| 1 |
% \iffalse meta-comment |
| 2 |
% bez123.dtx |
| 3 |
% Author: Peter Wilson, Herries Press |
| 4 |
% Maintainer: Will Robertson (will dot robertson at latex-project dot org) |
| 5 |
% Copyright 1998--2004 Peter R. Wilson |
| 6 |
% |
| 7 |
% This work may be distributed and/or modified under the |
| 8 |
% conditions of the LaTeX Project Public License, either |
| 9 |
% version 1.3c of this license or (at your option) any |
| 10 |
% later version: <http://www.latex-project.org/lppl.txt> |
| 11 |
% |
| 12 |
% This work has the LPPL maintenance status "maintained". |
| 13 |
% The Current Maintainer of this work is Will Robertson. |
| 14 |
% |
| 15 |
% This work consists of the files listed in the README file. |
| 16 |
% |
| 17 |
% |
| 18 |
% |
| 19 |
%<*driver> |
| 20 |
\documentclass{ltxdoc} |
| 21 |
\usepackage{bez123} |
| 22 |
\EnableCrossrefs |
| 23 |
\CodelineIndex |
| 24 |
\setcounter{StandardModuleDepth}{1} |
| 25 |
\begin{document} |
| 26 |
\DocInput{bez123.dtx} |
| 27 |
\end{document} |
| 28 |
%</driver> |
| 29 |
% |
| 30 |
% \fi |
| 31 |
% |
| 32 |
% \CheckSum{988} |
| 33 |
% |
| 34 |
% \DoNotIndex{\',\.,\@M,\@@input,\@addtoreset,\@arabic,\@badmath} |
| 35 |
% \DoNotIndex{\@centercr,\@cite} |
| 36 |
% \DoNotIndex{\@dotsep,\@empty,\@float,\@gobble,\@gobbletwo,\@ignoretrue} |
| 37 |
% \DoNotIndex{\@input,\@ixpt,\@m} |
| 38 |
% \DoNotIndex{\@minus,\@mkboth,\@ne,\@nil,\@nomath,\@plus,\@set@topoint} |
| 39 |
% \DoNotIndex{\@tempboxa,\@tempcnta,\@tempdima,\@tempdimb} |
| 40 |
% \DoNotIndex{\@tempswafalse,\@tempswatrue,\@viipt,\@viiipt,\@vipt} |
| 41 |
% \DoNotIndex{\@vpt,\@warning,\@xiipt,\@xipt,\@xivpt,\@xpt,\@xviipt} |
| 42 |
% \DoNotIndex{\@xxpt,\@xxvpt,\\,\ ,\addpenalty,\addtolength,\addvspace} |
| 43 |
% \DoNotIndex{\advance,\Alph,\alph} |
| 44 |
% \DoNotIndex{\arabic,\ast,\begin,\begingroup,\bfseries,\bgroup,\box} |
| 45 |
% \DoNotIndex{\bullet} |
| 46 |
% \DoNotIndex{\cdot,\cite,\CodelineIndex,\cr,\day,\DeclareOption} |
| 47 |
% \DoNotIndex{\def,\DisableCrossrefs,\divide,\DocInput,\documentclass} |
| 48 |
% \DoNotIndex{\DoNotIndex,\egroup,\ifdim,\else,\fi,\em,\endtrivlist} |
| 49 |
% \DoNotIndex{\EnableCrossrefs,\end,\end@dblfloat,\end@float,\endgroup} |
| 50 |
% \DoNotIndex{\endlist,\everycr,\everypar,\ExecuteOptions,\expandafter} |
| 51 |
% \DoNotIndex{\fbox} |
| 52 |
% \DoNotIndex{\filedate,\filename,\fileversion,\fontsize,\framebox,\gdef} |
| 53 |
% \DoNotIndex{\global,\halign,\hangindent,\hbox,\hfil,\hfill,\hrule} |
| 54 |
% \DoNotIndex{\hsize,\hskip,\hspace,\hss,\if@tempswa,\ifcase,\or,\fi,\fi} |
| 55 |
% \DoNotIndex{\ifhmode,\ifvmode,\ifnum,\iftrue,\ifx,\fi,\fi,\fi,\fi,\fi} |
| 56 |
% \DoNotIndex{\input} |
| 57 |
% \DoNotIndex{\jobname,\kern,\leavevmode,\let,\leftmark} |
| 58 |
% \DoNotIndex{\list,\llap,\long,\m@ne,\m@th,\mark,\markboth,\markright} |
| 59 |
% \DoNotIndex{\month,\newcommand,\newcounter,\newenvironment} |
| 60 |
% \DoNotIndex{\NeedsTeXFormat,\newdimen} |
| 61 |
% \DoNotIndex{\newlength,\newpage,\nobreak,\noindent,\null,\number} |
| 62 |
% \DoNotIndex{\numberline,\OldMakeindex,\OnlyDescription,\p@} |
| 63 |
% \DoNotIndex{\pagestyle,\par,\paragraph,\paragraphmark,\parfillskip} |
| 64 |
% \DoNotIndex{\penalty,\PrintChanges,\PrintIndex,\ProcessOptions} |
| 65 |
% \DoNotIndex{\protect,\ProvidesClass,\raggedbottom,\raggedright} |
| 66 |
% \DoNotIndex{\refstepcounter,\relax,\renewcommand,\reset@font} |
| 67 |
% \DoNotIndex{\rightmargin,\rightmark,\rightskip,\rlap,\rmfamily,\roman} |
| 68 |
% \DoNotIndex{\roman,\secdef,\selectfont,\setbox,\setcounter,\setlength} |
| 69 |
% \DoNotIndex{\settowidth,\sfcode,\skip,\sloppy,\slshape,\space} |
| 70 |
% \DoNotIndex{\symbol,\the,\trivlist,\typeout,\tw@,\undefined,\uppercase} |
| 71 |
% \DoNotIndex{\usecounter,\usefont,\usepackage,\vfil,\vfill,\viiipt} |
| 72 |
% \DoNotIndex{\viipt,\vipt,\vskip,\vspace} |
| 73 |
% \DoNotIndex{\wd,\xiipt,\year,\z@} |
| 74 |
% |
| 75 |
% \def\dtxfile{bez123.dtx} |
| 76 |
% \def\fileversion{v1.1} |
| 77 |
% \def\filedate{1998/10/14} |
| 78 |
% \def\fileversion{v1.1a} |
| 79 |
% \def\filedate{2004/04/16} |
| 80 |
% \def\fileversion{v1.1b} |
| 81 |
% \def\filedate{2009/09/02} |
| 82 |
% \newcommand*{\Lpack}[1]{\textsf {#1}} ^^A typest a package |
| 83 |
% \newcommand*{\Lopt}[1]{\textsf {#1}} ^^A typeset an option |
| 84 |
% \newcommand*{\file}[1]{\texttt {#1}} ^^A typeset a file |
| 85 |
% \newcommand*{\Lcount}[1]{\textsl {\small#1}} ^^A typeset a counter |
| 86 |
% \newcommand*{\pstyle}[1]{\textsl {#1}} ^^A typeset a pagestyle |
| 87 |
% \newcommand*{\Lenv}[1]{\texttt {#1}} ^^A typeset an environment |
| 88 |
% \newcommand{\eqref}[1]{equation~(\ref{#1})} ^^A typeset ref to an equation |
| 89 |
% |
| 90 |
% \title{The \Lpack{bez123} and \Lpack{multiply} packages\thanks{This |
| 91 |
% file (\texttt{\dtxfile}) has version number \fileversion, |
| 92 |
% last revised \filedate.}} |
| 93 |
% |
| 94 |
% \author{% |
| 95 |
% Author: Peter Wilson, Herries Press\\ |
| 96 |
% Maintainer: Will Robertson\\ |
| 97 |
% \texttt{will dot robertson at latex-project dot org} |
| 98 |
% } |
| 99 |
% \date{\filedate} |
| 100 |
% \maketitle |
| 101 |
% \begin{abstract} |
| 102 |
% The \Lpack{bez123} package provides for the drawing of linear, cubic, |
| 103 |
% and rational quadratic Bezier curves. The \Lpack{multiply} package |
| 104 |
% provides a command to multiply a length without numerical overflow. |
| 105 |
% \end{abstract} |
| 106 |
% \tableofcontents |
| 107 |
% \listoftables |
| 108 |
% \listoffigures |
| 109 |
% |
| 110 |
% \StopEventually{} |
| 111 |
% |
| 112 |
% |
| 113 |
% |
| 114 |
% \section{Introduction} |
| 115 |
% |
| 116 |
% This document provides the commented source for a \LaTeX{} |
| 117 |
% package file that extends the \LaTeX{} facilities for drawing |
| 118 |
% Bezier curves. The package was originally developed as part of |
| 119 |
% a suite designed for the typesetting of |
| 120 |
% documents according to the rules for ISO international |
| 121 |
% standards~\cite{PRW96i}. |
| 122 |
% This manual is typeset according to the conventions of the |
| 123 |
% \LaTeX{} \textsc{docstrip} utility which enables the automatic |
| 124 |
% extraction of the \LaTeX{} macro source files~\cite{GOOSSENS94}. |
| 125 |
% |
| 126 |
% Drawing a non-rational quadratic Bezier curve is provided as part |
| 127 |
% of the standard \LaTeX{} system. |
| 128 |
% Section~\ref{sec:usage} provides the user manual for the new commands |
| 129 |
% supplied by this package for drawing a variety of Bezier curves. |
| 130 |
% These include commands for drawing linear and cubic non-rational Bezier |
| 131 |
% curves and rational quadratic curves. |
| 132 |
% |
| 133 |
% Section~\ref{sec:bez} describes the implementation of the package. |
| 134 |
% As a side-effect of the implementation, a facility is also provided |
| 135 |
% for performing multiplication in \TeX{} without overflow. This is |
| 136 |
% described in Section~\ref{sec:mnoflow}. |
| 137 |
% |
| 138 |
% |
| 139 |
% |
| 140 |
% \section{Usage} \label{sec:usage} |
| 141 |
% |
| 142 |
% Leslie Lamport provided the means of drawing a quadratic Bezier curve |
| 143 |
% \emph{via} the \LaTeXe{} |\qbezier|~\cite[pp. 125--126]{LAMPORT94} command. |
| 144 |
% This package |
| 145 |
% extends the Bezier facility by providing commands to draw linear, |
| 146 |
% rational quadratic, and cubic Bezier curves. |
| 147 |
% |
| 148 |
% Bezier curves are named after Pierre Bezier who invented them. They |
| 149 |
% are widely used within Computer Aided Design (CAD) programs and other |
| 150 |
% graphics systems; descriptions can be found in many places, with varying |
| 151 |
% degrees of mathematical complexity, such |
| 152 |
% as~\cite{FandP,MORTENSON85,FARIN90}. |
| 153 |
% |
| 154 |
% The Bezier curve is a parameterized curve of degree $n$ and can |
| 155 |
% therefore be specified by $(n+1)$ points |
| 156 |
% (i.e., point $p_{0}$ through $p_{n}$). |
| 157 |
% Among its other properties, a Bezier curve of degree $n$ passes through |
| 158 |
% through the points $p_{0}$ and $p_{n}$ and passes close to the other |
| 159 |
% defining points. The general equation for a Bezier curve of degree $n$ with |
| 160 |
% parameter $t$ is |
| 161 |
% \begin{equation} |
| 162 |
% p(t) = a_{0} + a_{1}t + a_{2}t^{2} + \cdots + a_{n}t^{n} \label{eq:gen} |
| 163 |
% \end{equation} |
| 164 |
% where the coefficients $a_{i}$ depend on the defining points, and |
| 165 |
% traditionally $0 \leq t \leq 1$. |
| 166 |
% |
| 167 |
% For a linear (degree $1$) curve, the equation is |
| 168 |
% \begin{equation} |
| 169 |
% p(t) = p_{0} + (p_{1} - p_{0})t \label{eq:lin} |
| 170 |
% \end{equation} |
| 171 |
% By inspection, $p(0) = p_{0}$ and $p(1) = p_{1}$. |
| 172 |
% |
| 173 |
% Rearranging \eqref{eq:gen} slightly we get |
| 174 |
% \begin{equation} |
| 175 |
% p(t) - p_{0} = (p_{1} - p_{0})t \label{eq:lin2} |
| 176 |
% \end{equation} |
| 177 |
% In other words, we can march along the curve from the starting point to |
| 178 |
% the ending point by evaluating the right hand side of |
| 179 |
% \eqref{eq:lin2} for increasing values of the parameter $t$. |
| 180 |
% |
| 181 |
% In order to shorten the equations slightly, and also make them more |
| 182 |
% convenient to work with numerically, we will use the notation |
| 183 |
% \begin{displaymath} |
| 184 |
% l_{pq} = p_{p} - p_{q} |
| 185 |
% \end{displaymath} |
| 186 |
% Thus, the final form for the linear Bezier curve is |
| 187 |
% \begin{equation} |
| 188 |
% p(t) - p_{0} = l_{10}t \label{eq:lin3} |
| 189 |
% \end{equation} |
| 190 |
% |
| 191 |
% \DescribeMacro{\lbezier} |
| 192 |
% The command |\lbezier[|\meta{N}|](|\meta{p0}|)(|\meta{p1}|)| draws a |
| 193 |
