summaryrefslogtreecommitdiff
path: root/graphics/ellipse
diff options
context:
space:
mode:
authorNorbert Preining <norbert@preining.info>2019-09-02 13:46:59 +0900
committerNorbert Preining <norbert@preining.info>2019-09-02 13:46:59 +0900
commite0c6872cf40896c7be36b11dcc744620f10adf1d (patch)
tree60335e10d2f4354b0674ec22d7b53f0f8abee672 /graphics/ellipse
Initial commit
Diffstat (limited to 'graphics/ellipse')
-rw-r--r--graphics/ellipse/README39
-rw-r--r--graphics/ellipse/ellipse.dtx1238
-rw-r--r--graphics/ellipse/ellipse.ins71
-rw-r--r--graphics/ellipse/ellipse.pdfbin0 -> 418514 bytes
4 files changed, 1348 insertions, 0 deletions
diff --git a/graphics/ellipse/README b/graphics/ellipse/README
new file mode 100644
index 0000000000..5eab7c4f9a
--- /dev/null
+++ b/graphics/ellipse/README
@@ -0,0 +1,39 @@
+This is the README file for the ellipse package.
+
+VERSION
+
+Version 1.0, December 2015
+
+SUMMARY
+
+Adds the ability to draw ellipses and elliptical arcs on top of the standard LaTeX2e picture environment.
+
+LaTeX has many advanced graphics packages now, the most extensive are tikz and
+pstricks. However, these are also large packages that take long to load and
+may not always work on all drivers. The standard pict2e package removes many
+of the previous limitations of the `old' LaTeX |picture| environment and makes
+it a _lean and portable_ alternative to the more full featured packages.
+However, even though it can draw circles and circle arcs well, it lacks the
+ability to draw ellipses, ovals, and elliptical arcs. This package adds these
+functions on top of the standard pict2e primitives (i.e. the \cbezier command).
+
+LICENSE
+
+Copyright (c) 2015 Daan Leijen.
+
+This software is author-maintained. Permission is granted to copy,
+distribute and/or modify this software under the terms of the
+LaTeX Project Public License, version 1.3 or higher. This software
+is provided 'as it is', without warranty of any kind, either
+expressed or implied, including, but not limited to, the implied
+warranties of merchantability and fitness for any particular purpose.
+
+RELATED PACKAGES
+
+pict2e.sty
+
+AUTHOR
+
+Daan Leijen (daan@microsoft.com)
+
+
diff --git a/graphics/ellipse/ellipse.dtx b/graphics/ellipse/ellipse.dtx
new file mode 100644
index 0000000000..c22bfe2a58
--- /dev/null
+++ b/graphics/ellipse/ellipse.dtx
@@ -0,0 +1,1238 @@
+% \iffalse meta-comment
+%
+% Copyright (C) 2015 Daan Leijen
+% -------------------------------------------------------
+%
+% This file may be distributed and/or modified under the
+% conditions of the LaTeX Project Public License, either version 1.2
+% 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 1.2 or later is part of all distributions of LaTeX
+% version 1999/12/01 or later.
+%
+% \fi
+%
+% \iffalse
+%<*driver>
+\ProvidesFile{ellipse.dtx}
+%</driver>
+%<package>\NeedsTeXFormat{LaTeX2e}[1999/12/01]
+%<package>\ProvidesPackage{ellipse}
+%<*package>
+ [2004/11/05 v1.0 .dtx ellipse file]
+%</package>
+%<package>\RequirePackage{pict2e}
+%
+%<*driver>
+\documentclass{ltxdoc}
+\usepackage{graphicx}
+\usepackage{ellipse}[2004/11/05]
+\usepackage{amsmath}
+\usepackage{amssymb}
+\usepackage{hyperref}
+\usepackage{lmodern}
+\usepackage{xcolor}
+\usepackage{wrapfig}
+\newcommand\kk[1]{\iota_{#1}}
+\newcommand\sk[1]{\rho_{#1}}
+\newcommand\sint[1]{\textit{sint}_{#1}}
+\newcommand\cost[1]{\textit{cost}_{#1}}
+\newcommand\sign[1]{\pm_{#1}}
+\newcommand\csqrt{\textit{csqrt}}
+\renewcommand\cos{\textit{cos}}\renewcommand\sin{\textit{sin}}
+\renewcommand\arctan{\textit{arctan}}\renewcommand\tan{\textit{tan}}
+\renewcommand\max{\textit{max}}\renewcommand\sin{\textit{sin}}
+\providecolor{teal}{HTML}{008080}
+\providecolor{purple}{HTML}{800080}
+\providecolor{navy}{HTML}{000080}
+\providecolor{maroon}{HTML}{800000}
+\providecolor{floralwhite}{HTML}{FFFAF0}
+\providecolor{ivory}{HTML}{FFFFF0}
+\providecolor{white}{HTML}{FFFFFF}
+\providecolor{transparent}{named}{white}
+\providecolor{gainsboro}{HTML}{DCDCDC}
+\hypersetup{colorlinks=true,citecolor=navy,linkcolor=navy,urlcolor=navy,filecolor=navy,bookmarksdepth=3,bookmarksopenlevel=1}
+\makeatletter
+\DeclareRobustCommand*\package[2][]{%
+ \def\@tempa{#1}%
+ \ifx\@tempa\@empty
+ \textsf{#2}%
+ \else
+ \href{http://mirrors.ctan.org/macros/latex/#1}{\textsf{#2}}%
+ \fi
+}%
+\DeclareRobustCommand*\pkgpicte{\package[contrib/pict2e/pict2e.pdf]{pict2e}}
+\newsavebox{\@ebox}
+\newcommand*\@unit[1]{\strip@pt\dimexpr#1\relax}%
+\newcommand*\ellipbox[1]{%
+ \begingroup
+ \savebox{\@ebox}{#1}%
+ \setlength{\unitlength}{1pt}%
+ \hspace*{0.8ex}%
+ \begin{picture}(0,0)%
+ \put(\@unit{0.5\wd\@ebox},\@unit{0.5\ht\@ebox - 0.5\dp\@ebox}){%
+ \ellipse{\@unit{0.8ex + 0.5\wd\@ebox}}{\@unit{0.8ex + 0.5\ht\@ebox}}%
+ }%
+ \end{picture}%
+ \usebox{\@ebox}%
+ \hspace{0.8ex}%
+ \endgroup%
+}
+\makeatother
+\EnableCrossrefs
+\CodelineIndex
+\RecordChanges
+\begin{document}
+ \setlength\emergencystretch{3em}
+ \DocInput{ellipse.dtx}
+ \PrintChanges
+ \setcounter{IndexColumns}{2}%
+ \PrintIndex
+\end{document}
+%</driver>
+% \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 \~}
+%
+%
+% \changes{v1.0}{2015/03/24}{Initial version}
+%
+% \GetFileInfo{ellipse.dtx}
+%
+% \DoNotIndex{\newcommand,\renewcommand,\newenvironment,\renewenvironment}
+% \DoNotIndex{\providecommand,\def,\edef,\let,\gdef,\xdef,\global,\newtoks}
+% \DoNotIndex{\RequirePackage,\DeclareOption,\ProcessOptions,\ExecuteOptions}
+% \DoNotIndex{\input,\InputIfFileExists}
+% \DoNotIndex{\@ifdefinable,\@ifundefined,\@percentchar}
+% \DoNotIndex{\AtBeginDocument,\AtEndOfPackage}
+% \DoNotIndex{\PackageError,\PackageWarning,\PackageWarningNoLine,\PackageInfo}
+% \DoNotIndex{\MessageBreak,\typeout}
+%
+% \DoNotIndex{\z@,\z@skip,\p@,\@ne,\tw@,\thr@@,\@iv,\two@fourteen,\strip@pt}
+% \DoNotIndex{\the,\if,\else,\or,\fi,\ifnum,\ifdim,\ifcase,\ifodd}
+% \DoNotIndex{\advance,\multiply,\divide}
+% \DoNotIndex{\@tfor,\do}
+% \DoNotIndex{\csname,\endcsname,\begingroup,\endgroup}
+% \DoNotIndex{\expandafter,\afterassignment,\noexpand}
+% \DoNotIndex{\@ovxx,\@ovyy,\@ovdx,\@ovdy}
+% \DoNotIndex{\undefined,\dimexpr,\relax,\space,\protect,\begin}
+% \DoNotIndex{\@tempdima,\@tempdimb,\@tempdimc,\@tempdimd,\dimen@,\@tempa}
+% \DoNotIndex{\@tempswafalse,\@tempswatrue,\if@tempswa,\iffalse,\ifx,\ignorespaces}
+%
+% \title{The {\linethickness{0.6pt}\ellipbox{\textsf{ellipse}}} package}
+% \author{Daan Leijen \\ \texttt{daan@microsoft.com}}
+%
+% \maketitle
+%
+% \newcommand\pictexample[1]{%
+% \setlength{\unitlength}{10pt}%
+% \raisebox{-30pt}{\begin{picture}(6,8)%
+% \linethickness{0.4pt}\roundjoin%
+% \color{gainsboro}%
+% \put(0,0){\multiput(0,0)(1,0){9}{\line(0,1){6}}%
+% \multiput(0,0)(0,1){7}{\line(1,0){8}}%
+% }%
+% \put(0,0){\color{gray}%
+% {\vector(1,0){8}}\put(-0.5,6.5){$y$}%
+% {\vector(0,1){6}}\put(8.5,-0.5){$x$}%
+% }%
+% \color{black}%
+% \linethickness{0.8pt}%
+% #1\end{picture}}%
+% }
+%
+% \section{Introduction}
+%
+% \LaTeX{} has many advanced graphics packages now, the most extensive are
+% \href{https://www.ctan.org/pkg/pgf}{\textsf{tikz}}
+% and \href{https://www.ctan.org/pkg/pstricks-base}{\textsf{pstricks}}.
