工程符号（科学）作为带有 pgfplots 的刻度标签（可能通过 siunitx）？

Question 1

好的，我想我已经了解了大部分内容 - 除了现在有一个“指数往返”会导致舍入误差（见下图）；如果有人知道如何避免这种情况，请发表答案。

测试011.png

正是那些 500.12e-3 和 499.97 xtick 值显示了舍入误差；这里的问题是\tick这里的变量是对数的（实际上是 ln 的），所以我必须对它进行指数运算 - 这需要使用fpu，然后回到定点；当我们回到原始 xtick 值域时，那些值已经被“损坏”：显示\typeout：

tick -0.69316,  tmp 0.50012
tick  0.0,      tmp 1.0000000000
tick  1.60942,  tmp 4.99974000000000
tick  2.30258,  tmp 10.000000000
tick  3.912,    tmp 49.9974000000000
tick  4.60516,  tmp 100.00000000
tick  6.21458,  tmp 499.974000000000
tick  6.90775,  tmp 1000.0000000
tick  8.51717,  tmp 4999.74000000000
tick  9.21033,  tmp 10000.000000
tick 10.81975,  tmp 49997.4000000000
tick 11.51291,  tmp 100000.00000

否则，就会siunitx通过\num使用xticklabel图的键来完成工程符号格式化。

以下是 MWE：

\documentclass{article}
\usepackage{pgfplots}
\pgfplotsset{compat=1.5.1}
\usepackage{tikz}
\usepackage{siunitx}
\usepackage[pdftex,active,tightpage]{preview}

\begin{document}

\pgfkeys{
  /pgf/number format/sci generic={mantissa sep={\!\cdot\!},exponent={10^{#1}}},
}

% use \newcommand (not \def), so the one (and only)
% argument can be specified in {} (and not in []) brackets
\newcommand{\mynum}[1]{
  % =\pgfmathparse{e^\tick}\pgfmathresult : ! Dimension too large. ; - use exp(x)
  \pgfkeys{/pgf/fpu}% else dimension too large!
  \pgfmathparse{exp(#1)}%
  % \edef\tmp{\pgfmathresult}%
  % \pgfmathprintnumberto{\pgfmathresult}{\tmp}% nope, gives 1Y5.0012e-1]
  \pgfmathfloattofixed{\pgfmathresult}% also outputs in \pgfmathresult
  \edef\tmp{\pgfmathresult}%
  \pgfkeys{/pgf/fpu=false}%
  \typeout{tick #1, tmp \tmp}%
  \num[%
    scientific-notation=engineering,%
    round-mode=places,%
    round-precision=2,%
    zero-decimal-to-integer,%
    group-digits=false,%
    exponent-product=\!\cdot\!,%
  ]{\tmp} %
}

\begin{preview}
\begin{tikzpicture}
\begin{axis}[
  axis x line=bottom,
  axis y line=left,
  scaled ticks=false,
  xmode=log,
  /pgfplots/xtick={0,0.5,1,5,10,50,100,500,1000,5000,10000,50000,100000},
  x tick label style={rotate=-50, anchor=west, align=center, font=\small},
  xmin=0, xmax=200000,
  ymin=0,
  visualization depends on={rawx \as \myx}, % , \tick \as \myxt
  nodes near coords={($\pgfmathprintnumber\myx$, \textcolor{black}{\num[scientific-notation = engineering,round-mode=places,round-precision=2,zero-decimal-to-integer]{\myx}}\color{blue})},
  % cannot use options to \num here, we're already in options,
  % so extra [], like below, will cause "! Missing \endcsname inserted."
%   xticklabel=\num[scientific-notation=engineering,round-mode=places,round-precision=2,zero-decimal-to-integer]{123},
  % so, use a \newcommand ( \mynum ) here:
  xticklabel=\mynum{\tick}, % \myx undefined here, \tick is logarithm'd!
]

\addplot[color=blue,mark=*] coordinates
  { (0.5,1) (1,2) (5,3) (10,4) (50,5) (100,6) (500,7)
    (1000,8) (5000,9) (10000,10) (50000,11) (100000,12)};

\end{axis}
\end{tikzpicture}
\end{preview}

\end{document}

Answer

好的，我想我已经了解了大部分内容 - 除了现在有一个“指数往返”会导致舍入误差（见下图）；如果有人知道如何避免这种情况，请发表答案。