% linear Bezier curve with \meta{N} plotted points from the point \meta{p0} |
| 194 |
% (with coordinates \meta{x0,y0}) to the point \meta{p1} (with |
| 195 |
% coordinates \meta{x1,y1}). \meta{N} is an optional argument. If it is |
| 196 |
% either not given or is given with a value of zero, |
| 197 |
% then the command will calculate the number of points to be |
| 198 |
% plotted, subject to a maximum number. |
| 199 |
% There must be no spaces between the arguments to the |
| 200 |
% |\lbezier| command; this restriction also applies to the other Bezier |
| 201 |
% drawing commands provided by the \Lpack{bez123} package. |
| 202 |
% |
| 203 |
% Figure~\ref{fig:beta} shows an example of a dotted line drawn using |
| 204 |
% the |\lbezier| command. The actual code used is: |
| 205 |
% \begin{verbatim} |
| 206 |
% \lbezier[50](15,30)(30,0) |
| 207 |
% \end{verbatim} |
| 208 |
% thus drawing a straight line consisting of 50 points. |
| 209 |
% |
| 210 |
% \DescribeMacro{\qbeziermax} |
| 211 |
% The standard \LaTeX{} command |\qbeziermax| sets a maximum limit |
| 212 |
% on the number of points used to draw any of the Bezier curves. |
| 213 |
% |
| 214 |
% \DescribeMacro{\thinlines} |
| 215 |
% \DescribeMacro{\thicklines} |
| 216 |
% \DescribeMacro{\linethickness} |
| 217 |
% The `points' used in drawing the Bezier curves are small squares. The |
| 218 |
% size of these squares are controlled by the standard \LaTeX{} |
| 219 |
% |\thinlines|, |\thicklines| and/or |\linethickness| commands. |
| 220 |
% Consult Lamport~\cite{LAMPORT94} for descriptions of these, and |
| 221 |
% |\qbeziermax|, commands. |
| 222 |
% |
| 223 |
% |
| 224 |
% It is convenient to introduce some general properties of Bezier curves |
| 225 |
% at this point. |
| 226 |
% \begin{itemize} |
| 227 |
% \item A degree $n$ Bezier curve is defined by $(n+1)$ points which we |
| 228 |
% will label as $p_{0}$ through $p_{n}$. The lines joining the points |
| 229 |
% $p_{0}, p_{1}, \ldots , p_{n}$ are called the \emph{control polygon}. |
| 230 |
% The Bezier curve is parameterized by a variable we will call $t$, with |
| 231 |
% $0 \leq t \leq 1$. |
| 232 |
% \item A degree $n$ Bezier curve starts at point $p_{0}$ and ends at |
| 233 |
% point $p_{n}$. |
| 234 |
% \item At $t=0$ the curve passes through $p_{0}$ and is tangent to the |
| 235 |
% line $l_{10} = p_{1}-p_{0}$. |
| 236 |
% \item At $t=1$ the curve passes through $p_{n}$ and is tangent to the |
| 237 |
% line $l_{(n)(n-1)} = p_{n}-p_{(n-1)}$. |
| 238 |
% \item A \emph{non-rational} Bezier curve lies within the \emph{convex |
| 239 |
% hull}\footnote{The convex hull can be thought of as the shape that a rubber |
| 240 |
% band will take if it is stretched around pins placed at each point.} |
| 241 |
% of the points $p_{0}$ through $p_{n}$. For examples of convex hulls see |
| 242 |
% figure~\ref{fig:ch}. Note that the shape of a convex hull is independant |
| 243 |
% of the ordering of the points. |
| 244 |
% \end{itemize} |
| 245 |
% |
| 246 |
% \begin{figure} |
| 247 |
% \centering |
| 248 |
% \setlength{\unitlength}{1mm} |
| 249 |
% \begin{picture}(70,80) |
| 250 |
% ^^A degree 3 |
| 251 |
% \put(0,5){\begin{picture}(30,30) |
| 252 |
% \thinlines |
| 253 |
% \put(0,0){\circle{2}} |
| 254 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 255 |
% \put(10,30){\circle{2}} |
| 256 |
% \put(8,30){\makebox(0,0)[br]{1}} |
| 257 |
% \put(20,0){\circle{2}} |
| 258 |
% \put(22,0){\makebox(0,0)[bl]{2}} |
| 259 |
% \put(30,30){\circle{2}} |
| 260 |
% \put(32,30){\makebox(0,0)[bl]{3}} |
| 261 |
% ^^A convex hull |
| 262 |
% \put(0,0){\line(1,0){20}} |
| 263 |
% \put(20,0){\line(1,3){10}} |
| 264 |
% \put(30,30){\line(-1,0){20}} |
| 265 |
% \put(10,30){\line(-1,-3){10}} |
| 266 |
% ^^A control polygon |
| 267 |
% ^^A \put(0,0){\vector(1,3){10}} |
| 268 |
% ^^A \put(10,30){\vector(1,-3){10}} |
| 269 |
% ^^A \put(20,0){\vector(1,3){10}} |
| 270 |
% \thicklines |
| 271 |
% ^^A \cbezier[30](0,0)(10,30)(20,0)(30,30) |
| 272 |
% \thinlines |
| 273 |
% \end{picture}} |
| 274 |
% ^^A degree 3 |
| 275 |
% \put(0,45){\begin{picture}(30,30) |
| 276 |
% \thinlines |
| 277 |
% \put(0,0){\circle{2}} |
| 278 |
% \put(-2,0){\makebox(0,0)[br]{0, 3}} |
| 279 |
% \put(30,0){\circle{2}} |
| 280 |
% \put(32,0){\makebox(0,0)[bl]{1}} |
| 281 |
% \put(0,30){\circle{2}} |
| 282 |
% \put(2,30){\makebox(0,0)[bl]{2}} |
| 283 |
% ^^A convex hull |
| 284 |
% \put(0,0){\line(1,0){30}} |
| 285 |
% \put(30,0){\line(-1,1){30}} |
| 286 |
% \put(0,30){\line(0,-1){30}} |
| 287 |
% ^^A control polygon |
| 288 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 289 |
% ^^A \put(30,0){\vector(-1,1){30}} |
| 290 |
% ^^A \put(0,30){\vector(0,-1){30}} |
| 291 |
% \thicklines |
| 292 |
% ^^A \cbezier[30](0,0)(30,0)(0,30)(0,0) |
| 293 |
% \thinlines |
| 294 |
% \end{picture}} |
| 295 |
% ^^A degree 3 |
| 296 |
% \put(45,0){\begin{picture}(30,30) |
| 297 |
% \thinlines |
| 298 |
% \put(0,0){\circle{2}} |
| 299 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 300 |
% \put(10,30){\circle{2}} |
| 301 |
% \put(8,30){\makebox(0,0)[br]{2}} |
| 302 |
% \put(20,0){\circle{2}} |
| 303 |
% \put(22,0){\makebox(0,0)[bl]{3}} |
| 304 |
% \put(30,30){\circle{2}} |
| 305 |
% \put(32,30){\makebox(0,0)[bl]{1}} |
| 306 |
% ^^A convex hull |
| 307 |
% \put(0,0){\line(1,0){20}} |
| 308 |
% \put(20,0){\line(1,3){10}} |
| 309 |
% \put(30,30){\line(-1,0){20}} |
| 310 |
% \put(10,30){\line(-1,-3){10}} |
| 311 |
% ^^A control polygon |
| 312 |
% ^^A \put(0,0){\vector(1,1){30}} |
| 313 |
% ^^A \put(30,30){\vector(-1,0){20}} |
| 314 |
% ^^A \put(10,30){\vector(1,-3){10}} |
| 315 |
% \thicklines |
| 316 |
% ^^A \cbezier(0,0)(30,30)(10,30)(20,0) |
| 317 |
% \thinlines |
| 318 |
% \end{picture}} |
| 319 |
% \put(45,45){\begin{picture}(30,30) |
| 320 |
% \thinlines |
| 321 |
% \put(0,0){\circle{2}} |
| 322 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 323 |
% \put(30,0){\circle{2}} |
| 324 |
% \put(32,0){\makebox(0,0)[bl]{1}} |
| 325 |
% \put(0,30){\circle{2}} |
| 326 |
% \put(2,30){\makebox(0,0)[bl]{2}} |
| 327 |
% \put(10,10){\circle{2}} |
| 328 |
% \put(8,10){\makebox(0,0)[br]{3}} |
| 329 |
% ^^A convex hull |
| 330 |
% \put(0,0){\line(1,0){30}} |
| 331 |
% \put(30,0){\line(-1,1){30}} |
| 332 |
% \put(0,30){\line(0,-1){30}} |
| 333 |
% ^^A control polygon |
| 334 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 335 |
% ^^A \put(30,0){\vector(-1,1){30}} |
| 336 |
% ^^A \put(0,30){\vector(1,-2){10}} |
| 337 |
% \thicklines |
| 338 |
% ^^A \cbezier(0,0)(30,0)(0,30)(10,10) |
| 339 |
% \thinlines |
| 340 |
% \end{picture}} |
| 341 |
% \end{picture} |
| 342 |
% \setlength{\unitlength}{1pt} |
| 343 |
% \caption{Four sets of points and their convex hulls} \label{fig:ch} |
| 344 |
% \end{figure} |
| 345 |
% |
| 346 |
% The equation for cubic Bezier curves is |
| 347 |
% \begin{equation} |
| 348 |
% p(t) - p_{0} = 3l_{10}t + 3(l_{21} - l_{10})t^{2} + (l_{30} - 3l_{21})t^{3} |
| 349 |
% \label{eq:cubic} |
| 350 |
% \end{equation} |
| 351 |
% |
| 352 |
% \DescribeMacro{\cbezier} |
| 353 |
% The command |
| 354 |
% |\cbezier[|\meta{N}|](|\meta{p0}|)(|\meta{p1}|)(|\meta{p2}|)(|\meta{p3}|)| |
| 355 |
% draws a cubic Bezier curve, as defined by \eqref{eq:cubic}, |
| 356 |
% from point \meta{p0} to point \meta{p3}, where \meta{p1} and \meta{p2} |
| 357 |
% are the intermediate points defining the control polygon. |
| 358 |
% |
| 359 |
% \begin{figure} |
| 360 |
% \centering |
| 361 |
% \setlength{\unitlength}{1mm} |
| 362 |
% \begin{picture}(70,80) |
| 363 |
% ^^A degree 3 |
| 364 |
% \put(0,5){\begin{picture}(30,30) |
| 365 |
% \thinlines |
| 366 |
% \put(0,0){\circle{2}} |
| 367 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 368 |
% \put(10,30){\circle{2}} |
| 369 |
% \put(8,30){\makebox(0,0)[br]{1}} |
| 370 |
% \put(20,0){\circle{2}} |
| 371 |
% \put(22,0){\makebox(0,0)[bl]{2}} |
| 372 |
% \put(30,30){\circle{2}} |
| 373 |
% \put(32,30){\makebox(0,0)[bl]{3}} |
| 374 |
% ^^A convex hull |
| 375 |
% ^^A \put(0,0){\line(1,0){20}} |
| 376 |
% ^^A \put(20,0){\line(1,3){10}} |
| 377 |
% ^^A \put(30,30){\line(-1,0){20}} |
| 378 |
% ^^A \put(10,30){\line(-1,-3){10}} |
| 379 |
% ^^A control polygon |
| 380 |
% \put(0,0){\vector(1,3){10}} |
| 381 |
% \put(10,30){\vector(1,-3){10}} |
| 382 |
% \put(20,0){\vector(1,3){10}} |
| 383 |
% \thicklines |
| 384 |
% \cbezier[30](0,0)(10,30)(20,0)(30,30) |
| 385 |
% \thinlines |
| 386 |
% \end{picture}} |
| 387 |
% ^^A degree 3 |
| 388 |
% \put(0,45){\begin{picture}(30,30) |
| 389 |
% \thinlines |
| 390 |
% \put(0,0){\circle{2}} |
| 391 |
% \put(-2,0){\makebox(0,0)[br]{0, 3}} |
| 392 |
% \put(30,0){\circle{2}} |
| 393 |
% \put(32,0){\makebox(0,0)[bl]{1}} |
| 394 |
% \put(0,30){\circle{2}} |
| 395 |
% \put(2,30){\makebox(0,0)[bl]{2}} |
| 396 |
% ^^A convex hull |
| 397 |
% ^^A \put(0,0){\line(1,0){30}} |
| 398 |
% ^^A \put(30,0){\line(-1,1){30}} |
| 399 |
% ^^A \put(0,30){\line(0,-1){30}} |
| 400 |
% ^^A control polygon |
| 401 |
% \put(0,0){\vector(1,0){30}} |
| 402 |
% \put(30,0){\vector(-1,1){30}} |
| 403 |
% \put(0,30){\vector(0,-1){30}} |
| 404 |
% \thicklines |
| 405 |
% \cbezier[30](0,0)(30,0)(0,30)(0,0) |
| 406 |
% \thinlines |
| 407 |
% \end{picture}} |
| 408 |
% ^^A degree 3 |
| 409 |
% \put(45,0){\begin{picture}(30,30) |
| 410 |
% \thinlines |
| 411 |
% \put(0,0){\circle{2}} |
| 412 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 413 |
% \put(10,30){\circle{2}} |
| 414 |
% \put(8,30){\makebox(0,0)[br]{2}} |
| 415 |
% \put(20,0){\circle{2}} |
| 416 |
% \put(22,0){\makebox(0,0)[bl]{3}} |
| 417 |
% \put(30,30){\circle{2}} |
| 418 |
% \put(32,30){\makebox(0,0)[bl]{1}} |
| 419 |
% ^^A convex hull |
| 420 |
% ^^A \put(0,0){\line(1,0){20}} |
| 421 |
% ^^A \put(20,0){\line(1,3){10}} |
| 422 |
