np.array() vs np.column_stack

array-vs-columnstack.py

import perfplot
import numpy as np

rng = np.random.default_rng(0)


def setup(n):
    return [rng.random(n) for _ in range(10)]


def column_stack(data):
    return np.column_stack(data)


def array(data):
    return np.array(data)


b = perfplot.bench(
    setup=setup,
    kernels=[column_stack, array],
    n_range=[2 ** k for k in range(20)],
    equality_check=None,
)
b.save("out.png")
b.show()

When concatenating arrays, np.array is faster than np.column_stack by a factor of about 3 to 5.

Reason: For np.array, the memory doesn't need to rearranged. The array data is just copied into one place.