+% However, these are also large packages that take long to load and
+% may not always work on all drivers. The standard \pkgpicte{} package removes many of the previous
+% limitations of the `old' \LaTeX{} |picture| environment and makes it a \emph{lean
+% and portable} alternative to the more full featured packages. However, even though it can
+% draw circles and circle arcs well, it lacks the ability to draw ellipses and elliptical
+% arcs. This package adds these functions on top of the standard \pkgpicte{} primitives
+% (i.e. the |\cbezier| command).
+%
+% \section{Drawing ellipses}
+%
+% \noindent
+% \DescribeMacro\ellipse
+% \DescribeMacro{\ellipse*}
+% \marg{x-radius}\marg{y-radius}\\
+% \strut\\
+% These commands draw an ellipse with the specified radi\"\i. The |\ellipse| command draws
+% a stroked ellipse with the current |\linethickness| while |\ellipse*| draws a filled
+% ellipse with the current |\color|. For example:
+%
+% \noindent\begin{minipage}{0.7\linewidth}%
+% \begin{verbatim}
+% \setlength{\unitlength}{10pt}%
+% \begin{picture}(6,8)
+% \linethickness{0.8pt}%
+% \put(6,3){\color{teal}\ellipse*{2}{3}}%
+% \put(3,3){\color{blue}\ellipse{3}{2}}%
+% \end{picture}
+% \end{verbatim}\end{minipage}
+% \pictexample{%
+% \put(6,3){\color{teal}\ellipse*{2}{3}}%
+% \put(3,3){\color{blue}\ellipse{3}{2}}%
+% }\\[2ex]
+%
+% \noindent
+% \DescribeMacro\earc
+% \DescribeMacro{\earc*}
+% \oarg{start-angle$\rangle$|,|$\langle$end-angle}\marg{x-radius}\marg{y-radius}\\
+% \strut\\
+% These commands draw part of an ellipse with the specified radi\"\i.
+% The |\earc| command draws
+% a stroked elliptical arc with the current |\linethickness| while |\earc*| draws a filled
+% elliptical `pie slice' with the current |\color|. The optional argument specifies a start and
+% end-angle in degrees which must be between $-720$ and $720$ (but can be fractional).
+% The endings of the arcs are determined by the \emph{cap} setting: |\buttcap| (default),
+% |\roundcap| (add half disc), or |\squarecap| (add half square).
+%
+% \noindent\begin{minipage}{0.7\linewidth}%
+% \begin{verbatim}
+% \put(3,3){%
+% \color{blue}\roundcap\earc[135,330]{3}{2}}%
+% \put(6,3){%
+% \color{teal}\earc*[-45,90]{2}{3}}%
+% \end{verbatim}
+% \end{minipage}
+% \pictexample{%
+% \put(3,3){\color{blue}\roundcap\earc[135,330]{3}{2}}%
+% \put(6,3){\color{teal}\earc*[-45,90]{2}{3}}%
+% }\\[2ex]
+%
+% \noindent
+% \DescribeMacro\elliparc
+% \oarg{initial}\marg{center-x}\marg{center-y}\marg{x-rad}\marg{y-rad}\marg{start-angle}\marg{end-angle}\\
+%
+% \noindent The core elliptical arc routine. These are to be used with path commands, like
+% |\lineto|, |\moveto|, |\strokepath|, etc, and can draw an elliptical arc at any center point.
+% The optional argument specifies the initial drawing
+% action: the default is $0$ (|\lineto|) which draws a line to the arc starting point,
+% the value $1$
+% (|\moveto|) just moves to the starting point, and $2$ does nothing as an initial action.
+% If the start angle is larger than the end angle, the arc is drawn clockwise, and otherwise
+% anti-clockwise.\\[1ex]
+%
+% \noindent\begin{minipage}{0.7\linewidth}%
+% \begin{verbatim}
+% \elliparc[1]{3}{3}{3}{2}{90}{270}%
+% \elliparc{5}{3}{2}{2}{-90}{90}%
+% \closepath\strokepath
+% \color{teal}%
+% \moveto(1,3)
+% \elliparc{3}{3}{2}{1}{-135}{135}%
+% \closepath
+% \fillpath
+% \end{verbatim}\end{minipage}
+% \pictexample{%
+% \elliparc[1]{3}{3}{3}{2}{90}{270}%
+% \elliparc{5}{3}{2}{2}{-90}{90}%
+% \closepath\strokepath
+% \color{teal}%
+% \moveto(1,3)
+% \elliparc{3}{3}{2}{1}{-135}{135}%
+% \closepath
+% \fillpath
+% }\\[2ex]
+%
+% \noindent Note how the two initial arcs are automatically connected by a line
+% segment from $(3,1)$ to $(5,1)$ (due to the default optional argument of $0$ that
+% uses a |\lineto| command to the starting point of the arc). Similarly, we use
+% such initial line segment and a |\closepath| to draw the triangular side of the
+% inner ellipse.
+%
+% \subsection{Rotated ellipses}
+%
+% There is no direct command to rotate an ellipse but you can use the
+% standard |\rotatebox| command from the \package[required/graphics/grfguide.pdf]{graphicx} package. For example:
+%
+% \noindent\begin{minipage}{0.7\linewidth}%
+% \begin{verbatim}
+% \put(3,3){%
+% \rotatebox[origin=c]{45}{\ellipse{3}{2}}%
+% }%
+% \end{verbatim}\end{minipage}
+% \pictexample{%
+% \put(3,3){\rotatebox[origin=c]{45}{\ellipse{3}{2}}%
+% }%
+% }\\[2ex]
+%
+% \subsection{Using the picture environment inline}
+%
+% The standard \LaTeX{} |picture| environment is nowadays quite
+% powerful and convenient. Read the latest \pkgpicte{} documentation
+% and ``The unknown \emph{picture} environment'' \cite{picture} for more information.
+% One particularly nice feature is that we
+% can create a picture as |\begin{picture}(0,0)| to give it zero
+% space. This can be used for example to define an |\ellipbox| command
+% like:\\[1ex]
+%
+% \noindent\begin{minipage}{0.6\linewidth}%
+% \begin{verbatim}
+% Boxed numbers:
+% \ellipbox{1}, \ellipbox{123}.
+% \end{verbatim}
+% \end{minipage}
+% \begin{minipage}{0.38\linewidth}%
+% Boxed numbers: \ellipbox{1}, \ellipbox{123}.
+% \end{minipage}\\[1ex]
+%
+% \noindent We also used this command to draw the ellipse in the title
+% of this article, and it is defined as:
+% \begin{verbatim}
+%\newsavebox{\@ebox}
+%\newcommand*\@unit[1]{\strip@pt\dimexpr#1\relax}%
+%\newcommand*\ellipbox[1]{%
+% \begingroup
+% \savebox{\@ebox}{#1}%
+% \setlength{\unitlength}{1pt}%
+% \hspace*{0.8ex}%
+% \begin{picture}(0,0)%
+% \put(\@unit{0.5\wd\@ebox},\@unit{0.5\ht\@ebox - 0.5\dp\@ebox}){%
+% \ellipse{\@unit{0.8ex + 0.5\wd\@ebox}}{\@unit{0.8ex + 0.5\ht\@ebox}}%
+% }%
+% \end{picture}%
+% \usebox{\@ebox}\hspace{0.25ex}\endgroup}
+% \end{verbatim}
+% \noindent This is not the best code possible but it hopefully gives you a good
+% idea on how to implement your own boxes. Note the use of the |\@unit| macro
+% to convert dimensions to units, which is also why we need to set the |\unitlength|
+% to |1pt| here.
