Defining new colormap in matplotlib with colors "centered" around integers

Question

I’m trying to use a colormap in matplotlib where each color is "localized" about an integer. For example, when the data I am plotting is in the range [-1,1], I want it to be clear which data is around -1, 0, and 1. cmap = mpl.cm.coolwarm

fig, axs = plt.subplots(1, 1)
cmap = mpl.cm.coolwarm
levels = np.arange(-1.2,1.21,.01)
cs = axs.contourf(x_list,y_list,z_list,levels=levels, cmap=cmap)

bounds = [-1, 0, 1]
norm = mpl.colors.BoundaryNorm(bounds, cmap.N)
cbar = fig.colorbar(cs)

This works nicely, with data around -1 being blue, data around 0 being grey, and data around 1 to be red. However, I want to now consider data in the range [-2,2]. Now, I need to have the colormap yield different colors around -2 and 2 (say, magneta and green). How can I define a custom color map and edit the above to do this?

Asked By: Joshuah Heath

||

Source

Answer 1

One way to achieve that is with LinearSegmentedColormap. Read more about it at this documentation page.

The idea is very simple: from -2 to 2 there are 5 integers (2 and -2 included). So you create a list with 5 colors:

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

colors = ["r", "g", "b", "m", "c"]
cmap = LinearSegmentedColormap.from_list("mycmap", colors)

img = np.random.randint(-2, 3, (256, 256))
plt.figure()
i = plt.imshow(img, cmap=cmap)
plt.colorbar(i)

As you can see, each color is centered at a particular integer.

Answered By: Davide_sd

Defining new colormap in matplotlib with colors "centered" around integers

Question:

Answers: