Select elements from an array using another array as index

Question

Say I have an array

A = array([[1,2,3],
           [4,5,6],
           [7,8,9]])

Index array is

B = array([[1], # want [0, 1] element of A
           [0],  # want [1, 0], element of A
           [1]])  # want [2, 1] elemtn of A

By this index array B, I want a 3-by-1 array, whose elements are taken from array A, that is

C = array([[2],
           [4],
           [8]])

I tried numpy.choose, but I failed to that.

Jaime · Accepted Answer · 2014-07-11 14:06:48Z

5

For answer completeness... Fancy indexing arrays are broadcast to a common shape, so the following also works, and spares you that final reshape:

>>> A[np.arange(3)[:, None], B]
array([[2],
       [4],
       [8]])

answered Jul 11, 2014 at 14:06

Jaime

67.7k19 gold badges128 silver badges164 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Ashwini Chaudhary · Accepted Answer · 2014-07-11 07:39:27Z

3

You can do something like this:

>>> A[np.arange(len(A)), B.ravel()].reshape(B.shape)
array([[2],
       [4],
       [8]])

answered Jul 11, 2014 at 7:39

Ashwini Chaudhary

252k60 gold badges478 silver badges519 bronze badges

Nice, why didn't I think of it, ;-)

CT Zhu · Accepted Answer · 2014-07-11 07:55:53Z

2

You can do:

>>>np.diag(A[range(3),B]).reshape(B.shape)
array([[2],
       [4],
       [8]])

If you want to use choose you can do: np.choose(B.ravel(), A.T).reshape(B.shape).

answered Jul 11, 2014 at 7:46

CT Zhu

54.6k18 gold badges125 silver badges136 bronze badges

Thanks for the trick, the doc of choose is hard for me to understand.