+%
+% \StopEventually{}
+%
+% \begin{thebibliography}{9}
+% \raggedright
+%
+% \bibitem{abst} M.~Abramowitz and I.A.~Stegun: \textit{Handbook of Mathematical Functions},
+% \url{people.math.sfu.ca/~cbm/aands}, 1964
+%
+% \bibitem{picture} Claudio Beccari: \emph{The unknown \emph{picture} environment},
+% TUGBoat, vol. 33(1), 2012. \url{tug.org/TUGboat/tb33-1/tb103becc-picture.pdf}
+%
+% \bibitem{ellipse} Luc Maisonobe: \textit{Drawing an elliptical arc using polylines, quadratic or cubic B\'ezier lines.}
+% \url{www.spaceroots.org/documents/ellipse/elliptical-arc.pdf}, 2003
+%
+% \bibitem{rajan:atan} S.~Rajan, Sichun Wang, R.~Inkol, and A.~Joyal: \textit{Efficient approximations for the arctangent function}.
+% In Signal Processing Magazine, vol. 23(3), pages 108--111, May 2006
+%
+% \end{thebibliography}
+%
+% \clearpage
+%
+% \newcommand\abs[1]{\lvert #1\rvert}
+% \newcommand\rarg[1]{$\langle$\textit{#1}$\rangle$}
+% \newcommand\xellipse{\mathcal{E}}
+%
+% \section{Elliptical arcs as B\'ezier curves}
+% \begin{figure}\begin{center}
+% \setlength\unitlength{18pt}
+% \begin{picture}(12,9)(-1.5,-0.5)%
+% \linethickness{0.4pt}\roundjoin%
+% \iffalse
+% \color{gainsboro}%
+% \put(0,0){%
+% \multiput(-2,-1)(1,0){13}{\line(0,1){9}}%
+% \multiput(-2,-1)(0,1){10}{\line(1,0){12}}%
+% }%
+% \fi
+% \color{black}%
+% \put(0,0){%
+% \put(0,-0.5){\vector(0,1){8}}\put(8.5,-0.5){$x$}%
+% \put(-1.5,0){\vector(1,0){11}}\put(-0.5,6.5){$y$}%
+% }%
+% \put(4,3.3){\color{navy}\earc{5}{3}
+% \put(3,-2){$\xellipse$}%%
+% \color{teal}%
+% \put(-5.5,0){\line(1,0){11}}%
+% \put(0,-3.5){\line(0,1){8}}%
+% \put(3,0){\vector(-1,0){3}\vector(1,0){2}\raisebox{0.3ex}{$a$}}%
+% \put(0,-1.5){\vector(0,1){1.5}\vector(0,-1){1.5}$\,b$}%
+% \color{maroon}%
+% \put(0,0){\moveto(0,0)\elliparc{0}{0}{5}{3}{30}{120}\closepath\strokepath}%
+% {\linethickness{2pt}\earc[30,120]{5}{3}}%
+% \put(3.7,2.2){$p_1$}\put(-2,3.1){$p_2$}%
+% \put(2.25,2.92){\moveto(0,0)\lineto(1.35,-0.84)\strokepath\circle*{0.1}$\;q_1$}%
+% \put(-0.075,3.22){\moveto(0,0)\lineto(-1.56,-0.39)\strokepath\circle*{0.1}$\,q_2$}%
+% \color{black}
+% \linethickness{0.2pt}\arc[0,30]{1}\arc[0,120]{2}%
+% \put(1,0.2){$\alpha_1$}\put(1,1.8){$\alpha_2$}%
+% }%
+% \color{navy}\put(4,-0.3){$\,c_x$}\put(0,2.9){$\,c_y$}%
+% %
+% \end{picture}\end{center}
+% \caption{Approximating an elliptical arc with a cubic B\'ezier curve. The center of the ellipse
+% is at $(c_x,c_y)$ with a horizontal radius of $a$ and a vertical one $b$. The elliptical arc goes
+% from $\alpha_1$ to $\alpha_2$ and is approximated with a thick red cubic B\'ezier curve. The curve
+% starts at $p_1$ and ends in $p_2$ with two control points $q_1$ and $q_2$. The curve was drawn
+% using the command \texttt{\textbackslash{}elliparc\{4\}\{3.3\}\{5\}\{3\}\{30\}\{120\}}.}
+% \label{fig:ellipse}
+% \end{figure}
+% \begin{figure}
+% \setlength\unitlength{20pt}\begin{center}
+% \begin{picture}(12,8)(-1.5,-0.5)%
+% \linethickness{0.4pt}\roundjoin%
+% \iffalse
+% \put(0,0){\color{gainsboro}%
+% \multiput(-2,-1)(1,0){13}{\line(0,1){8}}%
+% \multiput(-2,-1)(0,1){9}{\line(1,0){12}}%
+% }%
+% \fi
+% \put(4,3){\color{navy}\earc{5}{3}\put(-4,1){$\xellipse$}\color{teal}%
+% \put(-5.5,0){\line(1,0){11}}%
+% \put(0,-3.5){\line(0,1){7}}%
+% \put(-2,0){\vector(-1,0){3}\vector(1,0){2}\raisebox{0.3ex}{$a$}}%
+% \put(0,-1.5){\vector(0,1){1.5}\vector(0,-1){1.5}$\,b$}%
+% \color{black}%
+% \put(0,0){\circle{6}}%
+% \put(0,0){\vector(3.6,2.08){3.6}\vector(2.16,2.08){2.16}}%
+% \put(0,0){\color{maroon}%
+% \put(3.6,0){\line(0,1){2.08}\circle*{0.1}}%
+% \put(3.2,-0.5){$a\cos(t_1)$}%
+% \put(2.16,0){\line(0,1){2.08}\circle*{0.1}}%
+% \put(1.3,-0.5){$\cos(t_1)$}%
+% }%
+% \linethickness{0.2pt}\arc[0,30]{1}\arc[0,43.9]{2}%
+% \put(1,0.2){$\alpha_1$}\put(1.65,1.25){$t_1$}%
+% }%
+% %
+% \end{picture}\end{center}
+% \caption{The relation between the parametric angle $t_1$ and the angle $\alpha_1$ to the point on the ellipse.
+% All points on the ellipse are defined by the parametric equation $\xellipse(t) = (c_x + a\cdot\cos(t), c_y + b\cdot\sin(t))$}
+% \label{fig:parametric}
+% \end{figure}
+% %
+% %
+% Drawing an ellipse or part of an ellipse (\emph{elliptical arc}) using B\'ezier
+% curves requires some math to determine the right control points of the B\'ezier curve.
+% Figure~\ref{fig:ellipse} establishes some notation. We do not consider rotated
+% ellipses here and always use $a$ for the $x$-radius and $b$ for the $y$-radius.
+% We are interested in finding the B\'ezier curve between the $\alpha_1$ and
+% $\alpha_2$ angles, which implies finding the starting
+% point $p_1$, the end point $p_2$ and the control points $q_1$ and $q_2$.
+%
+% Each point on an ellipse is determined by the following parametric equation:
+% \[ \xellipse(t) = (c_x + a\cdot\cos(t), c_y + b\cdot\sin(t)) \]
+% where $t$ is the
+% parametric angle. The parametric angle $t$ is just a property of the ellipse and has no
+% `real' counterpart. Figure~\ref{fig:parametric} gives some helpful intuition how
+% the $\alpha$ angles and $t$ angles are related: we can imagine drawing a unit circle inside
+% an ellipse where for every $t$ angle on the unit circle we have a corresponding point
+% and angle $\alpha$ on the ellipse. From the definition of $\xellipse$ it
+% is straightforward to derive a parametric angle $t_i$ for some $\alpha_i$:
+% \[ t_i = \arctan_2(\frac{\sin(\alpha_i)}{b},\frac{\cos(\alpha_i)}{a}) \]
+%
+% \noindent Given this relation, the start and end points of our curve are simply:
+% \begin{align*}
+% p_1 &= \xellipse(t_1)\\
+% p_2 &= \xellipse(t_2)
+% \end{align*}
+%
+% \noindent To be able to calculate optimal control points $q$ we need to also determine the
+% tangent of each point on the ellipse, which is given by the derivative of $\xellipse$:
+% \[ \xellipse'(t) = (-a\cdot\sin(t), b\cdot\cos(t)) \]
+%
+% \noindent
+% The derivation of the optimal B\'ezier control points for an ellipse is quite involved,
+% see~\cite{ellipse} for a nice overview.
+% For a quadratic B\'ezier curve, it turns out the optimal control points are determined as:
+% \begin{align*}
+% q_1 &= p_1 + \tan(\frac{t_2 - t_1}{2})\cdot\xellipse'(t_1)\\
+% &= p_2 - \tan(\frac{t_2 - t_1}{2})\cdot\xellipse'(t_2)
+% \end{align*}
+% %
+% \noindent while for a cubic B\'ezier curve, one solution for optimal control points is:
+% %
+% \begin{align*}
+% q_1 &= p_1 + \kappa\cdot\xellipse'(t_1)\\
+% q_2 &= p_2 - \kappa\cdot\xellipse'(t_2)\\
+% \kappa &= \sin(t_2 - t_1)\frac{\sqrt{4 + 3\tan^2(\frac{t_2 - t_1}{2})}-1}{3}
+% \end{align*}
+% \noindent We will use cubic bezier curves since they look best. However, a na\"\i{}ve implementation
+% may be too expensive in \LaTeX: if we count the expensive operations, we need about 11 $\cos$/$\sin$
+% operations, plus a $\sqrt{}$ and 2 $\arctan$ operations.