% ^^A \put(30,30){\line(-1,0){20}} |
| 423 |
% ^^A \put(10,30){\line(-1,-3){10}} |
| 424 |
% ^^A control polygon |
| 425 |
% \put(0,0){\vector(1,1){30}} |
| 426 |
% \put(30,30){\vector(-1,0){20}} |
| 427 |
% \put(10,30){\vector(1,-3){10}} |
| 428 |
% \thicklines |
| 429 |
% \cbezier(0,0)(30,30)(10,30)(20,0) |
| 430 |
% \thinlines |
| 431 |
% \end{picture}} |
| 432 |
% \put(45,45){\begin{picture}(30,30) |
| 433 |
% \thinlines |
| 434 |
% \put(0,0){\circle{2}} |
| 435 |
% \put(-2,0){\makebox(0,0)[br]{0}} |
| 436 |
% \put(30,0){\circle{2}} |
| 437 |
% \put(32,0){\makebox(0,0)[bl]{1}} |
| 438 |
% \put(0,30){\circle{2}} |
| 439 |
% \put(2,30){\makebox(0,0)[bl]{2}} |
| 440 |
% \put(10,10){\circle{2}} |
| 441 |
% \put(8,10){\makebox(0,0)[br]{3}} |
| 442 |
% ^^A convex hull |
| 443 |
% ^^A \put(0,0){\line(1,0){30}} |
| 444 |
% ^^A \put(30,0){\line(-1,1){30}} |
| 445 |
% ^^A \put(0,30){\line(0,-1){30}} |
| 446 |
% ^^A control polygon |
| 447 |
% \put(0,0){\vector(1,0){30}} |
| 448 |
% \put(30,0){\vector(-1,1){30}} |
| 449 |
% \put(0,30){\vector(1,-2){10}} |
| 450 |
% \thicklines |
| 451 |
% \cbezier(0,0)(30,0)(0,30)(10,10) |
| 452 |
% \thinlines |
| 453 |
% \end{picture}} |
| 454 |
% \end{picture} |
| 455 |
% \setlength{\unitlength}{1pt} |
| 456 |
% \caption{Four sets of points, the cubic Bezier curves and their control |
| 457 |
% polygons. Left --- curves plotted with $N=30$; Right --- |
| 458 |
% curves plotted with $N=0$} \label{fig:cp} |
| 459 |
% \end{figure} |
| 460 |
% |
| 461 |
% Figure~\ref{fig:cp} shows four such cubic Bezier curves, their |
| 462 |
% defining points and their control polygons. These are the same points |
| 463 |
% that were used in figure~\ref{fig:ch} to illustrate convex hulls. It is |
| 464 |
% easy to verify that a cubic Bezier curve does indeed lie within the convex |
| 465 |
% hull of its defining points. The curves on the left of the figure were |
| 466 |
% specified with a value of 30 for the argument \meta{N}, while those |
| 467 |
% on the right had no value given for \meta{N} and thus were drawn with |
| 468 |
% the number of plotted points calculated by the drawing algorithm. |
| 469 |
% The actual drawing commands used were: |
| 470 |
% \begin{verbatim} |
| 471 |
% \cbezier[30](0,0)(10,30)(20,0)(30,30) |
| 472 |
% \cbezier[30](0,0)(30,0)(0,30)(0,0) |
| 473 |
% \cbezier(0,0)(30,30)(10,30)(20,0) |
| 474 |
% \cbezier(0,0)(30,0)(0,30)(10,10) |
| 475 |
% \end{verbatim} |
| 476 |
% Note that points are plotted along the curve at equidistant values of the |
| 477 |
% of the parameter $t$. However, as the relationship between the actual |
| 478 |
% distance in $(x,y)$ coordinate space is a non-linear function of $t$, |
| 479 |
% the seperation between the plotted points is non-uniform. |
| 480 |
% |
| 481 |
% The equation for a \emph{non-rational} quadratic Bezier curve is |
| 482 |
% \begin{equation} |
| 483 |
% p(t) - p_{0} = 2l_{10}t + (l_{20} - 2l_{10})t^{2} \label{eq:quad} |
| 484 |
% \end{equation} |
| 485 |
% Using standard \LaTeX{} this can be drawn by the |\qbezier| command. |
| 486 |
% There is another form of a quadratic Bezier curve called a \emph{rational} |
| 487 |
% quadratic Bezier curve. Its equation is |
| 488 |
% \begin{equation} |
| 489 |
% p(t) - p_{0} = \frac^^A |
| 490 |
% {2w_{1}l_{10}t + (w_{2}l_{20} - 2w_{1}l_{10})t^{2}}^^A |
| 491 |
% {w_{0} + 2\omega_{10}t + (\omega_{20} - \omega_{10})t^{2}} |
| 492 |
% \label{eq:rqfull} |
| 493 |
% \end{equation} |
| 494 |
% where the $w_{i}$ are the \emph{weights} corresponding to the |
| 495 |
% points $p_{i}$ and $\omega_{pq} = w_{p} - w_{q}$. The shape of a |
| 496 |
% non-rational curve can be changed by changing the positions of the defining |
| 497 |
% points. The shape of a rational curve can also be modified by changing |
| 498 |
% the values of the weights. A rational curve |
| 499 |
% has the same general properties, outlined above, as a non-rational curve |
| 500 |
% with the exception that the curve may lie outside the convex hull of the |
| 501 |
% control polygon. |
| 502 |
% |
| 503 |
% For the purposes at hand, we use a more restricted form of a |
| 504 |
% rational quadratic Bezier curve, obtained by putting |
| 505 |
% $W = w_{1}/w_{0}$ and then making |
| 506 |
% $w_{0} = w_{2} = 1$ in \eqref{eq:rqfull}. Performing these |
| 507 |
% substitutions we end up with |
| 508 |
% \begin{equation} |
| 509 |
% p(t) - p_{0} = \frac^^A |
| 510 |
% {2Wl_{10}t + (l_{20} - 2Wl_{10})t^{2}}^^A |
| 511 |
% {1 + 2(1 - W)t + 2(1 - W)t^{2}} |
| 512 |
% \label{eq:rqfinal} |
| 513 |
% \end{equation} |
| 514 |
% Note that when $W=1$, (\ref{eq:rqfinal}) reduces to \eqref{eq:quad} |
| 515 |
% and when $W=0$ it effectively reduces to \eqref{eq:lin3}. |
| 516 |
% |
| 517 |
% It turns out that a non-rational quadratic Bezier curve is an arc of |
| 518 |
% a parabola, which is one of the conic curves. All the other conic curves |
| 519 |
% can be represented by the rational quadratic Bezier curve described |
| 520 |
% by \eqref{eq:rqfinal} by suitable choices for the value of $W$. |
| 521 |
% From now on, we will call $W$ the \emph{weight} of the rational quadratic |
| 522 |
% Bezier curve. Table~\ref{tab:rq} lists the value, or value range, |
| 523 |
% of $W$ for the various forms of the conic curve.\footnote{We do not deal |
| 524 |
% with the degenerate cases.} For the case of a circle, $\beta$ is the |
| 525 |
% angle between the lines $l_{10} = (p_{1} - p_{0})$ and |
| 526 |
% $l_{20} = (p_{2} - p_{0})$, as shown in figure~\ref{fig:beta}. |
| 527 |
% |
| 528 |
% \begin{table} |
| 529 |
% \centering |
| 530 |
% \caption{Conic forms of the rational quadratic Bezier curve} \label{tab:rq} |
| 531 |
% \begin{tabular}{lc} \hline |
| 532 |
% Conic form & Weight ($W$) \\ \hline |
| 533 |
% Hyperbola & $\|W\| > 1$ \\ |
| 534 |
% Parabola & $\|W\| = 1$ \\ |
| 535 |
% Ellipse & $0 < \|W\| < 1$ \\ |
| 536 |
% Circle & $\|l_{10}\| = \|l_{21}\|$ and $W = \cos \beta$ \\ |
| 537 |
% Straight line & $W = 0$ \\ \hline |
| 538 |
% \end{tabular} |
| 539 |
% \end{table} |
| 540 |
% |
| 541 |
% |
| 542 |
% \begin{figure} |
| 543 |
% \centering |
| 544 |
% \setlength{\unitlength}{1mm} |
| 545 |
% \begin{picture}(30,40) |
| 546 |
% \put(0,5){\begin{picture}(30,30) |
| 547 |
% \put(0,0){\circle{2}} |
| 548 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 549 |
% \put(15,30){\circle{2}} |
| 550 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 551 |
% \put(30,0){\circle{2}} |
| 552 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 553 |
% ^^A polygon |
| 554 |
% \thicklines |
| 555 |
% \put(15,30){\vector(-1,-2){15}} |
| 556 |
% \put(0,0){\vector(1,0){30}} |
| 557 |
% ^^A dashed line from 0 to 2 |
| 558 |
% ^^A \lbezier(15,30)(30,0) |
| 559 |
% \lbezier[50](15,30)(30,0) |
| 560 |
% \thinlines |
| 561 |
% \put(15,26){\makebox(0,0){$\beta$}} |
| 562 |
% \end{picture}} |
| 563 |
% \end{picture} |
| 564 |
% \setlength{\unitlength}{1pt} |
| 565 |
% \caption{The angle $\beta$} \label{fig:beta} |
| 566 |
% \end{figure} |
| 567 |
% |
| 568 |
% \DescribeMacro{\rqbezier} |
| 569 |
% The command |
| 570 |
% |\rqbezier[|\meta{N}|](|\meta{p0}|)(|\meta{p1}|)(|\meta{p2}|)(|\meta{W}|)| |
| 571 |
% draws a rational quadratic Bezier curve from \meta{p0} to \meta{p2} with |
| 572 |
% weight \meta{W}, according to \eqref{eq:rqfinal}. As in the |
| 573 |
% other Bezier commands, \meta{N} is optional and controls the number |
| 574 |
% of plotted points along the curve. Figure~\ref{fig:qrb} shows several |
| 575 |
% rational quadratic curves, all with the same control polygon but with |
| 576 |
% differing values for the weight $W$. The code is: |
| 577 |
% \begin{verbatim} |
| 578 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 579 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 580 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 581 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 582 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 583 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 584 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 585 |
% \end{verbatim} |
| 586 |
% When $W > 1$ the curve is pulled toward the point $p_{1}$. Conversely, |
| 587 |
% when $W < 1$ the curve is pushed away from the point $p_{1}$. In all |
| 588 |
% cases, though, the curve starts and stops at $p_{0}$ and $p_{2}$ |
| 589 |
% respectively. |
| 590 |
% |
| 591 |
% \begin{figure} |
| 592 |
% \centering |
| 593 |
% \setlength{\unitlength}{1mm} |
| 594 |
% \begin{picture}(30,40) |
| 595 |
% \put(0,5){\begin{picture}(30,30) |
| 596 |
% \put(0,0){\circle{2}} |
| 597 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 598 |
% \put(15,30){\circle{2}} |
| 599 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 600 |
% \put(30,0){\circle{2}} |
| 601 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 602 |
% ^^A polygon |
| 603 |
% \thinlines |
| 604 |
% \put(15,30){\vector(-1,-2){15}} |
| 605 |
% \put(0,0){\vector(1,0){30}} |
| 606 |
% ^^A dashed line from 0 to 2 |
| 607 |
% ^^A \lbezier(15,30)(30,0) |
| 608 |
% ^^A \lbezier[50](15,30)(30,0) |
| 609 |
% ^^A \thinlines |
| 610 |
% ^^A \put(15,26){\makebox(0,0){$\beta$}} |
| 611 |
% \thicklines |
| 612 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 613 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 614 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 615 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 616 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 617 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 618 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 619 |
% \end{picture}} |
| 620 |
% \end{picture} |
| 621 |
% \setlength{\unitlength}{1pt} |
| 622 |
% \caption{The effect of weight variation ($W \geq 0$) |
