使用Matplotlib make_axes_locatable时轴标签未对齐问题求助

问题背景 #

使用make_axes_locatable创建带直方图和色条的布局时，轴标签无法通过fig.align_labels()对齐，Matplotlib版本3.6.0。

原因 #

make_axes_locatable创建的附加轴（如顶部直方图、右侧直方图）属于AxesDivider管理的嵌套轴，fig.align_labels()默认不会自动处理这类轴的标签对齐；再加上轴设置了sharex/sharey或固定宽高比，进一步干扰了标签的位置计算逻辑。

解决方案 #

方案1：手动调整标签位置

先绘制一次获取主坐标轴标签的位置，再将其他轴的标签同步到该位置，确保对齐：

import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1 import make_axes_locatable

fig, ax = plt.subplots(constrained_layout=True)  # 改用constrained_layout替代tight_layout
ax.set_aspect("equal")
divider = make_axes_locatable(ax)

ax_histx = divider.append_axes("top", size="25%", pad="0.75%", sharex=ax)
ax_histy = divider.append_axes("right", size="25%", pad="0.75%", sharey=ax)
cax = divider.append_axes("right", size="4%", pad="0.75%")
hcax = divider.append_axes("top", size="4%", pad="0.75%")

ax.set_xlim(0,10000)
ax.set_ylim(0,10000)
ax.tick_params(axis='x', rotation=90)

# 设置标签
ax.set_xlabel("xlabel")
ax_histy.set_xlabel("xlabel")
ax.set_ylabel("ylabel")
ax_histx.set_ylabel("ylabel")

# 先绘制一次获取标签位置
fig.canvas.draw()
# 对齐x轴标签
x_label_pos = ax.xaxis.get_label().get_position()
ax_histy.xaxis.get_label().set_position(x_label_pos)
# 对齐y轴标签
y_label_pos = ax.yaxis.get_label().get_position()
ax_histx.yaxis.get_label().set_position(y_label_pos)

plt.show()

方案2：改用GridSpec布局替代make_axes_locatable

用matplotlib.gridspec.GridSpec更精准地控制轴的位置，让标签自然对齐：

import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec

fig = plt.figure(constrained_layout=True)
# 定义4行4列网格，分配各轴位置
gs = gridspec.GridSpec(4, 4, figure=fig)

ax = fig.add_subplot(gs[1:4, 0:3])
ax_histx = fig.add_subplot(gs[0, 0:3], sharex=ax)
ax_histy = fig.add_subplot(gs[1:4, 3], sharey=ax)
cax = fig.add_subplot(gs[1:4, 4])  # 可调整列数适配色条
hcax = fig.add_subplot(gs[0, 3])

ax.set_aspect("equal")
ax.set_xlim(0,10000)
ax.set_ylim(0,10000)
ax.tick_params(axis='x', rotation=90)

ax.set_xlabel("xlabel")
ax_histy.set_xlabel("xlabel")
ax.set_ylabel("ylabel")
ax_histx.set_ylabel("ylabel")

fig.align_labels()  # 此时align_labels可正常工作

plt.show()

方案3：禁用轴共享并强制对齐标签

如果不需要轴共享，去掉sharex/sharey后，直接指定要对齐的轴：

import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1 import make_axes_locatable

fig, ax = plt.subplots(constrained_layout=True)
ax.set_aspect("equal")
divider = make_axes_locatable(ax)

# 去掉sharex/sharey，手动同步轴范围
ax_histx = divider.append_axes("top", size="25%", pad="0.75%")
ax_histy = divider.append_axes("right", size="25%", pad="0.75%")
cax = divider.append_axes("right", size="4%", pad="0.75%")
hcax = divider.append_axes("top", size="4%", pad="0.75%")

ax.set_xlim(0,10000)
ax.set_ylim(0,10000)
ax_histx.set_xlim(0,10000)
ax_histy.set_ylim(0,10000)
ax.tick_params(axis='x', rotation=90)

ax.set_xlabel("xlabel")
ax_histy.set_xlabel("xlabel")
ax.set_ylabel("ylabel")
ax_histx.set_ylabel("ylabel")

# 强制对齐指定轴的标签
fig.align_xlabels([ax, ax_histy])
fig.align_ylabels([ax, ax_histx])

plt.show()

内容的提问来源于stack exchange，提问作者RaphBilder