+% %
+% \subsection{Optimizing elliptic arc equations}
+% Fortunately, we can improve upon this. First we note:
+% \begin{flalign*}
+% t_i &= \arctan_2(\frac{\sin(\alpha_i)}{b},\frac{\cos(\alpha_i)}{a}) & \hfill\\
+% &= \arctan(\frac{a}{b}\tan(\alpha_i)) \\
+% &= \arctan(\kk{i}) & \mbox{(introducing $\kk{i}$ for $\frac{a}{b}\tan(\alpha_i)$)}
+% \end{flalign*}
+% \noindent where we write $\kk{i}$ for $\frac{a}{b}\tan(\alpha_i)$. Now,
+% \begin{flalign*}
+% \cost{i} &= \cos(t_i)\\
+% &= \cos(\arctan(\kk{i})) & \mbox{(geometry and pythagorean theorem)}\\
+% &= \sign{i}\frac{1}{\sqrt{1 + \kk{i}^2}} \\
+% \mbox{with}\\
+% \sign{i} &= \textsf{if}\; \cos(\alpha_i) < 0\; \textsf{then}\; -\; \textsf{else}\; +
+% \end{flalign*}
+% Later we will see how we can efficiently calculate the square root term, but first
+% do the same derivation for the $\sin$ function:
+% \begin{flalign*}
+% \sint{i}&= \sin(t_i) \\
+% &= \sin(\arctan(\frac{a}{b}\tan(\alpha_i))) & \\
+% &= \sin(\arctan(\kk{i})) \\
+% &= \sign{i}\frac{\kk{i}}{\sqrt{1 + \kk{i}^2}}
+% \end{flalign*}
+% \noindent Note that the interaction between the $\sin$ and $\kk{i}$ term (whose sign is determined by $\tan(\alpha_i)$) allows us to reuse the sign function used for $\cost{i}$.
+% %
+% Using the previous equalities we can restate the parametric equations in terms of $\sint{i}$
+% and $\cost{i}$:
+% \begin{flalign*}
+% \xellipse_i &= (c_x + a\cdot\cost{i}), c_y + b\cdot\sint{i}) \\
+% \xellipse'_i &= (-a\cdot\sint{i}, b\cdot\cost{i})
+% \end{flalign*}
+% %
+% \noindent This takes care of $p_1$ and $p_2$. The control points $q$ still need $\sin(t_2 - t_1)$ and $\tan(\frac{t_2 - t_1}{2})$.
+% The halving rule on $\tan$ gives us:
+% \[
+% \tan(\frac{t_2 - t_1}{2}) = \frac{1 - \cos(t_2 - t_1)}{\sin(t_2 - t_1)} \mbox{\quad(\cite[page 71, 4.3.20]{abst})}
+% \]
+% \noindent So that leaves $\sin(t_2 - t_1)$ and $\cos(t_2 - t_1)$. Using the addition laws it follows:
+% \begin{flalign*}
+% \sin(t_2 - t_1) = \sint2\cost1 - \cost2\sint1 & \mbox{\quad(\cite[page 72, 4.3.16]{abst})}\\
+% \cos(t_2 - t_1) = \cost2\cost1 + \sint2\sint1 & \mbox{\quad(\cite[page 72, 4.3.17]{abst})}
+% \end{flalign*}
+%
+% \subsection{Circular square roots}
+% \noindent Now, we only need two $\tan$ operations to calculate the initial $\kk{1}$ and $\kk{2}$ terms but we still have
+% three square roots: $\sqrt{1 + \kk{i}^2}$ and $\sqrt{4 + 3\tan^2(\frac{t_2 - t_1}{2})}$.
+% Fortunately, both have the form $\sqrt{x^2 + y^2}$. For this form, we can make a very good
+% \begin{wrapfigure}[10]{o}{0.3\textwidth}
+% \setlength\unitlength{3pt}
+% \begin{picture}(30,30)(-15,-15)%
+% \linethickness{0.4pt}\roundjoin%
+% \iffalse
+% \put(0,0){\color{gainsboro}%
+% \multiput(-15,-15)(1,0){31}{\line(0,1){30}}%
+% \multiput(-15,-15)(0,1){31}{\line(1,0){30}}%
+% }%
+% \fi
+% \color{gainsboro}%
+% \put(0,0){\put(0,-15){\line(0,1){30}}%
+% \put(-15,0){\line(1,0){30}}%
+% }%
+% \color{black}%
+% \put(0,0){\circle{20}\put(-9,3){$\,x^2 + y^2$}%
+% \color{maroon}\moveto(14.14,0)\lineto(0,14.14)\lineto(-14.14,0)\lineto(0,-14.14)
+% \closepath\strokepath
+% \put(3,13){$\frac{1}{\sqrt{2}}(x+y)$}%
+% \color{navy}\moveto(10,10)\lineto(-10,10)\lineto(-10,-10)\lineto(10,-10)
+% \closepath\strokepath
+% \put(6,-12){$\,\max(x,y)$}%
+% }
+% %
+% \end{picture}\iffalse\caption{Estimating an initial value for $\sqrt{x^2+y^2}$}\fi
+% \end{wrapfigure}
+% initial guess for the square root, since this is the parametric equation for a circle.
+% The two good initial guesses form a `square' and `diamond' around this circle, namely
+% $\max(\abs{x},\abs{y})$ and $\frac{1}{\sqrt{2}}\abs{x+y}$. Each one can be superior depending if $x$ and $y$
+% are close or not, but it can be shown that the best choice is always the largest of these.
+% Using this guess as an initial seed, we can do a standard Newton-Raphson iteration to
+% find a the square root where we only need 2 or 3 steps to achieve the desired precision.
+% Let's define a `circular square root' function $\csqrt$ such that
+% $\csqrt(x,y) \approx \sqrt{x^2 + y^2}$ as:
+% \begin{align*}
+% \csqrt(x,y) = &\textsf{let} & sqr & = x^2 + y^2 & \hfill \\
+% & & x_0 & = \max(\abs{x},\abs{y},\frac{1}{\sqrt{2}}\abs{x+y}) & \\
+% & & x_1 & = (x_0 + \frac{sqr}{x_0})/2 & \hspace{\textwidth}\\
+% & & x_2 & = (x_1 + \frac{sqr}{x_1})/2 &\\
+% & \textsf{in} & x_2 &
+% \end{align*}
+%
+% \subsection{The optimized elliptical B\'ezier equations}
+% \noindent Taking it all together, we get the following equations for a cubic B\'ezier curve approximation of an elliptical arc,
+% where we assume as input the center point $(c_x,c_y)$, the $x$- and $y$-radius $(a,b)$, and a start and end angle
+% $\alpha_1$ and $\alpha_2$. % It is assumed that $\alpha_1 \ne \alpha_2$ and $a \ge 0, b \ge 0$. Of course,
+% with bezier curves one should build a full ellipse of parts where for each part $\rvert\alpha_1 - \alpha_2\lvert \le 90$.
+% Given these parameters, the start and end point $p_1$ and $p_2$, and the control points $q_1$ and $q_2$ are
+% defined as:
+% \begin{flalign*}
+% p_1 &= \xellipse_1 & \\
+% p_2 &= \xellipse_2\\
+% q_1 &= p_1 + \kappa\cdot\xellipse'_1\\
+% q_2 &= p_2 - \kappa\cdot\xellipse'_2\\
+% \xellipse_i &= (c_x + a\cdot\cost{i}, c_y + b\cdot\sint{i})\\
+% \xellipse'_i &= (-a\cdot\sint{i}, b\cdot\cost{i})
+% \end{flalign*}
+% The $\cost{i}$ and $\sint{i}$ are calculated as:
+% \begin{flalign*}
+% \sint{i} & = \sign{i}\frac{\kk{i}}{\sk{i}} & \\
+% \cost{i} & = \sign{i}\frac{1}{\sk{i}}
+% \end{flalign*}
+% with
+% \begin{flalign*}
+% \kk{i} &= \frac{a}{b}\tan(\alpha_i) & \\
+% \sk{i} &= \csqrt(1,\kk{i}) \;\;(\approx \sqrt{1 + \kk{i}^2}) \\
+% \sign{i} &= \textsf{if}\; \cos(\alpha_i) < 0\; \textsf{then}\; -\; \textsf{else}\; +
+% \end{flalign*}
+% And finally, the $\kappa$ term can be defined as:
+% \begin{flalign*}
+% \kappa &= \sint{21}\frac{\kappa_{sqrt}-1}{3} &
+% \end{flalign*}
+% with
+% \begin{flalign*}
+% \sint{21} & = \sint2\cost1 - \cost2\sint1 \;\;(=\sin(t_2 - t_1)) & \\
+% \cost{21} & = \cost2\cost1 + \sint2\sint1 \;\;(=\cos(t_2 - t_1))\\
+% \kappa_{tan} &= \frac{1 - \cost{21}}{\sint{21}} \;\;\mbox{(note: divides by zero if $\alpha_1 = \alpha_2$)}\\
+% \kappa_{sqrt} &= \csqrt(\sqrt{4},\sqrt{3}\cdot\kappa_{tan}) \;\;(\approx \sqrt{4 + 3\kappa_{tan}^2})
+% \end{flalign*}
+%
+%\section{Implementation}
+%
+% Generally, we use e-\TeX{} division to divide dimensions, where we
+% divide \rarg{dim$_1$} by \rarg{dim$_2$} using:
+% |\dimexpr 1pt * |\rarg{dim$_1$}|/|\rarg{dim$_2$}|\relax|
+% since it keeps a 64-bit intermediate result for such `scaling' expressions.
