我正在尝试制作一个包含使用渐近函数和 pgfplots 绘制的图的子图。正如人们所看到的，子图的大小不同。我希望使所有子图的大小相同，并使它们垂直和水平对齐。
对齐它们意味着绘图 2 和绘图 4 的 y 标签要对齐。绘图 1 和绘图 2 的 x 标签要对齐。
更新的代码：
\documentclass[12pt, a4paper,
parskip=half,
toc=bibliography,
numbers=noendperiod
]{scrbook}
\usepackage[hmargin={3.0cm, 2.5cm},
vmargin={2.5cm, 2.0cm},
includehead, includefoot
]{geometry}
\usepackage[UKenglish]{babel}
\usepackage[T1]{fontenc}
\usepackage{scrlayer-scrpage}
\usepackage{subcaption}
\usepackage{caption}
%\addtokomafont{caption}{\scriptsize}
\setkomafont{captionlabel}{\bfseries\rmfamily}
%\addtokomafont{caption}{\scriptsize}
\usepackage[dvipsnames,table,xcdraw]{xcolor}
\usepackage{pgfplots}
\pgfplotsset{compat=1.18}
\usepackage[inline]{asymptote}
%import graph;
%\usepackage[paper=a4paper,hmargin=2cm,vmargin=2.5cm]{geometry}
%\usepackage{parskip}
%\setlength{\columnsep}{3em}
%\title{Asymptote 3D graphics}
\begin{document}
\begin{figure}[ht]
\pgfplotsset{
width=\linewidth, height=0.9\linewidth,
%enlarge y limits=0.1,
%tick label style={font=\scriptsize},
%xticklabel style = {text width=0.4em},
%ylabel style = {text width=1em, inner sep=2pt},
%yticklabel style = {text width=0.2em},
label style={font=\bfseries\boldmath},
%tick label style={font=\bfseries\boldmath},
%
%tick align = outside,
%tick pos = left,
%every axis plot post/.append style={color=Blue, dashed, mark=*},
}
\begin{subfigure}[t]{0.48\linewidth}
\begin{tikzpicture}
\begin{loglogaxis}[
xmin=0.1, xmax=0.1e8,
ymin=1e-8, ymax=5e3,
xlabel={Number of Flops},
ylabel={Normalized MSE: $\log_{10} (e)$ },
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} },
]
\addplot[scatter, only marks,
scatter src=explicit symbolic]
table[meta=label] {
x y label
144 0.11587881 a
};
\end{loglogaxis}
\end{tikzpicture}
\caption{Plot 1}
\end{subfigure}
\hfill
\begin{subfigure}[t]{0.48\linewidth}
%\begin{tikzpicture}
%\begin{axis}[
% %domain=0:25,
% ymin= -3e-2, ymax= 3e-2,
% xmin=1, xmax=2,
% xlabel={Input $x$},
% ylabel={$f(x)$} ]
\begin{asy}
import graph;
size(7.5cm,8cm,IgnoreAspect);
real xmin=1,xmax=2, ymin=-.03,ymax=.03;
pair f(real x){return (x,Jn(0,170x)*cos(100x)/(x^2+1));}
guide gf=graph(f,xmin,xmax,n=1000);
(draw(gf,orange));
pair f(real x){return (x,Jn(0,90x)*cos(100x)/(x^2+1));}
guide gf2=graph(f,xmin,xmax,n=1000);
(draw(gf2,black));
label("$F(x)=\dfrac{\cos(100x)}{x^2+1}J_0(170x)$",(2,.02),2W); xaxis("Input",BottomTop,LeftTicks); yaxis("$f(x)$",LeftRight,RightTicks);
\end{asy}
%\end{axis}
%\end{tikzpicture}
\caption{Plot 2}
\end{subfigure}
\medskip
\begin{subfigure}[t]{0.48\linewidth}
\begin{tikzpicture}
\begin{loglogaxis}[
xmin=0.1, xmax=0.1e8,
ymin=1e-7, ymax=5e3,
xlabel={Number of Flops},
ylabel={Normalized MSE: $\log_{10} (e)$ },
%label style={font=\bfseries\boldmath},
%tick label style={font=\bfseries\boldmath},
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} },
]
\addplot[scatter, only marks,
scatter src=explicit symbolic]
