Numpy block matrix assignment with integer array indexing

Asked Sep 16 '18 at 20:21

Active Sep 16 '18 at 20:34

Viewed 516 times

Is there an elegant way (without loops) to do this kind of assignment with numpy indexing?

import numpy as np

m = 3
n = 5
small_matrix = np.random.randn(m, n)
big_matrix = np.zeros((100, 100))
row_indices = [3, 16, 9]
col_indices = [31, 45, 69, 71, 83]

for i in range(m):
    for j in range(n):
        big_matrix[row_indices[i], col_indices[j]] = small_matrix[i, j]

asked Sep 16 '18 at 20:21

janismac

2

There are lots of almost-duplicates: `np.ix_` is the way to go, i.e. `big_matrix[np.ix_(row_indices, column_indices)] = small_matrix`. – DSM Sep 16 '18 at 20:31
@DSM Thanks! I didn't find those. – janismac Sep 16 '18 at 20:42
I cheated because I knew what function to look for. ;-) – DSM Sep 16 '18 at 20:45

Numpy block matrix assignment with integer array indexing

0 Answers0