CSR matrices multiplication

Question

I have two sparse matrices, I need to bring them both to the CSR (Compressed Sparse Row) format and then multiply. But I also tried to first multiply the regular matrices and only then bring the result of this multiplication to the CSR format. And in the end I get dirrefent results. Why? Could someone pls explain?

Here's my code:

import numpy as np
from scipy.sparse import csr_matrix
np.random.seed(42)

J = np.eye(5)
J[4,:] = np.random.randint(low=0, high=4, size = (5))

K = np.eye(5)
K[2,:] = np.random.randint(low=0, high=4, size = (5))

print(J)
print('\n')
print(K)

Z = csr_matrix(J@K)
print('\n')
print(Z)

F = csr_matrix(J)
H = csr_matrix(K)

print('\n')
print(F@H)

Answer 1

oh, seems like I found the answer! :D I wasn't attentive enough. The result is technically same, the order of representation is different though.