用索引数组更改numpy数组 [英] Changing numpy array with array of indices

问题描述

我在numpy中有一个数组: A=np.zeros((5,10)) 我想根据另一数组N=np.array([7, 2, 9, 4, 5])

I have an array in numpy: A=np.zeros((5,10)) and I want to change one value in each row with ones according to another array N=np.array([7, 2, 9, 4, 5])

like:A[:,N]=1;

0   0   0   0   0   1   0   0   0   0
0   0   1   0   0   0   0   0   0   0
0   0   0   0   0   0   0   0   0   1
0   0   0   0   1   0   0   0   0   0
0   0   0   0   0   1   0   0   0   0

我该怎么做?

推荐答案

由于您要设置每行单个元素，因此需要使用arange(5)花式索引第一个轴.可以将其视为设置索引(I0[0], N[0])=(0,7)，(I0[1],N[1])=(1,2)，...

Since you want to set a single element per row, you need to fancy-index the first axis using arange(5). this can be thought of as setting indices (I0[0], N[0])=(0,7), (I0[1],N[1])=(1,2), ...

I0 = np.arange(A.shape[0])
A[I0, N] = 1
A
=> 
array([[ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  1.,  0.,  0.],
       [ 0.,  0.,  1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  1.],
       [ 0.,  0.,  0.,  0.,  1.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  0.,  1.,  0.,  0.,  0.,  0.]])
A.nonzero()
=> (array([0, 1, 2, 3, 4]), array([7, 2, 9, 4, 5]))

这篇关于用索引数组更改numpy数组的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！