+% Note that both \rarg{dim}
+% expressions occur in an integer context and \TeX{} will convert
+% them to numbers automatically (i.e. in |sp| units).
+%
+
+% \subsection{Generic math and trigonometry routines}
+%
+% \begin{macro}{\pIIe@csedef}
+% \marg{csname}\textit{pattern}\marg{body}\\
+% Define a macro by a \textit{csname}. Just like the |\csedef| function
+% from \textsf{etoolbox} package
+% \begin{macrocode}
+\providecommand*\pIIe@csedef[1]{\expandafter\edef\csname #1\endcsname}
+% \end{macrocode}
+% \end{macro}
+%
+%\begin{macro}{\pIIe@ellip@csqrt}
+% \marg{dimen$_x$}\marg{dimen$_y$}\marg{dimreg$_{res}$}\\
+% Calculates $res \approx \sqrt{x^2 + y^2}$ and caches previous results
+% for efficiency.
+%
+% Overwrites |\@ovxx|,|\@ovyy|,|\@ovdx|,|\@ovdy|,|\@tempa|, and |\dimen@|.
+% \begin{macrocode}
+\newcommand*\pIIe@ellip@csqrt[3]{%
+ \@ovxx=#1\relax
+ \ifdim\@ovxx<\z@\@ovxx-\@ovxx\fi
+ \@ovyy=#2\relax
+ \ifdim\@ovyy<\z@\@ovyy-\@ovyy\fi
+ \edef\pIIe@csname{@csqrt(\number\@ovxx,\number\@ovyy)}%
+ \expandafter\ifx\csname\pIIe@csname\endcsname\relax
+ \pIIe@ellip@csqrt@%
+ \pIIe@csedef{\pIIe@csname}{\the\dimen@}%
+ #3\dimen@
+ \else
+ #3\dimexpr\csname\pIIe@csname\endcsname\relax
+ \fi
+}
+% \end{macrocode}
+%\end{macro}
+%\begin{macro}{\pIIe@ellip@csqrt@}
+% Internal routine: calculates |\dimen@| $\approx \sqrt{x^2 + y^2}$.
+% where $x \ge 0$ and $y \ge 0$, and |\@ovxx| $=x$ and |\@ovyy| $=y$.
+%
+% Overwrites |\@ovdx|,|\@ovdy|, and |\@tempa|.
+% \begin{macrocode}
+\newcommand*\pIIe@ellip@csqrt@{%
+% \end{macrocode}
+% First determine $\max(x,y,\frac{1}{\sqrt{2}}(x+y))$ in |\dimen@|.
+% Put the sum $x+y$ in |\@ovdx|.
+% \begin{macrocode}
+ \@ovdx\@ovxx
+ \advance\@ovdx by \@ovyy
+% \end{macrocode}
+% Put initial guess in |\dimen@| $=\max(\abs{x},\abs{y},\frac{1}{\sqrt{2}}(x+y))$.
+% \begin{macrocode}
+ \dimen@0.7071067\@ovdx
+ \ifdim\dimen@<\@ovyy\dimen@\@ovyy\fi
+ \ifdim\dimen@<\@ovxx\dimen@\@ovxx\fi
+% \end{macrocode}
+% To prevent overflowing \TeX{} dimensions we only do
+% a further Newton-Raphson approximation if the sum $x+y$ is less than 128pt.
+% Otherwise, for our application, the initial guess is still very precise since $x \ll y$ in that case.
+% \begin{macrocode}
+ \ifdim\@ovdx<128\p@
+% \end{macrocode}
+% Set |\@ovxx| to $x^2 + y^2$
+% \begin{macrocode}
+ \edef\@tempa{\strip@pt\@ovxx}%
+ \@ovxx\@tempa\@ovxx
+ \edef\@tempa{\strip@pt\@ovyy}%
+ \@ovyy\@tempa\@ovyy
+ \advance\@ovxx by \@ovyy
+% \end{macrocode}
+% Do two steps of Newton-Raphson (should we do three?)
+% \begin{macrocode}
+ \advance\dimen@ by \dimexpr1pt * \@ovxx/\dimen@\relax
+ \divide\dimen@ by 2%
+ \advance\dimen@ by \dimexpr1pt * \@ovxx/\dimen@\relax
+ \divide\dimen@ by 2%
+ \fi
+% \end{macrocode}
+% Result is |\dimen@|.
+% \begin{macrocode}
+}
+% \end{macrocode}
+% \end{macro}
+%
+
+
+% \begin{macro}{\pIIe@atan@}
+%
+% Approximate the $\arctan$ using
+% \[x\cdot\frac{\pi}{4} - x \cdot (\abs{x} - 1) \cdot (0.2447 + 0.0663\cdot\abs{x})\]
+% This approximation was described by Rajan et al.~\cite{rajan:atan}.
+%
+% The \cmd{pIIe@atan@} computes the arctan of |\dimen@| which must be between $-1$ and $1$, and stores it in |\dimen@| again.
+% Overwrites |\@tempdim|(|a|,|b|,|c|,|d|),|\@tempa|, and |\dimen@|.
+% \begin{macrocode}
+\newcommand*\pIIe@atan@{%
+% \end{macrocode}
+% |\dimen@| contains $x$.
+% \begin{macrocode}
+ \@tempdima\dimen@
+% \end{macrocode}
+% Set |\@dimtmpb| to $\abs{x}$
+% \begin{macrocode}
+ \@tempdimb\@tempdima
+ \ifdim\@tempdimb<\z@\@tempdimb-\@tempdimb\fi
+ \dimen@0.0663\@tempdimb
+ \advance\dimen@ 0.2447pt\relax
+ \advance\@tempdimb -1pt\relax
+ \edef\@tempa{\strip@pt\@tempdimb}%
+ \dimen@\@tempa\dimen@
+ \edef\@tempa{\strip@pt\@tempdima}%
+ \dimen@\@tempa\dimen@
+ \dimen@-\dimen@
+% \end{macrocode}
+% Add $x\cdot\frac{\pi}{4}$ ($\approx 0.7853\cdot x$).
+% \begin{macrocode}
+ \advance\dimen@ 0.7853\@tempdima
+}
+% \end{macrocode}
+% \end{macro}
+
+% \begin{macro}{\pIIe@atantwo}
+% \marg{dimen$_y$}\marg{dimen$_x$}\marg{dimreg$_{res}$}\\
+% Calculate \rarg{res} $= \arctan_2(y,x)$ and caches the result for later use.
+%
+% Overwrites |\@tempdim|(|a|,|b|,|c|,|d|),|\@tempa|, and |\dimen@|.
+% Both $y$ and $x$ must be dimensions.
+% \begin{macrocode}
+\newcommand*\pIIe@atantwo[3]{%
+ \edef\pIIe@csname{@atan2(\number\dimexpr#1\relax,\number\dimexpr#2\relax)}%
+ \expandafter\ifx\csname\pIIe@csname\endcsname\relax
+ \pIIe@atantwo@{#1}{#2}{#3}%
+ \pIIe@csedef{\pIIe@csname}{\the\dimexpr#3\relax}%
+ \else
+ #3\dimexpr\csname\pIIe@csname\endcsname\relax
+ \fi
+}
+% \end{macrocode}
+% \end{macro}
+
+% \begin{macro}{\pIIe@atantwo@}
+% \marg{dimen$_y$}\marg{dimen$_x$}\marg{dimreg$_{res}$}\\
+% Calculate \rarg{res} $= \arctan_2(y,x)$.
+% Overwrites |\@tempdim|(|a|,|b|,|c|,|d|),|\@tempa|, and |\dimen@|.
