Matplotlib imshow colormap not covering full dynamic range of data

问题

I'm trying to produce colormaps of numpy arrays with large dynamic ranges. To that end, I've tried using imshow from matplotlib.pyplot, but there appears to be a problem with the colormap.

Specifically, every value in the array below a certain threshold is plotted as the same color as the minimum value in the array.

Below is a minimal working example. I produce an array containing values from 1e-4 to 1e+20, but in imshow all cells containing values less than 1e+3 appear to be set equal to 1e-4. However, there is no such issue using pcolormesh.

import matplotlib.pyplot as plt
from matplotlib.colors import LogNorm
import numpy as np

# 2d (5x5) array with large dynamic range: 1e-4, 1e-3, ..., 1e+20
a = np.logspace(-4, 20, num=25).reshape(5, 5)

# imshow producing wrong image
fig = plt.figure()
plt.imshow(a, norm=LogNorm(), origin='lower')
plt.colorbar()
fig.savefig("imshow_broken.png")

# pcolormesh producing correct image
fig = plt.figure()
plt.pcolormesh(a, norm=LogNorm())
plt.colorbar()
fig.savefig("pcolormesh_working.png")

I'm using matplotlib version 2.1.2, but the problem seems to disappear on older versions, e.g. 1.5.3, perhaps suggesting a bug in recent matplotlib versions.

imshow:

pcolormesh:

来源：https://stackoverflow.com/questions/50879082/matplotlib-imshow-colormap-not-covering-full-dynamic-range-of-data