Subtract Numpy Array by Column

Question

So I have an example array, say:

import numpy as np
np.array([[[ 85, 723]],
          [[ 86, 722]],
          [[ 87, 722]],
          [[ 89, 724]],
          [[ 88, 725]],
          [[ 87, 725]]])

What I want to do is subtract a number from only the second column, say 10 for example. What I hope to have the output look like is something like this:

np.array([[[ 85, 713]],
          [[ 86, 712]],
          [[ 87, 712]],
          [[ 89, 714]],
          [[ 88, 715]],
          [[ 87, 715]]])

I have tried using np.subtract, but it does not support subtraction along an axis (at least to my knowledge).

Answer 1

Slice and subtract -

a[...,1] -= 10

This would work for arrays of any number of dimensions to subtract from the second column.

Sample run -

In [582]: a
Out[582]: 
array([[[30, 23]],

       [[36, 88]],

       [[27, 15]],

       [[38, 61]],

       [[79, 14]]])

In [583]: a[...,1] -= 10

In [584]: a
Out[584]: 
array([[[30, 13]],

       [[36, 78]],

       [[27,  5]],

       [[38, 51]],

       [[79,  4]]])

Answer 2

Do an in-place subtraction on the specified index (in this case I index the whole column):

>>> arr[:, :, 1] -= 10

>>> arr
array([[[ 85, 713]],
       [[ 86, 712]],
       [[ 87, 712]],
       [[ 89, 714]],
       [[ 88, 715]],
       [[ 87, 715]]])

Also works with np.subtract when you specify out:

>>> np.subtract(arr[:, :, 1], 10, out=arr[:, :, 1])