使用子图放大时间序列或如何在轴边框外绘制线条

Question

我想用matplotlib生成这样的情节

时间表http://www.imagenetz.de/img.php?file=37d8879009.jpg&pid=

目前我只使用matplotlib做3个子图,并在inkscape中添加红线.我发现我可以创建虚线矩形Rectangle.

ax.add_patch(Rectangle((25, -1.4), 3, 1.3, edgecolor='red',
                       fill=False, linestyle='dashed'))

我还没有发现任何东西来绘制连接图的线条.是否有可以跨轴边界绘制的功能？

Answer 1

简答

我们可以利用 plt.annotate() 在图形坐标中在轴边界外绘制线条。

长答案

首先定义一个辅助函数：

from matplotlib.patches import Rectangle

def zoomingBox(ax1, roi, ax2, color='red', linewidth=2, roiKwargs={}, arrowKwargs={}):
    '''
    **Notes (for reasons unknown to me)**
    1. Sometimes the zorder of the axes need to be adjusted manually...
    2. The figure fraction is accurate only with qt backend but not inline...
    '''
    roiKwargs = dict([('fill',False), ('linestyle','dashed'), ('color',color), ('linewidth',linewidth)] + roiKwargs.items())
    ax1.add_patch(Rectangle([roi[0],roi[2]], roi[1]-roi[0], roi[3]-roi[2], **roiKwargs))
    arrowKwargs = dict([('arrowstyle','-'), ('color',color), ('linewidth',linewidth)] + arrowKwargs.items())
    srcCorners = [[roi[0],roi[2]], [roi[0],roi[3]], [roi[1],roi[2]], [roi[1],roi[3]]]
    dstCorners = ax2.get_position().corners()
    srcBB = ax1.get_position()
    dstBB = ax2.get_position()
    if (dstBB.min[0]>srcBB.max[0] and dstBB.max[1]<srcBB.min[1]) or (dstBB.max[0]<srcBB.min[0] and dstBB.min[1]>srcBB.max[1]):
        src = [0, 3]; dst = [0, 3]
    elif (dstBB.max[0]<srcBB.min[0] and dstBB.max[1]<srcBB.min[1]) or (dstBB.min[0]>srcBB.max[0] and dstBB.min[1]>srcBB.max[1]):
        src = [1, 2]; dst = [1, 2]
    elif dstBB.max[1] < srcBB.min[1]:
        src = [0, 2]; dst = [1, 3]
    elif dstBB.min[1] > srcBB.max[1]:
        src = [1, 3]; dst = [0, 2]
    elif dstBB.max[0] < srcBB.min[0]:
        src = [0, 1]; dst = [2, 3]
    elif dstBB.min[0] > srcBB.max[0]:
        src = [2, 3]; dst = [0, 1]
    for k in range(2):
        ax1.annotate('', xy=dstCorners[dst[k]], xytext=srcCorners[src[k]], xycoords='figure fraction', textcoords='data', arrowprops=arrowKwargs)

然后我们可以这样做：

import matplotlib.pyplot as plt

axs = plt.subplots(2, 2)[1]
axs[1,1].plot(rand(100))
zoomingBox(axs[1,1], [40,60,0.1,0.9], axs[0,0])
zoomingBox(axs[1,1], [10,30,0.1,0.9], axs[1,0], color='orange')