| 623 |
% on rational quadratic Bezier curves |
| 624 |
% (\texttt{weightscale = \theweightscale} (the default)) } |
| 625 |
% \label{fig:qrb} |
| 626 |
% \end{figure} |
| 627 |
% |
| 628 |
% Like the case of the cubic curve, points are plotted at equidistant |
| 629 |
% values of the parameter $t$. The relationship between parameter value |
| 630 |
% and coordinate positions in the rational case are highly non-linear. |
| 631 |
% Thus the distance between the plotted points can vary quite remarkably. |
| 632 |
% This is an inherent disadvantage with this type of curve. The user's remedy |
| 633 |
% is to increase the number of points to be plotted, but this can lead to |
| 634 |
% \TeX{} running out of memory, not to mention the increased time to |
| 635 |
% generate the drawing. |
| 636 |
% |
| 637 |
% \DescribeMacro{\setweightscale} |
| 638 |
% \DescribeMacro{\resetweightscale} |
| 639 |
% Because of the way in which \TeX{} performs arithmetic, and especially |
| 640 |
% division, it |
| 641 |
% is necessary to perform some scaling operations on the divisor when |
| 642 |
% evaluating \eqref{eq:rqfinal}. The optimum value for the |
| 643 |
% scaling is a complex function of the weight and the size and orientation |
| 644 |
% of the control polygon. The algorithm uses a heuristic approach to |
| 645 |
% calculate a `good' value but is not always successful. The |
| 646 |
% |\setweightscale{|\meta{number}|}| command can be used to specify |
| 647 |
% a scale factor. \meta{number} must be a positive integer. The |
| 648 |
% |\resetweightscale| command resets the scale factor to its default |
| 649 |
% value, which is currently 10000 (ten thousand). |
| 650 |
% |
| 651 |
% \begin{figure} |
| 652 |
% \centering |
| 653 |
% \setlength{\unitlength}{1mm} |
| 654 |
% \begin{picture}(70,80) |
| 655 |
% \put(0,5){\begin{picture}(30,30) |
| 656 |
% \put(0,0){\circle{2}} |
| 657 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 658 |
% \put(15,30){\circle{2}} |
| 659 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 660 |
% \put(30,0){\circle{2}} |
| 661 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 662 |
% ^^A polygon |
| 663 |
% \thinlines |
| 664 |
% ^^A \put(15,30){\vector(-1,-2){15}} |
| 665 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 666 |
% ^^A dashed line from 0 to 2 |
| 667 |
% ^^A \lbezier(15,30)(30,0) |
| 668 |
% ^^A \lbezier[50](15,30)(30,0) |
| 669 |
% ^^A \thinlines |
| 670 |
% ^^A \put(15,26){\makebox(0,0){$\beta$}} |
| 671 |
% \thicklines |
| 672 |
% \setweightscale{100} |
| 673 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 674 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 675 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 676 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 677 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 678 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 679 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 680 |
% \put(15,-5){\makebox(0,0)[b]{\texttt{weightscale = \theweightscale}}} |
| 681 |
% \end{picture}} |
| 682 |
% \put(0,45){\begin{picture}(30,30) |
| 683 |
% \put(0,0){\circle{2}} |
| 684 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 685 |
% \put(15,30){\circle{2}} |
| 686 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 687 |
% \put(30,0){\circle{2}} |
| 688 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 689 |
% ^^A polygon |
| 690 |
% \thinlines |
| 691 |
% ^^A \put(15,30){\vector(-1,-2){15}} |
| 692 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 693 |
% ^^A dashed line from 0 to 2 |
| 694 |
% ^^A \lbezier(15,30)(30,0) |
| 695 |
% ^^A \lbezier[50](15,30)(30,0) |
| 696 |
% ^^A \thinlines |
| 697 |
% ^^A \put(15,26){\makebox(0,0){$\beta$}} |
| 698 |
% \thicklines |
| 699 |
% \setweightscale{1000} |
| 700 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 701 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 702 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 703 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 704 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 705 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 706 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 707 |
% \put(15,-5){\makebox(0,0)[b]{\texttt{weightscale = \theweightscale}}} |
| 708 |
% \end{picture}} |
| 709 |
% \put(45,5){\begin{picture}(30,30) |
| 710 |
% \put(0,0){\circle{2}} |
| 711 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 712 |
% \put(15,30){\circle{2}} |
| 713 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 714 |
% \put(30,0){\circle{2}} |
| 715 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 716 |
% ^^A polygon |
| 717 |
% \thinlines |
| 718 |
% ^^A \put(15,30){\vector(-1,-2){15}} |
| 719 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 720 |
% ^^A dashed line from 0 to 2 |
| 721 |
% ^^A \lbezier(15,30)(30,0) |
| 722 |
% ^^A \lbezier[50](15,30)(30,0) |
| 723 |
% ^^A \thinlines |
| 724 |
% ^^A \put(15,26){\makebox(0,0){$\beta$}} |
| 725 |
% \thicklines |
| 726 |
% \resetweightscale |
| 727 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 728 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 729 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 730 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 731 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 732 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 733 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 734 |
% \put(15,-5){\makebox(0,0)[b]{\texttt{weightscale = \theweightscale}}} |
| 735 |
% \end{picture}} |
| 736 |
% \put(45,45){\begin{picture}(30,30) |
| 737 |
% \put(0,0){\circle{2}} |
| 738 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 739 |
% \put(15,30){\circle{2}} |
| 740 |
% \put(13,30){\makebox(0,0)[br]{0}} |
| 741 |
% \put(30,0){\circle{2}} |
| 742 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 743 |
% ^^A polygon |
| 744 |
% \thinlines |
| 745 |
% ^^A \put(15,30){\vector(-1,-2){15}} |
| 746 |
% ^^A \put(0,0){\vector(1,0){30}} |
| 747 |
% ^^A dashed line from 0 to 2 |
| 748 |
% ^^A \lbezier(15,30)(30,0) |
| 749 |
% ^^A \lbezier[50](15,30)(30,0) |
| 750 |
% ^^A \thinlines |
| 751 |
% ^^A \put(15,26){\makebox(0,0){$\beta$}} |
| 752 |
% \thicklines |
| 753 |
% \setweightscale{100000} |
| 754 |
% \rqbezier[100](15,30)(0,0)(30,0)(4) |
| 755 |
% \rqbezier[100](15,30)(0,0)(30,0)(2) |
| 756 |
% \rqbezier(15,30)(0,0)(30,0)(1) |
| 757 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.75) |
| 758 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.5) |
| 759 |
% \rqbezier[100](15,30)(0,0)(30,0)(0.25) |
| 760 |
% \rqbezier(15,30)(0,0)(30,0)(0) |
| 761 |
% \put(15,-5){\makebox(0,0)[b]{\texttt{weightscale = \theweightscale}}} |
| 762 |
% \end{picture}} |
| 763 |
% \end{picture} |
| 764 |
% \setlength{\unitlength}{1pt} |
| 765 |
% \resetweightscale |
| 766 |
% \caption{The effect of \texttt{weightscale} on the drawing |
| 767 |
% of rational quadratic Bezier curves} |
| 768 |
% \label{fig:qrw} |
| 769 |
% \end{figure} |
| 770 |
% |
| 771 |
% Figure~\ref{fig:qrw} illustrates the effect on changing the |
| 772 |
% \texttt{weightscale} used for drawing the same curves as shown |
| 773 |
% in figure~\ref{fig:qrb}. Note that the \texttt{weightscale} |
| 774 |
% has no effect when |
| 775 |
% $W = 1$ or $W = 0$ as in these cases the curves are drawn using the |
| 776 |
% algorithms for the |\qbezier| and |\lbezier| commands respectively. |
| 777 |
% |
| 778 |
% It is obvious that some choices give very poorly formed curves. In |
| 779 |
% other cases the curves may be poorly formed but do result in interesting |
| 780 |
% cross-stitch like patterns. |
| 781 |
% |
| 782 |
% Table~\ref{tab:rq} indicates that it is possible to draw circular |
| 783 |
% arcs using a rational quadratic Bezier curves. The two legs of the |
| 784 |
% control polygon define the tangents to the curve at the |
| 785 |
% end points. |
| 786 |
% Therefore, to draw a circular arc the two legs must be equal in length. |
| 787 |
% That is, the convex hull is an isosceles triangle. In the special case |
| 788 |
% when the convex hull forms an equilateral triangle, the required |
| 789 |
% weight ($\cos \beta$, see figure~\ref{fig:beta}) for drawing a circular |
| 790 |
% arc is $\cos \beta = 0.5$. Further, |
| 791 |
% for any given control polygon the the curves drawn with weights of |
| 792 |
% $\pm W$ are complementary. That is, the curve with weight $-W$ is |
| 793 |
% the `remainder' of the curve drawn with weight $W$. Thus, we have a |
| 794 |
% simple means of drawing a complete circle, as shown in figure~\ref{fig:qrc}. |
| 795 |
% The plotting commands of interest were: |
| 796 |
% \begin{verbatim} |
| 797 |
% \lbezier[25](0,0)(15,26) |
| 798 |
% \lbezier[25](0,0)(30,0) |
| 799 |
% \setweightscale{50000} |
| 800 |
% \rqbezier[100](15,26)(0,0)(30,0)(0.5) |
| 801 |
% \rqbezier[200](15,26)(0,0)(30,0)(-0.5) |
| 802 |
% \resetweightscale |
| 803 |
% \end{verbatim} |
| 804 |
% where the |\lbezier| drawing commands were used to draw the dotted outline |
| 805 |
% of the control polygon. |
| 806 |
% |
| 807 |
% \begin{figure} |
| 808 |
% \centering |
| 809 |
% \setlength{\unitlength}{1mm} |
| 810 |
% \setweightscale{50000} |
| 811 |
% \begin{picture}(60,62) |
| 812 |
% \put(0,5){\begin{picture}(30,30) |
| 813 |
% \thinlines |
| 814 |
% \put(0,0){\circle{2}} |
| 815 |
% \put(-2,0){\makebox(0,0)[br]{1}} |
| 816 |
% \put(15,26){\circle{2}} |
| 817 |
% \put(13,26){\makebox(0,0)[br]{0}} |
| 818 |
% \put(30,0){\circle{2}} |
| 819 |
% \put(32,0){\makebox(0,0)[bl]{2}} |
| 820 |
% ^^A polygon |
| 821 |
% \lbezier[25](0,0)(15,26) |
| 822 |
% \lbezier[25](0,0)(30,0) |
| 823 |
% ^^A RQBEZIER |
| 824 |
% \thicklines |
