不同形状的numpy乘法数组[重复]

This question already has an answer here:

这个问题已经有了答案:

Multiplying 2d array by 1d array 1 answer
用2d数组乘以1d数组1答案

I have an array A of shape (w,h) = 3000,2000 and another array B of shape d = 100

我有一个形状为(w,h)的数组A = 3000,2000，还有一个形状为d = 100的数组B

I want to multiply each value of A by B, and get the result in the form of an array C of shape (w,h,d) = 3000,2000,100

我想把A的每个值乘以B，得到一个形状为C的数组(w,h,d) = 3000,2000,100

Right now I am using the very slow code below, how can I vectorize this operation?

现在我正在使用下面非常慢的代码，我如何对这个操作进行矢量化?

w,h,d = 3000,2000,100

A = np.ones((w,h))

B = np.arange(d)

C = np.zeros((w,h,d))

for i in xrange(w):
    for j in xrange(h):
        C[i,j] = A[i,j] * B

Thank you

谢谢你！

1 个解决方案

#1

Use numpy broadcast.

使用numpy广播。

Try this

试试这个

from numpy.random import rand
a = rand(4,5)
b = rand(6)
c = a[...,None] * b
print (c.shape)

Or equivelently

或equivelently

c = a.reshape(4,5,1)*b

#1