How does matplotlib custom colormaps work

Question

I was trying to create custom color map with exaples from documentation but I have no idea how setting color range works.

https://matplotlib.org/2.0.2/examples/pylab_examples/custom_cmap.html

This is the closest to what I need: (Full green from 1.0 to 0.916, full yellow from 0.916 to 0.75 and full red below 0.75)

cdict1 = {'red':   ((0.0,  1.0, 1.0),
                   (0.75,  1.0, 1.0),
                   (1.0,  0.0, 0.0)),

         'green': ((0.0,  0.0, 0.0),
                   (0.75, 1.0, 1.0),
                   (0.91666666666, 1.0, 1.0),
                   (1.0,  1.0, 1.0)),

         'blue':  ((0.0,  0.0, 0.0),
                   (0.5,  0.0, 0.0),
                   (1.0,  0.0, 0.0))}

I don't undestand why this colormap is a smooth transition between colors.

Answer 1

To create a colormap with 3 fixed colors with unequal boundaries, the recommended approach uses a BoundaryNorm .

If you really only want to work with a colormap, you could create one from a list of colors .

A LinearSegmentedColormap makes smooth transitions with specific colors at given values. To make it work with fixed colors, these values can be set equal. The function either works the "old" way manipulating rgb values, or with a list of (value, color) pairs ( LinearSegmentedColormap.from_list() ).

The following example code shows how this can work:

import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap, BoundaryNorm, LinearSegmentedColormap
import numpy as np

x, y = np.random.rand(2, 100)
fig, (ax1, ax2, ax3) = plt.subplots(ncols=3, figsize=(14, 4))

# working with a BoundaryNorm
cmap1 = ListedColormap(['red', 'yellow', 'green'])
norm1 = BoundaryNorm([0, 0.75, 0.916, 1], ncolors=3)
scat1 = ax1.scatter(x, y, c=y, cmap=cmap1, norm=norm1)
plt.colorbar(scat1, ax=ax1, spacing='proportional')
ax1.set_title('working with BoundaryNorm')

# creating a special colormap
colors = ['green' if c > 0.916 else 'red' if c < 0.75 else 'yellow' for c in np.linspace(0, 1, 256)]
cmap2 = ListedColormap(colors)
scat2 = ax2.scatter(x, y, c=y, cmap=cmap2, vmin=0, vmax=1)
plt.colorbar(scat2, ax=ax2)
ax2.set_title('special list of colors')

cmap3 = LinearSegmentedColormap.from_list('', [(0, 'red'), (0.75, 'red'), (0.75, 'yellow'), (0.916, 'yellow'),
                                               (0.916, 'green'), (1, 'green')])
scat3 = ax3.scatter(x, y, c=y, cmap=cmap3, vmin=0, vmax=1)
plt.colorbar(scat3, ax=ax3)
ax3.set_title('LinearSegmentedColormap')

plt.tight_layout()
plt.show()

The spacing='proportional' option of plt.colorbar shows the boundaries at their proportional location. The default shows 3 equally-spaced boundaries together with the values.