| 825 |
% \rqbezier[100](15,26)(0,0)(30,0)(0.5) |
| 826 |
% \rqbezier[200](15,26)(0,0)(30,0)(-0.5) |
| 827 |
% \end{picture}} |
| 828 |
% \end{picture} |
| 829 |
% \setlength{\unitlength}{1pt} |
| 830 |
% \caption{Rational quadratics with weights of $\pm 0.5$ and an equilateral |
| 831 |
% triangular convex hull |
| 832 |
% (\texttt{weightscale = \theweightscale}) } \label{fig:qrc} |
| 833 |
% \resetweightscale |
| 834 |
% \end{figure} |
| 835 |
% |
| 836 |
% A more robust picture of the same circle is shown in |
| 837 |
% figure~\ref{fig:qr3c} where the complete circle is pieced together from |
| 838 |
% three non-complementary circular arcs. The drawing commands of interest |
| 839 |
% were |
| 840 |
% \begin{verbatim} |
| 841 |
% \rqbezier[100](15,26)(0,0)(30,0)(0.5) |
| 842 |
% \rqbezier[100](30,0)(60,0)(45,26)(0.5) |
| 843 |
% \rqbezier[100](45,26)(30,52)(15,26)(0.5) |
| 844 |
% \end{verbatim} |
| 845 |
% |
| 846 |
% \begin{figure} |
| 847 |
% \centering |
| 848 |
% \setlength{\unitlength}{1mm} |
| 849 |
% \begin{picture}(60,62) |
| 850 |
% \put(0,5){\begin{picture}(60,52) |
| 851 |
% \thinlines |
| 852 |
% \put(0,0){\circle{2}} |
| 853 |
% ^^A \put(-2,0){\makebox(0,0)[br]{1}} |
| 854 |
% \put(15,26){\circle{2}} |
| 855 |
% ^^A \put(13,30){\makebox(0,0)[br]{0}} |
| 856 |
% \put(30,0){\circle{2}} |
| 857 |
% ^^A \put(32,0){\makebox(0,0)[bl]{2}} |
| 858 |
% \put(60,0){\circle{2}} |
| 859 |
% \put(45,26){\circle{2}} |
| 860 |
% \put(30,52){\circle{2}} |
| 861 |
% ^^A polygon |
| 862 |
% \lbezier[50](0,0)(30,52) |
| 863 |
% \lbezier[50](30,52)(60,0) |
| 864 |
% \lbezier[50](0,0)(60,0) |
| 865 |
% ^^A RQBEZIER |
| 866 |
% \thicklines |
| 867 |
% \rqbezier[100](15,26)(0,0)(30,0)(0.5) |
| 868 |
% \rqbezier[100](30,0)(60,0)(45,26)(0.5) |
| 869 |
% \rqbezier[100](45,26)(30,52)(15,26)(0.5) |
| 870 |
% \end{picture}} |
| 871 |
% \end{picture} |
| 872 |
% \setlength{\unitlength}{1pt} |
| 873 |
% \caption{Three rational quadratics with weights of $0.5$ |
| 874 |
% (\texttt{weightscale = \theweightscale}) } |
| 875 |
% \label{fig:qr3c} |
| 876 |
% \end{figure} |
| 877 |
% |
| 878 |
% The astute reader will have realised that the divisor in |
| 879 |
% \eqref{eq:rqfinal} can go to zero, and can even be negative. |
| 880 |
% This has interesting consequences, both when trying to do computer |
| 881 |
% arithmetic, and also on the the kind of curve that results. Essentially, |
| 882 |
% the curve tends to $\infty$ as $W \rightarrow +0$. At $W = -0$ the curve |
| 883 |
% is at $-\infty$ and then it tends to $-0$ as $W \rightarrow -\infty$. |
| 884 |
% We will get a curve point at $\infty$ whenever $W = -1$ and a `negative' |
| 885 |
% curve for $W < -1$. |
| 886 |
% |
| 887 |
% \begin{figure} |
| 888 |
% \centering |
| 889 |
% \setlength{\unitlength}{1mm} |
| 890 |
% \begin{picture}(60,60) |
| 891 |
% \put(30,20){\begin{picture}(30,20) |
| 892 |
% \thinlines |
| 893 |
% \put(0,10){\circle{2}} |
| 894 |
% \put(30,0){\circle{2}} |
| 895 |
% \put(30,20){\circle{2}} |
| 896 |
% ^^A polygon |
| 897 |
% \lbezier[25](30,20)(0,10) |
| 898 |
% \lbezier[25](0,10)(30,0) |
| 899 |
% ^^A RQBEZIER |
| 900 |
% \thicklines |
| 901 |
% \rqbezier[100](30,20)(0,10)(30,0)(2) |
| 902 |
% \rqbezier[100](30,20)(0,10)(30,0)(-2) |
| 903 |
% \end{picture}} |
| 904 |
% \end{picture} |
| 905 |
% \setlength{\unitlength}{1pt} |
| 906 |
% \caption{A rational quadratic that has gone negative; weights of $\pm 2$ |
| 907 |
% (\texttt{weightscale = \theweightscale}) } |
| 908 |
% \label{fig:neg} |
| 909 |
% \end{figure} |
| 910 |
% |
| 911 |
% This effect is shown in figure~\ref{fig:neg} which draws the two branches |
| 912 |
% of a hyperbola. The basic code for the illustration was |
| 913 |
% \begin{verbatim} |
| 914 |
% \lbezier[25](30,20)(0,10) |
| 915 |
% \lbezier[25](0,10)(30,0) |
| 916 |
% \rqbezier[100](30,20)(0,10)(30,0)(2) |
| 917 |
% \rqbezier[100](30,20)(0,10)(30,0)(-2) |
| 918 |
% \end{verbatim} |
| 919 |
% where the control polygon was drawn using the |\lbezier| commands. |
| 920 |
% |
| 921 |
% |
| 922 |
% \section{The \Lpack{bez123} package implementation} \label{sec:bez} |
| 923 |
% |
| 924 |
% \LaTeX{} provides a facility for drawing quadratic Bezier curves. |
| 925 |
% This package provides additional facilities for drawing linear, |
| 926 |
% rational quadratic, and cubic Bezier curves. |
| 927 |
% |
| 928 |
% |
| 929 |
% Announce the name and version of the package, which requires \LaTeXe. |
| 930 |
% \begin{macrocode} |
| 931 |
%<*bez> |
| 932 |
\NeedsTeXFormat{LaTeX2e} |
| 933 |
\ProvidesPackage{bez123}[2009/09/02 v1.1b Bezier curves] |
| 934 |
% \end{macrocode} |
| 935 |
% \changes{v1.1}{1998/10/14}{Added call to include multiply package} |
| 936 |
% The package also requires the \Lpack{multiply} package. |
| 937 |
% \begin{macrocode} |
| 938 |
\RequirePackage{multiply}[2009/09/02] |
| 939 |
%</bez> |
| 940 |
% \end{macrocode} |
| 941 |
% |
| 942 |
% |
| 943 |
% \subsection{Arithmetic in \TeX} |
| 944 |
% |
| 945 |
% All arithmetic in \TeX{} is based on integer arithmetic, with a |
| 946 |
% maximum integer value of $M = \number\maxdimen$. For example, |
| 947 |
% \setcounter{weightscale}{8}\makeatletter\divide\c@weightscale by 3\makeatother |
| 948 |
% $8/3 = \theweightscale$, |
| 949 |
% \setcounter{weightscale}{9}\makeatletter\divide\c@weightscale by 3\makeatother |
| 950 |
% $9/3 = \theweightscale$, and |
| 951 |
% \setcounter{weightscale}{10}\makeatletter\divide\c@weightscale by 3\makeatother |
| 952 |
% $10/3 = \theweightscale$. |
| 953 |
% In other words, division always reduces the absolute value of the dividend, |
| 954 |
% and also possibly truncates the value. One consequence of this is that the |
| 955 |
% ordering of multiplication and division is important. For instance, |
| 956 |
% \setcounter{weightscale}{8}\makeatletter\multiply\c@weightscale by 3 \divide\c@weightscale by 3\makeatother |
| 957 |
% $(8 \times 3)/3 = \theweightscale$ but |
| 958 |
% \setcounter{weightscale}{8}\makeatletter\divide\c@weightscale by 3 \multiply\c@weightscale by 3\makeatother |
| 959 |
% $(8/3) \times 3 = \theweightscale$! |
| 960 |
% Thus, in arithmetic calculations involving both multiplication and |
| 961 |
% division, the dividend should be maximised and the divisor minimised, |
| 962 |
% with multiplication preceeding division; also remembering that there |
| 963 |
% is a limit on the size of an integer. To avoid multiplication overflow |
| 964 |
% when calculating say, $a \times b$, we must ensure that |
| 965 |
% $\|a\| \leq \|M/b\|$. |
| 966 |
% |
| 967 |
% When calculating polynomials, such as that in \eqref{eq:gen}, |
| 968 |
% we use a technique called Horner's schema, which is also known as nested |
| 969 |
% multiplication. A general cubic equation, for example, can be written as: |
| 970 |
% \begin{equation} |
| 971 |
% p(t) - a_{0} = t(a_{1} + t(a_{2} + ta_{3})) \label{eq:horn} |
| 972 |
% \end{equation} |
| 973 |
% The following pseudo-code shows one way to implement Horner's schema for |
| 974 |
% plotting $N$ points in the interval $0 \leq t \leq 1$ |
| 975 |
% of \eqref{eq:horn} using integer arithmetic. |
| 976 |
% \begin{verbatim} |
| 977 |
% procedure plot_cubic(a0, a1, a2, a3:vector; N:integer); |
| 978 |
% local p:vector; end_local; |
| 979 |
% a3 := a3/N; |
| 980 |
% repeat i := 0 to N by 1; |
| 981 |
% p := a3*i; |
| 982 |
% p := p + a2; p := p/N; p := p*i; |
| 983 |
% p := p + a1; p := p/N; p := p*i; |
| 984 |
% draw(p + a0); |
| 985 |
% end_repeat; |
| 986 |
% return; |
| 987 |
% end_procedure; |
| 988 |
% \end{verbatim} |
| 989 |
% We use the above algorithm, with suitable modifications according to the |
| 990 |
% degree of the polynomial, for plotting the points along Bezier curves. |
| 991 |
% |
| 992 |
% \subsection{Linear Bezier curves} |
| 993 |
% |
| 994 |
% \begin{macrocode} |
| 995 |
%<*bez> |
| 996 |
% \end{macrocode} |
| 997 |
% |
| 998 |
% As a linear curve is simpler than a quadratic curve there is no |
| 999 |
% need to declare extra variables from those used in the kernel by the |
| 1000 |
% |\qbezier| macro. |
| 1001 |
% |
| 1002 |
% \begin{macro}{\lbezier} |
| 1003 |
% The user command to draw a linear Bezier curve represented by |
| 1004 |
% \eqref{eq:lin3}. The form of the command is:\\ |
| 1005 |
% |\lbezier[|\meta{N}|]{(|\meta{p0}|)(|\meta{p1}|)| \\ |
| 1006 |
% where \meta{pN} is the comma seperated X and Y coordinate values of |
| 1007 |
% point \textit{pN}. |
| 1008 |
% |
| 1009 |
% \begin{macrocode} |
| 1010 |
\newcommand{\lbezier}[2][0]{\@lbez{#1}#2} |
| 1011 |
% \end{macrocode} |
| 1012 |
% \end{macro} |
| 1013 |
% |
| 1014 |
% \begin{macro}{\@lbez} |
| 1015 |
% The drawing macro. |
| 1016 |
% \begin{macrocode} |
| 1017 |
\gdef\@lbez#1(#2,#3)(#4,#5){% |
| 1018 |
%%%%\def\lbezier#1(#2,#3)(#4,#5){% |
| 1019 |
\ifnum #1<\@ne |
| 1020 |
% \end{macrocode} |
| 1021 |
% When the number of plotting points are not given, then we calculate |
| 1022 |
% how many are needed. First determine the X distance between the end points. |
| 1023 |
% \begin{macrocode} |
| 1024 |
\@ovxx = #4\unitlength |
| 1025 |
\advance\@ovxx by -#2\unitlength |
| 1026 |
\ifdim \@ovxx < \z@ |
| 1027 |
\@ovxx = -\@ovxx |
| 1028 |
\fi |
| 1029 |
% \end{macrocode} |
| 1030 |
% Similarly calculate the Y distance. |
| 1031 |
% \begin{macrocode} |
| 1032 |
\@ovyy = #5\unitlength |
| 1033 |
\advance\@ovyy by -#3\unitlength |
| 1034 |
\ifdim \@ovyy < \z@ |
| 1035 |
\@ovyy = -\@ovyy |
| 1036 |
\fi |
| 1037 |
% \end{macrocode} |
| 1038 |
% Temporarily store the maximum distance in |\@multicnt|. |
| 1039 |
% \begin{macrocode} |
| 1040 |
\ifdim \@ovxx > \@ovyy |
| 1041 |
\@multicnt = \@ovxx |
| 1042 |
\else |
| 1043 |
\@multicnt = \@ovyy |
| 1044 |
\fi |
| 1045 |
% \end{macrocode} |
| 1046 |
% We use a small square as the visual representation of a point. |
| 1047 |
% Calculate the number of points required to give 50\% overlap of adjacent |
| 1048 |
% squares, making |
| 1049 |
% sure that it doesn't exceed the limit. Store the result in |\@multicnt|. |
| 1050 |
% \begin{macrocode} |
| 1051 |
\@ovxx = 0.5\@halfwidth |
| 1052 |
\divide\@multicnt by \@ovxx |
| 1053 |
\ifnum \qbeziermax < \@multicnt |
| 1054 |
\@multicnt = \qbeziermax\relax |
| 1055 |
\fi |
| 1056 |
\else |
| 1057 |
% \end{macrocode} |
| 1058 |
% The number of points is given. |
| 1059 |
% \begin{macrocode} |
| 1060 |
\@multicnt = #1\relax |
| 1061 |
\fi |
| 1062 |
% \end{macrocode} |
| 1063 |
% |
| 1064 |
% Now we can prepare the constants for the plotting loop. |
| 1065 |
% \begin{macrocode} |
| 1066 |
\@tempcnta = \@multicnt |
| 1067 |
\advance\@tempcnta by \@ne |
| 1068 |
\@ovdx = #4\unitlength |
| 1069 |
\advance\@ovdx by -#2\unitlength |
| 1070 |
\divide\@ovdx by \@multicnt |
| 1071 |
\@ovdy = #5\unitlength |
| 1072 |
\advance\@ovdy by -#3\unitlength |
| 1073 |
\divide\@ovdy by \@multicnt |
| 1074 |
% \end{macrocode} |
| 1075 |
% The next bit of code defines the size of the square representing a point. |
| 1076 |
% \begin{macrocode} |
| 1077 |
\setbox\@tempboxa\hbox{\vrule \@height\@halfwidth |
| 1078 |
\@depth \@halfwidth |
| 1079 |
\@width \@wholewidth}% |
| 1080 |
% \end{macrocode} |
| 1081 |
% Start the plot at the first point. |
| 1082 |
% \begin{macrocode} |
| 1083 |
\put(#2,#3){% |
| 1084 |
\count@ = \z@ |
| 1085 |
\@whilenum{\count@ < \@tempcnta}\do |
| 1086 |
% \end{macrocode} |
| 1087 |
% Evaluate the polynomial (simple in this case) using Horner's schema. |
| 1088 |
% \begin{macrocode} |
| 1089 |
{\@xdim = \count@\@ovdx |
| 1090 |
\@ydim = \count@\@ovdy |
| 1091 |
% \end{macrocode} |
| 1092 |
% Plot this point. |
| 1093 |
% \begin{macrocode} |
| 1094 |
\raise \@ydim |
| 1095 |
\hb@xt@\z@{\kern\@xdim |
| 1096 |
\unhcopy\@tempboxa\hss}% |
| 1097 |
\advance\count@\@ne}}% |
| 1098 |
% \end{macrocode} |
| 1099 |
% The end of the definition of |\@lbez|. |
| 1100 |
% \begin{macrocode} |
| 1101 |
} |
| 1102 |
% \end{macrocode} |
| 1103 |
% \end{macro} |
| 1104 |
% |
| 1105 |
% \subsection{Cubic Bezier curves} |
| 1106 |
% |
| 1107 |
% As cubic curves are more complex than quadratic curves we need some |
| 1108 |
% extra variables. |
| 1109 |
% \begin{macro}{\@wxc} |
| 1110 |
% \begin{macro}{\@wyc} |
| 1111 |
% Lengths. |
| 1112 |
% \begin{macrocode} |
| 1113 |
\newlength{\@wxc} |
| 1114 |
\newlength{\@wyc} |
| 1115 |
% \end{macrocode} |
| 1116 |
% \end{macro} |
| 1117 |
% \end{macro} |
| 1118 |
% |
| 1119 |
% \begin{macro}{\cbezier} |
| 1120 |
% The user command for drawing a cubic Bezier curve as represented |
| 1121 |
% by \eqref{eq:cubic}. It is called as: \\ |
| 1122 |
% |\cbezier[|\meta{N}|](|\meta{p0}|)(|\meta{p1}|)(|\meta{p2}|)(|\meta{p3}|)|. |
| 1123 |
% |
| 1124 |
% \begin{macrocode} |
| 1125 |
\newcommand{\cbezier}[2][0]{\@cbez{#1}#2} |
| 1126 |
% \end{macrocode} |
| 1127 |
% \end{macro} |
| 1128 |
% |
| 1129 |
% \begin{macro}{\@cbez} |
| 1130 |
% The drawing macro for cubic Bezier curves. |
| 1131 |
% \begin{macrocode} |
| 1132 |
\gdef\@cbez#1(#2,#3)(#4,#5)(#6,#7)(#8,#9){% |
| 1133 |
\ifnum #1<\@ne |
| 1134 |
% \end{macrocode} |
| 1135 |
% We have to calculate the number of plotting points required. We will |
| 1136 |
% use the maximum of the box enclosing the convex hull as a measure. |
| 1137 |
% First do the X value, using |\@ovxx| to store the maximum X coordinate |
| 1138 |
% and |\@ovdx| the minimum. |
| 1139 |
% \begin{macrocode} |
| 1140 |
\@ovxx = #2\unitlength |
| 1141 |
\@ovdx = \@ovxx |
| 1142 |
\@ovdy = #4\unitlength |
| 1143 |
\ifdim \@ovdy > \@ovxx |
| 1144 |
\@ovxx = \@ovdy |
| 1145 |
\fi |
| 1146 |
\ifdim \@ovdy < \@ovdx |
| 1147 |
\@ovdx = \@ovdy |
| 1148 |
\fi |
| 1149 |
\@ovdy = #6\unitlength |
| 1150 |
\ifdim \@ovdy > \@ovxx |
| 1151 |
\@ovxx = \@ovdy |
| 1152 |
\fi |
| 1153 |
\ifdim \@ovdy < \@ovdx |
| 1154 |
\@ovdx = \@ovdy |
| 1155 |
\fi |
| 1156 |
\@ovdy = #8\unitlength |
| 1157 |
\ifdim \@ovdy > \@ovxx |
| 1158 |
\@ovxx = \@ovdy |
| 1159 |
\fi |
| 1160 |
\ifdim \@ovdy < \@ovdx |
| 1161 |
\@ovdx = \@ovdy |
| 1162 |
\fi |
| 1163 |
% \end{macrocode} |
| 1164 |
% Store the maximum X in |\@ovxx|. |
| 1165 |
% \begin{macrocode} |
| 1166 |
\advance\@ovxx by -\@ovdx |
| 1167 |
% \end{macrocode} |
| 1168 |
% Repeat the process for the maximum Y value, finally storing |
| 1169 |
% this in |\@ovyy|. |
| 1170 |
% \begin{macrocode} |
| 1171 |
\@ovyy = #3\unitlength |
| 1172 |
\@ovdy = \@ovyy |
| 1173 |
\@ovdx = #5\unitlength |
| 1174 |
\ifdim \@ovdx > \@ovyy |
| 1175 |
\@ovyy = \@ovdx |
| 1176 |
\fi |
| 1177 |
\ifdim \@ovdx < \@ovdy |
| 1178 |
\@ovdy = \@ovdx |
| 1179 |
\fi |
| 1180 |
\@ovdx = #7\unitlength |
| 1181 |
\ifdim \@ovdx > \@ovyy |
| 1182 |
\@ovyy = \@ovdx |
| 1183 |
\fi |
| 1184 |
\ifdim \@ovdx < \@ovdy |
| 1185 |
\@ovdy = \@ovdx |
| 1186 |
\fi |
| 1187 |
\@ovdx = #9\unitlength |
| 1188 |
\ifdim \@ovdx > \@ovyy |
| 1189 |
\@ovyy = \@ovdx |
| 1190 |
\fi |
| 1191 |
\ifdim \@ovdx < \@ovdy |
| 1192 |
\@ovdy = \@ovdx |
| 1193 |
\fi |
| 1194 |
\advance\@ovyy by -\@ovdy |
| 1195 |
% \end{macrocode} |
| 1196 |
% Temporarily store the max of X and Y in |\@multicnt|. |
| 1197 |
% \begin{macrocode} |
| 1198 |
\ifdim \@ovxx > \@ovyy |
| 1199 |
\@multicnt = \@ovxx |
| 1200 |
\else |
| 1201 |
\@multicnt = \@ovyy |
| 1202 |
\fi |
| 1203 |
% \end{macrocode} |
| 1204 |
% Calculate the number of points required to give 50\% overlap, making |
| 1205 |
% sure that it doesn't exceed the limit. Store the number of points in |
| 1206 |
% |\@multicnt|. |
| 1207 |
% \begin{macrocode} |
| 1208 |
\@ovxx = 0.5\@halfwidth |
| 1209 |
\divide\@multicnt by \@ovxx |
| 1210 |
\ifnum \qbeziermax < \@multicnt |
| 1211 |
\@multicnt = \qbeziermax\relax |
| 1212 |
\fi |
| 1213 |
\else |
| 1214 |
% \end{macrocode} |
| 1215 |
% The number of points is given. |
| 1216 |
% \begin{macrocode} |
| 1217 |
\@multicnt = #1\relax |
| 1218 |
\fi |
| 1219 |
% \end{macrocode} |
| 1220 |
% |
| 1221 |
% Now we can prepare the constants for the plotting loop. First the control |
| 1222 |
% counts. |
| 1223 |
% \begin{macrocode} |
| 1224 |
\@tempcnta = \@multicnt |
| 1225 |
\advance\@tempcnta by \@ne |
| 1226 |
% \end{macrocode} |
| 1227 |
% Then the cubic coefficients, firstly for X. |
| 1228 |
% \begin{macrocode} |
| 1229 |
\@ovdx = #4\unitlength \advance\@ovdx by -#2\unitlength |
| 1230 |
\@ovxx = #6\unitlength \advance\@ovxx by -\@ovdx |
| 1231 |
\multiply\@ovdx by \thr@@ |
| 1232 |
\advance\@ovxx by -#4\unitlength \multiply\@ovxx by \thr@@ |
| 1233 |
\@wxc = #4\unitlength \advance\@wxc by -#6\unitlength |
| 1234 |
\multiply\@wxc by \thr@@ \advance\@wxc by #8\unitlength |
| 1235 |
\advance\@wxc by -#2\unitlength \divide\@wxc by \@multicnt |
| 1236 |
% \end{macrocode} |
| 1237 |
% And similarly for Y. |
| 1238 |
% \begin{macrocode} |
| 1239 |
\@ovdy = #5\unitlength \advance\@ovdy by -#3\unitlength |
| 1240 |
\@ovyy = #7\unitlength \advance\@ovyy by -\@ovdy |
| 1241 |
\multiply\@ovdy by \thr@@ |
| 1242 |
\advance\@ovyy by -#5\unitlength \multiply\@ovyy by \thr@@ |
| 1243 |
\@wyc = #5\unitlength \advance\@wyc by -#7\unitlength |
| 1244 |
\multiply\@wyc by \thr@@ \advance\@wyc by #9\unitlength |
| 1245 |
\advance\@wyc by -#3\unitlength \divide\@wyc by \@multicnt |
| 1246 |
% \end{macrocode} |
| 1247 |
% Set up the plotting box. |
| 1248 |
% \begin{macrocode} |
| 1249 |
\setbox\@tempboxa\hbox{\vrule \@height\@halfwidth |
| 1250 |
\@depth \@halfwidth |
| 1251 |
\@width \@wholewidth}% |
| 1252 |
% \end{macrocode} |
| 1253 |
% Start the plot at the first point. |
| 1254 |
% \begin{macrocode} |
| 1255 |
\put(#2,#3){% |
| 1256 |
\count@ = \z@ |
| 1257 |
\@whilenum{\count@ < \@tempcnta}\do |
| 1258 |
{\@xdim = \count@\@wxc |
| 1259 |
\advance\@xdim by \@ovxx |
| 1260 |
\divide\@xdim by \@multicnt |
| 1261 |
\multiply\@xdim by \count@ |
| 1262 |
\advance\@xdim by \@ovdx |
| 1263 |
\divide\@xdim by \@multicnt |
| 1264 |
\multiply\@xdim by \count@ |
| 1265 |
\@ydim = \count@\@wyc |
| 1266 |
\advance\@ydim by \@ovyy |
| 1267 |
\divide\@ydim by \@multicnt |
| 1268 |
\multiply\@ydim by \count@ |
| 1269 |
\advance\@ydim by \@ovdy |
| 1270 |
\divide\@ydim by \@multicnt |
| 1271 |
\multiply\@ydim by \count@ |
| 1272 |
% \end{macrocode} |
| 1273 |
% Plot the point. |
| 1274 |
% \begin{macrocode} |
| 1275 |
\raise \@ydim |
| 1276 |
\hb@xt@\z@{\kern\@xdim |
| 1277 |
\unhcopy\@tempboxa\hss}% |
| 1278 |
\advance\count@\@ne}}% |
| 1279 |
% \end{macrocode} |
| 1280 |
% The end of the definition of |\@cbez|. |
| 1281 |
% \begin{macrocode} |
| 1282 |
} |
| 1283 |
% \end{macrocode} |
| 1284 |
% \end{macro} |
| 1285 |
% |
| 1286 |
% |
| 1287 |
% \subsection{Quadratic rational Bezier curve} |
| 1288 |
% |
| 1289 |
% This is the most complex of the Bezier curves that we deal with. |
| 1290 |
% We need yet more variables. |
| 1291 |
% |
| 1292 |
% \begin{macro}{\@ww} |
| 1293 |
% \begin{macro}{\@wwa} |
| 1294 |
% \begin{macro}{\@wwb} |
| 1295 |
% \begin{macro}{\@wwo} |
| 1296 |
% \begin{macro}{\@wwi} |
| 1297 |
% Variables for the weight calculations. |
| 1298 |
% \begin{macrocode} |
| 1299 |
\newlength{\@ww} |
| 1300 |
\newlength{\@wwa} |
| 1301 |
\newlength{\@wwb} |
| 1302 |
\newlength{\@wwo} |
| 1303 |
\newlength{\@wwi} |
| 1304 |
% \end{macrocode} |
| 1305 |
% \end{macro} |
| 1306 |
% \end{macro} |
| 1307 |
% \end{macro} |
| 1308 |
% \end{macro} |
| 1309 |
% \end{macro} |
| 1310 |
% |
| 1311 |
% \begin{macro}{\c@@pntscale} |
| 1312 |
% Scale factor for points. |
| 1313 |
% \begin{macrocode} |
| 1314 |
\newcounter{@pntscale} |
| 1315 |
% \end{macrocode} |
| 1316 |
% \end{macro} |