table[meta=label] {
x y label
320 0.608216725 a
\
};
\end{loglogaxis}
\end{tikzpicture}
\caption{Plot 3}
\end{subfigure}
\hfill
\begin{subfigure}[t]{0.48\linewidth}
%\begin{tikzpicture}
% \begin{axis}[
% domain=0:1,
% xlabel={Input $x$},
% ylabel={$f(x)$},
% ymin= 0, ymax=1e1,
% ]
% \addplot {e^x)};
% \addplot {e^(2*x))};
% \legend{$\exp(x)$,$\exp(2x)$}
% \end{axis}
%\end{tikzpicture}
\begin{tikzpicture}
\begin{axis}[%domain=0:1,
xlabel={Input $x$},
ylabel={$f(x)$},
ymin= -1e1, ymax=1e1 ]
\addplot[orange,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
\addplot[orange,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
%\legend{\textcolor{orange}{$\mathrm e^x\sin\left(100\cosh\left(x\right)\right)$}}
\addplot[black,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\addplot[black,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\legend{\textcolor{black}{$\mathrm e^x\sin\left(100\cosh\left(x\right)\right)$}, \textcolor{black}{$\mathrm e^x\sin\left(10\cosh\left(x\right)\right)$}}
\end{axis}
\end{tikzpicture}
\caption{Plot 4}
\end{subfigure}
\caption{caption for all plots}
\label{all_plots}
\end{figure}
\end{document}
答案1
首先说几句关于你问题的事情。我第一次打开它时，它没有编译。请确保始终发布一个可以编译的问题，这样帮助你的人就可以专注于真正解决你遇到的问题，而不是构建你遇到的问题。此外，尝试将代码缩减到绝对必要的最小值（最小工作示例），这样你的问题就一目了然了。
现在来谈谈你的问题：正如pgfplots 手册（p.456）groupplot可以被认为是环境的扩展axis。因此，以下代码
\begin{tikzpicture}
\begin{axis}[put your options for pic 1 here]
//your plots 1
\end{axis}
\end{tikzpicture}
\begin{tikzpicture}
\begin{axis}[put your options for pic 2 here]
//your plots 2
\end{axis}
\end{tikzpicture}
转换为
\begin{tikzpicture}
\begin{groupplot}[put your groupplot options here]
\nextgroupplot[put your options for pic 1 here]
\\your plots 1
\nextgroupplot[put your options for pic 2 here]
\\your plots 2
\end{axis}
\end{tikzpicture}
考虑到这一点，你可以尝试类似
\documentclass{scrbook}
\usepackage{amsmath}
\usepackage{pgfplots}
\usepgfplotslibrary{groupplots}
\pgfplotsset{compat=1.18}
\begin{document}
\begin{figure}[ht]
\pgfplotsset{label style={font=\bfseries\boldmath}}
\begin{tikzpicture}
\begin{groupplot}[group style={group size=2 by 2, vertical sep=1.5cm, horizontal sep = 1.8cm}, height=0.35\textheight, width=0.48\textwidth]
\nextgroupplot[
ymode=log,
xmode = log,
xmin=0.1, xmax=0.1e8,
ymin=1e-8, ymax=5e3,
xlabel={Number of Flops},
ylabel={Normalized MSE: $\log_{10} (e)$ },
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} },
]
\addplot[scatter, only marks, scatter src=explicit symbolic]
table[meta=label] {
x y label
144 0.11587881 a
};
\nextgroupplot[
xmin = 1, xmax = 2,
ymin = -0.03, ymax = 0.03,
xlabel = {Input},
ylabel = {$f(x)$},
]
\addplot[orange, samples = 5000] gnuplot {besj0(170*x)*cos(100*x)/(x^2+1)};
\addlegendentry{$F(x)=\dfrac{\cos(100x)}{x^2+1}J_0(170x)$}
\addplot[black, samples = 5000] gnuplot {besj0(90*x)*cos(100*x)/(x^2+1)};
\nextgroupplot[
ymode=log,
xmode=log,
xmin=0.1, xmax=0.1e8,
