus d’être associatif% \begin{footnote}[26]\sphinxAtStartFootnote Dans le cas des autres opérations matricielles présentées, elles étaient seulement associatives. Les calculs étaient, comme pour la sommation, indépendants les un des autres. Par contre, pour les autres opérations les résultats devaient être placés de manière ordonnée dans la matrice résultante. % \end{footnote}. La commutativité de l’addition permet à notre programme d’utiliser l’opérateur de réduction. \begin{sphinxadmonition}{note}{L’opérateur de réduction} Un opérateur de réduction permet de réduire les éléments d’un \sphinxhref{https://fr.wikipedia.org/wiki/Tableau\_(structure\_de\_donn\%C3\%A9es)}{tableau} à un seul résultat. \DUrole{bibtex}{{[}wikireducop{]}} \end{sphinxadmonition} En premier lieu, voici comme une addition séquentielle d’un tableau pourrait être réalisé. Assumons un tableau de 8 entiers comme suit: \sphinxcode{\sphinxupquote{tableau = {[}2,9,6,4,1,3,8,8{]}}}. L’addition pourrait alors être réalisée en ajoutant chaque nombre un par un jusqu’à obtenir le total. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{c+c1}{\PYGZsh{} Création du tableau.} \PYG{n}{tableau} \PYG{o}{=} \PYG{p}{[}\PYG{l+m+mi}{2}\PYG{p}{,}\PYG{l+m+mi}{9}\PYG{p}{,}\PYG{l+m+mi}{6}\PYG{p}{,}\PYG{l+m+mi}{4}\PYG{p}{,}\PYG{l+m+mi}{1}\PYG{p}{,}\PYG{l+m+mi}{3}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{]} \PYG{c+c1}{\PYGZsh{} Initiation de la variable `somme`.} \PYG{n}{somme} \PYG{o}{=} \PYG{l+m+mi}{0} \PYG{c+c1}{\PYGZsh{} Pour chaque chiffre dans le tableau.} \PYG{k}{for} \PYG{n}{chiffre} \PYG{o+ow}{in} \PYG{n}{tableau}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} Sommation de l\PYGZsq{}ancienne somme avec le nouveau chiffre.} \PYG{n}{somme} \PYG{o}{+}\PYG{o}{=} \PYG{n}{chiffre} \PYG{c+c1}{\PYGZsh{} Affichage de la somme.} \PYG{n+nb}{print}\PYG{p}{(}\PYG{n}{somme}\PYG{p}{)} \end{sphinxVerbatim} Cet exemple permettrait de réaliser une sommation séquentielle sur tous les chiffres contenus dans le tableau. Ce reviendrait à réaliser le calcul suivant: \((((((((2+9)+6)+4)+1)+3)+8)+8)\) Bien que la moindre performance de cette méthode ne se fait pas ressentir pour des petites sommations, ce programme ne s’adapte pas bien à de grands tableaux% \begin{footnote}[27]\sphinxAtStartFootnote Voir la \{section\} sur les résultats concrets. % \end{footnote}. Ensuite, si l’addition n’était qu’associative, l’opération pourrait tout de même être parallélisée. Les sommes partielles pourraient être calculées indépendamment les unes des autres comme pour les autres opérations matricielles. Le calcul serait similaire à celui ci: \(((2+1)+(9+3))+((6+8)+(4+8))\) Dans cet exemple, les sommes \(2+1\), \(9+3\), \(6+8\) et \(4+8\) sont calculées en même temps. Par contre, lorsque l’une des opérations est complétée avant une autre, il arrive que l’ordinateur ait à attendre \DUrole{bibtex}{{[}stackoverflowcommutativity{]}}. Le coeur ne pourrait alors pas se libérer pour faire d’autres opérations. Par exemple, assumons un ordinateur possédant deux unités de calcul disponibles et une addition non commutative. Si le calcul de \(2+9\) était complété avant celui de \(9+3\), l’ordinateur devrait attendre que les deux calculs soient complétés avant de calculer la somme partielle \((2+9)+(9+3)\). Heureusement, l’addition est associative. L’ordinateur ira donc écrire le résultat de la première opération complétée à la somme partielle, sans se soucier de la complétion de l’autre opération. L’unité de calcul sera alors libérée pour calculer, par exemple, la somme \(6+8\). Finalement , l’opérateur de réduction est beaucoup plus adapté aux échelles de l’intelligence artificielle. Encore une fois, l’ordinateur sépare la sommation en plusieurs petites opérations qui peuvent être exécutés en parallèle. De plus, l’opérateur profite de l’associativité pour optimiser la tâche au maximum. À la fin de la réduction, il ne reste qu’une seule addition. Le calcul mathématique serait par contre identique au précédent. \(((2+1)+(9+3))+((6+8)+(4+8))\) Il est possible de remarquer que l’addition est fait dans un ordre particulier. Cet ordre donne un meilleur modèle d’accès à la mémoire. Python utilise l’opérateur de réduction lors du calcul de sommations. Implémenter ce genre de solution s’avère donc assez simple. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{tableau} \PYG{o}{=} \PYG{p}{[}\PYG{l+m+mi}{2}\PYG{p}{,}\PYG{l+m+mi}{9}\PYG{p}{,}\PYG{l+m+mi}{6}\PYG{p}{,}\PYG{l+m+mi}{4}\PYG{p}{,}\PYG{l+m+mi}{1}\PYG{p}{,}\PYG{l+m+mi}{3}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{]} \PYG{n}{somme} \PYG{o}{=} \PYG{n+nb}{sum}\PYG{p}{(}\PYG{n}{tableau}\PYG{p}{)} \PYG{n+nb}{print}\PYG{p}{(}\PYG{n}{somme}\PYG{p}{)} \end{sphinxVerbatim} \sphinxcode{\sphinxupquote{NumPy}} possède aussi une fonction de sommation optimisée pour les \sphinxcode{\sphinxupquote{numpy arrays}}. Elle peut être implémentée tout aussi simplement. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{k+kn}{import} \PYG{n+nn}{numpy} \PYG{k}{as} \PYG{n+nn}{np} \PYG{n}{tableau} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{array}\PYG{p}{(}\PYG{p}{[}\PYG{l+m+mi}{2}\PYG{p}{,}\PYG{l+m+mi}{9}\PYG{p}{,}\PYG{l+m+mi}{6}\PYG{p}{,}\PYG{l+m+mi}{4}\PYG{p}{,}\PYG{l+m+mi}{1}\PYG{p}{,}\PYG{l+m+mi}{3}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{,}\PYG{l+m+mi}{8}\PYG{p}{]}\PYG{p}{)} \PYG{n}{somme} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{sum}\PYG{p}{(}\PYG{n}{tableau}\PYG{p}{)} \PYG{n+nb}{print}\PYG{p}{(}\PYG{n}{somme}\PYG{p}{)} \end{sphinxVerbatim} \subparagraph{En bref} \label{\detokenize{preprocessing:en-bref}} En bref, une majorité des opérations matricielles peuvent être parallélisées. Les matrices sont donc la représentation de choix pour les jeux de données dans le domaine de l’intelligence artificielle. La transformation du jeu de données en matrices est une partie majeure du \sphinxstyleemphasis{preprocessing}. Elle permet d’accélérer le calcul d’un facteur non\sphinxhyphen{}négligeable. \subsubsection{NumPy} \label{\detokenize{preprocessing:numpy}} C’est pour les opérations parallèles que la librairie \sphinxcode{\sphinxupquote{numpy}}, mentionnée lors de la section précédente, entre en jeu. Les opérations matricielles réalisées à l’aide de méthodes implémentés par \sphinxcode{\sphinxupquote{numpy}}profitent aussi de l’implémentation des \sphinxcode{\sphinxupquote{BLAS}}% \begin{footnote}[28]\sphinxAtStartFootnote \sphinxcode{\sphinxupquote{BLAS}} signifie \sphinxstylestrong{B}asic \sphinxstylestrong{L}inear \sphinxstylestrong{A}lgebra \sphinxstylestrong{S}ubroutines, ou sous\sphinxhyphen{}routines de base d’algèbre linéaire. % \end{footnote}. Les \sphinxcode{\sphinxupquote{BLAS}} permettent de grandement accélérer nos calculs sans même nécessiter de coeurs supplémentaires. Elles exploitent plutôt les différentes architectures de processeur ainsi que leur différents niveau de cache% \begin{footnote}[29]\sphinxAtStartFootnote Petite mémoire rapide allouée au processeur. % \end{footnote}. \paragraph{\sphinxstyleliteralintitle{\sphinxupquote{BLAS}}} \label{\detokenize{preprocessing:blas}} Les sous\sphinxhyphen{}routines d’algèbre linéaire permettent de nettement réduire l’ordre de complexité des opérations d’algèbre linéaire. Elles permettent, par exemple, de décomposer des matrices en blocs afin d’accélérer la multiplication. Ces sous\sphinxhyphen{}programmes sont extrêmement populaires. Ils sont implémentés dans une majorité des programmes de calcul scientifique \DUrole{bibtex}{{[}blaswebsite{]}}. \paragraph{Quelques résultats concrets} \label{\detokenize{preprocessing:quelques-resultats-concrets}} Voici quelques résultats plus concrets permettant d’obtenir une meilleure idée de l’ampleur de l’accélération des calculs. \subparagraph{Multiplication de matrices à l’aide de \sphinxstyleliteralintitle{\sphinxupquote{NumPy}}.} \label{\detokenize{preprocessing:multiplication-de-matrices-a-laide-de-numpy}} Dans ce programme, deux matrices de dimensions \(1000 x 1000\) sont multipliées. La méthode de base avec les itérations imbriquées est comparée avec l’implémentation de l’opération par la librairie \sphinxcode{\sphinxupquote{NumPy}}. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{k+kn}{import} \PYG{n+nn}{time} \PYG{k+kn}{import} \PYG{n+nn}{numpy} \PYG{k}{as} \PYG{n+nn}{np} \PYG{c+c1}{\PYGZsh{} Création de nos matrices} \PYG{n}{A} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{rand}\PYG{p}{(}\PYG{l+m+mi}{1000}\PYG{p}{,}\PYG{l+m+mi}{1000}\PYG{p}{)} \PYG{n}{B} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{rand}\PYG{p}{(}\PYG{l+m+mi}{1000}\PYG{p}{,}\PYG{l+m+mi}{1000}\PYG{p}{)} \PYG{n}{C} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{zeros}\PYG{p}{(}\PYG{p}{(}\PYG{l+m+mi}{1000}\PYG{p}{,}\PYG{l+m+mi}{1000}\PYG{p}{)}\PYG{p}{)} \PYG{c+c1}{\PYGZsh{} Test de la première implémentation} \PYG{n}{start\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{k}{for} \PYG{n}{i} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{A}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{k}{for} \PYG{n}{j} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{B}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{k}{for} \PYG{n}{k} \PYG{o+ow}{in} \PYG{n+nb}{range}\PYG{p}{(}\PYG{n+nb}{len}\PYG{p}{(}\PYG{n}{B}\PYG{p}{)}\PYG{p}{)}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} print(A[i][k])} \PYG{c+c1}{\PYGZsh{} print(B[k][j])} \PYG{c+c1}{\PYGZsh{} print(i,j)} \PYG{n}{C}\PYG{p}{[}\PYG{n}{i}\PYG{p}{]}\PYG{p}{[}\PYG{n}{j}\PYG{p}{]} \PYG{o}{+}\PYG{o}{=} \PYG{n}{A}\PYG{p}{[}\PYG{n}{i}\PYG{p}{]}\PYG{p}{[}\PYG{n}{k}\PYG{p}{]} \PYG{o}{*} \PYG{n}{B}\PYG{p}{[}\PYG{n}{k}\PYG{p}{]}\PYG{p}{[}\PYG{n}{j}\PYG{p}{]} \PYG{n}{end\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{time\PYGZus{}1} \PYG{o}{=} \PYG{n}{end\PYGZus{}time} \PYG{o}{\PYGZhy{}} \PYG{n}{start\PYGZus{}time} \PYG{c+c1}{\PYGZsh{} Test de l\PYGZsq{}implémentation avec Numpy} \PYG{n}{start\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{C} \PYG{o}{=} \PYG{n}{A}\PYG{o}{*}\PYG{n}{B} \PYG{n}{end\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{time\PYGZus{}2} \PYG{o}{=} \PYG{n}{end\PYGZus{}time} \PYG{o}{\PYGZhy{}} \PYG{n}{start\PYGZus{}time} \PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}}\PYG{l+s+s2}{Run time 1 = }\PYG{l+s+si}{\PYGZob{}\PYGZcb{}}\PYG{l+s+s2}{ seconds}\PYG{l+s+s2}{\PYGZdq{}}\PYG{o}{.}\PYG{n}{format}\PYG{p}{(}\PYG{n}{time\PYGZus{}1}\PYG{p}{)}\PYG{p}{)} \PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}}\PYG{l+s+s2}{Run time numpy = }\PYG{l+s+si}{\PYGZob{}\PYGZcb{}}\PYG{l+s+s2}{ seconds}\PYG{l+s+s2}{\PYGZdq{}}\PYG{o}{.}\PYG{n}{format}\PYG{p}{(}\PYG{n}{time\PYGZus{}2}\PYG{p}{)}\PYG{p}{)} \end{sphinxVerbatim} \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{Run} \PYG{n}{time} \PYG{l+m+mi}{1} \PYG{o}{=} \PYG{l+m+mf}{1838.9336512088776} \PYG{n}{seconds} \PYG{n}{Run} \PYG{n}{time} \PYG{n}{numpy} \PYG{o}{=} \PYG{l+m+mf}{0.005700111389160156} \PYG{n}{seconds} \end{sphinxVerbatim} Alors que l’implémentation de base prend plus de 30 minutes à faire le calcul, \sphinxcode{\sphinxupquote{Numpy}} n’a besoin de que moins de 6 millièmes de secondes% \begin{footnote}[30]\sphinxAtStartFootnote Testé sur un processeur \sphinxstyleemphasis{2.9 GHz Dual\sphinxhyphen{}Core Intel Core i5}. % \end{footnote}! \paragraph{Sommations de tableaux à l’aide de \sphinxstyleliteralintitle{\sphinxupquote{NumPy}}} \label{\detokenize{preprocessing:sommations-de-tableaux-a-laide-de-numpy}} Dans cet autre programme démontre quant à lui la différence entre notre implémentation de base de la sommation avec celle de \sphinxcode{\sphinxupquote{NumPy}} sur un \sphinxcode{\sphinxupquote{numpy array}}. La sommation se fait sur 1 milliard d’éléments aléatoires entre \(0\) et \(1\). \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{k+kn}{import} \PYG{n+nn}{time} \PYG{k+kn}{import} \PYG{n+nn}{numpy} \PYG{k}{as} \PYG{n+nn}{np} \PYG{c+c1}{\PYGZsh{} Création du tableau et initiation de la somme.} \PYG{n}{tableau} \PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{random}\PYG{o}{.}\PYG{n}{rand}\PYG{p}{(}\PYG{l+m+mi}{1000000000}\PYG{p}{)} \PYG{n}{somme} \PYG{o}{=} \PYG{l+m+mi}{0} \PYG{c+c1}{\PYGZsh{} Test de la première implémentation.} \PYG{n}{start\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{c+c1}{\PYGZsh{} Pour chaque chiffre dans le tableau.} \PYG{k}{for} \PYG{n}{chiffre} \PYG{o+ow}{in} \PYG{n}{tableau}\PYG{p}{:} \PYG{c+c1}{\PYGZsh{} Sommation de l\PYGZsq{}ancienne somme avec le nouveau chiffre.} \PYG{n}{somme} \PYG{o}{+}\PYG{o}{=} \PYG{n}{chiffre} \PYG{n}{end\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{time\PYGZus{}1} \PYG{o}{=} \PYG{n}{end\PYGZus{}time} \PYG{o}{\PYGZhy{}} \PYG{n}{start\PYGZus{}time} \PYG{c+c1}{\PYGZsh{} Test de l\PYGZsq{}implémentation avec Numpy.} \PYG{c+c1}{\PYGZsh{} Réinitialisation de la somme.} \PYG{n}{somme} \PYG{o}{=} \PYG{l+m+mi}{0} \PYG{n}{start\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{somme} \PYG{o}{+}\PYG{o}{=} \PYG{n}{np}\PYG{o}{.}\PYG{n}{sum}\PYG{p}{(}\PYG{n}{tableau}\PYG{p}{)} \PYG{n}{end\PYGZus{}time} \PYG{o}{=} \PYG{n}{time}\PYG{o}{.}\PYG{n}{time}\PYG{p}{(}\PYG{p}{)} \PYG{n}{time\PYGZus{}2} \PYG{o}{=} \PYG{n}{end\PYGZus{}time} \PYG{o}{\PYGZhy{}} \PYG{n}{start\PYGZus{}time} \PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}}\PYG{l+s+s2}{Run time 1 = }\PYG{l+s+si}{\PYGZob{}\PYGZcb{}}\PYG{l+s+s2}{ seconds}\PYG{l+s+s2}{\PYGZdq{}}\PYG{o}{.}\PYG{n}{format}\PYG{p}{(}\PYG{n}{time\PYGZus{}1}\PYG{p}{)}\PYG{p}{)} \PYG{n+nb}{print}\PYG{p}{(}\PYG{l+s+s2}{\PYGZdq{}}\PYG{l+s+s2}{Run time numpy = }\PYG{l+s+si}{\PYGZob{}\PYGZcb{}}\PYG{l+s+s2}{ seconds}\PYG{l+s+s2}{\PYGZdq{}}\PYG{o}{.}\PYG{n}{format}\PYG{p}{(}\PYG{n}{time\PYGZus{}2}\PYG{p}{)}\PYG{p}{)} \end{sphinxVerbatim} La différence entre les résultats est encore une fois majeure. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{Run} \PYG{n}{time} \PYG{l+m+mi}{1} \PYG{o}{=} \PYG{l+m+mf}{361.87627387046814} \PYG{n}{seconds} \PYG{n}{Run} \PYG{n}{time} \PYG{n}{numpy} \PYG{o}{=} \PYG{l+m+mf}{5.082181215286255} \PYG{n}{seconds} \end{sphinxVerbatim} \subsection{Représenter des images} \label{\detokenize{preprocessing:representer-des-images}} Pour notre programme, l’intrant de notre réseau neuronal sera constitué d’images. Par contre, ces images ne seront pas directement passées au travers de notre programme dans leur format d’origine. Comme discuté dans la section précédente, il serait préférable que les images soient représentées sous forme de matrices. C’est heureusement déjà le cas de nos données d’entrainement. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{In} \PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n+nb}{type}\PYG{p}{(}\PYG{n}{train\PYGZus{}data}\PYG{p}{)} \PYG{n}{Out}\PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n}{numpy}\PYG{o}{.}\PYG{n}{ndarray} \end{sphinxVerbatim} Dans cet exemple, \sphinxcode{\sphinxupquote{train\_data}} est un \sphinxcode{\sphinxupquote{array}} contenant l’ensemble de nos images. Pour obtenir le nombre d’éléments dans cet \sphinxcode{\sphinxupquote{array}}, la méthode \sphinxcode{\sphinxupquote{shape}} peut être utilisée. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{In} \PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n}{train\PYGZus{}data}\PYG{o}{.}\PYG{n}{shape} \PYG{n}{Out}\PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{p}{(}\PYG{l+m+mi}{60000}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{,} \PYG{l+m+mi}{28}\PYG{p}{)} \end{sphinxVerbatim} La première valeur correspond au nombre d’image dans nos données d’entrainement. Les deux valeurs suivantes sont le nombre de rangées et de colonnes des matrices utilisées pour représenter ces mêmes images. Ce sont donc des matrices carrées de dimensions \(n = 28\). Afin d’analyser précisément l’une de ces images, imprimons l’une des rangées de pixels. \begin{sphinxVerbatim}[commandchars=\\\{\}] \PYG{n}{In} \PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n}{train\PYGZus{}data}\PYG{p}{[}\PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{[}\PYG{l+m+mi}{20}\PYG{p}{]} \PYG{n}{Out}\PYG{p}{[} \PYG{p}{]}\PYG{p}{:} \PYG{n}{array}\PYG{p}{(}\PYG{p}{[} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{24}\PYG{p}{,} \PYG{l+m+mi}{114}\PYG{p}{,} \PYG{l+m+mi}{221}\PYG{p}{,} \PYG{l+m+mi}{253}\PYG{p}{,} \PYG{l+m+mi}{253}\PYG{p}{,} \PYG{l+m+mi}{253}\PYG{p}{,} \PYG{l+m+mi}{253}\PYG{p}{,} \PYG{l+m+mi}{201}\PYG{p}{,} \PYG{l+m+mi}{78}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{,} \PYG{l+m+mi}{0}\PYG{p}{]}\PYG{p}{,} \PYG{n}{dtype}\PYG{o}{=}\PYG{n}{uint8}\PYG{p}{)} \end{sphinxVerbatim} Les pixels sont représentés par des valeurs \sphinxstyleemphasis{grayscale} inversées. Traditionnellement, une valeur \sphinxstyleemphasis{grayscale} est élevée lorsque le pixel est très illuminé. La valeur maximale de \sphinxcode{\sphinxupquote{255}}signifie un blanc, alors que le \sphinxcode{\sphinxupquote{0}}correspond au noir. Dans notre jeu de données, ces deux valeurs sont inversées. Une valeur élevée signifie un pixel plus sombre. Le paramètre \sphinxcode{\sphinxupquote{dtype=uint8}}signifie que nos pixels sont représentés par des entiers de 8 bits. Chaque bit ne pouvant avoir une valeur que de 1 ou 0, le plus grand nombre pouvant être représenté par ce type d’entier est 255. \subsubsection{Explications plus détaillées sur la représentation des images.} \label{\detokenize{preprocessing:explications-plus-detaillees-sur-la-representation-des-images}} Cette section c’est seulement si j’ai le temps. \paragraph{Pourquoi le \sphinxstyleemphasis{grayscale}?} \label{\detokenize{preprocessing:pourquoi-le-grayscale}} \subparagraph{Pourquoi est\sphinxhyphen{}il inversé?} \label{\detokenize{preprocessing:pourquoi-est-il-inverse}} \paragraph{Pourquoi seulement 784 pixels?} \label{\detokenize{preprocessing:pourquoi-seulement-784-pixels}} \bigskip\hrule\bigskip \section{Notions de base} \label{\detokenize{notions_de_base:notions-de-base}}\label{\detokenize{notions_de_base::doc}} \subsection{Introduction au réseau neuronal} \label{\detokenize{notions_de_base:introduction-au-reseau-neuronal}} Un réseau neuronal est une forme d’intelligence artificielle, qui effectue des prédictions basées sur des valeurs qui sont entrées dans le système, afin d’accomplir une certaine tâche. Le réseau est constitué d’un ensemble de neurones interconnectés et distribués en plusieurs couches. Chaque neurone possède des paramètres qui peuvent être ajustés, afin d’obtenir des résultats plus fiables. C’est ce qu’on appele l’entrainement. Le réseau est entrainé à partir d’un jeu de données, qui contient des valeurs associées à une étiquette, qui consiste de la « réponse » attendue. Par exemple, un réseau neuronal ayant comme objectif de prédire l’achalandage dans un parc d’amusement pour une journée donnée pourrait recevoir comme intrant la température, le niveau d’ensoleillement ainsi que le pourcentage de précipitation et d’humidité. Le jeu de données serait alors constituée d’une liste ces quatres valeurs enregistrées à chaque jour des dernières années, avec comme étiquette le nombre de clients cette journée\sphinxhyphen{}là. Les réponses du réseau sont comparées aux étiquettes, et les paramètres des neurones sont individuellement modifiés de manière à se rapprocher de la réponse attendue. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{resneuronalsimp}.png} \caption{Ceci est un exemple simplifié d’un réseau neuronal. Les composantes du schéma seront expliquées en détail dans cette section.}\label{\detokenize{notions_de_base:reseau-neuronal}}\end{figure} \subsection{OCR} \label{\detokenize{notions_de_base:ocr}} Le terme OCR, ou ROC en français, signifie « Reconnaissance optique de caractères ». Cela désigne un processus aucours duquel du texte est extrait d’une image ou d’un document afin d’être transformé en fichier. Pour ce faire, un réseau neuronal reçoit les valeurs des pixels du document de source, \begin{quote} Note : La valeur d’un pixel en « grayscale » ou échelle de gris, est un nombre entier de format 8 bits et peut donc avoir une valeur comprise entre 0 et 255 (2\textasciicircum{}8 \sphinxhyphen{} 1), où 0 est noir et 255 est blanc. Un pixel en couleur est représenté sous la forme d’un vecteur de 3 nombres 8 bits, chaque nombre correspondant à une valeur de rouge, vert et bleu. \sphinxcite{zbib:hipr2} \end{quote} traitées afin de les rendre utilisables par le réseau. Ces données se propagent ensuite vers l’avant dans le réseau, de couche de neurone en couche de neurone, avant d’aboutir à la couche d’extrants, composée de 10 neurones dans le cas de notre programme, qui correspondent aux chiffres de 0 à 9. Un de ces neurones de cette couche finale s’active, donnant ainsi le résultat estimé par le réseau. Ensuite, divers paramètres sont ajustés par un algorithme d’optimisation afin d’augmenter la précision des réponses du réseau. \subsection{Le neurone} \label{\detokenize{notions_de_base:le-neurone}} Le neurone est l’unité de base d’un réseau neuronal. C’est un noeud parmis le réseau par lequel transitent des valeurs, qui sont modifiées au passage par un procédé qui sera expliqué plus en détail prochainement, avant d’être envoyées vers les prochains neurones. Essentiellement, un neurone reçoit une ou des valeurs comme intrant, effectue des opérations sur ces dernières, puis transmet la nouvelle valeur. La structure d’un neurone est relativement simple. Chaque neurone possède un coefficient, ou un \sphinxstylestrong{poids} \(p\) dans le jargon, associé à chaque \sphinxstylestrong{intrant} \(I\) qu’elle reçoit. La première opération que la neurone effectue est la somme des produits des intrants fois leur poids. À celà est ajouté un \sphinxstylestrong{biais} \(b\) propre à chaque neurone. Cette opération peut être représentée par la fonction \(Y = \sum_{i=1}^{n} I_i \times p_i + b\), où n correspond au nombre d’intrants. La dernière opération que les valeurs subissent avant d’être transmises est une fonction d’activation. La fonction d’activation est appliquée à chaque extrant de chaque neurone de la couche. Les fonctions d’activation, analogues à l’activation d’un neurone biologique, permettent généralement d’obtenir un extrant compris entre 0 et 1, ou \sphinxhyphen{}1 et 1. Elles ont plusieurs utilités, notamment pour la modélisation de fonctions non linéeaires, ainsi que pour l’entrainement du réseau, ce qui sera expliqué dans une section ultérieure. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{neurone}.png} \caption{Exemple des opérations effectuées au sein d’un neurone.}\label{\detokenize{notions_de_base:neurone}}\end{figure} La fonction la plus simple est la fonction à échelons. Elle retourne 1 si l’intrant \sphinxstyleemphasis{x} est plus grand qu’une valeur seuil \sphinxstyleemphasis{s}, et 0 s’il ne l’est pas. Cette fonction peut être représentée par l’équation \( E(x)= \begin{cases} 1 & \quad \text{si } x \text{ > s}\\ 0 & \quad \text{si } x \text{ <= s} \end{cases} \) Elle n’est néanmoins pas utilisée, puisqu’elle empêche l’entrainement du réseau. La fonction d’activation doit être dérivable en une autre fonction, et non en une constante, afin que le processus d’ajustement des paramètres puisse avoir lieu. Il est également impossible de représenter des situations non\sphinxhyphen{}linéeaires avec cette fonction, puisque seulement des fonctions linéaires sont présentes dans le réseau. La fonction d’activation la plus utilisée est la fonction Unité Linéaire Rectifiée, ou « ReLU » en anglais (Rectified Linear Unit). Cette fonction peut être représentée par l’équation : \( R(x)= \begin{cases} x & \quad \text{si } x \text{ > 0}\\ 0 & \quad \text{si } x \text{ <= 0} \end{cases} \) ou encore, \( R(x) = max(0, x)\). Cette fonction est peu demandante à calculer pour l’ordinateur, et se fait très rapidement. De plus, malgré son apparence linéaire, elle peut être dérivée, ce qui est nécessaire pour pouvoir entrainer le réseau. C’est pour ces raisons que c’est la fonction d’activation la plus répendue. Elle a toutefois comme désavantage de produire parfois une trop grande quantité de « 0 », ce qui peut entrainer une réaction en chaine, où ces zéros se propagent, empêchant le bon fonctionnement du réseau. Cette situation est appelée la « mort du réseau », où l’extrant de plusieurs neurones devient invariablement 0, ce qui diminue l’efficacité du réseau. Ce phénomène se produit surtout lorsque le réseau se fait entrainer de manière trop rigoureuse, et que le biais de certaines neurones devient une très grande valeur négative, ce qui fait que l’intrant dans la fonction d’activation est toujours en dessous de 0, et l’extrant reste ainsi invariablement 0. Une variation de cette fonction, nommée Leaky ReLU, a été créée afin de tenter de régler ce problème de mort du réseau : \( L(x)= \begin{cases} x & \quad \text{si } x \text{ > 0}\\ 0,01 \times x & \quad \text{si } x \text{ <= 0} \end{cases} \) Ici, les zéros sont remplacés par de très petits nombres négatifs, qui correspondent généralement à x multiplié par le coefficient 0,01. Une autre fonction commune est la sigmoide. Son équation est : \( \phi(x) = \frac{1}{1 + e^{-x}} \) La fonction retourne 0 lorsque x tend vers l’infini négatif, et 1 lorsque x tend vers l’infini positif. Cette fonction a comme avantage de s’approcher rapidement de 0 ou de 1, lorsque l’intrant \sphinxstyleemphasis{x} est plus petit que \sphinxhyphen{}2 ou plus grand que 2, respectivement. Cela permet d’envoyer un signal très fort aux prochains neurones. Cela peut toutefois devenir un désavantage lorsque les intrants sont très grands, puisque l’extrant reste pratiquement le même, ce qui peut nuire à l’entrainement. Cette fonction est également plus lourde pour l’ordinateur, ce qui peut ralentir considérablement le système lorsque ce calcul est effectué des centaines ou des milliers de fois. Une fonction similaire à la sigmoide et la TanH. Son équation est : \( tanh(x) = \frac{2}{1 + e^{-2x}} - 1 \) Elle retourne \sphinxhyphen{}1 lorsque x tend vers l’infini négatif, et 1 lorsque x tend vers l’infini positif. Elle a comme avantage de retourner en moyenne des valeurs proches de 0, ce qui rend la tâche plus facile pour les couches suivantes, puisque les valeurs auront moins tendance à devenir très grandes, ce qui ralentirait les opérations. \subsection{Couches de neurones} \label{\detokenize{notions_de_base:couches-de-neurones}} Comme mentionné précedemment, les neurones sont organisés en couches. Il y a 3 types de couches différentes. La première est la couche des intrants, dans laquelle les données sont rentrées dans le réseau. Dans le cas de notre programme, où les intrants sont des images de format 28x28, la première couche est composée de 784 (\(28\times28 = 784\)) neurones recevant chacun la valeur en échelle de gris d’un pixel de l’image. Plus concrètement, ces images sont des matrices carrées \(M_{28}\), qui se font vectoriser en un vecteur de taille 784. Par la suite, chacune de ces données est transmise à chacun des neurones de la couche cachée, puisque le réseau est densément connecté, et les neurones d’une couche sont connectés à tout ceux des couches adjacentes. Pour la suite de cette explication, le réseau neuronal provenant de la figure affichée plus haut sera utilisé, à des fins de clarté. Donc, les valeurs des trois neurones de la couche d’intrants sont contenus dans la matrice \(I_{1\times3}\). Les poids des neurones de la couche cachée 1 sont contenus dans la matrice \(C_{4\times3}\), où 4 correspond au nombre de neurones dans la couche, et 3 aux poids que possèdent chaque neurones de la couche (un poid par neurone de la couche précédente). Ici, l’opération à faire serait un produit matriciel \(A_{m\times p} \times B_{p\times n} = C_{m\times n}\) , afin de multiplier les intrants par chaque ensemble de poids. Toutefois, les matrices ne sont pas compatibles pour effectuer cette opération, puisque le nombre de colonnes de la première matrice n’est pas égal au nombre de rangées de la seconde. Il faut donc faire la transposée de la matrice \(C_{4\times3}\), qui devient alors \(C_{3\times4}^{t}\). L’opération \(I_{1\times3} \times C_{3\times4}^{t}\), où sont multipliés dans l’ordre, élément par élément, chaque élément d’une ligne de \sphinxstyleemphasis{I} par chaque élément d’une colonne de \sphinxstyleemphasis{C}, puis est effectué la somme de ces produits pour obtenir un nouvel élément de la matrice résultante \(R_{1\times4}\) \DUrole{bibtex}{{[}Alloprof{]}}. Par la suite, la matrice \(B_{1\times4}\) contenant les biais de chaque neurone de la couche est additionée à la matrice R, dans une opération où s’additionnent entre\sphinxhyphen{}eux les éléments correspondants de chaque matrice pour former une nouvelle matrice de même dimension. Finalement, dans une itération au travers de cette matrice, chaque élément passe par la fonction d’activation, pour former encore une nouvelle matrice de même dimensions contenant les résultats de cette dernière opération. Cette matrice résultante finale \(F_{1\times4}\) devient alors l’intrant de la couche suivante de neurones, et ainsi de suite. \subsection{Réseaux neuronaux et le cerveau humain} \label{\detokenize{notions_de_base:reseaux-neuronaux-et-le-cerveau-humain}} Plusieurs liens peuvent être faits entre les réseaux neuronaux et le cerveau humain. Le premier réseau neuronal était un système mécanique financé par la marine américaine qui tentait d’émuler les neurones biologiques. La fonction d’activation à échelons était utilisée, imitant les neurones biologiques qui s’activent \sphinxstyleemphasis{1} ou ne s’activent pas \sphinxstyleemphasis{0}. Le projet a rapidement été laissé de côté, principalement à cause du fait que le réseau était extrêmement difficile à entrainer, puisque, comme vu plus tôt, la fonction à échelon ne permet pas l’entrainement du réseau, et les paramètres devaient être ajustés au hasard. Voilà donc une première différence fondamentale entre les neurones artificiels et organiques. Les neurones artificiels peuvent sortir toutes sortes de valeurs, de manière à mieux servir les intérêts du système, alors que dans le cas d’un neurone organique, elles ne peuvent envoyer que le signal binaire \sphinxstyleemphasis{activé} ou \sphinxstyleemphasis{non\sphinxhyphen{}activé}. \begin{quote} Un neurone s’active lorsque son seuil d’excitation est atteint. Le potentiel de repos d’un neurone est d’environ \sphinxhyphen{}50mV. Lorsqu’il reçoit suffisament de neurotransmetteurs (des particules envoyées par d’autre neurones et qui possèdent une charge électrique), par ses dendrites et que le seuil d’excitation d’environ 15mV est atteint, le potentiel d’action se déclenche, et un influx nerveux se propage le long de l’axone sous forme de courant électrique. Une fois arrivé aux terminaisons axonales du neurone, d’autres neurotransmetteurs sont libérés par les synapses, poursuivant ainsi la transmission du signal. La quantité de neurotransmetteurs libérée ne dépend pas de l’intensité du stimulus initial ; c’est une situation de tout ou rien. \sphinxcite{zbib:futura-sciences} \end{quote} En d’autres termes, l’image de la fonction d’un neurone artificiel \sphinxstyleemphasis{A}, dépendamment de la fonction d’activation, peut être, par exemple \(\mathbb{R}\) , \(\mathbb{R^+}\), ou encore \([-1, 1]\), tandis que l’image de la fonction d’un neurone organique \sphinxstyleemphasis{O} est toujours limité à \(\text{{0, 1}}\). L’aspect où les réseaux neuronaux et le cerveau humain ont le plus en commun est leur état initial. Les deux commencent comme un canvas vierge, ne possédant aucune connaissances ou expériences. Les deux se font « entrainer » par des informations extérieurs, jusqu’à arriver au point ou ils deviennent autonomes. Les connaissances qu’ils amassent se trouvent d’une certaine manière encodées dans leur système, et influencent leurs actions futures. \section{L’entrainement d’un système neuronal} \label{\detokenize{training:l-entrainement-d-un-systeme-neuronal}}\label{\detokenize{training::doc}} L’entrainement d’un réseau à l’aide d’une certaine base de données (donnée d’entrainement) permet à celui\sphinxhyphen{}ci de prédire le résultat d’une autre base donnée. En effet, le but d’un réseau neuronal est de réduire l’erreur de l’entrainement ainsi que la différence entre l’erreur des données entrainées et l’erreur des données de test soient petites. Lorsque le réseau est sous\sphinxhyphen{}entrainé, le réseau de sera pas précis lors de ces résultats. Cependant, lorsque le réseau est sur\sphinxhyphen{}entrainé, celui\sphinxhyphen{}ci va prendre en compte tout le bruit des données. Ce bruit peut être, par exemple, le fait de prendre en compte les imperfections d’une image, reconnaitre seulement certains styles d’écriture, etc. Cela a comme impact d’augmenter l’erreur lorsque le système est exposé à une nouvelle base de données. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{overfitting}.png} \caption{Graphiques représentant l’effet de l’entrainement du réseau de neurone}\label{\detokenize{training:overfitting}}\end{figure} L’entrainement d’un réseau neuronal s’effectue à l’inverse. Visuellement, l’entrainement et l’ajustement des différents paramètres se font de la droite vers la gauche. Ce principe, appelé « backpropagation », va être expliqué à l’aide quelques démonstrations mathématiques complémentées par quelques explications écrites. \subsection{Fonction d’erreur} \label{\detokenize{training:fonction-d-erreur}} Une fonction d’erreur est une fonction permettant de connaitre la précision des résultats des extrants de la dernière couche. Il peut y avoir plusieurs fonctions d’erreur. En voici un exemple: \(E_{SS}=1/2\sum_{i=1}^nE_i^2 \) \sphinxstylestrong{(1.1)} \(E_{SS}=1/2\sum_{i=1}^nE_i^2 \) \sphinxstylestrong{(1.2)} où \(E_{SS}\)= « error sum of square ». Cela est tout simplement une de plusieurs fonctions d’erreur. \(E_i =|{t_i-I_i}|\) \sphinxstylestrong{(1.3)} où \(E_i\) correspond à l’erreur d’un neurone de la dernière couche (extrant). \(I_i\) correspond à la valeur numérique d’un extrant et \(t_i\) correspond à la valeur désirée provenant de la base de données fournies. Combiner les deux équations permet d’obtenir: \(E=1/2\sum_{i=1}^n({T_i-Y_i})^2\) \sphinxstylestrong{(1.4)} \subsection{Transmition de l’information} \label{\detokenize{training:transmition-de-l-information}}\begin{quote} Note: Afin de simplifier les explications, ces dernières seront faites en utilisant un réseau neuronal ayant seulement 1 neurone par couche. \end{quote} D’abord, il faut comprendre comment le réseau transmet son information de cellules en cellule. En effet, un neurone ayant contenant une certaine valeur \(Y\) transmet cette dernière à tous les autres neurones de la prochaine couche. Cependant, ces transmitions n’ont pas toutes les mêmes poids. Ces poids \(p\) diffèrent afin de favoriser certaines activations et en défavoriser d’autres. Chaque liaison entre chaque neurone possède un poid propre à chacune. Ces derniers sont multipliés avec l’extrant de la neurone en précédentes. \(Y_{i} = Y_{i-1}\times p_{i}\)\sphinxstylestrong{(2.0)} où \(p_{i}\) correspond au poid de la neurone de la couche i Ensuite, un biais \(b\) est additionné ou soustrait au résultat précédent \(Y_i = Y_{i-1}\times p_{i} + b_i\) \sphinxstylestrong{(2.1)} d’activation sera expliqué en détail plus loin.où \(b_i\) correspond au biais de la neurone de la couche i. Finalement, une fonction d’activation \(a\) est ajoutée au reste de la formule. L’utilité et le fonctionnement de la fonction d’activation sera expliqué en détail plus loin. \(Y_i = a\times(Y_{i-1}\times p_{i} + b_i)\) \sphinxstylestrong{(2.2)} \subsection{\sphinxstyleemphasis{Back propagation}} \label{\detokenize{training:back-propagation}} L’objectif est de comprendre comment le poids et le biais doit être ajuster en débutant de la fonction d’erreur et d’activation. Dabord, en utilisant la formule de base de transmission d’un neurone (sans le biais) : \(Y = \sum_{i=1}^{n} I_i \times p_i \) Il est possible de comprendre comment le changement d’une variable impact une autre. Les dérivés seront donc utilisées afin de démontrer ce principe. \(\frac{dY}{dI_i}=\frac{dY}{dI_i}\sum_{i=1}^{n} I_i \times p_{ji} \) \(\frac{dY}{dI_i} = p_i\) \(\frac{dY}{dp_i} = I_i\) Cela veut donc dire que le poid influence le résultat de l’extrant et que l’intrant influence le résultat de l’extrant. En utlisant la formule (1.4) et le concept de dérivée partielle, il est possible de comprendre l’impact d’un changement de la valeur de l’intrant \(I_i\) sur l’erreur: \(\frac{dE}{dI_i} = (2/2)(t_i - I_i)(-1) \) \(\frac{dE}{dI_i}= -(t_i-I_i)\) Maintenant, il faut calculer la dérivation de la fonction d’activation. La fonction sigmoïde sera utilisée pour cet exemple. \(a = \frac{1}{1 + e^{-Y}} =(1+e^{-Y})^{-1}\) \(\frac{da}{dY} = -1 (-e^{-Y})(1+e^{-Y})^{-2} \) \(= \frac{e^{-Y}}{(1+e^{-Y})^2} \) \(= \frac{1}{(1+e^{-Y})}\times\frac{e^{-Y}}{(1+e^{-Y})} \) \(= a \times \frac{e^{-Y}}{(1+e^{-Y})}\) \(= a \times \frac{1+e^{-Y}-1}{(1+e^{-Y})} \) \(= a \times (\frac{(1+e^{-Y})}{(1+e^{-Y})} + \frac{-1}{(1+e^{-Y})})\) \(\frac{da}{dY}= a \times (1-a)\) Maintenant il est possible, à l’aide de la règle de dérivation en chaine, de trouver l’impact qu’a \(Y\) sur l’erreur \(E\). Dans cet exemple, \(I_i = a \) puisque la fonction d’activation été appliquée au neurone en question. \(\frac{dE}{dY_i} = \frac{dE}{dI_i} \times \frac{dI_i}{dY_i}\) \(= \frac{dE}{dI_i} \times \frac{da}{dY_i}\) \(=-(t_i - I_i) I_i (1- I_i)\) \sphinxstylestrong{(3.0)} Ensuite il est possible de calculer la dérivation de l’erreur en fonction du poid \(p_{ji}\) d’une liaison entre deux neurones. \(\frac{dE}{dp_{ji}} =\frac{dE}{dY_i} \times \frac{dY_i}{dp_{ji}} \) \(= (-(t_i - I_i) \times I_i\times (1- I_i))\times I_i\) \(= -I_i I_j (1-I_i)(t_i-I_i)\)\sphinxstylestrong{(4.0)} Cette équation signifie que le changement de l’erreur influence le poid et cette influence correspond à l’extrant d’un neurone négatif multiplié par l’extrant du neurone précédent et ce tout est multplié par 1 moins la valeur du neurone. À ce résultat est multiplié la valeur de l’erreur soit : \((t_i-I_i)\) L’équation 3.0 sera représenté par la variable: \(\Delta p\) Le concept de « backpropagation » se résume donc a: \(p_{ji} = p_{ji} + \Delta p\) Le poids d’un neurone change légèrement en additionnant un \(\Delta p\) positif ou négatif. Ce changement est fait avec une plus grande importance plus le neurone est proche de la couche des extrants. Cela est dû au fait que l’apprentissage commence par la couche finale pour enfin se rendre jusqu’à la couche débutant le système neuronal. Les premiers ajustements, donc ceux des couches plus proches de la fin, sont plus importants. Les couches se situant plus au début du réseau vont plutôt avoir de petits changements à leur poids puisque rendu à l’ajustement de dce dernier, l’erreur est déja considérablement réduite. Ce concept se nomme descente de gradient stochastique. \#\#\#Origine du biais Certains problèmes peuvent survenir avec le concept de descente de gradiant dans un réseau neuronal. En effet, lorsqu’une couche « n’apprend plus » ou, en d’autres mots, lorsque le poids ne varie plus, on assiste a un problème se nommant la disparition du gradiant. Cela est un problème pour le réseau puisque le tout l’entrainement se fait uniquement dans les dernières couches. Un autre problème est que le gradiant dans une fonction de coût telle une sigoïde, le gradient se situe uniquement au milieu de la fonction comme le montre le graphique ci\sphinxhyphen{}dessous. \begin{figure}[htbp] \centering \capstart \noindent\sphinxincludegraphics{{tanh}.png} \caption{Fonction sigmoïde}\label{\detokenize{training:tanh}}\end{figure} En effet, les extrémités de la fonction forment un plateau. Il n’y a donc pas de changement possible puisque soit l’intrant est multiplié par 1, ce qui ne change pas le résultat, ou bien soit le résultat est multiplié par 0 ce qui rend la valeurr nul et cela est néfaste pour un réseau neuronal. En effet, une valeur égale a 0 empêche l’entrainement du réseau puisque peut importe la variation du poid, \(0\times p\) sera toujours être égale à 0. C’est pour remédier à cette erreur qu’un biais est ajouté à la fonction. \(Y =\sum_{i=1}^{n} I_i \times p_i + b\) L’ajout de ce biais va permettre de conserver un apprentissage même lorsque la valeur d’un neurone est figée à 0. \section{Bienfaits et inconvénients} \label{\detokenize{bienfaits_et_inconv_xe9nients:bienfaits-et-inconvenients}}\label{\detokenize{bienfaits_et_inconv_xe9nients::doc}} Il est important de parler des impacts positifs et négatifs de l’intelligence artificielle dans le but de bien comprendre ce que cette technologie peut apporter à l’évolution de la société. C’est pour cela que cette section va répondre à la sous\sphinxhyphen{}question de la thèse: Est\sphinxhyphen{} ce que l’intelligence artificielle peut constituer un bénéfice pour l’être humain. Pour commencer, l’intégration de l’intelligence artificielle à notre vie de tous les jours ainsi que dans plusieurs tâches pourrait amener plusieurs bénéfices importants. Qui dit programme, dit souvent un taux d’erreur qui diminue. En effet, l’erreur humaine est présente et joue un gros rôle dans plusieurs métiers. L’erreur est humaine , mais minimiser l’effort tout en maximisant la précision est primordial pour certaines tâches. L’utilisation d’une intelligence artificielle bien entraînée permet de réduire le taux d’erreurs et le temps requis à l’exécution d’une tâche. Cela est possible, car un programme assisté par l’intelligence artificielle peut parcourir un large jeu de données et appliquer des algorithmes en peu de temps, tandis que l’être humain prendrait des mois et des mois pour faire cela. Un exemple moderne serait l’utilisation de l’intelligence artificielle dans les prévisions météo. Ceux\sphinxhyphen{}ci deviennent donc de plus en plus précis avec l’aide de l’apprentissage machine ainsi que l’amélioration des algorithmes. L’IA peut assister l’être humain dans des tâches dangereuses qui pourraient risquer la vie d’individus, mais qui est sans risques pour l’intelligence artificielle. Un exemple serait le robot \sphinxhref{https://mars.nasa.gov/msl/home/}{Curiosity} lancé en novembre 2011 dans le but d’atteindre la planète Mars. Il se trouve toujours sur cette planète en date du 19 novembre 2020 dans des conditions impossibles pour qu’un être humain y soit resté aussi longtemps. Celui\sphinxhyphen{}ci est contrôlé par l’intelligence artificielle dans le but d’accomplir plusieurs tâches. Ses tâches consistent à examiner la biologie, la géologie et la radiation de Mars, afin de préparer une exploration humaine dans le futur. Les conditions de travail standard dictent qu’un être humain travaille environ 40 heures par semaine en incluant les pauses. De plus, ce ne sont pas toujours 40 heures productives. Les fins de journées et les fins de semaine sont souvent peu productives. En comparaison, un ordinateur n’a pas besoin de pause et peut travailler 24 heures sur 24, 7 jours sur 7. L’intelligence artificielle sera sûrement plus efficace qu’une majorité des travailleurs et aussi plus constante dans sa vitesse d’exécution. Cet ennui est souvent causé souvent par des travaux répétitifs. Ces tâches banales peuvent être automatisées par l’IA. Donnant plus de temps aux travailleurs pour se concentrer sur des tâches plus exigeantes et passionnantes. Comme les ordinateurs le font déjà dans plusieurs domaines, tel celui de la finance où un apprentissage machine s’occupe de détecter des fraudes potentielles dans les transactions des clients. Quand on parle de rapidité, l’être humain est très lent comparé à l’intelligence artificielle. En effet, l’utilisation de l’intelligence artificielle pour la prise de décision permet de prendre ceux\sphinxhyphen{}ci extrêmement rapidement avec une précision qui dépend de la qualité de leurs bases de données. Effectivement, un jeu de données fiable et bien construit permet à l’intelligence artificielle de prendre des bonnes décisions rapidement. Un être humain, de l’autre côté, doit analyser une plus petite quantité de données directement sous son nez en prenant plus de temps. Un des exemples serait l’utilisation de l’intelligence artificielle dans les jeux d’échecs. L’intelligence artificielle à accès à des milliers et des milliers de parties dans son jeu de données. Alors, lorsque son adversaire humain joue le coup Nc3 (Cavalier sur la case C3), l’ordinateur analyse toutes les parties où ce coup c’est fait et en ressort le meilleur coup pour contrer celui\sphinxhyphen{}ci, et ce, en un temps record. De plus, la prise de décision de l’intelligence artificielle ne va pas seulement s’arrêter au jeu d’échec. Nous voyons déjà l’industrie médicale se faire aider dans la prise de décision par l’intelligence artificielle en plus des nouveaux programmes qui se font développer en ce moment. Ceux\sphinxhyphen{}ci ont pour but d’aider les professionnels de la santé dans des diagnostics, surveillances de patients, et plusieurs autres. La compagnie Google est en train de développer leur projet \sphinxhref{https://health.google/}{Google Health} Pour se faire, ils étudient l’utilisation de l’intelligence artificielle dans le but d’assister au diagnostic de cancer, prévenir l’aveuglement et plusieurs autres. La croissance exponentielle de l’intelligence artificielle dans le monde a affecté de manière abrupte le nombre de travail qui requiert des capacités en intelligence artificielle, et donc, a créé plusieurs nouveaux emplois dans ce domaine. Grâce à cela, la part des emplois qui requièrent des capacités en intelligence artificielle a augmenté de 4,5 fois depuis 2013 aux États\sphinxhyphen{}Unis sur le site web \sphinxhref{https://ca.indeed.com/?r=us}{indeed.com}. Certes, ces merveilleux bénéfices ne viennent pas sans coût. Or, l’intelligence artificielle s’avère à être une arme à double tranchant et continents plusieurs problèmes Un de ces problèmes majeurs est le biais de l’intelligence artificielle intégrée dans le modèle de manière involontaire ou intentionnelle par la façon dont celui\sphinxhyphen{}ci est entraîné ou programmé. Un exemple flagrant serait le programme Correctional Offender Management Profiling for Alternative Sanctions (\sphinxhref{https://en.wikipedia.org/wiki/COMPAS\_(software)}{COMPAS}) qui a signalé à tort des personnes ayant la peau foncée presque deux fois plus que des personnes ayant la peau blanche(45 \% à 24 \%) devant la cour juridique aux États\sphinxhyphen{}Unis. Un autre exemple serait l’intelligence artificielle programmée par Microsoft Corporation dans le but d’interagir avec les internautes. Prénommé «\sphinxhref{https://twitter.com/tayandyou?lang=en}{Tay}», le bot s’est rapidement fait retirer après un entraînement avec l’interaction d’internautes racistes. Cela a mené le robot à dire des phrases vulgaires et inappropriées avant que celui\sphinxhyphen{}ci voit la fin de ses jours 16 heures après son arrivée sur Twitter. Le problème provient donc d’un jeu de données non fiable où mal construit qui induit l’apprentissage machine en erreur. Tel que mentionné plus haut, l’intelligence artificielle a apporté et va apporter plusieurs nouveaux emplois spécialisés dans l’intelligence artificielle, mais plusieurs autres emplois vont voir leur fin arriver à grands pas avec la montée de l’intelligence artificielle. Cela c’est déjà fait avec la révolution industrielle, qui, avec l’arrivée de la machine à vapeur, avait fait grimper le taux de chômage à une vitesse fulgurante. L’intelligence artificielle devrait, tout comme la machine à vapeur, augmenter le besoin de certains emplois et créer des emplois qui requièrent un plus haut niveau d’étude en éducation tout en éradiquant des emplois auxquels aucune caractéristique uniquement humaine est requise. Il reste à noter que l’être humain s’en est quand même bien sortie de la révolution industrielle, contrairement aux idées pessimistes quant au futur proliférées à l’époque. Par ailleurs, l’omniprésence de la technologie a créé une multitude d’interconnexions entre tous les pays. En plus, avec l’implémentation de l’intelligence artificielle, chaque pays va définir des lois et des règlements sur l’intelligence artificielle. Dans un monde idéal, tous les êtres humains s’entendent sur les mêmes lois et règlements pour éviter des conflits avec d’autres pays. Ces conflits seraient à cause de décisions par rapport à l’intelligence artificielle qu’un pays a choisi et qui mène à une contradiction avec un ou plusieurs autres pays. Malheureusement, comme l’Histoire nous l’a si bien démontré, essayer d’établir des règles communes s’avère difficile, et de les faire respecter, encore plus. Des différences se font déjà voir entre les puissances du monde face à l’intelligence artificielle avec l’Union Européenne qui est déjà entrain de pousser pour des mesures plus strictes pour le développement et l’utilisation de l’intelligence artificielle avec le ‘White Paper on Artificial Intelligence \textendash{} A European Approach to Excellence and Trust’ ,publié en 2020. Au contraire, les États \sphinxhyphen{} Unis et la Chine permettent à leur compagnie d’utiliser l’intelligence artificielle plus librement. L’intelligence artificielle va augmenter l’efficacité et la rapidité de plusieurs programmes. Cela n’exclut pas les programmes de piratages informatiques, qui eux aussi vont voir une amélioration drastique avec l’implémentation de l’intelligence artificielle. Cela va donc augmenter aussi la rapidité et l’efficacité des piratages informatiques menant sûrement à une augmentation de ceux\sphinxhyphen{}ci ce qui causera plusieurs problèmes majeurs, jusqu’à temps qu’une solution soit trouvée. L’utilisation de l’intelligence artificielle dans des buts de tuer des individus est un grave danger. Tel que mentionné dans le rapport, l’entraînement de l’intelligence se fait à partir d’un grand jeu de données fiables. Dans un contexte d’une guerre contre le terrorisme, avoir un jeu de données sur les terroristres s’avère très difficile et peu fiable, car la plupart d’entre eux s’habille comme des civiles. De plus, cette technologie, une fois tombée dans les mains des terroristes, pourrait semer terreur au sein d’un pays comme le démontre très bien la vidéo \sphinxhref{https://www.youtube.com/watch?v=HipTO\_7mUOw\&ab\_channel=FutureofLifeInstitute}{Slaughterbots} qui promeut l’interdiction de l’usage robot tueur. En conclusion, l’implémentation de l’intelligence artificielle ainsi que la croissance exponentielle de celle\sphinxhyphen{}ci va rapporter une tonne de bénéfices tel l’élimination de l’erreur humaine, l’utilisation de celle\sphinxhyphen{}ci dans des tâches dangereuses, la disponibilité 24/7 de celle\sphinxhyphen{}ci, la rapidité de prise de décision et les emplois dans ce domaine. De l’autre côté, si l’on veut que ces bénéfices portent fruit, il faut limiter ou éliminer les impacts négatifs de l’intelligence artificielle. Cela va se faire graduellement par l’implémentation de règles mondiales sur les jeux de données utilisés dans l’intelligence machine, l’intelligence machine en temps que tel, l’utilisation de l’intelligence artificielle dans un environnement de guerre, et ce, dans l’accord de la majorité des pays en plus de développer et améliorer la cybersécurité. Une idée intéressante pour régler le problème des jeux de données pourrait être de standardiser les jeux de données qui ne sont pas biaisés dans le but d’éviter les problèmes. C’est déjà ce que le NIST fait en fournissant des jeux de données fiables gratuitement. Finalement, quel est le fonctionnement de l’intelligence artificielle et comment devrait\sphinxhyphen{}elle être utilisée afin de bénéficier l’être humain? L’hypothèse émise était que si son développement se fait de manière éthique et s’il est bien encadré, nous pourrions en retirer plus d’avantages que d’inconvénients. Il était donc important de comprendre le fonctionnement de cette technologie pour pouvoir expliquer et rationaliser l’utilisation bénéfique de l’intelligence artificielle et d’ainsi le documenter. \section{Conclusion} \label{\detokenize{conclusion:conclusion}}\label{\detokenize{conclusion::doc}} Finalement, quel est le fonctionnement de l’intelligence artificielle et comment devrait\sphinxhyphen{}elle être utilisée afin de bénéficier l’être humain? L’hypothèse émise était que si son développement se fait de manière éthique et s’il est bien encadré, nous pourrions en retirer plus d’avantages que d’inconvénients. Il était donc important de comprendre le fonctionnement de cette technologie pour pouvoir expliquer et rationaliser l’utilisation bénéfique de l’intelligence artificielle et d’ainsi le documenter. Pour ce faire, il a fallu procéder à l’écriture d’un programme d’OCR, une forme simple d’intelligence artificielle qui a pour but de reconnaître des caractères écrits à la main. Grâce à l’information acquise lors de l’écriture de celui\sphinxhyphen{}ci ainsi que toute la documentation lue pour la préparation du programme, il a été plus facile de comprendre les aspects d’un réseau neuronal dans le but de documenter chaque composante du réseau neuronal. Ce que nous pouvons conclure d’un réseau neuronal après notre documentation, c’est que cette “intelligence” artificielle, n’est rien d’autre qu’un paquet de fonctions avec des paramètres qui ont été ajustés par une méthode d’optimisation dénommée “Back propagation”. Cette méthode est composée de fonctions qui utilisent des notions de mathématiques telles que les dérivées. Cet algorithme a lieu lors de chaque “epoch% \begin{footnote}[2]\sphinxAtStartFootnote Une “epoch” étant un cycle complet où l’algorithme a traité le jeu de données qui lui a été fournie une seule fois. % \end{footnote}” dans le but d’ajuster l’algorithme d’OCR graduellement. Donc, cet algorithme « intelligent ” n’est pas capable de prendre des décisions elle\sphinxhyphen{}même et doit être surveillé et entraîné par un être humain à l’aide d’un jeu de données.Cela dans le but de pouvoir répondre à la tâche précise à la\sphinxhyphen{}quel l’algorithme s’est fait assigner, qui est dans notre cas la reconnaissance optique des chiffres de 0 à 9. Alors, un algorithme qui est bon ou mauvais dans sa tâche et donc qui a une bonne ou mauvaise précision est principalement déterminé par la façon dont l’algorithme a été entraîné et si le jeu de données utilisé pour l’entraîner est fiable et diversifié. Dans notre cas, la précision est calculée par le nombre de prédiction réussite sur le nombre total de prédiction.C’est donc l’entraînement que provient une partie des biais de l’intelligence artificielle, et donc un des inconvénients qu’on avait cité dans notre hypothèse. L’autre inconvénient qui a été mentionné lors de l’hypothèse est l’importante quantité d’emploi qui risque de disparaître. Or, tel que mentionné dans la section sur les impacts de l’intelligence artificielle, il n’y a pas seulement cet inconvénient, et ces autres inconvénients sont accompagnés de plusieurs avantages intéressants. Comme il est décrit dans la section des bienfaits et inconvénients, l’usage de l’apprentissage machine peut amener beaucoup d’avantages ainsi que des inconvénients. Ces principaux avantages se résument à, l’élimination de l’erreur humaine, l’utilisation de l’IA dans des tâches dangereuses, la disponibilité 24/7 de celle\sphinxhyphen{}ci, la rapidité de prise de décision et les emplois dans ce domaine. Tandis que les inconvénient se résument au biais de l’IA, diminution de certains types d’emploi, les conflits mondiaux lié à celle\sphinxhyphen{}ci, l’amélioration des “hacks” et l’utilisation de celle\sphinxhyphen{}ci dans le but de tuer. Si l’on veut avoir les bénéfices de l’intelligence artificielle, il faut d’abord s’occuper de diminuer l’impact des inconvénients. La bonne nouvelle étant que ces inconvénients peuvent être quasi inexistants si l’IA est implémentée en suivant dans un bon cadre strict ainsi que des mesures de sécurité. Par exemple, des règles sur l’utilisation des bases de données et des réglementations sur celles\sphinxhyphen{}ci pourraient diminuer les biais dans les programmes d’apprentissage machine. Si ces conditions sont respectées, il est préférable de penser que l’usage de l’intelligence artificielle pourrait faire évoluer la société. Pour répondre à la thèse, l’intelligence artificielle fonctionne par l’étude de grosses bases de données par un programme. Celui\sphinxhyphen{}ci va ensuite faire varier ses paramètres clés dans le but d’ajuster certaines fonctions. Ces fonctions permettent de trouver des tendances dans le jeu de données et de les utiliser dans le but de répondre à la tâche attribuée. De plus, l’intelligence artificielle peut constituer un bénéfice pour l’être humain à condition que celle\sphinxhyphen{}ci soit bien encadrée. Il ne faut toutefois pas écarter le fait que le futur n’est jamais certain, et qu’on ne peut prédire à 100\% ce que l’intelligence artificielle va ressembler dans 50 ans. Mais, avec les avancées technologiques des dernières années qui ne semblent pas s’arrêter, il serait juste de croire que l’IA a un futur prometteur. Avec ces avancés prometteuses, ce pourrait\sphinxhyphen{}il qu’il y ait une sorte d’intelligence au\sphinxhyphen{}delà de celle humaine et artificielle dans le futur ? \bigskip\hrule\bigskip \section{Bibliographie} \label{\detokenize{zbib:bibliographie}}\label{\detokenize{zbib::doc}} \begin{sphinxthebibliography}{Les cont} \bibitem[InnovationQuebec, 2018]{zbib:gouvqc} et Innovation Québec, É. (2018). \sphinxstyleemphasis{Les avantages et inconvénients de l’intelligence artificielle}. \bibitem[futura\sphinxhyphen{}sciences, n.d.]