How to share secondary y-axis between subplots in matplotlib

Question

If you have multiple subplots containing a secondary y-axis (created using twinx), how can you share these secondary y-axes between the subplots? I want them to scale equally in an automatic way (so not set the y-limits afterwards by hand). For the primary y-axis, this is possible by using the keyword sharey in the call of subplot.

Below example shows my attempt, but it fails to share the secondary y-axis of both subplots. I'm using Matplotlib:

ax = []
#create upper subplot
ax.append(subplot(211))
plot(rand(1) * rand(10),'r')
#create plot on secondary y-axis of upper subplot
ax.append(ax[0].twinx())
plot(10*rand(1) * rand(10),'b')
#create lower subplot and share y-axis with primary y-axis of upper subplot
ax.append(subplot(212, sharey = ax[0]))
plot(3*rand(1) * rand(10),'g')
#create plot on secondary y-axis of lower subplot
ax.append(ax[2].twinx())
#set twinxed axes as the current axes again,
#but now attempt to share the secondary y-axis
axes(ax[3], sharey = ax[1])
plot(10*rand(1) * rand(10),'y')

This gets me something like:

Example of two subplots with failed sharing of secondary y-axis

The reason I used the axes() function to set the shared y-axis is that twinx doesn't accept the sharey keyword.

I am using Python 3.2 on Win7 x64. Matplotlib version is 1.2.0rc2.

Answer 1

You can use Axes.get_shared_y_axes() like so:

from numpy.random import rand
import matplotlib
matplotlib.use('gtkagg')
import matplotlib.pyplot as plt
# create all axes we need
ax0 = plt.subplot(211)
ax1 = ax0.twinx()
ax2 = plt.subplot(212)
ax3 = ax2.twinx()
# share the secondary axes
ax1.get_shared_y_axes().join(ax1, ax3)
ax0.plot(rand(1) * rand(10),'r')
ax1.plot(10*rand(1) * rand(10),'b')
ax2.plot(3*rand(1) * rand(10),'g')
ax3.plot(10*rand(1) * rand(10),'y')
plt.show()

Here we're just joining the secondary axes together.

• Puggie: Yes, that helped me further, thanks. I also joined the primary y-axes so that they have a shared y-axis as well.
• Mattia: It doesn't work, the second twin scale get updated, but the second twin plot stays were it is.
• ImportanceOfBeingErnest: Just to not have people get the wrong impression, the code as it is in this answer works fine.
• Ralph Versteegen: The problem @mattia might have encountered: If you call a function like ax1.set_ylim() before Axes.get_shared_y_axes().join() then the two axes will NOT have the same scales (until you shift the axes interactively). Call get_shared_y_axes().join() first. BTW it works fine if you call it after calling plot(), etc.
• piogor: join is deprecated in matplotlib 3.6.

Answer 2

@dmcdougall's solution doesn't work anymore because GrouperView doesn't define join method anymore. However, it defines joined, so replacing join by joined in their code makes it run as expected.

However, a better solution is to use the method recommended in the docs to share y-axis: Axes.sharey(Axes). A working example goes as follows:

import numpy as np
import matplotlib.pyplot as plt
# create all Axes
fig, (ax0, ax2) = plt.subplots(2, sharey=True) # share the primary y-axis
ax1 = ax0.twinx()
ax3 = ax2.twinx()
ax1.sharey(ax3) # share the secondary y-axis
ax0.plot(np.random.rand(10)*1., 'r')
ax1.plot(np.random.rand(10)*10, 'b')
ax2.plot(np.random.rand(10)*3., 'g')
ax3.plot(np.random.rand(10)*10, 'y')

which plots a figure like the following:

result