Bug summary

colors.PowerNorm ignores the (in this case) vmin values, and the resulting colorbar seems to b incorrect.

Code for reproduction

import matplotlib.pyplot as pl
import matplotlib.colors as colors
import numpy as np
    
data = np.random.random(100*100).reshape(100,100) * 600 + 398
    
pl.figure()
pl.imshow(data, cmap=pl.cm.terrain, norm=colors.PowerNorm(gamma=0.2, vmin=398, vmax=1000))
pl.colorbar()

Actual outcome

It seems that vmin is changed to some pretty number (in this case 300), but as a result, the colorbar seems to be incorrect. In the screenshot, my mouse is hoovering over the blue data point, which has a value of 398.062 (see value in top right corner of window). However, corresponding values in the colorbar are greenish, while blue colors correspond to much lower numbers.

Expected outcome

I would expect matplotlib to respect the vmin and vmax values that I set in colors.PowerNorm.

You can best see the correct outcome if you manually set vmin at the value that matplotlib chooses:

import matplotlib.pyplot as pl
import matplotlib.colors as colors
import numpy as np

data = np.random.random(100*100).reshape(100,100) * 600 + 398

pl.figure()
pl.subplot(121)
pl.imshow(data, cmap=pl.cm.terrain, norm=colors.PowerNorm(gamma=0.2, vmin=398, vmax=1000))
cb=pl.colorbar()

pl.subplot(122)
pl.imshow(data, cmap=pl.cm.terrain, norm=colors.PowerNorm(gamma=0.2, vmin=300, vmax=1000))
cb=pl.colorbar()

Additional information

No response

Operating system

Manjaro Linux

Matplotlib Version

3.6.2

Matplotlib Backend

Qt5Agg

Python version

3.10.9

Jupyter version

None

Installation

pip