{zbib:futura-sciences} futura\sphinxhyphen{}sciences (n.d.). \sphinxstyleemphasis{Potentiel d’action}. \bibitem[Goodfellow et al., 2016]{zbib:goodfellow-et-al-2016} ., ., \& . (2016). \sphinxstyleemphasis{Deep Learning}. MIT Press. http://www.deeplearningbook.org. \bibitem[Grother et al., 2019]{zbib:nistbias} ., ., \& . (2019). \sphinxstyleemphasis{Face Recognition Vendor Test (FRVT) Part 3: Demographic Effects}. NIST. \bibitem[LAURO, n.d.]{zbib:ubiquity} . M. (n.d.). \sphinxstyleemphasis{HUMAN BRAIN AND NEURAL NETWORK BEHAVIOR A COMPARISON}. \bibitem[Metz, 2020]{zbib:cnnportland} . (2020). \sphinxstyleemphasis{Portland passes broadest facial recognition ban in the US}. \bibitem[Nielsen, 2019]{zbib:michael} . A. (2019). \sphinxstyleemphasis{Neural Networl and Deep Learning}. \bibitem[Popper, 2016]{zbib:nytimes} Popper, N. (2016). \sphinxstyleemphasis{The Robots Are Coming for Wall Street}. \bibitem[RFisher \& Wolfart, 2003]{zbib:hipr2} . Perkins, A. W., \& . (2003). \sphinxstyleemphasis{Pixel Values}. \bibitem[Les contributeurs de Wikipedia, 2020]{zbib:wikitf} Les contributeurs de Wikipedia (2020). \sphinxstyleemphasis{TensorFlow}. \bibitem[The Numpy documentation team, 2020]{zbib:numpy} The Numpy documentation team (2020). \sphinxstyleemphasis{About us}. \bibitem[The TensorFlow developper team, 2015]{zbib:tfpaper} The TensorFlow developper team (2015). \sphinxstyleemphasis{TensorFlow: Large\sphinxhyphen{}Scale Machine Learning on Heterogeneous Distributed Systems}. \bibitem[The TensorFlow developper team, 2020]{zbib:tfmain} The TensorFlow developper team (2020). \sphinxstyleemphasis{TensorFlow}. \end{sphinxthebibliography} \renewcommand{\indexname}{Index} \printindex \end{document}doc/head.tex1-10 \usepackage{heuristica} \usepackage[heuristica,vvarbb,bigdelims]{newtxmath} \usepackage[T1]{fontenc} \renewcommand*\oldstylenums[1]{\textosf{#1}} lorisrossi/-Torrent \hypertarget{structt__udp_1_1connect__response}{}\section{t\+\_\+udp\+:\+:connect\+\_\+response Struct Reference} \label{structt__udp_1_1connect__response}\index{t\+\_\+udp\+::connect\+\_\+response@{t\+\_\+udp\+::connect\+\_\+response}} \subsection*{Public Attributes} \begin{DoxyCompactItemize} \item int32\+\_\+t \hyperlink{structt__udp_1_1connect__response_a72371c40b24c52782ef0d03ca61c57e3}{action} = 0 \item \mbox{\Hypertarget{structt__udp_1_1connect__response_af5f97e3b41bd74d238f78a2233bd53d5}\label{structt__udp_1_1connect__response_af5f97e3b41bd74d238f78a2233bd53d5}} int32\+\_\+t {\bfseries transaction\+\_\+id} \item \mbox{\Hypertarget{structt__udp_1_1connect__response_aa9e3aec7f9a33e92c59bcf548a1d83af}\label{structt__udp_1_1connect__response_aa9e3aec7f9a33e92c59bcf548a1d83af}} int64\+\_\+t {\bfseries connection\+\_\+id} \end{DoxyCompactItemize} \subsection{Member Data Documentation} \mbox{\Hypertarget{structt__udp_1_1connect__response_a72371c40b24c52782ef0d03ca61c57e3}\label{structt__udp_1_1connect__response_a72371c40b24c52782ef0d03ca61c57e3}} \index{t\+\_\+udp\+::connect\+\_\+response@{t\+\_\+udp\+::connect\+\_\+response}!action@{action}} \index{action@{action}!t\+\_\+udp\+::connect\+\_\+response@{t\+\_\+udp\+::connect\+\_\+response}} \subsubsection{\texorpdfstring{action}{action}} {\footnotesize\ttfamily int32\+\_\+t t\+\_\+udp\+::connect\+\_\+response\+::action = 0} Connection Value The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item lib/\hyperlink{tracker__udp_8hpp}{tracker\+\_\+udp.\+hpp}\end{DoxyCompactItemize} content/publication/moruzzi-2019-dysregulated/moruzzi-2019-dysregulated.bib @article{moruzzi2019dysregulated, author = { Nestor-Bergmann, }, doi = {10.1101/578146}, journal = {bioRxiv}, month = {May}, pages = {578146}, publisher = {Cold Spring Harbor Laboratory}, title = {Dysregulated wild-type cell division at the interface between host and oncogenic epithelium}, url = {https://www.biorxiv.org/content/10.1101/578146v1.abstract}, year = {2019} } dipteshkanojia/academic-kickstart @misc{sharma2022ensemble, title={An Ensemble Approach to Acronym Extraction using Transformers}, author={ and and and and }, year={2022}, eprint={2201.03026}, archivePrefix={arXiv}, primaryClass={cs.CL} } \documentclass[11pt]{article} \usepackage[left=2cm, right=2cm, top=2cm, bottom=2cm]{geometry} \usepackage{tcolorbox} \tcbuselibrary{breakable} \usepackage{amsfonts} \usepackage{amsmath} \usepackage{amssymb} \usepackage{newtxmath} \usepackage{hyperref} \linespread{1.3} \setlength{\parskip}{3mm} \setlength{\parindent}{2em} % \usepackage[backend=biber]{biblatex} % \addbibresource{my.bib} \numberwithin{equation}{section} % Problem \newcounter{exercise}[section] \newenvironment{exercise}[1][\textsc{Exercise }\thesection.\refstepcounter{exercise}\theexercise]{\begin{tcolorbox}[colback=black!15, colframe=black!80, breakable, title=\textsc{Exercise }#1]}{\end{tcolorbox}} % Example \newcounter{example}[section] \newenvironment{example}[1][\textsc{Example }\thesection.\refstepcounter{example}\theexample]{\begin{tcolorbox}[colback=black!15, colframe=black!80, breakable, title=#1]}{\end{tcolorbox}} % Theorem \newcounter{theorem}[section] \newenvironment{theorem}[1][\textsc{Theorem }\thesection.\refstepcounter{theorem}\thetheorem]{\begin{tcolorbox}[colback=black!15, colframe=red!60, breakable, title=#1]}{\end{tcolorbox}} \newenvironment{proof}{\begin{tcolorbox}[colback=white, colframe=black!50, breakable, title=Proof. ]\setlength{\parskip}{0.8em}}{\end{tcolorbox}} \newenvironment{solution}{\begin{tcolorbox}[colback=white, colframe=black!50, breakable, title=Solution. ]\setlength{\parskip}{0.8em}}{\end{tcolorbox}} \newcommand{\pder}{\partial\,} \newcommand{\der}{\,\mathbf{d}\,} \title{\textsc{Probability: Final Exam}} \author{\textsc{}} \date{\emph{\today}} \begin{document} \maketitle \begin{exercise}[2.3.19] Let $X_{n}$ be independent Poisson r.v.'s with $E X_{n}=\lambda_{n}$, and let $S_{n}=X_{1}+\cdots+X_{n}$. Show that if $\sum \lambda_{n}=\infty$, then $S_{n} / E S_{n} \rightarrow 1$ a.s. \end{exercise} \begin{solution} Because $X_n$ is sampled from poisson distribution, we can know that $Var(X_n)=\lambda_n=EX_n$. $X_n$ are independent, $Var(S_n)=\sum_nVar(X_n)$. So, \[ P\left(\left|\frac{S_n}{ES_n}-1\right|>\varepsilon\right)\leqslant\frac{Var(S_n/ES_n)}{\varepsilon^2}=\frac{\sum_{n}Var(X_n)}{(ES_n)^2\varepsilon^2}=\frac{1}{\sum_n\lambda_n\varepsilon^2}\to 0. \] Hence, $\frac{S_n}{ES_n}\to 1$ a.s. \end{solution} \begin{exercise}[3.2.1] Give an example of random variables $X_{n}$ with densities $f_{n}$ so that $X_{n} \Rightarrow$ a uniform distribution on $(0,1)$ but $f_{n}(x)$ does not converge to 1 for any $x \in[0,1]$. \end{exercise} \begin{solution} We consider a piecewise function as the density of $X_n$. \[ f_n(x)=a, x\in\left(\frac{2i}{2^n}, \frac{2i+1}{2^n}\right), 0\leqslant i\leqslant2^{n-1}-1. \] $f_n(x)=0$, for other situations. Because $\int_0^1f_n(x)\der x=1$, we have \[ \int_0^1f_n(x)\der x=\sum_{i=0}^{2^{n-1}-1}\int_{2i/2^n}^{(2i+1)/2^n}a\der x=1. \] Solve it, we can get that $a=2$. So, when $n\to\infty$, \[ f_n(x)\to f(x)=2, x\in(0,1). \] For any $x\in[0,1]$, $f_n(x)\neq1$. \end{solution} \begin{exercise}[3.3.1] Show that if $\varphi$ is a ch.f., then $Re(\varphi)$ and $|\varphi|^2$ are also. \end{exercise} \begin{solution} \begin{enumerate} \item $Re(\varphi)=Re(E(e^{itX}))=Re(E(\cos(tX)+i\sin(tX)))=E(\cos(tX))$, let $f(t)=Re(\varphi(t))$. \begin{itemize} \item $f(0)=1$. \item $f(-t)=Re(\varphi(-t))=E(\cos(-tX))=E(\cos(tX))=\overline{f(t)}$. \item $|E(\cos(tX))|\leqslant E|\cos(tX)|=1.$ \item \ \\ \vspace{-30pt}\[ \begin{aligned} \left|E(Re(e^{i(t+h) x}))-E(Re( e^{i t x})) \right| &=\left|E\left(Re( e^{i(t+h) x})-Re (e^{i t X})\right)\right| \\ &=\left|E( Re( e^{i t X}e^{i h x}-1))\right|\\ & \leqslant E\left|e^{i t X}\right|\left|e^{i h x}-1\right| \\ &=E\left|e^{i h x}-1\right| . \end{aligned} \] So, it is uniformly continuous. \item \ \\ \vspace{-30pt}\[ \begin{aligned} Re(E(e^{it(aX+b)}))=E ((e^{i t b}) Re (e^{i t a X})) &=e^{i t b} E \cos(t a X) \\ &=e^{i t b} Re( \varphi(a t)). \end{aligned} \] \end{itemize} So, $Re(\varphi)$ is a ch.f. \item $|\varphi|^2=\left(E\cos(tX)\right)^2+\left(E\sin(tX)\right)^2$. Similarly, we can verify it is a ch.f. \begin{itemize} \item Uniformly continuous: \[ \begin{aligned} \left|E e^{i(t+h) X}\right|^{2}-\left|E e^{itX}\right|^{2} &= \left(| E e^{i(t+h) X}|-| E e^{i t X} |\right)\left(|E e^{i(t+h) X}|+| E e^{i t X} |\right) \\ & \leqslant 2 E| e^{i(t+h) X}-e^{i t X} | \\ & \leqslant 2 E\left|e^{i t X}(e^{i h X}-1)\right|\\&=2 E| e^{i h x}-1 \mid . \end{aligned} \] \item Other conditions are easily to verify. \end{itemize} So, $|\varphi|^2$ is a ch.f. \end{enumerate} \end{solution} \begin{exercise}[3.4.4] Let $X_1, X_2, \cdots$ be i.i.d. with $X_i \geqslant 0$, $E(X_i) = 1$, and $Var(X_i)= \sigma^2\in(0,\infty)$. Show that \[ 2\left(\sqrt{S_n}-\sqrt{n}\right)\Rightarrow\sigma\,\chi. \] \end{exercise} \begin{solution} Let $2\sqrt{S_n}-\sqrt{n}$ as a r.v., and $F_n(x)$ as the distribution function. Then, \[ \begin{aligned} F_{n}(x) &=P\left(2\left(\sqrt{S_{n}}-\sqrt{n}\right) \leqslant x\right) \\ &=P\left(\sqrt{S_{n}} \leqslant \frac{x}{2}+\sqrt{n}\right) \\ &=P\left(S_{n} \leqslant n+\frac{x^{2}}{4}+x \sqrt{n}\right) \\ &=P\left(\frac{S_{n}-n}{\sqrt{n}} \leqslant x+\frac{x^{2}}{4 \sqrt{n}}\right) \end{aligned} \] From C.L.T., we know that \[ \frac{S_{n}-n}{\sqrt{n}} \Rightarrow \sigma \chi \] So, \[ \lim _{n \rightarrow \infty} P\left(\frac{S_{n}-n}{\sqrt{n}} \leqslant x\right)=P(\sigma \chi \leqslant x) \] Let $n\to\infty$, \[ \begin{aligned} \lim _{n \rightarrow \infty} F_{n}(x) &=\lim _{n \rightarrow \infty} P\left(\frac{S_{n}-n}{\sqrt{n}} \leqslant x+\frac{x^{2}}{4 \sqrt{n}}\right) \\ &=\lim _{n \rightarrow \infty} P\left(\frac{S_{n}-n}{\sqrt{n}} \leqslant x\right) \\ &=P\left(\frac{S_{n}-n}{\sqrt{n}} \leqslant x\right) \\ &=P(\sigma \chi \leqslant x) \end{aligned} \] Hence, $\lim_{n\to\infty}F_n(x)$ has the same distribution with $\frac{S_n-n}{\sqrt{n}}$. $2\left(\sqrt{S_{n}}-\sqrt{n}\right) \Rightarrow \sigma \chi$. \end{solution} \end{document} \documentclass[nociteref]{SIAM-GH-book} \usepackage{hyperref} \usepackage{import} \usepackage{amsmath, amsfonts, amscd, amssymb} \usepackage{epsfig} \usepackage{graphicx} \usepackage{url} \usepackage{mathrsfs} \usepackage{makeidx} \usepackage{multicol} \usepackage{algorithmicx} \usepackage[plain]{algorithm} \usepackage[noend]{algpseudocode} \usepackage{color} \usepackage{verbatim} \usepackage{listings} \usepackage{float} \usepackage{paralist} \usepackage{caption} \usepackage{subcaption} \usepackage{bbm} \usepackage{textcomp} \usepackage{tikz} \usepackage[framemethod=tikz]{mdframed} \usepackage[style=alphabetic,refsection=chapter,backref=true]{biblatex} \usepackage{mathtools} \usetikzlibrary{automata,positioning} \input{command} \makeindex \begin{document} %---------------------------------------------------------------- %Book cover and Front matter \thispagestyle{empty} \begin{center} {\huge \bf Labs for Foundations of Applied Mathematics} \\ \vspace{5mm} {\Large \bf Volume III: Modeling with Uncertainty and Data} \vspace{20mm} \includegraphics[scale = .25]{Cover} \end{center} \frontmatter \include{contributors} %------------------------------------------------------------------ %The preface, which will presumably be longer in the future \begin{thepreface} This lab manual is designed to accompany the textbook \emph{Foundations of Applied Mathematics} by Humpherys and Jarvis. \vfill \copyright{This work is licensed under the Creative Commons Attribution 3.0 United States License. You may copy, distribute, and display this copyrighted work only if you give credit to Dr.~J.~Humpherys. All derivative works must include an attribution to Dr.~J.~Humpherys as the owner of this work as well as the web address to \\\centerline{\url{https://github.com/byuimpact/numerical_computing}}\\ as the original source of this work.\\To view a copy of the Creative Commons Attribution 3.0 License, visit\\\centerline{\url{http://creativecommons.org/licenses/by/3.0/us/}} or send a letter to Creative Commons, 171 Second Street, Suite 300, San Francisco, California, 94105, USA.} \vfill \centering\includegraphics[height=1.2cm]{by} \vfill \end{thepreface} %----------------------------------------------------------------- \setcounter{tocdepth}{1} \tableofcontents \mainmatter \part{Measure Theory} \subimport{./Python/SQL/}{DatabaseEssentials} \subimport{./Python/AdvancedSQL/}{advanced_sql} \subimport{./Python/pandas/}{pandas} \subimport{./Python/PandasII/}{pandasII} \part{Random Spaces and Variables} \subimport{./Python/StateMachines/}{StateMachines} \subimport{./Python/Scrapy/}{scraping} \subimport{./Python/WebTechnology/}{WebTechnology} \part{Distributions} \part{Limit Theorems} \subimport{./Labs/MPITrapezoidalRule/}{MPITrapezoidalRule} \subimport{./Labs/MPICollectiveCommunication/}{MPICollectiveCommunication} \part{Markov Processes} \part{Poisson Distributions, Queues, and Renewal} \subimport{./Python/MongoDB/}{MongoDB} \part{Martingales and Diffusion} \part{Information Theory} \part{Multivariable Statistics} \subimport{./Labs/PCA/}{PCA} \subimport{./Labs/LSI/}{LSI} \part{State Estimation} \subimport{./Labs/KalmanFilter/}{kalman} \subimport{./Labs/ProjectileTracking/}{projectiletracking} \subimport{./Labs/HMMStateEstimation/}{hmmstates} \subimport{./Labs/HMMNaturalLanguage/}{englishvowels} \part{Time Series} \part{Expectation Maximization} \subimport{./Labs/HMMParameterEstimation/}{hmmtrain} \subimport{./Labs/HMMRussian/}{russianalphabet} \subimport{./Labs/CDHMM/}{cdhmm} \subimport{./Labs/SpeechRecognition/}{speechrecognition} \part{Bayesian Statistics I} \subimport{./Labs/BayesianUpdate/}{bayesianupdate} \subimport{./Labs/SearchAndRescue/}{bayesiansearch} \part{Bayesian Statistics II} \subimport{./Labs/Metropolis/}{metropolis} \subimport{./Labs/IsingModel/}{isingmodel} \subimport{./Labs/GibbsSampling/}{gibbs} \subimport{./Labs/LDA/}{lda} \part{Machine Learning I} \subimport{./Labs/KMeans/}{kmeans} \subimport{./Labs/GMM/}{gmm} \subimport{./Labs/CrimeMapping/}{crimemapping} \part{Machine Learning II} \subimport{./Labs/CART/}{cart} \subimport{./Labs/TitanicSurvival/}{randomforest} \subimport{./Labs/KNNSVM/}{knnsvm} \subimport{./Labs/ImageRecognition/}{imagerecognition} \end{document} doxygen--q5/latex/structnode.tex \hypertarget{structnode}{}\section{node Struct Reference} \label{structnode}\index{node@{node}} {\ttfamily \#include $<$tree.\+h$>$} Collaboration diagram for node\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=164pt]{structnode__coll__graph} \end{center} \end{figure} \subsection*{Public Attributes} \begin{DoxyCompactItemize} \item int \hyperlink{structnode_a2d890bb9f6af0ffd73fe79b21124c2a2}{data} \item struct \hyperlink{structnode}{node} $\ast$ \hyperlink{structnode_a3ce38490a651bfda86d88ff955e96abc}{left} \item struct \hyperlink{structnode}{node} $\ast$ \hyperlink{structnode_a875f75abfe22103500535b179828e4e3}{right} \item struct \hyperlink{structnode}{node} $\ast$ \hyperlink{structnode_a05e4fe9e0177ba2d8dbd2c487cfddd53}{parent} \item int \hyperlink{structnode_a3871d43e823ba9542b052912d01709dd}{level} \item int \hyperlink{structnode_a64dd8b65a7d38c632a017d7f36444dbb}{x} \item int \hyperlink{structnode_ae944a3a75efb9856fa5c6f2221e2b49e}{y} \item int \hyperlink{structnode_aeef7855fea382bfb671d7834aefa4b22}{offset} \item bool \hyperlink{structnode_afd9ff5fa3c3ab99d07cac2a7ad9d14a6}{thread} \end{DoxyCompactItemize} \subsection{Member Data Documentation} \mbox{\Hypertarget{structnode_a2d890bb9f6af0ffd73fe79b21124c2a2}\label{structnode_a2d890bb9f6af0ffd73fe79b21124c2a2}} \index{node@{node}!data@{data}} \index{data@{data}!node@{node}} \subsubsection{\texorpdfstring{data}{data}} {\footnotesize\ttfamily int node\+::data} \mbox{\Hypertarget{structnode_a3ce38490a651bfda86d88ff955e96abc}\label{structnode_a3ce38490a651bfda86d88ff955e96abc}} \index{node@{node}!left@{left}} \index{left@{left}!node@{node}} \subsubsection{\texorpdfstring{left}{left}} {\footnotesize\ttfamily struct \hyperlink{structnode}{node}$\ast$ node\+::left} \mbox{\Hypertarget{structnode_a3871d43e823ba9542b052912d01709dd}\label{structnode_a3871d43e823ba9542b052912d01709dd}} \index{node@{node}!level@{level}} \index{level@{level}!node@{node}} \subsubsection{\texorpdfstring{level}{level}} {\footnotesize\ttfamily int node\+::level} \mbox{\Hypertarget{structnode_aeef7855fea382bfb671d7834aefa4b22}\label{structnode_aeef7855fea382bfb671d7834aefa4b22}} \index{node@{node}!offset@{offset}} \index{offset@{offset}!node@{node}} \subsubsection{\texorpdfstring{offset}{offset}} {\footnotesize\ttfamily int node\+::offset} \mbox{\Hypertarget{structnode_a05e4fe9e0177ba2d8dbd2c487cfddd53}\label{structnode_a05e4fe9e0177ba2d8dbd2c487cfddd53}} \index{node@{node}!parent@{parent}} \index{parent@{parent}!node@{node}} \subsubsection{\texorpdfstring{parent}{parent}} {\footnotesize\ttfamily struct \hyperlink{structnode}{node}$\ast$ node\+::parent} \mbox{\Hypertarget{structnode_a875f75abfe22103500535b179828e4e3}\label{structnode_a875f75abfe22103500535b179828e4e3}} \index{node@{node}!right@{right}} \index{right@{right}!node@{node}} \subsubsection{\texorpdfstring{right}{right}} {\footnotesize\ttfamily struct \hyperlink{structnode}{node}$\ast$ node\+::right} \mbox{\Hypertarget{structnode_afd9ff5fa3c3ab99d07cac2a7ad9d14a6}\label{structnode_afd9ff5fa3c3ab99d07cac2a7ad9d14a6}} \index{node@{node}!thread@{thread}} \index{thread@{thread}!node@{node}} \subsubsection{\texorpdfstring{thread}{thread}} {\footnotesize\ttfamily bool node\+::thread} \mbox{\Hypertarget{structnode_a64dd8b65a7d38c632a017d7f36444dbb}\label{structnode_a64dd8b65a7d38c632a017d7f36444dbb}} \index{node@{node}!x@{x}} \index{x@{x}!node@{node}} \subsubsection{\texorpdfstring{x}{x}} {\footnotesize\ttfamily int node\+::x} \mbox{\Hypertarget{structnode_ae944a3a75efb9856fa5c6f2221e2b49e}\label{structnode_ae944a3a75efb9856fa5c6f2221e2b49e}} \index{node@{node}!y@{y}} \index{y@{y}!node@{node}} \subsubsection{\texorpdfstring{y}{y}} {\footnotesize\ttfamily int node\+::y} The documentation for this struct was generated from the following file\+:\begin{DoxyCompactItemize} \item Computer\+Graphics\+Project/\+Computer\+Graphics\+Project/\hyperlink{tree_8h}{tree.\+h}\end{DoxyCompactItemize} \documentclass[12pt]{standalone} \usepackage{amsmath,amssymb,tikz-cd,tikz,braids} \begin{document} $$ \begin{tikzcd} P^i \ar[r] \ar[d] & 0 \ar[r] \ar[d] & \cdots \\ P^{i+1} \ar[r] & P^{i+2} \ar[r] & \cdots \end{tikzcd} $$ \end{document} slliac/Ljus4Food \hypertarget{class_template_methods_test}{}\section{Template\+Methods\+Test Class Reference} \label{class_template_methods_test}\index{Template\+Methods\+Test@{Template\+Methods\+Test}} Inheritance diagram for Template\+Methods\+Test\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=3.303835cm]{class_template_methods_test} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_template_methods_test_afbf3ff88b322c6a7197ce02297cd23a0}{test\+One}} () \item \mbox{\hyperlink{class_template_methods_test_a4fb9974ce113d5d1db8075e0db0dc9b6}{test\+Two}} () \end{DoxyCompactItemize} \subsection*{Static Public Member Functions} \begin{DoxyCompactItemize} \item static \mbox{\hyperlink{class_template_methods_test_a80ef9eb20e7443b38276fb4647985fb7}{set\+Up\+Before\+Class}} () \item static \mbox{\hyperlink{class_template_methods_test_a6256aa1772f8d1a6b16f1df662d94433}{tear\+Down\+After\+Class}} () \end{DoxyCompactItemize} \subsection*{Protected Member Functions} \begin{DoxyCompactItemize} \item \mbox{\hyperlink{class_template_methods_test_a0bc688732d2b3b162ffebaf7812e78da}{set\+Up}} () \item \mbox{\hyperlink{class_template_methods_test_ad6402f56c691a56954229b7e6a2294aa}{assert\+Pre\+Conditions}} () \item \mbox{\hyperlink{class_template_methods_test_abf54422376f10d9cbe49a06676e39e86}{assert\+Post\+Conditions}} () \item \mbox{\hyperlink{class_template_methods_test_a80fe3d17e658907fc75346a0ec9d6fc7}{tear\+Down}} () \item \mbox{\hyperlink{class_template_methods_test_ad52d89ed0d081f2259d9ca6e398e2d3d}{on\+Not\+Successful\+Test}} (Exception \$e) \end{DoxyCompactItemize} \subsection*{Additional Inherited Members} \subsection{Member Function Documentation} \mbox{\Hypertarget{class_template_methods_test_abf54422376f10d9cbe49a06676e39e86}\label{class_template_methods_test_abf54422376f10d9cbe49a06676e39e86}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!assert\+Post\+Conditions@{assert\+Post\+Conditions}} \index{assert\+Post\+Conditions@{assert\+Post\+Conditions}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{assert\+Post\+Conditions()}{assertPostConditions()}} {\footnotesize\ttfamily assert\+Post\+Conditions (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \mbox{\Hypertarget{class_template_methods_test_ad6402f56c691a56954229b7e6a2294aa}\label{class_template_methods_test_ad6402f56c691a56954229b7e6a2294aa}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!assert\+Pre\+Conditions@{assert\+Pre\+Conditions}} \index{assert\+Pre\+Conditions@{assert\+Pre\+Conditions}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{assert\+Pre\+Conditions()}{assertPreConditions()}} {\footnotesize\ttfamily assert\+Pre\+Conditions (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \mbox{\Hypertarget{class_template_methods_test_ad52d89ed0d081f2259d9ca6e398e2d3d}\label{class_template_methods_test_ad52d89ed0d081f2259d9ca6e398e2d3d}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!on\+Not\+Successful\+Test@{on\+Not\+Successful\+Test}} \index{on\+Not\+Successful\+Test@{on\+Not\+Successful\+Test}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{on\+Not\+Successful\+Test()}{onNotSuccessfulTest()}} {\footnotesize\ttfamily on\+Not\+Successful\+Test (\begin{DoxyParamCaption}\item[{Exception}]{\$e }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \mbox{\Hypertarget{class_template_methods_test_a0bc688732d2b3b162ffebaf7812e78da}\label{class_template_methods_test_a0bc688732d2b3b162ffebaf7812e78da}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!set\+Up@{set\+Up}} \index{set\+Up@{set\+Up}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{set\+Up()}{setUp()}} {\footnotesize\ttfamily set\+Up (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \mbox{\Hypertarget{class_template_methods_test_a80ef9eb20e7443b38276fb4647985fb7}\label{class_template_methods_test_a80ef9eb20e7443b38276fb4647985fb7}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!set\+Up\+Before\+Class@{set\+Up\+Before\+Class}} \index{set\+Up\+Before\+Class@{set\+Up\+Before\+Class}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{set\+Up\+Before\+Class()}{setUpBeforeClass()}} {\footnotesize\ttfamily static set\+Up\+Before\+Class (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{class_template_methods_test_a80fe3d17e658907fc75346a0ec9d6fc7}\label{class_template_methods_test_a80fe3d17e658907fc75346a0ec9d6fc7}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!tear\+Down@{tear\+Down}} \index{tear\+Down@{tear\+Down}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{tear\+Down()}{tearDown()}} {\footnotesize\ttfamily tear\+Down (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [protected]}} \mbox{\Hypertarget{class_template_methods_test_a6256aa1772f8d1a6b16f1df662d94433}\label{class_template_methods_test_a6256aa1772f8d1a6b16f1df662d94433}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!tear\+Down\+After\+Class@{tear\+Down\+After\+Class}} \index{tear\+Down\+After\+Class@{tear\+Down\+After\+Class}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{tear\+Down\+After\+Class()}{tearDownAfterClass()}} {\footnotesize\ttfamily static tear\+Down\+After\+Class (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})\hspace{0.3cm}{\ttfamily [static]}} \mbox{\Hypertarget{class_template_methods_test_afbf3ff88b322c6a7197ce02297cd23a0}\label{class_template_methods_test_afbf3ff88b322c6a7197ce02297cd23a0}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!test\+One@{test\+One}} \index{test\+One@{test\+One}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{test\+One()}{testOne()}} {\footnotesize\ttfamily test\+One (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} \mbox{\Hypertarget{class_template_methods_test_a4fb9974ce113d5d1db8075e0db0dc9b6}\label{class_template_methods_test_a4fb9974ce113d5d1db8075e0db0dc9b6}} \index{Template\+Methods\+Test@{Template\+Methods\+Test}!test\+Two@{test\+Two}} \index{test\+Two@{test\+Two}!Template\+Methods\+Test@{Template\+Methods\+Test}} \subsubsection{\texorpdfstring{test\+Two()}{testTwo()}} {\footnotesize\ttfamily test\+Two (\begin{DoxyParamCaption}{ }\end{DoxyParamCaption})} The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item /\+Applications/\+X\+A\+M\+P\+P/xamppfiles/htdocs/\+Code\+Igniter/vendor/phpunit/phpunit/tests/\+\_\+files/\mbox{\hyperlink{_template_methods_test_8php}{Template\+Methods\+Test.\+php}}\end{DoxyCompactItemize} common/beamerthemeDLRG.sty0 % Copyright 2014 by % % This file may be distributed and/or modified % % 1. under the LaTeX Project Public License and/or % 2. under the GNU Public License. % \DeclareOptionBeamer{width} % {\PassOptionsToPackage{width=#1}{beamerouterthemesidebar}} % % \DeclareOptionBeamer{hideothersubsections} % {\PassOptionsToPackage{hideothersubsections=#1}{beamerouterthemesidebar}} % % \DeclareOptionBeamer{hideallsubsections} % {\PassOptionsToPackage{hideallsubsections=#1}{beamerouterthemesidebar}} % % \ProcessOptionsBeamer % \RequirePackage{tikz} \usetikzlibrary{calc} \RequirePackage[scaled]{helvet} \RequirePackage[T1]{fontenc} \RequirePackage{ifthen} \mode% \useoutertheme{DLRG} \usecolortheme{DLRG} \beamertemplatenavigationsymbolsempty \mode @inproceedings{10.1007/978-3-319-11918-2_1, author = { }, title = {Strategic Pattern Search in Factor-Compressed Text}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_1}, doi = {10.1007/978-3-319-11918-2_1}, abstract = {We consider the problem of pattern-search in compressed text in a context in which: (a) the text is stored as a sequence of factors against a static phrase-book; (b) decoding of factors is from right-to-left; and (c) extraction of each symbol in each factor requires Θ(log σ ) time, where σ is the size of the original alphabet. To determine possible alignments given information about decoded characters we introduce two Boyer-Moore-like searching mechanisms, including one that makes use of a suffix array constructed over the pattern. The new mechanisms decode fewer than half the symbols that are required by a sequential left-to-right search such as the Knuth-Morris-Pratt approach, a saving that translates directly into improved execution time. Experiments with a two-level suffix array index structure for 4\"{a}GB of English text demonstrate the usefulness of the new techniques.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {1–12}, numpages = {12}, keywords = {Burrows-Wheeler transform, pattern matching, experimental evaluation, succinct data structure, suffix array, disk-based algorithm, string search}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_2, author = {.}, title = {Relative Lempel-Ziv with Constant-Time Random Access}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_2}, doi = {10.1007/978-3-319-11918-2_2}, abstract = {Relative Lempel-Ziv (RLZ) is a variant of LZ77 that can compress well collections of similar genomes while still allowing fast random access to them. In theory, at the cost of using sublinear extra space, accessing an arbitrary character takes constant time. We show that even in practice this works quite well: e.g., we can compress 36 S. cerevisiae genomes from a total of 464 MB to 11 MB and still support random access to them in under 50 nanoseconds per character, even when the accessed substrings are short. Our theoretical contribution is an optimized representation of RLZ's pointers.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {13–17}, numpages = {5}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_3, author = { and }, title = {Efficient Compressed Indexing for Approximate Top-k String Retrieval}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_3}, doi = {10.1007/978-3-319-11918-2_3}, abstract = {Given a collection of strings (called documents), the top-k document retrieval problem is that of, given a string pattern p , finding the k documents where p appears most often. This is a basic task in most information retrieval scenarios. The best current implementations require 20—30 bits per character (bpc) and k to 4 k microseconds per query, or 12—24 bpc and 1—10 milliseconds per query. We introduce a Lempel-Ziv compressed data structure that occupies 5—10 bpc to answer queries in around k microseconds. The drawback is that the answer is approximate, but we show that its quality improves asymptotically with the size of the collection, reaching over 85% of the accumulated term frequency of the real answer already for patterns of length 4—6 on rather small collections, and improving for larger ones.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {18–30}, numpages = {13}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_4, author = { and Ord\'{o}\~{n}ez, Alberto}, title = {Grammar Compressed Sequences with Rank/Select Support}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_4}, doi = {10.1007/978-3-319-11918-2_4}, abstract = {Sequence representations supporting not only direct access to their symbols, but also rank/select operations, are a fundamental building block in many compressed data structures. In several recent applications, the need to represent highly repetitive sequences arises, where statistical compression is ineffective. We introduce grammar-based representations for repetitive sequences, which use up to 10% of the space needed by representations based on statistical compression, and support direct access and rank/select operations within tens of microseconds.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {31–44}, numpages = {14}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_5, author = { , . and and }, title = {Algorithms for Jumbled Indexing, Jumbled Border and Jumbled Square on Run-Length Encoded Strings}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_5}, doi = {10.1007/978-3-319-11918-2_5}, abstract = { Jumbled Indexing , the problem of indexing a text for histogram queries, has been of much interest lately. In this paper we consider jumbled indexing for run-length encoded texts. We refute a former conjecture and show an algorithm for general sized alphabets. We also consider Jumbled Borders , the extension of borders to jumbled strings. Borders are the basis for various algorithms. Finally, we consider Jumbled Squares , strings which are of the form $xbar{x}$ , where $bar{x}$ is a jumbling of x . We show efficient algorithms for these problems.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {45–51}, numpages = {7}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_6, author = { , Jouni}, title = {Relative FM-Indexes}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_6}, doi = {10.1007/978-3-319-11918-2_6}, abstract = {Intuitively, if two strings S 1 and S 2 are sufficiently similar and we already have an FM-index for S 1 then, by storing a little extra information, we should be able to reuse parts of that index in an FM-index for S 2. We formalize this intuition and show that it can lead to significant space savings in practice, as well as to some interesting theoretical problems.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {52–64}, numpages = {13}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_7, author = { and Konow, Roberto and }, title = {Efficient Indexing and Representation of Web Access Logs}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_7}, doi = {10.1007/978-3-319-11918-2_7}, abstract = {We present a space-efficient data structure, based on the Burrows-Wheeler Transform, especially designed to handle web sequence logs, which are needed by web usage mining processes. Our index is able to process a set of operations efficiently, while at the same time maintains the original information in compressed form. Results show that web access logs can be represented using 0.85 to 1.03 times their original (plain) size, while executing most of the operations within a few tens of microseconds.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {65–76}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_8, author = {. and Fari\~{n}a, }, title = {A Compressed Suffix-Array Strategy for Temporal-Graph Indexing}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_8}, doi = {10.1007/978-3-319-11918-2_8}, abstract = {Temporal graphs represent vertexes and binary relations that change over time. In this paper we consider a temporal graph as a set of 4-tuples ( v s , v e , t s , t e ) indicating that an edge from a vertex v s to a vertex v e is active during the time interval [ t s , t e ). Representing those tuples involves the challenge of not only saving space but also of efficient query processing. Queries of interest for these graphs are both direct and reverse neighbors constrained by a time instant or a time interval. We show how to adapt a Compressed Suffix Array ( CSA ) to represent temporal graphs. The proposed structure, called Temporal Graph CSA ( TGCSA ), was experimentally compared with a compact data structure based on compressed inverted lists, which can be considered as a fair baseline in the state of the art. Our experimental results are promising. TGCSA obtains a good space-time trade-off, owns wider expressive capabilities than other alternatives, obtains reasonable space usage, and it is efficient even when performing the most complex temporal queries.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {77–88}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_9, author = { .}, title = {Succinct Indexes for Reporting Discriminating and Generic Words}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_9}, doi = {10.1007/978-3-319-11918-2_9}, abstract = {We consider the problem of indexing a collection $cal{D}$ of D strings (documents) of total n characters from an alphabet set of size σ , such that whenever a pattern P (of p characters) and an integer \"{\i} ∈ [1, D ] comes as a query, we can efficiently report all (i) maximal generic words and (ii) minimal discriminating words as defined below: These problems were introduced by Kucherov et al.\"{a}[8], and they proposed linear space indexes occupying O ( n log n ) bits with query times O ( p + output ) and O ( p + loglog n + output ) for Problem (i) and Problem (ii) respectively. In this paper, we describe succinct indexes of n log σ + o ( n log σ ) + O ( n ) bits space with near optimal query times i.e., O ( p + loglog n + output ) for both these problems.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {89–100}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_10, author = {, .}, title = {Fast Construction of Wavelet Trees}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_10}, doi = {10.1007/978-3-319-11918-2_10}, abstract = {In this paper we describe a fast algorithm that creates a wavelet tree for a sequence of symbols. We show that a wavelet tree can be constructed in $O(nlceil{frac{log sigma}{sqrt{log n}}}rceil)$ time where n is the number of symbols and σ is the alphabet size.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {101–110}, numpages = {10}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_11, author = { and . and and }, title = {Order Preserving Prefix Tables}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_11}, doi = {10.1007/978-3-319-11918-2_11}, abstract = {In the Order Preserving Pattern Matching (OPPM) problem, we have a text T and a pattern P on an integer alphabet as input. And the goal is to locate a fragment which is order-isomorphic with the pattern. Two sequences over integer alphabet are order-isomorphic if the relative order between any two elements at the same positions in both sequences is the same. In this paper we present an efficient algorithm to construct an interesting and useful data structure, namely, prefix table, from the order preserving point of view.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {111–116}, numpages = {6}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_12, author = { and }, title = {Alphabet-Independent Algorithms for Finding Context-Sensitive Repeats in Linear Time}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_12}, doi = {10.1007/978-3-319-11918-2_12}, abstract = {The identification of repetitive sequences (repeats) is an essential component of genome sequence analysis, and there are dozens of algorithms that search for exact or approximate repeats. The notions of maximal and supermaximal (exact) repeats have received special attention, and it is possible to simultaneously compute them on index data structures like the suffix tree or the enhanced suffix array. Very recently, this research has been extended in two directions. Gall\'{e} and Tealdi devised an alphabet-independent linear-time algorithm that finds all context-diverse repeats (which subsume maximal and supermaximal repeats as special cases), while Taillefer and Miller gave a quadratic-time algorithm that simultaneously computes and classifies maximal, near-supermaximal, and supermaximal repeats. In this paper, we provide new alphabet-independent linear-time algorithms for both tasks.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {117–128}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_13, author = { }, title = {A 3-Approximation Algorithm for the Multiple Spliced Alignment Problem and Its Application to the Gene Prediction Task}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_13}, doi = {10.1007/978-3-319-11918-2_13}, abstract = {The Spliced Alignment Problem is a well-known problem in Bioinformatics with application to the gene prediction task. This problem consists in finding an ordered subset of non-overlapping substrings of a subject sequence g that best fits a target sequence t . In this work we present an approximation algorithm for a variant of the Spliced Alignment Problem, called Multiple Spliced Alignment Problem , that involves more than one target sequence. Under a metric, this algorithm is proved to be a 3-approximation for the problem and its good practical results compare to those obtained by four heuristics already developed for the Multiple Spliced Alignment Problem.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {129–138}, numpages = {10}, keywords = {multiple spliced alignment problem, Approximation algorithm, gene prediction}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_14, author = { and }, title = {Improved Filters for the Approximate Suffix-Prefix Overlap Problem}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_14}, doi = {10.1007/978-3-319-11918-2_14}, abstract = {Computing suffix-prefix overlaps for a large collection of strings is a fundamental building block for the analysis of genomic next-generation sequencing data. The approximate suffix-prefix overlap problem is to find all pairs of strings from a given set such that a prefix of one string is similar to a suffix of the other. V\"{a}lim\"{a}ki et al. (Information and Computation, 2012) gave a solution to this problem based on suffix filters. In this work, we propose two improvements to the method of V\"{a}lim\"{a}ki et al. that reduce the running time of the computation.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {139–148}, numpages = {10}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_15, author = { Ciardo, }, title = {Sequence Decision Diagrams}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_15}, doi = {10.1007/978-3-319-11918-2_15}, abstract = {Compact encoding of finite sets of strings is a classic problem. The manipulation of large sets requires compact data structures that allow for efficient set operations. We define sequence decision diagrams (SeqDDs), which can encode arbitrary finite sets of strings over an alphabet. SeqDDs can be seen as a variant of classic decision diagrams such as BDDs and MDDs where, instead of a fixed number of levels, we simply require that the number of paths and the lengths of these paths be finite. However, the main difference between the two is the target application: while MDDs are suited to store and manipulate large sets of constant-length tuples, SeqDDs can store arbitrary finite languages and, as such, should be studied in relation to finite automata. We do so, examining in particular the size of equivalent representations.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {149–160}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_16, author = { and and }, title = {Shortest Unique Queries on Strings}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_16}, doi = {10.1007/978-3-319-11918-2_16}, abstract = {Let D be a long input string of n characters (from an alphabet of size up to 2 w , where w is the number of bits in a machine word). Given a substring q of D , a shortest unique query returns a shortest unique substring of D that contains q . We present an optimal structure that consumes O ( n ) space, can be built in O ( n ) time, and answers a query in O (1) time. We also extend our techniques to solve several variants of the problem optimally.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {161–172}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_17, author = { and }, title = {Online Multiple Palindrome Pattern Matching}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_17}, doi = {10.1007/978-3-319-11918-2_17}, abstract = {A palindrome is a string that reads the same forward and backward. We say that two strings of the same length are pal-equivalent if for each possible center they have the same length of the maximal palindrome. Given a text\"{a} T of length\"{a} n and a set of patterns\"{a} P 1,', P k , we study the online multiple palindrome pattern matching problem that finds all pairs of an index\"{a} i and a pattern\"{a} P j such that T [ i '—'| P j | + 1: i ] and P j are pal-equivalent. We solve the problem in O ( m k M )\"{a}preprocessing time and O ( m k n )\"{a}query time using O ( m k M )\"{a}space, where M is the sum of all pattern lengths and m k is the longest pattern length.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {173–178}, numpages = {6}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_18, author = { and }, title = {Indexed Matching Statistics and Shortest Unique Substrings}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_18}, doi = {10.1007/978-3-319-11918-2_18}, abstract = {The unidirectional and bidirectional matching statistics between two strings s and t on alphabet Σ, and the shortest unique substrings of a single string t , are the cornerstone of a number of large-scale genome analysis applications, and they encode nontrivial structural properties of s and t . In this paper we compute for the first time the matching statistics between s and t in O ((| s | + | t |)log|Σ|) time and in O (| s |log|Σ|) bits of space, circumventing the need for computing the depths of suffix tree nodes that characterized previous approaches. Symmetrically, we compute for the first time the shortest unique substrings of a string t in O (| t |log|Σ|) time and in O (| t |log|Σ|) bits of space. A key component of our methods is an encoding of both the unidirectional and the bidirectional statistics that takes 2| t | + o (| t |) bits of space and that allows constant-time access to every position.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {179–190}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_19, author = { and and }, title = {I/O-Efficient Dictionary Search with One Edit Error}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_19}, doi = {10.1007/978-3-319-11918-2_19}, abstract = {This paper studies the 1-error dictionary search problem in external memory. The input is a set D of strings whose characters are drawn from a constant-size alphabet. Given a string q , a query reports the ids of all strings in D that are within 1 edit distance from q . We give a structure occupying O ( n / B ) blocks that answers a query in $O(1 + frac{m}{wB} + frac{k}{B})$ I/Os, where n is the total length of all strings in D , m is the length of q , k is the number of ids reported, w is the size of a machine word, and B is the number of words in a block.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {191–202}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_20, author = { }, title = {Online Pattern Matching for String Edit Distance with Moves}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_20}, doi = {10.1007/978-3-319-11918-2_20}, abstract = {Edit distance with moves (EDM) is a string-to-string distance measure that includes substring moves in addition to ordinal editing operations to turn one string to the other. Although optimizing EDM is intractable, it has many applications especially in error detections. Edit sensitive parsing (ESP) is an efficient parsing algorithm that guarantees an upper bound of parsing discrepancies between different appearances of the same substrings in a string. ESP can be used for computing an approximate EDM as the L 1 distance between characteristic vectors built by node labels in parsing trees. However, ESP is not applicable to a streaming text data where a whole text is unknown in advance. We present an online ESP (OESP) that enables an online pattern matching for EDM. OESP builds a parse tree for a streaming text and computes the L 1 distance between characteristic vectors in an online manner. For the space-efficient computation of EDM, OESP directly encodes the parse tree into a succinct representation by leveraging the idea behind recent results of a dynamic succinct tree. We experimentally test OESP on the ability to compute EDM in an online manner on benchmark datasets, and we show OESP's efficiency.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {203–214}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_21, author = {. and Konow, Roberto and }, title = {K2-Treaps: Range Top-k Queries in Compact Space}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_21}, doi = {10.1007/978-3-319-11918-2_21}, abstract = {Efficient processing of top- k queries on multidimensional grids is a common requirement in information retrieval and data mining, for example in OLAP cubes. We introduce a data structure, the K 2-treap, that represents grids in compact form and supports efficient prioritized range queries. We compare the K 2-treap with state-of-the-art solutions on synthetic and real-world datasets, showing that it uses 30% of the space of competing solutions while solving queries up to 10 times faster.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {215–226}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_22, author = { }, title = {Performance Improvements for Search Systems Using an Integrated Cache of Lists+Intersections}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_22}, doi = {10.1007/978-3-319-11918-2_22}, abstract = {Modern information retrieval systems use several levels of caching to speedup computation by exploiting frequent, recent or costly data used in the past. In this study we propose and evaluate a static cache that works simultaneously as list and intersection cache, offering a more efficient way of handling cache space. In addition, we propose effective strategies to select the term pairs that should populate the cache. Simulation using two datasets and a real query log reveal that the proposed approach improves overall performance in terms of total processing time, achieving savings of up to 40% in the best case.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {227–235}, numpages = {9}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_23, author = { }, title = {Information-Theoretic Term Selection for New Item Recommendation}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_23}, doi = {10.1007/978-3-319-11918-2_23}, abstract = {Recommender systems aim at predicting the preference of a user towards a given item (e.g., a movie, a song). For systems that must cope with continuously evolving item catalogs, there will be a considerable rate of new items for which no past preference is known that could otherwise inform preference-based recommendations. In contrast, pure content-based recommendations may suffer from noisy item descriptions. To overcome these problems, we propose an information-theoretic approach that exploits a taxonomy of categories associated with the cataloged items in order to select informative terms for an improved recommendation. Our experiments using two publicly available datasets attest the effectiveness of the proposed approach, which significantly outperforms state-of-the-art content-based recommenders from the literature.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {236–243}, numpages = {8}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_24, author = {, . and Radoszewski, and Wale\'{n}, Tomasz}, title = {On the String Consensus Problem and the Manhattan Sequence Consensus Problem}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_24}, doi = {10.1007/978-3-319-11918-2_24}, abstract = {In the Manhattan Sequence Consensus problem (MSC problem) we are given k integer sequences, each of length ℓ, and we are to find an integer sequence x of length ℓ (called a consensus sequence), such that the maximum Manhattan distance of x from each of the input sequences is minimized. For binary sequences Manhattan distance coincides with Hamming distance, hence in this case the string consensus problem (also called string center problem or closest string problem) is a special case of MSC. Our main result is a practically efficient $mathcal{O}(ell)$ -time algorithm solving MSC for k ≤ 5 sequences. Practicality of our algorithms has been verified experimentally. It improves upon the quadratic algorithm by Amir et al. (SPIRE 2012) for string consensus problem for k = 5 binary strings. Similarly as in Amir's algorithm we use a column-based framework. We replace the implied general integer linear programming by its easy special cases, due to combinatorial properties of the MSC for k ≤ 5. We also show that for a general parameter k any instance can be reduced in linear time to a kernel of size k !, so the problem is fixed-parameter tractable. Nevertheless, for k ≤ 4 this is still too much for any naive solution to be feasible in practice.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {244–255}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_25, author = {. and }, title = {Context-Aware Deal Size Prediction}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_25}, doi = {10.1007/978-3-319-11918-2_25}, abstract = {Daily deals sites, such as Groupon and LivingSocial, attract millions of customers in the hunt for products and services at substantially reduced prices (i.e., deals). An important aspect for the profitability of these sites is the correct prediction of how many coupons will be sold for each deal in their catalog–a task commonly referred to as deal size prediction. Existing solutions for the deal size prediction problem focus on one deal at a time, neglecting the existence of similar deals in the catalog. In this paper, we propose to improve deal size prediction by taking into account the context in which a given deal is offered. In particular, we propose a topic modeling approach to identify markets with similar deals and an expectation-maximization approach to model intra-market competition while minimizing the prediction error. A systematic set of experiments shows that our approach offers gains in precision ranging from 8.18% to 17.67% when compared against existing solutions.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {256–267}, numpages = {12}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } @inproceedings{10.1007/978-3-319-11918-2_26, author = {, }, title = {Simple and Efficient String Algorithms for Query Suggestion Metrics Computation}, year = {2014}, isbn = {9783319119175}, publisher = {Springer-Verlag}, address = {Berlin, Heidelberg}, url = {https://doi.org/10.1007/978-3-319-11918-2_26}, doi = {10.1007/978-3-319-11918-2_26}, abstract = {In order to make query suggestion mechanisms more efficient, it is important to have metrics that will estimate query suggestions quality well. Recently, Kharitonov et al.\"{a}[7] proposed a family of metrics that showed much better alignment with user satisfaction than previously known metrics. However, they did not address the problem of computing the proposed metrics. In this paper we show that the problem can be reduced to one of the two string problems which we call Top- k and Sorted-Top- k . Given an integer k and two sets of pairwise distinct strings (queries) with weights, Q and Q test , the Top- k problem is to find, for each query q ∈ Q test , its shortest prefix\"{a} q [1.. i ] such that q belongs to the list of k heaviest queries in Q starting with\"{a} q [1.. i ]. The Sorted-Top- k problem is to retrieve, for each q ∈ Q test and 1 ≤ i ≤ | q |, a position of\"{a} q in the sorted list of the k heaviest queries in Q starting with q [1.. i ]. We show several linear-time solutions to these problems and compare them experimentally.}, booktitle = {Proceedings of the 21st International Symposium on String Processing and Information Retrieval - Volume 8799}, pages = {268–278}, numpages = {11}, location = {Ouro Preto, Brazil}, series = {SPIRE 2014} } %% LaTeX2e class for student theses %% sections/abstract_en.tex %% %% Karlsruhe Institute of Technology %% Institute for Program Structures and Data Organization %% Chair for Software Design and Quality (SDQ) %% %% Dr.-Ing. %% %% %% Version 1.3.5, 2020-06-26 \Abstract English abstract.baseText/exercises/LinesAndPlanes-Lines.tex1-10 \section*{Exercises} \begin{ex} Find the vector equation for the line through $(-7,6,0)$ and $(-1,1,4)$. Then, find the parametric equations for this line. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Find parametric equations for the line through the point $(7,7,1)$ with direction vector $\vect{d} = \begin{mysmallmatrix}{r} 1 \\ 6 \\ 2 \end{mysmallmatrix}$. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Parametric equations of the line are \begin{equation*} \begin{array}{c} x = t+2, \\ y = 6-3t, \\ z = -t-6. \end{array} \end{equation*} Find a direction vector for the line and a point on the line. % \begin{sol} % \end{sol} \end{ex} \begin{ex} The equation of a line in two dimensions is written as $y=x-5$. Find a vector equation for this line. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Find parametric equations for the line through $(6, 5, -2, 3)$ and $(5, 1, 2, 1)$. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Consider the following vector equation for a line in $\R^3$: \begin{equation*} \begin{mymatrix}{c} x\\y\\z \end{mymatrix} = \begin{mymatrix}{r} 1\\2\\0 \end{mymatrix} + t\,\begin{mymatrix}{r} 1\\0\\1 \end{mymatrix}. \end{equation*} Find a new vector equation for the same line by doing the change of parameter $t=2-s$. \begin{sol} We have \begin{equation*} \begin{mymatrix}{c} x\\y\\z \end{mymatrix} = \begin{mymatrix}{r} 1\\2\\0 \end{mymatrix} + (2-s)\,\begin{mymatrix}{c} 1\\0\\1 \end{mymatrix} = \begin{mymatrix}{r} 3\\2\\2 \end{mymatrix} + s\,\begin{mymatrix}{r} -1\\0\\-1 \end{mymatrix}. \end{equation*} \end{sol} \end{ex} \begin{ex} Consider the line given by the following parametric equations: \begin{equation*} \begin{array}{c} x = 2t+2, \\ y = 5-4t, \\ z= -t-3. \end{array} \end{equation*} Find symmetric equations for the line. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Find the point on the line segment from $P = (-4, 7, 5)$ to $Q = (2, -2, -3)$ which is $\frac{1}{7}$ of the way from $P$ to $Q$. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Suppose a triangle in $\R^n$ has vertices at $P$, $Q$, and $R$. Consider the lines which are drawn from a vertex to the mid point of the opposite side. Show these three lines intersect in a point and find the coordinates of this point. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Determine whether the lines \begin{equation*} \begin{mymatrix}{c} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{c} 1 \\ 1 \\ 2 \end{mymatrix} + t \begin{mymatrix}{c} 1 \\ 2 \\ 2 \end{mymatrix} \quad\mbox{and}\quad \begin{mymatrix}{c} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{c} 1 \\ -1 \\ -4 \end{mymatrix} + s \begin{mymatrix}{c} 1 \\ 1 \\ -1 \end{mymatrix} \end{equation*} intersect. If yes, find the point of intersection. \end{ex} \begin{ex} Determine whether the lines \begin{equation*} \begin{mymatrix}{c} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{c} 2 \\ -1 \\ 0 \end{mymatrix} + t \begin{mymatrix}{c} 1 \\ 3 \\ 2 \end{mymatrix} \quad\mbox{and}\quad \begin{mymatrix}{c} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{c} 1 \\ 1 \\ 3 \end{mymatrix} + s \begin{mymatrix}{c} 1 \\ 2 \\ 0 \end{mymatrix} \end{equation*} intersect. If yes, find the point of intersection. \end{ex} \begin{ex} Find the angle between the two lines \begin{equation*} \begin{mymatrix}{r} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{r} 3 \\ 0 \\ 1 \end{mymatrix} + t \begin{mymatrix}{r} 3 \\ -3 \\ 0 \end{mymatrix} \quad\mbox{and}\quad \begin{mymatrix}{r} x \\ y \\ z \end{mymatrix} = \begin{mymatrix}{r} 3 \\ 0 \\ 1 \end{mymatrix} + s \begin{mymatrix}{r} -1 \\ 2 \\ 2 \end{mymatrix}. \end{equation*} % \begin{sol} % \end{sol} \end{ex} \begin{ex} Let $P = (1,2,3)$ be a point in $\R^3$. Let $L$ be the line through the point $P_0 = (1, 4, 5)$ with direction vector $\vect{d} = \begin{mysmallmatrix}{r} 1 \\ -1 \\ 1 \end{mysmallmatrix}$. Find the shortest distance from $P$ to $L$, and find the point $Q$ on $L$ that is closest to $P$. % \begin{sol} % \end{sol} \end{ex} \begin{ex} Let $P = (0,2,1)$ be a point in $\R^3$. Let $L$ be the line through the points $P_0 = (1, 1, 1)$ and $P_1 = (4, 1, 2)$. Find the shortest distance from $P$ to $L$, and find the point $Q$ on $L$ that is closest to $P$. % \begin{sol} % \end{sol} \end{ex} \begin{ex}\label{ex:angle-lines} When we computed the angle between two lines in Example~\ref{exa:angle-between-two-lines}, we calculated two different angles and took the smaller of the two. Show that one can get the same answer by taking the absolute value of the dot product, i.e., by solving \begin{equation*} \cos\theta = \frac{\abs{\vect{u}\dotprod\vect{v}}}{\norm{\vect{u}}\norm{\vect{v}}}. \end{equation*} \begin{sol} From trigonometry, we have the following properties of the cosine function: \begin{itemize} \item $\cos\theta$ is positive for $0\leq\theta\leq\frac{\pi}{2}$, and negative for $\frac{\pi}{2}\leq\theta\leq\pi$. \item $\cos(\pi-\theta) = -\cos\theta$. \end{itemize} By the method in Example~\ref{exa:angle-between-two-lines}, we calculated $\theta$ such that \begin{equation*} \cos\theta = \frac{\vect{u}\dotprod\vect{v}}{\norm{\vect{u}}\norm{\vect{v}}}. \end{equation*} If $0\leq\theta\leq\frac{\pi}{2}$, the answer is $\theta$. If $\frac{\pi}{2}\leq\theta\leq\pi$, the answer is $\phi=\pi-\theta$. But in the last case, the dot product is negative and we have \begin{equation*} \cos\phi = \cos(\pi-\theta) = -\cos\theta = \frac{-\vect{u}\dotprod\vect{v}~~}{\norm{\vect{u}}\norm{\vect{v}}} = \frac{\abs{\vect{u}\dotprod\vect{v}}}{\norm{\vect{u}}\norm{\vect{v}}}. \end{equation*} So in either case we get the correct answer by taking the absolute value. \end{sol} \end{ex} \hypertarget{class_o_p_immediate}{}\section{O\+P\+Immediate Class Reference} \label{class_o_p_immediate}\index{O\+P\+Immediate@{O\+P\+Immediate}} class representing an Immediate herited by \hyperlink{class_operand}{Operand} {\ttfamily \#include $<$O\+P\+Immediate.\+h$>$} Inheritance diagram for O\+P\+Immediate\+:\begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=2.000000cm]{class_o_p_immediate} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hypertarget{class_o_p_immediate_ab047dac5f3390947a21e4ab118c05857}{}\hyperlink{class_o_p_immediate_ab047dac5f3390947a21e4ab118c05857}{O\+P\+Immediate} (string)\label{class_o_p_immediate_ab047dac5f3390947a21e4ab118c05857} \begin{DoxyCompactList}\small\item\em Constructor of the Immediate Class. \end{DoxyCompactList}\item \hypertarget{class_o_p_immediate_ae940dcf9e9050227a94c759a0cae6861}{}\hyperlink{class_o_p_immediate_ae940dcf9e9050227a94c759a0cae6861}{O\+P\+Immediate} (int)\label{class_o_p_immediate_ae940dcf9e9050227a94c759a0cae6861} \begin{DoxyCompactList}\small\item\em Constructor of the Immediate Class. \end{DoxyCompactList}\item \hypertarget{class_o_p_immediate_af7f51ae61e075e02817d6ecd7441408f}{}virtual \hyperlink{class_o_p_immediate_af7f51ae61e075e02817d6ecd7441408f}{$\sim$\+O\+P\+Immediate} ()\label{class_o_p_immediate_af7f51ae61e075e02817d6ecd7441408f} \begin{DoxyCompactList}\small\item\em Destructor of the Immediate Class. \end{DoxyCompactList}\item virtual string \hyperlink{class_o_p_immediate_ad714fb614c0d8f4afa1157a34b2936fd}{get\+\_\+op} () \begin{DoxyCompactList}\small\item\em Get the string of the operand. \end{DoxyCompactList}\item virtual t\+\_\+\+Op\+Type \hyperlink{class_o_p_immediate_aed01353798ae57936a9f77dd05eafa88}{get\+\_\+op\+\_\+type} () \begin{DoxyCompactList}\small\item\em get the operator type \end{DoxyCompactList}\item virtual string \hyperlink{class_o_p_immediate_a12bc613de3bff73ead8632dafd8050a0}{to\+\_\+string} () \begin{DoxyCompactList}\small\item\em tostring \end{DoxyCompactList}\item \hypertarget{class_o_p_immediate_ae5d6c30c6bff17de4e7fabb24cf6bf59}{}virtual void \hyperlink{class_o_p_immediate_ae5d6c30c6bff17de4e7fabb24cf6bf59}{set\+\_\+op} (string)\label{class_o_p_immediate_ae5d6c30c6bff17de4e7fabb24cf6bf59} \begin{DoxyCompactList}\small\item\em set the string of the operand setter of the operand \end{DoxyCompactList}\end{DoxyCompactItemize} \subsection*{Additional Inherited Members} \subsection{Detailed Description} class representing an Immediate herited by \hyperlink{class_operand}{Operand} \subsection{Member Function Documentation} \hypertarget{class_o_p_immediate_ad714fb614c0d8f4afa1157a34b2936fd}{}\index{O\+P\+Immediate@{O\+P\+Immediate}!get\+\_\+op@{get\+\_\+op}} \index{get\+\_\+op@{get\+\_\+op}!O\+P\+Immediate@{O\+P\+Immediate}} \subsubsection[{get\+\_\+op()}]{\setlength{\rightskip}{0pt plus 5cm}virtual string O\+P\+Immediate\+::get\+\_\+op ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [virtual]}}\label{class_o_p_immediate_ad714fb614c0d8f4afa1157a34b2936fd} Get the string of the operand. \begin{DoxyReturn}{Returns} return the string of the Immediate \end{DoxyReturn} Implements \hyperlink{class_operand_a2bf3ad8b34d39cb35ff743ffcc0f4675}{Operand}. \hypertarget{class_o_p_immediate_aed01353798ae57936a9f77dd05eafa88}{}\index{O\+P\+Immediate@{O\+P\+Immediate}!get\+\_\+op\+\_\+type@{get\+\_\+op\+\_\+type}} \index{get\+\_\+op\+\_\+type@{get\+\_\+op\+\_\+type}!O\+P\+Immediate@{O\+P\+Immediate}} \subsubsection[{get\+\_\+op\+\_\+type()}]{\setlength{\rightskip}{0pt plus 5cm}virtual t\+\_\+\+Op\+Type O\+P\+Immediate\+::get\+\_\+op\+\_\+type ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [virtual]}}\label{class_o_p_immediate_aed01353798ae57936a9f77dd05eafa88} get the operator type \begin{DoxyReturn}{Returns} return the \hyperlink{class_operand}{Operand} type as enum \end{DoxyReturn} Implements \hyperlink{class_operand_afd469e305a467e2574f34ac9bd6c62b0}{Operand}. \hypertarget{class_o_p_immediate_a12bc613de3bff73ead8632dafd8050a0}{}\index{O\+P\+Immediate@{O\+P\+Immediate}!to\+\_\+string@{to\+\_\+string}} \index{to\+\_\+string@{to\+\_\+string}!O\+P\+Immediate@{O\+P\+Immediate}} \subsubsection[{to\+\_\+string()}]{\setlength{\rightskip}{0pt plus 5cm}virtual string O\+P\+Immediate\+::to\+\_\+string ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [virtual]}}\label{class_o_p_immediate_a12bc613de3bff73ead8632dafd8050a0} tostring \begin{DoxyReturn}{Returns} return the name of the Object as string \end{DoxyReturn} Implements \hyperlink{class_operand_a28aed96d5fafee66be81c30c1435ad00}{Operand}. The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item \hyperlink{_o_p_immediate_8h}{O\+P\+Immediate.\+h}\end{DoxyCompactItemize} Rocla/OverClouds %------------------------------------------------------- % DOCUMENT CONFIGURATIONS %------------------------------------------------------- %------------------------------------------------------- % START OF DECENTRALIZED APPS STATE OF THE ART %------------------------------------------------------- \subsubsection{Decentralized applications} \paragraph{YaCy} Description from the official website \cite{YaCyWebPeople} \blockquote{YaCy is a free distributed search engine, built on principles of peer-to-peer networks. Its core is a computer program written in Java distributed on several hundred computers, so-called YaCy-peers.} %------------------------------------------------------- % END OF DECENTRALIZED APPS STATE OF THE ART %-------------------------------------------------------\documentclass{fenicscourse} \begin{document} \fenicslecture{Lecture 14: From sensitivities to optimisation} {} \input{slides/from_sensitivity_to_optimisation} \end{document} 0 \documentclass{article} \usepackage[utf8]{inputenc} \usepackage[osf,sc]{mathpazo} \usepackage[scaled=0.90]{helvet} \usepackage[table,dvipsnames]{xcolor} \usepackage{enumitem,mathtools} % \usepackage[ruled,vlined]{algorithm2e} \usepackage[ruled,vlined,resetcount]{algorithm2e} % \SetAlFnt{\small} \usepackage{algorithmic} \usepackage{float} \usepackage{soul} \usepackage{subcaption} \usepackage{graphicx} \usepackage{lineno} \usepackage{url} % \def\UrlBreaks{\do\/\do-} % \usepackage{breakurl} \usepackage[breaklinks]{hyperref} \usepackage{tcolorbox} \usepackage{array} \usepackage{tabularx} \hypersetup{ colorlinks=true, linkcolor=cyan, % color1 : will be black filecolor=red, urlcolor=ForestGreen, citecolor=red, bookmarksopen=false, pdftitle={Title}, pdfauthor={Author}, } \usepackage[capitalise]{cleveref} \definecolor{darkpastelpurple}{rgb}{0.59, 0.44, 0.84} \def\code#1{\textbf{\texttt{#1}}} \def\codeRed#1{{\color{Maroon}{\textbf{\texttt{#1}}}}} \def\codeCyan#1{{\color{cyan}{\textbf{\texttt{#1}}}}} \def\vari#1{{\color{Cerulean}{\textbf{\texttt{#1}}}}} \def\func#1{{\color{Purple}{\textbf{\texttt{#1}}}}} \newenvironment{coded}{\color{blue}\code} \definecolor{Gray}{gray}{0.95} \colorlet{shadecolor}{gray!40} \definecolor{LightCyan}{rgb}{0.88, 1, 1} %\definecolor{LightCyan}{rgb}{0.92, 0.98, 0.98} \definecolor{brickred}{rgb}{0.8, 0.25, 0.33} \definecolor{mgreen}{rgb}{0.2, 0.8, 1} \definecolor{bleudefrance}{rgb}{0.19, 0.55, 0.91} \definecolor{aliceblue}{rgb}{0.94, 0.97, 1.0} \definecolor{azureWeb}{rgb}{0.94, 1.0, 1.0} \definecolor{beaublue}{rgb}{0.74, 0.83, 0.9} \definecolor{gainsboro}{rgb}{0.86, 0.86, 0.86} \definecolor{linen}{rgb}{0.98, 0.94, 0.9} \definecolor{oldlace}{rgb}{0.99, 0.96, 0.9} \definecolor{magnolia}{rgb}{0.97, 0.96, 1.0} \definecolor{moccasin}{rgb}{0.98, 0.92, 0.84} \definecolor{navajowhite}{rgb}{1.0, 0.87, 0.68} \definecolor{palecornflowerblue}{rgb}{0.67, 0.8, 0.94} \usepackage{enumitem} \setlist[description]{% topsep=10pt, % space before start / after end of list % itemsep=5pt, % space between items % font={\bfseries\sffamily}, % set the label font % font={\bfseries\sffamily\color{red}}, % if colour is needed font={\color{red}}, } \newcommand{\EVI}{E\!V\!I} \newcommand{\NDVI}{N\!D\!V\!I} \newcommand{\NIR}{N\!I\!R} \title{Google Earth Engine} \author{} \date{\today} \begin{document} \maketitle \tableofcontents \clearpage \section{Preface} Google Earth Engine sucks! Below (Fig.~\ref{fig:GEESucks}) we have a simple example to show GEE is very specific. Accessing to elements/entries of its object is not intuitive. Figuring out every single step is a challenge. \begin{figure}[H] \centering \includegraphics[width=1\textwidth]{figures/GEE_Sucks} \caption{GEE sucks.} \label{fig:GEESucks} \end{figure} \noindent Here is the code used for generation of Fig.~\ref{fig:GEESucks}. \begin{tcolorbox} % \textbf{Algorithm for gradient descent with momentum is given below} \begin{algorithm}[H] \label{alg:GEE_Sucks} \caption{GEE Sucks.} \SetAlgoLined % \KwResult{Faster convergence} 1. \func{print}(\codeRed{``Print \#1: 3+10''},~\vari{3+10}) \; 2. \textbf{var} \code{x=}\vari{3} \; 3. \textbf{var} \code{y=}\vari{10} \; 4. \func{print}(\codeRed{``Print \#2: x+y''}, \code{x+y}) \; \vspace{.1in} 5. \textbf{var} \code{big\_delta\_x} = \vari{3} \; 6. \func{print}(\codeRed{``Print \#3: big\_delta\_x''}, \code{big\_delta\_x}) \; \vspace{.1in} 7. \textbf{var} \code{x\_big} = \func{ee.List.sequence}(\vari{-125.0},~\vari{-111.3}, \code{big\_delta\_x}) \; \vspace{.1in} 8. \func{print} (\codeRed{``Print \#4: x\_big''}, \code{x\_big}) \; 9. \func{print}(\codeRed{``Print \#5: x\_big.get(1)''}, \code{x\_big}.\func{get}(\vari{1})) \; 10. \func{print}(\codeRed{``Print \#6''}, \func{\textbf{ee.Number}}(\code{big\_delta\_x}).\func{add} (\func{\textbf{ee.Number}}(\code{x\_big}.\func{get}(\vari{1})))) \; \vspace{.1in} 11. \textbf{var} \code{aaa} = \code{x\_big}.\func{get}(\vari{1}) + \code{big\_delta\_x} \; 12. // \func{print}(\codeRed{``Print \#7: aaa''}, \code{aaa}) \; 13. \func{print}(\codeRed{``Print \#8: ee.Number(aaa)''}, \func{\textbf{ee.Number}}(\code{aaa})) \; \end{algorithm} \end{tcolorbox} %%%%%%%---------------------------------------- %%%%%%% %%%%%%% JS or Python Interface %%%%%%% \section{JavaScript or Python Interface} I think Python should be avoided in this particular case for the following reasons: \begin{enumerate} \item The interface is too slow, \item The interface needs authentication every single time, \item Google does not maintain the Python. Therefore, the functions are first written/updated for the JavaScript (JS) by Google, and the Python equivalents/updates will not be provided in a timely manner (who knows when?). \item The tutorials for JS is already hard to find, it is much worse for Python. Again, since Google is responsible for JavaScript, it releases the tutorials for it, but not Python. P.S. tutorials for JS might be abundant, but finding your exact needs might be hard. Even when you find something you may not be sure if that is the best possible solution. \end{enumerate} %%%%%%%-------------------------------------------------------------------------- %%%%%%% %%%%%%% Landsat Products and Differences %%%%%%% \section{Landsat Products and Differences} There are different products\footnote{start here to collect some information. some of the products are deprecated and superseded and Google does not show them easily: \href{https://developers.google.com/earth-engine/datasets/catalog/landsat}{here}} that fall under different labels; tier 1 vs tier 2, collection 1 vs collection 2, level 1 and level 2. Some of these have the same description on Google developer pages. For example, \href{https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LC08_C01_T1_SR#description}{USGS Landsat 8 Surface Reflectance Tier 1} and \href{https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LC08_C01_T2_SR#description}{USGS Landsat 8 Surface Reflectance Tier 2} have the same description and identical bands. In this particular example we want to use Tier 1. But we need a deeper understanding of differences(?) Based on the information below and references therein, Collection 2 is an improvement over Collection 1\footnote{Is there any time period for which Collection 2 does not exist but 1 does?}. It seems Collection-2 Level-2 Tier-1 should be the best, but in our plots it was not different from T1\_SR (\cref{fig:C2L2Performance}). Also keep in mind \hl{Collection-2 Level-2 bands must be scaled.} \begin{figure}[H] \includegraphics[width=1\textwidth]{figures/00_merged_Landsats_Smoothed_and_raw} \caption{In this plot the data points from Landsat-5, -7, and -8 (Tier 1, Surface Reflectance, from GEE collection LANDSAT/LE07/C01/T1\_SR) are merged together to form one vector. The same is done to Landsat-5 and -7 Collection-2 Level-2 (from GEE collection LANDSAT/LE07/C02/T1\_L2).} We can see they all are performing well. \label{fig:C2L2Performance} \end{figure} Moreover, GEE~\cite{Landsat7T1SRBandWidths} says ``This dataset is the atmospherically corrected surface reflectance from the Landsat 7 ETM+ sensor.'' about ``USGS Landsat 7 Surface Reflectance Tier 1'' (LANDSAT/LE07/C01/T1\_SR). On the other hand, it also says ``Caution: This dataset has been superseded by LANDSAT/LC08/C02/T1\_L2.'' Collection-1 has only Level-1 data, however, Collection-2 has level-1 as well as Level-2. \begin{description} \item [Collection 1] Landsat Collection 1 was established in 2016 to improve archive management. \href{https://www.usgs.gov/core-science-systems/nli/landsat/landsat-collection-1?qt-science_support_page_related_con=1#qt-science_support_page_related_con}{Learn more about Collection 1 from the USGS.} Landsat Collection 1 consists of Level-1 data products generated from Landsat 8 Operational Land Imager (OLI)/Thermal Infrared Sensor (TIRS), Landsat 7 Enhanced Thematic Mapper Plus (ETM+), Landsat 4-5 Thematic Mapper (TM)*, and Landsat 1-5 Multispectral Scanner (MSS) instruments. \textbf{Collection 1 Tiers:} \begin{description} \item [Tier 1] ``Landsat scenes with the highest available data quality are placed into Tier 1 and are considered suitable for time-series analysis.''~\cite{C1Describe} \item [Tier 2] ``Landsat scenes not meeting Tier 1 criteria during processing are assigned to Tier 2. Tier 2 scenes adhere to the same radiometric standard as Tier 1 scenes, but do not meet the Tier 1 geometry specification due to less accurate orbital information (specific to older Landsat sensors), significant cloud cover, insufficient ground control, or other factors.''~\cite{C1Describe} \end{description} \item [Collection 2] Landsat Collection 2 marks the second major reprocessing effort on the Landsat archive by the USGS that results in several data product improvements that harness recent advancements in data processing, algorithm development, and data access and distribution capabilities. \href{https://www.usgs.gov/core-science-systems/nli/landsat/landsat-collection-2?qt-science_support_page_related_con=1#}{Learn more about Collection 2 from the USGS.} Collection-2 Level-1 has different processings for different satellites~\cite{C2L1Describe}. It seems Collection-2 level-1 is TOA and Collection-2 level-2 is Surface Reflectance. ``Collection-2 Level-2 science products are generated from Collection 2 Level-1 inputs that meet the <76 degrees Solar Zenith Angle constraint and include the required auxiliary data inputs to generate a scientifically viable product.''~\cite{C2L2Describe}. ``\textbf{Surface reflectance} (unitless) measures the fraction of incoming solar radiation that is reflected from the Earth's surface to the Landsat sensor. The LEDAPS and LaSRC surface reflectance algorithms correct for the temporally, spatially and spectrally varying scattering and absorbing effects of atmospheric gases, aerosols, and water vapor, which is necessary to reliably characterize the Earth’s land surface.''~\cite{C2L2Describe}. For the enhancement details please see~\cite{C2L2Describe}. \end{description} \section{Scaling the Bands} \label{sec:Scaling-the-Bands} The purpose of this section is to make a point. Since it is an important point, a section is devoted to it. If you look at the band tables on \href{https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S2#bands}{Sentinel-2}, there is a column called \emph{scale}. If you look at the band table of \href{https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LC08_C02_T1_L2#bands}{Landsat 8 Level 2, Collection 2, Tier 1}, there are two columns called \emph{scale} and \emph{offset}. But such columns do not exist on \href{https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LT05_C01_T1_TOA#bands}{Landsat 5 TM Collection 1 Tier 1 TOA Reflectance}. For some reason, Google Earth Engine has not scaled the bands and has made that your problem. So, you have to scale the bands properly during computations. If you forget to scale in case of Sentinel-2 and $\NDVI = \frac{\NIR - R}{\NIR + R}$ you will be lucky since scales cancel out but that will not happen in case of EVI because of the additional 1 in the denominator (or in case of Landsat an off-set parameter is present as well); \begin{gather}\label{eq:EVIeq} % gather and aligned leads to having one label for eq. \begin{aligned} \EVI &\coloneqq G \times \frac{\rho_{NIR} - \rho_R}{\rho_{NIR} + C_1 \rho_R - C_2 \rho_B + L} \\ &= 2.5 \times \frac{\rho_{NIR} - \rho_R}{\rho_{NIR} + 6 \rho_R - 7.5 \rho_B + 1} \\ \end{aligned} \end{gather} Moreover, if you search the web for masking clouds in Sentinel, you will find the function \href{https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S2}{maskS2clouds}. If you look closely, in the last line the function is dividing the result by 10,000. Therefore, you do not have to scale the bands again in computation of VIs. However, you have to apply the \func{maskS2clouds} functions to the image collection before computing the VIs. %%%%%%%-------------------------------------------------------------------------- %%%%%%% %%%%%%% Access a Feature/Entry of a FeatureCollection %%%%%%% \section{Access a Feature/Entry of a FeatureCollection} Suppose your \code{featurecollection} is called \code{SF}. In order to access its entries you have to convert it to a \vari{list} and then use \func{get(.)}:\\ \func{print} (\codeRed{``SF.get(0)''}, \code{SF}.\func{toList}(\vari{4}).\func{get}(\vari{0}));\\ \noindent where \vari{4} is the size of \code{SF} known in advance, and \vari{0} is index of first entry of \code{SF}. In general you can use:\\ \func{print} (\codeRed{``SF.get(0)''}, \code{SF}.\func{toList}(\vari{\code{SF}.\func{size}\code{()}}).\func{get}(\vari{index}));\\ \noindent Please note if you use \code{SF}.\func{get}(\codeCyan{0}) you will get an error. %%%%%%%---------------------------------------- %%%%%%% %%%%%%% %%%%%%% \section{Add a Property to a Feature} Suppose you have uploaded a shapefile \vari{SF} into your assets. The shapefiles usually have a component/slice called \vari{data} (which is of type datatable) that can be accessed via \vari{SF@data} in R. This component stores metadata corresponding to each polygon. Say each polygon is an agricultural field that has some attributes associated with it such as irrigation type, area of the field, etc. After some computations on GEE you may want to attach these metadata to the output to use later. These metadata is referred to by \vari{properties} on GEE. If you want to manually add a property to a feature you should use:\\ \code{a\_feature = a\_feature}.\func{set}(\codeRed{`my\_property'}, \codeRed{1});\\ If you want to copy \vari{properties} (metadata) of \vari{feature\_b} into \vari{feature\_a} you can do:\\ \code{feature\_a} = \code{feature\_a}.\func{copyProperties}(\code{feature\_b}, [\codeRed{`ID'}, \codeRed{`Irrigation\_type'}]);\\ \noindent where [\codeRed{`ID'}, \codeRed{`Irrigation\_type'}] is a subset of \vari{properties} of \vari{feature\_b} to be copied into \vari{feature\_a}. I guess if that argument is dropped, then all \vari{properties} will be copied. %%%%%%%---------------------------------------- %%%%%%% %%%%%%% Find Centroid of Polygons %%%%%%% \section{Find Centroid of Polygons} Suppose you have a shapefile that you have uploaded to GEE as an \emph{asset}. Here we will see how to find the centroids of the polygons in the shapefile. Let the name of shapefile be \vari{Our\_ShapeFile}. The function to compute centroids of the polygons in \vari{Our\_ShapeFile} is given by Alg.~\ref{alg:FindCentroidsAPoly}\footnote{This algorithm is accessible on GEE \href{https://code.earthengine.google.com/df1685205d4fbfd8def0efb12a75f8e4}{here}.}. Line 4 of the Alg.~\ref{alg:FindCentroidsAPoly} is keeping the columns of data slice in \vari{Our\_ShapeFile}; \vari{Our\_ShapeFile}@data. \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FindCentroidsAPoly} \caption{Find Centroids of Polygons in a Shapefile.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{getCentroid(feature)} \{ \\ \vspace{.1in} \hspace{.2in} 2. {\color{ForestGreen}{// Keep this list of properties.}}\; \hspace{.2in} 3. \textbf{var} \code{keepProperties =} [\codeRed{`ID'}, \codeRed{`county'}] \; \vspace{.2in} \hspace{.2in} 4. {\color{ForestGreen}{// Get the centroid of the feature's geometry.}}\; \hspace{.2in} 5. \textbf{var} \code{centroid =} \code{feature}.\func{geometry}().\func{centroid}(); \; \vspace{.2in} \hspace{.2in} 6. {\color{ForestGreen}{// Return a new Feature, copying properties from the \hspace{0.5in} old Feature.}}\; \hspace{.2in} 7. \textbf{return} \func{ee.Feature}(\code{centroid}).\func{copyProperties} (\code{feature}, \hspace{2.8in}\code{keepProperties})\; 8. \} \vspace{.1in} 9. \textbf{var} \code{SF =} \func{ee.FeatureCollection}(\code{Our\_ShapeFile})\; 10. \textbf{var} \code{centroids\_from\_GEE} = \code{SF}.\func{map}(\code{getCentroid}); \end{algorithm} \end{tcolorbox} {\color{red}{Warning:}} Imagine your polygon looks like a doughnut (non-convex shape). Then the centroid would be in the center of the disk in the center of the doughnut which is not part of the doughnut/polygon/region of interest. So, if you want to look at an area around the centroid, then that area (or parts of it, depending on how large the area is) would not belong to the polygon (See Fig.~\ref{fig:badBuffer}; it is not a doughnut, but it delivers the message!) \begin{figure}[!h] \centering \begin{subfigure}[b]{0.4\textwidth} % \centering \includegraphics[width=1.1\textwidth]{figures/badCentroid} \caption{Bad Centroid} \label{fig:badCentroid} \end{subfigure} \qquad \begin{subfigure}[b]{.4\textwidth} % \centering \includegraphics[width=1.06\textwidth]{figures/badBuffer} \caption{Bad Buffer} \label{fig:badBuffer} \end{subfigure} \caption{Centroids and buffers around the centroids of polygons in a shapefile.} \label{fig:badPolygon} \end{figure} By adding one line (line 5.5 in Alg.~\ref{alg:FindCentroidsBuffers}) to the function \func{getCentroid(.)} we can get a buffer (a rectangular or a circle area) around the centroids. \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FindCentroidsBuffers} \caption{Make a Buffer Around Centroids of Polygons.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{get\_rectangle\_arround\_centroid(feature)}\{ \\ \vspace{.1in} \hspace{.2in} 2. {\color{ForestGreen}{// Keep this list of properties.}}\; \hspace{.2in} 3. \textbf{var} \code{keepProperties} = [\codeRed{`ID'}, \codeRed{`county'}] \; \vspace{.2in} \hspace{.2in} 4. {\color{ForestGreen}{// Get the centroid of the feature's geometry.}}\; \hspace{.2in} 5. \textbf{var} \code{centroid} = \code{feature}.\func{geometry}().\func{centroid}(); \; \vspace{.1in} \hspace{.2in} 5.5 \small{\code{centroid} = \func{ee.Feature}(\code{centroid}.\func{buffer}(\codeCyan{200}).\func{bounds}())}\; \vspace{.1in} \hspace{.2in} 6. {\color{ForestGreen}{// Return a new Feature, copying properties from the \hspace{.2in} old Feature.}}\; \hspace{.2in} 7. \textbf{return} \func{ee.Feature}(\code{centroid}).\func{copyProperties}(\code{feature}, \hspace{2.8in} \code{keepProperties})\; 8. \} \vspace{.1in} 9. \textbf{var} \code{SF} = \func{ee.FeatureCollection}(\code{Our\_ShapeFile})\; 10. \textbf{var} \code{centroids\_from\_GEE} = \hspace{1.4in} \code{SF}.\func{map} (\code{get\_rectangle\_arround\_centroid}); \end{algorithm} \end{tcolorbox} %%%%%%%---------------------------------------- %%%%%%% %%%%%%% Cloud Filtering %%%%%%% \section{Cloud Filtering} Handling clouds for Sentinel and Landsat are different. Let us start by \textbf{Sentinel}.\\ \noindent First, the followings are equivalent: \begin{itemize}[leftmargin=0.5cm] \item \textbf{var} \code{filtered = my\_IC}.\func{filterMetadata}( \codeRed{`CLOUDY\_PIXEL\_PERCENTAGE'}, \hspace{2.5in} \codeRed{`less\_than'}, \codeRed{70}); \item \textbf{var} \code{filtered = my\_IC}.\func{filter}(\codeRed{`CLOUDY\_PIXEL\_PERCENTAGE < 70'}) \item \textbf{var} \code{filtered = my\_IC}.\func{filter}(\func{ee.Filter.lte}(\codeRed{`CLOUDY\_PIXEL\_PERCENTAGE'}, \hspace{2.9in} \codeRed{70})) \end{itemize} \noindent They all filter out \emph{images} with cloud cover less than or equal to 70\%. Those images will NOT be in our \code{filtered} collection. Said differently, our \code{filtered} collection may include images that are covered by cloud up t0 70\%. This is a pre-filtering step. Later, we can toss out the cloudy \emph{pixels} from every single image. \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FilterCloudyPixelsSentinel} \caption{Filter Cloudy Pixels for Sentinel.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{maskS2clouds(image)} \{ \\ \vspace{.2in} \hspace{.2in} 2. {\color{ForestGreen}{// Each Sentinel-2 image has a bitmask band with cloud \hspace{.55in} mask information QA60.}}\; \hspace{.2in} 3. \textbf{var} \code{qa = image}.\func{select}(\codeRed{`QA60'}); \vspace{.2in} \hspace{.2in} 4. {\color{ForestGreen}{// Bits 10 and 11 are clouds and cirrus, respectively.}}\; \hspace{.2in} 5. \textbf{var} \code{cloudBitMask =} \codeCyan{1} $\code{<<}$ \codeCyan{10} \; \hspace{.2in} 6. \textbf{var} \code{cirrusBitMask =} \codeCyan{1} $\code{<<}$ \codeCyan{11} \; \vspace{.2in} \hspace{.2in} 7. {\color{ForestGreen}{// Both flags should be set to zero, indicating clear \hspace{.6in} conditions.}}\; \hspace{.2in} 8. \textbf{var} \code{mask = qa}.\func{bitwiseAnd}(\code{cloudBitMask}) .\func{eq}(\codeCyan{0}).\func{and}( \hspace{1.1in} \code{qa}.\func{bitwiseAnd}(\code{cirrusBitMask}) .\func{eq}(\codeCyan{0}))\; \vspace{.2in} \hspace{.2in} 9. {\color{ForestGreen}{// Return the masked and scaled data, without \hspace{.7in} the QA bands.}} \hspace{.2in} 10. \textbf{return} \code{image}.\func{updateMask}(\code{mask}) \hspace{1.25in} .\func{divide}(\codeCyan{10000}) \hspace{1.25in} .\func{select}(\codeRed{``B.*''}) \hspace{.4in} .\func{copyProperties}( \code{image}, [\codeRed{``system:time\_start''}]); \hspace{.01in} 11. \} \end{algorithm} \end{tcolorbox} \noindent \textbf{{\color{red}{Note 1:}}} Please note the last line in Alg.~\ref{alg:FilterCloudyPixelsSentinel} is copying the system start time into the image which has nothing to do with clouds. It may be handy later.\\ \noindent \textbf{{\color{red}{Note 2:}}} Please note the three (equivalent) pre-filtering of images mentioned above do not exist for Landsat!\\ Landsat(s) is a different satellite, and therefore, the cloud filtering must be handled differently; the band names that includes cloud information are different between Sentinel and Landsat or even among different Landsats.\\ Landsat-8 \emph{Surface Reflectance} cloud mask~\cite{Landsat8CloudMask}: \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FilterCloudyPixelsLandsat8} \caption{Filter Cloudy Pixels for Landsat-8 \textbf{Tier 1 and 2} \emph{Surface Reflectance}.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{maskL8sr(image)} \{\\ \vspace{.1in} \hspace{.2in} 2. {\color{ForestGreen}{// Bits 3 and 5 are cloud shadow and cloud, \hspace{.55in} respectively.}}\; \hspace{.2in} 3. \textbf{var} \code{cloudShadowBitMask} = (\codeCyan{1} $\code{<<}$ \codeCyan{3})\; \hspace{.2in} 4. \textbf{var} \code{cloudsBitMask} = (\codeCyan{1} $\code{<<}$ \codeCyan{5})\; \vspace{.2in} \hspace{.2in} 5. {\color{ForestGreen}{// Get the pixel QA band.}}\; \hspace{.2in} 6. \textbf{var} \code{qa =} \code{image}.\func{select}(\codeRed{`pixel\_qa'})\; \vspace{.2in} \hspace{.2in} 7. {\color{ForestGreen}{// Both flags should be set to zero, indicating clear \hspace{.5in} conditions.}}\; \hspace{0.2in} 8. {\small{ \textbf{var} \code{mask = } \code{qa}.\func{bitwiseAnd}(\code{cloudShadowBitMask}) .\func{eq}(\codeCyan{0}) \hspace{1.25in} .\func{and}(\code{qa}.\func{bitwiseAnd}(\code{cloudsBitMask}).\func{eq}(\codeCyan{0}))}}\; \vspace{.2in} \hspace{.2in} 9. \textbf{return} \code{image}.\func{updateMask}(\code{mask})\; \hspace{.01in} 10. \} \end{algorithm} \end{tcolorbox} \noindent \textbf{{\color{red}{Note:}}} This is written for Landsat-8 (Surface Reflectance Tier 1 and 2).\\ The code for masking the cloudy pixels in Landsat-4, 5, and 7 \emph{Surface Reflectance} is given by~\cite{Landsat457CloudMask} that is given below by Alg.~\ref{alg:FilterCloudyPixelsLandsat457}: \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FilterCloudyPixelsLandsat457} \caption{Filter Cloudy Pixels for Landsat-4, 5, and 7 \textbf{Tier 1 and 2} \emph{Surface Reflectance}.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{cloudMaskL457(image)} \{ \vspace{.1in} \hspace{.2in} 2. \textbf{var} \code{qa = image}.\func{select}(\codeRed{`pixel\_qa'})\; \vspace{.1in} \hspace{.2in} 3. {\color{ForestGreen}{// If the cloud bit (5) is set and the cloud confidence (7) \hspace{0.6in} is high or the cloud shadow bit is set (3), \hspace{0.6in} then it's a bad pixel.}}\\ \hspace{.2in} 4. \textbf{var} \code{cloud = qa}.\func{bitwiseAnd}(\codeCyan{1} $\code{<<}$ \codeCyan{5})\\ \hspace{1.35in} .\func{and}(\code{qa}.\func{bitwiseAnd}(\codeCyan{1} $\code{<<}$ \codeCyan{7}))\\ \hspace{1.35in} .\func{or}(\code{qa}.\func{bitwiseAnd}(\codeCyan{1} $\code{<<}$ \codeCyan{3}))\; \vspace{.1in} \hspace{.2in} 5. {\color{ForestGreen}{// Remove edge pixels that don't occur in all bands}}\\ \hspace{.2in} 6. \textbf{var} \code{mask2 = image}.\func{mask}().\func{reduce}(\func{ee.Reducer.min}())\; \hspace{.2in} 7. \small{\textbf{return} \code{image}.\func{updateMask}(\code{cloud}.\func{not}()).\func{updateMask}(\code{mask2});} \hspace{.01in} 10. \} \end{algorithm} \end{tcolorbox} I have copied the cloud masking functions from GEE development/data-product pages into a script that can be found \href{https://code.earthengine.google.com/?scriptPath=users\%2Fhnoorazar\%2FGEE_Mini_Tutorial\%3ACloudMaskings}{here}~\cite{CloudandMaskingFunctionsonMyGEE}. More on masking clouds of Sentinel-2 and shadows are provided \href{https://developers.google.com/earth-engine/tutorials/community/sentinel-2-s2cloudless}{here} by GEE developers~\cite{CloudandShadowSentinel}. Another way of masking cloud used in ~\cite{rs11070820}: \begin{tcolorbox} \begin{algorithm}[H] \label{alg:FilterCloudyPixelsLandsat78TOA} \caption{Filter Cloudy Pixels for Landsat-7 and 8 \emph{TOA}; LANDSAT/LC08/C01/T1\_TOA and\\ LANDSAT/LE07/C01/T1\_TOA.} \SetAlgoLined % \KwResult{Faster convergence} 1. \textbf{function} \code{cloudMask(image)} \{ \vspace{.1in} \hspace{.2in} 2. \textbf{var} \code{cloudscore =} ee.Algorithms.Landsat\\\hspace{1.6in} \func{.simpleCloudScore}(image).\\ \hspace{1.6in} \func{.select}(`cloud')\; \vspace{.1in} \hspace{.2in} 3. \small{\textbf{return} \code{image}.\func{updateMask}(cloudscore.\func{lt}(50));} \hspace{.01in} 10. \} \end{algorithm} \end{tcolorbox} %%%%%%%---------------------------------------- %%%%%%% %%%%%%% Timelines %%%%%%% \section{Timelines} \label{sec:Timelines} Figure~\ref{fig:landsatTimeline} shows the timeline of Landsat satellites~\cite{LandsatTimelinesWiki} and~\cref{tab:ElevationTable} shows the exact dates. \begin{figure}[H] \includegraphics[width=1\textwidth]{figures/landsatTimeline1} \caption{Landsat Timeline.} \label{fig:landsatTimeline} \end{figure} \begin{table}[] \centering \caption{Landsat timeline table.} \label{tab:ElevationTable} \begin{tabular}{|l|l|l|} \hline \rowcolor{shadecolor} \small{Satellite} & \small{Launched} & \small{Terminated} \\ \hline Landsat 5 & 1 March 1984 & 5 June 2013 \\ \hline \rowcolor{aliceblue} Landsat 6 & 5 October 1993 & 5 October 1993 \\ \hline Landsat 7 & 15 April 1999 & Still active \\ \hline \rowcolor{aliceblue} Landsat 8 & 11 February 2013 & Still active \\ \hline Landsat 9 & 16 September 2021 (planned) & - \\ \hline \end{tabular} \end{table} %%%%%%%---------------------------------------- %%%%%%% %%%%%%% Band Names and Indices %%%%%%% \section{Band Names and Indices} \label{sec:Band_Names_and_Indices} Band names are different in each instrument (see~\cref{tab:BandNameTable}). Hence the indices must be defined differently using proper band names. Below we see some of indices. \cref{tab:SomeBandWavelengths} also provides more insight about the bandwidths of the satellites. The bandwidths are very similar. If their minimal differences makes any difference I am not aware of it and do not care. Go nuts if you wish; figure out why, what, how. Bandwidths of Sentinel-2 is found on Wikipedia~\cite{SentinelBandwidths} and Bandwidths of Landsats can be found on GEE pages (e.g.~\cite{Landsat7T1SRBandWidths}). \begin{table}[] \centering \caption{Some Band Names in Satellites.} \label{tab:BandNameTable} \begin{tabular}{|l|l|l|l|} \hline \rowcolor{shadecolor} \small{Satellite} & \small{NIR} & \small{Red} & \small{Blue} \\ \hline Sentinel & B8 & B4 & B2\\ \hline \rowcolor{aliceblue} Landsat-8 & B5 & B4 & B2\\ \hline Landsat-7 & B4 & B3 & B1\\ \hline \rowcolor{aliceblue} Landsat-5 & B4 & B3 & B1\\ \hline \end{tabular} \end{table} \begin{table}[] \centering \caption{Some Band Wavelengths. The bandwidths are very similar. If their minimal differences makes any difference I am not aware of it and do not care. Go nuts if you wish; figure out why, what, how.} \label{tab:SomeBandWavelengths} \begin{tabular}{|l|l|l|l|} \hline \rowcolor{shadecolor} \small{Satellite} & \small{NIR} & \small{Red} & \small{Blue} \\ \hline Sentinel-2A & B8:~~~~~~0.77 -- 0.88 $\mu m$ & B4:~~~~~~0.65 -- 0.68 $\mu m$ & B2:~~~~~~0.46 -- 0.52 $\mu m$\\ \hline \rowcolor{aliceblue} Sentinel-2B & B8:~~~~~~0.78 -- 0.88 $\mu m$ & B4:~~~~~~0.65 -- 0.68 $\mu m$ & B2:~~~~~~0.46 -- 0.52 $\mu m$ \\ \hline Landsat-8 & B5:~~~~~~0.85 -- 0.88 $\mu m$ & B4:~~~~~~0.64 -- 0.67 $\mu m$ & B2:~~~~~~0.45 -- 0.51 $\mu m$\\ \hline \rowcolor{aliceblue} Landsat-7 & B4:~~~~~~0.77 -- 0.90 $\mu m$ & B3:~~~~~~0.63 -- 0.69 $\mu m$ & B1:~~~~~~0.45 -- 0.52 $\mu m$\\ \hline Landsat-5 & B4:~~~~~~0.77 -- 0.90 $\mu m$ & B3:~~~~~~0.63 -- 0.69 $\mu m$ & B1:~~~~~~0.45 -- 0.52 $\mu m$\\ \hline \rowcolor{aliceblue} Landsat-7 C2 L2 & SR\_B4: 0.77 -- 0.90 $\mu m$ & SR\_B3: 0.63 -- 0.69 $\mu m$ & SR\_B1: 0.45 -- 0.52 $\mu m$\\ \hline Landsat-5 C2 L2 & SR\_B4: 0.77 -- 0.90 $\mu m$ & SR\_B3: 0.63 -- 0.69 $\mu m$ & SR\_B1: 0.45 -- 0.52 $\mu m$\\ \hline \end{tabular} \end{table} \begin{gather} \label{eq:EVILandsat8} \begin{aligned} \EVI &= G \times \frac{\NIR - R}{\NIR + C1 \times R - C2 \times B + L} \\ \EVI_S &= 2.5 \times \frac{B8 - B4}{B8 + 6 \times B4 - 7.5 \times B2 + 1}\\ \EVI_8 &= 2.5 \times \frac{B5 - B4}{B5 + 6 \times B4 - 7.5 \times B2 + 1} \\ \EVI_7 &= 2.5 \times \frac{B4 - B3}{B4 + 6 \times B3 - 7.5 \times B1 + 1} \end{aligned} \end{gather} \noindent where $NIR$ is near infrared, $R$ is Red, $B$ is blue, $\text{EVI}_8$ is the Enhanced Vegetation Index (EVI) in Landsat-8~\cite{Landsat8EVI}, and $\text{EVI}_S$ is the EVI in Sentinel; The NIR band in Landsat-8 is $B5$~\cite{L8BandNames} and for Sentinel is $B8$. ``EVI is similar to Normalized Difference Vegetation Index (NDVI) and can be used to quantify vegetation greenness. However, EVI corrects for some atmospheric conditions and canopy background noise and is more sensitive in areas with dense vegetation. It incorporates an ``$L$'' value to adjust for canopy background, ``$C$'' values as coefficients for atmospheric resistance, and values from the blue band ($B$). These enhancements allow for index calculation as a ratio between the $R$ and $NIR$ values, while reducing the background noise, atmospheric noise, and saturation in most cases''~\cite{Landsat8EVI}. Below are the NDVIs for Landsat-4 to Landsat-7~\cite{Landsat4NDVI}, Landsat-8~\cite{Landsat4NDVI}, and Sentinel: \begin{gather} \label{eq:NDVILandsat8} \begin{aligned} \NDVI &= \frac{\NIR - R}{\NIR + R}\\ \NDVI_S &= \frac{B5 - B4}{B5 + B4}\\ \NDVI_8 &= \frac{B8 - B4}{B8 + B4} \\ \NDVI_{4-7} &= \frac{B4 - B3}{B4 + B3} \\ \end{aligned} \end{gather} Landsat-7 has 8-day NDVI composite already provided by GEE~\cite{Landsat7NDVIComposite}. This product is based on TOA data which is not perfect! However, it seems running some smoothing methods on it can make it useful. \section{Tiny Tips, Big Problems} \label{sec:Tiny-Tips-Big-Problems} The tips in this section are useful for beginners and if you want to do something that is unusual. Some times you may find yourself in a situation for which you are using the biggest sledgehammer to deal with the tiniest nail. In these scenarios the empire of Google does not have a function (for good reasons most likely) to do the job. If brute force is the chosen approach then these tips may be handy. If you are the only person on the planet who wants to do a certain thing, maybe you need to think again, and let go of useless approaches. \begin{description} \item [Object Types] There are two types of objects or functions. Some are called server-side. Some are called client-side. Here is an \href{https://code.earthengine.google.com/?scriptPath=users\%2Fhnoorazar\%2FGEE_Mini_Tutorial\%3AobjectTypeForLoop}{example} that shows a client-side object does not work with server-side object. It is strongly advised to avoid using/writing client-side objects/functions. The client-side objects also make the server/code/interface be very slow, freeze at times. \item [Batch Export] This is an example that Google does not think is useful. But if you need to export a collection of images you can do it either using a for-loop for which you may need to look at the previous example. Or, you can use \func{batch.Download.ImageCollection.toDrive(.)}. Both of these approaches are demonstrated \href{https://code.earthengine.google.com/?scriptPath=users\%2Fhnoorazar\%2FGEE_Mini_Tutorial\%3ABatchExport}{here}. Two remarks in this regard. First, the function for downloading the image collection as a batch\footnote{\func{batch.Download.ImageCollection.toDrive(.)}} behaves strangely.\footnote{I was visualizing the images as RGB images and exporting them; \textbf{var} \code{imageRGB =} \codeCyan{an\_image}. \func{visualize}(vizParams). I am not too sure if the batch download's problem is specific to RGB images.} In~\cref{fig:strangeBatch} there are 4 parts. The top left shows two images in a folder; one is exported via for-loop and the other is exported via batch-download. In the batch-downloaded image, naked eye cannot see anything, it is black and white. After opening it, it turns all into white (lower left). But the image exported via for-loop can be seen with naked eye (top right). The strange event is that the batch-downloaded image, can be seen if it is opened via Python or GIS (lower right image)! \begin{figure}[H] \centering \includegraphics[width=.9\textwidth]{figures/strangeBatch} \caption{Strange Behavior of Batch-Download.} \label{fig:strangeBatch} \end{figure} The images I exported turned out to be black and white. Secondly, any time a data is exported on GEE interface, you need to click on the \textbf{Run} button on \textbf{Task} tab. Perhaps Python can be used to avoid this problem, as well as server-side/client-side problem altogether. \end{description} \section{Full Code Examples; Landsat and Sentinel} Here are two examples, one for Landsat-8~\cite{FullCodeLandsat8} and one for Sentinel-2~\cite{FullCodeSentinel}. They are both on the GEE Mini Tutorial repo on Google Earth Engine~\cite{GEEMiniTutorialRepoonEE}. I have had problems with sharing repo in the past. If that does not work, you can copy the codes from the GitHub repo where this PDF is located at~\cite{MiniTutorialOnGitHub}. There is also a shapefile on Google drive~\cite{ShapeFileOnDrive} that is used in some of these codes. \subsection{Code Example} Merging data on GEE; see appendix of the paper\\ ~\href{https://www.mdpi.com/2072-4292/11/7/820/htm}{https://www.mdpi.com/2072-4292/11/7/820/htm} \section{Beyond GEE} I like to advice once you are done with GEE, read your CSV files (Python/R/etc.) and round the digits to 2 (or 3?) decimal places if you will. That reduces the file sizes. Of course this metters when you are working with substantial number of fields. %%%%%%%---------------------------------------- %%%%%%% %%%%%%% references %%%%%%% \bibliographystyle{unsrt} \bibliography{GEE_References.bib} %%%%%%%---------------------------------------- %%%%%%% %%%%%%% The End %%%%%%% \end{document} \documentclass[12pt,fleqn,dvipdfmx]{jarticle} \usepackage{docmute} \input{./@settings/common} \input{./@settings/master} \input{./@settings/commands} \begin{document} \title{papers summary} \author{} \date{} \maketitle \tableofcontents \newpage % papers \include{./example/index} \include{./StarGAN/index} \end{document} % 画像挿入用 % \begin{figure}[htbp]\centering\includegraphics[width=9cm, bb=0 0 300 300]{photo.pdf}\caption{}\label{fig:photo}\end{figure} % 複数画像挿入用 % \begin{figure}[htbp]\begin{tabular}{c}\begin{minipage}{0.5 \hsize}\centering\includegraphics[width=5cm, bb=0 0 640 640]{photo.jpg}\caption{}\label{fig:}\end{minipage} % \begin{minipage}{0.5 \hsize}\centering\includegraphics[width=5cm, bb=0 0 640 640]{photo.jpg}\caption{}\label{fig:}\end{minipage}\end{tabular}\end{figure} % 表挿入用 % \begin{table}[htbp]\centering\caption{}\includegraphics[width=9cm, bb=0 0 300 300]{photo.pdf}\label{tab:photo}\end{table} notes/L16.tex10-100 \documentclass[11pt]{article} \usepackage{listings} \usepackage{tikz} \usepackage{alltt} \usepackage{hyperref} \usepackage{url} \usepackage{enumitem} %\usepackage{algorithm2e} \usetikzlibrary{arrows,automata,shapes} \tikzstyle{block} = [rectangle, draw, fill=blue!20, text width=5em, text centered, rounded corners, minimum height=2em] \tikzstyle{bt} = [rectangle, draw, fill=blue!20, text width=1em, text centered, rounded corners, minimum height=2em] \newtheorem{defn}{Definition} \newtheorem{crit}{Criterion} \newcommand{\true}{\mbox{\sf true}} \newcommand{\false}{\mbox{\sf false}} \newcommand{\handout}[5]{ \noindent \begin{center} \framebox{ \vbox{ \hbox to 5.78in { {\bf Software Testing, Quality Assurance and Maintenance } \hfill #2 } \vspace{4mm} \hbox to 5.78in { {\Large \hfill #5 \hfill} } \vspace{2mm} \hbox to 5.78in { {\em #3 \hfill #4} } } } \end{center} \vspace*{4mm} } \newcommand{\lecture}[4]{\handout{#1}{#2}{#3}{#4}{Lecture #1}} \topmargin 0pt \advance \topmargin by -\headheight \advance \topmargin by -\headsep \textheight 8.9in \oddsidemargin 0pt \evensidemargin \oddsidemargin \marginparwidth 0.5in \textwidth 6.5in \parindent 0in \parskip 1.5ex %\renewcommand{\baselinestretch}{1.25} %\renewcommand{\baselinestretch}{1.25} \begin{document} \lecture{16 --- February 25, 2019}{Winter 2019}{}{version 1} Recall that the goals of mutation testing are: \begin{enumerate}[noitemsep] \item mimic (and hence test for) typical mistakes; \item encode knowledge about specific kinds of effective tests in practice, e.g. statement coverage ($\Delta 4$ from Lecture 15), checking for 0 values ($\Delta 6$). \end{enumerate} Reiterating the process for using mutation testing: \begin{itemize}[noitemsep] \item \emph{Goal:} kill mutants \item \emph{Desired Side Effect:} good tests which kill the mutants. \end{itemize} These tests will help find faults (we hope). We find these tests by intuition and analysis. \section*{Mutation Operators} We'll define a number of mutation operators, although precise definitions are specific to a language of interest. Typical mutation operators will encode typical programmer mistakes, e.g. by changing relational operators or variable references; or common testing heuristics, e.g. fail on zero. Some mutation operators are better than others. You can find a more exhaustive list of mutation operators in the PIT documentation: \begin{center} \url{http://pitest.org/quickstart/mutators/} \end{center} How many (intraprocedural) mutation operators can you invent for the following code? { \Large \begin{lstlisting} int mutationTest(int a, b) { int x = 3 * a, y; if (m > n) { y = -n; } else if (!(a > -b)) { x = a * b; } return x; } \end{lstlisting} } \paragraph{Integration Mutation.} We can go beyond mutating method bodies by also mutating interfaces between methods, e.g. \begin{itemize} \item change calling method by changing actual parameter values; \item change calling method by changing callee; or \item change callee by changing inputs and outputs. \end{itemize} { \begin{lstlisting} class M { int f, g; void c(int x) { foo (x, g); bar (3, x); } int foo(int a, int b) { return a + b * f; } int bar(int a, int b) { return a * b; } } \end{lstlisting} } [Absolute value insertion, operator replacement, scalar variable replacement, statement replacement with crash statements\ldots] \paragraph{Mutation for OO Programs.} One can also use some operators specific to object-oriented programs. Most obviously, one can modify the object on which field accesses and method calls occur. {\small \begin{lstlisting} class A { public int x; Object f; Square s; void m() { int x; f = new Object(); this.x = 5; } } class B extends A { int x; } \end{lstlisting} } \vspace*{-1em} \paragraph{Exercise.} Come up with a test case to kill each of these types of mutants. \begin{itemize} \item {\bf ABS}: Absolute Value Insertion\\ {\tt x = 3 * a} $\Longrightarrow$ {\tt x = 3 * abs(a)}, {\tt x = 3 * -abs(a)}, {\tt x = 3 * failOnZero(a)}; \item {\bf ROR}: Relational Operator Replacement\\ {\tt if (m > n)} $\Longrightarrow$ {\tt if (m >= n)}, {\tt if (m < n)}, {\tt if (m <= n)}, {\tt if (m == n)}, {\tt if (m != n)}, {\tt if (false)}, {\tt if (true)} \item {\bf UOD}: Unary Operator Deletion\\ {\tt if (!(a > -b))} $\Longrightarrow$ {\tt if (a > -b)}, {\tt if (!(a > b))} \end{itemize} \vspace*{-1em} \paragraph{Summary of Syntax-Based Testing.}~\\ \begin{tabular}{l|ll} & Program-based & Input Space/Fuzzing \\ \hline Grammar & Programming language & Input languages / XML \\ Summary & Mutates programs / tests integration & Input space testing \\ Use Ground String? & Yes (compare outputs) & Sometimes \\ Use Valid Strings Only? & Yes (mutants must compile) & No \\ Tests & Mutants are not tests & Mutants are tests \\ Killing & Generate tests by killing & Not applicable \\ \end{tabular} Notes: \begin{itemize}[noitemsep] \item Program-based testing has notion of strong and weak mutants; applied exhaustively, program-based testing could subsume many other techniques. \item Sometimes we mutate the grammar, not strings, and get tests from the mutated grammar. \end{itemize} \paragraph{Tool support.} PIT Mutation testing tool: \url{http://pitest.org}. Mutates your program, reruns your test suite, tells you how it went. You need to distinguish equivalent vs. not-killed. \end{document} \hyt{kdejsou} \song{Kde jsou} \interpret{plihal}{} \refrainn{1}{ \chord{G}Kde jsou, kdepak \chord{C}jsou\\ naše \chord{G}velko\chord{Em}lepý \chord{D}plá\chord{G}ny?\chord{D}\\ Kde jsou, kdepak jsou --\\ na hřbitově zakopány. } \vers{1}{ \chord{F}Sedíš tu \chord{C}tiše, jako pěna,\chord{D\4}\nc\chord{D}\\ \chord{F}žijem svůj \chord{C}život, no jak se říká -- \chord{D\4}do ztra\chord{D}cena.\\ \chord{G}Kde jsou, kdepak \chord{C}jsou\\ naše \chord{G}velko\chord{Em}lepý \chord{D}plá\chord{G}ny?\chord{D} } \refrainn{2}{ Kde jsou, kdepak jsou\\ naše velkolepý plány?\\ Kde jsou, kdepak jsou\\ písně nikdy už nedopsány? } \vers{2}{ Život byl prima a neměl chybu,\\ teď je mi zima a nemám ani na Malibu.\\ Kde jsou, kdepak jsou\\ naše velkolepý plány? } \newpage haskell/db/v52/scutari16.bib @inProceedings{scutari16, title = {An Empirical-{B}ayes Score for Discrete {B}ayesian Networks}, author = {}, pages = {438-448}, abstract = {Bayesian network structure learning is often performed in a Bayesian setting, by evaluating candidate structures using their posterior probabilities for a given data set. Score-based algorithms then use those posterior probabilities as an objective function and return the \emph{maximum a posteriori} network as the learned model. For discrete Bayesian networks, the canonical choice for a posterior score is the Bayesian Dirichlet equivalent uniform (BDeu) marginal likelihood with a uniform (U) graph prior (Heckerman et al., 1995). Its favourable theoretical properties descend from assuming a uniform prior both on the space of the network structures and on the space of the parameters of the network. In this paper, we revisit the limitations of these assumptions and we introduce an alternative set of assumptions and the resulting score: the Bayesian Dirichlet sparse (BDs) empirical Bayes marginal likelihood with a marginal uniform (MU) graph prior. We evaluate its performance in an extensive simulation study, showing that MU+BDs is more accurate than U+BDeu both in learning the structure of the network and in predicting new observations, while not being computationally more complex to estimate.}, } 1-10 @article{MOURAD2019323, title = "A survey of models and algorithms for optimizing shared mobility", journal = "Transportation Research Part B: Methodological", volume = "123", pages = "323 - 346", year = "2019", issn = "0191-2615", doi = "https://doi.org/10.1016/j.trb.2019.02.003", url = "http://www.sciencedirect.com/science/article/pii/S0191261518304776", author = " and and ", keywords = "Optimization, Passenger and freight transportation, Prearranged and real-time ridesharing, Exact and heuristic methods", abstract = "The rise of research into shared mobility systems reflects emerging challenges, such as rising traffic congestion, rising oil prices and rising environmental concern. The operations research community has turned towards more sharable and sustainable systems of transportation. Shared mobility systems can be collapsed into two main streams: Those where people share rides and those where parcel transportation and people transportation are combined. This survey sets out to review recent research in this area, including different optimization approaches, and to provide guidelines and promising directions for future research. It makes a distinction between prearranged and real-time problem settings and their methods of solution, and also gives an overview of real-case applications relevant to the research area." }1-10 Traps.tex in trunk/Docs/Traps – OberonRu

