How to multiply numpy 2D array with numpy 1D array?

Question

The two arrays:

a = numpy.array([[2,3,2],[5,6,1]])
b = numpy.array([3,5])
c = a * b

What I want is:

c = [[6,9,6],
     [25,30,5]]

But, I am getting this error:

ValueError: operands could not be broadcast together with shapes (2,3) (2)

How to multiply a nD array with 1D array, where len(1D-array) == len(nD array)?

For users searching how to create a 3D array by multiplying a 2D array with a 1D array (the title of this question), note the solution is `a * b[:, None, None]` (or equivalently `numpy.multiply.outer(b, a)`), not `a * b[:, None]` which corresponds to additional details in the question body. — mins, Mar 23 '23 at 10:36

score 46 · Accepted Answer · answered Apr 26 '13 at 06:12

46

You need to convert array b to a (2, 1) shape array, use None or numpy.newaxis in the index tuple:

import numpy
a = numpy.array([[2,3,2],[5,6,1]])
b = numpy.array([3,5])
c = a * b[:, None]

Here is the document.

answered Apr 26 '13 at 06:12

HYRY

Thanks! What is the name of this syntax in Python: `[:, None]`? – Ashwin Nanjappa Apr 26 '13 at 06:14
1

@Ashwin, you can search `numpy.newaxis` for it. – HYRY Apr 26 '13 at 06:18
How does this work when `a` has an arbitrary number of dimensions? – Evidlo Mar 15 '23 at 20:57

Mike T · Answer 2 · 2016-12-15T21:49:00.887

3

Another strategy is to reshape the second array, so it has the same number of dimensions as the first array:

c = a * b.reshape((b.size, 1))
print(c)
# [[ 6  9  6]
#  [25 30  5]]

Alternatively, the shape attribute of the second array can be modified in-place:

b.shape = (b.size, 1)
print(a.shape)  # (2, 3)
print(b.shape)  # (2, 1)
print(a * b)
# [[ 6  9  6]
#  [25 30  5]]

edited Dec 15 '16 at 21:49

answered Dec 15 '16 at 21:07

Mike T

2 Answers2