+% Both $y$ and $x$ must be dimensions.
+% \begin{macrocode}
+\newcommand*\pIIe@atantwo@[3]{%
+ \@tempdima\dimexpr#2\relax
+ \@tempdimb\dimexpr#1\relax
+% \end{macrocode}
+% Handle extremes
+% \begin{macrocode}
+ \ifdim\@tempdima=\z@\relax
+ \ifdim\@tempdimb>\z@\relax\dimen@90\p@
+ \else\ifdim\@tempdimb<\z@\relax\dimen@-90\p@
+ \else\dimen@0\p@
+ \fi\fi
+ \else
+% \end{macrocode}
+% Save angle adjustment term in |\@tempdimd|.
+% \begin{macrocode}
+ \@tempdimd\z@
+ \ifdim\@tempdima<\z@\relax
+ \ifdim\@tempdimb<\z@\relax\@tempdimd-180\p@
+ \else\@tempdimd180\p@
+ \fi
+ \fi
+% \end{macrocode}
+% Divide $\frac{y}{x}$ and check if $-1 \le \frac{y}{x} \le 1$.
+% \begin{macrocode}
+ \dimen@\dimexpr1pt * \@tempdimb/\@tempdima\relax
+ \@tempdimc\dimen@
+ \ifdim\@tempdimc<\z@\relax\@tempdimc-\@tempdimc\fi
+ \ifdim\@tempdimc>\p@\relax
+% \end{macrocode}
+% Use the equality $\arctan(x) = \pm\frac{1}{2}\pi - \arctan(\frac{1}{x})$
+% to stay within the valid domain of |\pIIe@atan@|. The sign $\pm$ is
+% positive when $x \ge 0$ and negative otherwise.
+% \begin{macrocode}
+ \dimen@\dimexpr1pt * \@tempdima/\@tempdimb\relax
+ \ifdim\dimen@<\z@\relax\def\@tempsign{-}\else\def\@tempsign{}\fi
+ \pIIe@atan@
+ \dimen@-\dimen@
+ \advance\dimen@ by \@tempsign1.5707pt\relax
+ \else
+ \pIIe@atan@
+ \fi
+% \end{macrocode}
+% And convert back to degrees ($\frac{180}{\pi} \approx 57.29578$)
+% \begin{macrocode}
+ \dimen@57.29578\dimen@
+% \end{macrocode}
+% Apply angle adjustment
+% \begin{macrocode}
+ \advance\dimen@ by \@tempdimd
+ \fi
+ #3\dimen@%
+}
+% \end{macrocode}
+% \end{macro}
+
+% \subsection{Sub routines for drawing an elliptical arc}
+%
+%
+% \begin{macro}{\pIIe@noneto}
+% \marg{dimen$_x$}\marg{dimen$_y$}\\
+% Ignores its arguments. Used as a no-op instead of |\pIIe@lineto| or |pIIe@moveto|.
+% \begin{macrocode}
+\newcommand*\pIIe@noneto[2]{}
+% \end{macrocode}
+% \end{macro}
+%
+%\begin{macro}{\pIIe@ellip@sincost@}
+% \marg{$\alpha_i$}\marg{$i$ = |one| or |two|}\\
+% Calculate $\sint{i}$ and $\cost{i}$ into the |\@ellip|(|sin|/|cos|)$i$.
+% Assumes |\@ellipratio| $=\frac{a}{b}$.
+% \begin{macrocode}
+\newcommand*\pIIe@ellip@sincost@[2]{%
+% \end{macrocode}
+% Put the $\sin(\alpha_i)$ and $\cos(\alpha_i)$ into |\@tempdima| and |\@tempdimb|.
+% \begin{macrocode}
+ \CalculateSin{#1}%
+ \CalculateCos{#1}%
+ \@tempdima\UseSin{#1}\p@
+ \@tempdimb\UseCos{#1}\p@
+% \end{macrocode}
+% Check for extremes where $\tan = \pm\infty$.
+% \begin{macrocode}
+ \ifdim\@tempdima=\p@\relax
+ \pIIe@csedef{@ellipsin#2}{1}%
+ \pIIe@csedef{@ellipcos#2}{0}%
+ \else\ifdim\@tempdima=-\p@\relax
+ \pIIe@csedef{@ellipsin#2}{-1}%
+ \pIIe@csedef{@ellipcos#2}{0}%
+ \else
+% \end{macrocode}
+% Calculate $\kk{i}$ in |\@tempdimc| and $\sqrt{1 + \kk{i}^2}$ in |\@tempdimd|,
+% and derive $\sint{i}$ and $\cost{i}$.
+% \begin{macrocode}
+ \@tempdimc\@ellipratio\dimexpr1pt * \@tempdima/\@tempdimb\relax
+ %\typeout{ i#2=\the\@tempdimc, sin(#1)=\the\@tempdima}%
+ \pIIe@ellip@csqrt{\p@}{\@tempdimc}\@tempdimd
+ \ifdim\@tempdimb<\z@\relax\@tempdimd-\@tempdimd\fi
+ \pIIe@csedef{@ellipsin#2}{\strip@pt\dimexpr1pt * \@tempdimc/\@tempdimd\relax}%
+ \pIIe@csedef{@ellipcos#2}{\strip@pt\dimexpr1pt * \p@/\@tempdimd\relax}%
+ \fi\fi
+}
+% \end{macrocode}
+% \end{macro}
+
+%\begin{macro}{\pIIe@ellip@sincost}
+% \marg{$\alpha_1$}\marg{$\alpha_2$}\\
+% Calculate $\sint{i}$ and $\cost{i}$ into the |\@ellip|(|sin|/|cos|)(|one|/|two|).
+% Assumes |\@ovro|$=a$ and |\@ovri|$=b$ with $b \ne 0$.
+% \begin{macrocode}
+\newcommand*\pIIe@ellip@sincost[2]{%
+% \end{macrocode}
+% Set |\@ellipratio| to the ratio $\frac{a}{b}$.
+% \begin{macrocode}
+ %\typeout{ calc sin cos: angles (#1,#2), radii: (\the\@ovro,\the\@ovri)}%
+ \edef\@ellipratio{\strip@pt\dimexpr1pt * \@ovro/\@ovri\relax}%
+% \end{macrocode}
+% And calculate $\sint{i}$ and $\cost{i}$
+% \begin{macrocode}
+ \pIIe@ellip@sincost@{#1}{one}%
+ \pIIe@ellip@sincost@{#2}{two}%
+ %\typeout{ sincos(a=#1)=(\@ellipsinone,\@ellipcosone), sincos(a=#2)=(\@ellipsintwo,\@ellipcostwo), }%
+}
+% \end{macrocode}
+% \end{macro}
+%
+%
+%\begin{macro}{\pIIe@omega}
+% \marg{$i$ = |one| or |two|}\\
+% Calculates $\xellipse_i$ into |\@tempdima| and |\@tempdimb|.
+% Assumes |\@ovro|$=a$ and |\@ovri|$=b$.
+% \begin{macrocode}
+\newcommand*\pIIe@omega[3]{%
+ \@tempdima\csname @ellipcos#3\endcsname\@ovro
+ \advance\@tempdima by #1\relax
+ \@tempdimb\csname @ellipsin#3\endcsname\@ovri
+ \advance\@tempdimb by #2\relax
+}
+% \end{macrocode}
+% \end{macro}
+
+%\begin{macro}{\pIIe@omegai}
+% \marg{$i$ = |one| or |two|}\\
+% Calculates $\xellipse'_i$ into |\@tempdimc| and |\@tempdimd|.
+% Assumes |\@ovro|$=a$ and |\@ovri|$=b$.
+% \begin{macrocode}
+\newcommand*\pIIe@omegai[1]{%
+ \@tempdimc\csname @ellipsin#1\endcsname\@ovro
+ \@tempdimc-\@tempdimc
+ \@tempdimd\csname @ellipcos#1\endcsname\@ovri
+}
+% \end{macrocode}
+% \end{macro}
+
+%\begin{macro}{\pIIe@ellip@kappa}
+% Calculates $\kappa$, expects |\@ellip|(|sin|/|cos|)(|one|/|two|) to be defined.
+% \begin{macrocode}
+\newcommand*\pIIe@ellip@kappa{%
+% \end{macrocode}
+% Calculate $\sint{21}$ and $\cost{21}$ in |\@tempdima| and |\@tempdimb|.
+% \begin{macrocode}
+ \@ovyy\@ellipsinone\p@
+ \@ovxx\@ellipcosone\p@
+ \@tempdima\@ellipcostwo\@ovyy
+ \@tempdima-\@tempdima
+ \advance\@tempdima by \@ellipsintwo\@ovxx
+ \@tempdimb\@ellipcostwo\@ovxx
+ \advance\@tempdimb by \@ellipsintwo\@ovyy
+% \end{macrocode}
+% First test if $\sint{21} = 0$ to prevent division by zero. In that
+% case, it must have been that $\alpha_1 = \alpha_2$ and we set $\kappa$ to zero
+% so it the control points become equal to the start and end point.