source: trunk/Docs/Traps/Traps.tex @ 229

Revision 96, 4.6 KB checked in by lit-uriy, 6 years ago (diff)

Документация:
Закинул всю документацию из официального хранилища А2
(Появилась документация на Сборщик Мусора-RTGC!).
Line 
1\documentclass[a4paper,11pt]{article}
2\usepackage{color}
3\usepackage{hyperref}
4% listings: http://mirror.switch.ch/ftp/mirror/tex/help/Catalogue/entries/listings.html
5\usepackage{listings}
6\usepackage{xspace}
7% longtable: http://mirror.switch.ch/ftp/mirror/tex/help/Catalogue/entries/longtable.html
8\usepackage{longtable}
9\usepackage{array}
10% --------------------------------- page layout --------------------------------------
11\pagestyle{headings}
12% other font styles:
13%\usepackage[math]{iwona} % iwona, kurier
14%\usepackage{cmbright}
15
16\oddsidemargin -0 cm
17\evensidemargin 0 cm
18%\topmargin -1.5cm
19\textwidth 16   cm
20%\textheight 25 cm
21
22\parskip 5pt
23\parindent 0cm
24
25% --------------------------------- colors --------------------------------------
26\definecolor{lightgrey}{rgb}{0.5,0.5,0.5}
27\definecolor{darkgrey}{rgb}{0.4,0.4,0.4}
28
29% --------------------------------- own commands --------------------------------------
30\newcommand{\changefont}[3]{\fontfamily{#1}\fontseries{#2}\fontshape{#3}\selectfont}
31
32\newcommand{\progfont}{\changefont{pcr}{m}{n}}
33\newcommand{\kwfont}{\changefont{pcr}{b}{n}}
34\renewcommand{\lstlistingname}{Fig.}
35\newcommand{\todo}[1]{\setlength{\fboxrule}{2pt}\fcolorbox{red}{yellow}{\begin{minipage}{\textwidth} \color{blue}$todo:$ #1 \end{minipage}}}
36
37\newcommand{\pc}[1]{\makebox{\progfont #1}}
38\newcommand{\kw}[1]{\makebox{\kwfont #1}}
39\newcommand{\AZ}{\ensuremath{\mathcal{A}_{2}}\xspace}
40
41% --------------------------------- tables --------------------------------------
42\newcolumntype{v}[1]{>{\raggedright\hspace{0pt}}p{#1}} % line breaking but align text left (not block)
43
44% --------------------------------- listings --------------------------------------
45\lstdefinelanguage{ebnf}[]{}
46{morekeywords={},
47sensitive=true,
48comment=[l]{//},
49comment=[s]{(*}{*)},
50morestring=[b]',
51morestring=[b]",
52basicstyle=\scriptsize\changefont{pcr}{m}{n},
53stringstyle=\changefont{pcr}{b}{n},
54columns = fixed
55}
56\lstdefinelanguage{Oberon}[]{Oberon-2}%
57  {morekeywords={OBJECT,SELF,%
58   HUGEINT,% Basic Types
59   AWAIT},% Built in functions
60   %sensitive=f,%
61   %alsoother={},% list here in lower case if keyword some where else wrongly highlighted
62    morecomment=[s][\color{red}]{(*!}{!*)}
63  }[keywords]%
64\lstset{language=Oberon,
65basicstyle=\small\progfont,keywordstyle=\kwfont ,identifierstyle=\progfont,
66commentstyle=\color{darkgrey}, stringstyle=, showstringspaces=false, %keepspaces=true,
67numbers=none, numberstyle=\tiny, stepnumber=1, numbersep=5pt, captionpos=b,
68columns=flexible, % flexible, fixed, fullflexible
69framerule=1mm,frame=shadowbox, rulesepcolor=\color{blue}, % frame = shadowbox
70xleftmargin=2mm,xrightmargin=2mm,
71breaklines=true,                                % break long lines
72breakatwhitespace=true,                 % break lines only at white space
73}
74\renewcommand{\lstlistingname}{Fig.}
75
76
77\begin{document}
78\title{\AZ Traps}
79\author{}
80\maketitle
81Blah Blah
82
83\tableofcontents
84
85\section{Traps}\label{Traps}
86When a run-time error is detected, the system generates a numbered trap. The information included with the trap can be used to diagnose the problem. Especially useful is the module and procedure name and PC location where the trap occured. This allows a programmer to find the exact location in the source code.
87
88\subsection{Example}\label{section:ModulesAndCommands}
89
90\begin{lstlisting}[language=Oberon,frame=none,numbers=left]
91  MODULE TrapDemo;
92
93  PROCEDURE Proc2();
94  VAR string : POINTER TO ARRAY OF CHAR;
95  BEGIN
96    string := NIL;
97    ASSERT(string # NIL);
98  END Proc2;
99
100  PROCEDURE Proc1;
101  VAR a, b : LONGINT;
102  BEGIN
103    a := 99; b := 11;
104    Proc2();
105  END Proc1;
106
107  PROCEDURE Demo*;
108  VAR string : ARRAY 8 OF CHAR;
109  BEGIN
110    string := "Demo!";
111    Proc1();
112  END Demo;
113
114  END TrapDemo.
115\end{lstlisting}
116
117\begin{lstlisting}[language=Oberon,frame=none,numbers=left]
118  [1] TRAP 8 PL 3 8  ASSERT failed WinAos Revision 2081 (19.02.2009)
119  CS:=00000023 DS:=0000002B ES:=0000002B SS:=0000002B PC=0ECA7F92
120  ESI=0CCB679A EDI=05FBFF46 ESP=05FBFF20 PID=000017A8
121  EAX=00000000 EBX=00000000 ECX=75443D09 EDX=00000000
122  EBP=05FBFF28 FS:=00000053 GS:=0000002B TMR=00A97689
123  FLAGS: cPaZstIdo iopl0 {1..2, 6, 9}
124  Process: 6056 run 0 20ECE9BF0:Commands.Runner NIL {0, 28}
125  TrapDemo.Proc2 pc=34 [00000022H]
126    string=00000000H (NIL)
127  State TrapDemo:
128    @Self=0ECE9950H (Modules.Module)
129  TrapDemo.Proc1 pc=65 [00000041H]
130    a=99 (00000063H)
131    b=11 (0000000BH)
132  TrapDemo.Demo pc=95 [0000005FH]
133    string="Demo!"
134  Commands.Runner.@Body pc=1042 [00000412H]
135    @Self=0ECE9BF0H (Commands.Runner)
136\end{lstlisting}
137
138\end{document}
139
Note: See TracBrowser for help on using the repository browser.
\hypertarget{classcurlpp_1_1NoValueOptionTrait}{\section{curlpp\-:\-:No\-Value\-Option\-Trait$<$ option $>$ Class Template Reference} \label{classcurlpp_1_1NoValueOptionTrait}\index{curlpp\-::\-No\-Value\-Option\-Trait$<$ option $>$@{curlpp\-::\-No\-Value\-Option\-Trait$<$ option $>$}} } This class is just a wrapper around \hyperlink{classcurlpp_1_1OptionTrait}{curlpp\-::\-Option\-Trait}, in order to be able to have \char`\"{}\-No value\char`\"{} option, like Ssl\-Default\-Engine. {\ttfamily \#include $<$Option.\-hpp$>$} Inheritance diagram for curlpp\-:\-:No\-Value\-Option\-Trait$<$ option $>$\-:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=550pt]{classcurlpp_1_1NoValueOptionTrait__inherit__graph} \end{center} \end{figure} Collaboration diagram for curlpp\-:\-:No\-Value\-Option\-Trait$<$ option $>$\-:\nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[height=550pt]{classcurlpp_1_1NoValueOptionTrait__coll__graph} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item virtual \hyperlink{classcurlpp_1_1NoValueOptionTrait}{No\-Value\-Option\-Trait} $\ast$ \hyperlink{classcurlpp_1_1NoValueOptionTrait_af1e088d6a56673c18eaebe13b87fa743}{clone} () const \begin{DoxyCompactList}\small\item\em Return a copy of the current option. \end{DoxyCompactList}\end{DoxyCompactItemize} \subsection*{Additional Inherited Members} \subsection{Detailed Description} \subsubsection*{template$<$C\-U\-R\-Loption option$>$class curlpp\-::\-No\-Value\-Option\-Trait$<$ option $>$} This class is just a wrapper around \hyperlink{classcurlpp_1_1OptionTrait}{curlpp\-::\-Option\-Trait}, in order to be able to have \char`\"{}\-No value\char`\"{} option, like Ssl\-Default\-Engine. \subsection{Member Function Documentation} \hypertarget{classcurlpp_1_1NoValueOptionTrait_af1e088d6a56673c18eaebe13b87fa743}{\index{curlpp\-::\-No\-Value\-Option\-Trait@{curlpp\-::\-No\-Value\-Option\-Trait}!clone@{clone}} \index{clone@{clone}!curlpp::NoValueOptionTrait@{curlpp\-::\-No\-Value\-Option\-Trait}} \subsubsection[{clone}]{\setlength{\rightskip}{0pt plus 5cm}template$<$C\-U\-R\-Loption option$>$ {\bf No\-Value\-Option\-Trait}$<$ option $>$ $\ast$ {\bf curlpp\-::\-No\-Value\-Option\-Trait}$<$ option $>$\-::clone ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} ) const\hspace{0.3cm}{\ttfamily [virtual]}}}\label{classcurlpp_1_1NoValueOptionTrait_af1e088d6a56673c18eaebe13b87fa743} Return a copy of the current option. Note that the option value is copied too. Reimplemented from \hyperlink{classcurlpp_1_1OptionTrait_a2e0254ab4187bf413d8c53a0114a191d}{curlpp\-::\-Option\-Trait$<$ bool, option $>$}. The documentation for this class was generated from the following files\-:\begin{DoxyCompactItemize} \item include/curlpp/Option.\-hpp\item include/curlpp/Option.\-inl\end{DoxyCompactItemize} 1-10 % EXCL SO EMA-3 MACD-SMA EMA-7 EMA-11 RSI HA EMA-9 ZZS-kind MACD-EMA EMA-5 ZZS-level % INCL SMA-3 SMA-5 SMA-7 SMA-9 SMA-11 %{\sc USDCNY} & 1 & 0.01 & 316 & 37 & 330 & {\em 0.42} & 0.69 & --- & 30 & 14 & 28 & 18 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.49 & 2.31 & 1.10 & 0.06 \\ %{\sc USDCNY} & 1 & 0.02 & 298 & 65 & 320 & {\em 0.41} & 0.69 & --- & 30 & 20 & 25 & 21 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.07 & 1.22 & 1.08 & 0.03 \\ %{\sc USDCNY} & 1 & 0.03 & 279 & 100 & 304 & {\em 0.34} & 0.65 & --- & 30 & 22 & 19 & 20 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.78 & 1.51 & 1.09 & 0.07 \\ %{\sc USDCNY} & 1 & 0.04 & 262 & 136 & 285 & {\em 0.42} & 0.64 & --- & 30 & 23 & 17 & 24 & 19 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.77 & 1.36 & 1.07 & 0.03 \\ %{\sc USDCNY} & 1 & 0.05 & 248 & 165 & 270 & {\em 0.32} & 0.67 & --- & 30 & 22 & 18 & 24 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.38 & 1.31 & 1.11 & 0.07 \\ %{\sc USDCNY} & 1 & 0.1 & 191 & 310 & 182 & {\em 0.37} & 0.64 & --- & 30 & 23 & 11 & 20 & 21 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.15 & 1.88 & 1.08 & 0.04 \\ {\sc USDCNY} & 1 & 0.2 & 112 & 451 & 120 & {\em 0.40} & 0.71 & --- & 30 & 19 & 13 & 17 & 21 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 14.78 & 1.92 & 1.10 & 0.07 \\ %{\sc USDCNY} & 1 & 0.3 & 55 & 564 & 64 & {\em 0.35} & 0.66 & --- & 30 & 19 & 12 & 16 & 29 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.55 & 1.42 & 1.07 & 0.04 \\ %{\sc USDCNY} & 1 & 0.4 & 31 & 616 & 36 & {\em 0.40} & 0.65 & --- & 30 & 21 & 12 & 14 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.25 & 1.46 & 1.07 & 0.03 \\ % insufficient data for H1 T0.5 % insufficient data for H1 T1.0 % insufficient data for H1 T1.5 % insufficient data for H1 T2.0 % insufficient data for H1 T2.5 % insufficient data for H1 T3.0 {\sc USDCNY} & 2 & 0.01 & 316 & 22 & 345 & 0.74 & 0.84 & --- & 30 & 17 & 13 & 21 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.31 & 1.88 & 1.06 & 0.03 \\ %{\sc USDCNY} & 2 & 0.02 & 304 & 41 & 338 & {\em 0.23} & 0.56 & --- & 30 & 17 & 22 & 18 & 27 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.87 & 1.52 & 1.08 & 0.03 \\ %{\sc USDCNY} & 2 & 0.03 & 298 & 61 & 324 & {\em 0.30} & 0.54 & --- & 30 & 15 & 16 & 20 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.37 & 1.31 & 1.08 & 0.03 \\ %{\sc USDCNY} & 2 & 0.04 & 287 & 82 & 314 & {\em 0.22} & 0.53 & --- & 30 & 16 & 14 & 18 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.27 & 1.75 & 1.10 & 0.06 \\ %{\sc USDCNY} & 2 & 0.05 & 278 & 102 & 303 & {\em 0.32} & 0.54 & --- & 30 & 21 & 20 & 17 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.86 & 1.41 & 1.08 & 0.06 \\ %{\sc USDCNY} & 2 & 0.1 & 239 & 190 & 254 & {\em 0.33} & 0.55 & --- & 30 & 22 & 15 & 25 & 17 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.78 & 1.61 & 1.07 & 0.04 \\ %{\sc USDCNY} & 2 & 0.2 & 152 & 369 & 162 & {\em 0.36} & 0.56 & --- & 30 & 21 & 16 & 18 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.67 & 1.53 & 1.06 & 0.02 \\ %{\sc USDCNY} & 2 & 0.3 & 100 & 472 & 111 & {\em 0.42} & 0.58 & --- & 30 & 23 & 13 & 18 & 20 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.72 & 1.79 & 1.13 & 0.09 \\ %{\sc USDCNY} & 2 & 0.4 & 66 & 551 & 66 & {\em 0.34} & 0.58 & --- & 30 & 11 & 18 & 18 & 20 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 14.67 & 1.33 & 1.07 & 0.02 \\ %{\sc USDCNY} & 2 & 0.5 & 40 & 603 & 40 & {\em 0.34} & 0.60 & --- & 29 & 17 & 16 & 13 & 19 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 14.53 & 2.09 & 1.07 & 0.03 \\ % insufficient data for H2 T1.0 % insufficient data for H2 T1.5 % insufficient data for H2 T2.0 % insufficient data for H2 T2.5 % insufficient data for H2 T3.0 {\sc USDCNY} & 3 & 0.01 & 324 & 17 & 342 & {\em 0.44} & 0.77 & --- & 29 & 20 & 17 & 22 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.57 & 2.00 & 1.10 & 0.05 \\ %{\sc USDCNY} & 3 & 0.02 & 317 & 34 & 332 & {\em 0.17} & 0.47 & --- & 29 & 25 & 11 & 16 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.60 & 1.29 & 1.07 & 0.03 \\ %{\sc USDCNY} & 3 & 0.03 & 313 & 46 & 324 & {\em 0.27} & 0.50 & --- & 30 & 18 & 16 & 21 & 20 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.27 & 1.56 & 1.06 & 0.04 \\ %{\sc USDCNY} & 3 & 0.04 & 308 & 65 & 310 & {\em 0.22} & 0.45 & --- & 30 & 24 & 11 & 17 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.24 & 1.73 & 1.08 & 0.02 \\ %{\sc USDCNY} & 3 & 0.05 & 297 & 87 & 299 & {\em 0.29} & 0.45 & --- & 30 & 15 & 17 & 20 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.18 & 1.75 & 1.08 & 0.03 \\ %{\sc USDCNY} & 3 & 0.1 & 257 & 166 & 260 & {\em 0.26} & 0.50 & --- & 30 & 21 & 19 & 20 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.97 & 1.58 & 1.08 & 0.04 \\ %{\sc USDCNY} & 3 & 0.2 & 184 & 310 & 189 & {\em 0.38} & 0.56 & --- & 30 & 20 & 19 & 15 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.85 & 1.76 & 1.07 & 0.06 \\ %{\sc USDCNY} & 3 & 0.3 & 129 & 430 & 124 & {\em 0.29} & 0.53 & --- & 30 & 20 & 16 & 20 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.58 & 1.30 & 1.10 & 0.05 \\ %{\sc USDCNY} & 3 & 0.4 & 95 & 505 & 83 & {\em 0.25} & 0.54 & --- & 30 & 14 & 21 & 15 & 20 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 14.90 & 1.60 & 1.10 & 0.04 \\ %{\sc USDCNY} & 3 & 0.5 & 74 & 552 & 57 & {\em 0.28} & 0.53 & --- & 30 & 17 & 22 & 16 & 20 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.14 & 1.72 & 1.10 & 0.02 \\ % insufficient data for H3 T1.0 % insufficient data for H3 T1.5 % insufficient data for H3 T2.0 % insufficient data for H3 T2.5 % insufficient data for H3 T3.0 {\sc USDCNY} & 4 & 0.01 & 337 & 14 & 332 & {\em 0.41} & 0.81 & --- & 30 & 20 & 14 & 18 & 21 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.26 & 1.65 & 1.06 & 0.05 \\ {\sc USDCNY} & 4 & 0.02 & 331 & 28 & 324 & {\em 0.46} & 0.76 & --- & 30 & 20 & 11 & 14 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.04 & 1.14 & 1.08 & 0.03 \\ %{\sc USDCNY} & 4 & 0.03 & 326 & 43 & 314 & {\em 0.24} & 0.61 & --- & 30 & 23 & 24 & 20 & 17 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.07 & 1.17 & 1.12 & 0.04 \\ %{\sc USDCNY} & 4 & 0.04 & 315 & 59 & 309 & {\em 0.26} & 0.56 & --- & 30 & 18 & 25 & 22 & 21 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.07 & 1.30 & 1.09 & 0.03 \\ %{\sc USDCNY} & 4 & 0.05 & 311 & 70 & 302 & {\em 0.24} & 0.58 & --- & 30 & 22 & 24 & 22 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.56 & 1.90 & 1.11 & 0.04 \\ %{\sc USDCNY} & 4 & 0.1 & 271 & 144 & 268 & {\em 0.26} & 0.50 & --- & 29 & 19 & 19 & 19 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.57 & 1.80 & 1.05 & 0.03 \\ %{\sc USDCNY} & 4 & 0.2 & 204 & 267 & 212 & {\em 0.24} & 0.55 & --- & 29 & 23 & 21 & 19 & 27 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.53 & 1.24 & 1.10 & 0.02 \\ %{\sc USDCNY} & 4 & 0.3 & 156 & 364 & 163 & {\em 0.25} & 0.51 & --- & 30 & 15 & 19 & 20 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.66 & 1.55 & 1.06 & 0.03 \\ %{\sc USDCNY} & 4 & 0.4 & 121 & 449 & 113 & {\em 0.24} & 0.47 & --- & 30 & 15 & 15 & 17 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.34 & 1.64 & 1.07 & 0.03 \\ %{\sc USDCNY} & 4 & 0.5 & 91 & 506 & 86 & {\em 0.26} & 0.52 & --- & 30 & 18 & 16 & 15 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.23 & 1.60 & 1.08 & 0.04 \\ % insufficient data for H4 T1.0 % insufficient data for H4 T1.5 % insufficient data for H4 T2.0 % insufficient data for H4 T2.5 % insufficient data for H4 T3.0 {\sc USDCNY} & 5 & 0.01 & 345 & 11 & 327 & {\em 0.42} & 0.72 & --- & 30 & 24 & 17 & 15 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.80 & 1.41 & 1.07 & 0.03 \\ {\sc USDCNY} & 5 & 0.02 & 341 & 18 & 324 & {\em 0.38} & 0.72 & --- & 29 & 22 & 16 & 19 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.55 & 1.93 & 1.07 & 0.03 \\ %{\sc USDCNY} & 5 & 0.03 & 334 & 31 & 318 & {\em 0.21} & 0.55 & --- & 30 & 15 & 25 & 17 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.97 & 1.23 & 1.06 & 0.04 \\ %{\sc USDCNY} & 5 & 0.04 & 333 & 34 & 316 & {\em 0.26} & 0.53 & --- & 30 & 19 & 28 & 16 & 17 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.72 & 1.57 & 1.06 & 0.02 \\ %{\sc USDCNY} & 5 & 0.05 & 327 & 49 & 307 & {\em 0.20} & 0.52 & --- & 30 & 23 & 23 & 22 & 18 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 16.11 & 1.65 & 1.09 & 0.04 \\ %{\sc USDCNY} & 5 & 0.1 & 305 & 112 & 266 & {\em 0.21} & 0.54 & --- & 30 & 21 & 17 & 16 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.98 & 1.11 & 1.09 & 0.04 \\ %{\sc USDCNY} & 5 & 0.2 & 237 & 239 & 207 & {\em 0.22} & 0.46 & --- & 30 & 20 & 15 & 23 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.89 & 1.23 & 1.09 & 0.04 \\ %{\sc USDCNY} & 5 & 0.3 & 188 & 335 & 160 & {\em 0.26} & 0.51 & --- & 30 & 17 & 11 & 16 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.09 & 1.20 & 1.07 & 0.03 \\ %{\sc USDCNY} & 5 & 0.4 & 142 & 415 & 126 & {\em 0.23} & 0.47 & --- & 30 & 19 & 16 & 22 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.79 & 1.73 & 1.09 & 0.02 \\ %{\sc USDCNY} & 5 & 0.5 & 109 & 482 & 92 & {\em 0.26} & 0.50 & --- & 30 & 20 & 17 & 23 & 17 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 683 & 15.41 & 1.75 & 1.10 & 0.06 \\ % insufficient data for H5 T1.0 % insufficient data for H5 T1.5 % insufficient data for H5 T2.0 % insufficient data for H5 T2.5 % insufficient data for H5 T3.0 %{\sc USDCNY} & 10 & 0.01 & 361 & 8 & 313 & {\em 0.31} & 0.67 & --- & 30 & 19 & 17 & 17 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.32 & 1.80 & 1.08 & 0.05 \\ %{\sc USDCNY} & 10 & 0.02 & 355 & 19 & 308 & {\em 0.35} & 0.68 & --- & 30 & 20 & 14 & 16 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.09 & 1.75 & 1.09 & 0.09 \\ %{\sc USDCNY} & 10 & 0.03 & 351 & 24 & 307 & {\em 0.25} & 0.66 & --- & 30 & 18 & 17 & 14 & 27 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.60 & 1.23 & 1.06 & 0.02 \\ %{\sc USDCNY} & 10 & 0.04 & 347 & 34 & 301 & {\em 0.23} & 0.46 & --- & 30 & 27 & 20 & 18 & 17 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 16.10 & 1.31 & 1.06 & 0.03 \\ %{\sc USDCNY} & 10 & 0.05 & 340 & 45 & 297 & {\em 0.14} & 0.48 & --- & 30 & 25 & 20 & 13 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.88 & 1.50 & 1.08 & 0.03 \\ %{\sc USDCNY} & 10 & 0.1 & 324 & 75 & 283 & {\em 0.21} & 0.48 & --- & 30 & 20 & 22 & 17 & 27 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 16.14 & 1.65 & 1.07 & 0.03 \\ %{\sc USDCNY} & 10 & 0.2 & 286 & 155 & 241 & {\em 0.19} & 0.46 & --- & 30 & 22 & 17 & 18 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.91 & 2.14 & 1.16 & 0.09 \\ %{\sc USDCNY} & 10 & 0.3 & 241 & 242 & 199 & {\em 0.16} & 0.47 & --- & 29 & 22 & 19 & 19 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 16.23 & 1.11 & 1.09 & 0.05 \\ %{\sc USDCNY} & 10 & 0.4 & 208 & 309 & 165 & {\em 0.24} & 0.51 & --- & 30 & 21 & 17 & 18 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.96 & 1.55 & 1.08 & 0.04 \\ %{\sc USDCNY} & 10 & 0.5 & 175 & 363 & 144 & {\em 0.22} & 0.52 & --- & 30 & 22 & 11 & 16 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.44 & 1.25 & 1.07 & 0.03 \\ %{\sc USDCNY} & 10 & 1.0 & 67 & 555 & 60 & {\em 0.28} & 0.50 & --- & 30 & 22 & 15 & 14 & 29 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 682 & 15.59 & 1.46 & 1.06 & 0.04 \\ % insufficient data for H10 T1.5 % insufficient data for H10 T2.0 % insufficient data for H10 T2.5 % insufficient data for H10 T3.0 %{\sc USDCNY} & 15 & 0.01 & 364 & 6 & 311 & {\em 0.23} & 0.63 & --- & 30 & 21 & 8 & 20 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.16 & 1.91 & 1.05 & 0.03 \\ %{\sc USDCNY} & 15 & 0.02 & 362 & 15 & 304 & {\em 0.36} & 0.67 & --- & 30 & 17 & 21 & 16 & 23 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.53 & 1.61 & 1.08 & 0.02 \\ %{\sc USDCNY} & 15 & 0.03 & 361 & 18 & 302 & {\em 0.27} & 0.64 & --- & 30 & 19 & 17 & 19 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.53 & 1.92 & 1.08 & 0.03 \\ %{\sc USDCNY} & 15 & 0.04 & 360 & 25 & 296 & {\em 0.30} & 0.66 & --- & 30 & 18 & 17 & 16 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.55 & 1.37 & 1.08 & 0.03 \\ %{\sc USDCNY} & 15 & 0.05 & 358 & 29 & 294 & {\em 0.38} & 0.62 & --- & 30 & 18 & 17 & 19 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.73 & 1.75 & 1.08 & 0.02 \\ %{\sc USDCNY} & 15 & 0.1 & 338 & 58 & 285 & {\em 0.27} & 0.45 & --- & 30 & 24 & 16 & 17 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.53 & 1.41 & 1.07 & 0.03 \\ %{\sc USDCNY} & 15 & 0.2 & 310 & 121 & 250 & {\em 0.24} & 0.40 & --- & 30 & 22 & 17 & 17 & 24 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.74 & 1.57 & 1.06 & 0.04 \\ %{\sc USDCNY} & 15 & 0.3 & 282 & 182 & 217 & {\em 0.18} & 0.46 & --- & 30 & 21 & 13 & 18 & 26 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.67 & 1.77 & 1.07 & 0.03 \\ %{\sc USDCNY} & 15 & 0.4 & 252 & 240 & 189 & {\em 0.20} & 0.44 & --- & 30 & 24 & 17 & 22 & 22 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.85 & 1.37 & 1.08 & 0.04 \\ %{\sc USDCNY} & 15 & 0.5 & 220 & 297 & 164 & {\em 0.23} & 0.39 & --- & 30 & 24 & 15 & 18 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 16.07 & 1.32 & 1.10 & 0.07 \\ %{\sc USDCNY} & 15 & 1.0 & 103 & 488 & 90 & {\em 0.25} & 0.45 & --- & 30 & 21 & 17 & 18 & 25 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.67 & 1.31 & 1.07 & 0.08 \\ %{\sc USDCNY} & 15 & 1.5 & 36 & 598 & 47 & {\em 0.28} & 0.47 & --- & 30 & 17 & 19 & 7 & 28 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.17 & 1.80 & 1.07 & 0.04 \\ %{\sc USDCNY} & 15 & 2.0 & 13 & 634 & 34 & {\em 0.39} & 0.63 & --- & 30 & 18 & 21 & 17 & 21 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & 681 & 15.64 & 1.08 & 1.09 & 0.03 \\ % insufficient data for H15 T2.5 % insufficient data for H15 T3.0 {\sc USDCNY } & \multicolumn{9}{|r|}{Frequency of inclusion in feature selection (\%)} & 100 & 66 & 58 & 60 & 77 & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & --- & \multicolumn{5}{|l|}{\phantom{total}} \\ docs/wstepna/tex/ocena_modeli.tex Część danych (ok. 10-20\%) zostanie wydzielona jako zbiór testujący, pozostała część będzie stanowiła zbiór uczący. Pozwoli to ocenić, jak dobrze model radzi sobie z uogólnianiem pojęcia. Ocena modelu predykcyjnego odbędzie się na podstawie poniższych parametrów: \begin{itemize} \item suma kwadratów różnicy wartości predykcyjnej od wartości rzeczywistej \begin{equation} MSE = {\frac{1}{n}\sum_{i=1}^{n}(Y_{i} - y_{i})^{2}} \end{equation} \item pierwiastek sumy kwadratów różnicy wartości predykcyjnej od wartości rzeczywistej \begin{equation} RMSE = \sqrt{(\frac{1}{n})\sum_{i=1}^{n}(Y_{i} - y_{i})^{2}} \end{equation} \item sprawdzenie confusion matrix - czyli informacja o klasyfikacji próbek fałszywie i prawdziwie sklasyfikowanych dla poszczególnych klas, co pozwoli ocenić jak model radzi sobie z niezbalansowanymi klasami; \item sprawdzenie dokładności modelu przez użycie walidacji krzyżowej; \item w modelu zastosujemy bootstrapping, czyli K-krotne użycie pierwotnego zbioru danych do uczenia modelu, w tym wypadku zbadamy zagregowany błąd predykowany w agregacie do zagregowanego błędu rzeczywistego; \end{itemize} AYCH-Inc/aych.hyperwp \subsection{Identity} \input{HighlightedFeatures/identity.tex}lambdalainen/metropolia-thesis-latex % Appendix to demonstrate R integration \chapter{R examples} TODO \clearpage %force the next chapter/appendix to start on a new page. Keep that as the last line of your appendix! %--------------------------------------------------------------------- % % TeXiS_bib.tex % %--------------------------------------------------------------------- % % TeXiS_bib.tex % Copyright 2009 , % % This file belongs to TeXiS, a LaTeX template for writting % Thesis and other documents. The complete last TeXiS package can % be obtained from http://gaia.fdi.ucm.es/projects/texis/ % % This work may be distributed and/or modified under the % conditions of the LaTeX Project Public License, either version 1.3 % of this license or (at your option) any later version. % The latest version of this license is in % http://www.latex-project.org/lppl.txt % and version 1.3 or later is part of all distributions of LaTeX % version 2005/12/01 or later. % % This work has the LPPL maintenance status `maintained'. % % The Current Maintainers of this work are % and % %--------------------------------------------------------------------- % % Fichero que contiene la generaci�n de la bibliograf�a. En principio, % no har�a falta tenerlo en un fichero separado, pero como permitimos % a�adir una frase c�lebre antes de la primera cita, la configuraci�n % ya no es trivial. Para "ocultar" la fontaner�a de LaTeX, est� % separada la configuraci�n de los par�metros de la generaci�n % concreta de la bibliograf�a. La configuraci�n est� en el "directorio % del usuario" (Cascaras), mientras que la generaci�n se encuentra en % el directorio TeXiS (este fichero). % %--------------------------------------------------------------------- %%% % Gesti�n de la configuraci�n %%% % Ficheros .bib \def\ficherosBibliografia{bibliographify} \newcommand{\setBibFiles}[1]{ \def\ficherosBibliografia{#1} } % Frase c�lebre \def\citaBibliografia{} \newcommand{\setCitaBibliografia}[1]{ \def\citaBibliografia{#1} } %%% % Configuraci�n terminada %%% %%% %% COMANDO PARA CREAR LA BIBLIOGRAF�A. %% CONTIENE TODO EL C�DIGO LaTeX %%% \newcommand{\makeBib}{ % % Queremos que tras el t�tulo del cap�tulo ("Bibliograf�a") aparezca % una frase c�lebre, igual que en el resto de cap�tulos. El problema % es que aqu� no ponemos nosotros a mano el \chapter{Bibliograf�a}, % sino que lo mete �l autom�ticamente. % % Afortunadamente, la gente de bibtex hace las cosas bien ;-) y % despu�s de insertar el t�tulo de la secci�n ejecuta un comando % denominado \bibpreamble que por defecto no hace nada. Pero si % sobreescribimos ese comando, podremos ejecutar c�digo arbitrario % justo despu�s de la inserci�n del t�tulo, y antes de la primera % referencia. Por tanto, lo que hacemos es sobreescribir ese comando % (normalmente se conocen como "hooks") para a�adir la cita justo % despu�s del t�tulo. % % Desgraciadamente, dependiendo de la versi�n de Natbib, hay que % definir o redefinir el comando (es decir, utilizar newcommand o % renewcommand)... como eso es un l�o, utilizamos let y def, pues def % no falla si ya estaba definido. \let\oldbibpreamble\bibpreamble \def\bibpreamble{% \oldbibpreamble % A�adimos a la tabla de contenidos la bibliograf�a. Si no lo hacemos % aqu�, sale mal o el n�mero de p�gina (grave) o el enlace en el PDF % que te lleva a un sitio cercano (no tan grave) \ifx\generatoc\undefined \else \addcontentsline{toc}{chapter}{Bibliography} \fi % A�adimos tambi�n una etiqueta, para poder referenciar el n�mero % de p�gina en el que comienza \label{bibliografia} % Frase c�lebre configurada por el usuario \citaBibliografia } % Fin definici�n "bibpreamble" % % Cambiamos el estilo de la cabecera. Hay que hacerlo porque por % defecto el paquete que estamos usando (fancyhdr) pone en may�sculas % el t�tulo completo del cap�tulo. Con los cap�tulos normales esto se % pudo evitar en el pre�mbulo redefiniendo el comando \chaptermark, % pero con la bibliograf�a no se puede hacer. Se define la cabera para % que aparezca la palabra "Bibliograf�a" en ambas p�ginas. % % \cabeceraEspecial{Bibliograf�a} % Creamos la bibliograf�a. Lo hacemos dentro de un bloque (entre la % pareja \begingroup ... \endgroup) porque dentro vamos a anular la % sem�ntica que da babel a la tilde de la e�e (que hace que un ~N se % convierta autom�ticamente en una �). Esto es debido a que en el % bibtex aparecer�n ~ para separar iniciales de los nombres con % espacios no separables en varias lineas, y aquellos nombres que % tengan una N como inicial ser�an puestos como �. Al anular la % sem�ntica al ~ que da babel, deshacemos este % comportamiento. Naturalmente, para que esto no tenga repercusiones % negativas, en ning�n .bib deber�amos utilizar ~N (o ~n) para % representar una � ... tendremos que utilizar o una �/� directamente % (no aconsejable porque asume que hemos usado inputenc) o, mejor, % usamos la versi�n larga \~n o \~N que no falla nunca. Para m�s % informaci�n, consulta el TeXiS_pream.tex en el punto donde se % incluye natbib y babel. \begingroup %\sp%anishdeactivate{~} \bibliographystyle{abbrv} %\bibliographystyle{abbrv} \bibliography{\ficherosBibliografia} \endgroup } %\newcommand{\makeBib} % Variable local para emacs, para que encuentre el fichero maestro de % compilaci�n y funcionen mejor algunas teclas r�pidas de AucTeX %%% %%% Local Variables: %%% mode: latex %%% TeX-master: "../Tesis.tex" %%% End: @article{vahldiek2018thesis, author = {Vahldiek-Oberwagner, }, publisher = {Saarländische Universitäts-und Landesbibliothek}, title = {Techniques to Protect Confidentiality and Integrity of Persistent and In-Memory Data}, url = {https://publikationen.sulb.uni-saarland.de/bitstream/20.500.11880/27354/1/thesis.pdf}, year = {2018} } tufts-ml-courses/comp136-spr-20s-assignments- \documentclass[12pt]{article} \usepackage{fullpage} \usepackage{microtype} % microtypography \usepackage{array} \usepackage{amsmath,amssymb,amsfonts} \usepackage{amsthm} %% Header \usepackage{fancyhdr} \fancyhf{} \fancyhead[C]{COMP 136 - 2020s - HW2 Submission} \fancyfoot[C]{\thepage} % page number \renewcommand\headrulewidth{0pt} \pagestyle{fancy} %% Hyperlinks always blue, no weird boxes \usepackage[hyphens]{url} \usepackage[colorlinks=true,allcolors=black,pdfborder={0 0 0}]{hyperref} %%% Doc layout \usepackage{parskip} \usepackage{times} %%% Write out problem statements in blue, solutions in black \usepackage{color} \newcommand{\officialdirections}[1]{{\color{blue} #1}} %%% Avoid automatic section numbers (we'll provide our own) \setcounter{secnumdepth}{0} \begin{document} ~~\\ %% add vert space {\Large{\bf Student Name: TODO}} {\Large{\bf Collaboration Statement:}} Turning in this assignment indicates you have abided by the course Collaboration Policy: \url{www.cs.tufts.edu/comp/136/2020s/index.html#collaboration-policy} Total hours spent: TODO I consulted the following resources: \begin{itemize} \item TODO \item TODO \item $\ldots$ \end{itemize} FYI Official instructions for all problems can be found at: \url{www.cs.tufts.edu/comp/136/2020s/hw2.html} \tableofcontents \newpage \officialdirections{ \subsection*{1a: Problem Statement} Consider the estimator $\hat{\sigma}^2$: \begin{align} \hat{\sigma}^2(x_1, \ldots x_N) = \frac{1}{N} \sum_{n=1}^N (x_n - \mu_{\text{true}})^2 \end{align} Compute the expected value of this estimator under the sampling distribution of the observations $x$. You should assume each $x_n$ are drawn i.i.d from $x_n \sim \mathcal{N}( \mu_{\text{true}}, \sigma^2_{\text{true}})$. } \subsection{1a: Solution} \newpage \officialdirections{ \subsection*{1b: Problem Statement} Given your result above, describe if the estimator is biased or unbiased. } \subsection{1b: Solution} TODO \officialdirections{ \subsection*{1c: Problem Statement} How does your answer about bias in 1b compare to what we learned about whether the maximum likelihood estimator for the variance is biased? Provide some justification. } \subsection{1c: Solution} TODO \newpage \officialdirections{ \subsection*{2a: Problem Statement} Show that the SGD estimator is an *unbiased* estimator of the whole dataset gradient. That is, show: \begin{align} \mathbb{E}_{i \sim U}[ \hat{G}(\theta, x_i) ] = G(\theta, x) \end{align} } \subsection{2a: Solution} \newpage \officialdirections{ \subsection*{2b: Problem Statement} How will the $\theta$ value estimated by whole dataset gradient descent (which has no randomness) typically compare to resulting $\theta$ from SGD? Why is it important that the SGD estimator is unbiased? } \subsection{2b: Solution} \newpage \officialdirections{ \subsection*{3a: Problem Statement} Given the log PDF below, show that $x$ has a univariate Gaussian distribution. \begin{align} \log p(x) = \text{const} - \frac{1}{2} a x^2 + bx \end{align} where $a > 0$ and $b$ is any number. } \subsection{3a: Solution} \newpage \officialdirections{ \subsection*{3b: Problem Statement} Given the log PDF below, show that vector $x$ has a multivariate Gaussian distribution. \begin{align} \log p(x) = \text{const} - \frac{1}{2} x^T A x + b^T x \end{align} where $A$ symmetric and positive definite, and $b$ any vector. } \subsection{3b: Solution} \newpage \officialdirections{ \subsection*{4a: Problem Statement} For the case $M=1$, show that we can write $S_{N+1}^{-1} = S_N^{-1} + v^2$ for some scalar value $v$. } \subsection{4a: Solution} \newpage \officialdirections{ \subsection*{4b: Problem Statement} For the case $M=1$, write an expression for $\sigma^2_{N+1} - \sigma^2_{N}$, simplifying as much as possible. } \subsection{4b: Solution} \newpage \officialdirections{ \subsection*{4c: Problem Statement} By studying the terms of your expression from **4b**, prove the following statement: $$ \sigma_{N+1}^2(x_*) \leq \sigma_N^2(x_*) $$ } \subsection{4c: Solution} \newpage \officialdirections{ \subsection*{5a: Problem Statement} Show that we can write $S_{N+1}^{-1} = S_N^{-1} + vv^T$ for some vector $v$. } \subsection{5a: Solution} \newpage \officialdirections{ \subsection*{5b: Problem Statement} \begin{align} (A + vv^T)^{-1} &= A^{-1} - \frac{ (A^{-1}v)(v^T A^{-1})}{ 1 + v^T A^{-1} v} \end{align} Substitute $A = S_N^{-1}$ and $v$ from 5a, then simplify to write an expression for $S_{N+1}$ in terms of $S_{N}$. } \subsection{5b: Solution} \officialdirections{ \subsection*{5c: Problem Statement} Show that $\sigma^2_{N+1}(x_*) - \sigma^2_{N}(x_*) = \phi(x_*)^T \left[ S_{N+1} - S_{N} \right] \phi(x_*)$ } \subsection{5c: Solution} \newpage \officialdirections{ \subsection*{5d: Problem Statement} Finally, plug your result from 5c into 5d, plus the fact that $S_N$ must be positive definite, to show that: \begin{align} \sigma_{N+1}^2(x_*) \leq \sigma_N^2(x_*), \quad \text{or equivalently} \quad \sigma_{N+1}^2(x_*) - \sigma_N^2(x_*) \leq 0 \end{align} } \subsection{5d: Solution} \end{document} 06_table_of_contents/file.tex1-10 \documentclass{article} \usepackage{setspace} % \setcounter{tocdepth}{0} show nothing % \setcounter{tocdepth}{1} show sections % \setcounter{tocdepth}{2} + subsections % \setcounter{tocdepth}{3} + subsubsections % \setcounter{tocdepth}{4} + paragraphs % \setcounter{tocdepth}{5} + subparagraphs \setcounter{tocdepth}{2} \title{My first document} \date{2020-07-01} \author{Christopher} \begin{document} \doublespacing % needs to be compiled twice \tableofcontents \singlespacing \section{Section} Dummy text \subsection{Subsection} Dummy text % empty dummies \begin{figure} \caption{Dummy figure} \end{figure} \begin{table} \caption{Dummy table} \end{table} % change the depth on the fly \addtocontents{toc}{\setcounter{tocdepth}{0}} \section{Section2} Dummy text \begin{appendix} \listoffigures \listoftables \end{appendix} \end{document} \hypertarget{classamunmt_1_1ThreadPool}{}\section{amunmt\+:\+:Thread\+Pool Class Reference} \label{classamunmt_1_1ThreadPool}\index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} {\ttfamily \#include $<$threadpool.\+h$>$} Collaboration diagram for amunmt\+:\+:Thread\+Pool\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d2/d4e/classamunmt_1_1ThreadPool__coll__graph} \end{center} \end{figure} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hyperlink{classamunmt_1_1ThreadPool_a3bc1ace57ef28ce5c78ee1672ae9c382}{Thread\+Pool} (size\+\_\+t threads, size\+\_\+t \hyperlink{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{bound}=0) \item {\footnotesize template$<$class F , class... Args$>$ }\\auto \hyperlink{classamunmt_1_1ThreadPool_a1bb89599cb11af51958c771c437aff22}{enqueue} (F \&\&f, Args \&\&...args) -\/$>$ std\+::future$<$ typename std\+::result\+\_\+of$<$ F(Args...)$>$\+::type $>$ \item \hyperlink{classamunmt_1_1ThreadPool_a03f83dd78a39dbe485326a50497d6906}{$\sim$\+Thread\+Pool} () \item size\+\_\+t \hyperlink{classamunmt_1_1ThreadPool_a56f770af85865ea1eb35f3d2ade28043}{get\+Num\+Tasks} () const \end{DoxyCompactItemize} \subsection*{Private Attributes} \begin{DoxyCompactItemize} \item std\+::vector$<$ std\+::thread $>$ \hyperlink{classamunmt_1_1ThreadPool_abd917167eb8a227f9aa7faaaec23745e}{workers} \item std\+::queue$<$ std\+::function$<$ void()$>$ $>$ \hyperlink{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546}{tasks} \item std\+::mutex \hyperlink{classamunmt_1_1ThreadPool_a4b1865f03dab554b2e933dfe6f662f48}{queue\+\_\+mutex} \item std\+::condition\+\_\+variable \hyperlink{classamunmt_1_1ThreadPool_a2f3b5194456806a842817062870b4907}{condition} \item std\+::size\+\_\+t \hyperlink{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{bound} \item std\+::condition\+\_\+variable \hyperlink{classamunmt_1_1ThreadPool_ade8c762e453107fd5e3e3097be95732e}{bounded\+\_\+condition} \item bool \hyperlink{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{stop} \end{DoxyCompactItemize} \subsection{Detailed Description} Definition at line 43 of file threadpool.\+h. \subsection{Constructor \& Destructor Documentation} \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!Thread\+Pool@{Thread\+Pool}} \index{Thread\+Pool@{Thread\+Pool}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{Thread\+Pool(size\+\_\+t threads, size\+\_\+t bound=0)}{ThreadPool(size_t threads, size_t bound=0)}}]{\setlength{\rightskip}{0pt plus 5cm}amunmt\+::\+Thread\+Pool\+::\+Thread\+Pool ( \begin{DoxyParamCaption} \item[{size\+\_\+t}]{threads, } \item[{size\+\_\+t}]{bound = {\ttfamily 0}} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [inline]}, {\ttfamily [explicit]}}\hypertarget{classamunmt_1_1ThreadPool_a3bc1ace57ef28ce5c78ee1672ae9c382}{}\label{classamunmt_1_1ThreadPool_a3bc1ace57ef28ce5c78ee1672ae9c382} Definition at line 71 of file threadpool.\+h. \begin{DoxyCode} 72 : \hyperlink{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{stop}(\textcolor{keyword}{false}), \hyperlink{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{bound}(in\_bound) \{ 73 \textcolor{keywordflow}{for} (\textcolor{keywordtype}{size\_t} i = 0;i task; 78 \{ 79 std::unique\_lock lock(this->queue\_mutex); 80 this->condition.wait(lock, 81 [this]\{ return this->stop || !this->tasks.empty(); \}); 82 if (this->stop && this->tasks.empty()) 83 return; 84 task = std::move(this->tasks.front()); 85 this->tasks.pop(); 86 \} 87 this->bounded\_condition.notify\_one(); 88 89 task(); 90 \} 91 \} 92 ); 93 \} \end{DoxyCode} \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!````~Thread\+Pool@{$\sim$\+Thread\+Pool}} \index{````~Thread\+Pool@{$\sim$\+Thread\+Pool}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{$\sim$\+Thread\+Pool()}{~ThreadPool()}}]{\setlength{\rightskip}{0pt plus 5cm}amunmt\+::\+Thread\+Pool\+::$\sim$\+Thread\+Pool ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} )\hspace{0.3cm}{\ttfamily [inline]}}\hypertarget{classamunmt_1_1ThreadPool_a03f83dd78a39dbe485326a50497d6906}{}\label{classamunmt_1_1ThreadPool_a03f83dd78a39dbe485326a50497d6906} Definition at line 122 of file threadpool.\+h. \begin{DoxyCode} 122 \{ 123 \{ 124 std::unique\_lock lock(\hyperlink{classamunmt_1_1ThreadPool_a4b1865f03dab554b2e933dfe6f662f48}{queue\_mutex}); 125 \hyperlink{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{stop} = \textcolor{keyword}{true}; 126 \} 127 \hyperlink{classamunmt_1_1ThreadPool_ade8c762e453107fd5e3e3097be95732e}{bounded\_condition}.notify\_all(); 128 \hyperlink{classamunmt_1_1ThreadPool_a2f3b5194456806a842817062870b4907}{condition}.notify\_all(); 129 \textcolor{keywordflow}{for} (std::thread &worker: \hyperlink{classamunmt_1_1ThreadPool_abd917167eb8a227f9aa7faaaec23745e}{workers}) \{ 130 worker.join(); 131 \} 132 \} \end{DoxyCode} \subsection{Member Function Documentation} \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!enqueue@{enqueue}} \index{enqueue@{enqueue}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{enqueue(\+F \&\&f, Args \&\&...\+args) -\/$>$ std\+::future$<$ typename std\+::result\+\_\+of$<$ F(\+Args...)$>$\+::type $>$}{enqueue(F &&f, Args &&...args) -> std::future< typename std::result_of< F(Args...)>::type >}}]{\setlength{\rightskip}{0pt plus 5cm}template$<$class F , class... Args$>$ auto amunmt\+::\+Thread\+Pool\+::enqueue ( \begin{DoxyParamCaption} \item[{F \&\&}]{f, } \item[{Args \&\&...}]{args} \end{DoxyParamCaption} ) -\/$>$ std\+::future$<$typename std\+::result\+\_\+of$<$F(Args...)$>$\+::type$>$}\hypertarget{classamunmt_1_1ThreadPool_a1bb89599cb11af51958c771c437aff22}{}\label{classamunmt_1_1ThreadPool_a1bb89599cb11af51958c771c437aff22} Definition at line 97 of file threadpool.\+h. \begin{DoxyCode} 99 \{ 100 \textcolor{keyword}{using} return\_type = \textcolor{keyword}{typename} std::result\_of::type; 101 102 \textcolor{keyword}{auto} task = std::make\_shared< std::packaged\_task >( 103 std::bind(std::forward(f), std::forward(args)...) 104 ); 105 106 std::future res = task->get\_future(); 107 \{ 108 std::unique\_lock lock(\hyperlink{classamunmt_1_1ThreadPool_a4b1865f03dab554b2e933dfe6f662f48}{queue\_mutex}); 109 this->\hyperlink{classamunmt_1_1ThreadPool_ade8c762e453107fd5e3e3097be95732e}{bounded\_condition}.wait(lock, [\textcolor{keyword}{this}] \{ \textcolor{keywordflow}{return} this-> \hyperlink{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546}{tasks}.size() < this->\hyperlink{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{bound} || this->\hyperlink{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{bound} == 0 || this->\hyperlink{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{stop}; \}); 110 \textcolor{comment}{// don't allow enqueueing after stopping the pool} 111 \textcolor{keywordflow}{if} (\hyperlink{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{stop}) \{ 112 \textcolor{keywordflow}{throw} std::runtime\_error(\textcolor{stringliteral}{"enqueue on stopped ThreadPool"}); 113 \} 114 115 \hyperlink{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546}{tasks}.emplace([task]()\{ (*task)(); \}); 116 \} 117 \hyperlink{classamunmt_1_1ThreadPool_a2f3b5194456806a842817062870b4907}{condition}.notify\_one(); 118 \textcolor{keywordflow}{return} res; 119 \} \end{DoxyCode} Here is the caller graph for this function\+: \nopagebreak \begin{figure}[H] \begin{center} \leavevmode \includegraphics[width=350pt]{d9/de5/classamunmt_1_1ThreadPool_a1bb89599cb11af51958c771c437aff22_icgraph} \end{center} \end{figure} \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!get\+Num\+Tasks@{get\+Num\+Tasks}} \index{get\+Num\+Tasks@{get\+Num\+Tasks}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{get\+Num\+Tasks() const }{getNumTasks() const }}]{\setlength{\rightskip}{0pt plus 5cm}size\+\_\+t amunmt\+::\+Thread\+Pool\+::get\+Num\+Tasks ( \begin{DoxyParamCaption} {} \end{DoxyParamCaption} ) const\hspace{0.3cm}{\ttfamily [inline]}}\hypertarget{classamunmt_1_1ThreadPool_a56f770af85865ea1eb35f3d2ade28043}{}\label{classamunmt_1_1ThreadPool_a56f770af85865ea1eb35f3d2ade28043} Definition at line 52 of file threadpool.\+h. \begin{DoxyCode} 52 \{ 53 \textcolor{keywordflow}{return} \hyperlink{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546}{tasks}.size(); 54 \} \end{DoxyCode} \subsection{Member Data Documentation} \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!bound@{bound}} \index{bound@{bound}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{bound}{bound}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::size\+\_\+t amunmt\+::\+Thread\+Pool\+::bound\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc}{}\label{classamunmt_1_1ThreadPool_aa85a12e692be6c514ffd8443b80ba0dc} Definition at line 65 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!bounded\+\_\+condition@{bounded\+\_\+condition}} \index{bounded\+\_\+condition@{bounded\+\_\+condition}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{bounded\+\_\+condition}{bounded_condition}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::condition\+\_\+variable amunmt\+::\+Thread\+Pool\+::bounded\+\_\+condition\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_ade8c762e453107fd5e3e3097be95732e}{}\label{classamunmt_1_1ThreadPool_ade8c762e453107fd5e3e3097be95732e} Definition at line 66 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!condition@{condition}} \index{condition@{condition}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{condition}{condition}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::condition\+\_\+variable amunmt\+::\+Thread\+Pool\+::condition\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_a2f3b5194456806a842817062870b4907}{}\label{classamunmt_1_1ThreadPool_a2f3b5194456806a842817062870b4907} Definition at line 64 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!queue\+\_\+mutex@{queue\+\_\+mutex}} \index{queue\+\_\+mutex@{queue\+\_\+mutex}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{queue\+\_\+mutex}{queue_mutex}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::mutex amunmt\+::\+Thread\+Pool\+::queue\+\_\+mutex\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_a4b1865f03dab554b2e933dfe6f662f48}{}\label{classamunmt_1_1ThreadPool_a4b1865f03dab554b2e933dfe6f662f48} Definition at line 63 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!stop@{stop}} \index{stop@{stop}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{stop}{stop}}]{\setlength{\rightskip}{0pt plus 5cm}bool amunmt\+::\+Thread\+Pool\+::stop\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7}{}\label{classamunmt_1_1ThreadPool_abf9a006374a3c361401bf98ecbfeabf7} Definition at line 67 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!tasks@{tasks}} \index{tasks@{tasks}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{tasks}{tasks}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::queue$<$ std\+::function$<$void()$>$ $>$ amunmt\+::\+Thread\+Pool\+::tasks\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546}{}\label{classamunmt_1_1ThreadPool_ab4e36abc1369312be77d1cfb04396546} Definition at line 60 of file threadpool.\+h. \index{amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}!workers@{workers}} \index{workers@{workers}!amunmt\+::\+Thread\+Pool@{amunmt\+::\+Thread\+Pool}} \subsubsection[{\texorpdfstring{workers}{workers}}]{\setlength{\rightskip}{0pt plus 5cm}std\+::vector$<$std\+::thread$>$ amunmt\+::\+Thread\+Pool\+::workers\hspace{0.3cm}{\ttfamily [private]}}\hypertarget{classamunmt_1_1ThreadPool_abd917167eb8a227f9aa7faaaec23745e}{}\label{classamunmt_1_1ThreadPool_abd917167eb8a227f9aa7faaaec23745e} Definition at line 58 of file threadpool.\+h. The documentation for this class was generated from the following file\+:\begin{DoxyCompactItemize} \item \hyperlink{threadpool_8h}{threadpool.\+h}\end{DoxyCompactItemize} 0 %!TEX root = paper.tex % \looseness-1 The reliability of inferences made by data-driven systems hinges on the data's continued conformance to the systems' initial settings and assumptions. When serving data (on which we want to apply inference) deviates from the profile of the initial training data, the outcome of inference becomes unreliable. % We introduce \emph{conformance constraints}, a new data profiling primitive tailored towards quantifying the degree of \emph{non-conformance}, which can effectively characterize if inference over that tuple is \emph{untrustworthy}. % \Dis are constraints over certain arithmetic expressions (called \emph{projections}) involving the numerical attributes of a dataset, which existing data profiling primitives such as functional dependencies and denial constraints cannot model. % Our key finding is that projections that incur \emph{low variance} on a dataset construct effective \dis. This principle yields the surprising result that low-variance components of a principal component analysis, which are usually discarded for dimensionality reduction, generate stronger \dis than the high-variance components. Based on this result, we provide a highly scalable and efficient technique---linear in data size and cubic in the number of attributes---for discovering conformance constraints for a dataset. To measure the degree of a tuple's non-conformance with respect to a dataset, we propose a \emph{quantitative semantics} that captures how much a tuple violates the \dis of that dataset. % We demonstrate the value of \dis on two applications: \emph{trusted machine learning} and \emph{data drift}. We empirically show that \dis offer mechanisms to (1)~reliably detect tuples on which the inference of a machine-learned model should not be trusted, and (2)~quantify data drift more accurately than the state of the art.joegeisz/pylith1-10 \chapter{Installation and Getting Help} \label{cha:installation} Figure~\ref{fig:install:choices} provides a guide to select the appropriate method for installing PyLith. Installation of PyLith on a desktop or laptop machine is, in most cases, very easy. Binary packages have been created for Linux and Mac OS X (Darwin) platforms. For Windows 10 users, we recommend installing the Windows Subsystem for Linux and using the Linux binary (see instructions in Section~\ref{sec:install:windows}). You can also run PyLith inside a Docker container, which provides a virtual Linux environment on any platform that Docker supports, including Linux, Mac OS X, and Windows. Installation of PyLith on other operating systems -- or installation on a cluster -- requires building the software from the source code, which can be difficult for inexperienced users. We have created a small utility called PyLith Installer that makes installing PyLith and all of its dependencies from source much easier. \begin{figure}[htbp] \includegraphics[scale=0.8]{install/figs/installchoices} \caption{Guide for selecting the appropriate installation choice based on a hardware and intended use. The installation options are discussed in more detail in the following sections.} \label{fig:install:choices} \end{figure} Help for installing and using PyLith is available from both a CIG mailing list and the GitHub issue tracking system \url{https://github.com/geodynamics/pylith/issues}. See Section~\vref{sec:help} for more information. \section{Installation of Binary Executable} The binaries are intended for users running on laptops or desktop computers (as opposed to clusters). The binaries contain the compilers and header files, so users wishing to extend the code can still use the binary and do not need to build PyLith and its dependencies from source. See Chapter~\vref{cha:extending} for more information on extending PyLith. Binary executables are available for Linux (glibc 2.12 and later) and Mac OS X (Intel 10.10 and later) from the PyLith web page \url{geodynamics.org/cig/software/packages/short/pylith/}. Users running Windows 10 build 14316 and later can install a Linux bash environment and use the PyLith binary for Linux (see Section~\vref{sec:install:windows} for more information). \tip{On Linux systems you can check which version of glibc you have by running \filename{ldd --version}}. \tip{On Darwin systems running OS X, you can check the operating system version by clicking on the Apple icon and \menu{About this Mac}.} \subsection{Linux and Mac OS X (Darwin)} \begin{enumerate} \item Open a terminal window and change to the directory where you want to place the distribution. \begin{shell} $ cd $HOME $ mkdir pylith $ cd pylith \end{shell} \item Download the Linux or Mac OS X (Darwin) tarball from the PyLith web page \url{geodynamics.org/cig/software/packages/short/pylith/}, and save it to the desired location, e.g., \filename{\$HOME/pylith}. \item Unpack the tarball. \begin{shell} # Linux 32-bit $ tar -xzf pylith-2.2.1-linux-i686.tgz # Linux 64-bit $ tar -xzf pylith-2.2.1-linux-x86_64.tgz # Mac OS X $ tar -xzf pylith-2.2.1-darwin-10.11.6.tgz \end{shell} \item Set environment variables. The provided \filename{setup.sh} script only works if you are using bash shell. If you are using a different shell, you will need to alter how the environment variables are set in \filename{setup.sh}. \begin{shell} $ source setup.sh \end{shell} \end{enumerate} \warning{The binary distribution contains PyLith and all of its dependencies. If you have any of this software already installed on your system, you need to be careful in setting up your environment so that preexisting software does not conflict with the PyLith binary. By default the \filename{setup.sh} script will prepend to the PATH and PYTHONPATH (for Darwin and Linux) and LD\_LIBRARY\_PATH (for Linux) environment variables. This will prevent most conflicts.} \warning{The PyLith binary distribution for {\bf Darwin} systems is built using the system clang compiler suite and the system Python. {\bf This means the system Python must be in your path to use the PyLith binary executable}; ensure \filename{/bin} and \filename{/usr/bin} are at the beginning of the PATH environment variable, which is done automatically if you use the \filename{setup.sh} script. {\bf This condition is often violated if you have Python installed from Anaconda, HomeBrew, MacPorts, etc. and set the PATH variable in your bash configuration file.}} \subsection{Windows 10} \label{sec:install:windows} PyLith is developed within the Unix/Linux framework, and we do not provide a native PyLith binary distribution for Windows. The preferred approach to installing PyLith on a computer running Windows 10 is to enable use of a Linux subsystem. This permits use of the PyLith Linux x86\_64 binary within the bash environment. To enable the Linux subsystem on Windows 10 build 14316 and later (users running an earlier Windows build should use the PyLith Docker container): \begin{enumerate} \item Go to \menu{Settings} $\rightarrow$ \menu{Security}. \item Under \menu{For developers} select \menu{Developer mode}. This step should not be required for Windows build 16215 and later. \item Go to \menu{Control Panel} $\rightarrow$ \menu{Programs} $\rightarrow$ \menu{Turn Windows Features On or Off}. \item Enable \menu{Windows Subsystem for Linux} and click \menu{OK}. \item Restart the computer. \item Go to \menu{Start} $\rightarrow$ \menu{bash}. You will be prompted to download "Bash on Ubuntu on Windows" from the Windows Store. Create a user account and password for the bash environment. \item Install the PyLith Linux x86 binary within the bash environment following the instructions for installing the PyLith binary for Linux. You will run PyLith within the bash environment just like you would for a Linux operating system. \end{enumerate} \subsection{Extending PyLith and/or Integrating Other Software Into PyLith} \newfeature{v.2.2.0} We have constructed the binary package so that you can extend PyLith and/or build additional software for integration with PyLith using the binary distribution. \begin{description} \item[Darwin] The binary package includes the header files for PyLith and all of its dependencies. Use the clang compiler and Python provided with the operating system. You will need to install XTools. \item[Linux] The binary package includes the GNU compilers, Python, as well as header files for PyLith and all of its dependencies. \end{description} \tip{We encourage anyone extending PyLith to fork the PyLith repository and build from source using the PyLith Installer Utility to facilitate contributing these features back into the CIG repository via pull requests.} \section{Installation of PyLith Docker Container} As an alternative to installing a binary package, we provide a Docker container for running PyLith in a self-contained virtual environment. Docker containers provide a self-contained virtual environment that are a smaller, simpler alternative to a virtual machine. The PyLith Docker container provides a Debian Linux environment with a pre-built PyLith executable, vim text editor, iceweasel (GNU version of Firefox) web-browser, and the matplotlib Python module. \tip{In nearly all cases, installing a PyLith binary provides easier integration with mesh generation and post-processing tools, so binaries are the preferred approach to using the PyLith Docker container. This installation method targets users running Windows versions earlier than Windows 10 build 14316.} \subsection{Setup (first time only)} \begin{enumerate} \item Install Docker (See \url{https://www.docker.com/products/docker}) \item Create a container to store persistent user data\\ This container, called pylith-data, will hold a directory where all your user data can be stored for use with PyLith within Docker. The data can persist for different versions of PyLith; that is, you can update to a newer version of PyLith and your user data will still be available. This directory is not directly accessible from your host computer. However, you can copy files to/from your host filesystem using ``docker cp'' (see below). \end{enumerate} \begin{shell}[] # Create the container $ docker create --name pylith-data geodynamics/pylith-data # Run the docker container and copy examples to the persistent storage. $ docker run -ti --volumes-from pylith-data geodynamics/pylith # This next command is run WITHIN the docker container. $ cp -R $HOME/pylith-VERSION/examples $HOME/data \end{shell} \subsection{Run Unix shell within Docker to use PyLith.} To run the container with a text only interface: \begin{shell} $ docker run -ti --volumes-from pylith-data geodynamics/pylith \end{shell} To run the container and allow display of windows on the host computer (requires that X-Windows be installed): \begin{shell} # Darwin: Allow X connections $ xhost +YOUR_IP_ADDRESS; DISPLAY=YOUR_IP_ADDRESS:0 # Linux: Allow X connections $ xhost +local:root # For Linux and Darwin, continue with the follow lines. $ XSOCK=/tmp/.X11-unix $ docker run -ti --volumes-from pylith-data \ -e DISPLAY=$DISPLAY -v $XSOCK:$XSOCK geodynamics/pylith \end{shell} In addition to a minimalist Debian Linux distribution and PyLith and all of its dependencies, the container includes the following useful utilities: \begin{description} \item[vim] Lightweight text editor \item[matplotlib] Python plotting module \item[iceweasel] GNU version of Firefox \end{description} \important{We do not yet include ParaView due to difficulties associated with setting up rendering on the host display outside the container. You will need to copy the output files to your host machine to view them in ParaView as described later.} \subsubsection{Using Docker containers} \begin{itemize} \item To ``pause'' a container: \texttt{Control-p Control-q} \item To attach to a ``paused'' or ``running'' container. \begin{shell} # Get the container id. $ docker ps # Attach to the container $ docker attach CONTAINER_ID \end{shell} \item To restart an existing container after it exited. \begin{shell} # Get the container id. $ docker ps -a # Start and then attach to the container $ docker run CONTAINER_ID $ docker attach CONTAINER_ID \end{shell} \end{itemize} \subsection{Copy data to/from persistent storage volume.} These commands are run on the local host outside the container, not inside the Docker container. These commands are used to move files from your host machine into the PyLith Docker container and vice versa. For example, you will generate your mesh on the host, copy the mesh file into the Docker container, run PyLith within the container, and then copy the output files to the host to display in ParaView. \begin{shell} # Copy data FROM persistent storage volume TO local host $ docker cp pylith-data:/data/pylith-user/PATH/FILENAME LOCAL_PATH # Copy data FROM local host TO persistent storage volume $ docker cp LOCAL_PATH pylith-data:/data/pylith-user/PATH/ \end{shell} \subsection{Docker Quick Reference} \begin{shell} # List local docker images. $ docker images # List all docker containers. $ docker ps -a # List running docker containers. $ docker ps # Remove docker container $ docker rm CONTAINER_ID # Remove docker image $ docker rmi IMAGE_ID \end{shell} \section{Installation from Source} PyLith depends on a number of other packages (see Figure \vref{fig:pylith-dependencies}). This complicates building the software from the source code. In many cases some of the packages required by PyLith are available as binary packages. On the one hand, using the binary packages for the dependencies removes the burden of configuring, building, and installing these dependencies, but that can come with its own host of complications if consistent compiler and configuration settings are not used across all of the packages on which PyLith depends. This is usually not an issue with Linux distributions, such as Fedora, Ubuntu, and Debian that have good quality control; it can be an issue with Darwin package managers, such as Fink, MacPorts, and Homebrew, where there is limited enforcement of consistency across packages. Nevertheless, PyLith can be built on most systems provided the instructions are followed carefully. PyLith is developed and tested on Linux and Mac OS X. A small utility, PyLith Installer, removes most of the obstacles in building PyLith and its dependencies from source. For each package this utility downloads the source code, configures it, builds it, and installs it. This insures that the versions of the dependencies are consistent with PyLith and that the proper configure arguments are used. The minimum requirements for using the PyLith installer are a C compiler, \filename{tar}, and \filename{wget} or \filename{curl}. Detailed instructions for how to install PyLith using the installer are included in the installer distribution, which is available from the PyLith web page \url{geodynamics.org/cig/software/packages/short/pylith/}. \section{Verifying PyLith is Installed Correctly} The easiest way to verify that PyLith has been installed correctly is to run one or more of the examples supplied with the binary and source code. In the binary distribution, the examples are located in \filename{src/pylith-\pylithVersionNumber/examples} while in the source distribution, they are located in \texttt{pylith-\pylithVersionNumber/examples}. Chapter \vref{cha:examples} discusses how to run and visualize the results for the examples. To run the example discussed in Section \vref{sec:example:3dhex8-static}: \begin{shell} $ cd examples/3d/hex8 $ pylith step01.cfg # A bunch of stuff will be written to stdout. The last few lines should be: WARNING! There are options you set that were not used! WARNING! could be spelling mistake, etc! Option left: name:-snes_atol value: 1.0e-9 Option left: name:-snes_converged_reason (no value) Option left: name:-snes_error_if_not_converged (no value) Option left: name:-snes_linesearch_monitor (no value) Option left: name:-snes_max_it value: 100 Option left: name:-snes_monitor (no value) Option left: name:-snes_rtol value: 1.0e-10 \end{shell} If you run PyLith in a directory without any input, you will get the error message: \begin{shell} $ pylith >> {default}:: -- pyre.inventory(error) -- meshimporter.meshioascii.filename <- '' -- Filename for ASCII input mesh not specified. To test PyLith, run an example as discussed in the manual. >> {default}:: -- pyre.inventory(error) -- timedependent.homogeneous.elasticisotropic3d.label <- '' -- Descriptive label for material not specified. >> {default}:: -- pyre.inventory(error) -- timedependent.homogeneous.elasticisotropic3d.simpledb.label <- '' -- Descriptive label for spatial database not specified. >> {default}:: -- pyre.inventory(error) -- timedependent.homogeneous.elasticisotropic3d.simpledb.simpleioascii.filename <- '' -- Filename for spatial database not specified. pylithapp: configuration error(s) \end{shell} This indicates that a number of default settings must be set in order to run PyLith, including setting the filename for the finite-element mesh. \section{Configuration on a Cluster} If you are installing PyLith on a cluster with a batch system, you can configure Pyre such that the \filename{pylith} command automatically submits jobs to the batch queue. Pyre contains support for the LSF, PBS, SGE, and Globus batch systems. The command to submit a batch job depends upon the particular batch system used. Further, the command used in a batch script to launch an MPI program varies from one cluster to the next. This command can vary between two clusters, even if the clusters use the same batch system! On some systems, \filename{mpirun} is invoked directly from the batch script. On others, a special wrapper is used instead. Properly configured, Pyre can handle job submissions automatically, insulating users from the details of the batch system and the site configuration. This feature has the most value when the system administrator installs a global Pyre configuration file on the cluster (under \filename{/etc/pythia-0.8}), for the benefit of all users and all Pyre-based applications. \subsection{Launchers and Schedulers} \label{sec:launchers:schedulers} If you have used one of the batch systems, you will know that the batch system requires you to write a script to launch a job. Fortunately, launching a parallel PyLith job is simplified by Pyre's \texttt{launcher} and \facility{scheduler} facilities. Many properties associated with \facility{launcher} and \facility{scheduler} are pertinent to the cluster you are on, and are best customized in a configuration file. Your personal PyLith configuration file (\filename{\$HOME/.pyre/pylithapp/pylithapp.cfg}) is suitable for this purpose. On a cluster, the ideal setup is to install a system-wide configuration file under \filename{/etc/pythia-0.8}, for the benefit of all users. Pyre's \facility{scheduler} facility is used to specify the type of batch system you are using (if any): \begin{cfg} [pylithapp] # The valid values for scheduler are 'lsf", 'pbs', 'globus', and 'none. scheduler = lsf # Pyre's launcher facility is used to specify the MPI implementation. # The valid values for launcher include 'mpich' and 'lam-mpi'. launcher = mpich \end{cfg} You may find the 'dry' option useful while debugging the \facility{launcher} and \facility{scheduler} configuration. This option causes PyLith to perform a ``dry run,'' dumping the batch script or mpirun command to the console, instead of actually submitting it for execution (the output is only meaningful if you're using a batch system). \begin{shell} # Display the bash script that would be submitted. $ pylith --scheduler.dry # Display the mpirun command. $ pylith --launcher.dry \end{shell} \subsection{Running without a Batch System} On a cluster without a batch system, you need to explicitly specify the machines on which the job will run. Supposing the machines on your cluster are named n001, n002, \ldots, etc., but you want to run the job on machines n001, n003, n004, and n005 (maybe n002 is down for the moment). To run an example, create a file named \filename{mymachines.cfg} which specifies the machines to use: \begin{cfg} [pylithapp.launcher]

nodegen

= n%03d

nodelist

= [1,3-5] \end{cfg} The \property{nodegen} property is a printf-style format string, used in conjunction with \property{nodelist} to generate the list of machine names. The \texttt{nodelist} property is a comma-separated list of machine names in square brackets. Now, invoke the following: \begin{shell} $ pylith example.cfg mymachines.cfg \end{shell} This strategy gives you the flexibility to create an assortment of \filename{cfg} files (with one \filename{cfg} file for each machine list) which can be easily paired with different parameter files. If your machine list does not change often, you may find it more convenient to specify default values for \property{nodegen} and \property{nodelist} in \filename{\$HOME/.pyre/pylithapp/pylithapp.cfg} (which is read automatically). Then, you can run any simulation with no additional arguments: \begin{shell} $ pylith example.cfg \end{shell} \warning{This assumes your machine list has enough nodes for the simulation in question.} You will notice that a machine file \filename{mpirun.nodes} is generated. It will contain a list of the nodes where PyLith has run. \subsection{Using a Batch System} Many clusters use some implementation of a PBS (e.g., TORQUE/Maui) or LSF batch system. The examples below illustrate use of some of the more important settings. You may need to make use of more options or adjust these to submit jobs on various cluster. These settings are usually placed in \filename{\$HOME/.pyre/pylithapp/pylithapp.cfg} or in a system-wide configuration file. They can be overridden on the command line, where one typically specifies the number of compute nodes and number of processes per compute node, the job name, and the allotted time for the job: \begin{shell} $ pylith example1.cfg \ --job.queue=debug \ --job.name=example1 \ --job.stdout=example1.log \ --job.stderr=example1.err \ --job.walltime=5*minute \ --nodes=4 \end{shell} \important{The value for nodes is equal to the number of compute nodes times the number of processes (usually the number of cores) requested per compute node. Specifying the number of processes per compute node depends on the batch system. For more information on configuring Pyre for your batch system, see CIG's Pythia page \url{geodynamics.org/cig/software/packages/cs/pythia}.} \subsubsection{LSF Batch System} \begin{cfg} [pylithapp] scheduler = lsf ; the type of batch system [pylithapp.lsf]

bsub-options

= [-a mpich_gm] ; special options for 'bsub' [pylithapp.launcher]

command

= mpirun.lsf ; 'mpirun' command to use on our cluster [pylithapp.job]

queue

= normal ; default queue for jobs \end{cfg} \subsubsection{PBS Batch System} \begin{cfg} [pylithapp] scheduler = pbs ; the type of batch system [pylithapp.pbs]

shell

= /bin/bash ; submit the job using a bash shell script # Export all environment variables to the batch job # Send email to when the job begins, ends, or aborts

qsub-options

= -V -m bea -M [pylithapp.launcher]

command

= mpirun -np ${nodes} -machinefile ${PBS_NODEFILE} \end{cfg} For most PBS batch systems you can specify N processes per compute node via the command line argument \commandline{-{}-scheduler.ppn=N}. \section{Getting Help and Reporting Bugs} \label{sec:help} The CIG Short-Term Crustal Dynamics Mailing List \url{} is dedicated to CIG issues associated with short-term crustal dynamics, including the use of PyLith. You can subscribe to the mailing list and view messages at cig-short Mailing List \url{geodynamics.org/cig/lists/cig-short}. CIG uses \object{GitHub} for source control and bug tracking. If you find a bug in PyLith, please submit a bug report to the GitHub issue tracking system for PyLith \url{https://github.com/geodynamics/pylith/issues}. Of course, it is helpful to first check to see if someone else already submitted a report related to the issue; one of the CIG developers may have posted a work around to the problem. You can reply to a current issue by clicking on the issue title. To submit a new issue, click on the \object{New Issue} button. % End of file 10-100 \usepackage{graphicx} \usepackage{balance} % for \balance command ON LAST PAGE (only there!) \usepackage{url} \usepackage{hyperref} \usepackage{microtype} \usepackage{calc} \usepackage[final]{listings} \usepackage[table]{xcolor} \usepackage{xcolor} \usepackage[utf8]{inputenc} \usepackage{upquote} \usepackage{fixme} \usepackage{xfrac} \usepackage{minted} \usepackage{amsmath,amssymb} \usepackage{siunitx} \usepackage{subfig} \usepackage{pmboxdraw} % Dark boxes for text histograms \usepackage{array,multirow} \usepackage{tikz} \usetikzlibrary{patterns} \makeatletter \global\let\tikz@ensure@dollar@catcode=\relax \makeatother \fxsetup{nomargin,inline} % Images in draft mode \setkeys{Gin}{draft=false} \newlength{\barlength} \newcommand{\singlebar}[1]{ \setlength{\barlength}{\dimexpr #1em / 2 \relax} \begin{tikzpicture} \draw[pattern=north east lines,pattern color=.] (0,0) rectangle (\barlength,0.5em); \end{tikzpicture} } \lstset{ % basicstyle=\ttfamily\small\color{black}, % breaklines=true, % keywordstyle=\textbf, % literate={█}{{\singlebar{1}}}{1} {██}{{\singlebar{2}}}{2} {███}{{\singlebar{3}}}{3} {████}{{\singlebar{4}}}{4} {█████}{{\singlebar{5}}}{5} {██████}{{\singlebar{6}}}{6} {███████}{{\singlebar{7}}}{7} {████████}{{\singlebar{8}}}{8} {█████████}{{\singlebar{9}}}{9} {██████████}{{\singlebar{10}}}{10} {███████████}{{\singlebar{11}}}{11} {████████████}{{\singlebar{12}}}{12} {█████████████}{{\singlebar{13}}}{13} {██████████████}{{\singlebar{14}}}{14} {███████████████}{{\singlebar{15}}}{15} {████████████████}{{\singlebar{16}}}{16} {█████████████████}{{\singlebar{17}}}{17} {██████████████████}{{\singlebar{18}}}{18} {███████████████████}{{\singlebar{19}}}{19} {████████████████████}{{\singlebar{20}}}{20} {█████████████████████}{{\singlebar{21}}}{21} {██████████████████████}{{\singlebar{22}}}{22} {███████████████████████}{{\singlebar{23}}}{23} {████████████████████████}{{\singlebar{24}}}{24} {█████████████████████████}{{\singlebar{25}}}{25} {▌}{{\singlebar{0.5}\hspace{\dimexpr0.5em-0.5em/2\relax}}}{1} {•}{{$\bullet$}}{1}, moredelim=[is][\color{red}\underbar]{\$}{\$}, % moredelim=[is][\color{red}]{/}{/}, % upquote=true, % keepspaces=true, % basewidth=0.5em % } \lstnewenvironment{lstnobreak}[1][] {\noindent\minipage{\linewidth}\medskip\lstset{#1}} {\endminipage} \hypersetup{breaklinks=true, bookmarks=true, pdfauthor={}, pdftitle={}, colorlinks=true, citecolor=blue, urlcolor=blue, linkcolor=magenta, pdfborder={0 0 0}} \DeclareMathOperator{\Var}{Var} \DeclareMathOperator{\Covar}{Covar} \newlength{\figurepadding} \setlength{\figurepadding}{0.5em} \newcommand{\paddedgraphics}[2][]{\includegraphics[#1]{#2}\vspace{\figurepadding}} \newcommand{\timing}[4][\second]{analysis time: \SI{#2}{#1}, training time: \SI{#3}{\second}, total runtime: \SI{#4}{#1}} \def\dBoost/{\emph{dBoost}} % Local Variables: % mode: latex % End: \section{Synthesis} \subsection{Resource Graph} A resource graph is a graph $G_R = (V_R, E_R), V_R = V_S \cup V_T$, where $V_S$ are the nodes from the sequence graph and $V_T$ are the different resource types. It is bipartite, $E_R = V_S \times V_T, (v_s, v_t) \in E_R$ means $v_s$ can be executed on resource type $v_t$. Additionally there is a cost function $c: V_T \to \mathbb{Z}$ and a execution time function $w: E_R \to \mathbb{Z}^{\geq 0}$ Furthermore $\alpha(v_t)$ denotes the number of available instances of resource type $v_t$, $\beta(v_s)$ denotes on which resource type $v_s$ is running, and $\gamma(v_s)$ denotes on which instance of this resource type it is running. \subsection{Scheduling} A schedule is a function $\tau: V_S \to \mathbb{Z}^{>0}$ that determines the starting times of operations. It is feasible if \begin{equation*} \forall (v_i, v_j) \in E_S . \tau(v_j) - \tau(v_i) \geq w(v_i) \end{equation*} where $w(v_i) = w(v_i, \beta(v_i))$ denotes the execution time of $v_i$. The latency $L$ is $L = \tau(v_n) - \tau(v_0)$, the differnce between the starting times. \subsubsection{ASAP} \begin{lstlisting}[escapeinside={(*}{*)}] ASAP((*$V_S$*), (*$E_S$*), w) { \tau((*$v_0$*)) = 1 do { (*$v_i \leftarrow \set{v \in V_S | \text{v's predecessors are planned}}$*); (*$\tau(v_i) = \max\set{\tau(v_j) + w(v_j) | (v_j, v_i) \in E_S}$*); } while (unplanned operations exist); } \end{lstlisting} \subsection{ALAP} \begin{lstlisting}[escapeinside={(*}{*)}] ALAP((*$V_S$*), (*$E_S$*), w, (*$L_{max}$*)) { (*$\tau(v_n)$*) = (*$L_{max}$*) + 1; do { (*$v_i \leftarrow \set{v \in V_S | \text{v's successors are planned}}$*); (*$\tau(v_i) = \min\set{\tau(v_j) | (v_i, v_j) \in E_S} - w(v_i)$*); } while (unplanned operations exist); } \end{lstlisting} \subsection{Bellman Ford} One might add additional constraints as weighted edges in the sequence graph, which can be resolved using Bellman Ford to find the single source longest path. Set the weight of the normal edges to their execution times. \begin{lstlisting}[escapeinside={(*}{*)}] BELLMAN-FORD((*$V_S$*), (*$E_S$*), W) { (*$\tau_0^0$*) = 0; for (int i = 0; i < (*$\abs{V_S}$*); i++) (*$\tau_i^0 = w(v_0, v_i)$*); for (int j = 0; i < (*$\abs{V_S}$*); i++) { for (int i = 0; i < (*$\abs{V_S}$*); i++) { (*$\tau_i^{j+1}$*) = (*$\max\set{\tau_i^j, \tau_k^j + w(v_k, v_i) | k \neq i}$*); } if ((*$\forall i \,.\, \tau_i^{j+1} = \tau_i^j$*)) return true; // success } return false; // failed } \end{lstlisting} \subsubsection{Example constraints} \begin{tikzpicture} [node distance=1.5cm] \tikzset{nop/.append style={minimum width=5mm}} \tikzset{inner/.append style={minimum width=5mm}} \node[nop] (0) {0}; \node[inner] (1) [below left of=0] {1}; \node[inner] (2) [below of=1] {2}; \node[inner] (3) [below right of=0] {3}; \node[inner] (4) [below of=3] {4}; \node[nop] (5) [below right of=2] {5}; \path (0) edge[dashed] node[above] {0} (1) (0) edge[dashed] node[above] {0} (3) (1) edge[->] node[right] {2} (2) (1) edge[->] node[above] {2} (4) (3) edge[->] node[right] {2} (4) (2) edge[dashed] node[above] {1} (5) (4) edge[dashed] node[above] {1} (5) ; \end{tikzpicture} We add the following constraints \begin{itemize} \item Between 1 and 2 there are at most 3 time units \item Between 0 and 4 there are at least 4 time units \end{itemize} \begin{tikzpicture} [node distance=1.5cm] \tikzset{nop/.append style={minimum width=5mm}} \tikzset{inner/.append style={minimum width=5mm}} \node[nop] (0) {0}; \node[inner] (1) [below left of=0] {1}; \node[inner] (2) [below of=1] {2}; \node[inner] (3) [below right of=0] {3}; \node[inner] (4) [below of=3] {4}; \node[nop] (5) [below right of=2] {5}; \path (0) edge[dashed] node[above] {0} (1) (0) edge[dashed] node[above] {0} (3) (1) edge[->] node[right] {2} (2) (1) edge[->] node[above] {2} (4) (3) edge[->] node[right] {2} (4) (2) edge[dashed] node[above] {1} (5) (4) edge[dashed] node[above] {1} (5) (2) edge[bend left, ->] node[left] {-3} (1) (0) edge[->] node[left] {4} (4) ; \end{tikzpicture} \subsection{List Scheduling} \begin{lstlisting}[escapeinside={(*}{*)}] LIST((*$V_S$*), (*$E_S$*), (*$V_R$*), (*$E_R$*), (*$\alpha$*), (*$\beta$*), priorities) { (*$V_T$*) = (*$V_R - V_S$*); t = 1; do { foreach((*$v_k \in V_T$*)) { (*$U_k$*) = candidates to be scheduled; (*$T_k$*) = running operations; (*$S_k$*) = subset of (*$U_k$*) with maximal priority and (*$\abs{S_k} + \abs{T_k} \leq \alpha(v_k)$*); foreach((*$v_i \in S_k$*)) { (*$\tau(v_i)$*) = t; } } t = t + 1; } while ((*$v_n$*) unplanned); } \end{lstlisting} \subsection{Integer Linear Programming} To get optimal results one can use ILP. First, for each $v_i \in V_S$ we have to determine $l_i$ and $h_i$, the earliest and latest starting time respectively, using ASAP and ALAP with a suitable $L_{max}$. $x_{i,t} = 1 \iff$ operation $v_i$ starts at time $t$. \begin{align*} \min \tau(v_n) - \tau(v_0) &\qquad \text{subject to} \\ \forall v_i \in V_S \forall l_i \leq t \leq h_i &\,.\,x_{i,t} \in \set{0, 1} \\ \forall v_i \in V_S &\,.\, \sum_{t = l_i}^{h_i} x_{i,t} = 1 \\ \forall v_i \in V_S &\,.\, \sum_{t = l_i}^{h_i} t \cdot x_{i,t} = \tau(v_i) \\ \forall (v_i, v_j) \in E_S &\,.\, \tau(v_j) - \tau(v_i) \geq w(v_i) \\ \forall v_k \in V_T \forall 1 \leq t \leq \max_i\set{h_i} &\,.\, \sum_{\forall i. (v_i, v_k) \in E_R} \sum_{p' = \max(0, t - h_i)}^{\min(w(v_i) - 1, t - l_i)} x_{i, t - p'} \leq \alpha(v_k) \end{align*} \subsubsection{Modifications} To adapt the ILP to iterative algorithms (marked graphs, pipelining), replace \begin{align*} \forall (v_i, v_j) \in E_S &\,.\, \tau(v_j) - \tau(v_i) \geq w(v_i) \\ \forall v_k \in V_T \forall 1 \leq t \leq \max_i\set{h_i} &\,.\, \sum_{\forall i. (v_i, v_k) \in E_R} \sum_{p' = \max(0, t - h_i)}^{\min(w(v_i) - 1, t - l_i)} x_{i, t - p'} \leq \alpha(v_k) \end{align*} by \begin{align*} \forall (v_i, v_j) \in E_S &\,.\, \tau(v_j) - \tau(v_i) \geq w(v_i) - d_{i,j} \cdot P\\ \forall v_k \in V_T \forall 1 \leq t \leq \max_i\set{h_i} &\,.\, \sum_{\forall i. (v_i, v_k) \in E_R} \sum_{p' = 0}^{w(v_i) - 1} \sum_{\forall p \,.\, l_i \leq t - p' + p \cdot P \leq h_i} x_{i, t - p' + p \cdot P} \leq \alpha(v_k) \end{align*} and where $d_{i,j}$ is the amount of tokens on edge $(i,j)$. \subsection{DVS ILP} \begin{align*} \min \sum_{k \in K} \sum_{v_i \in V_S} y_{ik} \cdot e_k(v_i) &\qquad\text{subject to} \\ \forall v_i \in V_S, k \in K &\,.\, y_{ik} \in \set{0,1} \\ \forall v_i \in V_S &\,.\, \sum_{k \in K} y_{ik} = 1 \\ \forall (v_i, v_j) \in E_S &\,.\, \tau(v_j) - \tau(v_i) \geq \sum_{k \in K} y_{ik} \cdot w_k(v_i) \\ \forall v_i \in V_S & \,.\, \tau(v_i) + \sum_{k \in K} y_{ik} \cdot w_k(v_i) \leq d(v_i) \end{align*} where $K$ is the set of voltage levels and there are no resource constraints. ergo-cms/theme-future-imperfect_partials/menu.tex * "Home":/ * "About":/about.html Manuscript/Quantum_Chaos/q_chaos.tex \documentclass[../thesis.tex]{subfiles} %!TeX spellcheck = en-GB \theoremstyle{definition} \newtheorem*{def*}{Definition} \begin{document} \chapter{Quantum Chaos} \label{chap:quantum-chaos} \section{Fundamental notions} In this section will briefly present the basic principles of quantum mechanics and introduce the concepts which are required for the characterisation of energy spectra. We begin with the concept of the physical state. A \emph{physical state} contains all the information that can be learned about the system. We associate to our physical system a \emph{Hilbert space}, denoted with \(\mathbb{H}\), which will contain all the possible states of the system. A vector in the Hilbert space will describe the state of the system (more rigorously only the direction of the vector will describe the state since by convention any two vectors that differ only by a constant describe the same physical state). In the Dirac formalism~\cite{Dirac1967} the vectors in the Hilbert space corresponding to physical states are represented by \emph{ket vectors}, denoted with \(\ket{\Psi}\). The elements of the dual of the Hilbert space are called \emph{bra vectors} and are denoted with \(\bra{\Psi}\). Due to the isomorphism between the Hilbert space and its dual, the bra vectors equally describe the physical state and the correspondence between the ket vectors and the bra vectors is given by dual conjugation. \subsubsection{The observables postulate} In quantum mechanics the properties of the physical system, called \emph{observables} are described by linear hermitian operators. As a result of a measurement, the obtained values of the observable are among the eigenvalues of the associated hermitian operator. \subsubsection{The measurement postulate} Let us consider a physical system in a state described by the state vector \(\ket{\Psi} \in \mathbb{H}\) and an observable described by the linear hermitian operator \(A\). If a measurement of the observable is performed, then following the measurement the system will jump in an uncontrollable manner in one of the eigenstates of the operator associated with the measured observable and the result of the measurement will be given by the corresponding eigenvalue. The probability of obtaining a given eigenvalue is given by the square of the absolute value of the scalar product between the final state associated with the given eigenvalue and the initial state. % {\color{red}discrete / continuous discussion?} \subsubsection{Fundamental commutation relations} In analogy with classical mechanics, a quantum version of the Poisson brackets can be defined, postulating that they have the same properties as the classical Poisson brackets, namely anti-symmetry, linearity, the product rule and the Jacobi identity. These properties uniquely define the form of the quantum Poisson brackets the commutator of two observables, \(\comm{A}{B} = AB - BA\), as being proportional to the quantum Poisson brackets \[ \comm{A}{B} = \ii \hbar {\{A,B\}}_{QM}. \] Thus we obtain the fundamental commutation relations in quantum mechanics \begin{align*} \comm{Q_i}{Q_j} &= 0 \\ \comm{P_i}{P_j} &= 0 \\ \comm{Q_i}{P_j} &= \ii \hbar \,\delta_{i,j}. \end{align*} \subsubsection{Time evolution postulate} In quantum mechanics time is just a parameter and not an observable. The time evolution can be postulated in different ways. If we consider the states as time dependent entities and observables as time independent ones, we can view the time evolution as a temporal displacement, similar to a spatial displacement. Then we can use the time evolution operator identity \[ \ii \hbar \pdv{t} U(t,t_0) = H U(t,t_0) \] to derive the Schrödinger equation: \[ \ii \hbar \pdv{t} \ket{\Psi} = H \ket{\Psi}. \] If we consider the states as time independent and the observables as time dependent, then the time evolution of the observables will be given by \[ \ii \hbar \dv{t} A(t) = [A(t), H]. \] We can observe the link between this equation and Hamilton's equations is given by the correspondence between the commutator in quantum mechanics and the Poisson bracket in classical mechanics, as mentioned in the previous chapter. The first approach is called the Schrödinger picture of quantum mechanics, while the second is called the Heisenberg picture. There is also another formulation, namely the interaction picture also named the Dirac picture. In the following we will only consider the Schrödinger picture. In the particular case when the Hamiltonian is time-independent, Schrödinger's equation reduces to \[ H \ket{\Psi} = E \ket{\Psi}, \] also called the time-independent Schrödinger equation and the time evolution of the state is given by \[ \ket{E(t)} = \ee^{-\frac{\ii}{\hbar} (t-t_0) H} \ket{E(t_0)} = \ee^{-\frac{\ii}{\hbar} (t-t_0) E} \ket{E}. \] Thus if the system is initially in an energy eigenstate, it will remain in the same state, having at most a phase modulation. Such states are called \emph{stationary states}. \section{From symmetry to degeneracy} Let us consider the above mentioned time-independent case. If we suppose that the energy spectrum is discrete, \[ H \ket{\Psi_i} = E_i \ket{\Psi_i}. \] If for an eigenvalue \(E_i\) repeats itself for different eigenstates \(\ket{\Psi_i}\) we have what is called a \emph{degeneracy}. In order to emphasise this, we can use a second index for the eigenvectors, \(j=1,\dotsc,m\), where \(m\) is the number of times the eigenvalue repeats. Thus for the \(i\)-th eigenvalue, \[ H \ket{\Psi_{i,j}} = E_i \ket{\Psi_{i,j}}. \] If the Hamiltonian is invariant to a set of unitary transformations, \(T^\dagger H T = H\), then this set forms a group since the transformation given by \(T T'\), with \(T'\) an arbitrary transformation from the set, will also leave the Hamiltonian invariant. Since symmetry operations form a group, the unitary operators which correspond to the symmetries of the Hamiltonian will form a group to which the Hamiltonian is invariant. Thus, \[ T H = T T^\dagger H T = T T^{-1} H T = H T, \] or \(\comm{T}{H} = 0\). We can once again observe how classical mechanics and quantum mechanics are linked through the Poisson bracket-commutator structure. In classical mechanics Noether's theorem asserts that to any symmetry of the action will correspond a constant of the motion. This will have a vanishing Poisson bracket with the Hamiltonian. For example, the invariance to rotations around a given axis will lead to angular momentum conservation. Since the operators commute, they share a common set of eigenvectors \[ H (T \ket{\Psi_i}) = H T \ket{\Psi_i} = T H \ket{\Psi_i} = T E_i \ket{\Psi_i} = E_i (T \ket{\Psi_i}). \] Thus if the Hamiltonian is invariant to the set of unitary transformations \( \{T_j\} \), with \(j=1,\dotsc,m\) \[ T_j H \ket{\Psi_i} = T E_i \ket{\Psi_i} = E_i (T \ket{\Psi_i}) = E_i \ket{\Psi_{i,j}} \] and we will not change the value of the eigenvalue \(E_i\) by applying \(T_j\), we only transform the eigenstates \(\ket{\Psi_i}\) to a linear combination \(\ket{\Psi_{i,j}}\). \section{Level repulsion} As is the case with classical mechanics there are few situations in which the analytic solution is known. In order to find approximate solutions we can use perturbation theory if we can assume that the Hamiltonian can be viewed as to have a part for which the solutions are known and another part that can be considered a small perturbation. The following concepts are mainly following the notations from~\cite{Sakurai2011} and the lecture notes~\cite{Baran,Zus}. In the framework of time-independent perturbation theory, we want to obtain an approximate solution to the problem \begin{equation} H \ket{\Psi^{[j]}} = E_j \ket{\Psi^{[j]}} \label{eq:q-pert-th-ex-pr} \end{equation} and we consider that the Hamiltonian can be written as \(H = H_0 + \lambda V\), where the solution to the unperturbed problem \[ H_0 \ket{\Psi^0_{n\alpha}} = E_n^0 \ket{\Psi^0_{n\alpha}} \] is known and \( \alpha \) gives the degeneracy of the \(n\)-th level. Since the unperturbed eigenvectors form a basis in the Hilbert space, we can expand the perturbed eigenvectors in that basis. \[ \ket{\Psi^{[j]}} = \sum_{m,\beta} c_{m,\beta}^{[j]} \ket{\Psi_{m,\beta}^0}. \] Inserting into eq.~\eqref{eq:q-pert-th-ex-pr} we obtain \[ (H_0 + \lambda V) \sum_{m,\beta} c_{m,\beta}^{[j]} \ket{\Psi_{m,\beta}^0} = \sum_{m,\beta} E_j\, c_{m,\beta}^{[j]} \ket{\Psi_{m,\beta}^0}. \] Using the solution to the unperturbed problem the above equation becomes \[ \sum_{m,\beta} (E_m^0 + \lambda V) c_{m,\beta}^{[j]} \ket{\Psi_{m,\beta}^0} = \sum_{m,\beta} E_j\, c_{m,\beta}^{[j]} \ket{\Psi_{m,\beta}^0}. \] To simplify the discussion we will now consider what happens with two unperturbed states \(\ket{\Psi_1^0}\) with the energy \(E_1^0\) and \(\ket{\Psi_2^0}\) with the energy \(E_2^0\) when \(\lambda=1\). \[ \sum_{m,\beta} (E_m^0 + V) c_{m}^{[j]} \ket{\Psi_{m}^0} = \sum_{m=1,2} E_j\, c_{m}^{[j]} \ket{\Psi_{m}^0} .\] By taking the scalar product with the state \(\bra{\Psi_{k}^0}\) and using the orthogonality relation \(\braket{\Psi_{k}^0}{\Psi_{m}^0} = \delta_{k,m}\) we obtain \[ E_k^0\, c_{k}^{[j]} + \sum_{m=1,2} \bra{\Psi_{k}^0} V \ket{\Psi_{m}^0} c_{m}^{[j]} = E_j\, c_{k}^{[j]}, \] or \[ c_{k}^{[j]} \left( E_k^0 - E_j + V_{k,k} \right) + \sum_{m \neq k} V_{k,m}\, c_{m}^{[j]} = 0, \] where \(\bra{\Psi_{k}^0} V \ket{\Psi_{m}^0} \equiv V_{k,m}\). The above equation becomes \begin{align*} (E_1^0 - E_j + V_{1,1})\, c_1^{[j]} + V_{1,2}\, c_2^{[j]} = 0, \text{ for } k=1 \\ V_{2,1}\, c_1^{[j]} + (E_2 - E_j + V_{2,2})\, c_2^{[j]} = 0, \text{ for } k=2. \end{align*} We use the following notations: \[ H_{1,1} \equiv E_1^0 + V_{1,1}, H_{2,2} \equiv E_2^0 + V_{2,2}, H_{1,2} \equiv V_{1,2}, H_{2,1} \equiv V_{2,1}, \delta \equiv H_{1,1} - H_{2,2}, \tan{\beta} \equiv \frac{2\abs{H_{1,2}}}{\delta}. \] The above system of equations has non-trivial solutions if the determinant vanishes. \[ (H_{1,1} - E_j)(H_{2,2} - E_j) - H_{1,2} H_{2,1} = 0 \] or \[ E_j^2 - (H_{1,1} + H_{2,2}) E_j + H_{1,1} H_{2,2} - H_{1,2} H_{2,1} = 0. \] This equation has the solutions \begin{align} E_j &= \frac{(H_{1,1} + H_{2,2}) \pm \sqrt{ {(H_{1,1} + H_{2,2})}^2 - 4(H_{1,1} H_{2,2} - H_{1,2} H_{2,1})}}{2} \\ &= \frac{H_{1,1} + H_{2,2}}{2} \pm \frac{1}{2} \sqrt{\delta^2 + 4 \abs{H_{1,2}}^2}. \label{eq:lvl-repulsion-e} \end{align} Since \((H_{1,1} - E_j)\, c_1^{[j]} + H_{1,2}\, c_2^{[j]} = 0\), \[ \frac{c_1^{[j]}}{c_2^{[j]}} = \frac{H_{1,2}}{E_j - H_{1,1}}. \] If we rewrite eq.~\eqref{eq:lvl-repulsion-e} as \[ E_j = \frac{1}{2} \left[ H_{1,1} + H_{2,2} \mp (H_{2,2} - H_{1,1}) \sqrt{1 + \frac{4\abs{H_{1,2}}^2}{\delta^2}}\right] \] we obtain \[ \frac{c_1^{[j]}}{c_2^{[j]}} = \frac{2 H_{1,2}}{H_{2,2} - H_{1,1}} \left[ 1 \mp \sqrt{1 + \frac{4\abs{H_{1,2}}^2}{\delta^2}} \right]^{-1} = -\tan{\beta}\left(1 \mp \sqrt{1+\tan^2\beta}\right). \] This ratio can be also expressed as \begin{align*} \frac{c_1^{[j]}}{c_2^{[j]}} &= \frac{-\tan{\beta}}{1 \mp \frac{1}{\cos{\beta}}} = \frac{-\sin{\beta}}{\cos{\beta} \mp 1} \\ &= \frac{-2\sin{\frac{\beta}{2}} \cos{\frac{\beta}{2}}} {\cos^2{\frac{\beta}{2}} - \sin^2{\frac{\beta}{2}} \mp \left(\cos^2{\frac{\beta}{2}} + \sin^2{\frac{\beta}{2}}\right)} = \begin{cases} \cot{\frac{\beta}{2}} \\ -\tan{\frac{\beta}{2}} \end{cases}. \end{align*} Thus, \begin{align*} \ket{\Psi^{[1]}} &= \cos{\frac{\beta}{2}} \ket{\Psi_1^0} + \sin{\frac{\beta}{2}} \ket{\Psi_2^0} \\ \ket{\Psi^{[2]}} &= -\sin{\frac{\beta}{2}} \ket{\Psi_1^0} + \cos{\frac{\beta}{2}} \ket{\Psi_2^0}. \end{align*} If the matrix elements of the interaction which mix distinct states are relatively small, that is \(\abs{H_{1,2}} \ll \delta \), \(\beta \simeq 0\) \[ E_{1,2} = \frac{H_{1,1} + H_{2,2}}{2} \pm \frac{\delta}{2} \sqrt{1 + \frac{4 \abs{H_{1,2}}^2}{\delta^2}} \] Thus \[ E_1 \simeq H_{1,1} + \frac{\abs{H_{1,2}}^2}{\delta} \] and \[ E_2 \simeq H_{2,2} - \frac{\abs{H_{1,2}}^2}{\delta}. \] and the perturbed energy levels are close to the unperturbed ones. The states will also be approximatively the unperturbed ones \begin{align*} \ket{\Psi^{[1]}} &\simeq \ket{\Psi_1^0} \\ \ket{\Psi^{[2]}} &\simeq \ket{\Psi_2^0}. \end{align*} If on the other hand, the matrix elements levels which mix distinct states are relatively strong, that is \(\abs{H_{1,2}} \gg \delta \), \(\beta \simeq \frac{\pi}{2}\) \[ E_{1,2} = \frac{H_{1,1} + H_{2,2}}{2} \pm \sqrt{\frac{\delta^2}{4} + \abs{H_{1,2}}^2} \simeq \frac{H_{1,1} + H_{2,2}}{2} \pm \left(\abs{H_{1,2}} + \frac{\delta^2}{8\abs{H_{1,2}}}\right) \] and \begin{align*} \ket{\Psi^{[1]}} &\simeq \frac{\sqrt{2}}{2} \ket{\Psi_1^0} + \frac{\sqrt{2}}{2} \ket{\Psi_2^0} \\ \ket{\Psi^{[2]}} &\simeq -\frac{\sqrt{2}}{2} \ket{\Psi_1^0} + \frac{\sqrt{2}}{2} \ket{\Psi_2^0}. \end{align*} In this case we observe that if \(\delta \simeq 0\), then \(E_1 - E_2 \simeq 2\abs{H_{1,2}}\). If the unperturbed energy levels were degenerated, then the perturbed energy levels will not remain so. The removal of the degeneracy is called \emph{level repulsion}. We can illustrate this phenomena by considering an arbitrary \(2 \cross 2\) hermitian matrix and computing its eigenvalues. \[ H = \begin{pmatrix} H_{1,1} & H_{1,2} \\ H_{1,2}^* & H_{2,2} \end{pmatrix} \] The eigenvalues will be given by eq.~\eqref{eq:lvl-repulsion-e}. The first case, \(\abs{H_{1,2}} \ll \delta \), can be viewed as the case when the off-diagonal elements are small and the matrix can be approximated with a diagonal matrix. In this case we expect that the energy levels will be close to the diagonal levels or the unperturbed levels. The perturbed eigenstates will also be approximatively equal with the unperturbed eigenstates. The level repulsion emphasised by the second case, \(\abs{H_{1,2}} \gg \delta \), corresponds to the case when the off-diagonal elements are significant. When we expand the perturbed eigenstates in the basis given by the unperturbed eigenstates we will have significant components from each element in the basis and the new states will be a mixture of the unperturbed states. We can say that the energy levels are no longer independent. In the case when the unperturbed energy levels are degenerated, the difference between the perturbed energy levels is given by \(2 \abs{H_{1,2}}\), which is a measure of the mixing of the unperturbed states. \section{Probability notions} \begin{def*}[Joint probability] Given two events $A$ and $B$, their joint probability, \(P(A \cap B)\), is the probability of the two events to occur simultaneously. \end{def*} \begin{def*}[Conditional probability] The conditional probability of an event $A$ given an event $B$ with \(P(B)>0\), denoted \(P(A\,|\,B)\) is given by \[ P(A\,|\,B) = \frac{P(A \cap B)}{P(B)}. \] \end{def*} The joint probability of $A$ and $B$ can be expressed as \(P(A \cap B) = P(A\,|\,B) P(B)\). \begin{def*}[Continuous random variable] A continuous random variable is a function from the set of all outcomes to the set of real numbers \(X:\Omega \to \mathbb(R)\) such that \[ P(a \leq X(\omega) \leq b) = \int_a^b f(x) \dd{x}, \] where \(f(x) \geq 0\) and \(\int_{-\infty}^{+\infty}f(x)\dd{x}=1\). \end{def*} The above function \(f\) is called the \emph{probability density function}. The probability for a continuous random variable $X$ to take a value in the infinitesimal interval of length \(\dd{x}\) is given by \begin{equation} \label{eq:prob-rv-in-int} P(x \in [x, x + \dd{x}]) = \int_x^{x+\dd{x}} f(z) \dd{z} \approx f(x)\dd{x} \end{equation} \section{Nearest neighbour distributions} % \emph{Random matrices} are matrices which have random variables as elements, % with their randomness is restricted by the symmetries of the whole matrix. Nearest neighbour spacing distributions show how the differences between consecutive energy levels fluctuate around the average. In order to better understand this concept we shall begin with the simpler case of real random numbers as presented in~\cite{Timberlake2006} and then continue with a general case as in~\cite{Brody1981}. \subsection{The nearest neighbour spacing distribution of random numbers} We consider a sequence of uniformly distributed, ordered, real, random numbers. We define the \emph{spacing} of an ordered sequence as the sequence of differences between consecutive elements. For an interval of length \(s, s>0\), we will denote with \( P(n \in s) \) the probability for the interval to contain $n$ numbers and with \( P(n \in \dd{s} |\; m \in s) \) the conditional probability for the interval of length \( \dd{s} \) to contain $n$ numbers given that the interval of length $s$ contains $m$ numbers. If $E$ is a given number in the sequence, we are interested in the probability \( P(s)\dd{s} \) to have the next number between \( E+s \) and \( E+s+\dd{s} \). Since we are interested in the next number after $E$, we know that in the interval of length $s$ there is no other number and the next number is somewhere in the infinitesimal interval \(\dd{s}\). Thus the joint probability of the events \(1 \in \dd{s}\) and \(0 \in s\) is given by: \begin{equation} \label{eq:jpr-next} P(s)\dd{s} = P(1 \in \dd{s} |\; 0 \in s) P(0 \in s). \end{equation} Since random numbers are not correlated, the probability of a random number to be found in the interval \( \dd{s} \) does not depend on the number of random numbers in $s$, so \[ P(1 \in \dd{s} |\; 0 \in s) = P(1 \in \dd{s}). \] The random numbers are uniformly distributed, so their probability density function \(f\) is a constant. Hence, according to eq.~\eqref{eq:prob-rv-in-int}, the probability of finding a number in the interval of length \( \dd{s} \) is given by \[ P(1 \in \dd{s}) \equiv P(1 \in [0, 0+\dd{s}]) \approx f(s) \dd{s} \sim \dd{s}. \] If we denote the constant probability density function with $a$ \[ P(s)\dd{s} = a \dd{s} P(0 \in s). \] \( P(0 \in s) \) can be expressed using the complementary probability as \( {1 - \int_0^s P(s') \dd{s'}} \). Now we can express \( P(s)\dd{s} \) as follows: \[ P(s)\dd{s} = a \dd{s} \left( 1 - \int_0^s P(s') \dd{s'} \right). \] In order to differentiate with respect to $s$, we will use the Leibniz rule for differentiating integrals, namely \begin{equation} \label{eq:leibnitz} \dv{x} \int\limits_{G(x)}^{H(x)} F(x, t) \dd{t} = \int\limits_{G(x)}^{H(x)} \pdv{F}{x} \dd{t} + F(x, H(x))\, \dv{H}{x} - F(x, G(x))\, \dv{G}{x} \end{equation} Using this rule, we obtain \[ \dv{s}P(s) = -aP(s). \] This differential equation can be solved by separation of variables, yielding \[ P(s) = \mathcal{C} \ee^{-as} \] In order to determine the constant \(\mathcal{C}\), we use the normalisation condition for the probability density function \[ \int_{-\infty}^{\infty} P(s) \dd{s} = 1 \] Since \(s>0\), this reduces to \[ \int_{0}^{\infty} P(s) \dd{s} = \int_{0}^{\infty} \mathcal{C} \ee^{-as} \dd{s} = -\frac{\mathcal{C}}{a} \eval{\ee^{-as}}_0^{\infty} = \frac{\mathcal{C}}{a} \] Thus \(\mathcal{C} = a\) and \(P(s) = a \ee^{-as}\). We can further simplify the formula if we set that the average spacing to unity. The average spacing is given by \[ \mean{s} = \int_{0}^{\infty} s P(s) \dd{s} = -\int_{0}^{\infty} s \dv{s} \left( \ee^{-as} \right) \dd{s} = \int_{0}^{\infty} \ee^{-as} \dd{s} = \frac{1}{a}, \] so setting it to unity results in \(a=1\). Thus the probability density function becomes \begin{equation} \label{eq:poisson-dist} P(s) = \ee^{-s}. \end{equation} This function is known as the \emph{Poisson distribution}. \subsection{The Wigner distribution} We will now consider a more complicated situation, by considering the probability density function for the sequence of random numbers to be arbitrary. We can start from eq.~\eqref{eq:jpr-next} since the discussion up to that point did not include any details related to the distribution of the random numbers. In this case \(P(1 \in \dd{s} |\; 0 \in s) = f_{1,0}(s) \dd{s}\), where \(f_{n,m}(s)\) is function which describes how the probability of having $n$ numbers in \(\dd{s}\) is influenced by the $m$ numbers in $s$. Thus \[ P(s)\dd{s}=f_{1,0}(s)\dd{s} \left( 1 - \int_0^s P(s') \dd{s'} \right) \] By solving % {\color{red}\large????} this integral equation, we obtain the solution \[ P(s) = \mathcal{C} f_{1,0}(s) \exp(-\int_0^s f_{1,0}(x) \dd{x}) \] We observe that if we take the probability density function constant, \(f_{1,0}(s) = \frac{1}{a}\), we obtain the above case of the Poisson distribution. For a % {\color{red}linear (why?)} probability density function, \(f_{1,0}(s) = \alpha s\) we obtain \[ P(s) = \mathcal{C} \alpha s \exp(-\alpha \frac{s^2}{2}) \] From the normalisation condition we obtain \[ \int_{0}^{\infty} P(s) \dd{s} = \mathcal{C} = 1 \] The average spacing is given by \[ \mean{s} = \int_{0}^{\infty} s P(s) \dd{s} = \alpha \int_{0}^{\infty} s^2 \exp(-\alpha \frac{s^2}{2}) \dd{s} = \frac{1}{\sqrt{\alpha}} \sqrt{\frac{\pi}{2}}. \] If we set the average spacing to unity, we obtain \(\alpha = \frac{\pi}{2}\) and \[ P(s) = \frac{\pi}{2} s \exp(-\frac{\pi}{4} s^2) \] This function is known as the \emph{Wigner distribution}. \section{From Classical Chaos to Quantum Chaos} The classical concept of sensitivity to initial conditions loses its meaning in the quantum realm since the trajectory cannot be defined due to Heisenberg's uncertainty principle. However, there are some other ways in which we can link classically chaotic dynamics to quantum features. These bridges between classical mechanics and quantum mechanics allows us to give a meaning to quantum chaos~\cite{Berry1989}. % In the terms of classical mechanics, such as the sensitivity to initial conditions % quantum chaos does not exist~\cite{Berry1989} because of the Heisenberg's uncertainty principle, % the absence of trajectories and many other reasons. In turn, there are some other % ways in which we can link classically chaotic systems to their % quantum counterparts. These bridges between classical mechanics and quantum % mechanics allows us to give a meaning to quantum chaos. % {\color{red} (Berry $\to$ quantum chaology)} Specifically, there are two important conjectures that allow us to connect classical systems and quantum systems as mentioned above. \subsubsection{The Berry-Tabor conjecture}% {\color{red}(or theorem?)}} This conjecture states that the quantum counterpart of a classically integrable system has a Poissonian nearest neighbour distribution. \subsubsection{The Bohigas-Gianoni-Schmit conjecture} This conjecture states that the nearest neighbour distribution of a quantum system with a classically chaotic counterpart is given by the Wigner distribution. \end{document} @article{Ribalet2015, abstract = {Theoretical studies predict that competition for limited resources reduces biodiversity to the point of ecological instability, whereas strong predator/prey interactions enhance the number of coexisting species and limit fluctuations in abundances. In open ocean ecosystems, competition for low availability of essential nutrients results in relatively few abundant microbial species. The remarkable stability in overall cell abundance of the dominant photosynthetic cyanobacterium Prochlorococcus is assumed to reflect a simple food web structure strongly controlled by grazers and/or viruses. This hypothesized link between stability and ecological interactions, however, has been difficult to test with open ocean microbes because sampling methods commonly have poor temporal and spatial resolution. Here we use continuous techniques on two different winter-time cruises to show that Prochlorococcus cell production and mortality rates are tightly synchronized to the day/night cycle across the subtropical Pacific Ocean. In warmer waters, we observed harmonic oscillations in cell production and mortality rates, with a peak in mortality rate consistently occurring ∼6 h after the peak in cell production. Essentially no cell mortality was observed during daylight. Our results are best explained as a synchronized two-component trophic interaction with the per-capita rates of Prochlorococcus consumption driven either directly by the day/night cycle or indirectly by Prochlorococcus cell production. Light-driven synchrony of food web dynamics in which most of the newly produced Prochlorococcus cells are consumed each night likely enforces ecosystem stability across vast expanses of the open ocean.}, author = { and Swalwell, Jarred and Clayton, Sophie and Jiménez, Valeria and Sudek, Sebastian and Lin, Yajuan and Johnson, . and Worden, . and }, doi = {10.1073/pnas.1424279112}, file = {:Users/ribalet/Downloads/pnas.1424279112.sapp.pdf:pdf;:Users/ribalet/Library/Application Support/Mendeley Desktop/Downloaded/Ribalet et al. - 2015 - Light-driven synchrony of Prochlorococcus growth and mortality in the subtropical Pacific gyre(2).pdf:pdf}, issn = {0027-8424}, journal = {Proceedings of the National Academy of Sciences}, keywords = {Cell division,Cyanobacteria,Flow cytometry,Mortality,SeaFlow}, month = {jun}, number = {26}, pages = {8008--8012}, title = {Light-driven synchrony of Prochlorococcus growth and mortality in the subtropical Pacific gyre}, url = {http://www.pnas.org/lookup/doi/10.1073/pnas.1424279112}, volume = {112}, year = {2015} } 6DL/jianqingCZNN.tex \subsection{Jinchao's summary} @article{liu2017microgrid, title={Microgrid optimal scheduling with chance-constrained islanding capability}, author={ and }, journal={Electric Power Systems Research}, volume={145}, pages={197--206}, year={2017}, publisher={Elsevier} } \hypertarget{interfacesirius_1_1sirius__get__energy__vha}{}\section{sirius\+:\+:sirius\+\_\+get\+\_\+energy\+\_\+vha Interface Reference} \label{interfacesirius_1_1sirius__get__energy__vha}\index{sirius\+::sirius\+\_\+get\+\_\+energy\+\_\+vha@{sirius\+::sirius\+\_\+get\+\_\+energy\+\_\+vha}} \subsection*{Public Member Functions} \begin{DoxyCompactItemize} \item \hypertarget{interfacesirius_1_1sirius__get__energy__vha_afe83bb90f0174d36dba634f60d54aa74}{}subroutine {\bfseries sirius\+\_\+get\+\_\+energy\+\_\+vha} (val)\label{interfacesirius_1_1sirius__get__energy__vha_afe83bb90f0174d36dba634f60d54aa74} \end{DoxyCompactItemize} \subsection{Detailed Description} Definition at line 729 of file sirius.\+f90. The documentation for this interface was generated from the following file\+:\begin{DoxyCompactItemize} \item sirius.\+f90\end{DoxyCompactItemize} \chapter{Použité prostředky a~hardware}\label{kap:prostredky} \section{Interpolátor} Jednotku interpolátoru jsem se rozhodl postavit na STM32F4 Dicovery kitu. Tento vývojový kit je postaven na STM32F407VGT6 ARM mikrokontroléru\cite{discovery}. Kit obsahuje integrovaný debugger. Deska je osazena i dalšími zajímavými komponentami (akcelerometrem, audio DAC, či mikrofonem), ty však nejsou pro můj projekt využité. Tento vývojový kit jsem se rozhodl použít pro jeho výkonný mikrokontrolér s~bohatými perifériemi. Jádro mikrokontroléru je schopné běžet až na 168~MHz, což jej společně s~integrovanou FPU (floating-point unit) a~192~kB RAM předurčuje pro aplikace vyžadující vysoký výkon. Tento kit také disponuje osazeným PHY pro USB FS aplikace využívající integrovanou USB-OTG periférii na mikrokontroléru, což byl jeden z~mých požadavků. Jako poslední rozhodující faktor pro tuto desku byla její příznivá cena -- 329~Kč\footnote{15. 2. 2013 ve~Farnell electronics \url{http://cz.farnell.com/stmicroelectronics/stm32f4discovery/stm32f407-usb-otg-discovery-kit/dp/2009276}}. Pro vývoj aplikací na kitu bylo použito IDE Atollic TrueSTUDIO Lite 2.2.0. (Nejedná se o~nejnovější verzi tohoto IDE -- při započetí vývoje byla nejnovější, avšak na novou verzi jsem neupgradoval, protože v~ní přibylo omezení velikosti výsledného kódu). Toto IDE je přímo navrženo pro spolupráci s~tímto vývojovým kitem. Ve verzi Lite (2.2.0), která je dostupná zdarma, podporuje pouze překladač jazyka C (nikoliv C++) a~omezuje maximální počet breakpointů při debuggování na dva, avšak bez omezení velikosti výsledného kódu. Celá aplikace pro interpolátor byla napsána v~jazyce~C s~využítím StdPeriph knihovny od~STMicroelectronics pro obsluhu periferií mikrokontroléru a~knihovny USB-Host-Device pro obsluhu USB-OTG periférie. \section{Aplikace pro počítač} Aplikaci pro počítač jsem se rozhodl vyvíjet pro Windows. Na základě tohoto rozhodnutí jsem zvolil použité technologie. Aplikaci jsem vyvíjel v~jazyce C++ s~použitím frameworku WxWidgets\footnote{\url{http://www.wxwidgets.org/}} a~knihovny {WinUSB}\footnote{\url{http://msdn.microsoft.com/en-us/library/windows/hardware/ff540196(v=vs.85).aspx}} pro komunikaci s~interpolátorem. Jazyk C++ jsem použil, jelikož s~ním mám největší zkušenosti, navíc se v~něm velice dobře implementoval relativně nízkoúrovňový protokol komunikace s~interpolátorem (popsaný v~kapitole \ref{kap:protokol}). Knihovnu WxWidgets jsem použil, abych zachoval \uv{look'n'feel} mé platformy (Windows). Jelikož mým primárním cílem nebylo vyvíjet multiplatformní aplikaci, zvolil jsem knihovnu WinUSB před knihovnou LibUSB díky její nativní podpoře v~systému Windows (od~verze XP SP2\cite{winusb}). \section{Testovací stroj} Aby bylo možné zařízení odzkoušet, rozhodl jsem si postavit jednoduchý \uv{plotter}. Tento plotter (obrázek \ref{fot:plotter} a~\ref{fot:plotter2}) je postaven na dřevotřískové desce. Jako vedení slouží broušené tyče ze~staré inkoustové tískárny, odkud byla použita i bronzová kluzná pouzdra na tyto tyče. K~pohonu slouží dva bipolární krokové motory opět ze staré inkoustové tiskárny. Tyto motory pohybují strojem za pomoci ozubených řemenů. S~návrhem a~výrobou tohoto plotteru mi pomáhal můj otec, za což mu děkuji. \begin{figure}[h] \centering \includegraphics[width=0.8\textwidth]{img/plotter.jpg} \caption{Fotografie plotteru, na kterém byl systém testován}\label{fot:plotter} \end{figure} \begin{figure}[h] \centering \includegraphics[width=0.8\textwidth]{img/plotter2.jpg} \caption{Fotografie plotteru, na kterém byl systém testován}\label{fot:plotter2} \end{figure} Motory jsou řízeny čínskými \uv{no-name} jednoosými drivery postavenými na integrovaném obvodu Toshiba TB6560\cite{TB6560}. Mezi uživateli C-N-C.cz fóra se běžně označují jako \uv{čínská zelená jednoosá deska}. Existuje více mutací tohoto driveru, pro jeho přesnou identifikaci přikládám fotografii (obrázek \ref{fot:tb6560}). Tyto drivery jsou řízeny step-dir signálem. \begin{figure}[h] \centering \includegraphics[width=0.8\textwidth]{img/tb6560.jpg} \caption{Fotografie použitých driverů krokových motorů}\label{fot:tb6560} \end{figure} \chapter{Software pro interpolátor} Software pro interpolátor je primárně závislý na příchozí USB komunikaci a~interpolaci. Veškerou sérii úkonů jsem se rozhodl rozdělit do jednolivých komponent. Komponenty nejsou reprezentovány objekty, jelikož software je napsán v~jazyce C. Jelikož existuje pouze jediná instance od~každé komponenty, rozhodl jsem se každou komponentu reprezentovat jako soubor globálních funkcí pracujících nad svými globálními proměnnými. Rozvržení a~provázání jednotlivých komponent je znázorněno na schématu \ref{nak:rozvrzeni}. \begin{figure}[h] \centering \includegraphics[width=1\textwidth]{img/rozvrzeni2.pdf} \caption{Schéma znázorňující rozvržení softwaru pro interpolátor}\label{nak:rozvrzeni} \end{figure} Základní komponentou celé aplikace je komponenta {\tt Receive}. Tato komponenta zachytává přícházející pakety z~USB, skládá je dohromady a~nasledně tato přijatá data předává dále příslušným komponentám. Komponenta {\tt Config} příjímá zprávy týkající se nastavení a~na základě přijatých dat volá funkce pro nastavení příšlušných komponent. Komponenta {\tt State} uchovává informace o současném stavu interpolátoru a~podává zpětnou vazbu počítači. Obdobně komponenta {\tt ErrState} uchovává informace o chybovém stavu jednotlivých komponent a~podává zpětnou vazbu. Jednou z~nejdůležitějších komponent je {\tt CommandStack}. Tato komponenta přebírá veškeré příkazy pro stroj a~dle příznaku je buď řadí do fronty a~postupně vykonává nebo je vykoná přímo hned po přijetí. Příkazy pohybu zpracovává komponenta Axis, která provádí samotnou interpolaci. Zbývající dvě komponenty ({\tt GPIO} a~{\tt Global}) zpracovávají ostatní druhy příkazů -- např. prodlevu v~programu a~podobně. \section{Implementace USB a~komunikační protokol} Pro komunikaci mezi počítačem a~interpolátorem používám integrovanou periférii v~mikrokontroléru. Pro její obsluhu jsem použil knihovnu dodávanou výrobcem -- STMicroelectronics -- USB Host-DeviceLibrary. Jelikož počítač umí v~USB komunikaci vystupovat pouze jako host, nakonfiguroval jsem kit jako device. Mým cílem bylo vytvořit vendor-defined device, které bude kromě control enpointu pro komunikaci využívat jeden vstupní a~druhý výstupní bulk endpoint -- tedy bude tvořit sériovou linku. Kit s~počítačem komunikuje rychlostí Full Speed (pro rychlost High Speed není na kitu příslušné PHY). Délku paketu jsem použil nejdelší možnou pro Full Speed komunikaci -- 64 bytů. Jelikož jsem měl problémy s nastavením projektu tak, aby se knihovna pro obsluhu USB periférie zkompilovala, vyšel jsem pro mou aplikaci z~projektu vytvořeného v~článku \uv{Začínáme s~STM32F4Discovery 7}\cite{mcu} od~uživatele s~přezdívkou Mard na serveru mcu.cz. Vstupní projekt jsem upravil tak, aby se zařízení tvářilo jako vendor-defined. O této úpravě a~mém prvním sestrojeném demu USB komunikace byl napsán článek na témže serveru\footnote{\url{http://mcu.cz/comment-n2904.html}}. \subsection{Komunikační protokol a~komponenta Receive}\label{kap:protokol} Jak vyplývá z~hiearchie jednotlivých komponent, komunikační protokol je vrstvený. V~nejvyšší vrstvě, kterou zpracovává komponenta {\tt Receive}, je ošetřeno rozdělení dat do více paketů (v~případě, že celková délka dat přesáhne 64 bytů). Veškerá komunikace probíhá binárně -- jednak abych ušetřil přenášená data, tak také, aby zpracování dat v~interpolátoru proběhlo co nejrychleji. Pokud budu čísla posílat v~binárním tvaru (což si můžu dovolit, jelikož platforma x86, resp. x64, je little-endian a~platforma ARM umí pracovat v~obou režimech -- little i big-endian\cite{wiki:end}), tak mi stačí pouhé ukládání dat a~nemusím parsovat textový formát. První byte prvního přijatého paketu obsahuje informaci o počtu paketů, do kterých je rozdělena právě příjímaná zpráva. Za tímto bytem následuje další byte, který definuje, které komponentě je zpráva určena. Seznam těchto kódů je definován v~hlavičkovém souboru {\tt CommunicationEnumerations.h}. Všechny pakety, které jsou přijaty, jsou předány funkci ProcessPacket. Tato funkce při přijetí prvního paketu vyčte délku zprávy $n$ v~paketech, nakopíruje ji do bufferu a~dalších $n-1$ přijatých paketů nakopíruje do bufferu za sebou. Jakmile je přijato všech $n$ paketů, zavolá funkci DistributeData, která předá přijatou zprávu dalším komponentám (na~základě druhého bytu). Každá komponenta definuje svou funkci ProcessMessage (tedy např. \texttt{CommandStackProcessMessage}), která zajistí zpracování zprávy dané komponenty. Tvar zprávy pro dané komponenty popisuji v~následujících sekcích. Pro komunikaci s~počítačem (pro zpětnou vazbu) je použit stejný princip na rozdělení zprávy do paketů. Druhý byte zprávy však nyní neurčuje danou komponentu (tato hiearchie na straně počítače neexistuje), nýbrž identifikátor zprávy. Všechny možné zprávy pro počítač jsou definovány v~hlavičkovém souboru {\tt MessageCodeToPC.h}. \section{Komponenta CommandStack} Komponenta {\tt CommandStack} má za úkol spouštět přijaté příkazy. Název této komponenty jsem bohužel zvolil jako trochu zavádějící. CommandStack reprezentuje datový typ fronty -- příkaz přijatý jako první se vykoná jako první. Příkazy jsou dvojího druhu -- jedny jsou určeny pro okamžité vykonání po jejich přijetí a~druhé jsou prvně umístěny do~bufferu a~postupně se vykonávají. {\tt CommandStack} je tedy implementován jako kruhový buffer. Do~tohoto bufferu jsou ukládány prvky typu {\tt CommandStruct}. Tato struktura reprezentuje jakýkoliv vykonatelný příkaz. {\tt CommandStack} také udržuje informace o~aktuálně vykonávaném a~posledním vykonaném příkazu. Rozhraním pro tuto komponentu je funkce StartPerforming. Tato funkce je po inicializaci všech komponent po startu interpolátoru volána v~nekonečné smyčce. V~této funkci je vždy vyjmut z~bufferu první příkaz a~je poslán ke zpracování. \subsection{Struktura CommandStruct} Struktura CommandStruct je definována následně: \begin{verbatim} typedef struct { uint8_t receiver;//Receiver component for command uint8_t type;//Type of command CommandID ID;//Unique ID for command union { uint32_t delay;//Delay in ms for waiting uint8_t LED;//Led argument for GPIO AxisLine line;//line argument for Axis component AxisSine sine;//sine arguments for Axis component AxisCircle circle;//circle argument for Axis component }; } CommandStruct; \end{verbatim} V první části struktury jsou deklarovány členy, které jsou společné pro všechny příkazy. Mezi ně patří cílová komponenta, typ příkazu (enumerace pro tyto členy jsou opět definovány v~souboru {\tt CommunicationEnumerations.h}) a~ID každého příkazu. V~druhé části jsou pak v~unionu uloženy datové struktury pro jednotlivé typy příkazů. {\tt ID} je definováno jako 32-bitové neznaménkové číslo. Každý příkaz má jedinečný identifikátor přidělený počítačem -- používá se např. když interpolátor posílá počítači zprávu, který příkaz je právě vykonáván. Identifikátory není třeba recyklovat -- pokud by každou milisekundu byl vykonán jeden příkaz, k~přetečení tohoto datového typu by došlo za 49 dní, což je dostatečná rezerva. \subsection{Zpracování zpráv} CommandStack příjmá následující zprávy: \paragraph{COMSTACK\_RET\_SPACE} se dotazuje na prázdné místo v~bufferu. Nemá žádné parametry. Odpovědí je {\tt COMSTACK\_FREE\_SPACE\_MESS}, která vrací 16-bitové číslo vyjadřující počet prázdných míst v~bufferu. \paragraph{COMSTACK\_ADD\_COM} přidává do bufferu příkazy. Zpráva je složena z~8-bitového čísla určující počet zpráv k~přidání. Za ním následuje $n$ příkazů v~podobě {\tt CommandStruct}. Pokud není dostatek volného místa v~bufferu, nastaví se chybový stav {\tt ERR\_COMSTACKOWERFLOW} a~odešle se zpráva počítači. \paragraph{COMSTACK\_LAST\_COM\_ID} se dotazuje na ID posledního vloženého příkazu. Odpovědí je {\tt COMSTACK\_LAST\_ID}, která vrací ID posledního příkazu. \paragraph{COMSTACK\_LAST\_ITEM\_PROC} vrací identifikátor posledního úspěšně dokončeného příkazu. Odpovědí je {\tt LAST\_PROC\_ITEM\_STACK}, která vrací ID posledního příkazu. \paragraph{COMSTACK\_CURRENT\_ITEM\_PROC} vrací identifikátor právě prováděného příkazu. Odpovědí je {\tt CUR\_PROC\_ITEM\_STACK}, která vrací ID posledního příkazu. \paragraph{COMSTACK\_CLEAR} vyprázdní celý buffer. \paragraph{COMSTACK\_START} spustí vykonávání příkazů v~bufferu. \paragraph{COMSTACK\_PAUSE} zastaví vykonávání příkazů v~bufferu. \section{Komponenta Axis} Komponenta {\tt Axis} vykonává provádění veškerých příkazů pro pohyb. V~této komponentě jsou definovány buffery, které uchovávají aktuální \uv{virtuální} pozici všech os. Na základě obsahu těchto bufferů řídí komponenta {\tt Stepper}, popř. {\tt Servo}, drivery pohonů. Interpolace probíhá v~předem dané obnovovací frekvenci. To má na starosti timer {\tt TIM14}, který na základě nastavení, které je posláno z~počítače, pravidelně volá funkci {\tt UpdatePosition} pro vykonání dalšího kroku interpolace. Příkaz pro pohyb vyvolává komponenta {\tt CommandStack}, která volá funkci {\tt ProcessAxisCommand}. Tato funkce jako argument přebírá strukturu {\tt CommandStruct}. Na základě jejího obsahu připraví funkce {\tt ProcessAxisCommand} příslušný druh interpolace. Funkce poté před svým ukončením čeká, než funkce {\tt UpdatePosition} nastaví příznak označující dokončení pohybu. \subsection{Použití datového typu float} Před implementací této komponenty bylo důležité vhodně vybrat datové typy, ve kterých budou probíhat interpolační výpočty. Ačkoliv jsem zvažoval použití celočíselných datových typů a~následnou implementaci fixed-point aritmetiky, rozhodl jsem se nakonec použít čísla s~plovoucí desetinnou čárkou -- konkrétně datový typ {\tt float}. Čísla s~plovoucí desetinnou čárkou se pro podobné výpočty zpravidla nepoužívají, jelikož při jejich nesprávném použití můžou vznikat velké chyby a~práce s nimi na mikrokontrolérech bývá pomalá. Jelikož však mnou vybraný mikrokontrolér disponuje hardwarovou FPU, je práce s~datovým typem float téměř stejně rychlá jako s~celočíselným datovým typem. Navíc nemusím řešit změnu řádu při operacích jako je násobení a~dělení, či operace se dvěma čísly v~rozdílných řádech. Veškeré výpočty, které provádím, jsou počítány absolutně -- jejich vstupem jsou předem daná přesná data, nikdy se nezakládají na dříve vypočtené hodnotě -- výpočty tedy nejsou iterativní. V~případě interpolace nepřičítám k~aktuální pozici změnu, ale do bufferu ukládám součet počáteční pozice a~uražené dráhy. Existují však případy, kdy se nelze vyhnout iteračnímu výpočtu -- jedním z~případů je postupné obnovování hodnoty aktuálního času, kdy se k~proměnné přičítá velmi malá hodnota. V~tomto případě uchovávám a~pracuji s~hodnotou času jako s~celočíselným typem a~až těsně před použitím ve výpočtu převedu celočíselný datový typ na float. Stejně tak pracuji s~uloženou pozicí krokového motoru. \subsection{Implementace funkce sinus} Celý fyzikální model je založen na funkcích sinus a~cosinus. Tyto funkce jsou ve standardní knihovně implementovány za pomoci řad. Tato implementace je přesná, avšak velmi pomalá. Rozhodl jsem se proto naimplementovat tyto funkce pomocí vyhledávací tabulky. Mou prioritou bylo dosažení co nejrychlejšího výpočtu a~rozumné přesnosti. Jelikož má mnou vybraný mikrokontrolér dostatek flash paměti, připravil jsem si tabulku obsahující hodnoty pro kladnou půlvlnu sinu po krocích $\frac{\pi}{4000}$. Tato tabulka zabírá v~paměti necelých 16~kB. Funkce cosinus je realizována za pomoci posunu v~této tabulce. Touto tabulkou jsem získal velmi rychlý výpočet pro hodnoty sinu s~dostatečnou přesností. \subsection{Lineární interpolace} Lineární interpolace je využita pro realizaci funkcí G00 a~G01, popř. jako část cyklů (např. G73 vrtací cyklus)\cite{gcode}. Tato interpolace je předávána jako prvek line struktury CommandStruct. Line je typ AxisLine, který je definován následovně: \begin{verbatim} typedef struct { float axis[3]; float v0, v, vb; float As, Ab; } AxisLine; \end{verbatim} Pole {\tt axis} reprezentuje novou polohu os, neboli cílový bod interpolace ze současné pozice. Členy {\tt v0, v, vb} reprezentují počáteční rychlost, respektive cílovou a~brzdnou. {\tt As a~Ab} reprezentují zrychlení použité pro rozjezd a~brzdění. Veškeré jednotky, které stroj používá, používají jako jednotku délky 1~mm, tedy rychlost je v~jednotkách 1~mm$\cdot$s$^{-1}$, zrychlení v~jednotkách 1~mm$\cdot$s$^{-2}$. Tato struktura je zpracována funkcí {\tt ProcessAxisLine} (voláno funkcí {\tt ProcessAxisCommand}). Tato funkce připraví nezbytná data tak, aby interpolace každé osy mohla probíhat nezávisle na~ostatních. V~této funkci nejprve určím vektor ve směru pohybu jako rozdíl konečného a~počátečního bodu. Tento vektor znormalizuji a~jeho vynásobením se zadaným zrychlením dostanu maximální velikost zrychlení pro každou osu. Stejnou proceduru provedu i s~jednotlivými rychlostmi. Nyní dopočítám dle odvozených vztahů v~kapitole \ref{kap:pohybpousecce} čas a~dráhu nutnou k~rozjezdu. Na~základě těchto dvou údajů poté určuji v~jaké fázi se pohyb nachází, abych mohl jeho polohu počítat dle správných vztahů. Pro interpolaci využívám vztah pro dráhu (vztah \ref{rov:lindraha}), který mi vyjadřuje okamžitou vzdálenost od~počátku pohybu. V tomto vztahu je spousta konstantních výrazů pro daný pohyb, proto si tyto konstanty před začátkem pohybu předpočítám, abych je nemusel při každém kroku interpolace znovu počítat. Všechny předpočítané údaje jsou uloženy ve dvou strukturách. Struktura {\tt AxisMovementLineCommon} reprezentuje data shodná pro všechny osy. \begin{verbatim} typedef struct { float T, TB, t; } AxisMovementLineCommon; \end{verbatim} Zde {\tt T} vyjadřuje čas nutný k~rozjezdu, {\tt TB} čas nutný ke zbrzdění a~{\tt t} čas, jak dlouho trvá pohyb konstantní rychlostí. Struktura {\tt AxisMovementBufferLine} je přiřazena ke každé ose a~uchovává jejich specifická data. \begin{verbatim} typedef struct { float* position; float ATpi, ATpiB; float start, constant, brake, end; float v, v0; } AxisMovementBufferLine; \end{verbatim} {\tt position} je ukazatel na float uchovávající aktuální pozici osy, {\tt ATpi} ({\tt ATpiB}) je předpočítaná hodnota výrazu $\frac{AT}{\pi^2}$ (respektive jeho varianty pro brzdění). {\tt start, constant, brake, end} znamenají postupně: pozici osy pro počátek; místo, odkud se osa pohybuje konstantní rychlostí; místo, odkud začíná osa brzdit; cílová pozice. {\tt v} reprezentuje konstantní rychlost, {\tt v0} počáteční. Interpolace probíhá následovně. Prvně se pro všechny osy inkrementuje čas. Dále probíhá interpolace pro každou osu samostatně -- na základě času se určí, v~jaké fázi se pohyb nachází. Následně se spočte jeho dráha podle vztahu \ref{rov:lindraha} a~přičte se k~ní počátek celého pohybu. Pokud na dané ose došlo ke změně polohy, je zavolána fukce NotifyMovement, která aktualizuje polohu pohonu. Jakmile čas dosáhne zadané hranice, je nastaven příznak dokončeného pohybu a~funkce {\tt ProcessAxisLine} se ukončí. \subsection{Oblouková interpolace}\label{kap:oblinter} Oblouková interpolace je využita pro realizaci funkcí G02, G03, G12 a~G13\cite{gcode}, popř. je využita při cyklech a~kompenzaci nástroje. Tato interpolace má podobnou mechaniku a~hiearchii jako lineární, proto se v~následujícím textu zaměřím pouze na rozdíly. Příkaz pro obloukovou interpolaci je předáván jako prvek {\tt circle} struktury {\tt CommandStruct}. {\tt circle} je typ {\tt AxisCircle}, který je definován následovně: \begin{verbatim} typedef struct { float B[3], C[3]; float v, v0, vb; float As, Ab; } AxisCircle; \end{verbatim} {\tt B} a~{\tt C} zde označují body; B je koncový bod oblouku a~C je bod ležící kdekoliv na něm (s~podmínkou, že současná poloha stroje a~body B, C nesmí ležet na jedné přímce). {\tt v, v0, vb} opět označují příslušné rychlosti, {\tt As} a~{\tt Ab} opět označují použitá zrychlení. Zde je nutno podotknout, že současná podoba interpolace je v~rozporu se standardem G-kódu. Můj kód je schopný projet kruhový oblouk ležící v~obecné rovině (proto se zadává za pomoci tří bodů), zatímco G-kód definuje jako přípustné roviny pouze XY, XZ a~YZ. Tato chyba vznikla mým prvopočátečním špatným pochopením standardu. Dále dle standardu je možné pomocí kruhové interpolace interpolovat šroubovici. Toho lze docílit zadáním z-ové (respektive y-ové nebo x-ové) souřadnice koncového bodu tak, aby ležel mimo danou rovinu. Tento druh interpolace můj systém zatím nezvládá a~chybně se snaží ze zadaných údajů sestrojit kružnici v~obecné rovině. Z~tohoto důvodu probíhá samotná interpolace zbytečně složitě. Výše uvedená struktura je zpracovávaná funkcí {\tt ProcessAxisCircle} (volané funkcí {\tt ProcessAxisCommand}). V~této funkci jsou připravena veškerá data. Kružnice je interpolována společně pro všechny osy -- výstupem interpolace je úhlová dráha (v~kódu označována jako {\tt alph}a), na~základě které se dopočítává poloha jednotlivých os. Funkce {\tt ProcessAxisCircle} nejprve dopočte střed kružnice na základě jejích třech bodů (sestrojením normálového vektoru roviny kružnice, díky němuž je možné vytvořit osy tětiv ze~tří bodů a~nalezením průsečíků těchto os). Poté připraví data. Příjatá data ve struktuře {\tt AxisCircle} vyjadřují obvodové hodnoty. Jak jsem napsal v~předchozím odstavci, pro mou interpolaci potřebuji úhlové hodnoty. Proto všechny hodnoty rychlostí a~zrychlení před použitím vydělím poloměrem oblouku, čímž získám úhlovou rychlost a~úhlové zrychlení, na základě nichž dopočítám potřebné konstanty stejně jako v~případě lineární interpolace (nyní však pouze pro jednu \uv{osu} -- úhlovou). Po dopočítání úhlové dráhy potřebuji tuto dráhu převést na kružnici v~kartézských souřadnicích. Pro tento převod používám parametrickou rovnici kružnice v~obecné rovině (vztah \ref{rov:kruznice}\cite{kruznice}) \begin{equation} \label{rov:kruznice} \vec{P} = R\;\vec{u}\cos(t) + R\sin(t) \;\;\vec{n}\times\vec{u} + \vec{c} \end{equation} V tomto vztahu $\vec{P}$ je polohový vektor hledaného bodu, $R$ poloměr oblouku, $t$ parametr, $\vec{n}$ jednotkový vektor kolmý na rovinu kružnice, $\vec{u}$ jednotkový vektor směřující ze středu k~počátečnímu bodu oblouku a~$\vec{c}$ polohový vektor středu kružnice. Vektorovou rovnici jsem si pro potřeby převodu rozdělil na jednotlivé osy, jim předpočítal konstantní hodnoty a~jako paramet předávám okamžitou úhlovou dráhu. Potřebná data jsou uložena obdobně jako u~lineární interpolace v~následujících strukturách: \begin{verbatim} typedef struct { float w, w0; float alpha; float constant, brake, end; float ATpi, ATpiB, T, TB; } AxisMovementCircleCommon; typedef struct { float* position; float final; float crossProduct, center, R, u; } AxisMovementBufferCircle; \end{verbatim} Vynechané části průběhu interpolace probíhají podobně jako u lineární, proto je dále nepopisuji. \subsection{Interpolace ideální křivky} V kódu lze spatřit i odkazy na interpolaci ideální křivky z~kapitoly \ref{kap:krivka}. Jak jsem v~ní zmínil, s~křivkou jsem počítal pouze na~počátku a~v~pozdější fázi vývoje jsem se ji rozhodl vypustit. Interpolace této křivky je tedy ve funkčním stavu, avšak na této křivce není omezen ryv a~neexistuje možnost, jak křivku vložit z~G-kódu (není podporovaná na straně počítače). V~kódu interpolátoru jsem ji však do budoucna pro případnou plnou implementaci ponechal. \section{Komponenty Movement a~Stepper} Komponenta {\tt Movement} zajišťuje převod pozice jednotlivých os na výstupní signál pro drivery. V~začátcích projektu jsem měl v~plánu implementovat jak step-dir řízení v~podobě komponenty {\tt Stepper}, tak i PID smyčku se zpětnou vazbou pro přímé řízení servomotoru v~podobě komponenty {\tt Servo}. V~současnosti je implementována pouze komponenta {\tt Stepper} a~interpolátor tedy umí generovat pouze step-dir signál. Úkolem komponenty {\tt Movement} je zajistit jednotné rozhraní pro generování výstupních signálů pro komponentu {\tt Axis}, která volá její funkci {\tt NotifyMovement}. Pro každou osu je vytvořen jeden prvek následující struktury: \begin{verbatim} typedef enum {STEPPER = 1, SERVO = 2, UNUSED = 0} MovementType; typedef struct { MovementType type; union { Stepper stepper; Servo servo; }; } MovementStruct; \end{verbatim} Při zavolání funkce {\tt NotifyMovement} (tedy když se změnila poloha osy), je na základě členu type rozhodnuto, zdali se bude volat funkce pro upozornění komponenty {\tt Stepper} nebo {\tt Servo}. To, která osa bude obsahovat komponentu {\tt Stepper} a~která {\tt Servo}, je určeno nastavením z~počítače. Toto nastavení je provedeno pomocí příchozí zprávy {\tt CONFIG\_MOVEMENT\_SETTYPE}. První byte této přijaté zprávy určuje index osy, pro kterou se nastavuje typ, a~druhý byte určuje druh generátoru výstupního signálu (Stepper/Servo) pro tuto osu. Druhou zprávou, kterou je nutno poslat, je zpráva {\tt CONFIG\_MOVEMENT\_INDIVIDUAL} předávající dané ose, respektive generátoru, jeho příslušná nastavení. První byte této zprávy určuje index osy a~za ním následují konfigurační data (konfigurační zpráva), která jsou předána již příslušné komponentě ke zpracování. \subsection{Komponenta Stepper} Komponenta {\tt Stepper} sloužící ke generování step-dir signálu je reprezentována následující strukturou: \begin{verbatim} typedef struct { uint32_t mmPerStep; int32_t position; uint8_t pulsePin; GPIO_TypeDef* GPIO; uint16_t directionPin; bool inverted; bool lastDirection; } Stepper; \end{verbatim} Tato struktura uchovává veškerá data a~nastavení. Člen {\tt mmPerStep} vyjadřuje vzdálenost, o kterou je posunuta osa při jednom kroku. Tato hodnota je uložena v~desetitisícinách milimetru ($10^{-4}~$mm). Člen {\tt position} vyjadřuje současnou pozici osy v~desetitisícinách milimetru (tedy maximální délka osy je $\pm214$~metrů). Člen {\tt pulsePin} je index výstupního pinu pro komponentu {\tt PulseGenerator}, na kterém probíhá generování signálu step. Členy {\tt GPIO} a~{\tt directionPin} určují výstupní pin pro signál dir. Člen {\tt inverted} určuje, zda-li je třeba prohodit směr otáčení osy. Člen {\tt lastDirection} není nastavením, nýbrž ukládá směr pohybu při poslední aktualizaci pohybu. Pokud nastane změna pohybu osy, je skrz univerzální rozhraní komponenty {\tt Movement} zavolána funkce komponenty {\tt Stepper NotifyStepper}. Tato funkce nejprve zjistí odchylku fyzické pozice osy (na základě členu {\tt position} ve struktuře {\tt Stepper}) od~pozice osy vypočtené komponentou {\tt Axis}. Pokud je tato odchylka větší než velikost jednoho kroku (člen {\tt mmPerStep}), zjistí jednotka směr (tedy polaritu signálu dir) a~za pomoci komponenty {\tt PulseGenerator} vygeneruje signál step. Poté přičte, či odečte délku jednoho kroku od~aktuální pozice. \subsection{Komponenta PulseGenerator} Komponenta {\tt PulseGenerator} má za úkol generovat pulzy. Původní myšlenka byla, aby byla schopna generovat pulzy libovolné délky v~daném rozsahu s~největší možnou přesností. Toho jsem chtěl docílit použitím timerů {\tt TIM3} a~{\tt TIM12} v~PWM módu. V~PWM módu generují timery na pinech {\tt PA6, PA7, PB0, PB1} a~{\tt PB14, PB15} PWM signál\cite{refman}, jehož střídu lze určit hodnotou příslušných compare registrů. Frekvence PWM je dána předděličkou k~timeru. Procesor umí generovat přerušení při přetečení timeru\cite{refman}. Těchto vlastností jsem využil ke generování jednoho pulzu -- nastavil jsem si do fronty délku jednoho pulzu, při přetečení timeru jsem nastavil správnou hodnotu šířky pulzu a~PWM mód jsem spustil. Při opětovném přetečení jsem timer vypnul. Tím jsem docílil generování jednoho pulzu. V~průběhu testování se však ukázalo, že vznikají parazitní pulzy -- pohony dělaly kroky navíc. Podezříval jsem, že se díky vnořování přerušení nestíhá PWM mód v~čas vypínat -- tuto doměnku se mi však nepodařilo ověřit. Rozhodl jsem se proto oželet možnost generovat proměnlivou délku pulzu a~za pomoci timeru generovat pulz konstantní šířky. Bohužel při tomto generování pulzu vznikal stejný jev -- objevovaly se pulzy navíc. Snažil jsem se chybu objevit, avšak neúspěšně. Z nedostatku času a~díky nutnosti funkčnosti této komponenty (bez ní nelze řídicí systém testovat), jsem se uchýlil k~dočasnému hacku. Nyní generuji 100~ns pulz a~čekání zajišťuji sérií instrukcí nop. Nyní mi již nevznikají žádné parazitní pulzy. Toto řešení je dočasné a~mělo by být co nejdříve nahrazeno jiným. \section{Implementace ostatních komponent} V předchozích sekcích byly popsány nejdůležitější komponenty, které tvoří jádro celého interpolátoru. Tyto komponenty mají na starost pohyb stroje. G-kód však definuje další druhy příkazů, které se netýkají pohybu a~proto bylo nutné do návrhu přidat ještě dvě komponenty. \subsection{Komponenta GPIOControl} Tato kompenta má za úkol obsluhovat vstupy a~výstupy mikrokontroléru. Skrze ni by mělo být např. zapínáno vřeteno (funkce M03, M04\cite{gcode}). Záměrně byla tato komponenta navžena úzce spjatá s~hardwarem, aby si zachovala univerzálnost, a~tak bylo možné v~počítači co nejpodrobněji nakonfigurovat chování. Tato komponenta není zatím dokončena a~pro demonstrativní účely zatím umí ovládat jednotlivé LED na kitu, které jsem používal pro demonstraci funkčnosti. \subsection{Komponenty SystemComponents} {\tt SystemComponents} zastřešuje příkazy spíše systémového rázu, které nesouvisí s~výstupem. Příkladem takového příkazu (a zatím jediného implementovaného z této katogorie) je funkce prodlevy -- P parametr příkazů G-kódu\cite{gcode} realizován pomocí {\tt DelayComponent}. Prodleva je implementována pomocí přerušení {\tt SysTick} \cite{refman}. Frekvence tohoto přerušení byla nastavena na 1~kHz -- přerušení nastává každou milisekundu. {\tt DelayComponent} implementuje seznam až 32 prodlev. Pokaždé, když nastane přerušení, je hodnota každé prodlevy právě o~jednu milisekundu zmenšena. Jakmile dosáhne hodnota prodlevy nuly, je prodleva ze seznamu odstraněna a~je zavolána případná callback funkce. \chapter{Software pro počítač} Jak jsem již zmínil v~kapitole \ref{kap:prostredky} (\nameref{kap:prostredky}), aplikace pro počítač je napsána v~jazyce C++ s~využitím prvků z~nového standardu C++11. Pro GUI jsem použil framework WxWidgets. V~následujícím textu záměrně vynechávám některé detaily implementace (především co se GUI a~výstavby aplikace týče), jelikož se jedná vesměs o~standardní postupy, a~zaměřuji se na specifika řídicího systému -- interpretaci G-kódu, výpočet rychlosti pro stroj a~následnou komunikaci s~interpolátorem. \section{Uživatelské rozhraní} Uživatelské rozhraní mé aplikace je tvořeno jedním oknem, rozděleným na tři panely (viz obrázek \ref{obr:okno}). \begin{figure}[h] \centering \includegraphics[width=0.99\textwidth]{img/okno2.png} \caption{Screenshot uživatelského rozhraní}\label{obr:okno} \end{figure} Pravý panel zobrazuje aktuálně načtený program v~G-kódu a~za jeho běhu zvýrazňuje aktuálně prováděný řádek. Úplně levý panel zatím není využit, do budoucna by však měl zobrazovat další ovládací prvky a~informace. Prostřední panel zobrazuje náhled drah stroje. V~současné verzi zobrazuje pouze 2D náhled drah, které leží v~rovině XY. Náhled je možné libovolně posouvat myší a~kolečkem jej přibližovat/oddalovat. Do tohoto náhledu se také zobrazuje zpětná vazba z~interpolátoru (zvýrazňují se již projeté dráhy). Dráhy zadané přímo G-kódem se zobrazují barevně odlišené podle druhu interpolace (rychloposuv je naznačen přerušovaně). Barvy jednotlivých typů interpolace lze navolit v~konfiguračním souboru. Dráhy, které byly dopočteny (vznikly kompenzací nástroje), jsou zobrazeny bíle. Dole v~taskbaru aplikace jsou zobrazeny následující údaje: stav připojeno/odpojeno od~interpolátoru, otevřený program, okamžitá pozice os a~současný (chybový) stav interpolátoru. V~toolbaru se nacházejí tři tlačítka pro spuštění, pozastavení a~úplné zastavení běhu programu. V~menu aplikace Soubor lze otevřít program v~G-kódu. V~menu náhled je možné vypnout zobrazování již provedených pohybů strojem a~obnovit zobrazení náhledu drah. Uživatelské rzohraní je v~současném stádiu vývoje velmi chudé a~veškerá nastavení jak aplikace, tak interpolátoru jsou prováděna úpravou konfiguračního INI souboru. Aplikace využívá jediný konfigurační soubor {\tt settings.ini}. V tomto souboru jsou definovány jednotlivé sekce, obsahující příslušná nastavení. \section{Intepretace G-kódu} Jednou z~klíčových vlastností řídicího systému je parsování G-kódu na interní formát, se kterým lze uvnitř programu pracovat. \subsection{G-kód a~jeho formát} \label{kap:gcodeform} G-kód je univerzální jazyk pro programování CNC strojů\cite{wiki:gcode}. Syntaxe G-kódu je ralativně jednoduchá. Program je dělen na tzv. bloky. Blok je jeden řádek programu. V~každém bloku jsou uvedeny hodnoty jednotlivých registrů (A--Z). Registr G může být uveden vícekrát, čímž mu je přiděleno více hodnot\cite{wiki:gcode}. G-kód ignoruje mezery a~obecně jakékoliv bílé znaky. Program zpravidla začíná a~končí znakem procenta (\%). Komentáře v~G-kódu jsou umístěny do obyčejných jednoduchých závorek a~nesmí přesáhnout hranici jednoho bloku. Nejpoužívanějšími registry jsou registry G a~M, které uchovávají typ funkce k~vykonání. Ostatní registry slouží jako parametry těchto funkcí. Registry X, Y, Z uchovávají souřadnice lineárních os (koncové body). Vedle těchto registrů se používají i registry A, B, C pro polohu rotačních os, avšak jelikož můj interpolátor podporuje pouze 3 osy, nebudu tyto registry dále zmiňovat. Registry I, J, K a~R slouží pro zadávání parametrů obloukové interpolace. Registr F určuje rychlost stroje (posuv). Ostatní registry jsou méně používané, jejich kompletní seznam lze nalézt na stránkách \cite{wiki:gcode}. Seznam všech G-funkcí, které můj systém v~současné verzi podporuje, a~jejich popis je uveden v~tabulce~\ref{tab:funkce}. Funkce G zpravidla vykonávají pohyb stroje a~nastavují parametry. Funkce M mají speciální význam, který je dán specificky pro každý stroj. Existují však některé ustálené M-funkce (např. M03 pro zapnutí vřetena\cite{wiki:gcode}). Pokud je v~jednom bloku uvedeno více funkcí, určuje pořadí jejich vykonání stroj. G-kód podporuje i podprogramy, zápis matematických výrazů či polární zadávání souřadnic. Tyto funkce však dnes už nejsou (obvzláště v~hobby sféře) příliš rozšířené. Dříve podprogramy umožňovaly zkrácení programu a~jeho snažší zápis pro člověka. Tyto funkce však již dnes s~nástupem CAM programů nejsou zapotřebí. Proto jsem se tyto funkce rozhodl do mého systému neimplementovat. Některé hodnoty některých registrů si systém pamatuje do doby, dokud nejsou znovu nastaveny. Příkladem můžou být zadané cílové souřadnice, či posuv. Není tedy nutné zadávat pokaždé znovu všechny 3 souřadnice, stačí pouze ty, které se změnily. Také pokud byla použita funkce G01 (lineární interpolace), tak lze pro sérii pohybů po úsečce napsat pouze do prvního bloku G01 a~do dalších bloků uvést pouze souřadnice. Systém pak automaticky použije funkci G01. Část programu v G-kódu může vypadat následovně: \begin{verbatim} G00 X1 Y31 Z50 G01 Z0.1 F100 X0.19 Y 29.58 G02 X2.1 Y26.89 I6.64 J3.47 X5.53 Y25.61 I4.26 J6.16 X2.1 Y26.89 I6.64 J3.47 \end{verbatim} \begin{table} \begin{tabularx}{\textwidth}{|c|X|c|} \hline Funkce & Význam & Argumenty \\ \hline \hline G00 & Rychloposuv. Funkce není určená pro obrábění. Stroj nemusí pohyb interpolovat lineárně. & X, Y, Z \\ \hline G01 & Lineární interpolace ze současné pozice na zadanou posuvem F & X, Y, Z, F \\ \hline G02 & Oblouková interpolace ve směru hodinových ručiček. Oblouk je zadán současnou polohou a~koncovým bodem. Jeho střed je určen buď registry I, J, K, které určují odsazení středu od~počáteční polohy, nebo prametrem R učujícím poloměr oblouku. Učebnice G-kódu doporučují zadávání pomocí odsazení. & X, Y, Z, I, J, K, R, F \\ \hline G03 & Oblouková interpolace proti směru hodinových ručiček. Zadáváno stejně jako G02 & X, Y, Z, I, J, K, R, F \\ \hline G20 & Nastaví pracovní jednotky na palce & ~ \\ \hline G21 & Nastaví pracovní jednotky na milimetry & ~ \\ \hline G40 & Funkce zruší kompenzaci nástroje & ~ \\ \hline G41 & Zapne kompenzaci nástroje doleva ve směru obrábění. Průměr nástroje je zadán jako D & D \\ \hline G42 & Zapne kompenzaci nástroje doprava ve směru obrábění. Průměr nástroje je zadán jako D & D \\ \hline G43 & Zapně kompenzaci délky nástroje kladného směru osy~Z. Délka nástroje je zadána jako H & H \\ \hline G44 & Zapně kompenzaci délky nástroje do záporného směru osy~Z. Délka nástroje je zadána jako H & H \\ \hline G49 & Vypne délkovou kompenzaci nástroje & ~ \\ \hline \end{tabularx} \caption{Tabulka G-funkcí podporovaných mým systémem. Definice funkcí převzaty z~\cite{gcode}.