How to change rescale tick label for imshow in python?

Question

I am not able to change the tick labels. For example, I create a heatmap using imshow, and the indices for the heatmap don't correspond to the intensity values. Here is a snippet of code to demonstrate my question:

iters=100
A=np.arange(0,iters+1)/(iters)
B=np.arange(0,iters+1)/(iters)
C = np.zeros((len(A),len(B)))
for i in range(0,len(A)):
    for j in range(0,len(B)):
        a = A[i]
        b = B[j]
        C[j][i]=a-b
        
fig, ax = plt.subplots()
img = ax.imshow(C, cmap='hot', interpolation='none')
plt.colorbar(img)
ax.invert_yaxis()
ax.set_xlabel('A')
ax.set_ylabel('B')
ax.set_xticklabels(A)
ax.set_yticklabels(B)
plt.show()

The values for the X and Y axes in the plot should be labeled according to the values populated in A and B. If I set the ticks according to the arrays A and B, I only get the first few elements. How associate the tick labels with the full range of A and B?

your code would produce a warning, which is explained here: https://stackoverflow.com/questions/63723514/userwarning-fixedformatter-should-only-be-used-together-with-fixedlocator, which could solve your problem — Z Li, Feb 22 '23 at 21:37
Instead of manually setting tick labels, you could set the extent: `ax.imshow(..., extent=[x0,x1,y0,y1])`. Maybe `extent=[0,1,0,1]`, if my guess of the code's intent isn't mistaken. — JohanC, Feb 22 '23 at 21:42

plasma_r · Accepted Answer · 2023-02-23T01:27:45.280

Better solution

No doubt the more elegant solution, as suggested in the comments.

import numpy as np
import matplotlib.pyplot as plt

iters=100
A=np.arange(0,iters+1)/(iters)
B=np.arange(0,iters+1)/(iters)
C = np.zeros((len(A),len(B)))

fig, ax = plt.subplots()
img = ax.imshow(C, cmap='hot', interpolation='none',
                extent=(A[0], A[-1], B[-1], B[0]))
plt.colorbar(img)
ax.invert_yaxis()
ax.set_xlabel('A')
ax.set_ylabel('B')

plt.show()

Original solution

I guess the use case for this solution would be when you want custom labels, for example:

labels=["a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k"]

for i in range(0,len(A)):
    for j in range(0,len(B)):
        a = A[i]
        b = B[j]
        C[j][i]=a-b
        
fig, ax = plt.subplots()
img = ax.imshow(C, cmap='hot', interpolation='none')
plt.colorbar(img)
ax.invert_yaxis()
ax.set_xlabel('A')
ax.set_ylabel('B')

ax.set_xticks(np.arange(0, 101, 10), labels)
ax.set_yticks(np.arange(0, 101, 10), B[0:101:10])

plt.show()

Using ax.set_xticks(np.arange(0, 101, 10), A[0:101:10]) you set the tick locations with np.arange(0, 101, 10), and then labels.

It's recommended not to change ticks to numerical values if the same can be achieved setting the correct scale (as in using `extent=` for `imshow(...)`). Setting these new labels can be very confusing, e.g. when you would want to put text at a given position, e.g. `ax.text(0.5, 0.5, 'Center???')` would behave very surprisingly. — JohanC, Feb 22 '23 at 22:01
Thank you! The extent argument is what I was looking for. Every search I found more complex solutions, but I felt there was a simple way. Even your original solution was more simple than what I could find by searching — Ralff, Feb 23 '23 at 02:38

How to change rescale tick label for imshow in python?

1 Answers1

Better solution

Original solution