| 1317 |
% |
| 1318 |
% \begin{macro}{\c@weightscale} |
| 1319 |
% Scale factor for divisor. |
| 1320 |
% \begin{macrocode} |
| 1321 |
\newcounter{weightscale} |
| 1322 |
% \end{macrocode} |
| 1323 |
% \end{macro} |
| 1324 |
% |
| 1325 |
% \begin{macro}{\botscale} |
| 1326 |
% Scale factor for bottom weights. |
| 1327 |
% \begin{macrocode} |
| 1328 |
\newlength{\botscale} |
| 1329 |
% \end{macrocode} |
| 1330 |
% \end{macro} |
| 1331 |
% |
| 1332 |
% \begin{macro}{\setweightscale} |
| 1333 |
% User level command |\setweightscale{|\meta{number}|}| for setting the |
| 1334 |
% divisor scaling. |
| 1335 |
% \begin{macrocode} |
| 1336 |
\newcommand{\setweightscale}[1]{\setcounter{weightscale}{#1}} |
| 1337 |
% \end{macrocode} |
| 1338 |
% \end{macro} |
| 1339 |
% |
| 1340 |
% \begin{macro}{\resetweightscale} |
| 1341 |
% User level command for setting the divisor scaling to its default |
| 1342 |
% value ($10^{4}$). We also ensure that the scaling is set to this value. |
| 1343 |
% \begin{macrocode} |
| 1344 |
\newcommand{\resetweightscale}{\setcounter{weightscale}{10000}} |
| 1345 |
\resetweightscale |
| 1346 |
% \end{macrocode} |
| 1347 |
% \end{macro} |
| 1348 |
% |
| 1349 |
% \begin{macro}{\rqbezier} |
| 1350 |
% The user level command for drawing a rational quadratic Bezier curve |
| 1351 |
% as represented by \eqref{eq:rqfinal}. The form of the command is \\ |
| 1352 |
% |\rqbezier[|\meta{N}|](|\meta{p0}|)(|\meta{p1}|)(|\meta{p2}|)(|\meta{W}|)| \\ |
| 1353 |
% where the arguments are as per the other Bezier drawing commands, but with |
| 1354 |
% the final argument being the weight. |
| 1355 |
% |
| 1356 |
% \begin{macrocode} |
| 1357 |
\newcommand{\rqbezier}[2][0]{\@rqbez{#1}#2} |
| 1358 |
% \end{macrocode} |
| 1359 |
% \end{macro} |
| 1360 |
% |
| 1361 |
% |
| 1362 |
% \begin{macro}{\@rqbez} |
| 1363 |
% The drawing macro for a rational quadratic Bezier curve. If the weight |
| 1364 |
% is such that the curve is either rational quadratic ($W = 1$) or |
| 1365 |
% linear ($W = 0$), we use the simpler drawing macro. |
| 1366 |
% \begin{macrocode} |
| 1367 |
\gdef\@rqbez#1(#2,#3)(#4,#5)(#6,#7)(#8){% |
| 1368 |
\@ovxx = #8\unitlength |
| 1369 |
\ifdim\@ovxx = \unitlength |
| 1370 |
\PackageWarning{bez123}{Rational quadratic denerates to quadratic} |
| 1371 |
\qbezier[#1](#2,#3)(#4,#5)(#6,#7) |
| 1372 |
\else |
| 1373 |
\ifdim\@ovxx = \z@ |
| 1374 |
\PackageWarning{bez123}{Rational quadratic degenerates to linear} |
| 1375 |
\lbezier[#1](#2,#3)(#6,#7) |
| 1376 |
\else |
| 1377 |
% \end{macrocode} |
| 1378 |
% Calculate the maximum length of the control polygon's bounding box. |
| 1379 |
% Store the result in |\@wwi|. |
| 1380 |
% \begin{macrocode} |
| 1381 |
\@ovxx = #4\unitlength |
| 1382 |
\advance\@ovxx by -#2\unitlength |
| 1383 |
\ifdim \@ovxx < \z@ |
| 1384 |
\@ovxx = -\@ovxx |
| 1385 |
\fi |
| 1386 |
\@ovdx = #6\unitlength |
| 1387 |
\advance\@ovdx by -#4\unitlength |
| 1388 |
\ifdim \@ovdx < \z@ |
| 1389 |
\@ovdx = -\@ovdx |
| 1390 |
\fi |
| 1391 |
\ifdim \@ovxx < \@ovdx |
| 1392 |
\@ovxx = \@ovdx |
| 1393 |
\fi |
| 1394 |
\@ovyy = #5\unitlength |
| 1395 |
\advance\@ovyy by -#3\unitlength |
| 1396 |
\ifdim \@ovyy < \z@ |
| 1397 |
\@ovyy = -\@ovyy |
| 1398 |
\fi |
| 1399 |
\@ovdy = #7\unitlength |
| 1400 |
\advance\@ovdy by -#5\unitlength |
| 1401 |
\ifdim \@ovdy < \z@ |
| 1402 |
\@ovdy = -\@ovdy |
| 1403 |
\fi |
| 1404 |
\ifdim \@ovyy < \@ovdy |
| 1405 |
\@ovyy = \@ovdy |
| 1406 |
\fi |
| 1407 |
\ifdim \@ovxx > \@ovyy |
| 1408 |
\@multicnt = \@ovxx |
| 1409 |
\else |
| 1410 |
\@multicnt = \@ovyy |
| 1411 |
\fi |
| 1412 |
\@wwi = \@multicnt sp |
| 1413 |
% \end{macrocode} |
| 1414 |
% Now determine the number of points to be plotted. |
| 1415 |
% \begin{macrocode} |
| 1416 |
\ifnum #1<\@ne |
| 1417 |
\@ovxx = 0.5\@halfwidth |
| 1418 |
\divide\@multicnt by \@ovxx |
| 1419 |
\ifnum\qbeziermax < \@multicnt |
| 1420 |
\@multicnt = \qbeziermax\relax |
| 1421 |
\fi |
| 1422 |
\else |
| 1423 |
% \end{macrocode} |
| 1424 |
% Number of points is a given. |
| 1425 |
% \begin{macrocode} |
| 1426 |
\@multicnt = #1\relax |
| 1427 |
\fi |
| 1428 |
% \end{macrocode} |
| 1429 |
% We are going to plot the curve in two halves in an attempt to reduce |
| 1430 |
% roundoff problems. At |
| 1431 |
% a minimum this should at least make a symmetrical curve look symmetric |
| 1432 |
% about its mid point. |
| 1433 |
% \begin{macrocode} |
| 1434 |
\@tempcnta = \@multicnt |
| 1435 |
\advance\@tempcnta by \@ne |
| 1436 |
\divide\@tempcnta by \tw@ |
| 1437 |
\advance\@tempcnta by \@ne |
| 1438 |
% \end{macrocode} |
| 1439 |
% We now have to deal with a possible multiplication overflow problem due |
| 1440 |
% to multiplication by the weight. In \eqref{eq:rqfinal} the potentially |
| 1441 |
% largest term is the coefficient of $t^{2}$ (i.e., $(l_{20}-2Wl_{10})$). |
| 1442 |
% The maximum length likely to be encountered is, say, 10 inches for a |
| 1443 |
% drawing on either A4 or US letterpaper. This is approximately |
| 1444 |
% $4.8\times10^{8}$sp. Doing a little arithmetic, and remembering that the |
| 1445 |
% maximum length in \TeX{} is $M = \number\maxdimen$sp, it means that we must |
| 1446 |
% have $\|W\| \leq 1$ to prevent overflow. However, a typical range for |
| 1447 |
% $W$ is $-10 \leq W \leq 10$. Therefore we might have to do some scaling. |
| 1448 |
% Being pessimistic, we'll assume that $l_{20} = - l_{10}$ and that $l_{10}$ |
| 1449 |
% is the largest dimension in the drawing. To prevent overflow we then have to |
| 1450 |
% meet the condition $\|W\| \leq (M - l_{20})/2l_{20}$, where all lengths are |
| 1451 |
% positive. We will use |\c@@pntscale| as a scale factor on $W$ to meet this |
| 1452 |
% condition. Earlier we set |\@wwi| to be the positive value of the largest |
| 1453 |
% dimension in the drawing. |
| 1454 |
% |
| 1455 |
% Set the distance scale factor. First evaluating the test condition. |
| 1456 |
% \begin{macrocode} |
| 1457 |
\@wwo = \maxdimen |
| 1458 |
\advance\@wwo by -\@wwi |
| 1459 |
\divide\@wwo by \tw@ |
| 1460 |
\divide\@wwo by \@wwi |
| 1461 |
% \end{macrocode} |
| 1462 |
% Now perform the check and set the scale factor. We have to get a positive |
| 1463 |
% integer value for $W$ as it may be a fraction. Actually, we only need to |
| 1464 |
% be concerned if $\|W\| > 1$. |
| 1465 |
% \begin{macrocode} |
| 1466 |
\@wwi = 10sp |
| 1467 |
\@wwi = #8\@wwi |
| 1468 |
\ifdim\@wwi < \z@ |
| 1469 |
\@wwi = -\@wwi |
| 1470 |
\fi |
| 1471 |
\divide\@wwi by 10\relax |
| 1472 |
\ifdim\@wwi < \@wwo |
| 1473 |
\c@@pntscale = \@ne |
| 1474 |
\else |
| 1475 |
\divide\@wwi by \tw@ |
| 1476 |
\ifdim\@wwi < \@wwo |
| 1477 |
\c@@pntscale = \tw@ |
| 1478 |
\else |
| 1479 |
\divide\@wwi by \tw@ |
| 1480 |
\ifdim\@wwi < \@wwo |
| 1481 |
\c@@pntscale = 4\relax |
| 1482 |
\else |
| 1483 |
\divide\@wwi by \tw@ |
| 1484 |
\ifdim\@wwi < \@wwo |
| 1485 |
\c@@pntscale = 8\relax |
| 1486 |
\else |
| 1487 |
\c@@pntscale = 16\relax |
| 1488 |
\fi |
| 1489 |
\fi |
| 1490 |
\fi |
| 1491 |
\fi |
| 1492 |
% \end{macrocode} |
| 1493 |
% Calculate the constants for the top line of the function. |
| 1494 |
% \begin{macrocode} |
| 1495 |
\@ovxx = #4\unitlength \advance\@ovxx by -#2\unitlength |
| 1496 |
\multiply\@ovxx by \tw@ |
| 1497 |
\divide\@ovxx by \c@@pntscale |
| 1498 |
\@ovdx = #8\@ovxx |
| 1499 |
\@ovxx = #6\unitlength \advance\@ovxx by -#2\unitlength |
| 1500 |
\divide\@ovxx by \c@@pntscale |
| 1501 |
\advance\@ovxx by -\@ovdx |
| 1502 |
\divide\@ovxx by \@multicnt |
| 1503 |
\@ovyy = #5\unitlength \advance\@ovyy by -#3\unitlength |
| 1504 |
\multiply\@ovyy by \tw@ |
| 1505 |
\divide\@ovyy by \c@@pntscale |
| 1506 |
\@ovdy = #8\@ovyy |
| 1507 |
\@ovyy = #7\unitlength \advance\@ovyy by -#3\unitlength |
| 1508 |
\divide\@ovyy by \c@@pntscale |
| 1509 |
\advance\@ovyy by -\@ovdy |
| 1510 |
\divide\@ovyy by \@multicnt |
| 1511 |
% \end{macrocode} |
| 1512 |
% Now the constants for the bottom line. We also need to do some scaling |
| 1513 |
% here. This scaling can be set by the user. |
| 1514 |
% \begin{macrocode} |
| 1515 |
\setlength{\botscale}{\c@weightscale sp} |
| 1516 |
\@wwo = \botscale |
| 1517 |
\@wwi = #8\@wwo |
| 1518 |
\@wwa = \@wwo \advance\@wwa by -\@wwi |
| 1519 |
\multiply\@wwa by \tw@ |
| 1520 |
\@wwb = \@wwa |
| 1521 |
\divide\@wwb by \@multicnt |
| 1522 |
% \end{macrocode} |
| 1523 |
% Prepare for the drawing. |
| 1524 |
% \begin{macrocode} |
| 1525 |
\@wwi = \botscale |
| 1526 |
\setbox\@tempboxa\hbox{\vrule \@height\@halfwidth |
| 1527 |
\@depth \@halfwidth |
| 1528 |
\@width \@wholewidth}% |
| 1529 |
% \end{macrocode} |
| 1530 |
% Draw the first half of the curve. |
| 1531 |
% \begin{macrocode} |
| 1532 |
\put(#2,#3){% |
| 1533 |
\count@ = \z@ |
| 1534 |
\@whilenum{\count@ < \@tempcnta}\do |
| 1535 |
{\@xdim = \count@\@ovxx |
| 1536 |
\advance\@xdim by \@ovdx |
| 1537 |
\divide\@xdim by \@multicnt |
| 1538 |
\multiply\@xdim by \count@ |
| 1539 |
\@ydim = \count@\@ovyy |
| 1540 |
\advance\@ydim by \@ovdy |
| 1541 |
\divide\@ydim by \@multicnt |
| 1542 |
\multiply\@ydim by \count@ |
| 1543 |
\@ww = \count@\@wwb |
| 1544 |
\advance\@ww by -\@wwa |
| 1545 |
\divide\@ww by \@multicnt |
| 1546 |
\multiply\@ww by \count@ |
| 1547 |
\advance\@ww by \@wwo |
| 1548 |
\divide\@ww by \c@@pntscale |
| 1549 |
\ifdim\@ww = \z@ |
| 1550 |
% \end{macrocode} |
| 1551 |
% We are about to divide by |\@ww| which is zero. Treat |\@ww| as unity. |
| 1552 |
% \begin{macrocode} |
| 1553 |
\else |
| 1554 |
\divide\@xdim by \@ww |
| 1555 |
\divide\@ydim by \@ww |
| 1556 |
\fi |
| 1557 |
% \end{macrocode} |
| 1558 |
% For reasons I don't understand, the \% signs at the end of the next few |
| 1559 |
% lines are important! |
| 1560 |
% \begin{macrocode} |