ymin=1e-7, ymax=5e3,
xlabel={Number of Flops},
ylabel={Normalized MSE: $\log_{10} (e)$ },
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} }]
\addplot[scatter, only marks, scatter src=explicit symbolic]
table[meta=label] {
x y label
320 0.608216725 a
};
\nextgroupplot[
xlabel={Input $x$},
ylabel={$f(x)$},
xmin = 0, xmax = 2,
ymin= -1e1, ymax=1e1
]
\addlegendimage{orange}
\addlegendimage{black}
\legend{\textcolor{black}{$\mathrm e^x\sin\left(100\cosh\left(x\right)\right)$}, \textcolor{black}{$\mathrm e^x\sin\left(10\cosh\left(x\right)\right)$}}
\addplot[orange,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
\addplot[orange,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
\addplot[black,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\addplot[black,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\end{groupplot}
\end{tikzpicture}
\caption{caption for all plots; \textbf{top left:} text; \textbf{top right:} more text; \textbf{bottom left:} still more text; \textbf{bottom right:} last text}
\label{all_plots}
\end{figure}
\end{document}
是什么让你
我还做了一些改变：
- 将您的情节转换
asymptote为pgfplots使用gnuplot，因为我找不到解决方案asymptote并groupplot一起工作 - 将贝塞尔图的样本数量改为 5000，因为它们看起来不太清晰（你可能需要进一步调整
- 修复了上一个图例，因为之前的图例显示了两条橙色的线
- 为图 4 添加了 x 轴限制
- 删除了问题中所有不必要的包和选项以及所有注释的代码
编辑：
- 使用 XeLaTeX 编译文档没有任何问题。我在 Overleaf 中尝试过，并将编译器更改为 XeLaTeX，完全没有问题。您的文档的其他部分可能有错误？
- 的对齐方式
ylabels以 的宽度为导向yticks。由于右侧两个图中的尺寸不同，因此标签位于不同的位置。您可以通过手动设置标签的位置来更正此问题。只需添加类似 的内容即可y label style={at={(-0.15,0.5)}},。坐标是相对于绘图比例大小的相对坐标。例如，这意味着高度为 50%，宽度为 -15%。所有四个绘图的宽度和高度均相同（因为这些值是在所有绘图的全局选项中设置的： ），因此在垂直和水平对齐中导致\begin{groupplot}[..., height=0.35\textheight, width=0.48\textwidth]相同的相对位置。ylabel - 关于副标题，我猜你指的是这个问题。您可能会注意到，在这个问题中，
groupplot环境被赋予了名称my plots，以便您可以解决它。如果您添加相应的名称，例如\begin{groupplot}[group style={group name=my plots,...},...]命令工作正常。但是我尝试了一种不同的方法，因为子标题与图的重叠xlabel。相反，我通过添加来命名所有图的 x 标签xlabel style = {name = label_plot_x},。现在您可以相对于它定位子标题。例如，通过添加以下代码
\node[text width=0.48\textwidth, align=left, anchor=north] at (label_plot_x.south) {\subcaption{Caption of plot x \label{subplot:plot_x}}};
- 请注意，图表和子标题节点的宽度应该匹配（我将该值更改为
0.45\textwidth与我的第一个答案相比） - 我通过添加以下代码将右上角图表的样式更改
yticks为与您问题中的样式相匹配。
yticklabel style={
/pgf/number format/fixed,
/pgf/number format/precision=2,
},
scaled y ticks = false,
- 我将右上角图例移到了顶部，因为它与图例重叠，并且我将框稍微放大一点，以便所有图例都适合它。这是通过添加
legend style = {
at={(0.5,1.05)},
anchor=south,
minimum height=1.2cm,
},
所以最终的代码是这样的
\documentclass{scrbook}
\usepackage{amsmath}
\usepackage{pgfplots}
\usepgfplotslibrary{groupplots}
\pgfplotsset{compat=1.18}
\usepackage{caption, subcaption}
\begin{document}
\begin{figure}[ht]
\pgfplotsset{label style={font=\bfseries\boldmath}}
\begin{tikzpicture}
\begin{groupplot}[group style={group name=my plots, group size=2 by 2, vertical sep=2.5cm, horizontal sep =1.8cm}, height=0.32\textheight, width=0.45\textwidth]
\nextgroupplot[
ymode=log,
xmode = log,
