Why NumPy Transpose Is Almost Instant

Sun, 10 May 2026 17:12:09 +0000

Most operations on large NumPy arrays are computationally expensive, so I naturally assumed that transposing one would involve rearranging huge chunks of data in memory. But when I tried it and saw it happen almost instantly, I was surprised.

How could a large transformation of this kind happen without any noticeable delay?

Imagine we have a matrix,

import numpy as np
arr = np.array([
 [1,2,3],
 [4,5,6]
])

arr
## returns below output ##
array([[1, 2, 3],
 [4, 5, 6]])

After applying transpose, we get

akscape

Why NumPy Transpose Is Almost Instant