numpy中的norm用法

　　np.linalg.norm() computes the norm of a NumPy array according to an order, ord, which specifies the metric by which the norm takes. For example, if we are given an array

　　　　　　[𝑥1,...,𝑥𝑛]

 　　with numbers 𝑥𝑖xi then we can compute the Frobenius Norm or more commonly called the 2-norm by doing:

In NumPy you can also use np.linalg.norm() to compute the norm of a matrix, or a matrix's columns or rows, treating each as their own array.

　　example:

import numpy as np  
from numpy.linalg import norm
np.set_printoptions(threshold='nan')

a1 = np.array([1,2,3])
a2 = np.array([0,0,-3])
testa = np.array([[ 1.76405235,  0.40015721,  0.97873798],
    [ 2.2408932 ,  2.2677152 , -0.57712067],
    [ 0.95008842,  0.79873121, -0.68033952],
    [ 0.4105985 ,  0.55464207,  0.77393398]])

testb = np.array([[ 1.76405235,  0.40015721,  0.97873798],
    [ 2.2408932 ,  2.2677152 , -0.57712067],
    [ 0.95008842,  0.79873121, -0.68033952],
    [ 0.4105985 ,  0.55464207,  0.77393398]])
dist=lambda x, y: norm(x - y, ord=1)

print np.linalg.norm([2,-1,3,-4], np.inf) # returns 2,

print np.linalg.norm(a1 - a2, ord=1) # returns 2,

 

Refer:

1 https://plot.ly/numpy/norm/