}\label{tab:funkce} \end{table} \subsection{Interní formát a~třída GCodeInterpreter} Abych mohl G-kód dále zpracovat (dopočítat rychlost, provést korekce, odeslat do interpolátoru), musím jej neprve převést na interní formát mého programu, se kterým lze dobře pracovat. Veškeré zpracování G-kódu má na starosti třída {\tt GCodeInterpreter}. Tato třída implementuje veškerou funkcionalitu pro převod vstupního řetězce obsahujícího G-kód na příkazy pro interpolátor. Tento proces v~sobě zahrnuje i provedení korekcí nástroje a~výpočet rychlosti. Převedení G-kódu na výsledek je spuštěno funkcí {\tt ProcessString}. Tato metoda přebírá {\tt std::string}, který obsahuje G-kód. Výsledkem je seznam objektů pro vykreslení na obrazovce a~seznam připravených příkazů pro interpolátor. Třída {\tt GCodeInterpreter} v~sobě zahrnuje i načtení konfigurace týkající se zpracování z~daných konfiguračních souborů a~jejich následné předání příslušným metodám či objektům. Také uchovává stav jednotlivých parametrů v~průběhu převodu. \subsection{Převod G-kódu} Parsování G-kódu probíhá v metodě {\tt GCodeInterpreter::ParseCodeToPath} ve dvou krocích -- v~prvním kroku z~G-kódu vyčtu hodnoty jednotlivých registrů pomocí metody {\tt GCodeInterpreter::ParseGCodeLine}. Jelikož některé registry mohou obsahovat více hodnot (např. registr G), vrací tato funkce asociativní pole, kde je klíčem char označující registr a~hodnotou je seznam čísel s plovoucí desetinnou čárkou ({\tt std::map$<$char, vector$<$float$>>$}). V~druhém kroku jsou z~tohoto asociativního pole vybrány jednotlivé funkce a~jejich argumenty podle prototypů. To provádí metoda {\tt GCodeInterpreter::SeparateCommandsFromLine}. Zároveň tato metoda zajišťuje správné pořádí provádění funkcí v~případě, kdy je jich v~bloku uvedeno více. Metoda {\tt ParseGCodeLine} v~cyklu postupně načítá jeden znak ze stringu. Pokud je tento znak bílý, tak jej přeskočí. Pokud narazí na znak~\uv{{\tt (}}, začne načítat a~zahazovat další znaky, dokud nenarazí na znak~\uv{{\tt )}}. Pokud je načtený znak písmeno, načte pomocí třídy stringstream z~následujích znaků číslo a~uloží je do asociativního pole, které je na konci navráceno jako výsledek. Metoda {\tt SeparateCommandsFromLine} využívá seznam prototypů funkcí G-kódu uvedených v~souboru {\tt gcode.dat}. Tento soubor zároveň definuje pořadí vykonání funkcí, pokud by všechny byly uvedeny v~jednom bloku. Toto pořadí jsem převzal z~online učebnice G-kódu \cite{gcode2}. Prochází jeden prototyp funkce za druhým a~zjišťuje, jestli pasuje na hodnoty registrů předané v~tomto bloku. Pokud ano, zkopíruje tyto hodnoty do struktury {\tt GCodeLine}, která vždy obsahuje přesně jednu funkci pouze s~jejími parametry. Tato metoda zároveň zjišťuje, zda-li je možné převzít některý z~předchozích registů G -- např. pro řetězení funkcí G01 bez jejich opisování, jak bylo zmíněno v~kapitole \ref{kap:gcodeform}. Soubor {\tt gcode.dat} je textový soubor, který na každém řádku obsahuje definici jednoho prototypu. Je složen ze sloupců oddělených tabulátorem. První sloupec obsahuje druh registru -- buď M, nebo G. Druhý sloupec obsahuje číslo funkce, popř. -1 pokud na tomto čísle nesejde (např. posuv je realizován jako funkce). V~dalších sloupcích jsou vždy uloženy jednotlivé kombinace registrů, které fungují jako argumenty. Jednotlivé registry jsou od~sebe odděleny mezerou a~za~posledním je umístěn středník. \section{Třída PathPart} G-kód je v~mém systému převáděn na trasu ({\tt path}). Trasa vyjadřuje všechny úseky pohybu proložené všemi modálními příkazy (např. změna posuvu, zapnutí vřetene, atd.). Jednotlivé úseky jsou poté převáděny na příkazy pro interpolátor. Trasa je definována jako {\tt list$<$unique\_ptr$<$PathPart$>>$}. {\tt list} (tedy řetězený seznam) jsem zvolil, abych mohl přidávat prvky uprostřed seznamu s~malou časovou náročností. Cenou za to je ztráta nahodilého přístupu jako u kontejneru {\tt vector}. Avšak trasu vždy potřebuji procházet pouze sekvenčně, takže zde pro mě {\tt list} neskýtá žádné omezení. Třída {\tt PathPart} je abstraktní základní třída, od~které dědí všechny ostatní části. Abych mohl využít polymorfického chování, je umístěna v~kontejneru skrz třídu {\tt unique\_ptr} ze standardní knihovny C++11. Tato třída je deklarována v~hlavičkovém souboru {\tt PathPart.h}. Třída obsahuje deklaraci metod pro všechny úkony, které lze s~úsekem trasy provést. Definuje funkce pro získání jednotlivých rychlostí, pro získání tečného vektoru na svém začátku či konci. Také definuje metodu {\tt ComputeSpeeds}, která má za úkol provést výpočet rychlosti na daném úseku. Metoda {\tt ProcessCompensationStarting}, respektive {\tt ProcessCompensationEnding}, řeší provedení kompenzace nástroje na začátku tohoto úseku, respektive na jeho konci. Jelikož je při některých operacích nutné přistupovat k~předcházejícímu či následujícímu prvku, obsahuje třída PathPart atribut {\tt list$<$unique\_ptr$<$PathPart$>>$::iterator item}. Tento atribut uchovává iterátor ukazující na tento objekt třídy {\tt PathPart}. Jeho inkrementováním či dekrementováním můžu získat následující úsek, respektive předcházející. Toto řešení je založeno na tom, že iterátor kontejneru {\tt list} zůstává po celou dobu existence daného prvku platným, nehledě na přidané či odebrané prvky. Z třídy {\tt PathPart} jsou děděny dvě třídy, které slouží jako základ pro implementaci konkrétních typů úseku. Jsou to třídy {\tt PathPartMovable} a~{\tt PathPartModable}. Třída {\tt PathPartMovable} je základem pro části trasy, které skutečně reprezentují pohyb stroje a~je nutné pro ně počítat rychlost a~provádět korekci nástroje. Naopak potomci třídy {\tt PathPartModable} nereprezentují pohyb stroje, ale ostatní příkazy. Pro tyto úseky se rychlost nepočítá a~ani se neprovádí korekce nástroje. Obě výše uvedené třídy mají stejný základ a~musí definovat všechny metody. Třída {\tt PathPart} definuje metodu {\tt IsMovable}, která vrací typ třídy. Tato metoda je definována v~{\tt PathPartMovable} tak, aby vracela {\tt true}, v~{\tt PathPartModable} vrací {\tt false}. Třída {\tt PathPartMovable} také definuje všechny pro ni nesmyslené funkce (např. výpočet rychlosti) tak, že ve svém těle vyhodí výjimku, což usnadňuje debugování. Během práce s~úseky tedy volám funkci {\tt IsMovable}, abych věděl, jestli pro daný úsek mohu volat všechny metody. Toto řešení není nejelegantnější, ale použil jsem ho kvůli jednoduchosti implementace a~také kvůli tomu, že drtivá většina úseků je typu {\tt PathPartMovable} a~jen mizivá část je typu {\tt PathPartModable}. Z~třídy {\tt PathPartMovable} jsou odvozeny třídy {\tt PathPartLine} (reprezentuje lineání interpolaci), {\tt PathPartRapid} (rychloposuv), {\tt PathPartCircle} (reprezentuje obloukovou interpolaci) a~třídy {\tt PathPartOffsetLine} a~{\tt PathPartOffsetCircle} (reprezentují úseky trasy, které vznikly např. při kompenzaci nástroje a~nemají odkaz na sebe v~G-kódu). Z~třídy {\tt PathPartModable} byla prozatím odvozena pouze jediná třída {\tt PathPartUnproc}, která funguje jako forma návěstidla -- označuje např. konec či začátek programu. \section{Převod GCodeLine na PathPart} Tento převod má na starosti metoda {\tt GCodeInterpreter::ConvertCommandToPath}. Jako argument přebírá funkci G-kódu v~podobě třídy {\tt GCodeLine}. Podstata této metody spočívá v~asociativním poli {\tt processingFunctions}. Jako klíč zde vystupuje kód funkce {\tt GFunction} a~hodnotou je ukazatel na metodu třídy {\tt GCodeInterpreter}, která vykonává převod. Toto pole je deklarováno následovně: \begin{verbatim} //Definice datových typů typedef void (GCodeInterpreter::*GCodeProcessingFunction)(GCodeLine&); typedef pair GFunction; //Definice pole map processingFunctions; \end{verbatim} V~konstruktoru třídy {\tt GCodeInterpreter} je toto pole naplněno příslušnými ukazateli na metody. Tyto metody pro převod jsem nazval stejně jako příslušné G-funkce -- např. G00, G01 apod. Metoda {\tt ConvertCommandToPath} prvně zkusí najít příšlušnou funkci v~asociativním poli. Pokud ji nenalezne, vyhodí výjimku. Pokud ji nalezne, zavolá ji a~příslušná funkce se postará o~převod a~zařazení úseku do trasy. \subsection{Funkce G00} Tato funkce má za úkol vytvořit pohyb rychloposuvem. Ačkoliv standard tvrdí, že v~tomto případě není nutné, aby interpolace probíhala lineárně (může probíhat maximální rychlostí pro každou osu), implementuji jej jako lineární interpolaci s~rychlostí pro rychloposuv nastavenou v~konfiguračním souboru. Jedině tak můžu zajistit, že rychloposuv probíhá na základě fyzikálního modelu a~respektuje všechna omezení. V~prvním kroku funkce ještě na základě aktuálně nastaveného režimu souřadnic (absolutní a~inkrementální) provede jejich přepočet tak, aby vždy byly v~absolutních hodnotách. Na základě těchto údajů se zavolá konstruktor třídy {\tt PathPartRapid}, která je odvozena od~třídy PathPartLine. Takto vytvořený úsek je následně vložen do trasy. \subsection{Funkce G01} Funkce G01 slouží pro lineární interpolaci. Tato interpolace je realizována stejně jako funkce G00, pouze se jako hodnota posuvu použije naposledy zadaná hodnota funkce F a~jako úsek trasy je vytvořen objekt třídy {\tt PathPartLine}. \subsection{Funkce G02 a~G03} Funkce G02 a~G03 slouží pro zadání obloukové interpolace. Navzájem se liší pouze ve směru. V~mé konkrétní implementaci to tedy znamená, jestli výsledný úsek obdrží identifikátor {\tt CIRCLE\_CW}, nebo {\tt CIRCLE\_CCW}. Proto tento druh úseku vytváří funkce {\tt CircleFunction}, která je volána jak ve funkci G02, tak i ve funkci G03 -- pouze je jí předán jiný identifikátor. {\tt CircleFunction} projde předané argumenty v~{\tt GCodeLine} a~buď zavolá konstruktor {\tt PathPartCircle}, který přebírá poloměr, nebo konstruktor, který přebírá souřadnice středu kružnice. Zde se implementace G-kódu na různých strojích rozcházejí -- některé stroje berou argumenty I, J a~K jako odsazení od~počátku oblouku, jiné je berou jako absolutní hodnotu. Zde jsem se inspiroval implementací, kterou používá LinuxCNC -- standardně jsou tyto argumenty přebírány jako odsazení od~počátku oblouku, avšak pomocí funkce G90.1 nebo změnou nastavení v~konfiguračním souboru je lze přepnout na absolutní hodnoty. \subsection{Funkce G20 a~G21} Funkce G20 nastavuje pracovní jednotky na palce, naopak G21 na milimetry. Můj systém standardně pracuje v~milimetrech. Pro implementaci jsem zavedl do třídy {\tt GCodeInterpreter} atribut {\tt unitMultiply}, který uchovává aktuální jednotku. Tímto atributem jsou všechny souřadnice zadané v~G-kódu násobeny před jejich použitím pro kontrukci úseků trasy. Funkce G20 mu tedy přiřazuje hodnotu $2,54$; funkce G21 hodnotu $1$. \subsection{Funkce F} Funkce F nastavuje aktuální rychlost. Činí tak natavení atributu {\tt GCodeInterpreter::standard-Feed}. Tato hodnota uchovává posuv -- jednotkou je tedy buď mm$\cdot$min$^{-1}$ nebo in$\cdot$min$^{-1}$. Před jejich použitím je tedy nutné je převést na mm$\cdot$s$^{-1}$. Funkce F zároveň do trasy přidává návěstidlo pro debugovací účely. Toto návěstidlo je typu {\tt PathPartUnproc}. \subsection{Funkce G40, G41 a~G42} Funkce G40 ruší jakoukoliv aplikovanou korekci průměru nástroje. Funkce G41 zapíná kompenzaci průměru nástroje směrem doleva ve směru obrábění, naopak funkce G43 doprava. Tyto funkce jsou relizovány nastavením příšlušných hodnot objektu třídy {\tt PathPartOffset}. Objekt této třídy je předáván každému úseku typu {\tt PathPartMovable} v~jeho konstruktoru. Na~základě něj potom úsek vytvoří příslušnou úpravou trasu, která respektuje kompenzaci nástroje. \subsection{Funkce G43, G44 a~G49} Tyto funkce jsou analogické k~funkcím uvedeným v~předchozím odstavci -- tyto funkce pouze nastavují délkovou korekci nástroje. Jsou i naprosto stejně implementované. Funkce G49 délkovou kompenzaci vypíná, G43 ji zapíná do kladného směru osy Z a~funkce G44 ji zapíná do záporného směru osy Z. \section{Implementace korekce nástroje}\label{kap:korekce} Poté, co je G-kód převeden na jednotlivé úseky trasy, je tato trasa dále zpracovávána. Prvním stupněm zpracování je vyřešešení korekce nástroje. Tato korekce je vyvolána v~metodě {\tt GCodeInterpreter::PostProcessPath}, kde je postupně pro každý úsek volána jeho metoda {\tt PathPart::ProcessToolCompensation}. Zpracování každého úseku se skládá ze dvou částí -- prvně je vyřešen počáteční bod tohoto úseku a~následně koncový bod. Během tohoto zpracování je třeba někdy vložit do trasy nové úseky. Jelikož však tyto úseky nemají přímý odkaz na G-kód, nevychází z~něj a~neobsahují informace o původní trase, jsou reprezentovány speciálními úseky -- {\tt PathPartOffsetLine} a~{\tt PathPartOffsetCircle}. Standard G-kódu nijak podrobně nespecifikuje, jak má korekce nástroje probíhat a~jak mají být řešeny jednotlivé přechody. Různé stroje implementují korekce mírně odlišně. Pro mou implementaci jsem vycházel z~chování systému LinuxCNC (viz dokument \cite{emckor}), systému Sinumerik\cite{sinumerik} (i přesto, že se jedná o systém pro soustruh, bral jsem jej jako porovnání s~novějšími systémy) a~systému Heidenheim (viz manuál \cite{heid}). V následujícím textu se odprostím od~implementace tohoto řešení do kódu, která je relativně jednoduchá, a~zaměřím se na teoretickou část problému -- jak má korekce vypadat a~jak dopočítat souřadnice korigované trasy. \subsection{Korekce průměru} Korekce průměru nástroje probíhá v~rovině XY. Tedy veškeré úvahy se vztahují pouze k~průmětu aktuální trasy do roviny XY. \subsubsection{Základní korekce, tečné úseky} Nejjednodušším případem je korekce úseku, který se nachází uprostřed již korigované oblasti. Situace je zobrazena na obrázku \ref{obr:kor-zak}. Pro korekce jsem zavedl konvenci, že kladná hodnota korekce značí korekci doprava ve~směru obrábění, záporná doleva. Pro provedení odsazení na úsečce mi tedy stačí pouhé posunutí jejího počátečního a~koncového bodu. Vektor $\vec{n}$ je kolmý jednotkový vektor na daný úsek, $k$ označuje velikost korekce a~nabývá hodnot $\pm r$ ($r$ je poloměr nástroje). Posun $\vec{A}$, lze tedy zapsat jako: \begin{equation} \label{rov:odsz} \vec{A}' = \vec{A}+k\cdot\vec{n} \end{equation} Obdobně lze provést i korekci na kruhovém oblouku. Pouze vektor $\vec{n}$ nyní není stejný pro počáteční i koncový bod a~je nutné jej pro každý bod určit sólo. Zkorigovaný oblouk je s~původním soustředný. \begin{figure}[h] \centering \begin{subfigure}[b]{0.30\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-lin-jedn.pdf} \caption{Korekce na úsečce} \end{subfigure} ~~~~~~~~~~~~~~~~~~~ \begin{subfigure}[b]{0.25\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-kruh-jedn.pdf} \caption{Korekce na oblouku} \end{subfigure} \caption{Znázornění základní korekce. Černou čarou je znázorněna naprogramovaná kontura (tedy trasa zadaná G-kódem), přerušovaně pak dopočtená korigovaná trasa.}\label{obr:kor-zak} \end{figure} Tuto korekci lze provést i na tečně navazujících úsecích se stejnou hodnotou korekce -- jelikož jsou tečné, jejich kolmý vektor v~koncovém bodě prvního a~počátečním bodě druhého úseku je stejný, tudíž zkorigované body splynou a~po korekci dostávám spojitou trasu. \subsubsection{Korekce pro úseky nenavazující tečně} Pokud bych na úseky, které na sebe tečně nenavazují použil výše uvedenou korekci, dostal bych nespojitou zkorigovanou trasu. V~tomto případě navázání je nutné rozlišit několik případů. Z~pohledu korekce je nutné hlavně rozlišit, zdali se jedná o vnitřní nebo vnější roh. Z~pohledu výpočtu je ještě nutné rozlišit, mezi jakými druhy úseků korekce nastává. Možné situace jsou znázorněny na obrázcích \ref{obr:kor-netec}. \begin{figure}[h] \centering \begin{subfigure}[b]{0.3\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-lin-dva.pdf} \caption{Korekce na úsečkách} \end{subfigure} ~ \begin{subfigure}[b]{0.3\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-kruh-dva.pdf} \caption{Korekce na obloucích} \end{subfigure} ~ \begin{subfigure}[b]{0.3\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-komb-dva.pdf} \caption{Kombinovaná korekce} \end{subfigure} \caption{Znázornění korekcí na tečně nenavazujících úsecích. Tlustou čarou je zde znázorněna naprogramovaná kontura (původní trasa zadaná G-kódem), přerušovaně pak dopočítaná korekce trasy.}\label{obr:kor-netec} \end{figure} Provedení odsazení vnějšího rohu je relativně jednoduché. Pro zkrácení obráběcího času se neprojíždí plné odsazení, nýbrž se roh opíše obloukem\cite{heid}. Tím je dosaženo zkrácení trasy z~$2r$ na $\frac{\pi}{2}r$. Zároveň se tím však zjednodušuje výpočet korigované trasy. Oba úseky stačí zpracovat stejně jako v~základním případě podle vztahu \ref{rov:odsz} a~mezi koncový odsazený bod prvního úseku a~počáteční odsazený bod druhého úseku vložit oblouk se středem v~koncovém neodsazeném bodu prvního úseku (popř. počátečním neodsazeném bodu druhého úseku, jelikož tyto body jsou totožné). Při obrábění vnějšího rohu nevznikají žádné nepřesnosti. Pro zpracování vnitřního rohu je třeba nalézt průsečík již zkorigovaných úseků. Jak lze vidět na obrázcích \ref{obr:kor-netec}, vzniká při obrábění vnitřního rohu nepřesnost, kterou nelze nijak korigovat. Lze ji pouze minimalizovat použitím menšího nástroje. Nejjednodušší situace nastává u dvou netečně navazujících úseček. Průsečík zde hledám tak, že si parametricky vyjádřím zkorigované úsečky (k~parametrickému vyjádření původní trasy přičtu $k\cdot\vec{n}$). Tato soustava vypadá následovně: \begin{equation} \begin{cases} \vec{P}=\vec{A}+p\cdot\vec{u}+k\cdot\vec{n} \\ \vec{P}=\vec{B}+q\cdot\vec{v}+k\cdot\vec{m}, \end{cases} \end{equation} kde $p$ a~$q$ jsou parametry, $u$ a~$v$ směrové vektory, $n$ a~$m$ přislušné normálové vektory a~$k$ je hodnota korekce nástroje. Porovnáním těchto dvou rovnic a~následným vyřešením získám hodnotu parametru, ze kterého lze jednoduše dopočítat průsečík, což je koncový a~počáteční bod příslušných zkorigovaných úseků. \begin{equation} p=\frac{-A_1+A_2+B_1-B_2+m_1-m_2-n_1+n_2}{u_1-u_2} \end{equation} Obdobně lze naléz i průsečík dvou oblouků -- pro zjednodušení výpočtu však všechny útvary posunu tak, aby střed jedné z~kružnici ležel v~bodě $[0;0]$ a~po provedení výpočtu výsledek posunu opačným směrem (zpět). Tímto posunem se mi výrazně zjednodušší výpočet, jelikož dostanu jednoduší středovou rovnici jedné z~kružnic: \begin{equation} \begin{cases} x^2+y^2=r_1^2 \text{~~~~~~~~--~rovnice zjednodušená posunem}\\ (x-m)^2+(y-n)^2=r_2^2, \end{cases} \end{equation} kde $x$ a~$y$ jsou souřadnice hledaného průsečíku $P[x;y]$, $m$ a~$n$ jsou souřadnice středu druhé kružnice $S[m;n]$, $r_1$ a~$r_2$ jsou zkorigované poloměry odsazených kružnic (k~poloměru původních oblouků zadaných G-kódem je přičtena korekce $k$). Řešením této soustavy je tento relativně složitý výraz: \begin{equation} P[x;y]\begin{cases} x = \frac{m^3+m\left(n^2+r_1^2-r_2^2\right)\pm\sqrt{n^2\left(-\left(m^2+n^2-r_1^2\right)^2+2r_2^2\left(m^2+n^2+r_1^2\right)-r_2^4\right)}}{2\left(m^2+n^2\right)}\\ y = \frac{n^4+n^2\left(m^2+r_1^2-r_2^2\right)\mp m \sqrt{n^2\left(-\left(m^2+n^2-r_1^2\right)^2+2r_2^2\left(m^2+n^2+r_1^2\right)-r_2^4\right)}}{2n\left(m^2+n^2\right)} \end{cases} \end{equation} Existují dva průsečíky těchto kružnic, bohužel se mi nepodařilo zjistit, jak předem rozlišit, který průsečík je ten mnou hledaný. V~mém kódu tedy počítám oba dva a~za správný považuji ten, který leží blíže původnímu napojení nekorigovaných oblouků. Poslední případ, který může nastat je vnitřní roh mezi obloukem a~úsečkou. Zde opět využiji parametrického vyjádření úsečky, které následně dosadím do obecné rovnice kružnice. Opět v~zájmu zjednodušení byl proveden posun tétou soustavy tak, aby střed oblouku ležel v~bodě $[0;0]$. \begin{equation} \begin{cases} \vec{P}=\vec{A}+p\cdot\vec{u}+k\cdot\vec{n} \\ x^2+y^2=r^2 \end{cases} \end{equation} Pro zjednodušení výrazu sloučím v~rovnici přímky konstantní výrazy v~konstantu $\vec{k}$: \begin{equation} \vec{A}+k\cdot\vec{n}=\vec{k} \end{equation} Po rozdělení rovnice přímky na dvě (pro $x$-ové a~$y$-ové souřadnice) a~jejich dosazení do rovnice kružnice, dostávám následující vztah pro parametr $p$: \begin{equation} p=\frac{\pm k_1u_1\pm k_2u_2\mp\sqrt{u_1^2(r^2-k_2^2)+2k_1k_2u_1u_2+u_2^2(r^2-k_1^2)}}{u_1^2+u_2^2}, \end{equation} z~něhož snadno dopočítám oba průsečíky. Stejně jako v~předchozím případě, i~zde jsem nepřišel na způsob, jak předem určit, který z~parametrů je mnou hledaný. Proto opět počítám oba dva a~vybírám bod, který je blíže průsečíku původních nekorigovaných úseků. \subsubsection{Korekce pro úseky se změnou její hodnoty} To, jak se má korekce chovat, pokud je nastavena, je relativně logické. Standard však nedefinuje, jak se má systém zachovat po spuštění korekce. Rozhodl jsem se respektovat chování systémů LinuxCNC\cite{emckor} a~Heidenhein\cite{heid}. Tyto systémy považují první pohyb po aplikování korekce za \uv{lead-in}\footnote{termín používaný v~dokumentaci LinuxuCNC\cite{emckor}, nenašel jsem adekvátní překlad, proto v~textu používám anglický výraz}. Tento pohyb by neměl být určen k~obrábění a~jeho cílem je aplikovat korekci -- na svém konci by měl být nástroj již ve správném odsazení od~původní trasy. Jako vzor chování stroje během lead-in pohybu jsem si vzal chování systému LinuxCNC (popsáno v~\cite{emckor}). Toto chování se však ukázalo v~jednom případě jako nevhodné (viz dále). Výše uvedené systémy (LinuxCNC, Heidenhein, Sinumerik), akceptují jako lead-in pohyb pouze lineární pohyb (funkci G01), avšak já jsem se rozhodl do mého systému zaimplementovat i podporu pro obloukový lead-in pohyb (funkce G02, G03). Celkem tedy mohou nastat čtyři případy: lead-in pohyb je lineární a~navazuje na vnitřní roh; lead-in pohyb je lineární a~navazuje na vnější roh; lead-in pohyb je obloukový a~korekce se zmenšuje; lead-in pohyb je obloukový a~korekce se zvětšuje. Nejjedodušším případem je lineární lead-in pohyb navazující na vnitřní roh. Zde jsem se také po konzultaci na diskusním fóru C-N-C.cz\footnote{\url{http://www.c-n-c.cz/viewtopic.php?f=47\&t=9669}} rozešel s~implementací LinuxuCNC. Implementace systémem LinuxCNC je znázorněna na nákresu \ref{nak:EMCk}. Logikou tohoto systému je, že nástroj nesmí odebrat materiál mimo zadanou konturu, proto raději zanechá neobrobenou část na kontuře\cite{emckor}. Po konzultaci na diskusním fóru jsem si stanovil pro můj systém stejnou prioritu jako systém Heidenheim\cite{heid} -- systém musí, co nejpřesněji obrobit zadanou konturu. Na nákresech \ref{nak:leadin1} je znázorněno chování mého systému. Na nákresu b si lze povšimnout, že systém porušuje konturu lead-in pohybu. Jak jsem však již dříve zmínil, lead-in pohyb by neměl obrábět. Proto porušení jeho kontury nevadí. Ve správně napsaném programu by v~prostoru lead-in pohybu neměl být žádný materiál. Z této úvahy vyplývá, že pro zpracování lineárního lead-in pohybu stačí počáteční bod zpracovat stejně, jako by se jednalo o~klasickou konturu, a~jeho koncový bod má souřadnice: \begin{equation} \vec{P} = \vec{B}+k\cdot\vec{n}, \end{equation} kde $\vec{B}$ je počáteční bod navazujícího úseku, $\vec{n}$ normálový vektor navazujícího úseku a~$k$ je hodnota korekce. \begin{figure}[h] \centering \includegraphics[width=0.6\textwidth]{img/comp02.png} \caption{Implementace chování lineárního lead-in pohybu v~systému LinuxCNC. Převzato z~\url{http://www.linuxcnc.org/docs/2.4/html/comp02.png}}\label{nak:EMCk} \end{figure} \begin{figure}[h] \centering \begin{subfigure}[b]{0.42\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-leadin1.pdf} \caption{Mé řešení situace z~dokumentace LinuxCNC} \end{subfigure} ~~~ \begin{subfigure}[b]{0.3\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-leadin2.pdf} \caption{Obecný případ} \end{subfigure} \caption{Implementace lineárního lead-in pohybu v~mém systému. Tlustá čára znázorňuje trasu zadanou G-kódem, přerušovaná pak dopočtenou korekci.}\label{nak:leadin1} \end{figure} Pokud lineární lead-in pohyb navazuje na vnější roh, je situace obdobná jako v~případě klasického korigovaného pohybu -- vnější roh je obroben vloženým obloukem, který jej opisuje. Pokud bych však na počátku a~na konci lead-in pohybu použil jinou hodnotu korekce, nenavazoval by tento oblouk tečně, tudíž by těsně před ním musel stroj zabrzdit. Ideálně by měl oblouk navazovat tečně, jak znázorňuje obrázek \ref{nak:leadin3}. Výpočet pozice bodu C jsem v~tomto případě realizoval skrz vektor odsazení $\vec{o}$. Mé řešení vychází z~pravoúhlého trojúhelníku ABC. Pokud budu uvažovat vektor mezi body A a~B jako~$\vec{u}$, mohu úhel $\alpha$ vyjádřit následovně: \begin{equation} \vec{u}\cdot\vec{o}=\cos\alpha \end{equation} Jelikož je délka vektoru $\vec{o}$ rovna korekci $\abs{k}$, tak můžu vyjářit $\cos\alpha$ i pomocí těchto délek. Získám tedy soustavu rovnic: \begin{equation} \begin{cases} \vec{u}\cdot\vec{o} = \frac{\abs{k}}{\abs{\vec{u}}} \\ \abs{\vec{o}}=\abs{k} \implies o_1^2+o_2^2=k^2 \end{cases} \end{equation} Pokud označím $\abs{\vec{u}}=d$, řešením této soustavy je \begin{equation} \vec{o} \begin{cases} o_1=\frac{dku_1\pm\sqrt{d^2k^2u_2^2 \left(d^2\left(u_1^2+u_2^2\right)-1\right)}}{d^2\left( u_1^2+u_2^2\right)} \\ o_2=\frac{dku_2^2\mp u_1\sqrt{d^2k^2u_2^2\left(d^2\left(u_1^2+u_2^2\right)-1\right)}}{d^2u_2(u_1^2+u_2^2)} \end{cases} \end{equation} Tento výraz je použitelný i pro situaci, kdy původní korekce nebyla nulová, či dokonce měla opačný směr. Pouze se jako bod A vezme již jeho zkorigovaná hodnota. Nadále zde zůstává pravoúhlý trojúhelník. U~tohoto výrazu se mi podařilo určit znaménko ve vzorci pro určení požadovaného řešení. Pokud se nová korekce nachází na stejné straně od~kontury jako předcházející, je ve výrazu pro $o_1$ plus a~ve výrazu $o_2$ mínus. Pokud se původní korekce nacházela na opačné straně od~kontury, jsou znaménka přesně opačná. \begin{figure}[h] \centering \includegraphics[width=0.5\textwidth]{img/korekce-leadin3.pdf} \caption{Znázornění implementace lead-in pohybu, který navazuje na vnější roh.}\label{nak:leadin3} \end{figure} Jak jsem zmínil v~úvodu této sekce, rozhodl jsem se do mého systému zaimplementovat podporu pro obloukový lead-in pohyb. Jsou dvě možnosti korekce -- výsledkem první je oblouk o menším poloměru, výsledkem druhé je oblouk o větším poloměru. Obě situace znázorňují nákresy~\ref{nak:leadinobl}. Při korekci na oblouk o větším poloměru přidám do trasy úsečku délky $k$ ve směru tečném k~počátečnímu bodu oblouku. Následně za tento lineární úsek přidám oblouk o původním poloměru a~úhlové délce 90~$^\circ$. V~jeho koncovém bodě se již nástroj dostává do zkorigované pozice a~následuje oblouk již o zkorigovaném poloměru. Při korekci na oblouk o menším poloměru je postup opačný -- pohyb začíná obloukem se~zkorigovaným poloměrem a~úhlové délce 90~$^\circ$. Za něj vkládám lineární úsek o délce $k$. V~tomto bodě se již nástroj nachází ve~zkorigované pozici a~následuje zkorigovaný původní oblouk kontury. Z~mnou použitého řešení, jak implementovat oblouk jako lead-in pohyb, vyplývá jedno omezení -- jako lead-in pohyb nelze použít oblouk, který má úhlovou délku kratší než 90~$^\circ$. Na obrázku \ref{nak:korvys} si lze prohlédnout výstup již implementované korekce průměru v~mém systému. Korekce je znázorněna bílou čarou. Jedná se o ukázkový příklad, kdy byly do jednoduchého programu v G-kódu na náhodná místa přidány funkce G40, G1 a~G42. \begin{figure}[h!] \centering \begin{subfigure}[b]{0.40\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-leadin4.pdf} \caption{Korekce na větší poloměr} \end{subfigure} ~~~ \begin{subfigure}[b]{0.40\textwidth} \centering \includegraphics[width=\textwidth]{img/korekce-leadin5.pdf} \caption{Korekce na menší poloměr} \end{subfigure} \caption{Znázornění implementace obloukového lead-in pohybu}\label{nak:leadinobl} \end{figure} \begin{figure}[h!] \centering \includegraphics[width=0.7\textwidth]{img/korekce.png} \caption{Ukázka zpracování korekce průměru nástroje mým systémem. Na počátku kontury je nastavena korekce doprava ve směru obrábění, na třetím vrcholu hvězdy, je změněna korekce na opačnou (doleva). Za čtvrtým vrcholem je korekce vypnuta.}\label{nak:korvys} \end{figure} \subsection{Korekce délky} Oproti korekci poloměru nástroje je implementace korekce délky nástroje jednoduchou záležitostí. Jedná se totiž pouze o korekci ve směru jedné osy (Z), narozdíl od~dvou os (X a~Y) v~případě korekce poloměru nástroje. Počáteční bod každého úseku je nutno zkorigovat o korekci nastavenou pro úsek předcházející a~koncový bod každého úseku je nutno zkorigovat korekcí nastavenou pro aktuálně zpracovávaný úsek. Zde je nutno podotknout, že tato korekce zatím nebyla v~mém systému plně implementována a~může způsobovat chyby. Na vině je implementace kruhové interpolace, která není v~souladu se~standardem G-kódu (viz kapitola \ref{kap:oblinter}). Pokud by nastala změna délkové korekce na kruhovém oblouku, měl by počáteční a~koncový bod jiné Z-ové souřadnice. Dle standardu by se tato změna měla interpolovat po šroubovici, avšak můj systém se pokusí nalézt kruhový oblouk v~obecné rovině, který tomuto zadání vyhovuje. \section{Výpočet rychlosti}\label{kap:rychlost} Poslední činností, kterou vykonává třída {\tt GCodeInterpreter} před tím, než vytvoří seznam příkazů pro interpolátor, je dopočtení rychlostí pro každý úsek. Rychlost na každém úseku je charakterizována počáteční, cílovou a~brzdnou rychlostí a~také maximálním zrychlením pro rozjezd a~brzďění. Na základě těchto údajů je schopen interpolátor provést pohyb. {\tt GCodeInterpreter} na základě posuvu v~G-kódu, maximálního zrychlení a~ryvu zadaného v~konfiguračním souboru a~tvaru úseku určuje vhodnou rychlost. Výpočet rychlosti probíhá pro každý úsek samostatně. Pro vypočtení rychlosti celé trasy tedy procházím každý prvek trasy od~konce po začátek. Při provádění výpočtu rychlosti od~konce je totiž menší šance na vznik cyklické závislosti -- může se stát, že vypočtená počáteční rychlost je pro daný prvek příliš velká, a~musím se proto tedy vrátit na $n$ předcházejících prvků a~ty znovu spočítat. Při procházení trasy od~konce však tato situace nastává pouze zřídka, např. při seskupení krátkých oblouků o malém poloměru za sebou. Výpočet rychlosti daného prvku má na starosti metoda {\tt PathPart::ComputeSpeeds}. V následujícím textu používám pojmy předcházející a~následující úsek. Ačkoliv úseky zpracovávám v~opačném pořadí (od posledního po první), pojmem následující úsek mám na mysli úsek, který následuje během pohybu (tedy pořadí od~prvního po poslední), resp. pojmem předchozí úsek mám na mysli úsek, který předchází během pohybu. Pro jednotlivé rychlosti platí, že počáteční a~koncová rychlost je menší nebo rovna požadované. Tato podmínka vychází z~logiky, že posuvem je dána maximální možná rychlost na daném úseku, která nesmí být překročena. A jelikož posuv určuje požadovanou rychlost, tak počáteční i koncová rychlost musejí být menší. Postup výpočtu, který je popsán v~následujících odstavcích je také znázorněn na schématu~\ref{nak:vypocet}. \begin{figure}[h!] \centering \includegraphics[width=0.7\textwidth]{img/vypocetrychlosti.pdf} \caption{Znázornění postupu výpočtu rychlosti daného úseku.}\label{nak:vypocet} \end{figure} V každém úseku před zpracováním je počáteční a~koncová rychlost nastavena na nulu a~cílová rychlost je nastavena dle posuvu. Při výpočtu se koncová rychlost úseku nastaví na počáteční rychlost následujícího úseku. Poté se testuje, zda-li předcházející úsek navazuje tečně. Pokud ano, je současnému prvku nastavena počáteční rychlost rovna požadované rychlosti předchozího prvku (za podmínky, že je menší nebo rovna požadované rychlosti aktuálního prvku). Nyní je třeba otestovat, zda-li je možno požadovaných rychlostí dosáhnout na aktuálním úseku (jeho délce). Pro tento test je nutno spočítat rozjezdovou a~brzdnou dráhu (podrobněji popsáno v~kapitolách \ref{kap:drahlin} a~\ref{kap:rych-obl}). Test rychlosti probíhá v~několika úrovních. Pokud je počáteční rychlost rovna požadované a~zárověň jsou tyto rychlosti větší než koncová, je proveden test navíc - je totiž nutné určit, zda-li je možno na zadaném úseku ze zadané počáteční rychlosti zbrzdit na koncovou. Pokud ne, je nutné vhodně upravit počáteční (a požadovanou) rychlost. Následně se určí jak brzdná, tak rozjezdová dráha a~zkoumají se jejich hodnoty. Pokud je brzdná dráha nulová a~tedy požadovaná rychlost rovna koncové, zkoumá se, zda-li je možno požadované rychlosti dosáhnout. Pokud ne, je třeba snížit koncovou rychlost. Jelikož jsem změnil hodnotu koncové rychlosti, musím změnit hodnotu počáteční rychlosti následujícího úseku, který již byl zpracován, a~znovu jej zpracovat. Tento případ však nenastává často (to je důvod, proč zpracovávám úseky od~konce). Pokud brzdná dráha není nulová, nastává obecný případ. Pokud je součet obou drah větší než délka úseku, je nutné pomocí bisekce najít novou požadovanou rychlost. Tím je výpočet rychlosti na úseku vyřešen. \subsection{Určení rozjezdové a~brzdné dráhy při lineárním pohybu}\label{kap:drahlin} Určení délky rozjedové (brzné) dráhy při lineárním pohybu má na starosti metoda {\tt PathPartLine::GetSFromStartMovement}, resp. {\tt GetSFromEndMovement}. Tato metoda volá metodu {\tt PathPartLine::GetATFromMovement}, která na základě předané počáteční a~požadované rychlosti dopočte hodnoty maximální amplitudy zrychlení $A$ a~doby pohybu $T$ při zadaných omezeních. Poté z~těchto hodnot podle vztahu \ref{rov:lindrah} z~kapitoly \ref{kap:linrozbrzd} (resp. vztahu \ref{rov:lindrahb} pro brzdění) dopočte dráhu. Zároveň uloží vypočtené zrychlení do interního atributu {\tt Astart} (resp. {\tt Aend}) třídy {\tt PathPartMovable}, které je poté součástí příkazu pro interpolátor. Metoda {\tt GetATFromMovement} prvně testuje, zda-li nejsou předané rychlosti stejné. Pokud ano, vrátí nulový čas a~maximální možné zrychlení. Pokud ne, pokusí se prvně dopočítat čas pohybu za omezení ryvem. Využívá k~tomu vztah \ref{rov:linTJ}. Z~vypočteného času poté určí potřebnou amplitudu zrychlení. Pokud je tato amplituda menší než konfigurací nastavená maximální hodnota, vrátí funkce již vypočtený čas a~příslušné zrychlení. Pokud by při omezení ryvem bylo přesáhnuto maximální zrychlení, je dopočten čas pohybu za omezení zrychlením dle vztahu \ref{rov:lincas}. Poté je navrácen tento nově vypočtený čas a~maximální přípustná hodnota zrychlení. \subsection{Určení rozjezdové a~brzdné dráhy při pohybu po oblouku}\label{kap:rych-obl} I u obloukového pohybu jsou použity pro výpočet drah metody {\tt GetSFromStartMovement} a~{\tt GetSFromEndMovement}. Tyto metody volají přetíženou metodu {\tt GetATFromMovement}, která vrací příslušné hodnoty $A$ a~$T$ pro pohyb na oblouku pro předané rychlosti za daných omezení. Metoda {\tt GetSFromStartMovement} (resp. {\tt GetSFromEndMovement}) využívá pro určení dráhy stejné vztahy jako její obdoba pro lineární pohyb. Metoda {\tt GetATFromMovement} prvně zkontroluje, zdali je možno požadované rychlosti dosáhnout -- podmínka vyplývající ze~vztahu \ref{rov:oblpodm}. Tato podmínka je zde pro debuggovací účely -- již v dřívějších krocích by mělo být zajištěno, aby požadovaná rychlost nepřekračovala tuto mezní hodnotu. Funkce dále samostatně řeší případ, kdy je požadovaná rychlost rovna mezní hodnotě. V~tomto případě prvně počítám hodnoty $A$ a~$T$ za omezení ryvem s~jeho maximální hodnotou v~$t=0$, jelikož výpočet s~maximem v~$t=\frac{3T}{4}$ je výpočetně náročnější (je řešen numericky), proto jej počítám pouze v~krajním případě. K~výpočtu těchto hodnot jsou použity vztahy \ref{rov:linTJ} a~\ref{rov:oblA}. Na základě spočítaného $A$ a~$T$ dopočítám hodnotu ryvu v~čase $t=\frac{3T}{4}$ (dle prvního vztahu ze soustavy \ref{rov:oblJT}). Pokud je tato hodnota vyšší než hodnota v~čase $t=0$, musím omezit pohyb ryvem s~maximem v~$t=\frac{3T}{4}$. Tento numerický výpočet provádí metoda {\tt ComputeATLimitedByJerk} popsaná v~sekci \ref{kap:newton}. Pokud požadovaná rychlost nedosahuje mezní hodnoty, jsou hodnoty $A$ a~$T$ určeny obdobně. Pouze je na závěr ještě nutno provést test, zda-li nebylo překročeno maximální zrychlení. Tento test není nutno v~mezním případě provádět, protože maximum zrychlení automaticky nastává v~čase $t=T$ a~dosahuje přesně mezní hodnoty. Maximum celkového zrychlení zde nastává v~čase $t=\frac{T}{2}$ a~pro jeho výpočet je použit vztah \ref{rov:oblak}. Pokud je překročeno, je nutné pohyb limitovat zrychlením. Příslušné hodnoty $A$ a~$T$ v~tomto případě určují vztahy \ref{rov:oblt2a}. \subsubsection{Řešení soustavy rovnic Newtonovou metodou}\label{kap:newton} Pro výpočet rychlosti, kdy maximum ryvu nastává v~čase $t=\frac{3T}{4}$, je třeba vyřešit soustavu rovnic \ref{rov:oblJT}. Tato soustava však není řešitelná obecně, je třeba využít numerické metody. K~jejímu řešení v~mé implementaci používám Newtonovu metodu, jelikož se snadno aplikuje na soustavu rovnic a~rychle konverguje\cite{wiki:newton}. Postup, jak řešit soustavu rovnic Newtonovou metodou jsem čerpal z~dokumentu~\cite{newton}. Princip řešení soustavy rovnic Newtonovou metodou spočívá v~převedení hledaného kořene na vektor a~nahrazení derivace příšlušné funkce maticí parciálních derivací daných funkcí\cite{newton}. Soustavu rovnic tedy musím převést na funkci\cite{newton} \begin{equation} \vec{f(x)}= \begin{pmatrix} f_1(x_1, x_2,~\hdots~, x_n) \\ f_2(x_1, x_2,~\hdots~, x_n) \\ \vdots \\ f_n(x_1, x_2,~\hdots~, x_n) \end{pmatrix}, \end{equation} jejíž derivací je matice parciálních derivací\cite{newton} \begin{equation} D\vec{f(x)}= \begin{pmatrix} \frac{\partial f_1}{\partial x_1}(x) & \frac{\partial f_1}{\partial x_2}(x) & \cdots & \frac{\partial f_1}{\partial x_n}(x)\\ \frac{\partial f_2}{\partial x_1}(x) & \frac{\partial f_2}{\partial x_2}(x) & \cdots & \frac{\partial f_2}{\partial x_n}(x)\\ \vdots & \vdots & \ddots & \vdots\\ \frac{\partial f_n}{\partial x_1}(x) & \frac{\partial f_n}{\partial x_2}(x) & \cdots & \frac{\partial f_n}{\partial x_n}(x)\\ \end{pmatrix} \end{equation} Mnou hledaným kořenem soustavy je vektor $\vec{p}$, pro nějž platí\cite{newton} \begin{equation} \vec{f(p)}=0 \end{equation} Rekurzivní vztah pro kořen soustavy tedy potom vypadá následovně\cite{newton}: \begin{equation} \vec{p_{n+1}}=\vec{p_n}-\left(D\vec{f(p_n)}\right)^{-1}\vec{f(p_n)} \end{equation} Implementace mého konkrétního problému Newtonovou metodou není složitá, jelikož mám pouze soustavu dvou rovnic, tudíž pracuji s~dvojrozměrným vektorem a~maticí 2$\times$2. Tím se mi výrazně zjednodušuje hledání inverzní matice parciálních derivací. Pokud mám matici 2$\times$2 \begin{equation} A=\begin{pmatrix} a & b \\ c & d \end{pmatrix}, \end{equation} tak její inverzní matici můžu vypočítat následovně\cite{inverze}: \begin{equation} A^{-1}=\frac{1}{\det A}\begin{pmatrix} d & -b \\ -c & a \end{pmatrix}=\frac{1}{ad-bc}\begin{pmatrix} d & -b \\ -c & a \end{pmatrix} \end{equation} Následné přepsání do kódu už je pouze rutinní záležitost. Díky jednoduchosti hledání inverzní matice není nutno použít speciální knihovnu pro práci s~maticemi a~celý problém lze naimplementovat s~použitím několika málo proměnných. Soustava \ref{rov:oblJT} reprezentuje v~kladných hodnotách (jiné nemají pro můj problém smysl) dvě křivky se dvěma průsečíky, jak lze vidět na grafu \ref{graf:newton}. Rovnice má tedy dva reálné kořeny v~kladných hodnotách (další reálné řešení se nachází v~záporném čase). První řešení s~nižší hodnotou času pohybu $T$, vždy přesahuje maximální zadanou hodnotu $A$ (experimentálně ověřeno na vzorku reálných dat). Za relevatní tedy považuji druhé řešení s~vyšší hodnotou periody $T$. Odtud se odvíjí můj odhad kořene pro Newtonovu metodu. Jako počáteční hodnotu volím dostatečně velkou hodnotu doby pohybu $T$ -- konkrétně stonásobek hodnoty vypočtené pomocí vztahu, který omezuje ryv v~čase $t=0$. Tím dosáhnu toho, že se k~hodnotě blížím zprava a~jako první naleznu pro mě relevantní řešení. \begin{figure}[h] \centering \includegraphics[width=0.5\textwidth]{img/graf_newton.pdf} \caption{Graf znázorňující řešení soustavy rovnic \ref{rov:oblJT}. Vodorovná osa reprezentuje dobu pohybu $T$, svislá amplitudu zrychlení $A$.}\label{graf:newton} \end{figure} \subsection{Určení nové rychlosti pomocí bisekce}\label{kap:bisekce} Jak jsem zmínil v~předcházejícím textu, pro výpočet mezní rychlosti jsem použil metodu bisekce. Bisekce je též známá jako metoda půlení intervalů\cite{bisekce}. Tato metoda se zpravidla používá pro hledání kořene funkce mezi dvěmi zadanými mezemi. Metoda pracuje tak, že určuje, v~které polovině daného intervalu se nachází hledaná hodnota. Jakmile určí správný interval, rekurzivně se opakuje a~opět půlí tentokrát již menší interval\cite{bisekce}. Rekurze je ukončena, jakmile je buď nalezena správná hodnota, nebo je dosaženo požadované přesnosti. V praxi však není hojně používána, jelikož konverguje relativně pomalu (např. oproti Newtonově metodě). Avšak pro mě má bisekce výhodu v~tom, že ji lze použít nejen pro jednoduché hledání kořene, ale i k~hledání parametru v~určitém vztahu s~relativně složitou podmínkou, což je právě můj případ pro hledání rychlosti. Ačkoliv jsem v~kapitole \ref{kap:linryv} uvedl vztah, ze kterého lze Newtonovou metodou určit maximální dosažitelnou rychlost pro lineární pohyb, nepoužívám jej v~mém řídicím systému. Problémem tohoto řešení je, že je nutno vytvořit 4 varianty tohoto vztahu -- pro různé kombinace omezení rozjezdu a~brzdění omezeného ryvem či zrychlením. Také je nutné určit, který vztah je třeba použít. Výsledkem byl složitý kód, který měl podobnou časovou náročnost jako má současná implementace (avšak měl o něco lepší přesnost -- zde jsem generoval rychlost s~přesností 0,01~mm$\cdot$s$^{-1}$, nyní generuji rychlost s~přesností 0,1~mm$\cdot$s$^{-1}$, což je více než dostatečné). Situace pro výpočet rychlosti na oblouku byla ještě složitější a~počet variant zde prudce vzrostl. Proto jsem se rozhodl přiklonit k~metodě bisekce. Bisekci v~kódu implementuji pomocí cyklu. Tento cyklus probíhá, dokud velikost intervalu je větší než $0,1$ a~nebylo provedeno více než 10 iterací. V~tomto cyklu spočítám dráhu pro rychlost v~polovině daného intervalu. Na základě této dráhy a~podmínky rozhodnu, který interval budu dále půlit. Jako jedna hranice počátečního intervalu slouží nyní nevyhovující požadovaná rychlost a~jako druhá slouží počáteční nebo koncová rychlost -- podle toho, která z~nich je větší. Na reálných programech je rychlost s~přesností na 0,1~mm$\cdot$s$^{-1}$ dosažena během 4 iterací. \section{Komunikace s interpolátorem} Pro komunikaci s~interpolátorem přes USB používám knihovnu WinUSB. Tato knihovna obsahuje driver, který mi umožní z~user space operačního systému otevřít jednotlivé endpointy a~zapisovat na ně, popř. z~nich číst. Odprostí mě tak od~nutnosti psát vlastní driver v~kernel space\cite{winusb}. Ačkoliv je WinUSB dílem Microsoftu, musí tento ovladač projít stejným certifikačním procesem jako jakýkoliv jiný driver. To je pravděpodobně způsobeno tím, že pro použití musím upravit INF soubor ovladače, čímž poruším integritu dat a~již dříve udělený certifikát by se stal neplatným. Problematika ovladačů a~jejich certifikování na Windows je poměrně rozsáhlá a~bohužel se mi do ní nepodařilo zcela proniknout. Je možné, že mnou vytvořený certifikát a~vlastně i celý balíček ovladače nesplňuje všechny požadavky Microsoftu, avšak podařilo se mi jej nainstalovat na několika počítačích a~různých systémech (Windows XP, Windows 7 x86 i Windows 7 x64). \subsection{Tvorba balíčku ovladačů WinUSB} Aby WinUSB spolupracoval s~mým zařízením, je nutné upravit vzorový INF soubor\cite{winusb}. Tento soubor slouží jako popis driveru, resp. zařízení. V~sekci {\tt Version} je nutné vyplnit jméno poskytovatele driveru, uvést třídu zařízení, verzi a~datum vydání driveru, jeho GUID a~jako poslední také jméno souboru katologu. Katalog obsahuje hash všech souborů, které ovladač obsahuje a~tím slouží k~ověření, zda-li soubory nebyly modifikovány. K~ověření pravosti katalogu slouží certifikát, jak zmiňuji níže. GUID (Globally unique identifier) je 128 bitové číslo\cite{wiki:guid}. Toto číslo slouží jako identifikátor COM objektu, na základě něhož si je možné z~mé aplikace tento objekt vyvolat a~skrz něj komunikovat se zařízením. GUID se generuje náhodně -- vzhledem k~velikosti čísla je šance, že by se podařilo vygenerovat dvě stejná GUID velmi malá\cite{winusb}. V~sekci {\tt Manufacturer} je třeba nastavit identifikátor, resp. cestu k zařízení, pro jednotlivé platformy. Tato cesta je ve tvaru {\tt USB\textbackslash VID\_xxxx\&PID\_yyyy}, kde xxxx je příslušné vendor ID zařízení a~yyyy je příslušné product ID zařízení. Úpravou sekce {\tt SourceDisksNames} je možné docílit vlastního pojmenování instalačních disků (toto jméno se zobrazuje během instalace ovladače). Poslední sekcí, kterou je třeba upravit, je sekce {\tt Strings}. Touto sekcí lze nastavit jméno zařízení tak, jak se bude zobrazovat ve správci zařízení a~specifikovat výrobce, kterého lze zobrazit ve vlastnostech daného zařízení. Jakmile je INF soubor upraven, je třeba vytvořit katalog ovladače\cite{kernel}. Katalog zajišťuje integritu celého ovladače a~umožňuje systému před instalací ověřit, zda-li nebyl driver modifikován. Microsoft poskytuje utilitu zvanou Inf2Cat\cite{kernel}, která na základě předaného INF souboru vytvoří katalog všech souborů, na které INF soubor odkazuje. V~případě problémů je možné také použít utilitu MakeCat\cite{kernel}, která vytvoří katalog na základě předaného seznamu souborů. Následně je nutno katalog podepsat\footnote{Windows XP podepsání nevyžadují, avšak zobrazí varování o nedůvěryhodném driveru. Windows Vista a~výše však již podepsání vyžadují, jinak odmítnou ovladač nainstalovat.}, čímž je zajištěna jeho pravost. Toto podepsání lze provést pomocí utility SingTool\cite{kernel}. Pro podepsání je však třeba mít certifikát. Aby driver při instalaci do systému prošel bez chyby ověřovacím procesem, je třeba mít certifikát vydaný institucí schválenou Microsoftem\cite{kernel}. Tato možnost je relativně drahá a~hlavně pro fyzickou osobu nedostupná. Naštěstí Microsoft umožňuje pro vývojové účely vygenerovat si vlastní certifikát, který umožní katalog podepsat\cite{kernel}. Avšak při instalaci ovladače se zobrazí, že certifikát pochází od~nedůvěryhodné osoby. Tento certifikát je možné vygenerovat si pomocí utility MakeCert\cite{kernel}. Mnou upravený balíček ovladačů WinUSB, společně s~katalogem a~příslušnými utilitami lze nalézt na přiloženém CD. \subsection{Třída MyDevice} Třída {\tt MyDevice} reprezentuje rozhraní pro komunikaci s~interpolátorem. Zaobaluje veškerou funkcionalitu okolo WinUSB -- stará se o inicializaci zařízení (otevření endpointů), jeho správnou deinicializaci při ukončení aplikace a~reaguje na jeho připojení čí odpojení. Posílání zpráv do interpolátoru je realizováno metodou {\tt SendRawData}, která přebírá data k odeslání buď jako ukazatel na pole a~jeho délku nebo jako {\tt std::vector}. Tato metoda zkontroluje, zda-li je zařízení připojeno a~případně odešle data do interpolátoru pomocí funkce {\tt WinUsb\_WritePipe} přes příslušný endpoint. Jelikož je obrovské množství zpráv, které je možné do zařízení odeslat, nedefinuje tato třída pro každou zprávu speciální metodu, ale raději poskytuje obecnou metodu pro zápis libovolných dat. V~mém návrhu existuje pouze několik málo speciálních zpráv (požadavků, na které je očekávána odpověď), pro které je definována speciální metoda. Tyto požadavky jsou popsány v~následujícím odstavci. Třída {\tt MyDevice} také uchovává objekt třídy {\tt ReceiveFromDevice}, který se stará o příchozí data. Tato třída je popsána v~následující podkapitole. Ve spolupráci s~ní definuje třída {\tt MyDevice} speciální metody {\tt GetFreeSpaceInQueue}, {\tt GetLastProcessedItemStack}, {\tt GetCurrentlyProcessed-ItemStack}. Tyto metody odešlou příslušnou zprávu pro intepolátor (dle svého názvu) a~čekají na odpověď. Jsou tedy definovány jako blokující (proto by neměly být volány ve vláknu uživatelského rozhraní). Jako návratovou hodnotu mají již zpracovanou odpověď na danou zprávu. \subsection{Třída ReceiveFromDevice} Třída {\tt ReceiveFromDevice} je potomkem třídy {\tt wxThread} -- reprezentuje tedy vlákno. Uvnitř tohoto vlákna je v~nekonečném cyklu volána blokující funkce {\tt WinUsb\_ReadPipe}, která čte z~daného endpointu. Tato třída skládá dohromady zprávy rozdělené na více paketů stejně jako to dělá interpolátor (popsáno v~kapitole \ref{kap:protokol}). Pro každou přijatou zprávu pak volá metodu {\tt ProcessMessage}. V~této metodě je definováno zpracování všech možných zpráv. \subsection{Třídy ProgramControl a~ProgramRun} Nejdůležitějším typem komunikace je odesílání příkazů interpolátoru do jeho komponenty {\tt CommandStack}, odkud jsou postupně vykonávány. Další důležitou funkcí je zpětná vazba o poloze stroje, která je následně zobrazována uživateli. Tyto operace má na starosti třída {\tt ProgramRun}, která je ovládána třídou {\tt ProgramControl}. Třída {\tt ProgramRun} je potomkem třídy {\tt wxThread} -- reprezentuje tedy vlákno. Proto jsem pro interakci tohoto vlákna se zbytkem aplikace implementoval třídu {\tt ProgramControl}. Třída {\tt ProgramControl} při spuštění vykonávání programu v~G-kódu vytvoří novou instanci třídy {\tt ProgramRun}. Tuto třídu, respektive vlákno, poté ovládá pomocí příznaků. Příznak může mít tři stavy -- stav signalizující běžící program, stav signalizující pozastavení vykonávání a~stav zastavení programu. Během běhu programu třída {\tt ProgramRun} se neustále dotazuje interpolátoru na volné místo v~zásobníku a~pokud nějaké volné místo je, odešle patřičný počet příkazů do interpolátoru. Zároveň také posílá dotaz na aktuální pozici stroje. Tento dotaz je posílán v~neblokujícím režimu, jelikož jeho výsledek není nutný pro další běh vlákna (na~rozdíl od~dotazu na volné místo). Odpověď na pozici stroje zpracuje třída ReceiveFromDevice, která zajistí předání dat na správné místo tak, aby byla pozice stroje uživateli znázorněna. Pokud je běh programu pozastaven, pošle vlákno interpolátoru příkaz pro zastavení aktuálně vykonávané činnosti a~čeká na znovuobnovení běhu programu. Pokud je program zastaven, odešle vlákno interpolátoru příkaz pro zastavení aktuálně vykonávané činnosti a~příkaz pro vyprázdnění zásobníku příkazů. Poté se vlákno ukončí. \documentclass[../thesis.tex]{subfiles} \newcommand{\W}{\mathcal{W}} % world %\newcommand{\S}{\mathcal{S}} % Sensors/Inputs \def\approxindep{\mathrel{% \mathchoice{\APPROXINDEP}{\APPROXINDEP}{\scriptsize\APPROXINDEP}{\tiny\APPROXINDEP}% }} \def\APPROXINDEP{{% \setbox0\hbox{$\independent$}% \rlap{\hbox to \wd0{\hss$\sim$\hss}}\box0 }} \begin{document} % , Cornell Tech \\ % , Carnegie Melon University \\ % , ICSI %} \paragraph{Abstract} Many privacy and data protection policies stipulate restrictions on the flow of information based on that information's original source. We formalize this concept of privacy as Origin Privacy. This formalization shows how information flow security can be represented using causal modeling. Causal modeling of information security leads to general theorems about the limits of privacy by design as well as a shared language for representing specific privacy concepts such as noninterference, differential privacy, and authorized disclosure. These considerations raise questions for future work about whether policies should be design with respect to the feasibility of automating their enforcement. %\keywords{information flow security, Bayesian networks, noninterference, differential privacy, causality} \section{Introduction} \label{sec:orgheadline2} %\mct{The introduction is missing some high-level vision. Why are you focusing on origins? What limitation exists in the prior work to require this work?} % sb: will hold off on revising since in future versions we may drop % 'origin' angle all together Many policies (e.g. HIPAA, GLBA, FERPA, and Executive Order 13526 in the United States and the GDPR in the European Union) place restrictions on the collection, flow, and processing of personal information. When engineers build technical systems that collect and use personal data, they are under business and social pressure to translate prescriptive privacy policies, fitted to their case by lawyers, ethicists, and other policy-makers, into engineering requirements \cite{barth07csf,fisler2010embracing,swire14iapp,sen14sp}. The goal of engineering a privacy policy is to enable automated enforcement of some or all of the policy. This reduces the cost of protecting privacy. To automate enforcement of a privacy policy, that policy must first be translated into a machine-readable language with a precise syntax and semantics. Prior research has explored the feasibility of translating classes of privacy clauses into formal logic for enforcement. Some attempt to formalize a wide range of policies, such as those expressible within the framework of Contextual Integrity \cite{barth07csf,shvartzshnaider2017vaccine}. Others focus on particular kinds of clauses, such as those restricting information based on its purpose \cite{tschantz13esorics} or its use \cite{datta2017use}. This article is concerned with clauses in privacy policies that restrict information based on its \emph{origin}. We consider the origin of data to be the processes that created or transmitted that data to the system governed by the privacy policy, that is, the data's provenance. Section~\ref{sec:policy} will show how existing policies motivate this work by defining restricted classes of information in terms of the processes that originated them. This policy analysis reveals that origin clauses are most often used to identify a broad class or type of information that is then subject to restrictions or exemptions. In addition, information topic (what the information refers to or is about) is also frequently used alongside information origin. We show that the distinction between information origin and information topic is subtle but important for the purposes of determining the conditions under which a formalized policy can be enforced. In Section~\ref{sec:ontology}, we derive, from the policies analyzed, a general ontology of systems, processes, and messages. This ontology is intended to bridge between policies as intuitively articulated in natural language and the mathematical formalisms we will use in the rest of the paper. Using this ontology, we propose Origin Privacy as a framework for understanding privacy requirements and the knowledge necessary to enforce them by design. Section~\ref{sec:causality} shows how this informal specification can be mapped to a well established formal representation of causal models~\cite{pearl1988probabilistic}. In addition to presenting the formal theory, we show that causal modeling makes clear distinctions between two elements of information flow that are sometimes conflated: causal flow and nomic association, where ``nomic'' means law-like, or regular. These two aspects of information flow correspond to information origin and information topic. In Section~\ref{sec:security}, we combine the ontology with causal modeling to develop the Embedded Causal System (ECS) model. This model represents a computational system embedded in a larger environment. This is motivated by the need to consider technical systems in their environments (possibly interacting with third parties) when assessing their privacy, fairness, and security properties~\cite{veale2017fairer}. We show demonstrate conditions under which an ECS model is secure according to the well-established formal security model of noninterference \cite{gm82security}. We also formalize semantic security for an ECS model and reproduce the result that it is, in general, impossible for a system designer to guarantee semantic security on a statistical database given auxiliary knowledge. It is well known that information can be revealing of other sensitive information given auxiliary knowledge. Our theorem reflects the conditions under which auxiliary knowledge is possible. The contibutions from the model are due to explicit causal modeling of the generative process of data input into the system as well as the operations of the system itself. In Section~\ref{sec:robustness}, we build on these results to develop security models for cases where information is restricted based on its origin. We find these models analogous to noninterference and semantic security, and demonstrate sufficient conditions under which an ECS has these properties. Section~\ref{sec:usecase} shows a case study of using Origin privacy. We show that in a case of biometric sensing with Internet of Things devices, origin privacy specification can be used to enforce GDPR compliance. Section~\ref{sec:differential} shows how differential privacy is a special case of origin privacy. Section \ref{sec:incentives} demonstrates how a game theoretic layer can be added to the ECS model to show how system security properties relate to the impact on system users. Section~\ref{sec:future} addresses directions for future work. %\mct{Is there no prior work section? %I think the noninterference composition results should be covered.} % sb: Good point. Let's revisit this when we are preparing for publication. % \textbf{Contributions}. The contributions of this Chapter include: \begin{itemize} \item An analysis of privacy policies, specifically with respect to how they determine protected classes of information through information topic and information origin. \item An informal ontology and articulation of Origin Privacy appropriate for use by policy designers. \item Disambiguation of the concept of ``information flow'' into causal flow and nomic association components through causal modeling. \item The Embedded Causal System (ECS) model, a model of a causal system embedded in its environment suitable for proving properties of information flow security under these conditions. \item Proofs of conditions for noninterference and semantic security in causal and embedded causal systems. \item Formal security models for origin noninterference and origin semantic security, with proofs of sufficient conditions. \item Demonstration of the use of Origin Privacy in a biometric Internet of Things use case. \item Relaxed security models based on mutual information and proofs of their formal relationship to differential privacy. \item A demonstration of the use of ECS models in game theoretic modeling using Multi-Agent Influence Diagrams. \end{itemize} This chapter refers to two technical appendices. Appendix \ref{appendix:information-theory-theorems} proves several supplemental theorems in information theory that are used in a proof of the relationship between causal modeling and differential privacy. Appendix \ref{appendix:maid} outlines the major findings of \citet{koller2003multi} on Multi-Agent Influence Diagrams, and introduces a new concept: tactical independence. \section{Policy motivations} \label{sec:policy} To motivate a consideration of Origin Privacy as a flavor of privacy specification, we look to existing policies that include rules that restrict information flow based on its origin. We build on prior work in logical specification of privacy laws as inspiration for this approach~\cite{barth06sp,DeYoungGJKD10}. \subsection{Policy example: HIPAA Psychotherapy Notes} \label{sec:orgheadline3} Some laws define a class of information in terms of the process that creates it. A straightforward example from law are psychotherapy notes as defined under HIPAA\footnote{45 CFR \S 164.501.}: \begin{quote} Psychotherapy notes means notes recorded (in any medium) by a health care provider who is a mental health professional documenting or analyzing the contents of conversation during a private counseling session or a group, joint, or family counseling session and that are separated from the rest of the individual's medical record. Psychotherapy notes excludes medication prescription and monitoring, counseling session start and stop times, the modalities and frequencies of treatment furnished, results of clinical tests, and any summary of the following items: Diagnosis, functional status, the treatment plan, symptoms, prognosis, and progress to date. \end{quote} In this definition, there is a reference to a process involving ``documenting or analyzing [\ldots] a [\ldots] counseling session''. Any information with provenance beginning with its creation by a health care provider who is a mental health professional documenting a conversation during a counseling session separately from an individual's medical record, barring some exceptions, are psychotherapy notes. \subsection{Policy example: GLBA} \label{sec:orgheadline4} Some laws include references to information origin in the definition of what information is protected. We find this in particular in the Privacy Rule of the Gramm--Leach--Bliley Act, which applies to ``nonpublic personal information'' (NPI)\footnote{GLBA, 15 U.S. Code \S 6809}. This class of information is defined as personally identifiable financial information that is \begin{enumerate} \item provided by a consumer to a financial institution; \item resulting from any transaction with the consumer or any service performed for the consumer; or \item otherwise obtained by the financial institution. \end{enumerate} Reading progressively through each of these criteria, the concept of \emph{origin} helps us understand the differences between them and how that affect their enforceability. % \mct{Before getting to the criteria, there's general condition that % the information must be ``personally identifiable financial % information''. Maybe that should be discussed first.} Criterion~1 explicitly refers to the channel of transmission and does not refer to any specific meaning or category of the information transmitted (though examples are provided in the law, these are constrained only by what is normally transmitted in the process of procuring a financial service). It sets clear guidelines as to the process of creation and transmission. Criterion~2 refers to broad classes of ways in which the information is transmitted to the governed entity. Criterion~3 is written as if to cover all other cases where a financial institution could collect individualized information, regardless of the process of transmission. It is agnostic to the process of transmission. It raises the question of whether information can be personal financial information without having the specific provenance of a transaction or service with a financial institution. For example, information about a person's income may be personal financial information no matter how it is discovered. %\mct{It appears that Criterion~3 is a superset of 1 and 2. So, why % are 1 and 2 even stated? In fact, Criterion~3 doesn't even seem to % be a limitation. Why not just say it's all ``personally % identifiable financial information''?} % sb: I don't know. Do you think we need to address this? % \subsection{Policy example: PCI DSS} \label{sec:pci-dss} Though not a law, we consider the Payment Card Industry Data Security Standard (PCI DSS) to be a security regulation~\cite{bradleypayment}. Established as a proprietary information security standard by the Payment Card Industry Security Standards Council, it applies to organizations that use major branded credit cards such as Visa, Mastercard, and American Express. Though referenced in the laws of some U.S. states, it is mandated mainly by Visa and Mastercard themselves through their dealings with merchants\footnote{Minnesota Session Laws - CHAPTER 108--H.F.No. 1758. Nevada Revised Statutes, Chap. 603A \S 215. Wash. Rev. Code \S 19.255.020 (2011). See~\cite{pcisecuritystandardscouncilFAQ}}. We note two aspects of the PCI DSS that make it an effective regulatory standard. First, the PCI DSS governs only types of data the enforcing industry is responsible for generating: cardholder data and sensitive authentication data~\cite{pcisecuritystandardscouncil2016DSS}. This data is generally not meaningful outside of the operations that the payment card industry enables, because the potential uses of the data are a function of the system that created the data in the first place. This allows the payment card industry to straightforwardly enforce contractual obligations on those that use the data. This is in contrast with legal regimes that regulate the flow of more generally meaningful information between persons and organizations. A second feature of PCI DSS is that it is explicit about the application of the standard to networks of technology and business processes, which it calls the \textit{cardholder data environment} (CDE) \cite{pcisecuritystandardscouncil2016DSS}: \begin{quote} The PCI DSS security requirements apply to all system components included in or connected to the cardholder data environment. The cardholder data environment (CDE) is comprised of people, processes and technologies that store, process, or transmit cardholder data or sensitive authentication data. \end{quote} which PCI DSS further defines as \cite{pcisecuritystandardscouncil2016DSS}: \begin{quote} The first step of a PCI DSS assessment is to accurately determine the scope of the review. At least annually and prior to the annual assessment, the assessed entity should confirm the accuracy of their PCI DSS scope by identifying all locations and flows of cardholder data, and identify all systems that are connected to or, if compromised, could impact the CDE (for example, authentication servers) to ensure they are included in the PCI DSS scope. All types of systems and locations should be considered as part of the scoping process, including backup/recovery sites and fail-over systems. \end{quote} This is a clear example of how a privacy policy can be specific about the technical system to which it applies. \subsection{Other policies} The examples above have been chosen for their representativeness of the concepts developed in this paper. Other information protection policies do not define protected information solely in terms of its origin but rather depend in whole or in part on a definition of information topic. For example, the Family Educational Rights and Privacy Act (FERPA) defines ``education records'' as \begin{quote} those records, files, documents, and other materials which-- (i) contain information directly related to a student; and (ii) are maintained by an educational agency or institution or by a person acting for such agency or institution. \end{quote} The Children's Online Privacy Protection Rule (COPPA) includes a broad definition of ``personal informal'' as any individually identifiable information collected on-line, but includes special restrictions on personal information collected from a child, which could be read as a restriction based on origin. The United States Executive Order 13526 of 2009 gives certain government officials authority to classify documents they determine pose a risk to national security. In some cases, the information may be classified as soon as it is produced. In all cases, the information's classification prevents unauthorized access. Derivative information carries the classification of the original information. Information may be declassified when the conditions for classification no longer hold. Information restrictions on information therefore depend partly on its procedural history, but also on predictions made about the effects of disclosure. Our analysis of Origin Privacy explores the limits of privacy by design and what policies can be automatically enforced on a system bound by laws of statistical causation. We will demonstrate the ambiguity of the term ``information'' and how this renders the application of many policies that restrict information based on information topic indeterminate. \section{Origin Privacy} \label{sec:ontology} We will now provide an informal definition of Origin Privacy. First, we will introduce an ontology appropriate to the design of technical systems and inspired by the policy examples above. Three concepts are introduced in this section. \emph{Systems} are assemblages of people and technologies through which information flows and is transformed. \emph{Processes} are regular events, implemented as a person or technology's behavior, which act on information. Processes pass information to other processes as data. The \emph{origin} of data is the history of processes that led to its creation. Origin privacy includes any privacy restrictions made on information flow conditioned on that information's origin. \subsection{Systems} Regulations mark out spaces or domains in which information may flow more freely than others, or where restrictions specifically apply. For example, HIPAA applies to ``covered entities'', including health care providers\footnote{``Covered entity means: (1) A health plan. (2) A health care clearinghouse. (3) A health care provider who transmits any health information in electronic form in connection with a transaction covered by this subchapter.'' 45 CFR \S 160.103}. These spaces may be defined by legal jurisdiction, or they may be defined through networks of contractual relations. More often than not, the regulated space is not bounded geographically but rather by relationships between personnel, institutions, and technical systems, all of which are participating in information flows and processing. The Payment Card Industry Data Security Standard (PCI DSS) is explicit about this in its definition of the covered system in terms of the ``people, processes, and technologies that store, process, or transmit'' data (Section \ref{sec:pci-dss} provides further details) \cite{pcisecuritystandardscouncil2016DSS}. We generalize this concept and refer to assemblages of people, processes, and technologies such as these \emph{information processing systems}, or just \emph{systems}. There will inevitably be people, processes, and technologies that interact with a governed system without being included within it. We refer to these external factors as the \emph{environment} of the system. Systems have inputs and outputs, which are events that pass data from and to the environment. We will find it useful to model the world of a system within its environment as a larger, superordinate system. This causally embedded system framework is formalized in Section \ref{sec:ecs}. The use of the term ``system'' here is abstract and intended to provide a precise analytic frame. How it gets applied to an empirical case can vary. For example, in healthcare, we might consider a single hospital with its procedures as a system, or a network of covered entities collectively as a system. A complete discussion of systems and how they may be nested or interconnected is beyond the scope of this paper. \subsection{Processes} For the purpose of these definitions, we will assume that technical systems and their environments consist of many different components implementing information processes. People in their professional roles are, for the purposes of this framework, included as another way of implementing processes. The outputs of a process may be the input to another. These messages between processes are \emph{data}. When a process $A$ sends data to another process $B$ under one or more conditions, we say that there is a channel from $A$ and $B$, or, equivalently, that (the state of) $A$ directly causes (the state of) $B$. The structure of processes and their dependence on each other implies a causal graph \cite{pearl1988probabilistic} with directed edges representing the channels between processes. We will discuss the implications of causal modeling of systems more thoroughly in Section \ref{sec:causality}. % % NOTE: Contiguity is mentioned here, but not formally defined % in the mathematical section of this paper. % Systems are composed of contiguous processes, meaning that for any two processes in the system there will be an (undirected) path between the two through channels that includes only other processes in the system. While causal modeling has been used in a security context (e.g., \citet{feng2014security}), formal security research more often relies on static analysis of programs (e.g., \citet{mclean90sp}). We choose a causal models in this chapter because these models can represent both programs and their subprograms, as well as non-technical environmental processes that generate data.\footnote{While it may be the case that programs are as expressive as causal models, this discussion is beyond the scope of this paper.