1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
|
\section{Memory and Speed considerations}
\label{sec:pgfplots:optimization}
\PGFPlots\ can typeset plots with several thousand points if memory limits of \TeX\ are configured properly. Its runtime is roughly proportional to the number of input points\footnote{In fact, the runtime is pseudo--linear: starting with about $100{,}000$ points, it will become quadratic. This limitation applies to the path length of \PGF\ paths as well. Furthermore, the linear runtime is not possible yet for stacked plots.}.
\message{ATTENTION: I am now about to typeset really huge pictures. You will need to ENLARGE YOUR TeX MEMORY CAPACITIES if this fails...}%
\begin{codeexample}[]
\begin{tikzpicture}
\begin{axis}[
enlargelimits=0.01,
title style={yshift=5pt},
title=Scatter plot with $2250$ points]
\addplot[blue,
mark=*,only marks,mark options={scale=0.3}]
file[skip first]
{plotdata/pgfplots_scatterdata3.dat};
\end{axis}
\end{tikzpicture}
\end{codeexample}
\begin{codeexample}[]
\begin{tikzpicture}
\begin{axis}[
enlarge x limits=0.03,
title=Ornstein-Uhlenbeck sample
($13000$ time steps),
xlabel=$t$]
\addplot[blue] file{plotdata/ou.dat};
\end{axis}
\end{tikzpicture}
\end{codeexample}
\message{ok, passed.}%
\PGFPlots\ relies completely on \TeX\ to do all typesetting. In uses the front-end-layer and basic layer of \PGF\ to perform all drawing operations. For complicated plots, this may take some time, and you may want to read section~\ref{sec:pgfplots:importexport} for how to write single figures to external graphics files. Externalization is the best way to reduce typesetting time.
However, for large scale plots with a lot of points, limitations of \TeX's capacities are reached easily.
\subsection{Memory limitations}
The default settings of most \TeX-distributions are quite restrictive, so it may be necessary to adjust them. For Mik\TeX, this can be done using simple command line switches:
\begin{codeexample}[code only]
pdflatex
--stack-size=n --save-size=n
--main-memory=n --extra-mem-top=n --extra-mem-bot=n
--pool-size=n --max-strings=n
\end{codeexample}
\noindent Experiment with these settings if Mik\TeX\ runs out of memory.
Usually, the log--file contains a summary about the used ressources, giving a hint which parameter needs to be increased. For Unix installations, one needs to adjust config files. This can be done as follows:
\begin{enumerate}
\item Locate |texmf.cnf| on your system. On my Ubuntu installation, it is in
|/usr/share/texmf/web2c/texmf.cnf|.
\item Either change |texmf.cnf| directly, or copy it to some convenient place. If you copy it, here is how to proceed:
\begin{itemize}
\item keep only the changed entries in your local copy to reduce conflicts. \TeX\ will always read \emph{all} config files found in its search path.
\item Adjust the search path to find your local copy. This can be done using the environment variable |TEXMFCNF|. Assuming your local copy is in |~/texmf/mytexcnf/texmf.cnf|, you can write
\begin{codeexample}[code only]
export TEXMFCNF=~/texmf/mytexcnf:
\end{codeexample}
to search first in your directory, then in all other system directories.
\end{itemize}
\item You should change the entries
\begin{codeexample}[code only]
main_memory = n
extra_mem_top = n
extra_mem_bot = n
max_strings = n
param_size = n
save_size = n
stack_size = n
\end{codeexample}
The log--file usually contains information about the parameter which needs to be enlarged.
\end{enumerate}
Unfortunately, \TeX\ does not allow arbirtrary memory limits, there is an upper bound hard coded in the executables.
|
