Calculating norm of columns as vectors in a matrix

Question

I am looking for the best way of calculating the norm of columns as vectors in a matrix. My code right now is like this but I am sure it can be made better(with maybe numpy?):

import numpy as np
def norm(a):
    ret=np.zeros(a.shape[1])
    for i in range(a.shape[1]):
        ret[i]=np.linalg.norm(a[:,i])
    return ret

a=np.array([[1,3],[2,4]])
print norm(a)

Which returns:

[ 2.23606798  5.        ]

Thanks.

Duplicate of http://stackoverflow.com/questions/7741878/how-to-apply-numpy-linalg-norm-to-each-row-of-a-matrix Can be done with np.linalg.norm(a, axis=0). — Remi LP, Jan 07 '17 at 20:17

score 5 · Accepted Answer · answered Mar 31 '13 at 02:14

5

You can calculate the norm by using ufuncs:

np.sqrt(np.sum(a*a, axis=0))

answered Mar 31 '13 at 02:14

HYRY

94,853
25
187
187

Thank, may I add that `np.square(a)` may be more comfy when using something complicated instead of `a` – Matthias Beaupère Apr 23 '19 at 08:54
Note that this doesn't work for complex matrices. The correct solution is to use `np.linalg.norm` as in the next answer. – Miguel Sep 27 '20 at 12:19

Miguel · Answer 2 · 2019-04-24T14:06:50.923

3

Direct solution using numpy:

x = np.linalg.norm(a, axis=0)

edited Apr 24 '19 at 14:06

answered Dec 23 '17 at 15:15

Miguel

658
1
6
19

this method does not exist with my numy install, and I can't find any documentation about it online. Could you provide some reference ? – Matthias Beaupère Apr 23 '19 at 08:55
@mattiasbe see [https://docs.scipy.org/doc/numpy/reference/generated/numpy.linalg.norm.html](https://docs.scipy.org/doc/numpy/reference/generated/numpy.linalg.norm.html) – Miguel Apr 24 '19 at 14:08

Calculating norm of columns as vectors in a matrix

2 Answers2