xmin=0.1, xmax=0.1e8,
ymin=1e-8, ymax=5e3,
xlabel={Number of Flops},
xlabel style = {name = label_upper_left},
ylabel={Normalized MSE: $\log_{10} (e)$ },
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} },
]
\addplot[scatter, only marks, scatter src=explicit symbolic]
table[meta=label] {
x y label
144 0.11587881 a
};
\nextgroupplot[
xmin = 1, xmax = 2,
ymin = -0.03, ymax = 0.03,
xlabel = {Input},
ylabel = {$f(x)$},
y label style={at={(-0.15,0.5)}},
xlabel style = {name = label_upper_right},
yticklabel style={
/pgf/number format/fixed,
/pgf/number format/precision=2,
},
scaled y ticks = false,
legend style = {
at={(0.5,1.05)},
anchor=south,
minimum height=1.2cm,
},
]
\addplot[orange, samples = 5000] gnuplot {besj0(170*x)*cos(100*x)/(x^2+1)};
\addlegendentry{$F(x)=\dfrac{\cos(100x)}{x^2+1}J_0(170x)$}
\addplot[black, samples = 5000] gnuplot {besj0(90*x)*cos(100*x)/(x^2+1)};
\nextgroupplot[
ymode=log,
xmode=log,
xmin=0.1, xmax=0.1e8,
ymin=1e-7, ymax=5e3,
xlabel={Number of Flops},
xlabel style = {name = label_lower_left},
ylabel={Normalized MSE: $\log_{10} (e)$ },
scatter/classes={ a={mark=square*, blue}, b={mark=square*, red}, c={mark=square, black}, d={mark=triangle*, blue}, e={mark=triangle*, red},f={mark=triangle*, black},g={mark=x, black}, h={mark= diamond*, pink} }]
\addplot[scatter, only marks, scatter src=explicit symbolic]
table[meta=label] {
x y label
320 0.608216725 a
};
\nextgroupplot[
xlabel={Input $x$},
xlabel style = {name = label_lower_right},
ylabel={$f(x)$},
xmin = 0, xmax = 2,
ymin= -1e1, ymax=1e1,
y label style={at={(-0.15,0.5)}},
]
\addlegendimage{orange}
\addlegendimage{black}
\legend{\textcolor{black}{$\mathrm e^x\sin\left(100\cosh\left(x\right)\right)$}, \textcolor{black}{$\mathrm e^x\sin\left(10\cosh\left(x\right)\right)$}}
\addplot[orange,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
\addplot[orange,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(100*cosh(\x) r)});
\addplot[black,samples=500,domain=0:1.5]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\addplot[black,samples=1000,domain=1.5:2]plot (\x, {exp(\x)*sin(10*cosh(\x) r)});
\end{groupplot}
\node[text width=0.48\textwidth, align=center, anchor=north] at (label_upper_left.south) {\subcaption{Caption of upper left plot \label{subplot:upper_left}}};
\node[text width=0.48\textwidth, align=center, anchor=north] at (label_upper_right.south) {\subcaption{Caption of upper right plot \label{subplot:upper_right}}};
\node[text width=0.48\textwidth, align=center, anchor=north] at (label_lower_left.south) {\subcaption{Caption of lower left plot \label{subplot:lower_left}}};
\node[text width=0.48\textwidth, align=center, anchor=north] at (label_lower_right.south) {\subcaption{Caption of lower right plot \label{subplot:lower_right}}};
\end{tikzpicture}
\caption{Caption for all plots}
\label{all_plots}
\end{figure}
\end{document}
编译结果如下