| 1561 |
\multnooverflow{\@xdim}{\botscale}% |
| 1562 |
\multnooverflow{\@ydim}{\botscale}% |
| 1563 |
\raise \@ydim |
| 1564 |
\hb@xt@\z@{\kern\@xdim |
| 1565 |
\unhcopy\@tempboxa\hss}% |
| 1566 |
\advance\count@\@ne}} |
| 1567 |
% \end{macrocode} |
| 1568 |
% |
| 1569 |
% We now repeat the above process for plotting the second half of the |
| 1570 |
% curve, starting at the end point. |
| 1571 |
% |
| 1572 |
% Calculate the constants for the top line of the function. |
| 1573 |
% \begin{macrocode} |
| 1574 |
\@ovxx = #4\unitlength \advance\@ovxx by -#6\unitlength |
| 1575 |
\multiply\@ovxx by \tw@ |
| 1576 |
\divide\@ovxx by \c@@pntscale |
| 1577 |
\@ovdx = #8\@ovxx |
| 1578 |
\@ovxx = #2\unitlength \advance\@ovxx by -#6\unitlength |
| 1579 |
\divide\@ovxx by \c@@pntscale |
| 1580 |
\advance\@ovxx by -\@ovdx |
| 1581 |
\divide\@ovxx by \@multicnt |
| 1582 |
\@ovyy = #5\unitlength \advance\@ovyy by -#7\unitlength |
| 1583 |
\multiply\@ovyy by \tw@ |
| 1584 |
\divide\@ovyy by \c@@pntscale |
| 1585 |
\@ovdy = #8\@ovyy |
| 1586 |
\@ovyy = #3\unitlength \advance\@ovyy by -#7\unitlength |
| 1587 |
\divide\@ovyy by \c@@pntscale |
| 1588 |
\advance\@ovyy by -\@ovdy |
| 1589 |
\divide\@ovyy by \@multicnt |
| 1590 |
% \end{macrocode} |
| 1591 |
% The constants for the bottom line are the same as before as the function |
| 1592 |
% is symmetric. Similarly we don't need to recalculate the size of the |
| 1593 |
% rule box. |
| 1594 |
% |
| 1595 |
% Draw the second half of the curve. |
| 1596 |
% \begin{macrocode} |
| 1597 |
\put(#6,#7){% |
| 1598 |
\count@ = \z@ |
| 1599 |
\@whilenum{\count@ < \@tempcnta}\do |
| 1600 |
{\@xdim = \count@\@ovxx |
| 1601 |
\advance\@xdim by \@ovdx |
| 1602 |
\divide\@xdim by \@multicnt |
| 1603 |
\multiply\@xdim by \count@ |
| 1604 |
\@ydim = \count@\@ovyy |
| 1605 |
\advance\@ydim by \@ovdy |
| 1606 |
\divide\@ydim by \@multicnt |
| 1607 |
\multiply\@ydim by \count@ |
| 1608 |
\@ww = \count@\@wwb |
| 1609 |
\advance\@ww by -\@wwa |
| 1610 |
\divide\@ww by \@multicnt |
| 1611 |
\multiply\@ww by \count@ |
| 1612 |
\advance\@ww by \@wwo |
| 1613 |
\divide\@ww by \c@@pntscale |
| 1614 |
\ifnum\@ww = \z@ |
| 1615 |
% \end{macrocode} |
| 1616 |
% We are about to divide by |\@ww| which is zero. Treat |\@ww| as unity. |
| 1617 |
% \begin{macrocode} |
| 1618 |
\else |
| 1619 |
\divide\@xdim by \@ww |
| 1620 |
\divide\@ydim by \@ww |
| 1621 |
\fi |
| 1622 |
% \end{macrocode} |
| 1623 |
% For reasons I don't understand, the \% signs at the end of the next few |
| 1624 |
% lines are important! |
| 1625 |
% \begin{macrocode} |
| 1626 |
\multnooverflow{\@xdim}{\botscale}% |
| 1627 |
\multnooverflow{\@ydim}{\botscale}% |
| 1628 |
\raise \@ydim |
| 1629 |
\hb@xt@\z@{\kern\@xdim |
| 1630 |
\unhcopy\@tempboxa\hss}% |
| 1631 |
\advance\count@\@ne}} |
| 1632 |
% \end{macrocode} |
| 1633 |
% End of definition of |\@rqbez|. |
| 1634 |
% \begin{macrocode} |
| 1635 |
\fi\fi} |
| 1636 |
% \end{macrocode} |
| 1637 |
% \end{macro} |
| 1638 |
% |
| 1639 |
% The end of this package. |
| 1640 |
% \begin{macrocode} |
| 1641 |
%</bez> |
| 1642 |
% \end{macrocode} |
| 1643 |
% |
| 1644 |
% \section{Multiplication without overflow: The \Lpack{multiply} package} \label{sec:mnoflow} |
| 1645 |
% |
| 1646 |
% \TeX{} provides for integer arithmetic, subject to an upper limit |
| 1647 |
% given by |\maxdim|. For at least the \Lpack{bez123} package we need to |
| 1648 |
% be able to multiply without overflow. |
| 1649 |
% |
| 1650 |
% Announce the name of the package. |
| 1651 |
% \changes{v1.1}{1998/10/14}{Put multnooverflow into a seperate package} |
| 1652 |
% \begin{macrocode} |
| 1653 |
%<*mult> |
| 1654 |
\ProvidesPackage{multiply}[1998/10/14 v1.1 Multiplication of lengths without overflow] |
| 1655 |
% \end{macrocode} |
| 1656 |
% |
| 1657 |
% \begin{macro}{\n@fl@wa} |
| 1658 |
% \begin{macro}{\n@fl@wb} |
| 1659 |
% \begin{macro}{\n@fl@wc} |
| 1660 |
% \begin{macro}{\ifch@nge} |
| 1661 |
% We need three length variables for this function. |
| 1662 |
% We also need a boolean flag for dealing with negative numbers. |
| 1663 |
% \begin{macrocode} |
| 1664 |
\newlength{\n@fl@wa} |
| 1665 |
\newlength{\n@fl@wb} |
| 1666 |
\newlength{\n@fl@wc} |
| 1667 |
\newif\ifch@nge |
| 1668 |
% \end{macrocode} |
| 1669 |
% \end{macro} |
| 1670 |
% \end{macro} |
| 1671 |
% \end{macro} |
| 1672 |
% \end{macro} |
| 1673 |
% |
| 1674 |
% \begin{macro}{\multnooverflow} |
| 1675 |
% |
| 1676 |
% The routine |\multnooverflow{|\meta{a}|}{|\meta{b}|}| sets |
| 1677 |
% $a$ to the minimum of $ab$ and |\maxdimen|, preserving signs. \meta{a} |
| 1678 |
% must be a length; it must not be a number literal. |
| 1679 |
% |
| 1680 |
% \begin{macrocode} |
| 1681 |
\newcommand{\multnooverflow}[2]{% |
| 1682 |
\n@fl@wa = #1\relax% |
| 1683 |
\n@fl@wb = #2\relax% |
| 1684 |
\ch@ngefalse% |
| 1685 |
% \end{macrocode} |
| 1686 |
% Easy if $-1 \leq b \leq 1$. |
| 1687 |
% \begin{macrocode} |
| 1688 |
\ifnum\n@fl@wb = \@ne% |
| 1689 |
\else% |
| 1690 |
\ifnum\n@fl@wb = \z@% |
| 1691 |
\n@fl@wa = \z@% |
| 1692 |
\else% |
| 1693 |
\ifnum\n@fl@wb = \m@ne% |
| 1694 |
\ch@ngetrue% |
| 1695 |
\else% |
| 1696 |
% \end{macrocode} |
| 1697 |
% Also easy if $-1 \leq a \leq 1$. |
| 1698 |
% \begin{macrocode} |
| 1699 |
\ifnum\n@fl@wa = \z@% |
| 1700 |
\else% |
| 1701 |
\ifnum\n@fl@wa = \@ne% |
| 1702 |
\n@fl@wa = \n@fl@wb% |
| 1703 |
\else% |
| 1704 |
\ifnum\n@fl@wa = \m@ne% |
| 1705 |
\n@fl@wa = -\n@fl@wb% |
| 1706 |
\else% |
| 1707 |
% \end{macrocode} |
| 1708 |
% We have to check for potential overflow. First make sure that we deal |
| 1709 |
% only with positive values. |
| 1710 |
% \begin{macrocode} |
| 1711 |
\ifnum\n@fl@wa < \z@% |
| 1712 |
\ch@ngetrue% |
| 1713 |
\n@fl@wa = -\n@fl@wa% |
| 1714 |
\fi% |
| 1715 |
\ifnum\n@fl@wb < \z@% |
| 1716 |
\n@fl@wb = -\n@fl@wb% |
| 1717 |
\ifch@nge% |
| 1718 |
\ch@ngefalse% |
| 1719 |
\else% |
| 1720 |
\ch@ngetrue% |
| 1721 |
\fi% |
| 1722 |
\fi% |
| 1723 |
% \end{macrocode} |
| 1724 |
% Check for overflow. |
| 1725 |
% \begin{macrocode} |
| 1726 |
\n@fl@wc = \maxdimen% |
| 1727 |
\divide\n@fl@wc by \n@fl@wb% |
| 1728 |
\advance\n@fl@wc by -1sp% \m@ne |
| 1729 |
\ifnum\n@fl@wa > \n@fl@wc% |
| 1730 |
% \end{macrocode} |
| 1731 |
% We have overflow. Set the multiplication result to |\maxdimen|. |
| 1732 |
% \begin{macrocode} |
| 1733 |
\n@fl@wa = \maxdimen% |
| 1734 |
\PackageWarning{multiply}{Multiplication overflow}% |
| 1735 |
\else% |
| 1736 |
% \end{macrocode} |
| 1737 |
% It is safe to do the multiplication. |
| 1738 |
% \begin{macrocode} |
| 1739 |
\multiply\n@fl@wa by \n@fl@wb% |
| 1740 |
\fi% |
| 1741 |
\fi% |
| 1742 |
\fi% |
| 1743 |
\fi% |
| 1744 |
\fi% |
| 1745 |
\fi% |
| 1746 |
\fi% |
| 1747 |
% \end{macrocode} |
| 1748 |
% The result of $ab$ is in |\n@fl@wa|. Adjust the sign if necessary. |
| 1749 |
% \begin{macrocode} |
| 1750 |
\ifch@nge% |
| 1751 |
\n@fl@wa = -\n@fl@wa% |
| 1752 |
\fi% |
| 1753 |
% \end{macrocode} |
| 1754 |
% Return the result in the first argument variable. |
| 1755 |
% \begin{macrocode} |
| 1756 |
#1 = \n@fl@wa% |
| 1757 |
} |
| 1758 |
% \end{macrocode} |
| 1759 |
% \end{macro} |
| 1760 |
% |
| 1761 |
% |
| 1762 |
% The end of this package. |
| 1763 |
% \begin{macrocode} |
| 1764 |
%</mult> |
| 1765 |
% \end{macrocode} |
| 1766 |
% |
| 1767 |
% |
| 1768 |
% \bibliographystyle{alpha} |
| 1769 |
% |
| 1770 |
% \begin{thebibliography}{GMS94} |
| 1771 |
% |
| 1772 |
% \bibitem[Far90]{FARIN90} |
| 1773 |
% Gerald Farin. |
| 1774 |
% \newblock {\em Curves and Surfaces for Computer Aided Geometric Design --- A |
| 1775 |
% Practical Guide}. |
| 1776 |
% \newblock Academic Press, Inc., second edition, 1990. |
| 1777 |
% |
| 1778 |
% \bibitem[FP81]{FandP} |
| 1779 |
% I.~D. Faux and M.~J. Pratt. |
| 1780 |
% \newblock {\em Computational Geometry for Design and Manufacture}. |
| 1781 |
% \newblock Ellis Horwood, 1981. |
| 1782 |
% |
| 1783 |
% \bibitem[GMS94]{GOOSSENS94} |
| 1784 |
% Michel Goossens, Frank Mittelbach, and Alexander Samarin. |
| 1785 |
% \newblock {\em The LaTeX Companion}. |
| 1786 |
% \newblock Addison-Wesley Publishing Company, 1994. |
| 1787 |
% |
| 1788 |
% \bibitem[Lam94]{LAMPORT94} |
| 1789 |
% Leslie Lamport. |
| 1790 |
% \newblock {\em LaTeX: A Document Preparation System}. |
| 1791 |
% \newblock Addison-Wesley Publishing Company, second edition, 1994. |
| 1792 |
% |
| 1793 |
% \bibitem[Mor85]{MORTENSON85} |
| 1794 |
% Michael~E. Mortenson. |
| 1795 |
% \newblock {\em Geometric Modeling}. |
| 1796 |
% \newblock John Wiley \& Sons, Inc., 1985. |
| 1797 |
% |
| 1798 |
% \bibitem[Wil96]{PRW96i} |
| 1799 |
% Peter~R. Wilson. |
| 1800 |
% \newblock {\em {LaTeX for standards: The LaTeX package files user manual}}. |
| 1801 |
% \newblock NIST Report NISTIR, June 1996. |
| 1802 |
% |
| 1803 |
% \end{thebibliography} |
| 1804 |
% |
| 1805 |
% |
| 1806 |
% \Finale |
| 1807 |
% \PrintIndex |
| 1808 |
% |
| 1809 |
\endinput |
| 1810 |
|
| 1811 |
%% \CharacterTable |
| 1812 |
%% {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 |
| 1813 |
%% 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 |
| 1814 |
%% Digits \0\1\2\3\4\5\6\7\8\9 |
| 1815 |
%% Exclamation \! Double quote \" Hash (number) \# |
| 1816 |
%% Dollar \$ Percent \% Ampersand \& |
| 1817 |
%% Acute accent \' Left paren \( Right paren \) |
| 1818 |
%% Asterisk \* Plus \+ Comma \, |
| 1819 |
%% Minus \- Point \. Solidus \/ |
| 1820 |
%% Colon \: Semicolon \; Less than \< |
| 1821 |
%% Equals \= Greater than \> Question mark \? |
| 1822 |
%% Commercial at \@ Left bracket \[ Backslash \\ |
| 1823 |
%% Right bracket \] Circumflex \^ Underscore \_ |
| 1824 |
%% Grave accent \` Left brace \{ Vertical bar \| |
| 1825 |
%% Right brace \} Tilde \~} |
| 1826 |
|
| 1827 |
|