} An example of such a process is a medical examination conducted via an interpersonal interaction between patient and doctor. \subsection{Origin and provenance} The data resulting from a process depends causally on a history of prior processes. Sometimes data that is an input to a system is generated by a process that is not immediately part of the system. Data can flow through a series of relays before it reaches the system on which a privacy policy is being enforced. The entire history of the information as it flows from its creation to the system input is the information's \emph{provenance}. The governed system may or may not have access to assured metadata (such as cryptographic certificates) about the provenance of its data. We consider the origin of data to be the processes that have caused it, either directly or indirectly. For the purposes of enforcement of a policy, these processes may be either in the governed system or outside of it, in an originating system or the system's environment. \subsection{Origin privacy} Given the above ontology, we can now provide a definition of origin privacy. Origin privacy includes any and only those information flow restrictions implemented in a system that are conditioned on the provenance of system inputs. \section{Information flow and causal models} \label{sec:causality} We have motivated Origin Privacy as a concept of privacy that is useful when considering how to design information processing systems to be compliant with laws and other rules regarding the flow of personal information. The ontology in Section~\ref{sec:ontology} is motivated by policies that specifically mention systems and restrict information flow based on its origin. In this section we will introduce philosophical and formal concepts with which we will make our definitions and claims about origin privacy precise. Philosophically, privacy depends on appropriate information flow~\cite{nissenbaum09book}, where information is defined as that which allows somebody to learn about something else based on its regular associations with it. We find a formalization of this idea in Bayesian networks, a common formalism in statistics which represents the relationships between random variables with a directed acyclic graph. Bayesian networks have two attractive properties which we will explain. First, it is easy to derive some independence relations between variables from the graph structure of a Bayesian network. Second, this formalism supports an intervention operation that gives it robust causal semantics. All of these conceptual tools will be used in proofs later in this paper. We will close this section by showing how this formalism rigorously clarifies an ambiguity in the term `information flow', which refers to both causal flow and nomic associations between variables. We adopt the term \textit{situated information flow} for this sense of information flow in causal context. \subsection{Philosophy: contextual integrity and information flow} Origin Privacy is intended to be broadly consistent with the contextual integrity \cite{nissenbaum09book} philosophy of privacy in so far as it defines privacy as \emph{appropriate information flow}. Specifically, contextual integrity maintains that privacy expectations can be characterized by norms about how information about persons flows in particular social contexts, or spheres. In this article, we restrict our analysis of Origin Privacy to cases where expectations of privacy have been articulated as \emph{policies} in natural language and endowed with a social system of enforcement. We consider laws and contracts as kinds of policies. Policies may or may not express social norms as per contextual integrity; addressing the conditions under which policies reflect social norms is beyond the scope of this article. However, we maintain that some policies are specifically \emph{privacy} policies because they, like privacy norms, prescribe personal information flows. As a way of bridging from contextual integrity through privacy policies to the specification of privacy-preserving mechanisms, we address \emph{information flows} in general. Despite its wide use, the phrase ``information flow'' is rarely given a precise definition. However, philosophical and formal work on information flows have provided general insights that can bridge between social, legal, and technical theories of privacy. We will build on these insights to make arguments about origin privacy. There is a long history of literature on information flow in computer security and privacy research \cite{mclean90sp,gray91sp,barthe04csf,tschantz15csf,smith15lics}. %\cite{csftschantz13blackboxtr} \cite{tschantz14arxiv} These take their inspiration from Shannon's classic formulation of information theory~\cite{shannon1948mathematical}. Dretske's philosophical formulation of information flow~\cite{dretske1983epistemology} also draws on Shannon's information theory. In this work, we are explicitly bridging between philosophy and engineering principles, finding common ground between. According to Dretske's theory, a message carries information about some phenomenon if a suitably equipped observer could learn about the phenomenon from the message. In other words, a message carries information about anything that can be learned from it. For an observer to learn from it, the message must have a \emph{nomic} connection with its subject, where here "nomic" means ``law-like'' or ``regular''~\cite{dretske1981knowledge}. %\mct{How does this differ from the information as mere association view?} % sb: Do you have a citation for this view that I could use % in order to make the comparison? Messages, in this understanding, get their meaning from the processes that generate them, because these processes connect the content of messages reliably to other events. There is a logical connection between the definition of information flow and the structure of the regular dependence between events. The formal theory of the causal dependence between events has been worked out in the literature on causal graphical models \cite{pearl1988probabilistic}. \subsection{Causal probabilistic graphical models} \label{sec:orgheadline18} There is the a well known formalism for representing the causal relationship between uncertain events: the \emph{Bayesian network}, or probabilistic graphical model, framework \cite{pearl1988probabilistic}. As we will see, this form of modeling can be used to represent processes within a system, as well as in its environment. Before showing the relationship between Bayesian networks and Origin Privacy, we will present them and a few of their formal properties, drawing heavily on \citet{koller2003multi} for our choice of formal notation and wording. \subsubsection{Bayesian networks} \label{sec:orgheadline16} A Bayesian network represents the joint probability distribution of a set of random variables with a graph. Consider variables \(X_1, ..., X_n\) where each \(X_i\) takes on values in some set \(dom(X_i)\). We use \(\mathcal{X}\) to refer to the set \(X_1, ..., X_n\) and \(dom(\mathcal{X})\) to refer to their joint domain. A Bayesian network (BN) represents the distribution using a graph whose nodes represent the random variables and whose edges represent direct influence of one variable on another. \begin{dfn}[Bayesian network] A Bayesian network over variables $\mathcal{X} = X_1, ..., X_n$ is a pair (G,Pr). G is a directed acyclic graph with $n$ nodes, each labeled for one of the variables in $\mathcal{X}$. We use $Pa(X)$ to denote the parents of $X$ in the graph. Pr is a mapping of each node $X$ to a conditional probability distribution (CPD), $Pr(X \vert Pa(X))$. \end{dfn} \begin{figure} \begin{center} \begin{tikzcd} A \arrow[rrd] & & \\ C \arrow[r] & T \arrow[r] & W \\ D \arrow[ur] & & \end{tikzcd} \end{center} \caption{Alice's commute to work.} \end{figure} \begin{exm} (Figure 4.1) %alice-commute Alice will be on time for work $W$ if she sets her alarm $A$ early enough and traffic $T$ allows. Bad traffic can be caused by construction $C$ or an accident on the road $D$. \end{exm} Given the discussion of information flows above, we can see why this is relevant to privacy. The conditional dependence functions between random variables are the \emph{nomic relations} between events and messages. If two variables are conditionally dependent on each other, and this conditional dependence is known to the observer of one of the variables, then the observer can infer something (have knowledge of) the other variables. Hence, by our definitions, the variables carry information about each other. If privacy is appropriate information flow, then the privacy of a system will depend on the causal relationships between its components and the environment. A directed edge between one variable and another indicates a possible conditional dependence between them. Strictly speaking, it does not necessitate that there is a conditional dependence between them, it only necessitates that there is a conditional probability distribution function defined between them. But it does guarantee that at least one such conditional probability distribution does exist, and under reasonable conditions \emph{most} possible functions (in a measure-theoretic sense) will exhibit the conditional independence~\cite{meek1995strong}. As functions ensuring independence are quite rare in the space of all possible conditional probability functions, this quirk in the notation has not prevented this formalism from being useful in identifying independence in practice. %%% Seems to be repeated below % One property of causal models is that the % conditional independence of its variables can % to some extent be read off of the model's % graph structure. \subsubsection{D-separatedness} A useful property of probabilistic graphical models is that some aspects of the joint probability distribution of all variables represented in the graph can be read easily from the graph's structure. Of particular interest in the analysis of the joint probability distribution is when and under what conditions two random variables are independent. \begin{dfn}[Path] A \emph{path} between two nodes \(X_1\) and \(X_2\) in a graph to be a sequence of nodes starting with \(X_1\) and ending with \(X_2\) such that successive nodes are connected by an edge (traversing in either direction). \end{dfn} \begin{dfn}[Head-to-tail, tail-to-tail, head-to-head] For any three nodes (\(A, B,C\)) in succession on a path, they may be \emph{head-to-tail} (\(A \rightarrow B \rightarrow C\) or \(A \leftarrow B \leftarrow C\)), \emph{tail-to-tail} (\(A \leftarrow B \rightarrow C\)), or \emph{head-to-head} (\(A \rightarrow B \leftarrow C\)). \end{dfn} We will find it useful to refer to a special kind of paths, \emph{direct paths}. \begin{dfn}[Direct path] A \emph{direct path} from \(X_1\) to \(X_2\) is a path starting with \(X_1\) and ending with \(X_2\) such that all triples are head-to-tail. \end{dfn} \begin{dfn}[Ancestors and descendants] If there is a direct path from \(X_1\) to \(X_2\), then $X_1$ is an ancestor of $X_2$ and $X_2$ is a descendant of $X_1$. Let $descendants(X)$ be the set of descendants of $X$. \end{dfn} There are two ways in which a variable \(A\) can be conditionally dependent on another variable \(B\) without one of them being a descendant of the other. The variables may share an unobserved common cause or they may share an observed common effect. \begin{exm} One building in a neighborhood loses power, $B_1$. One can guess that other buildings $B_i$ around nearby lost power, because power in each building is dependent on the electric grid $G$. All the buildings may be affected by the common cause of a grid failure. \end{exm} \begin{center} \begin{tikzcd} & B_1 \\ G \arrow[ur] \arrow[r] \arrow[dr]& B_2 \\ & B_3& \end{tikzcd} \end{center} \begin{exm}(Figure 4.1) % alice-commute Suppose we observe that Alice is late for work $W$, as per our earlier example. This could be due to many reasons, including traffic $T$ and missing her alarm $A$. Traffic may be due to construction $C$ or an accident $D$. The probability of any particular cause is conditionally dependent on the others, because if any one cause is ruled out, the others are more likely. \end{exm} The existence of a path between two nodes is necessary for their probabilistic dependence on each other. It is not sufficient, particularly when considering their dependence \emph{conditional on other variables}. For this reason, paths in a Bayesian network can be blocked or unblocked based on a set of variables that is otherwise known or observed, the \emph{conditioning set}. \begin{dfn}[Blocked path] A path is considered to be \emph{blocked} if either: \begin{itemize} \item it includes a node that is in the conditioning set \(C\) where the arrows point to it do not meet head-to-head, or \item it includes a node where arrows do meet head to head, but neither this node nor any of its descendants is in the conditioning set \end{itemize} \end{dfn} \begin{dfn}[D-separation] If every path from \(X_1\) to \(X_2\) given conditioning set \(C\) is blocked, then \(X_1\) and \(X_2\) are d-separated. \end{dfn} \begin{thm}\label{thm:d-seperated} If \(X_1\) and \(X_2\) are d-separated conditioned on set \(C\), then $X_1 \independent X_2 \vert C$. \end{thm} \begin{proof} Proof is discussed in \citet{koller2003multi}. \end{proof} \hfill \hfill The converse (that independence implies d-separatedness) is not true in general because specific conditional distribution functions can imply independence. Similarly, it is not generally true that the absence of d-separatedness implies conditional dependence. However, is has been shown that conditional distribution functions implying conditional independence are rare in a measure-theoretic sense \cite{geiger1990logical,meek1995strong,koller2003multi}. %When there is uncertainty about which of a wide range of functions %characterizes the probability distribution over many variables, %absence of d-separation is strong evidence of the probability of %independence. %\mct{This is hard to follow.} \subsubsection{Intervention} We have used the terms \emph{Bayesian network} and \emph{causal model} interchangeably. This is because Bayesian networks support a causal interpretation through one additional construct, \emph{intervention} \cite{pearl1993bayesian}. An intervention on a Bayesian network sets the values of one or more of its values. Unlike an observation of a variable, an intervention effectively creates a new graphical model that cuts off the influence of a set variable on its parents and vice versa. Descendants of the set variable are affected by the intervention according to the probability distribution of the original model. \begin{dfn}[Intervention] An \emph{atomic intervention} setting variable $X_i$ to $x'_i$ on a Bayesian network $\W$ creates a new network $\W'$ with post-intervention probability distribution $Pr_{x'_i}$ $$Pr_{x'_i}(X_1,X_2,...,X_n) = \begin{cases} \frac{Pr(X_1,X_2,...,X_n)}{Pr(X_i = x'_i \vert Pa(X_i))} & \text{if} X_i = x'_i \\ 0, & \text{otherwise}\\ \end{cases}$$ \end{dfn} Theories of causation based on manipulation and intervention have been influential in philosophy~\cite{woodward2005making} and have been shown to be effective theories in psychology of causation~\cite{sloman2005causal} including the role of causation in moral reasoning~\cite{sloman2009causal}, suggesting interventionist causation as a potential bridge between computer science and ethical domains such as privacy and fairness. Cowgill and Tucker~\cite{cowgillalgorithmic} discuss the evaluation of algorithmic impact using counterfactuals, which draws on a different but compatible theory of causality~\cite{rubin1974estimating}. \subsection{Ambiguity of information flow} \label{sec:ambiguity} We have drawn a connection between information flow in the philosophical sense relevant to Contextual Integrity and Bayesian networks. A Bayes network is a way of representing the nomic dependencies between phenomena. They are ``nomic'' because they describe probability distributions that generalize over particular instances of a system's functioning. These nomic relations are factored out as an explicit structure of causal relationships. This reveals an ambiguity in the very concept of \emph{information flow}, illustrated in the following example. \begin{exm} Alice, a teacher tells every student privately their test score's rank $R$ (first in class, second in class, etc.) after every test, with class participation used as a tie-breaker. Alice sends a message $B$ to Bob with the information that he has the second highest rank in the class. Alice also sends a message $E$ to Eve that she has the highest rank in the class. From her message and knowledge of the test environment, Eve learns from her message that Bob was told that he was, at best, second in class. Did information about Bob flow to Eve? \end{exm} A formal representation of this example makes the root of the ambiguity clear. Consider a three node Bayesian network where $R$ is the test results, $B$ is the message sent to Bob, and $E$ is the message sent to Eve (Fig 4.2). \begin{figure} \label{fig:school} \begin{center} \begin{tikzcd}[row sep=tiny] & B \\ R \arrow[ur] \arrow[dr]& \\ & E \end{tikzcd} \end{center} \caption{Test score ranks ($R$) distributed to Bob ($B$) and Eve ($E$).} \end{figure} There is causal flow along the edges from $R$ to $B$ and from $R$ to $E$. But an observer of a single variable aware of the system's laws (nomic connections, graphical structure) can learn nomic associations of a message that inform about variables that are not in the message's causal history. Despite $E$ neither causing nor being caused by $B$, $E$ reveals information about $B$. The phrase ``information flow'' is ambiguous because the word ``information'' is ambiguous \cite{nunberg1996farewell}: it can refer to both a message and the contents of a message. We do not favor either sense. Rather, we propose that to resolve this ambiguity, one has to recognize how the systematic creation and transfer of messages--represented in general by a graph of causal flows--gives each message its meaningful contents. In our formalism, a situated information flow is a causal flow that, by virtue of its place in a larger causal structure, has nomic associations. Our analysis of privacy policies shows how they variously restrict the flow of information based on its contents as well as its causal history or origin. This is consistent with our analysis of ``information flow'' as refering to one causal flow within a larger system of causes that give it contents. This scientific formulation of information flows is not yet native to the the language of the law. That the law refers variously to aspects of information flow based on contents and causal history reflects how both are essential to the meaning of the term. In the following sections we will precisely model a system in its environment in order to disambiguate the different aspects of information flow and understand the conditions under which a system can be free of information leaks. We will measure the strength of nomic associations using a well-understood measure, \emph{mutual information}. Mutual information captures how much about one random variable can be learned from observing another. \begin{dfn}[Mutual information] The mutual information of two discrete random variables $X$ and $Y$ is $$I(X,Y) = \sum_{x \in X} \sum_{y \in Y} p(x,y) log \frac{p(x,y)}{p(x)p(y)}$$ \end{dfn} In particular, $I(X,Y) = 0$ if and only if $X \independent Y$. %space Mutual information is a technical term with a specific mathematical meaning. It is no etymological accident that it forms part of the analytic definition of ``information flow'' that we have developed in this section. In the Appendix \ref{appendix:information-theory-theorems}, we derive several theorems relating to the mutual information of variables organized in a Bayesian Network. We will use these theorems in proofs about system security in the following sections. \section{Bayesian networks and information flow security} \label{sec:security} In this section, we formalize the ontology from Section~\ref{sec:ontology} that we derived from privacy policies. Our formalization uses the causal graphical modeling tools outlined in Section~\ref{sec:causality}. We show that several known results in information flow security have dual results within our formal causal modeling of systems in their environments. We demonstrate this for concepts of noninterference~\cite{gm82security} and the impossibility of guaranteeing secrecy given the possibility of arbitrary auxiliary knowledge~\cite{dwork06icalp,dwork08jpc}. Where our analysis goes beyond these known results in information flow security are that our models explicitly take into account the relationship between a technical system and its environment. In each case, our theorems prove conditions of the security properties that could not be discovered by static program analysis in isolation. For example, it is well known that information can be revealing of other sensitive information given auxiliary knowledge. Our Theorem \ref{thm:semantic} reflects the conditions under which auxiliary knowledge is possible. Because the CPD defined by \(Pr\) between each random variable and its parents can be an arbitrary function, including deterministic logical operations, it is possible to encode a system of computational components, including sensors, data processors, and actuators, as a BN. Earlier we defined Origin Privacy in terms of systems, processes, and messages. These concepts map easily into this formalism: systems are Bayesian networks; processes are random variables or events; the inputs and outputs of processes are determined by links connecting them to other processes; messages are the instantiation of particular random variables, which are available as inputs to later variables. %\mct{I think key points of this section is the following:\\ %CS people have been using noninterference, which models systems as a computation. This model is limited in what it can say since it doesn't have a model of the environment in which the system operates. We will instead models systems as ECSes, expressed with causal graphs that can extend out to model the environment. ECSes have a property that implies noninterference. To do this, we have cut out part of the ECS to model the system found in noninterference, leaving the rest as the environment. We present a property that these cut-out graphical sub-models corresponding to systems can have that implies the computation modeling the same system having noninterference. %\\ %I'm I right? If so, I'm missing the part where you show that your model is the same as the old model. %} % %sb: I'm not sure what you're getting at, but it sounds important. %We should discuss before publication. % \subsection{Embedded Causal System (ECS) Model} \label{sec:ecs} In the subsequent sections we will use the following standard notation and model, which we will refer to as the Embedded Causal System (ECS) Model. \begin{dfn}[World] A \emph{world} $\W$ is a set of random variables with a conditional probability distribution that may be modeled as a Bayesian network $(G_\W,Pr_\W)$. \end{dfn} \begin{center} \begin{tikzcd} \W \end{tikzcd} \end{center} % %\begin{dfn}[Contiguous] % Given a Bayesian network $(G,Pr)$ over $\mathcal{X}$, % a subset $\Y \subset \mathcal{X}$ is \emph{contiguous} % if for all $Y_1, Y_2 \in \Y$, there is a path between % $Y_1, Y_2$ such that all nodes on the path are members % of $\Y$. %\end{dfn} % \begin{dfn}[System] A subset of the world $\Y \subset \W$ is the \emph{system}. \end{dfn} %% A system must be contiguous. \begin{dfn}[Environment] The \emph{environment} of $\Y$ is the set of nodes in the world that are not in the system, $\mathcal{E} = \W - \Y$ \end{dfn} \begin{center} \begin{tikzcd} \Y \arrow[d, bend left=50]\\ \mathcal{E} \arrow[u, bend left=50] \end{tikzcd} \end{center} (In this and a few other diagrams, we will include cycles because these are diagrams of blockmodels over other networks. In a blockmodel of $G$, a partition $\{P_1, P_2, \ldots\}$ of the set of original nodes $\mathcal{X}$ is treated as a new set of nodes, with an edge between $P_1$ and $P_2$ iff there exists $X_1$ and $X_2$ such that $X_1$ is in $P_1$, $X_2$ is in $P_2$, and $(X_1,X_2)$ is in $G_{\msf{edges}}$) It is common in security research to consider \emph{systems} as units of analysis; these systems contain \emph{programs}; system input is data and system output is the result of the programs operating on the data~\cite{mclean90sp}. These programs are represented using a formal approximation of a programming language in order to prove security properties of systems. Systems in our formalism also have inputs and outputs, which are defined by their position relative to the environment. \begin{dfn}[Sensors and inputs] A \emph{sensor} is an edge $(A,B) \in G_\W$ such that $A \in \mc{E}$ and $B \in \Y$, An \emph{input} is the head node of a sensor, $B$. Denote the set of all inputs with $\mathcal{S}$. \end{dfn} % % MCT: Shouldnt' inputs be represented as I? % SB: Maybe. We can revisit this before publication. % \begin{dfn}[Actuators and outputs] An \emph{actuator} is an edge $(A,B) \in G_\W$ such that $A \in \Y$ and $B \in \mc{E}$. An \emph{output} is the tail node of an actuator, $A$. Denote the set of all outputs with $\A$. \end{dfn} \begin{center} \begin{tikzcd}[column sep=tiny] \mathcal{S} \arrow[r] & \Y \setminus (\mathcal{S} \cup \A) \arrow[r] & \A \arrow[dl, bend left=50]\\ & \mathcal{E} \arrow[ul, bend left=50] & \end{tikzcd} \end{center} See the definition of \emph{orderly} (Definition \ref{def:orderly-system}) for a set of further constraints on inputs and outputs that are necessary for proving security properties of ECS models. For some security related applications of this model, it is necessary to distinguish between ``high-side'' and ``low-side'' inputs and outputs. High-side variables denote sensitive variables that should not be leaked. \begin{dfn}[High and low sides] Inputs $\mathcal{S}$ are partitioned into high $\mathcal{S}_H$ and low $\mathcal{S}_L$ side variables. Similarly, outputs $\A$ are each partitioned high-side $\A_H$ and low-side $\A_L$ variables. \end{dfn} \begin{center} \begin{tikzcd}[column sep=tiny] \mathcal{S_H} \arrow[r] & \Y \setminus (\mathcal{S} \cup \A) \arrow[r] \arrow[dr] & \A_H \arrow[ddl, bend left=100]\\ \mathcal{S_L} \arrow[ur] & & \A_L \arrow[dl, bend left=50]\\ & \mathcal{E} \arrow[uul, bend left=100] \arrow[ul, bend left=50] & \end{tikzcd} \end{center} Note that in the above diagram and throughout this paper, we will sometimes refer to a set of random variables such as the set of all high-side inputs $\mc{S}_H$ as if it is a single random variable. This is well-motivated, because for any set of random variables $\mc{X} = \{X_0, X_1, X_2, ...\}$ one can define a new random variable whose domain $Dom(\mc{X})$ is the cross product $Dom{X_0} \times Dom{X_1} \times ...$ and whose probability distribution is the joint probability distribution $Pr(X_0, X_1, ...)$. \begin{exm} A hospital uses a system to manage its medical records. It takes input from many health care professionals through many different forms of treatment. Most medical records are considered a low-side input because they can be accessed by other professionals treating the patient. Psychotherapy notes are a high-side input because they have special restrictions on their use. \end{exm} \begin{exm} An intelligence agency has many classified sensors, such as satellites and drone imagery, which contain information that is critical for national security. These are high-side inputs. They also use many data sets that are available publicly and commercially. These are low-side inputs. \end{exm} \subsection{System design} \label{sec:design} In many cases what we are interested in is the possibility of a \emph{system designer} inventing a system subject to certain constraints. %An example of such a constraint is the system's %meeting a \emph{specification}, which is a desired %relationship between inputs and outputs. % %\begin{dfn}[Specification] % A \emph{specification} is a function % $f \from \mathcal{S}, \A \to [0,1]$ % \mct{Do you mean $f \from \mathcal{S} \times \A \to [0,1]$?} % that is a joint probability distribution % over $\{\mathcal{S}, \A\}$. % The system $\Y$ fulfills the specification iff % $Pr(\mathcal{S},\A) = f(\mathcal{S},\A)$. % \mct{I'm not sure what $f(\mathcal{S},\A)$. If $f$ is a function % from $\mathcal{S} \times \A$, then it has to take elements of them % as inputs, not the whole sets.} % \mct{If $\Y$ is fulfilling something, it must be found in the % equation somewhere. I suspect it's implicitly determining $Pr$, % but maybe that should be made explicit as $Pr_{\Y}$.} %\end{dfn} % % I dont' think I use this definition of nontriviality anywhere in the draft... % %\begin{dfn}[Nontriviality] % A specification $f \from \mathcal{S}, \A \to [0,1]$ % is \emph{nontrivial} iff given $Pr = f$, % $Pr(\A \vert \mathcal{S}) \neq Pr(\A)$. %\end{dfn} % %\mct{I think you're just saying that $f$ depends upon the input % $\mc{S}$: there exists $s_1$, $s_2$, and $a$ such that % $f(s_1,a) \neq f(s_2,a)$.} % %\mct{Making $f$ be a joint distribution over inputs and outputs is % nonstandard. Typically, we make it a function from inputs to % distributions over outputs. This isn't just currying, but looks % similar. We should discuss.} % %The general theory of Bayesian networks shows %that the extent to which nomic associations %between variables allow information to flow %between them depends not only on the causal %structure of the network, but also on which %other variables have already been conditioned %on or, in a Bayesian sense, are observed. % We are interested in the ways that an ECS enables inferences, and how these inferences depend on what is known or observable about the system and its environment. We define some terms here to denote properties of the conditioning set that we will use in later proofs. Intuitively, conditioning sets can be interpreted as observed states of the world that are available to an adversary trying to learn high-side information. The condition of being \emph{present} captures the intuition that system designers cannot account for the ways that downstream uses of system outputs may be used to reveal sensitive information. \begin{dfn}[Present] A system $\mathcal{Y}$ with conditioning set $\mc{C}$ is \emph{present} iff \begin{itemize} \item No descendants of $\mathcal{A}$ are in the conditioning set $\mc{C}$, and \item No descendant of $\mathcal{A}$ is in $\mathcal{S}$. \end{itemize} \end{dfn} The term ``present'' indicates that the attacker is able to condition on variables prior to and during the operation of the system, but not variables in the ``future'' of the system. Requiring that the system outputs are not ancestors of the the system inputs guarantees that the system is in fact positioned in a particular place in time, so to speak. The condition of being \emph{covered} captures the intuition that in general we do not expect attackers to have the ability to observe systems \emph{as they are functioning}, even if we allow them to know exactly how a system works because they know the causal relationships between the system components. \begin{dfn}[Covered] \label{def:covered-system} A system is \emph{covered} if no $Y \in \mathcal{Y}$ is in the conditioning set $\mc{C}$. \end{dfn} % %\mct{A somewhat subtle point here is that people don't see $\A$ % directly, but rather see the effects of $\A$. If the reader isn't % paying attention they might take your model as ruling out seeing % outputs altogether.} % We also specify a condition on the relationship between sensors, actuators, the system, and the environment. In some cases we will not allow an input to be caused by a system variable. We will also not allow an output to cause a system variable. When both these conditions hold, we call a system \emph{orderly}. \begin{dfn}[Orderly] \label{def:orderly-system} A system is \emph{orderly} iff: \begin{itemize} \item $\forall X \in \mathcal{W}, S \in \mathcal{S}, X \in Pa(S) \Longrightarrow X \in \mathcal{E}$ \item $\forall X \in \mathcal{W}, A \in \mathcal{A}, A \in Pa(X) \Longrightarrow X \in \mathcal{E}$ \end{itemize} \end{dfn} Given only the subgraph represented by $\mathcal{Y}$, under some condition it will be the case that the high-side inputs and the low-side outputs are conditionally independent. We will name this property \emph{safety}. \begin{dfn}[Safe] A system $\mathcal{Y}$ is \emph{safe} given conditioning set $\mc{C}$ iff when considering it as a subgraph, there are no unblocked paths between $\mathcal{A}_L$ and $\mathcal{S}_H$. \end{dfn} If there are no unblocked paths between $\A_L$ and ${S}_H$ in the system subgraph, then these variables are d-separated and so the system is safe. We assume for our purposes that a system designer can guarantee its safety through sound engineering alone. For example, consider the system defined by the graph in Figure 4.3. \begin{figure} \label{fig:safe-system} \begin{center} \begin{tikzcd} \mathcal{S}_H \arrow[r] & \mathcal{A}_H \\ \mathcal{S}_L \arrow[r] \arrow[ur]& \mathcal{A}_L \end{tikzcd} \end{center} \caption{A system that is safe so long as its high-side actuators are not observed.} \end{figure} When $\mathcal{A}_H$ is not in the conditioning set, the path between $\mathcal{S}_H$ and $\mathcal{A}_L$ is blocked, so the high-side input and low-side outputs are independent. In general, system designers can guarantee safety by removing any direct paths from $\mathcal{S}_H$ to $\mathcal{A}_L$ and then ensuring that the system is covered. \begin{thm} \label{thm:safety} If a system subgraph $\Y$ has no direct paths from $\mathcal{S}_H$ to $\mathcal{A}_L$ and is covered and orderly, then it is safe. \end{thm} \begin{proof} Assume there are no direct paths from $\mathcal{S}_H$ to $\mathcal{A}_L$ in the system subgraph. So any unblocked path between $\mathcal{S}_H$ and $\mathcal{A}_L$ must be indirect. Suppose there is a path $p$ that begins with an incoming edge to $\mathcal{S}_H$, as in $\mc{S}_H \leftarrow \cdots \mc{A}_L$. Because the system is orderly, incoming edges to into input $\mathcal{S}_H$ must come from nodes that are not in the system $\Y$, as in $\mc{S}_H \leftarrow \mc{E} \cdots \mc{A}_L$. Because these nodes are not in the system subgraph, they cannot be in $p$. Therefore, the path $p$ must begin with an outgoing edge to $\mathcal{S}_H$. By a parallel argument, the path must end with an incoming edge to $\mathcal{A}_L$, as in $\mc{S}_H \rightarrow \cdots \rightarrow \mc{A}_L$. Because the path $p$ is indirect and it begins with an outgoing edge and ends with an incoming edge, there must be some node $X$ such that $X$ is on the path and $X$ is a common effect node, as in $\mc{S}_H \rightarrow \cdots \rightarrow X \leftarrow \cdots \rightarrow \mc{A}_L$. %\mct{Why?} Because the system is covered, $X$ must be unobserved. This implies that the path $p$ is blocked. Therefore, there are no unblocked paths between $\mathcal{S}_H$ and $\mathcal{A}_L$. Thus, by Theorem~\ref{thm:d-seperated}, these variables are independent and the system is safe. \end{proof} Information flow security literature often considers systems or programs in isolation from their environment. In practice, systems are always connected with an environment, which is why we have developed ECS. So Theorem \ref{thm:safety} is not enough to show the conditions of security in an ECS model because its inputs and outputs are not connected with variables in an environment. For this, we turn to a well established formal security model, noninterference. \subsection{Noninterference} \label{sec:noninterference} \emph{Noninterference}, introduced by \cite{gm82security}, is a security policy model widely used in computer science. Sabelfeld and Myers~\cite{sabelfeld03journal} define noninterference informally as ``a variation of confidential (high) input does not cause a variation of public (low) output.'' More formally, model a program $C$ as taking an input state $s = (s_h,s_l)$, as producing an output in a set $S \cup \{ \bot \}$, where $\bot$ stands for nontermination and $\bot \notin S$. We use $\lbb C\rbb : S \rightarrow S \cup \{ \bot \}$ to denote such a semantics. An equivalence relation $=_L$ holds when two inputs they agree on the low values ($s=_Ls'$ iff $s_l = s_l'$). The attacker's power is characterized by a relation $\approx_L$ such that if two behaviors are related by $\approx_L$ they are indistinguishable to an attacker. Following \citet{tschantz15csf} and \citet{datta2017use}, we expand the definition of the operator $\lbb \cdot\rbb $ to be a function to and from probability distributions over states, which affords a probabilistic definition of noninterference. \begin{dfn}[(Probabilistic) noninterference] For a given semantic model, $C$ is exhibits \emph{noninterference} or \emph{is secure} iff for all $s_1$ and $s_2$ in $S$, $s_1 =_L s_2$ implies $\lbb C\rbb(s_1) \approx_L \lbb C\rbb(s_2)$. \end{dfn} This definition admits a wide range of possible semantics for the attacker's equivalence relation $\approx_L$. We will choose a particular semantics relevant to the ECS model. We impose a probability distribution over inputs $\mathcal{S}$. With it we can construct the variable $\A = \lbb C\rbb(\mathcal{S})$. We use $\Y$ to denote the minimal Bayesian network relating $\mathcal{S}$ to $\A$ and treat it as the system model. As per the ECS model, we partition the inputs and outputs into high and low sides, $(\mathcal{S}_H, \mathcal{S}_L)$ and $(\mathcal{A}_H, \mathcal{A}_L)$, respectively. %\mct{This should be done in a way respecting $=_L$, right?} %sb: I don't know what this means. Define attacker indistinguishability $\approx_L$ as probabilistic indistinguishability of the low-side outputs when conditioned on inputs: \begin{dfn}[ECS attacker indistinguishability] $$\A \approx_L \A' \text{ iff } Pr(\A_L) = Pr (\A_L')$$ \end{dfn} Conceptually, we are modeling the execution of the program $C$ as the realization of the random variables in $\Y$. $C$ implies a probability distribution $Pr(\mc{A} \vert \mc{S})$. It also implies a probability distribution of $\mc{A}$ conditional on $\mc{S} = s$ realized. $\lbb C\rbb(s) = Pr(\A \vert \mathcal{S} = s)$, where $s$ is an instantiation of $\mathcal{S}$. For $s \in \mathcal{S}, a \in \A$, let $(s_h,s_l,a_h,a_l) =_L (s_h',s_l',a_h',a_l')$ iff $s_l = s_l'$. \begin{dfn}[ECS Noninterference] For a given ECS model, $\Y$ exhibits \emph{noninterference} or \emph{is secure} iff $$\forall s_1,s_2 \in \mathcal{S}, s_1 =_L s_2 \Longrightarrow P(\A \vert \mathcal{S} = s_1) \approx_L P(\A \vert \mathcal{S} = s_2)$$ \end{dfn} \begin{cor}\label{cor:noninterference-ind} $\Y$ is secure by noninterference iff $$\A_L \independent \mc{S}_H \given \mc{S}_L$$ \end{cor} \begin{proof} $\Y$ is secure by noninterference iff $$\forall s_1,s_2 \in \mathcal{S}, s_1 =_L s_2 \Longrightarrow P(\A \vert \mathcal{S} = s_1) \approx_L P(\A \vert \mathcal{S} = s_2)$$ iff \begin{equation} \begin{split} & \forall s_H,s_H',s_L,s_L' \in \mathcal{S}, s_L = s_L' \\ & \Longrightarrow P(\A \vert \mathcal{S}_H = s_H, \mathcal{S}_L = s_L) \approx_L P(\A \vert \mathcal{S}_H = s_H', \mathcal{S}_L = s_L') \end{split} \end{equation} iff \begin{equation} \begin{split} & \forall s_H,s_H',s_L \in \mathcal{S} \\ & P(\A_L \vert \mathcal{S}_H = s_H, S_L = s_L) = P(\A_L \vert \mathcal{S}_H = s_H', S_L = s_L) \end{split} \end{equation} iff $$A_L \independent S_H \vert S_L$$ \end{proof} Under what conditions is a system secure by noninterference? We can prove that if the system designer can guarantee that the system is present and safe, then it is secure by noninterference. \begin{lem} \label{lem:present} If an ECS model is present, then there can be no unblocked path between $\mc{S}_H$ and $\A_L$ that includes an outgoing edge from $\A_L$. \end{lem} \begin{proof} Proof by contradiction. Suppose an unblocked path exists between $\mathcal{S}_H$ and $\A_L$ such that the edge connecting to $\A_L$ was outgoing. That is, suppose the path was of the form: $$S_H \cdots \leftarrow \A_L$$ for some $S_H$ in $\mc{S}_H$. Consider the sequence of nodes on the path $S_H, X_1, X_2, \ldots, X_n, \A_L$ and the direction of the arrows between them, with $X_n \rightarrow \A_L$, and labeling $S_H$ with $X_0$. Because the system is present, no descendant of $\A_L$, is in $\mathcal{S}$. Therefor the must be at least one edge on the path such that $X_{i-1} \rightarrow X_i$. Count down from $n$ to $1$ and identify the first $X_i$ such that $X_{i-1} \rightarrow X_i$, $X_i$ will be a descendant of $\A_L$ because there is a direct path between $\A_L$ and it. $X_i$ cannot be $X_0 = S_H$, which is in $\mc{S}$. Therefore node $X_i$ will be the common cause of a head-to-head connection. That is, \[S_H \cdots X_{i-1} \rightarrow X_i \leftarrow X_{i+1} \leftarrow \cdots \leftarrow \A_L \] Because the system is present, this node $X_i$ must not be in the conditioning set. The path must therefore have a head-to-head connecting node that is not in the conditioning set. So it is a blocked path, resulting in a contradiction. \end{proof} \begin{thm} \label{thm:ecs-noninterference} Given an ECS model, if the system is present, safe, and orderly then the system is secure by noninterference. \end{thm} \begin{proof} Consider possible paths between $\A_L$ and $\mc{S}_H$ while $\mc{S}_L$ is in the conditioning set. By Lemma \ref{lem:present}, a present system can have no unblocked paths from $\mc{S}_H$ to $\A_L$ that end with an outgoing edge from $\A_L$. So any unblocked path from $\mc{S}_H$ to $\A_L$ must end with an incoming edge into $\A_L$. Because the system is orderly, if $X \rightarrow A$ for any $A$ in $\A$, then $X \in \Y$. Therefore, unblocked paths must include nodes in the $\Y$ subgraph. Because the system is safe, there are no unblocked paths between $\mathcal{S}_H$ and $\A_L$ consisting of only nodes in the system $\Y$ subgraph. So any unblocked path between $\mc{S}_H$ and $\A_L$ must include both nodes that are in $\Y$ and nodes the are in $\mathcal{E}$. We have already ruled out paths that include descendants of $\A$. So such a path must include ancestors of $\mathcal{S}$. The path must begin with $\mc{S}_H$, go into the environment, then re-enter the system via $\mc{S}_L$, then go to $\A_L$. Because the system is orderly, incoming edges into $\mc{S}_L$ must be $E \in \mathcal{E}$ and outgoing edges must be $Y in \mathcal{Y}$. That is, $$S_H \leftarrow \cdots E \rightarrow S_L \rightarrow Y \cdots \leftarrow A_L$$ for some $S_H$ in $\mc{S}_H$, $S_L$ in $\mc{S}_L$, and $A_L$ in $\A_L$. $S_L$ is in the conditioning set and part of a head-to-tail structure $E \rightarrow S_L \rightarrow Y$ on the path. Therefore the path is blocked. So with $\mathcal{S}_L$ in the conditioning set, all paths between $\A_L$ and $\mathcal{S_H}$ must be blocked. So $\A_L \independent \mc{S}_H \vert \mc{S}_L$. By Corollary~\ref{cor:noninterference-ind}, the system is secure by noninterference. \end{proof} It is therefore possible to implement an ECS that is secure in the sense of noninterference as long as a few conditions of the system (covered, safe, and orderly) are guaranteed. Privacy policies that restrict information flow (e.g. by guaranteeing confidentiality) of data based on how it was inputted into the system can be modeled in this framework. In the next section, we will show that policies that impose restrictions on information flow based on the content of information cannot be as easily restricted; to be effective there must be independence relations in the environment of the system. Thus the viability of privacy policies depends on the distinction noted in Section \ref{sec:ambiguity} between causal flow and association. \subsection{Preventing associations} \label{sec:prevention} We have proven that under certain conditions an ECS is secure by noninterference (Theorem \ref{thm:ecs-noninterference}). Noninterference is a widely respected formal security model in among computer security researchers. One reason for this is that it is a criterion that depends only one the internals of the system. Computer scientists can guarantee that a system, such as a program, is secure by noninterference without considering how that program will be used in practice. What if we wanted to hold systems to a higher standard that takes into consideration the processes that generate a system's data? For this we need a stronger security property. We can be more specific and introduce a security policy model that is strictly stronger than noninterference. \begin{dfn}[ECS semantic security] %%%% For a given ECS model, $\Y$ is exhibits \emph{semantic security} iff $$ \mc{S}_H \independent \A_L $$ \end{dfn} Semantic security is a well known property of cryptographic systems that means, intuitively, that an attacker intending to determine the contents of a message might as well not look at an encrypted signal. The term has taken on a wider use in the differential privacy literature as it has been introduced as a desideratum for statistical databases along the lines of that proposed by Dalenius in 1977 \cite{dalenius77statistik, dwork06icalp}. Dwork and Naor \cite{dwork06icalp,dwork08jpc} show that it is impossible to guarantee the semantic security of such a database given arbitrary auxiliary knowledge. We draw on the spirit of this literature in our definition of ECS semantic security. The principle difference between ECS semantic security and noninterference is that the latter is concerned with the Independence of system outputs from sensitive inputs \emph{conditioned} on the inputs, whereas the former takes into consideration how environmental correlations may allow system outputs to reveal system inputs. Noninterference does not imply semantic security. Put another way, the same system can be secure by noninterference but semantically insecure. Consider the system in Figure 4.4. \begin{figure} \label{fig:not-propitious} \begin{center} \begin{tikzcd} E \arrow[r] \arrow[dr] & S_H \arrow[r] & A_H\\ & S_L \arrow[r] \arrow[ur] & A_L \end{tikzcd} \end{center} \caption{A system that is not propitious when $E$ is unobserved.} \end{figure} This system is safe when $A_H$ is not in the conditioning set. It is secure by noninterference because when $S_L$ is in the conditioning set, the path from $S_H$ to $A_L$ that goes through $E$ is blocked. But when $S_L$ is not in the conditioning set, this path is open and therefore $A_L$ can be conditionally dependent on $S_H$. %\iffalse % %PROBLEM: A possible path is opened by $S_L$ being %a common effect. So the system has to be present? % %Semantic security implies noninterference security %under rather general conditions. % %*** PROOF NEEDED -- WHAT CONDITIONS? *** % %\begin{thm} % If a system is both orderly and % ECS semantically secure, % then it is secure by noninterference. %\end{thm} %\begin{proof} % *** Check edge cases! *** % If a system is semantically secure, then % $$A_L \independent S_H$$ % % By the definition of d-separation, % adding nodes to the conditioning set will only % unblock paths if those nodes are common % effects on the path, or are descendants % of a common path. % % *** actually need ANOTHER condition, % which is that sensors are not % descendants from actuators?!? *** % % $$A_L \independent S_H \vert S_L$$ %\end{proof} %\fi % Our conjecture is that semantic security cannot be guaranteed by the system designer alone. We are able to prove sufficient conditions for semantic security by including a general property of the world, including the environment outside the system. \begin{dfn}[Propitious] The world $\mathcal{W}$ is \emph{propitious} iff there is no unblocked path between $S_H$ and $S_L$. \end{dfn} \begin{thm} \label{thm:semantic} If a system is present, safe, and orderly, and the world is propitious, then the system is semantically secure. \end{thm} \begin{proof} By Theorem \ref{thm:ecs-noninterference}, the system is secure by noninterference, implying that $$S_H \independent A_L \vert S_L$$ So no unblocked paths run from $S_H$ to $A_L$ when $S_L$ is in the conditioning set. Recall from the proof of Theorem \ref{thm:ecs-noninterference} that this was because we ruled out all possible paths from $S_H$ to $A_L$. Paths running from $S_H$ through $S_L$ to $A_L$ were blocked because $S_L$ was in the conditioning set. Consider any such path, now unblocked as $S_L$ is not in the conditioning set. If it is unblocked, then there is an unblocked path between $S_H$ and $S_L$, which contradicts the assumption that the world is propitious. Therefore there are no unblocked paths between $S_H$ and $A_L$ and so $A_L \independent S_H$. \end{proof} %\mct{It feels like the punchline of this section should be that noninterference can be designed for whereas semantic security cannot. However, this isn't made clear for two reasons. %\\ %First, it's not all the clear how much harder the conditions of Thm.~\ref{thm:semantic} are to meet than the conditions of Thm.~\ref{thm:ecs-noninterference}. The harder one just requires adding one extra one, propitious, to the set of three required for the easier one. A 25\% increase isn't that bad, right? Well, that's, of course, not the right way of looking at this. We need to look at how hard each of these conditions are to ensure. I don't see such a discussion anywhere. Indeed, more can be done to make each of these conditions intuitive (starting with their names). %\\ %Second, people know that semantic security is something promised in crypto papers, so it can't be that hard, right? We need some explanation about why it's possible to achieve it in that special case, but not in general.} % %\mct{Much of this section reminds me of work on composition properties for noninterference. We should compare what we're doing to that work.} % \section{Formalizing origin privacy} \label{sec:robustness} We have defined origin privacy as privacy policies that place restrictions on information based on its provenance. This is in contrast to policies that restrict information based on its content. Another way to put this difference is that origin-based privacy policies restrict information based on the structure of causal flows, while information content based policies restrict information based on its nomic associations. The problem with restricting information flows based on information content is well illustrated by the problem of guaranteeing semantic security in an ECS system. Any sensitive information content can potentially have nomic associations with otherwise inocuous inputs due to causal paths in the environment. Guaranteeing the absence of associations depends on properties of the environment that may be outside the system designer's control. Noninterference, on the other hand, is an achievable property for a system designer. However, it is defined in such a way that it can be guaranteed even when some kinds of harmful information leaks are probable in practice. In our legal analysis in Section \ref{sec:policy} we identified some policies that restrict information based on its provenance, or origin, rather than its information content. In Section \ref{sec:ontology}, we have identified the origin of information as the chain of processes causing its input into the system. Taking the concept of ``high-side'' input as those inputs to a system that are treated with special sensitivity, we can model an example world that meets the most basic requirement of an origin based policy roughly like so: \begin{center} \begin{tikzcd} O \arrow[r] & R_0 \arrow[r] & \cdots R_n \arrow[r] & S_H \arrow[r] & A_H\\ & & & S_L \arrow[r] \arrow[ur] & A_L \\ \end{tikzcd} \end{center} In this model, the original value $O$ is connected only to the high-side input $S_H$ by a direct path of relays $R_0, \ldots, R_n$. We can define the origin property as: \begin{dfn}[Origin restricted] A system $\Y$ with inputs $\mc{S}$ is \emph{origin-restricted} for a protected variable $O$ iff all direct paths from $O$ to $\mathcal{S}$ end in $\mathcal{S}_H$, and there is at least one such path. \end{dfn} In what sense is an origin restricted system secure? We would like the low-side output of an origin restricted system to be independent of the sensitive variable. As we have seen in Sections \ref{sec:noninterference} and \ref{sec:prevention}, there are multiple security models that make different assumptions about the conditions of security. We can use analogous security models for origin privacy. \begin{dfn}[Origin noninterference] $\Y$ is secure by origin noninterference with respect to a sensitive variable $O \in \mathcal{E}$ iff $$A_L \independent O \vert S_L$$ \end{dfn} \begin{thm} \label{thm:origin-noninterference} Given an ECS model, if the system is present, safe, orderly, and origin restricted then the system is secure by origin noninterference. \end{thm} \begin{proof} Because the system is origin restricted, there is at least one direct path from $O$ to $S_H \in \mathcal{S}_H$. If there were an unblocked path from $A_L$ to $O$ that included an outgoing edge from $A_L$, this would extend into an unblocked path to $S_H$, violating the condition imposed by Lemma \ref{lem:present}. Therefore there is no unblocked path from $A_L$ to $O$ that includes an outgoing edge from $A_L$. Because the system is orderly, incoming edges to $A_L$ must go to nodes in the system. Therefore, any unblocked path from $O$ to $A_L$ must go through $\mathcal{S}$ Because the system is safe, there is no unblocked path from $\mathcal{S}_H$ to $A_L$. Because the system is orderly, any path from $E \in \mathcal{E}$ to $Y \in \Y$ going through $\mathcal{S}_L$ will include a head-to-tail triplet centered on $S_L$. Conditioning on this node $S_L$ blocks the path. Therefore there is no unblocked path between $O$ and $A_L$, and the system is secure by origin noninterference. \end{proof} \begin{dfn}[Origin semantic security] $\Y$ is secure by origin noninterference with respect to a sensitive variable $O \in \mathcal{E}$ iff $$A_L \independent O$$ \end{dfn} \begin{thm} Given an ECS model, if the system is present, safe, orderly, and origin-restricted and the world is propitious, then the system is secure by origin semantic security. \end{thm} \begin{proof} By Theorem \ref{thm:origin-noninterference}, the system is secure by origin noninterference. The system is origin restricted, implying that there is at least one direct path from $O$ to $S_H$. $S_L$ cannoth be on this path because the system is orderly. As no node on this path is in the conditioning set, it is not blocked. Mirroring the proof to \ref{thm:semantic}, we consider any path $\phi$ between $O$ and $A_L$ that was blocked by conditioning on $S_L$. Such path must have a node $S_L$ either within a head-to-tail triplet or as a common cause. Suppose $\phi$ includes $S_L$ in a head-to-tail triplet. Then there is a subpath of $\phi$ there is an unblocked path between $S_L$ and $O$. But there is also an unblocked path from $O$ to $S_H$, implying that there is an unblocked path from $S_L$ to $S_H$. This contradicts the condition that the world is propitious. Suppose $\phi$ includes a $S_L$ as a common cause node. Because the system is orderly, both outgoing edges must go to nodes in $\Y$. The path $\phi$ must therefore enter the system through a node in $S_H$. That implies a subpath of $\phi$ within the system runs unblocked from $S_H$ to $S_L$. That contradicts the condition that the world is propitious. Because no unblocked path between $O$ and $A_L$ is possible, $O \independent A_L$ and the system has origin semantic security. \end{proof} This demonstrates that origin restrictions do prevent associations between low-side outputs and the sensitive environmental variable under the condition that the systems are otherwise secure. % %\mct{It would be nice if the above theorems were if-and-only-ifs. I % suspect showing such a thing would be complicated by the possibility % of not having d-separation but the distributions having independence % anyhow.}% % % sb: that would be nice. We would need to 'faithfulness' property, % or use the 'there exists one network with this structure such that...' % techniques to establish soundness. % % %Note that while it is considered true in privacy literature %that it is impossible it is impossible to prevent %inferences from being drawn from data because %of the possible presence of auxiliary knowledge %\cite{dwork06icalp,dwork08jpc}, we can see an exception %to this rule when considering a system embedded in %its environment. %If there is in fact independence between the system and %and environmental variable, there is no auxiliary information %that would allow one to infer that environmental variable %from system state. % %If we consider an embedded causal system, %the structure of the world determines what %auxiliary knowledge is possible. %More granularly, this has implications %for what combinations of structural knowledge and %observations of particular variables are necessary %for an adversary to draw sensitive inferences from %system outputs. %This granularity has let us define a positive privacy %property, origin privacy, that allows for guarantees %about inferences about sensitive variables on the %condition that system designers have reliable %knowledge of the structure of the world. % \section{Use case: IoT and biometric data} \label{sec:usecase} In this section we introduce a use case of Origin Privacy that we have identified through legal analysis and conversations with stakeholders. \begin{exm}[Smart building biometric sensing] In an ``Internet of Things'' instrumented building, many sensors collect information about the contents of rooms, including photograph and other imagery such as infrared scanning to identify the number and size of people present. This information is useful to control the environment in the room (heating, ventilliation). However, this data can also be potentially identified using auxiliary information, such as a facial recognition database. This processed data reveals the identities of persons in the room. In some cases this may be intentional, as when it is used for building security. In other cases these revelations may be unexpected and constitute an invasion of privacy. \end{exm} We chose this example because it highlights the way smart building technology interacts with privacy policies around photography and biometric data. \subsection{GDPR biometric data} Here we focus particularly on the EU's General Data Protection Regulation (GDPR) (Regulation (EU) 2016/679). In general, the GDPR places a number of restrictions on the processing of \emph{personal data}, which it defines thus: \begin{quote} `personal data' means any information relating to an identified or identifiable natural person (`data subject'); an identifiable natural person is one who can be identified, directly or indirectly, in particular by reference to an identifier such as a name, an identification number, location data, an online identifier or to one or more factors specific to the physical, physiological, genetic, mental, economic, cultural or social identity of that natural person; (Article 4 \S 1, GDPR) \end{quote} We interpret this definition as referring to the topic or content of information; personal data is any information \emph{relating to} a natural person. As we have argued, a system designer cannot guarantee that a system does not process information relating to a natural person since these relations may be caused by nomic associations that are external to the system itself. Noting this difficulty with ensuring compliance, we can nevertheless continue to work with the more specific requirements relating to biometric data. In particular, the GDPR makes a distinction between photographs and biometric data: \begin{quote} The processing of photographs should not systematically be considered to be processing of special categories of personal data as they are covered by the definition of biometric data only when processed through a specific technical means allowing the unique identification or authentication of a natural person. (Recital 51, GDPR) \end{quote} By definition under the GDPR, biometric data is a form of personal data that results from particular kinds of processes: \begin{quote} 'biometric data' means personal data resulting from specific technical processing relating to the physical, physiological or behavioural characteristics of a natural person, which allow or confirm the unique identification of that natural person, such as facial images or dactyloscopic data; (Article 4 \S 14, GDPR) \end{quote} Unlike the definition of personal data, the definition of biometric data is an origin requirement because it refers to the causal flow of data from a class of processes. Using these legal requirements, we can now use Origin Privacy to formalize their semantics with respect to a system. \subsection{Formalizing GDPR requirements} Consider the smart building example as an ECS. Let $S_P$ be the photographic input to the system. Let $S_D$ be a database of identified photographs, originating from an external process $E_F$. Let $Y_B$ be a component of the system $\Y$ caused by $S_P$ and $S_D$; it includes imagery from $S_P$ that has been identified using the database $S_D$. \begin{center} \begin{tikzcd} \\ E_F \arrow[r] & S_D \arrow[r] & Y_B \arrow[r] & A_H\\ & S_P \arrow[ur] \arrow[rr] & & A_L\\ \end{tikzcd} \end{center} We note that the photographic input $S_P$ may indeed ``relate to'' natural persons in a systematic way if, for example, certain persons frequent the smart building on a regular schedule. Since these regularities (nomic associations) are unknown to the system designer, there is little she can do to guarantee the treatment of this information as personal data. What the system designer does have access to is the identified faces database, $S_D$. The \emph{process of identification} that results in the biometric building surveillance data $Y_B$ requires data from an identified source such as $S_D$. The system designer knows about the origin of $S_D$. Specifically, she knows that this data is sourced from $E_F$, a process that personally identifies the facial images within it. Knowing the environmental source is sensitive, they can impose the conditions for noninterference between $E_F$ and $A_L$: that no unblocked path exist within $Y$ between inputs originating in $E_F$ and $A_L$. This implies that both $S_D$ and $Y_B$ be d-separated from $A_L$ within $\Y$. Note that in the diagram above, $E_F$ is indeed d-separated from $A_L$ when the system is \emph{covered}, i.e. when none of its components are in the conditioning set. Intuitively, $Y_B$ is subject to restricted flow because it originates from the sensitive process $E_F$; it inherits this origin from one of its parent components, $S_D$. We build on this result to model more complicated aspects of GDPR compliance. For example, processing of personal information, including biometric information, is generally legal given the consent of the identified person. We can introduce identified sources into the ECS model by denoting the set of natural persons $\mathcal{I}$, and denoting a process that generates data from an identified person $X_i$ for $i \in I$. We can then place conditions on any data that is a result of causal flow from this source, as in this example specification: \begin{exm}[Disclosure specification] In the system, all outputs $A \in \A_L$ such that $A$ is a descendant of $X_i$ must also be a descendant of $Z_i$, where $i \in I$ is the identifier of a natural person, $X_i$ is personally identifiable information, and $Z_i$ is a disclosure agreement identifiable with that person. \end{exm} To the extent the GDPR controls on biometric information are use restriction or topic restrictions as opposed to an origin restriction, they cannot be automatically enforced based on origin alone. However, considering GDPR through the rigor of Origin Privacy clarifies some of its requirements as formal specification shows what knowledge is needed by the system designer for origin based policy enforcement. \section{Relationship to Differential Privacy} \label{sec:differential} In this section we will show the connection between origin privacy and differential privacy. Thus far we have defined origin privacy strictly in terms of conditional independence. This has been inspired by the formal security model of \emph{noninterference}. When assessing computational systems that process personal information, it is possible for privacy requirements to be looser than this strict security requirement. The particular case of privacy preserving statistical analysis of databases of personal information has motivated differential privacy as a formal privacy model \cite{dwork06icalp,dwork08jpc}. Formally, an algorithm $\A$ is $\epsilon$-differentially private if for all subsets $S \subset image(\A)$ and datasets $D_1$ and $D_2$ that differ on a single element, the probability of the output of the algorithm run on the datasets being in $S$ differs by at most a multiplicative factor of $e^\epsilon$ and an additive factor $\delta$ \cite{dwork2014algorithmic}. \begin{dfn}[Differential Privacy (($\epsilon,\delta$)-DP)] $$Pr[\A(D_1) \in S] \leq e^\epsilon Pr[\A(D_2) \in S] + \delta$$ \end{dfn} When $\delta = 0$, then $\A$ is $\epsilon$-differentially private ($\epsilon$-DP). Consistent with our origin privacy approach, we will investigate how differential privacy can be assessed given a model of system and its environment as a causal Bayes network. We will draw on prior results relating differential privacy and causality and variations on differential privacy expressed in terms of mutual information. \subsection{Mutual information differential privacy} \citet{cuff2016differential} demonstrate that there is an equivalence between differential privacy and what they define as mutual information differential privacy: \begin{dfn}[Mutual information differential privacy ($\epsilon$-MI-DP)] A randomized mechanism $P_{Y \vert X^n}$ satisfies $\epsilon$-mutual information differential privacy if, $$sup_{i, P_{X^n}} I(X_i, Y \vert X^{-i}) \leq \epsilon$$ where $X^{-i}$ is the dataset $X$ excluding variable $X_i$. \end{dfn} They prove that $\epsilon$-MI-DP is equivalent to differential privacy in the sense that $\epsilon$-MI-DP implies ($\epsilon,\delta$)-DP) for some $\delta$, though $\epsilon$-MI-DP is weaker than $\epsilon$-DP. \citet{mcsherry_2017} argues that that $\epsilon$-MI-DP falls short of the desiderata of $\epsilon$-DP. Nevertheless, we will proceed with the $\epsilon$-MI-DP because it is suggestive of probabilistic structure may be used to infer a privacy relevant bound. That the mutual information limit is conditioned on every other member of the database is an indication of a disappointing fact about differential privacy, which is that its beneficial properties with respect to preserving privacy are not robust to cases when entries in the database are correlated with each other. The value of using MI-DP for our purposes is that the properties of mutual information are well-understood, and we can derive a number of useful theorems about mutal information between variables in a Bayesian network. \subsection{Randomizing database inputs} We can now combine the previous results to show how a system designer can develop a Bayesian network model that guarantees differential privacy. A common way of achieving differential privacy is by randomizing inputs to the database prior to aggregation; this is done in practice in \citet{erlingsson2014rappor}. We can model randomization explicitly as in the following example. Let $X_i$ be a set of variables representing the personal information to be aggregated. Let $Y_i$ be a random variable over the same domain as $X_i$ that is almost but not quite independent of $X_i$; we'll say that the mutual information between $X_i$ and $Y_i$ is bounded by $\epsilon_i$. Then aggregate all the $Y_i$ variables into a database, $Z$, that is available for querying. \begin{center} \begin{tikzcd} X_0 \arrow[r,"\epsilon"] & Y_0 \arrow[ddr] & \\ X_1 \arrow[r,"\epsilon"] & Y_1 \arrow[dr] & \\ X_2 \arrow[r,"\epsilon"] & Y_2 \arrow[r] & Z \\ X_3 \arrow[r,"\epsilon"] & Y_3 \arrow[ur] & \\ X_4 \arrow[r,"\epsilon"] & Y_4 \arrow[uur] & \\ \end{tikzcd} \end{center} In the above diagram, we annotate an arrow between variables $A$ and $B$ with the upper bound on the mutual information $I(A,B)$ afforded by the conditional probability distribution. In this case, we have set all the $\epsilon_i$ equal to each other, $\epsilon$. We can now use this graphical structure to prove that this system is $2\epsilon$-MI-DP). We can prove this using the Data Processing Inequality, \begin{prp}[Data Processing Inequality] If three variables are in a Markov chain $$X \rightarrow Y \rightarrow Z$$ where $X \independent Z \vert Y$, then $I(X,Y) \geq I(X,Z)$ \end{prp} A standard proof of this is in the appendix. We will also use the Path Mutual Information Theorem (PMIT), that we prove as Theorem \ref{thm:path-mutual-information} in Appendix \ref{appendix:information-theory-theorems}. \begin{exm} For the structure described above, $Z$ is $2\epsilon$-MI-DP. \end{exm} \begin{proof} Note that because $X_i$ and $X^{-i}$ are joined only by a path with a common effect node, $I(X_i,X^{-i}) = 0$. It follows that: \begin{equation} \begin{split} I(X_i, Z \vert X^{-i})\\ = I(X_i;Z,X^{-i}) - I(X_i,X^{-i}) \\ = I(X_i;Z,X^{-i}) \\ = I(X_i;Z) + I(X_i;X^{-1} \vert Z) \end{split} \end{equation} By DPI and the graphical structure, we know that for all $i$ $$I(X_i,Z) \leq I(X_i,Y_i) = \epsilon$$ By PMIT, we know the mutual information of two variables connected by a path with all of its common effect nodes observed is bounded by the mutual information of steps along the path. In this case, it entails that: $$I(X_i;X^{-1} \vert Z) \leq I(X_i,Y_i) = \epsilon$$ By substitution, we know that: $$I(X_i, Z \vert X^{-i}) = I(X_i;Z) + I(X_i;X^{-1} \vert Z) \leq 2\epsilon$$ As this holds for all $i$, it follows that $Z$ is $2\epsilon$-MI-DP. \end{proof} \subsection{Generalizing $\epsilon$-security} We can generalize the two formal security models that we introduced in Section \ref{sec:noninterference} to models that allow for $\epsilon$ mutual information between sensitive inputs and low-side outputs. \begin{dfn}[$\epsilon$-noninterference] A system is secure by $\epsilon$-noninterference iff $$I(A_L, S_H \vert S_L) \leq \epsilon$$. \end{dfn} \begin{dfn}[$\epsilon$-semantic security] A system is $\epsilon$-semantically secure iff $$I(A_L, S_H] \leq \epsilon$$. \end{dfn} Recall that two variables are perfectly independent if and only if their mutual information is zero, implying that $0$-noninterference is equivalent to noninterference, and $0$-semantic security is equivalent to ECS semantic security. We can now show that differential privacy follows from an application of $\epsilon$-noninterference with one important caveat. We have defined noninterference in terms of a system's high-side and low-side inputs. Schematically, we have considered the ``high side'' to be a part of the system in need of special information flow restrictions, while the ``low side'' is what's available to less restricted access. This model is intended to be generalized to cases where there are multiple categories of restricted information. In particular, to prove that $\epsilon$-noninterference implies differential privacy, we must consider \emph{each entry individually} to be noninterferent with respect to the other entries in the data set. \begin{thm} For an ECS model, if for all $D_i$, ECS is secure by $\epsilon$-noninterference with respect to $D_i$ as a high-side input and $D^{-i}$ as low-side input, then $A$ is mutual information differentially private with respect to data set $D$ \end{thm} \begin{center} \begin{tikzcd} D_i \arrow[rd,"\epsilon"] & \\ D^{-i} \arrow[r] & A \\ \end{tikzcd} \end{center} \begin{proof} If for all $D_i$, ECS is secure by $\epsilon$-noninterference with respect to $D_i$ as a high-side input and $D^{-i}$ as low-side input, then $$\forall i, I(D_i, A_L \vert D^{-i}) \leq \epsilon$$ which implies that $$sup_{i, P_{X^n}} I(D_i, A_L \vert D^{-i}) \leq \epsilon$$ which is the condition for $\epsilon$-MI-DP. \end{proof} It is not generally the case that if each of a database's entries $D_i$ is $\epsilon$-semantically secure from the output $A$ that the output will be $\epsilon$-differentially private. A bound on $I(D_i;A)$ does not imply a bound on $I(D_i; A \vert D^{-i}$), as is apparent from Equation \ref{eq:mutual-with-semantic}. \begin{equation} \label{eq:mutual-with-semantic} I(D_i;A \vert D^{-i}) = I(D_i,D^{-1}; A) - I(D_i;A) \end{equation} \section{Incentives in ECS} \label{sec:incentives} We have motivated ECS modeling by showing how it captures the implicit ontology of privacy policies and enables reasoning about the security properties of systems embedded in an environment. We have used Bayesian networks as a modeling tool because they clarifying the relationship between two aspects of information flow. Bayesian networks also provide a robust, qualitative means of determining dependency relations between their variables, which we have used in our proofs about the relationship between various system and privacy properties. In this section, we will show how the same ECS framework can be extended to include strategic actors and their incentives. To accomplish this, we will use the Multi-Agent Influence Diagram (MAID) framework developed by \citet{koller2003multi}. In brief, a MAID is a Bayesian network with two important extensions. \begin{itemize} \item Some of the variables are reserved as \emph{decision variables} and assigned to one of several agents. An agent's assignment of CPDs to its decision variables is that agent's \emph{strategy}; replacing each decision variable with the CPD from a \emph{strategy profile} transforms the MAID into a Bayesian network. \item Some of the variables are reserved as \emph{utility variables}. These are assigned to one agent each, and are summed when realized into the agent's total utility. Utility variables must not have children. \end{itemize} A formal definition of MAIDS, strategies, and other useful properties is given in Appendix \ref{appendix:maid}, which includes an account of the graphical notation we will use in this Section. \subsection{Expert services model} \label{sec:expert-services-ECS} As an illustration of an information game that can be represented as a MAID, consider the following diagram, which is a generalized model of an expert service. (This model will be analyzed in more detail in the following chapter, in Section \ref{sec:expertise}.) The services, which include health services, legal services, as well as some software based services like search engines, involve a client, $c$, who presents knowledge of their situation to an expert, $e$. The expert has access to general knowledge relevant to the client's condition, and recommends some action based on this knowledge. The client can choose to take this recommendation. In the idealized case considered here, the client and the expert have perfectly aligned incentives and so the expert will use their knowledge to the best of their ability. \begin{center} \begin{tikzcd} & W \arrow[ddl, bend right = 40] \arrow[ddr, bend left = 40]& \\ & C \arrow[dl] \arrow[dd] \arrow[dr, bend left = 20, dotted]& \\ V \arrow[dd] \arrow[ddr] & & \tilde{R}_e \arrow[dl] \\ & \tilde{A}_c \arrow[dl] \arrow[d] &\\ \breve{U_e} & \breve{U_c} &\\ \end{tikzcd} \end{center} In this model, $W$ is the generalized knowledge of the world that is available to $e$ at their decision variable representing their recommendation, $\tilde{R}_e$. The action taken by the client $\tilde{A}_c$ and the action value function $V$ determine the utility of both the expert and the client, $\breve{U_e}$ and $\breve{U_c}$. The value function is influence both by general facts about the world $W$ and the particular situation of the client, $C$. The client knows their situation but not the general expert knowledge. The client's action is informed by the general knowledge only through the recommendation of the expert. The expert may or may not know the client's specific situation; this is represented by the dotted arrow between $C$ and $\tilde{R}_e$. The value of such a model is that a qualitative analysis can readily provide insights into the influence of an information flow on outcomes. Since we know the expert's utility depends on influencing the client to make the best possible action at $\tilde{A}_c$, and that the value of this action depends on $V$, the expert's effectiveness will be limited by how well they can predict $V$ given the knowledge availabel to them at $\tilde{R}_e$. Without information about their client's specific situation $C$, their advice can at best be perfectly general. But with access to information $C$, the expert can improve their recommendation and outcomes for both players. \subsection{Expert ECS Model} \label{sec:expert-ecs-model} We now combine the expert service model with the ECS model. We will embed the expert insider an ECS and give them access to personal information of the client via the high-side input. They will also have access to general knowledge through the low-side input. An adversary will have access to the low-side output, but not the high-side output. Using this model, we will be able to test how the security properties we have analyzed in Section \ref{sec:security} and Section \ref{sec:robustness} can be motivated in terms of the way they affect the incentives of interacting with the system. \begin{center} \begin{tikzcd} & & W \arrow[dddll, bend right = 40] \arrow[dl, dotted, "e_1"] \arrow[ddd] \arrow[dddr, bend left = 30] & \\ & C \arrow[d] \arrow[ddrr] \arrow[ddl] & & \\ & \tilde{E}_a \arrow[d] & & \\ V \arrow[dddd] \arrow[ddddr, bend right = 20] & \mc{S}_H \arrow[d] \arrow[ddr, dotted, "e_2"] & \mc{S}_L \arrow[dl] \arrow[dd] & V' \arrow[dddd] \arrow[ddddl, bend left = 20] \\ & \tilde{D}_b \arrow[d] & & \\ & \mc{A}_H \arrow[d] & \mc{A}_L \arrow[d] & \\ & \tilde{D}_a \arrow[d] \arrow[dl] & \tilde{D}_e \arrow[d] \arrow[dr]& \\ \breve{U}^+_a & \breve{U}_b & \breve{U}_e & \breve{U}^-_a\\ \end{tikzcd} \end{center} This game has three players, Alice ($a$), Bob ($b$), and Eve ($e$). Alice and Bob have perfectly aligned incentives, and Eve is an attacker who is adversarial to Alice. We specify that the following relations hold: $$\breve{U}^+_a = \breve{U}_b$$ $$\breve{U}^-_a = -\breve{U}_e$$ $$\breve{U}_a = \breve{U}_b - \breve{U}_e$$ At the center of this model is an ECS, with high- and low- side sensors ($\mc{S}_H, \mc{S}_L$) and actuators ($\mc{A}_H, \mc{A}_L$). In this model, Alice is aware of her personal information $C$ and decides at $\tilde{E}_a$ what if any of it to divulge secretly (via $\mc{S}_H$) into the system because she wants an expert recommendation from Bob. Bob has access to general expertise $W$ through a low-side input ($\mc{S}_L$). These inputs are both available to Bob at his decision node $\tilde{D}_b$, at which he chooses a recommended action for Alice, which he passes through the high-side output $\mc{A}_H$. Alice will use the information about the recommendation taken from the high-side output $\mc{A}_H$ to choose an action. The utility of this action will depend on the action values $V$, which are a function of two variables: the personal characteristics $C$ of Alice and other general information about the world, $W$. %\mct{Dummy cite: \cite{gm82security}} Eve will make a decision $\tilde{D}_e$ based on the low-side output of the system, $A_L$. Eve's utility $U_e$ depends on this decision and an action value function $V'$ that is analogous to Alice's action value function $V$, in that it depends on $C$ and $W$. The system in this diagram is $\mc{Y} = \{\mc{S}_H, \mc{S}_L, \tilde{D}_b, \mc{A}_H, \mc{A}_L\}$. The diagram has two dotted edges, $e_1$ and $e_2$ Each dotted edge may be either included in the graph (open) or excluded from the graph (closed). The diagram therefore describes four distinct models: none open, $e_1$ open, $e_2$ open, and both open. We will analyze each case in Section \ref{sec:expert-ecs-analysis}. \subsubsection{Analysis} \label{sec:expert-ecs-analysis} First, we can analyze the expert ECS model presented in Section \ref{sec:expert-ecs-model} in terms of the system design properties introduced in Section \ref{sec:design}. Note that no descendent of $\mc{A}$ is in $\mc{S}$. Therefore, the expert ECS model is present if none of $\tilde{D}_a, \tilde{D}_e, \breve{U}^+_a, \breve{U}_b, \breve{U}_e,\breve{U}^-_a$ are in the conditioning set $\mc{C}$. The system is covered if none of $\mc{S}_H, \mc{S}_L, \tilde{D}_b, \mc{A}_H, \mc{A}_L$ are in the conditioning set $\mc{C}$. It is plain from observation that the system is orderly. It is also clear that the system is origin-restricted with respect to the personal characteristics $C$: there is one direct path from $C$ to $\mc{S}$ and it goes to $\mc{S}_H$. We are left with several candidates for the conditioning set: $W, C, V, V', \tilde{E}_a$. Recall that the world is propitious if there are no unblocked paths from $\mc{S}_H$ to $\mc{S}_L$. There are two ways an unblocked path can happen under the conditions discussed so far. One is that either $V$ or $V'$ is in the conditioning set. Another is that the edge $e_1$ is open. It is clear that the system is safe if edge $e_2$ is closed and not safe if edge $e_2$ is open. Suppose that there are no variables in the conditioning set. Then by the reasoning above, the following properties of the expert ECS system hold: \begin{itemize} \item If $e_1$ and $e_2$ are closed, then the system is origin secure with respect to $C$ both semantically and by noninterference. \item If $e_1$ is open, then the system will be origin secure with respect to $C$ by noninterference, but may not be semantically secure. \item If $e_2$ is open, then the system may not be origin secure with respect to $C$ either by noninterference nor semantically. \end{itemize} Though we have been able to show that these security properties hold on the expert ECS model, this model also reveals how these security properties do not provide all desireable guarantees a system might provide in terms of the incentives of Alice and Bob. What can be shown is that given that the expert ECS system is semantically secure, it is also the case that $\tilde{E}_a$ and $\tilde{D}_e$ are tactically independent (see Definition \ref{dfn:tactical-independence}), meaning that for any strategy specifying decision rules to each decision variable, in the induced probability distribution $\tilde{E}_a$ and $\tilde{D}_e$ are independent. In other words, at the level of tactics, Alice's choice to reveal her information to the ECS will not depend on Eve's choice of how to use the system's low-side outputs adversarially. However, despite these security properties, we can show with this model that $\tilde{E}_a$ may \emph{strategically rely} on $\tilde{D}_e$ (see Definition \ref{dfn:strategic-reliance}). This means that Alice's choice of decision rule at $\tilde{E}_a$ can depend on Eve's choice of decision rule at $\tilde{D}_e$. This can be a problem for system designers if their goal is to guarantee that the presence of Eve has no deterring effect on Alice's choice to reveal her data to the ECS. Further exploration of the relationship between system security properties and incentives of players in this causal formalism is left to future work. \section{Discussion and future work} \label{sec:future} We have analyzed privacy policies and discovered that they variously restrict information based on origin and topic. We developed an informational ontology that reflects the assumptions underlying many of these policies. We have shown that this informal ontology can be formalized in terms of causal graphical models. These models show that the two aspects of information flow correspond to precisely defined concepts of causal flow and nomic association. Both senses of information flow are accomodated by an understanding of situated information flow as a causal flow in causal context, as represented by a Bayesian network. We developed a model of system security, the ECS model, which represents a system embedded in its environment. This model can demonstrate and extend known results in computer security, such as those concerning noninterference, semantic security, and differential privacy. It also allows us to formally define a new privacy property, origin privacy, which assumes that system designers have some control over the paths through which information enters their systems. We demonstrate how the ECS model can be used to elucidate a case of implementing GDPR compliance on biometric data. We demonstrated preliminery results on how the ECS model can be extended into game theoretic form to account for how strategically acting agents interact with systems with or without relevant security properties. These contributions are suggestive of several lines of future work. \subsection{Specifiability Criterion} One direction for future work is to question what these results mean for the design and interpretation of legal policies. Are privacy policies based on information topic harder for consumers to understand than policies based on information origin? How do end users interpret ambiguous language in privacy policies? One benefit of using causal models as opposed to program analysis for considering information flow security of technical systems is that it shows that information flow security is not only a technical problem. Because information leaks based on nomic associations may be properties of any mathematically defined causal system, these results extend to human institutions as well. In general, mathematical results about the limits of computational enforceability will generalize to the possibility of enforceability through non-computational means.\footnote{This claim assumes the Church-Turing thesis.} Privacy policies and social expectations that restrict information based on its contents may be unenforceable or ambiguous in general. Concretely, the brittleness of these kinds of restrictions is exposed by advances in large-scale data analysis, or ``big data'', which routinely confound our expectations about what information is about. Data about a person's purchases of scent-free hand lotion, which we might have assumed to be innocuous, has been shown to be correlated with sensitive information about early-stage pregnancy \cite{hill12forbes}. Ad targeting systems that use all available correlational information in the data they collect risk violating people's privacy due to unexpected discoveries from automated learning processes. By demonstrating the limits of what security properties can be enforced, and by whom, this framework can shed light on how privacy policies should be written and which parties should be held liable in the event of a violation. This may build on other work in identifying legal liability in cases of perceived inappropriate behavior of a sociotechnical system \cite{datta2018discrimination}. \subsection{Observer capabilities} \label{sec:observer-capabilities} We have shown that nomic associations between system outputs and sensitive environmental variables can lead to violations of privacy policies. In order for these threats to be material, nomic associations must be known to attackers as auxiliary information. A natural next step in this line of inquiry is a more systematic study of how observer's capabilities for learning nomic associations factor into information flow security considerations. In our models in this article, we have used Bayesian networks as models of the objective frequencies of variable outcomes. Possible inferences from observed variables have been interpreted as those inferences possible \emph{in fact} from the causal structure of the world. Information flow security guarantees were possible when the system designer was assumed to have some true knowledge about the system's environment, such as the origin of its inputs. In practice, most systems will be embedded in environments that are only partially known. In the cases of fraud and spam detection, and other adaptive machine learning systems, a model of the origin of inputs is trained continuously from collected data. Probabilistic graphical models are indeed one of the many machine learning paradigms used in these applications \cite{bishop2006pattern}. A direction for future work is developing a theory of information flow security and privacy under conditions where observer knowledge of nomic associations is itself a function of system inputs. In simple cases this may reduce to single program analyses already established in work on use privacy \cite{datta2017use}. In cases where multiple systems interact, there may be novel problems. \subsection{Incentives based policy design} Prior work has been done on game theoretic models of information flow policies \cite{barth07csf}, mechanism design with differential privacy \cite{mcsherry2007mechanism}, and semantic interpretations of differential privacy framed in terms of the incentives of data subjects \cite{kasivisiwanathan14jpc}. We see potential to continue this line of work using statistical models of systems, their internal processes, and their environment, including the process that generate their input data. We have shown how Multi-Agent Influence Diagrams (MAIDS) \cite{koller2003multi} can be used in game theoretic modeling of security problems. \citet{feng2014security} have used Bayesian networks to model security risks and perform a security vulnerability propagation analysis. They build their causal model from observed cases and domain experts. We anticipate new frameworks for aligning the incentives of system designers and data subjects which are sensitive to risk of data misuse (see \cite{brooks15nist}). An application of this work is automated policy design for smart buildings and cities, where interacting data subjects and control systems must share information while minimizing data misuse. Our work in Section \ref{sec:incentives} is a first step in developing a new way of assessing the value of security properties based on their impact on game outcomes. This method of modeling information value through data games is the subject of the next chapter. \hfill \hfill We have unpacked the assumptions of privacy policies to develop a general ontology of systems, processes, and information. We have then formalized this ontology using the assumption that these systems are subject to the laws of probability and an interventionist account of causation \cite{woodward2005making}. Using these simple assumptions, we have confirmed well-known results about information flow security about programs and systems in isolation. We have also gone beyond these results by showing explicitly what their consequences are when systems are embedded in an environment, developing a general security modeling framework for this purpose. We prove the relative effectiveness of origin based information flow restrictions over association based information flow restrictions using the causally embedded system framework. We show how origin privacy would be applied in a GDPR and Internet of Things use case. We anticipate new lines of inquiry extending from the intersection of causal modeling and information flow security, including evaluation of policy enforceability, modeling the role of observer knowledge in privacy and security, and automated policy generation through incentives-based design. \end{document} docs/1st term/Algebra/Aut_Mid/Body/P1_s.tex0 \section{1 Диофантовы уравнения} \subsection{1 вариант} 1)\\ Рассмотрим $ax+by = k$, разделим все, если возможно(иначе корней нет), на $\gcd(a,b)$,\\ Получим $\frac{a}{\gcd(a,b)} x+ \frac{b}{\gcd(a,b)} y = \frac{k}{\gcd(a,b)} \ = \ a_0 x + b_0 y = k_0$, теперь $\gcd(a_0,b_0) = 1$\\ По алгоритму евклида найдем 1 пару $a_1,\ b_1$ при которой равенство выполнено, тогда все решения можно записать как:\\ \begin{gather*} \begin{cases} x_n \ = \ a_1 + n \cdot \frac{a}{\gcd(a,b)} \ = \ a_1 + n \cdot a_0\\ y_n \ = \ b_1 - n \cdot \frac{b}{\gcd(a,b)} \ = \ b_1 - n \cdot b_0 \end{cases} \quad n \in \mathbb{Z}\\ \text{gcd(a,b) \ = \ НОД(a,b) \ = \ (a,b)} \end{gather*} 2)\\ Рассмотрим $ax+by+cz = k$. Если $k \mod \gcd(a, b, c) = 0$, то у уравнения есть решения, иначе их нет.\\ Пусть $p = \gcd(a, b)$, и $a^{\star} = \frac{a}{p} \quad b^{\star} = \frac{b}{p}$\\ Тогда решим уравнение $a^{\star} u + b^{\star} v = c$ -- его решения $u_0$ и $v_0$ (по (1) пункту)\\ $z_0$ и $t_0$ -- решения $cz + pt = d$ (по (1) пункту)\\ $x_0$ и $y_0$ -- решения $a^{\star} x + b^{\star} y = t_0$ (по (1) пункту)\\ Тогда решения системы это: \begin{gather*} \begin{cases} x \ = \ x_0 + b^{\star} k - u_0 m \\ y \ = \ y_0 - a^{\star} k - v_0 m \\ z \ = \ z_0 + p m \end{cases} \\ k, m \in \mathbb{Z} \end{gather*} \subsection{2 вариант} \begin{gather*} ax + by = k,\ c = (a, b)\\ \frac{a}{c} \cdot cx + \frac{b}{c} \cdot cy = k,\ a_0 = \frac{a}{c},\ b_0 = \frac{b}{c}\\ c(\frac{a}{c} \cdot x + \frac{b}{c} \cdot y) = k\\ c(a_0 x + b_0 y) = k \ \Rightarrow c | k,\ k_0 = \frac{k}{c}\\ a_0 x + b_0 y = k_0\\ a_0 x_1 + b_0 y_1 = 1 \Leftrightarrow (a_0, b_0) = 1\\ x = x_1 \cdot k_0 + b_0 \cdot n\\ y = y_1 \cdot k_0 - a_0 \cdot n\\ \end{gather*} \\ $ax + by + cz = k,\ d = (a, b)$\\ Заметим, что $((a, b), c) = (a, b, c)$\\ $k_0 \cdot (a, b) = ax + by \ \Rightarrow \ (a, b)k_0 + cz = k$ Свели к предыдущей задаче. \newpage \section{2} \subsection{1 вариант} Рассмотрим поле $L$ разложения многочлена $x^{p^n} - x$ над полем $\mathbb{Z}_p$.\\ У данного многочлена нет кратных корней (так как его производная равна $-1$ и взаимно просто с самим многочленом), поэтому все корни многочлена $x^{p^n} - x$, лежащие в $L$, различны.\\ Колическтво таких корней равно $q = p^n$.\\ Докажем, что множество образует поле.\\ Действительно, если $a^q = a$; $b^q = b$, то $(ab)^q = ab$. Значит множество замкнуто по умножению, тоже самое для сложения. $\Rightarrow$ нашли искомое поле из $p^n$ элементов.\\ Докажем, что все поля из $p^n$ элементов изоморфны.\\ Как мы показали выше, поле из $p^n$ элементов обязательно является полем разложения многочлена $x^{p^n} - x$. Так мы доказали, что поле разложение многочлена единственно с точностью до изоморфизма, единственность доказана полностью. \subsection{2 вариант} Докажем что все конечные поля одинакового порядка изоморфны\\ Рассмотрим поля $A$ и $B$ порядка $p^n$. Пусть $a \in A$ и $b \in B$ -- примитивные элементы полей. Ненулевых элементов в $A$ и $B$ ровно $p^n - 1$. \\ У многочлена $x^{p^n - 1} - 1$ ровно $p^n - 1$ ненулевых корней. Все эти корни различны и лежат как в $A$, так и в $B$. Тогда, так как порядки полей совпадают, то некий $\alpha \in A$ перешел в $\beta \in B$. И тогда $\alpha^k = \beta$, а это отоношение задает изоморфизм полей. \\ 2: \begin{center} \begin{tabular}{|c|c|c|} \hline $+ $&$ 0 $&$ 1 $\\ \hline $0 $&$ 0 $&$ 1 $\\ \hline $1 $&$ 1 $&$ 0 $\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|} \hline $\times $&$ 0 $&$ 1 $\\ \hline $0 $&$ 0 $&$ 0 $\\ \hline $1 $&$ 0 $&$ 1 $\\ \hline \end{tabular} \end{center} 4: \begin{center} \begin{tabular}{|c|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ x $&$ x+1 $\\ \hline $0 $&$ 0 $&$ 1 $&$ x $&$ x+1$\\ \hline $1 $&$ 1 $&$ 0 $&$ x+1 $&$ x$\\ \hline $x $&$ x $&$ x+1 $&$ 0 $&$ 1$\\ \hline $x+1 $&$ x+1 $&$ x $&$ 1 $&$ 0$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ x $&$ x+1 $\\ \hline $0 $&$ 0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ x $&$ x+1$\\ \hline $x $&$ 0 $&$ x $&$ 0 $&$ 1$\\ \hline $x+1 $&$ 0 $&$ x+1 $&$ 1 $&$ 0$\\ \hline \end{tabular} \end{center} 8: \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ x $&$ x+1 $&$ x^2 $&$ x^2+1 $&$ x^2+x $&$ x^2+x+1$\\ \hline $0 $&$ 0 $&$ 1 $&$ x $&$ x+1 $&$ x^2 $&$ x^2+1 $&$ x^2+x $&$ x^2+x+1$\\ \hline $1 $&$ 1 $&$ 0 $&$ x+1 $&$ x $&$ x^2+1 $&$ x^2 $&$ x^2+x+1 $&$ x^2+x$\\ \hline $x $&$ x $&$ x+1 $&$ 0 $&$ 1 $&$ x^2+x $&$ x^2+x+1 $&$ x^2 $&$ x^2+1$\\ \hline $x+1 $&$ x+1 $&$ x $&$ 1 $&$ 0 $&$ x^2+x+1 $&$ x^2+x $&$ x^2+1 $&$ x^2$\\ \hline $x^2 $&$ x^2 $&$ x^2+1 $&$ x^2+x $&$ x^2+x+1 $&$ 0 $&$ 1 $&$ x $&$ x+1$\\ \hline $x^2+1 $&$ x^2+1 $&$ x^2 $&$ x^2+x+1 $&$ x^2+x $&$ 1 $&$ 0 $&$ x+1$&$ x$\\ \hline $x^2+x $&$ x^2+x $&$ x^2+x+1 $&$ x^2 $&$ x^2+1 $&$ x $&$ x+1 $&$ 0 $&$ 1$\\ \hline $x^2+x+1 $&$ x^2+x+1 $&$ x^2+x $&$ x^2+1 $&$ x^2 $&$ x+1 $&$ x $&$ 1 $&$ 0$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ x $&$ x+1 $&$ x^2 $&$ x^2+1 $&$ x^2+x $&$ x^2+x+1$\\ \hline $0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ x $&$ x+1 $&$ x^2 $&$ x^2+1 $&$ x^2+x $&$ x^2+x+1$\\ \hline $x $&$ 0 $&$ x $&$ x^2 $&$ x^2+x $&$ x+1 $&$ 1 $&$ x^2+x+1 $&$ x^2+1$\\ \hline $x+1 $&$ 0 $&$ x+1 $&$ x^2+x $&$ x^2+1 $&$ x^2+x+1 $&$ x^2 $&$ 1 $&$ x$\\ \hline $x^2 $&$ 0 $&$ x^2 $&$ x+1 $&$ x^2+x+1 $&$ x^2+x $&$ x $&$ x^2+1 $&$ 1$\\ \hline $x^2+1 $&$ 0 $&$ x^2+1 $&$ 1 $&$ x^2 $&$ x $&$ x^2+x+1 $&$ x+1 $&$ x^2+x$\\ \hline $x^2+x $&$ 0 $&$ x^2+x $&$ x^2+x+1 $&$ 1 $&$ x^2+1 $&$ x+1 $&$ x $&$ x^2$\\ \hline $x^2+x+1 $&$ 0 $&$ x^2+x+1 $&$ x^2+1 $&$ x $&$ 1 $&$ x^2+x $&$ x^2 $&$ x+1$\\ \hline \end{tabular} \end{center} 3: \begin{center} \begin{tabular}{|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ 2$\\ \hline $0 $&$ 0 $&$ 1 $&$ 2$\\ \hline $1 $&$ 1 $&$ 2 $&$ 0$\\ \hline $2 $&$ 2 $&$ 0 $&$ 1$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ 2$\\ \hline $0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ 2$\\ \hline $2 $&$ 0 $&$ 2 $&$ 1$\\ \hline \end{tabular} \end{center} 9: \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ 2 $&$ x $&$ x+1 $&$ x+2 $&$ 2x $&$ 2x+1 $&$ 2x+2$\\ \hline $0 $&$ 0 $&$ 1 $&$ 2 $&$ x $&$ x+1 $&$ x+2 $&$ 2x $&$ 2x+1 $&$ 2x+2$\\ \hline $1 $&$ 1 $&$ 2 $&$ 0 $&$ x+1 $&$ x+2 $&$ x $&$ 2x+1 $&$ 2x+2 $&$ 2x$\\ \hline $2 $&$ 2 $&$ 0 $&$ 1 $&$ x+2 $&$ x $&$ x+1 $&$ 2x+2 $&$ 2x+1 $&$ 2x$\\ \hline $x $&$ x $&$ x+1 $&$ x+2 $&$ 2x $&$ 2x+1 $&$ 2x+2 $&$ 0 $&$ 1 $&$ 2$\\ \hline $x+1 $&$ x+1 $&$ x+2 $&$ x $&$ 2x+1 $&$ 2x+2 $&$ 2x $&$ 1 $&$ 2 $&$ 0$\\ \hline $x+2 $&$ x+2 $&$ x $&$ x+1 $&$ 2x+2 $&$ 2x $&$ 2x+1 $&$ 2 $&$ 0 $&$ 1$\\ \hline $2x $&$ 2x $&$ 2x+1 $&$ 2x+2 $&$ 0 $&$ 1 $&$ 2 $&$ x $&$ x+1 $&$ x+2$\\ \hline $2x+1 $&$ 2x+1 $&$ 2x+2 $&$ 2x $&$ 1 $&$ 2 $&$ 0 $&$ x+1 $&$ x+2 $&$ x$\\ \hline $2x+2 $&$ 2x+2 $&$ 2x $&$ 2x+1 $&$ 2 $&$ 0 $&$ 1 $&$ x+2 $&$ x $&$ x+1$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ 2 $&$ x $&$ x+1 $&$ x+2 $&$ 2x $&$ 2x+1 $&$ 2x+2$\\ \hline $0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ 2 $&$ x $&$ x+1 $&$ x+2 $&$ 2x $&$ 2x+1 $&$ 2x+2$\\ \hline $2 $&$ 0 $&$ 2 $&$ 1 $&$ 2x $&$ 2x+2 $&$ 2x+1 $&$ x $&$ x+2 $&$ x+1$\\ \hline $x $&$ 0 $&$ x $&$ 2x $&$ 2 $&$ x+2 $&$ 2x+2 $&$ 1 $&$ x+1 $&$ 2x+1$\\ \hline $x+1 $&$ 0 $&$ x+1 $&$ 2x+2 $&$ x+2 $&$ 2x $&$ 1 $&$ 2x+1 $&$ 2 $&$ x$\\ \hline $x+2 $&$ 0 $&$ x+2 $&$ 2x+1 $&$ 2x+2 $&$ 1 $&$ x $&$ x+1 $&$ 2x $&$ 2$\\ \hline $2x $&$ 0 $&$ 2x $&$ x $&$ 1 $&$ 2x+1 $&$ x+1 $&$ 2 $&$ 2x+2 $&$ x+2$\\ \hline $2x+1 $&$ 0 $&$ 2x+1 $&$ x+2 $&$ x+1 $&$ 2 $&$ 2x $&$ 2x+2 $&$ x $&$ 1$\\ \hline $2x+2 $&$ 0 $&$ 2x+2 $&$ x+1 $&$ 2x+1 $&$ x $&$ 2 $&$ x+2 $&$ 1 $&$ 2x$\\ \hline \end{tabular} \end{center} 5: \begin{center} \begin{tabular}{|c|c|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $\\ \hline $0 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4$\\ \hline $1 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 0$\\ \hline $2 $&$ 2 $&$ 3 $&$ 4 $&$ 0 $&$ 1$\\ \hline $3 $&$ 3 $&$ 4 $&$ 0 $&$ 1 $&$ 2$\\ \hline $4 $&$ 4 $&$ 0 $&$ 1 $&$ 2 $&$ 3$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $\\ \hline $0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4$\\ \hline $2 $&$ 0 $&$ 2 $&$ 4 $&$ 1 $&$ 3$\\ \hline $3 $&$ 0 $&$ 3 $&$ 1 $&$ 4 $&$ 2$\\ \hline $4 $&$ 0 $&$ 4 $&$ 3 $&$ 2 $&$ 1$\\ \hline \end{tabular} \end{center} 7: \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|} \hline $+ $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6$\\ \hline $0 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6$\\ \hline $1 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6 $&$ 0$\\ \hline $2 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6 $&$ 0 $&$ 1$\\ \hline $3 $&$ 3 $&$ 4 $&$ 5 $&$ 6 $&$ 0 $&$ 1 $&$ 2$\\ \hline $4 $&$ 4 $&$ 5 $&$ 6 $&$ 0 $&$ 1 $&$ 2 $&$ 3$\\ \hline $5 $&$ 5 $&$ 6 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4$\\ \hline $6 $&$ 6 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5$\\ \hline \end{tabular} \end{center} \begin{center} \begin{tabular}{|c|c|c|c|c|c|c|c|} \hline $\times $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6$\\ \hline $0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0 $&$ 0$\\ \hline $1 $&$ 0 $&$ 1 $&$ 2 $&$ 3 $&$ 4 $&$ 5 $&$ 6$\\ \hline $2 $&$ 0 $&$ 2 $&$ 4 $&$ 6 $&$ 1 $&$ 3 $&$ 5$\\ \hline $3 $&$ 0 $&$ 3 $&$ 6 $&$ 2 $&$ 5 $&$ 1 $&$ 4$\\ \hline $4 $&$ 0 $&$ 4 $&$ 1 $&$ 5 $&$ 2 $&$ 6 $&$ 3$\\ \hline $5 $&$ 0 $&$ 5 $&$ 3 $&$ 1 $&$ 6 $&$ 4 $&$ 2$\\ \hline $6 $&$ 0 $&$ 6 $&$ 5 $&$ 4 $&$ 3 $&$ 2 $&$ 1$\\ \hline \end{tabular} \end{center} \newpage \section{3} \subsection{1 вариант} Приводимые в $\mathbb{F}_3$ приводимы и в $\mathbb{F}_9$.\\ В $\mathbb{F}_3$ неприводимые это $x^2 + 1$; $x^2 + x + 2$; $x^2 + 2x + 2$\\ В $\mathbb{F}_9$ $y^2 + 1 = 0$; $y^2 + y + 2$ имеет корень $(x + 1)$; $y^2 + 2y + 2$ имеет корень $(x + 2)$. \subsection{2 вариант} Рассмотрим элементы $F_9$. Определим, что для каждого элемента $x \in F_3$\\ \begin{gather*} \sqrt{x} = \begin{cases} 0 \quad \text{при} \quad x = 0 \\ 1 \quad \text{при} \quad x = 1 \\ \overline{x} \quad \text{при} \quad x = -1 \end{cases} \end{gather*} Заметим, что $\sqrt{x}^2 = x$.\\ Проверим, что $y = \frac{-b + \sqrt{b^2 - 4ac} }{2a}$ является корнем $ax^2 + bx + c = 0$\\ $\frac{(-b + \sqrt{b^2 - 4ac})^2}{4a^2} \cdot a + \frac{-b + \sqrt{b^2 - 4ac}}{2a} \cdot b + c = \frac{b^2 + b^2 - 4ac - 2\cdot b\cdot\sqrt{b^2 - 4ac} - 2\cdot b^2 + 2\cdot b\cdot\sqrt{b^2 - 4ac} + 4ac}{4a} = 0$. \newpage \section{4} \subsection{1 вариант} Покажем, что в $F_{16} = F[y]/(y^2 + y\overline{(x+1)} + 1)$ есть корни.\\ Заметим, что если любое уравнение можно свести к $x^2 + x + c = 0$ или $x^2 + c = 0$. Докажем это: пусть уравнение вида $ax^2 + bx + c = 0$ (считаем $a$ не нулевым), тогда оно эквивалентно уравнению $x^2 + \frac{b}{a}x + \frac{c}{a}$. Сделаем замену переменных: $z \cdot \frac{b}{a} = x$. Тогда уравнение эквивалентно $z^2 + z + \frac{c \cdot a^2}{b^2} = 0$ с заменой корней (при $b \ne 0$, иначе эквивалентно уравнению вида $x^2 + c = 0$). Заметим, что $x^2 + x + c = 0$ имеют корни в $F16$: \\ \begin{gather*} x^2 + x = 0 \qquad \ \ x = 0; \quad x = 1 \\ x^2 + x + 1 = 0 \quad x = \overline{x}; \quad x = \overline{x+1} \end{gather*} Примечание: \\$(y \cdot \overline{x} + \alpha)^2 = y^2 \cdot \overline{x+1} + \alpha^2 = y \cdot \overline{x} + \overline{x+1} + \alpha^2$, поэтому решение двух оставшихся уравнений сводится к двум первым: $(y \cdot \overline{x} + \alpha)^2 + (y \cdot \overline{x} + \alpha) + c = \alpha^2 + \alpha + \overline{x+1} + c$\\ \begin{gather*} x^2 + x + \overline{x} = 0 \qquad \ \ x = y \cdot \overline{x} + \overline{x}; \quad x = y \cdot \overline{x} + \overline{x+1} \\ x^2 + x + \overline{x+1} = 0 \quad x = y \cdot \overline{x}; \qquad \ \ x = y \cdot \overline{x} + 1 \end{gather*} Покажем, что у уравнений вида $x^2 + c = 0$ есть решения: \\ $0 \cdot 0 = 0$, откуда $x^2 + 0 = 0$ имеет корень. \\ $1 \cdot 1 = 1$, откуда $x^2 + 1 = 0$ имеет корень. \\ $\overline{x} \cdot \overline{x} = \overline{x+1}$, откуда $x^2 + \overline{x+1} = 0$ имеет корень. \\ $\overline{x+1} \cdot \overline{x+1} = \overline{x}$, откуда $x^2 + \overline{x} = 0$ имеет корень. \subsection{2 вариант} $\mathbb{Z}_4 = \mathbb{Z}_2[x]/(x2 + x + 1)$ \begin{gather*} \begin{matrix} y^2,\ 0 & y^2 + y,\ 0 & y^2 + yx,\ 0 & y^2 + y(x + 1),\ x + 1 \\ y^2 + 1,\ 1 & y^2 + y + 1,\ x & y^2 + yx + 1,\ \varnothing & y^2 + y(x + 1) + 1,\ \varnothing \\ y^2 + x,\ x + 1 & y^2 + y + x,\ \varnothing & y^2 + yx + x,\ \varnothing & y^2 + y(x + 1) + x,\ 1\\ y^2 + x + 1,\ x & y^2 + y + x + 1,\ \varnothing & y^2 + yx + x + 1,\ \varnothing & y^2 + y(x + 1) + x + 1,\ \varnothing \end{matrix} \\ (\mathbb{Z}_2[x]/(x2 + x + 1))[y]/(y2 + y + x)\\ z^2 + z + x + 1, z = y + x \Rightarrow = (y + x)^2 + (y + x) + x + 1 = \\ = y^2 + x^2 + y + x + x + 1 = y + x + x + 1 + y + x + x + 1 = 0 \\ \\ z^2 + zx + 1, z = xy \Rightarrow = x^2y^2 + x^2y + 1 = (x + 1)(x + y) + (x + 1)y + 1 = \\ = x^2 + xy + x + y + xy + y + 1 = x + 1 + xy + x + y + xy + y + 1 = 0 \\ \\ z^2 + zx + x, z = yx + 1 \Rightarrow = (yx + 1)^2 + (yx + 1)x + x = \\ = y^2x^2 + 1^2 + x(yx + 1 + 1) = y^2x^2 + 1 + yx^2 = x^2(y^2 + y) + 1 = (x + 1)x + 1 = 0 \\ \\ z^2 + z(x + 1) + 1, z = y(x + 1) + 1 \Rightarrow = (y(x + 1) + 1)^2 + y(x + 1)^2 + (x + 1) + 1 = \\ = (y + x)x + 1 + yx + x = yx + x + 1 + 1 + yx + 1 = 0 \\ \\ z^2+z(x+1)+x+1, z = y(x+1) \Rightarrow = x(y+x)+yx+x+1 = yx+x+1+yx+x+1 = 0 \end{gather*} \newpage \section{5} \subsection{1 вариант} Рассмотрим группу обратимых для $n = 12 \quad \mathbb{Z}/ 12 \mathbb{Z}$, это $\{ 1,\ 5,\ 7,\ 11 \}$.\\ Заметим, что $5 \cdot 5 = 25 = 1,\quad 7 \cdot 7 = 49 = 1,\quad 11 \cdot 11 = -1 \cdot -1 = 1$, откуда следует, что эта группа не циклична. \subsection{2 вариант} Рассмотрим группу обратимых для $n = 8 \quad \mathbb{Z}/ 8 \mathbb{Z}$, это $\{ 1,\ 3,\ 5,\ 7 \}$.\\ Заметим, что $3 \cdot 3 = 9 = 1,\quad 5 \cdot 5 = 25 = 1,\quad 7 \cdot 7 = -1 \cdot -1 = 1$, откуда следует, что эта группа не циклична. \newpage \section{6} \subsection{1 вариант} Докажем, что существует первообразный корень в $Z/pZ$.\\ \textbf{(1) Т.Ферма}: $\alpha^{p-1} = 1$ при $\alpha \ne 0$. Доказательство:\\ Рассмотрим всевозможные произведения $\alpha$ на другие элемента поля. Так как это поле, значит в нём нет делителей $0$, откуда не может быть такого, что $\alpha*x = \alpha * y$ при $x \ne y$, поэтому всевозможные произведения различны. Откуда следует, что $\alpha \dot 2\alpha \dot ... \dot (p-1)\alpha = (p-1)! <=> a^{p-1}(p-1)! = (p-1)! <=> a^{p-1} = 1$ $((p-1)! \ne 0)$.\\ \\ \textbf{(2) Лемма}: $n = \sum \phi (i)$, где $i$ пробегает по всем делителям $n$ (Здесь мы работаем в натуральных числах).\\ Доказательство:\\ Будем говорить, что $\alpha \in [1,n]$ принадлежит множеству $M_i$ (где $i$ -- делитель $n$), если $\frac{\alpha}{\frac{n}{i}}$ целое, меньше $i$ и взаимнопросто с $i$. Нетрудно видеть, что каждое $\alpha$ может принадлежать не более $1$ множеству, так как то, что $\alpha \in M_{i_1} \ \Leftrightarrow \ (\alpha,\ n) = \frac{n}{i_1}$. Также видно, что любое $\alpha$ принадлежит хоть какому то множеству.\\ Теперь заметим, что в каждом множестве $M_i$ ровно $\phi(i)$ элементов, так как таких $\alpha:\ \frac{\alpha}{n/i} \in Z$ - $i$ $(\frac{n}{i}, \frac{2n}{i},\ ... \: , \frac{in}{i})$, при этом среди чисел в промежутке $[1,i]$ взаимнопростых $\phi(i)$. Откуда следует то, что и требовалось доказать.\\ \\ \textbf{(3)} Заметим, что элементов порядка $k$ либо $0$, либо $\phi(k)$. Доказательство:\\ Предположим есть хотя бы $1$. (элемент $g$) Тогда элементы вида $g,\ g^2,\ g^3,\ ... ,\ g^k$ различны и являются корнями уравнения $x^k - 1 = 0$, откуда следует, что других корней нет, при этом если $\alpha \in [1,k]$ не взаимнопросто с $k$ (пусть $(\alpha,k) = y$), то порядок у $g^{\alpha} = \frac{k}{y}$, что не равно k. Поэтому элементов порядка $k$ ровно $\phi(k)$.\\ \\ Следствие \textbf{(1)}, \textbf{(2)} и \textbf{(3)}:\\ Заметим, что если $k$ - не делитель $p-1$, то чисел порядка $k$ - 0, так как порядок больше чем $p-1$ быть не может (так как иначе среди чисел $g,\ g^2,\ ...,\ g^i$ найдутся $2$ одинаковых, и тогда разделим одно на другое и получим, что порядок меньше, чем предполагался, откуда следует противоречие), при этом любое число в степени $p-1 = 1$. Теперь рассмотрим все $k$, которые делят $p-1$. Заметим, что для всякого $k$ количество чисел порядка $k$ не больше чем $\phi(k)$, при этом сумма всех $\phi(i)$, где $i$ делит $p-1$, равна $p-1$, и всякое ненулевой элемент принадлежит хоть какому то порядку, откуда следует, что для всякого k количество чисел порядка k ровно $\phi(k)$, откуда есть элементы порядка $p-1$, что и требовалось доказать.\\ \\ \\ Пусть это $g$, и $g^{p-1} = 1 + pk$. Рассмотрим числа вида $(g + pt)^{p-1} \quad \forall t \in Z$. Тогда $(g + pt)^{p-1} = 1 + p \cdot (k + (p-1)g^{p-2} \cdot t + p \cdot X)$. Заметим, что существует такое $t_1$, что $k + (p-1)g^{p-2} \cdot t_1 + p \cdot X = 1$ ($\mod \quad \mathbb{Z}/p^{n-1} \mathbb{Z}$), так как существует обратное у $(p-1) \cdot g^{p-2}$ (назовем его $t_0$). Рассмотрим $t_1 = t_0 \cdot (1 - k - p \cdot X)$, заметим, что $1+p$ принадлежит показателю вида $p^\alpha$, так как $g + pt_1$ принадлежит показателю вида $p^\beta \cdot (p-1)$, так как все возможные непустые показатели являются делителями $p^{n-1} \cdot (p-1)$, при этом $(g + pt_1)^{p-1} \ne 1$. \\ \\ Рассмотрим $(1+p) ^{p^{\alpha}} = 1 + p^{\alpha + 1} \cdot (1 + p \cdot Y) = 1 + p^{\alpha + 1} \cdot u_{\alpha}$, где $u_{\alpha}$ взаимнопросто с $p$. Предположим, что $(1+p) ^{p^{\alpha}} = 1$, тогда $p^{\alpha + 1} = 0$, откуда $\alpha + 1 = n$, поэтому $1 + p$ принадлежит показателю $p^{n-1}$, следовательно $g + pt_1$ принадлежит показателю $p^{n-1} \cdot (p-1)$, и тогда $g + pt_1$ -- первообразный корень.\\ \\ \subsection{2 вариант} \textbf{Лемма 1}\\ Пусть $G$ -- конечная абелева группа и $a$ -- элемент наибольшего порядка в ней. Тогда порядок любого элемента группы $G$ делит порядок $a$.\\ \\ \textbf{Доказательство}\\ $x^{ord(x)} = 1$ Пусть $x$ -- произвольный элемент из $G$. Если $ord(x)$ не делит $ord(a)$, то существует такое простое $q^\alpha$ и показатель $\alpha \geq 1$, что $q$ делит $ord(x)$ и не делит $ord(a)$. Пусть $\beta \geq 0$ -- наибольшее число такое, что $q^\beta$ делит $ord(a)$, тогда $\alpha > \beta$.\\ Положим $y = x^{ord(x) / q^\alpha} = q$ и $b = a^{q^{\beta}}$. Тогда $ord(y) = q^\alpha$ и $ord(b) = ord(a) = q / q^\beta$. Так как $ord(y)$ и $ord(b)$ взаимно просты и группа $G$ абелева, то $ord(yb) = ord(y) \cdot ord(b) = ord(a)q^{\alpha - \beta} > ord(a)$ -- противоречие.\\ \\ По \textbf{Th Лагранжа}\\ порядок подгруппы конечной группы делит порядок этой группы.\\ \textbf{Следствие}\\ Порядок элемемнта конечной группы делит порядок этой группы.\\ \\ \textbf{Докажем Теорему}:\\ Мультипликативная группа конечного поля является циклической.\\ \\ \textbf{Доказательство}:\\ Пусть $\mathbb{K}$ -- конечное поле. Докажем, что $\mathbb{K} = \mathbb{K} \text{(без нуля)}$ -- циклическая. Пусть $x_1,\ x_2,\ ... \: ,\ x_n$ -- все элементы группы $\mathbb{K}^{\star}$ и пусть $x_1$ -- элемент наибольшего порядка $d$ в ней. По Лемме 1 порядки всех элементов $x_i$ делят $d$, в частности, все $x_i$ удовлетворяют уравнению $x^d - 1 = 0$ в $\mathbb{K}$. Знаем, что корней уравнения не больше $d$ $\Rightarrow$ $d \geq n$. Однако по следствию из Лагранжа $d \bigm| n \quad \Rightarrow \quad n = d$ значит $x_1$ порождает $\mathbb{K}^{\star}$.\\ \\ \textbf{Следствие}:\\ Если $p$ -- простое, то $Z_p$ -- циклическая группа порядка $p - 1$.\\ \\ \textbf{Предположение}:\\ Если $p > 2$ -- простое, то группа $Z^{\star}_{p^n}$ циклическая для всех $n > 0$.\\ \\ \textbf{Доказательство}:\\ По следствию существует существует такое натуральное число $g$, что $g^{p - 1} \equiv 1 \ (\mod p)$. Если $g^{p-1} \equiv 1 \ (\mod p^2)$, то $(g + p)^{p-1} = g^{p - 1} + (p - 1)g^{p - 2} + p^2(...) = 1 + (p - 1)g^{p - 2}p \equiv \overline{1} \ (\mod p^2)$. Поэтому можно считать что $g^{p - 1} \not\equiv 1 \ (\mod p^2)$ Таким образом $g^{p - 1} = 1 + p \cdot u$, где $u$ не делится на $p$.\\ \\ Докажем, что вычет $g$ порождает $Z^{\star}_{p^n}$\\ Докажем, что $g^{(p-1)p^k} = 1 + p^{k+1}u_k$ по индукции, база была доказана ранее.\\ Шаг: $g^{(p-1)p^{k+1}} = (1 + p^{k+1}u_k)^p = 1 + p \cdot p^{k+1}u_k + \sum^p_{i = 2} {p \choose i} (p^{k+1}u_k)^i$\\ $\sum^p_{i = 2} {p \choose i} (p^{k+1}u_k)^i$ делится на $p^{k+3}$, а $ p \cdot p^{k+1}u_k$ на $p^{k+2} \quad \Rightarrow \quad \frac{\sum^p_{i = 2} {p \choose i} (p^{k+1}u_k)^i}{p^{k+2}} + u_k$ не делится на $p$, так как $\frac{\sum^p_{i = 2} {p \choose i} (p^{k+1}u_k)^i}{p^{k+2}}$ делится на $p$, а $u_k$ нет, что и требовалось доказать\\ $\Rightarrow$ $g$ порождает $Z^{\star}_{p^n}$\\ \\ \textbf{Лемма 2}\\ $\forall n \in \mathbb{N} \ \exists ! v$ не делящееся на $p:\ (p + 1)^{p^{n - 1}} = v \cdot p^n + 1$\\ База:\\ $n = 1 \quad \Rightarrow \quad v = 1$\\ $(p + 1)^{p^n} = (v \cdot p^n + 1)^p = \sum^p_{i = 0} {p \choose i} (v^i p^{ni}) = 1 + pvp^n + \sum^p_{i = 2} {p \choose i} (v^i p^{ni}),\ pvp^n + \sum^p_{i = 2} {p \choose i} (v^i p^{ni})$ делится на $p^{n+1}$, что и требовалось доказать\\ \\ Докажем, что $ord(p + 1) \geq p^{n - 1}$. Пусть нет, тогда $ord(p + 1) < d = p^{n - 1}$ $d \bigm| p^{n - 1} \quad d = p^k,\ k < n - 1$, тогда $(p + 1)^{p^k} = v \cdot p^n + 1$, но по лемме 2, $\exists! u \: : \: (p + 1)^{p^k} = up^{k + 1} + 1 \quad \Rightarrow \quad k = n - 1$. Противоречие. $\Rightarrow \quad d \geq p^{n - 1}$\\ \\ Докажем, что $d = p^{n - 1}$\\ Пусть $(p + 1)^{p^{n-1}} \equiv 1 \quad (\mod p^n)$\\ Докажем $(p + 1)^{p^{n}} \equiv 1 \quad (\mod p^{n+1})$\\ $(p + 1)^{p \cdot p^{n - 1}} = ((p + 1)^{p^{n - 1}})^p = (1 + lp^n)^p = 1 + \sum^p_{i = 1} {p \choose i} (lp^n)^i \equiv 1 (\mod p^{n+1})$, что и требовалось доказать\\ 0 \begin{figure} \centering \begin{subfigure}{.4\textwidth} \centering \includegraphics[height=0.4\textheight]{figures/dit/all_pig} \caption{Including outliers}\label{fig:dit:all:pig_outlier} \end{subfigure}% \begin{subfigure}{.4\textwidth} \centering \includegraphics[height=0.4\textheight]{figures/dit/all_pig_no_outlier} \caption{Excluding outliers}\label{fig:dit:all:pig_no_outlier} \end{subfigure} \caption{Overview: Developer Identification effectiveness measures for Pig v0.14.0} \label{fig:dit:all:pig} \end{figure} %------------------------------------------------------------------------------------------------------------------------------------------ % Úvod %------------------------------------------------------------------------------------------------------------------------------------------ \chapter*{ÚVOD} \addcontentsline{toc}{chapter}{ÚVOD} \par V~aktuální době je velké množství dokumentů ve firmách zpracováváno nástroji, které do jisté míry nepodporují spolupráci a často tak dochází k~vytváření velkých a nepřehledných dokumentů, které se časem zvětšují a po několika letech se musí kompletně přepsat a případně zrušit. \par S~nástupem internetu tento problém částečně vymizel díky tomu, že se nyní dají takové dokumenty velice snadno sdílet, nicméně stále je zde problém, že velká část nástrojů pracujících s~dokumenty nenabízí jednoduché a přehledné provázání. Proto se v~rámci této práce zaměříme na možná řešení některých problémů, které vznikají při sdílení dokumentů a při jejich propojování. \par Cílem této práce je tedy analýza současných nástrojů a následně návrh a vytvoření specifického nástroje, který se bude zaměřovat z~velké části na jeho snadné používání uživateli. \par Diplomová práce nás postupně provede několika kapitolami, kdy se nejdříve budeme věnovat teoretickým východiskům, kde si vysvětlíme některé pojmy zabývající se prací s~dokumenty pomocí webových aplikací, poté se zaměříme na aktuální stav nástrojů a aplikací zabývajících se touto problematikou. Část popisu aktuálního stavu se budeme věnovat nástrojům, které nám mohou pomoci při samotném vývoji aplikace. Na samotný návrh a vývoj aplikace se zaměříme v~poslední kapitole, ve které si rozebereme také případná rizika vývoje takového nástroje. Od čtenáře se očekávají alespoň základní technické znalosti vývoje aplikací a prací s~webovými technologiemi. %------------------------------------------------------------------------------------------------------------------------------------------ % Cíl a metodika práce %------------------------------------------------------------------------------------------------------------------------------------------ \chapter*{CÍL DIPLOMOVÉ PRÁCE} Hlavním cílem této práce je vytvořit platformu, která bude schopná zpracovat data zadaná uživatelem, analyzovat je a na základě vnitřní logiky a informace nesené v~těchto datech je uložit do logických celků. Aplikace poté umožní tyto celky zobrazit uživateli, nechá jej dále definovat dodatečné informace, provádět reporting a případně zobrazit v~přehledných grafech. Platforma se bude zaměřovat převážně na uživatelské rozhraní, tak aby její používání bylo co nejintuitivnější a nejjednodušší. \addcontentsline{toc}{chapter}{CÍL DIPLOMOVÉ PRÁCE} \chapter*{METODIKA PRÁCE} \addcontentsline{toc}{chapter}{METODIKA PRÁCE} \section*{Metody} \par Nejdůležitějším zaměřením této platformy je uživatelská přívětivost a jednoduchost na používání, proto bude při vývoji kladen důraz na spokojenost uživatelů. Tohoto bude dosaženo použitím agilních metodik při vývoji, kdy bude postupně dodávaný produkt předáván úzkému kruhu uživatelů, kteří se budou vyjadřovat k~uživatelskému rozhraní. Platforma bude psána jako webová aplikace, která bude přistupovat do databáze přes rozhraní napsané v~jazyce Java. \par V~části zpracování dat bude použito několik ETL metodik a data mining technik, které povedou ke získání logických informací ze zadaných dat. Platforma bude vyvíjena s~možností škálovatelnosti a použití nad velkým objemem dat. \addcontentsline{toc}{section}{Metody} \section*{Postupy} \addcontentsline{toc}{section}{Postupy} \par K~vytvoření co nejpřívětivější platformy budou využity naše zkušenosti a knižní publikace zabývající se tímto tématem. Dále budou analyzovány jednotlivé postupy zadávání dat uživatelů do tohoto systému, které povedou ke zpřehlednění a zjednodušení používání. \par Pro komunikaci se serverem bude použit standard REST, který usnadní komunikaci se serverem a umožní případné navázání nových aplikací. V~případě, že bude vytvořena mobilní aplikace pro získávání dat, nebude nutné psát znovu stejnou nebo podobnou logiku. \par Pro zabezpečené přihlášení do aplikace bude použit autentizační server, který bude zajišťovat vytváření a správu uživatelů spolu s~jejich právy. estradacarles11/EightySecondsCV0 \makebacksidebar %------------------------------------------------------------------------------- % TABLE ENTRIES RIGHT COLUMN SECOND PAGE %------------------------------------------------------------------------------- \cvsection{section} \cvsubsection{Subsection} \begin{cvtable}[2] \cvitem{}{}{}{} \end{cvtable} \cvsection{cvitem} \cvsubsection{Multi-line with longer description} \begin{cvtable}[2] \cvitem{date}{Description}{location}{Some longer and more detailed description, that takes two lines of space instead of only one.} \cvitem{date}{Description}{location}{Some longer and more detailed description, that takes two lines of space instead of only one.} \cvitem{date}{Description}{location}{Some longer and more detailed description, that takes two lines of space instead of only one.} \end{cvtable} \cvsubsection{One-line without description} \begin{cvtable}[2] \cvitem{Award}{One-line description}{Sponsor}{} \cvitem{Award}{One-line description}{Sponsor}{} \cvitem{Award}{One-line description}{Sponsor}{} \end{cvtable} \cvsection{cvitemshort} \cvsubsection{One-line} \begin{cvtable}[2] \cvitemshort{Key}{Some further description} \cvitemshort{Key}{Some further description} \cvitemshort{Key}{Some further description} \end{cvtable} \cvsubsection{Multi-line with longer description} \begin{cvtable}[2] \cvitemshort{Key}{Some further description. Can fill even more than only one single line while still keeping the correct indendation level.} \cvitemshort{Key}{Some further description. Can fill even more than only one single line while still keeping the correct indendation level.} \cvitemshort{Key}{Some further description. Can fill even more than only one single line while still keeping the correct indendation level.} \end{cvtable} \cvsection{cvpubitem} \begin{cvtable}[2] \cvpubitem{Publication title}{Authors}{Journal}{Year} \cvpubitem{Publication title}{Authors}{Journal}{Year} \cvpubitem{Publication title that is spanning over multiple lines and still does not look too bad}{Authors}{Journal}{Year} \end{cvtable} \cvsignaturefidsusj/SmartWarehouse % ------------------------------------------------------------ % LaTeX Template für die DHBW zum Schnellstart! % Original: https://github.wdf.sap.corp/D064996/LaTeX-Template-DHBW % ------------------------------------------------------------ % ---- Präambel mit Angaben zum Dokument \input{Inhalt/00_Latex/praeambel} % ---- Elektronische Version oder Gedruckte Version? % ---- Unterschied: Die elektronische Version enthält keinen Platzhalter für die Unterschrift \usepackage{ifthen} \newboolean{e-Abgabe} \setboolean{e-Abgabe}{false} % false=gedruckte Fassung % ---- Persönlichen Daten: \newcommand{\titel}{Machbarkeitsstudie} \newcommand{\titelcaption}{- Smart Warehouse -} \newcommand{\subtitel}{Echtzeit-Objektdetektoren im Vergleich} \newcommand{\titelheader}{Smart Warehouse} \newcommand{\arbeit}{Studienarbeit} \newcommand{\studiengang}{Angewandte Informatik} \newcommand{\studienjahr}{2017} \newcommand{\autor}{ und } \newcommand{\autorReverse}{Hausberger, Felix und Kuck, Robin} \newcommand{\verfassungsort}{Karlsruhe} \newcommand{\matrikelnr}{2773463, 4409176} \newcommand{\kurs}{TINF17B2} \newcommand{\bearbeitungsmonat}{Oktober 2019 - Juni 2020} \newcommand{\abgabe}{01. Juni 2020} \newcommand{\bearbeitungszeitraum}{30.09.2019 - 01.06.2020} \newcommand{\firmaName}{SAP SE} \newcommand{\firmaStrasse}{Dietmar-Hopp-Allee 16} \newcommand{\firmaPlz}{69190 Walldorf, Deutschland} \newcommand{\betreuerFirma}{} \newcommand{\betreuerDhbw}{PD Dr.-Ing. } \input{Inhalt/00_Latex/kopfundFusszeile} % ---- Hilfreiches \newcommand{\zB}{z.\,B. } % "z.B." mit kleinem Leeraum dazwischen (ohne wäre nicht korrekt) \newcommand{\dash}{d.\,h. } % ---- Silbentrennung (falls LaTeX defaults falsch / nicht gewünscht sind) \hyphenation{HANA} % anstatt HA-NA \hyphenation{Graph-Script} % anstatt GraphS-cript % ---- Beginn des Dokuments \begin{document} \setlength{\parindent}{0pt} % Keine Paragraphen Einrückung. % Dafür haben wir den Abstand zwischen den Paragraphen. \setcounter{secnumdepth}{2} % Nummerierungstiefe fürs Inhaltsverzeichnis \setcounter{tocdepth}{1} % Tiefe des Inhaltsverzeichnisses. Ggf. so anpassen, % dass das Verzeichnis auf eine Seite passt. \sffamily % Serifenlose Schrift verwenden. % ---- Vorspann % ------ Titelseite \singlespacing \include{Inhalt/01_Standard/titelseite} % Titelseite \newcounter{savepage} \pagenumbering{arabic} % Römische Seitenzahlen \onehalfspacing % ------ Erklärung, Sperrvermerk, Abstact \include{Inhalt/01_Standard/erklaerung} \include{Inhalt/02_Abstract/abstract-en} \include{Inhalt/02_Abstract/abstract-de} % ------ Inhaltsverzeichnis \singlespacing \tableofcontents % ------ Verzeichnisse \renewcommand*{\chapterpagestyle}{plain} \pagestyle{plain} \setlength{\cftsecnumwidth}{3em} \setlength{\cfttabnumwidth}{3em} %\include{Inhalt/03_Verzeichnisse/formelgroessen} \include{Inhalt/03_Verzeichnisse/abkuerzungen} \include{Inhalt/03_Verzeichnisse/abbildungen} \include{Inhalt/03_Verzeichnisse/formeln} %\listoftables % Erzeugen des Tabellenverzeichnisses \renewcommand{\lstlistlistingname}{Listenverzeichnis} \lstlistoflistings % Erzeugen des Listenverzeichnisses % \setcounter{savepage}{\value{page}} % ---- Inhalt der Arbeit \cleardoublepage % \pagenumbering{arabic} % Arabische Seitenzahlen für den Hauptteil \setlength{\parskip}{0.5\baselineskip} % Abstand zwischen Absätzen \rmfamily \renewcommand*{\chapterpagestyle}{scrheadings} \pagestyle{scrheadings} \onehalfspacing \setcounter{chapter}{-1} \input{Inhalt/04_Vorwort/vorwort} \input{Inhalt/05_Einleitung/Einfuehrung/einfuehrung} \input{Inhalt/05_Einleitung/Problemstellung/problemstellung} \input{Inhalt/05_Einleitung/Zielsetzung/zielsetzung} \input{Inhalt/06_Forschungsstand/einfuehrung} \input{Inhalt/06_Forschungsstand/neuronale_netze} \input{Inhalt/06_Forschungsstand/hyperparameter} \input{Inhalt/06_Forschungsstand/datensatzlehre} \input{Inhalt/06_Forschungsstand/faltungsnetze} \input{Inhalt/06_Forschungsstand/map} \input{Inhalt/06_Forschungsstand/rcnn} \input{Inhalt/06_Forschungsstand/ssd} \input{Inhalt/06_Forschungsstand/yolo} \input{Inhalt/06_Forschungsstand/datensatzformate} \input{Inhalt/06_Forschungsstand/cloud} \input{Inhalt/06_Forschungsstand/drohnen} \input{Inhalt/07_Konzeption/trainingsdatensatz} \input{Inhalt/07_Konzeption/bewertungskriterien} \input{Inhalt/07_Konzeption/objektdetektoren} \input{Inhalt/07_Konzeption/trainingsinfrastruktur} \input{Inhalt/07_Konzeption/drohne} \input{Inhalt/07_Konzeption/software} \input{Inhalt/08_Realisierung/realisierung} \input{Inhalt/09_Ergebnisse/ergebnisse} \input{Inhalt/10_Bewertung/bewertung} \input{Inhalt/11_Zusammenfassung_Ausblick/zusammenfassung_ausblick} % ---- Literaturverzeichnis \cleardoublepage \renewcommand*{\chapterpagestyle}{plain} \pagestyle{plain} % \pagenumbering{Roman} % Römische Seitenzahlen % \setcounter{page}{\numexpr\value{savepage}+1} \printbibliography[title=Literaturverzeichnis] %\chapter*{Literaturverzeichnis} %\addcontentsline{toc}{chapter}{Literaturverzeichnis} %\printbibheading %\printbibliography[keyword=book,heading=subbibliography,title={Literaturquellen}] % Ausgabe der Literaturquellen %\printbibliography[keyword=online,heading=subbibliography,title={Internetquellen}] % Ausgabe der Internetquellen % ---- Anhang \onehalfspacing \appendix \chapter{Anhang} %\captionsetup{list=false} \input{Inhalt/12_Anhang/anhang} %\clearpage %\pagenumbering{Roman} % römische Seitenzahlen für Anhang \end{document} ManualUsuario/PRATSMOVILES/06_terminarPrestamo/encargado/03_terminarPrestamo.tex Para terminar un préstamo se selecciona cuál de los préstamos en curso mostrados se desea y se siguen los siguientes pasos. \subsection{Subpaso 1-A: Terminar préstamo} \begin{enumerate} \item Presione el botón \textbf{Terminar} en la tarjeta del préstamo correspondiente en la interfaz \textbf{IUGS-10 Bienvenida de encargado} \end{enumerate} \begin{figure}[hbtp] \centering \includegraphics[scale=0.3]{images/InterfazMovil/IUGS06_terminarPrestamo.png} \caption{Bienvenida para encargado} \end{figure}kobylky/savlozuby_zpevnik1 \beginsong{Napojen}[by={}] \beginverse \chordson \[E]Vzpomeň, jaks' očima dít\[A]ěte \[E]hleděl na letící p\[H]táky, \chordsoff jak krásné bylo být na Světě s vědomím, že umíš to taky. \chordson Dnes už \[F#]nelétáš, neboť oko\[A]lí zmátlo Tě \[F#]strachem, že pády \[A]bo\[H]lí. \endverse \beginverse \chordsoff Zastav se, čas nikam nespěchá, jen Rozum nutí ho k běhu. Věř však, Ty, synu Člověka, žes' nedílná součást příběhu. Ty jsi já, jako já Tebou jsem, společná je nám planeta Zem. \endverse \beginrefrain \chordson \[E]Ničeho nelituj, \[A]život, co máš, je tvůj, \[E]žij a měj rád, co \[H]je. \chordsoff Nechť srdce se rozbijí láskou, tak přijmi ji, cítíš, jak jsi napojen, jak jsi napojen, přes všechny pochyby, život je krásný, jsi napojen a Svět je Tvá továrna na sny. \endrefrain \beginverse \chordsoff Nebe až najde jas ve vodní hladině, až Měsíc opře se o břeh a padne Hvězda, zkus si přát jediné, to aby dopadla dobře. Ztiš se a poprvé k Sobě se obrať, bez zbraní a krve porazíš Obra. \endverse \beginverse \chordsoff Obra, co založil Duhy stín, když před léty zotročil Zem. A člověka, Bůh ví proč, znejistil v otázce, kdo vlastně jsem. Poznej, že být je jednoduché, ucítíš klid, že vše jedno je Duchem. \endverse \emptyrefrain \endsong % Reference Card for GNU Emacs % Copyright (C) 1987, 1993, 1996-1997, 2002-2004, 2006-2012 % Free Software Foundation, Inc. % Author: <> % Portuguese translation: <> % This file is part of GNU Emacs. % GNU Emacs is free software: you can redistribute it and/or modify % it under the terms of the GNU General Public License as published by % the Free Software Foundation, either version 3 of the License, or % (at your option) any later version. % GNU Emacs is distributed in the hope that it will be useful, % but WITHOUT ANY WARRANTY; without even the implied warranty of % MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the % GNU General Public License for more details. % You should have received a copy of the GNU General Public License % along with GNU Emacs. If not, see . % This file is intended to be processed by plain TeX (TeX82). % % The final reference card has six columns, three on each side. % This file can be used to produce it in any of three ways: % 1 column per page % produces six separate pages, each of which needs to be reduced to 80%. % This gives the best resolution. % 2 columns per page % produces three already-reduced pages. % You will still need to cut and paste. % 3 columns per page % produces two pages which must be printed sideways to make a % ready-to-use 8.5 x 11 inch reference card. % For this you need a dvi device driver that can print sideways. % Which mode to use is controlled by setting \columnsperpage. % % To compile and print this document: % tex refcard.tex % dvips -t landscape refcard.dvi % Thanks to for the opinions. %**start of header \newcount\columnsperpage \newcount\letterpaper % This file can be printed with 1, 2, or 3 columns per page. % Specify how many you want here. \columnsperpage=3 % Set letterpaper to 0 for A4 paper, 1 for letter (US) paper. Useful % only when columnsperpage is 2 or 3. \letterpaper=0 % PDF output layout. 0 for A4, 1 for letter (US), a `l' is added for % a landscape layout. \input pdflayout.sty \pdflayout=(0l) \def\versionemacs{24} % version of Emacs this is for \def\year{2012} % latest copyright year % Nothing else needs to be changed below this line. \def\shortcopyrightnotice{\vskip 1ex plus 2 fill \centerline{\small \copyright\ \year\ Free Software Foundation, Inc. Permissions on back.}} \def\copyrightnotice{ \vskip 1ex plus 2 fill\begingroup\small \centerline{Copyright \copyright\ \year\ Free Software Foundation, Inc.} \centerline{For GNU Emacs version \versionemacs} \centerline{Designed by } \centerline{Translated by } Permission is granted to make and distribute copies of this card provided the copyright notice and this permission notice are preserved on all copies. For copies of the GNU Emacs manual, see: {\tt http://www.gnu.org/software/emacs/\#Manuals} \endgroup} % make \bye not \outer so that the \def\bye in the \else clause below % can be scanned without complaint. \def\bye{\par\vfill\supereject\end} \newdimen\intercolumnskip %horizontal space between columns \newbox\columna %boxes to hold columns already built \newbox\columnb \def\ncolumns{\the\columnsperpage} \message{[\ncolumns\space column\if 1\ncolumns\else s\fi\space per page]} \def\scaledmag#1{ scaled \magstep #1} % This multi-way format was designed by October 1986. % Note that the 1-column format is fontfamily-independent. \if 1\ncolumns %one-column format uses normal size \hsize 4in \vsize 10in \voffset -.7in \font\titlefont=\fontname\tenbf \scaledmag3 \font\headingfont=\fontname\tenbf \scaledmag2 \font\smallfont=\fontname\sevenrm \font\smallsy=\fontname\sevensy \footline{\hss\folio} \def\makefootline{\baselineskip10pt\hsize6.5in\line{\the\footline}} \else %2 or 3 columns uses prereduced size \hsize 3.2in \if 1\the\letterpaper \vsize 7.95in \else \vsize 7.65in \fi \hoffset -.75in \voffset -.745in \font\titlefont=cmbx10 \scaledmag2 \font\headingfont=cmbx10 \scaledmag1 \font\smallfont=cmr6 \font\smallsy=cmsy6 \font\eightrm=cmr8 \font\eightbf=cmbx8 \font\eightit=cmti8 \font\eighttt=cmtt8 \font\eightmi=cmmi8 \font\eightsy=cmsy8 \textfont0=\eightrm \textfont1=\eightmi \textfont2=\eightsy \def\rm{\eightrm} \def\bf{\eightbf} \def\it{\eightit} \def\tt{\eighttt} \if 1\the\letterpaper \normalbaselineskip=.8\normalbaselineskip \else \normalbaselineskip=.7\normalbaselineskip \fi \normallineskip=.8\normallineskip \normallineskiplimit=.8\normallineskiplimit \normalbaselines\rm %make definitions take effect \if 2\ncolumns \let\maxcolumn=b \footline{\hss\rm\folio\hss} \def\makefootline{\vskip 2in \hsize=6.86in\line{\the\footline}} \else \if 3\ncolumns \let\maxcolumn=c \nopagenumbers \else \errhelp{You must set \columnsperpage equal to 1, 2, or 3.} \errmessage{Illegal number of columns per page} \fi\fi %% \intercolumnskip=.46in \intercolumnskip=.65in \def\abc{a} \output={% %see The TeXbook page 257 % This next line is useful when designing the layout. %\immediate\write16{Column \folio\abc\space starts with \firstmark} \if \maxcolumn\abc \multicolumnformat \global\def\abc{a} \else\if a\abc \global\setbox\columna\columnbox \global\def\abc{b} %% in case we never use \columnb (two-column mode) \global\setbox\columnb\hbox to -\intercolumnskip{} \else \global\setbox\columnb\columnbox \global\def\abc{c}\fi\fi} \def\multicolumnformat{\shipout\vbox{\makeheadline \hbox{\box\columna\hskip\intercolumnskip \box\columnb\hskip\intercolumnskip\columnbox} \makefootline}\advancepageno} \def\columnbox{\leftline{\pagebody}} \def\bye{\par\vfill\supereject \if a\abc \else\null\vfill\eject\fi \if a\abc \else\null\vfill\eject\fi \end} \fi % we won't be using math mode much, so redefine some of the characters % we might want to talk about \catcode`\^=12 \catcode`\_=12 \chardef\\=`\\ \chardef\{=`\{ \chardef\}=`\} \hyphenation{mini-buf-fer} \parindent 0pt \parskip 1ex plus .5ex minus .5ex \def\small{\smallfont\textfont2=\smallsy\baselineskip=.8\baselineskip} % newcolumn - force a new column. Use sparingly, probably only for % the first column of a page, which should have a title anyway. \outer\def\newcolumn{\vfill\eject} % title - page title. Argument is title text. \outer\def\title#1{{\titlefont\centerline{#1}}\vskip 1ex plus .5ex} % section - new major section. Argument is section name. \outer\def\section#1{\par\filbreak \vskip 3ex plus 2ex minus 2ex {\headingfont #1}\mark{#1}% \vskip 2ex plus 1ex minus 1.5ex} \newdimen\keyindent % beginindentedkeys...endindentedkeys - key definitions will be % indented, but running text, typically used as headings to group % definitions, will not. \def\beginindentedkeys{\keyindent=1em} \def\endindentedkeys{\keyindent=0em} \endindentedkeys % paralign - begin paragraph containing an alignment. % If an \halign is entered while in vertical mode, a parskip is never % inserted. Using \paralign instead of \halign solves this problem. \def\paralign{\vskip\parskip\halign} % \<...> - surrounds a variable name in a code example \def\<#1>{{\it #1\/}} % kbd - argument is characters typed literally. Like the Texinfo command. \def\kbd#1{{\tt#1}\null} %\null so not an abbrev even if period follows % beginexample...endexample - surrounds literal text, such a code example. % typeset in a typewriter font with line breaks preserved \def\beginexample{\par\leavevmode\begingroup \obeylines\obeyspaces\parskip0pt\tt} {\obeyspaces\global\let =\ } \def\endexample{\endgroup} % key - definition of a key. % \key{description of key}{key-name} % prints the description left-justified, and the key-name in a \kbd % form near the right margin. \def\key#1#2{\leavevmode\hbox to \hsize{\vtop {\hsize=.75\hsize\rightskip=1em \hskip\keyindent\relax#1}\kbd{#2}\hfil}} \newbox\metaxbox \setbox\metaxbox\hbox{\kbd{M-x }} \newdimen\metaxwidth \metaxwidth=\wd\metaxbox % metax - definition of a M-x command. % \metax{description of command}{M-x command-name} % Tries to justify the beginning of the command name at the same place % as \key starts the key name. (The "M-x " sticks out to the left.) \def\metax#1#2{\leavevmode\hbox to \hsize{\hbox to .75\hsize {\hskip\keyindent\relax#1\hfil}% \hskip -\metaxwidth minus 1fil \kbd{#2}\hfil}} % threecol - like "key" but with two key names. % for example, one for doing the action backward, and one for forward. \def\threecol#1#2#3{\hskip\keyindent\relax#1\hfil&\kbd{#2}\hfil\quad &\kbd{#3}\hfil\quad\cr} %**end of header \title{GNU Emacs: Cart\~ao de Refer\^encia} \centerline{(para vers\~ao \versionemacs)} \section{Iniciando o Emacs} Para entrar no GNU Emacs, digite: \kbd{emacs} \section{Saindo do Emacs} \key{suspende ou minimiza o Emacs}{C-z} \key{encerra o Emacs}{C-x C-c} \section{Arquivos} \key{{\bf abre} um arquivo}{C-x C-f} \key{{\bf salva} um arquivo em disco}{C-x C-s} \key{salva {\bf todos} arquivos abertos}{C-x s} \key{{\bf insere} outro arquivo neste buffer}{C-x i} \key{substitui este arquivo por outro}{C-x C-v} \key{salva o buffer em um arquivo especificado}{C-x C-w} \key{alterna o estado de somente leitura do buffer}{C-x C-q} \section{Ajuda (Help)} Tecle \kbd{C-h} (ou \kbd{F1}) e siga as instru{\c{c}}{\~o}es. \key{remove a janela de ajuda}{C-x 1} \key{rola a janela de ajuda}{C-M-v} \key{apropos: mostra comandos que casam com a string}{C-h a} \key{descreve fun{\c{c}}{\~a}o associada a teclas}{C-h k} \key{descreve uma fun{\c{c}}{\~a}o}{C-h f} \key{busca informa{\c{c}}{\~o}es espec{\'\i}ficas do modo}{C-h m} \section{Recuperando-se de Erros} \key{{\bf aborta} uma opera{\c{c}}{\~a}o}{C-g} \metax{{\bf recupera} arquivos ap{\'o}s crash}{M-x recover-session} \metax{desfaz uma altera{\c{c}}{\~a}o ({\bf undo})}{C-x u, C-_ {\rm or} C-/} \metax{restaura um buffer para o arquivo}{M-x revert-buffer} \key{redesenha a tela}{C-l} \section{Busca Incremental} \key{busca para frente}{C-s} \key{busca para tr{\'a}s}{C-r} \key{busca por express{\~a}o regular}{C-M-s} \key{busca por express{\~a}o regular para tr{\'a}s}{C-M-r} \key{seleciona a string de pesquisa anterior}{M-p} \key{seleciona a string seguinte de pesquisa}{M-n} \key{sai da busca incremental}{RET} \key{desfaz o efeito do {\'u}ltimo caracter}{DEL} \key{encerra a busca}{C-g} Use \kbd{C-s} ou \kbd{C-r} novamente para repetir a busca. \kbd{C-g} cancela apenas o que ainda n{\~a}o foi feito. \shortcopyrightnotice \section{Movimenta{\c{c}}{\~a}o} \paralign to \hsize{#\tabskip=10pt plus 1 fil&#\tabskip=0pt&#\cr \threecol{{\bf avan{\c{c}}ar}}{{\bf tr{\'a}s}}{{\bf frente}} \threecol{um caracter}{C-b}{C-f} \threecol{uma palavra}{M-b}{M-f} \threecol{uma linha}{C-p}{C-n} \threecol{para in{\'\i}cio ou fim de linha}{C-a}{C-e} \threecol{senten{\c{c}}a}{M-a}{M-e} \threecol{par{\'a}grafo}{M-\{}{M-\}} \threecol{p{\'a}gina}{C-x [}{C-x ]} \threecol{sexp}{C-M-b}{C-M-f} \threecol{fun{\c{c}}{\~a}o}{C-M-a}{C-M-e} \threecol{para in{\'\i}cio ou fim do buffer}{M-<}{M->} } \key{rolar para pr{\'o}xima tela}{C-v} \key{rolar para tela anterior}{M-v} \key{rolar para esquerda}{C-x <} \key{rolar para direita}{C-x >} \key{rolar a linha corrente para o centro da tela}{C-u C-l} \section{Cortando e Apagando} \paralign to \hsize{#\tabskip=10pt plus 1 fil&#\tabskip=0pt&#\cr \threecol{{\bf entidade a cortar}}{{\bf tr{\'a}s}}{{\bf frente}} \threecol{caracter (apaga, n{\~a}o corta)}{DEL}{C-d} \threecol{palavra}{M-DEL}{M-d} \threecol{linha (at{\'e} o final)}{M-0 C-k}{C-k} \threecol{senten{\c{c}}a}{C-x DEL}{M-k} \threecol{sexp}{M-- C-M-k}{C-M-k} } \key{corta {\bf regi{\~a}o}}{C-w} \key{copia a {\bf regi{\~a}o}}{M-w} \key{cortar at{\'e} a pr{\'o}xima ocorr{\^e}ncia de {\it char}}{M-z {\it char}} \key{colar a {\'u}ltima coisa cortada}{C-y} \key{substitui a {\'u}lt. colagem pela c{\'o}pia anterior}{M-y} \section{Marcando} \key{posiciona a marca aqui}{C-@ {\rm or} C-SPC} \key{troca a marca pelo ponto e vice-versa}{C-x C-x} \key{coloca a marca {\it arg\/} {\bf palavras} adiante}{M-@} \key{marca o {\bf par{\'a}grafo}}{M-h} \key{marca a {\bf p{\'a}gina}}{C-x C-p} \key{marca a {\bf sexp}}{C-M-@} \key{marca uma {\bf fun{\c{c}}{\~a}o}}{C-M-h} \key{marca todo {\bf buffer}}{C-x h} \section{Busca e Substitui{\c{c}}{\~a}o} \key{Substitui interativamente uma string}{M-\%} % query-replace-regexp is bound to C-M-% but that can't be typed on % consoles. \metax{usando express{\~a}o regular}{M-x query-replace-regexp} Respostas v{\'a}lidas no modo de busca e substitui{\c{c}}{\~a}o \key{{\bf substitui} esta, e prossegue}{SPC} \key{substitui esta e entrada e n{\~a}o avan{\c{c}}a}{,} \key{{\bf pula} para a pr{\'o}xima sem substituir}{DEL} \key{substitui em todo o texto restante}{!} \key{{\bf volta} para a palavra anterior}{^} \key{{\bf encerra}}{RET} \key{entra na edi{\c{c}}{\~a}o recursiva (\kbd{C-M-c} para sair)}{C-r} \section{M{\'u}ltiplas Janelas} Quando forem mostrados 2 comandos, o segundo tem comportamento similar para frame. {\setbox0=\hbox{\kbd{0}}\advance\hsize by 0\wd0 \paralign to \hsize{#\tabskip=10pt plus 1 fil&#\tabskip=0pt&#\cr \threecol{elimina todas outras janelas}{C-x 1\ \ \ \ }{C-x 5 1} \threecol{divide a janela, acima e abaixo}{C-x 2\ \ \ \ }{C-x 5 2} \threecol{elimina esta janela}{C-x 0\ \ \ \ }{C-x 5 0} }} \key{divide a janela, lado a lado}{C-x 3} \key{rola a outra janela}{C-M-v} {\setbox0=\hbox{\kbd{0}}\advance\hsize by 2\wd0 \paralign to \hsize{#\tabskip=10pt plus 1 fil&#\tabskip=0pt&#\cr \threecol{leva o cursor para outra janela}{C-x o}{C-x 5 o} \threecol{seleciona um buffer em outra janela}{C-x 4 b}{C-x 5 b} \threecol{mostra um buffer em outra janela}{C-x 4 C-o}{C-x 5 C-o} \threecol{busca um arquivo em outra janela}{C-x 4 f}{C-x 5 f} \threecol{busca arquivo (ro) em outra janela}{C-x 4 r}{C-x 5 r} \threecol{executa Dired em outra janela}{C-x 4 d}{C-x 5 d} \threecol{busca tag em outra janela}{C-x 4 .}{C-x 5 .} }} \key{aumenta a janela na vertical}{C-x ^} \key{estreita a janela}{C-x \{} \key{alarga a janela}{C-x \}} \section{Formatando} \key{identa a {\bf linha} corrente (modo)}{TAB} \key{identa a {\bf regi{\~a}o} (modo)}{C-M-\\} \key{identa a {\bf sexp} (modo)}{C-M-q} \key{identa regi{\~a}o rigidamente {\it arg\/} colunas}{C-x TAB} \key{insere uma nova linha ap{\'o}s o ponto}{C-o} \key{move o restante da linha para baixo}{C-M-o} \key{apaga linhas em branco em torno do ponto}{C-x C-o} \key{junta a linha com a anterior}{M-^} \key{apaga todos brancos em torno do ponto}{M-\\} \key{insere um espa{\c{c}}o em branco}{M-SPC} \key{preenche o par{\'a}grafo}{M-q} \key{define a coluna limite de preenchimento}{C-x f} \key{define um prefixo para cada linha}{C-x .} \key{formata fonte}{M-o} \section{Mai{\'u}sculas e Min{\'u}sculas} \key{Palavra para mai{\'u}sculas}{M-u} \key{Palavra para min{\'u}sculas}{M-l} \key{Primeira letra mai{\'u}scula (capitalize)}{M-c} \key{Regi{\~a}o para mai{\'u}sculas}{C-x C-u} \key{Regi{\~a}o para min{\'u}sculas}{C-x C-l} \section{O Minibuffer} As teclas seguintes s{\~a}o definidas no minibuffer. \key{complete o m{\'a}ximo possi{\'\i}vel}{TAB} \key{complete at{\'e} uma palavra}{SPC} \key{complete e execute}{RET} \key{mostre as op{\c{c}}{\~o}es para completar}{?} \key{busca a entrada anterior no minibuffer}{M-p} \key{busca a pr{\'o}xima entrada no minibuffer ou o default}{M-n} \key{busca regexp no hist{\'o}rico para tr{\'a}s}{M-r} \key{busca regexp no hist{\'o}rico para frente}{M-s} \key{encerra o comando}{C-g} Tecle \kbd{C-x ESC ESC} para editar e repetir o {\'u}ltimo comando utilizado. Tecle \kbd{F10} para ativar o menu. \newcolumn \title{GNU Emacs: Cart\~ao de Refer\^encia} \centerline{(para vers\~ao \versionemacs)} \section{Buffers} \key{seleciona outro buffer}{C-x b} \key{lista todos buffers}{C-x C-b} \key{mata um buffer}{C-x k} \section{Transposi{\c{c}}{\~a}o} \key{transp{\~o}e {\bf caracteres}}{C-t} \key{transp{\~o}e {\bf palavras}}{M-t} \key{transp{\~o}e {\bf linhas}}{C-x C-t} \key{transp{\~o}e {\bf sexps}}{C-M-t} \section{Verifica{\c{c}}{\~a}o Ortogr{\'a}fica} \key{verifica a palavra corrente}{M-\$} \metax{verifica todas palavras de uma regi{\~a}o}{M-x ispell-region} \metax{verifica todo o buffer}{M-x ispell-buffer} \section{Tags} \key{busca uma tag (uma defini{\c{c}}{\~a}o)}{M-.} \key{encontra a pr{\'o}xima ocorr{\^e}ncia da tag}{C-u M-.} \metax{especifica um novo arquivo de tags}{M-x visit-tags-table} \metax{busca por regexp em todos arquivos}{M-x tags-search} \metax{busca e subst. em todos arquivos}{M-x tags-query-replace} \key{continua a {\'u}ltima busca ou busca e substitui{\c{c}}{\~a}o}{M-,} \section{Shells} \key{executa um comando do shell}{M-!} \key{executa um comando do shell na regi{\~a}o}{M-|} \key{filtra uma regi{\~a}o por um comando do shell}{C-u M-|} \key{inicia um shell na janela \kbd{*shell*}}{M-x shell} \section{Ret{\^a}ngulos} \key{copia o ret{\^a}ngulo para o registrador}{C-x r r} \key{corta o ret{\^a}ngulo}{C-x r k} \key{cola o ret{\^a}ngulo}{C-x r y} \key{abre o ret{\^a}ngulo, move o texto para direita}{C-x r o} \key{troca por espa{\c{c}}os o conte{\'u}do do ret{\^a}ngulo}{C-x r c} \key{antep{\~o}e uma linha a string}{C-x r t} \section{Abreviaturas} \key{adiciona uma abreviatura global}{C-x a g} \key{adiciona abreviatura ao modo local}{C-x a l} \key{adiciona globalmente expans{\~a}o de abrev.}{C-x a i g} \key{adiciona localmente expans{\~a}o de abrev.}{C-x a i l} \key{explicitamente expande uma abrev}{C-x a e} \key{completa com base em palavras anteriores}{M-/} \section{Express{\~o}es Regulares} \key{qualquer caracter exceto nova linha}{. {\rm(dot)}} \key{zero ou mais repeti{\c{c}}{\~o}es}{*} \key{uma ou mais repeti{\c{c}}{\~o}es}{+} \key{zero ou uma repeti{\c{c}}{\~a}o}{?} \key{protege o caracter especial {\it c\/}}{\\{\it c}} \key{(``or'')}{\\|} \key{agrupamento}{\\( {\rm$\ldots$} \\)} \key{mesmo texto que {\it n\/}-{\'e}simo grupo}{\\{\it n}} \key{quebra de palavra}{\\b} \key{sem quebra de palavra}{\\B} \paralign to \hsize{#\tabskip=10pt plus 1 fil&#\tabskip=0pt&#\cr \threecol{{\bf entidade}}{{\bf casa in{\'\i}cio}}{{\bf casa fim}} \threecol{linha}{^}{\$} \threecol{palavra}{\\<}{\\>} \threecol{buffer}{\\`}{\\'} \threecol{{\bf classe de caracteres}}{{\bf casa esses}}{{\bf casa os outros}} \threecol{conjunto expl{\'\i}cito}{[ {\rm$\ldots$} ]}{[^ {\rm$\ldots$} ]} \threecol{caracter de sintaxe de palavra}{\\w}{\\W} \threecol{caracter de sintaxe de {\it c}}{\\s{\it c}}{\\S{\it c}} } \section{Conjuntos de Carac. Internacionais} \key{especifica uma l{\'\i}ngua principal}{C-x RET l} \metax{mostra todos m{\'e}todos de inser{\c{c}}{\~a}o}{M-x list-input-methods} \key{habilita/desabilita um m{\'e}todo de inser{\c{c}}{\~a}o}{C-\\} \key{determina o sistema de codifica{\c{c}}{\~a}o}{C-x RET c} \metax{mostra sistemas de codifica{\c{c}}{\~a}o}{M-x list-coding-systems} \metax{escolhe a codifica{\c{c}}{\~a}o preferida}{M-x prefer-coding-system} \section{Info} \key{entra no leitor de Info}{C-h i} \key{busca fun{\c{c}}{\~a}o ou arquivo no Info}{C-h S} \beginindentedkeys Movimenta{\c{c}}{\~a}o em um nodo: \key{rola para frente}{SPC} \key{rola para tr{\'a}s}{DEL} \key{in{\'\i}cio do nodo}{. {\rm (dot)}} Movimenta{\c{c}}{\~a}o entre nodos: \key{{\bf pr{\'o}ximo} nodo}{n} \key{nodo {\bf anterior}}{p} \key{mover cima {\bf cima}}{u} \key{seleciona item do menu pelo nome}{m} \key{seleciona {\it n\/}-{\'e}simo item do menu}{{\it n}} \key{segue refer{\^e}ncia cruzada (retorna com \kbd{l})}{f} \key{retorna {\'u}ltimo nodo visitado}{l} \key{retorna ao diret{\'o}rio de nodos}{d} \key{ir para o topo do arquivo Info}{t} \key{ir para qualquer nodo por nome}{g} Outros: \key{executar {\bf tutorial} do Info}{h} \key{busca pelo assunto no {\'\i}ndice}{i} \key{busca por express{\~a}o regular}{s} \key{{\bf sair} Info}{q} \endindentedkeys \section{Registrador} \key{salva regi{\~a}o em um registrador}{C-x r s} \key{insere o conte{\'u}do do registrador no buffer}{C-x r i} \key{salva valor do ponto no registrador}{C-x r SPC} \key{salta para o ponto salvo no registrador}{C-x r j} \section{Macros de Teclado} \key{{\bf inicia} a defini{\c{c}}{\~a}o de uma macro}{C-x (} \key{{\bf encerra} a defini{\c{c}}{\~a}o de uma macro}{C-x )} \key{{\bf executa} a {\'u}ltima macro definida}{C-x e} \key{adiciona a {\'u}ltima macro definida}{C-u C-x (} \metax{nomeia a {\'u}ltima macro definida}{M-x name-last-kbd-macro} \metax{insere uma defini{\c{c}}{\~a}o em Lisp}{M-x insert-kbd-macro} \section{Lidando com Emacs Lisp} \key{avalia {\bf sexp} antes do ponto}{C-x C-e} \key{avalia a {\bf defun} corrente}{C-M-x} \metax{avalia a {\bf regi{\~a}o}}{M-x eval-region} \key{l{\^e} e avalia o minibuffer}{M-:} \metax{carrega do diret{\'o}rio padr{\~a}o do sistema}{M-x load-library} \section{Personaliza{\c{c}}{\~a}o Simples} \metax{personaliza vari{\'a}veis e fontes}{M-x customize} % The intended audience here is the person who wants to make simple % customizations and knows Lisp syntax. Fazendo teclas de atalho globais em Emacs Lisp (exemplos): \beginexample% (global-set-key "\\C-cg" 'goto-line) (global-set-key "\\M-\#" 'query-replace-regexp) \endexample \section{Escrevendo Comandos} \beginexample% (defun \ (\) "\" (interactive "\