python 中是否有办法使用 matplotlib 创建带有子图的子图？

Question

I'm trying to display a figure that contains 3 plots, and each of the plots is a plot of (8,1)-shaped subplots.

Essentially, I want one big figure with three sections each containing (8,1)-shaped subplots.

I'm looking for a way to do this without having to manually set all the proportions and spacings. The reason I'm doing this is to visualize an 8-channel neural signal compared to three other pre-defined signals, each signal being 8 channels.

If it makes any sense this way, I'm trying for something like this (ficticious code):

fig, ax = plt.subplots(n_figures = 3, n_rows = 8, n_cols = 1)
ax[figure_i, row_j, col_k].imshow(image)

Is there a way to do this?

---

Here is an example of what I am talking about. Ideally it would three subplots, and in each of the subplots there is a set of subplots of shape 8x1. I understand how to plot this all out by going through all the margins and setting the proportions, but I'm wondering if there's a simpler way to do this without having to go through all the additional code and settings as described in the above example code I've written.

Answer 1

You can create this kind of figure by first creating a subplot grid with the appropriate layout using the plt.subplots() function and then looping through the array of axes to plot the data, like in this example:

import numpy as np                 # v 1.19.2
import matplotlib.pyplot as plt    # v 3.3.2

# Create sample signal data as a 1-D list of arrays representing 3x8 channels
signal_names = ['X1', 'X2', 'X3']
nsignals = len(signal_names)  # ncols of the subplot grid
nchannels = 8  # nrows of the subplot grid
nsubplots = nsignals*nchannels
x = np.linspace(0, 14*np.pi, 100)
y_signals = nsubplots*[np.cos(x)]

# Set subplots width and height
subp_w = 10/nsignals  # 10 corresponds the figure width in inches
subp_h = 0.25*subp_w

# Create figure and subplot grid with the appropriate layout and dimensions
fig, axs = plt.subplots(nchannels, nsignals, sharex=True, sharey=True,
                        figsize=(nsignals*subp_w, nchannels*subp_h))

# Optionally adjust the space between the subplots: this can also be done by
# adding 'gridspec_kw=dict(wspace=0.1, hspace=0.3)' to the above function
# fig.subplots_adjust(wspace=0.1, hspace=0.3)

# Loop through axes to create plots: note that the list of axes is transposed
# in this example to plot the signals one after the other column-wise, as
# indicated by the colors representing the channels
colors = nsignals*plt.get_cmap('tab10').colors[:nchannels]
for idx, ax in enumerate(axs.T.flat):
    ax.plot(x, y_signals[idx], c=colors[idx])
    if ax.is_first_row():
        ax.set_title(signal_names[idx//nchannels], pad=15, fontsize=14)

plt.show()