+% \begin{macrocode}
+ \ifdim\@tempdima=\z@\relax
+ \edef\@ellipkappa{0}%
+ \else
+% \end{macrocode}
+% Calculate $\kappa_{tan}$ in |\dimen@|
+% \begin{macrocode}
+ \dimen@\dimexpr1pt - \@tempdimb\relax
+ \dimen@\dimexpr1pt * \dimen@/\@tempdima\relax
+% \end{macrocode}
+% Calculate $\kappa_{sqrt}$ in |\dimen@|
+% \begin{macrocode}
+ \pIIe@ellip@csqrt{2\p@}{1.73205\dimen@}{\dimen@}%
+% \end{macrocode}
+% Calculate $\kappa$ in |\dimen@|
+% \begin{macrocode}
+ \advance\dimen@ by -\p@
+ \divide\dimen@ by 3%
+ \edef\@tempa{\strip@pt\@tempdima}%
+ \dimen@\@tempa\dimen@
+ \edef\@ellipkappa{\strip@pt\dimen@}%
+ \fi
+ %\typeout{ calculated kappa: \@ellipkappa}%
+}
+% \end{macrocode}
+% \end{macro}
+
+% \subsection{Core routines for drawing elliptical arcs}
+%
+%\begin{macro}{\pIIe@elliparc@}
+% \oarg{start}\marg{$c_x$}\marg{$c_y$}\marg{$\alpha_1$}\marg{$\alpha_2$}\\
+% Assumes that the radii are set as |\@ovro|$=a$ and |\@ovri|$=b$.
+% This is the main routine for drawing an elliptic arc, where $\abs{\alpha_2 - \alpha_1}\,\le 90$.
+% \begin{macrocode}
+\newcommand*\pIIe@elliparc@[5]{%
+ %\typeout{elliparc: #1, center: (#2, #3), radius (\the\@ovro, \the\@ovri),angle (#4, #5)}%
+% \end{macrocode}
+% Define initial action: 0 (lineto), 1(moveto), or 2 (nothing)
+% \begin{macrocode}
+ \ifcase #1\relax
+ \let\@ellip@startto\pIIe@lineto
+ \or \let\@ellip@startto\pIIe@moveto
+ \or \let\@ellip@startto\pIIe@noneto%
+ \else\PackageWarning{ellipse}{Illegal initial action in \protect\elliparc: %
+ must be one of 0 (lineto), 1 (moveto) or 2 (do nothing) but I got: #1}%
+ \fi
+% \end{macrocode}
+% Perform just the start action if the radii are zero
+% \begin{macrocode}
+ \ifdim\@ovro=\z@\relax\@ovri\z@\fi
+ \ifdim\@ovri=\z@\relax
+ \@ellip@startto{#2}{#3}%
+ \else
+% \end{macrocode}
+% Calculate $\sint{i}$ and $\cost{i}$ first into the |\@ellip|(|sin|/|cos|)(|one|/|two|) registers.
+% \begin{macrocode}
+ \pIIe@ellip@sincost{#4}{#5}%
+% \end{macrocode}
+% And draw..
+% \begin{macrocode}
+ \pIIe@elliparc@draw{#2}{#3}%
+ \fi
+}
+% \end{macrocode}
+% \end{macro}
+%
+%\begin{macro}{\pIIe@elliparc@t}
+% \oarg{start}\marg{$c_x$}\marg{$c_y$}\marg{$t_1$}\marg{$t_2$}\\
+% Assumes that the radii are set as |\@ovro|$=a$ and |\@ovri|$=b$.
+% Moreover, this routine take $t_1$ and $t_2$ as the angles of the ellipse equation (instead of real angles $\alpha_i$).
+% This routine is mainly for other libraries that may already have computed the $t$ angles
+% and need a bit more efficiency.
+% \begin{macrocode}
+\newcommand*\pIIe@elliparc@t[5]{%
+% \end{macrocode}
+% Define initial action: 0 (lineto), 1(moveto), or 2 (nothing)
+% \begin{macrocode}
+ \ifcase #1\relax
+ \let\@ellip@startto\pIIe@lineto
+ \or \let\@ellip@startto\pIIe@moveto
+ \or \let\@ellip@startto\pIIe@noneto%
+ \else\PackageWarning{ellipse}{Illegal initial action in \protect\elliparc: %
+ must be one of 0 (lineto), 1 (moveto) or 2 (do nothing) but I got: #1}%
+ \fi
+% \end{macrocode}
+% Perform just the start action if the radii are zero
+% \begin{macrocode}
+ \ifdim\@ovro=\z@\relax\@ovri\z@\fi
+ \ifdim\@ovri=\z@\relax
+ \@ellip@startto{#2}{#3}%
+ \else
+% \end{macrocode}
+% Calculate $\sint{i}$ and $\cost{i}$ first into the |\@ellip|(|sin|/|cos|)(|one|/|two|) registers.
+% \begin{macrocode}
+ \CalculateSin{#4}\CalculateCos{#4}%
+ \edef\@ellipsinone{\UseSin{#4}}%
+ \edef\@ellipcosone{\UseCos{#4}}%
+ \CalculateSin{#5}\CalculateCos{#5}%
+ \edef\@ellipsintwo{\UseSin{#5}}%
+ \edef\@ellipcostwo{\UseCos{#5}}%
+% \end{macrocode}
+% And draw..
+% \begin{macrocode}
+ \pIIe@elliparc@draw{#2}{#3}%
+ \fi
+}
+% \end{macrocode}
+% \end{macro}
+%
+% \begin{macro}{pIIe@elliparc@draw}
+% \marg{$c_x$}\marg{$c_y$}\\
+% Expects $a =$|\@ovro|, $b$ = |\@ovri|, and |\@ellip|(|sin|/|cos|)(|one|/|two|) defined.
+% |\@ellipstarto| should contain the initial drawing action and is called with an initial
+% $x$ and $y$ coordinate (usually equal to |\pIIe@lineto|,|\pIIe@moveto|, or |pIIe@noneto|).
+% \begin{macrocode}
+\newcommand*\pIIe@elliparc@draw[2]{%
+% Calculate $\kappa$.
+% \begin{macrocode}
+ \pIIe@ellip@kappa%
+% \end{macrocode}
+% Now we are ready to compute the control points. First $p_1$.
+% \begin{macrocode}
+ \pIIe@omega{#1}{#2}{one}%
+ %\typeout{ point one: (\the\@tempdima,\the\@tempdimb)}%
+% \end{macrocode}
+% The coordinates are added to the path if and how necessary:
+% \begin{macrocode}
+ \@ellip@startto\@tempdima\@tempdimb
+% \end{macrocode}
+% Add control point $q_1$
+% \begin{macrocode}
+ \pIIe@omegai{one}%
+ \advance\@tempdima by \@ellipkappa\@tempdimc
+ \advance\@tempdimb by \@ellipkappa\@tempdimd
+ \pIIe@add@nums\@tempdima\@tempdimb
+ %\typeout{ control one: (\the\@tempdima,\the\@tempdimb)}%
+% \end{macrocode}
+% Calculate $p_2$
+% \begin{macrocode}
+ \pIIe@omega{#1}{#2}{two}%
+% \end{macrocode}
+% Add control point $q_1$
+% \begin{macrocode}
+ \pIIe@omegai{two}%
+ \@tempdimc\@ellipkappa\@tempdimc
+ \@tempdimd\@ellipkappa\@tempdimd
+ \@tempdimc-\@tempdimc
+ \@tempdimd-\@tempdimd
+ \advance\@tempdimc by \@tempdima
+ \advance\@tempdimd by \@tempdimb
+ \pIIe@add@nums\@tempdimc\@tempdimd
+ %\typeout{ control two: (\the\@tempdimc,\the\@tempdimd)}%
+% \end{macrocode}
+% And finally add $p_2$ to the path
+% \begin{macrocode}
+ \pIIe@add@CP\@tempdima\@tempdimb
+ %\typeout{ point two: (\the\@tempdima,\the\@tempdimb)}%
+ \pIIe@addtoGraph\pIIe@curveto@op
+}
+% \end{macrocode}
+% \end{macro}
+%
+%
+% \subsection{Normalizing elliptical arcs}
+%
+% \begin{macro}{pIIe@elliparc}
+% \begin{macro}{pIIe@@elliparc}
+% \oarg{start}\marg{$c_x$}\marg{$c_y$}\marg{a}\marg{b}\marg{$\alpha_1$}\marg{$\alpha_2$}\\
+% \strut\\
+% These two macros check the arguments and normalize the angles.
