1
\$\begingroup\$

I have a list of N dimensional NumPy arrays.

num_vecs = 10
dims = 2

vecs = np.random.normal(size=(num_vecs, dims))

I want to normalize them, so the magnitude/length of each vector is 1. I can easily do this with a for-loop.

norm = np.linalg.norm(vecs, axis=1)

for dd in range(dims):
    vecs[:, dd] /= norm

assert np.allclose(np.linalg.norm(vecs, axis=1), 1.)

But how do I get rid of the for-loop?

\$\endgroup\$
2
\$\begingroup\$

The trick is to use the keepdims parameter.

import numpy as np

num_vecs = 10
dims = 2

vecs = np.random.normal(size=(num_vecs, dims))


vecs /= np.linalg.norm(vecs, axis=1, keepdims=True)

assert np.allclose(np.linalg.norm(vecs, axis=1), 1.)
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.