+% \begin{macrocode}
+\newcommand*\pIIe@elliparc[7][0]{%
+% \end{macrocode}
+% Store the radii in registers, where |\@ovro|$=a$ and |\@ovri|$=b$.
+% \begin{macrocode}
+ \@ovro #4\relax
+ \@ovri #5\relax
+ \iffalse%dim\@ovro=\@ovri
+% \end{macrocode}
+% Call the circular arc routine if the x- and y-radius are equal
+% \begin{macrocode}
+ \pIIe@arc[#1]{#2}{#3}{#4}{#6}{#7}
+ \else
+% \end{macrocode}
+% Normalize angles such that the arc angle $\abs{\alpha_2 - \alpha_1}\,\le 720$.
+% Store the arc angle in |\@arclen|.
+% \begin{macrocode}
+ \ifdim \@ovro<\z@ \pIIe@badcircarg\else
+ \ifdim \@ovri<\z@ \pIIe@badcircarg\else
+ \@arclen #7\p@ \advance\@arclen -#6\p@
+ \ifdim \@arclen<\z@ \def\@tempsign{-}\else\def\@tempsign{}\fi
+ \ifdim \@tempsign\@arclen>720\p@
+ \PackageWarning {ellipse}{The arc angle is reduced to -720..720}%
+ \@whiledim \@tempsign\@arclen>720\p@ \do {\advance\@arclen-\@tempsign360\p@}%
+ \@tempdima #6\p@ \advance\@tempdima \@arclen
+ \edef\@angleend{\strip@pt\@tempdima}%
+ \pIIe@@elliparc{#1}{#2}{#3}{#6}{\@angleend}%
+ \else
+ \pIIe@@elliparc{#1}{#2}{#3}{#6}{#7}%
+ \fi
+ \fi
+ \fi
+ \fi
+}
+% \end{macrocode}
+% |\pIIe@@elliparc| divides the total angle in parts of at most $90$ degrees.
+% Assumes |\@ovro|$=a$ and |\@ovri|$=b$, and |\@arclen| the arc angle, with |\@tempsign|
+% sign of the arc angle.
+% \begin{macrocode}
+\newcommand*\pIIe@@elliparc[5]{%
+ \begingroup
+ \ifdim \@tempsign\@arclen>90\p@
+% \end{macrocode}
+% If the arc angle is too large, the arc is recursively
+% divided into 2 parts until the arc angle is at most 90~degrees.
+% \begin{macrocode}
+ \divide\@arclen 2%
+ \@tempdima #4\p@\advance\@tempdima by \@arclen
+ \edef\@anglemid{\strip@pt\@tempdima}%
+ \def\@tempa{\pIIe@@elliparc{#1}{#2}{#3}{#4}}%
+ \expandafter\@tempa\expandafter{\@anglemid}%
+ \def\@tempa{\pIIe@@elliparc{2}{#2}{#3}}%
+ \expandafter\@tempa\expandafter{\@anglemid}{#5}%
+ \else
+% \end{macrocode}
+% The arc angle is smaller than 90 degrees.
+% \begin{macrocode}
+ \pIIe@elliparc@{#1}{#2}{#3}{#4}{#5}%
+ \fi
+ \endgroup
+}%
+% \end{macrocode}
+% \end{macro}
+%\end{macro}
+
+% \subsection{Drawing elliptical arcs}
+%
+%\begin{macro}{\elliparc}
+%\begin{macro}{\pIIeelliparc}
+% \oarg{start}\marg{center-x}\marg{center-y}\marg{radius-x}\marg{radius-y}\marg{start-angle}\marg{end-angle}\\
+% \strut\\
+% The main elliptical arc drawing routine. We start with |\pIIeelliparc| to avoid conflicts with
+% other packages.
+% \begin{macrocode}
+\newcommand*\pIIeelliparc[7][0]{%
+ \@killglue
+ \pIIe@elliparc[#1]{#2\unitlength}{#3\unitlength}{#4\unitlength}{#5\unitlength}{#6}{#7}%
+ \ignorespaces%
+}
+\ifx\undefined\elliparc\else
+ \PackageWarning{ellipse}{\protect\elliparc\space is redefined}%
+\fi
+\let\elliparc\pIIeelliparc
+% \end{macrocode}
+% \end{macro}
+% \end{macro}
+
+% \begin{macro}{\earc}
+% \begin{macro}{\earc*}
+% |[|\rarg{$\alpha_0$}|,|\rarg{$\alpha_1$}|]|\marg{radius-x}\marg{radius-y}\\
+% \strut\\
+% The \cmd{\earc} command generalizes the standard \cmd{\arc} with both
+% a $x$- and $y$-radius. The |\earc*| version draws a filled elliptical arc while
+% |\earc| only strokes the elliptical arc. Both take an optional comma separated pair of
+% angles which specify the initial and final angle ($0$ and $360$ by default).
+% We start with \cmd{\pIIeearc} to avoid conflicts with otherpackages.
+% \begin{macrocode}
+\newcommand*\pIIeearc
+ {\@ifstar{\@tempswatrue\pIIe@earc@}{\@tempswafalse\pIIe@earc@}}
+\newcommand*\pIIe@earc@[3][0,360]{\pIIe@earc@@(#1){#2}{#3}}
+\def\pIIe@earc@@(#1,#2)#3#4{%
+ \if@tempswa
+ \pIIe@moveto\z@\z@
+ \pIIe@elliparc{\z@}{\z@}{#3\unitlength}{#4\unitlength}{#1}{#2}%
+ \pIIe@closepath\pIIe@fillGraph
+ \else
+ \pIIe@elliparc[1]{\z@}{\z@}{#3\unitlength}{#4\unitlength}{#1}{#2}%
+ \pIIe@strokeGraph
+ \fi}
+\ifx\undefined\earc\else
+ \PackageWarning{ellipse}{\protect\earc\space is redefined}%
+\fi
+\let\earc\pIIeearc
+% \end{macrocode}
+% \end{macro}
+% \end{macro}
+
+% \begin{macro}{\ellipse}
+% \begin{macro}{\ellipse*}
+% \marg{radius-x}\marg{radius-y}\\
+% \strut\\
+% The \cmd{\ellipse} draws an ellipse with the specified $x$- and $y$-radius.
+% The |\ellipse*| version draws a filled ellipse.
+% We start with \cmd{\pIIeellipse} to avoid conflicts with other packages.
+% The implementation redirects immediately to |earc| which generalized this command.
+% \begin{macrocode}
+\newcommand*\pIIeellipse
+ {\@ifstar{\@tempswatrue\pIIe@earc@}{\@tempswafalse\pIIe@earc@}}
+\let\ellipse\pIIeellipse
+% \end{macrocode}
+% \end{macro}
+% \end{macro}
+
+%
+% \Finale
+\endinput
+ \ No newline at end of file
diff --git a/graphics/ellipse/ellipse.ins b/graphics/ellipse/ellipse.ins
new file mode 100644
index 0000000000..51abc729e9
--- /dev/null
+++ b/graphics/ellipse/ellipse.ins
@@ -0,0 +1,71 @@
+%%
+%% Copyright (C) 2015
+%% Daan Leijen
+%%
+%% This file is part of the LaTeX `ellipse' package.
+%% ------------------------------------------------
+%%
+%% 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 2003/12/01 or later.
+%%
+%% This work has the LPPL maintenance status "author-maintained".
+%%
+%% In particular, NO PERMISSION is granted to modify the contents of this
+%% file since it contains the legal notices that are placed in the files
+%% it generates.
+%%
+%% This work consists of all files listed in manifest.txt.
+%%
+%%
+%% --------------- start of docstrip commands ------------------
+%%
+
+\input docstrip
+
+\preamble
+
+Copyright (C) 2015
+Daan Leijen
+
+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 2003/12/01 or later.
+
+This work has the LPPL maintenance status "author-maintained".
+
+\endpreamble
+
+\keepsilent
+\askforoverwritefalse
+
+\usedir{tex/latex/pict2e}
+
+\generate{\file{ellipse.sty}{\from{ellipse.dtx}{package}}}
+
+%\usepreamble\empty
+%\usepostamble\empty
+
+
+\Msg{***********************************************************}
+\Msg{*}
+\Msg{* To finish the installation you have to move the following}
+\Msg{* files into a directory searched by TeX:}
+\Msg{*}
+\Msg{* All the files with extension `.sty' and `.def'}
+\Msg{*}
+\Msg{* To produce the documentation run the .dtx files through LaTeX.}
+\Msg{*}
+\Msg{* Happy TeXing}
+\Msg{***********************************************************}
+
+
+\endbatchfile \ No newline at end of file
diff --git a/graphics/ellipse/ellipse.pdf b/graphics/ellipse/ellipse.pdf
new file mode 100644
index 0000000000..8a043e3176
--- /dev/null
+++ b/graphics/ellipse/ellipse.pdf
Binary files differ