How to reshape numpy array with reorder?

Question

I have a 1 x 2 x 3 array:

>>> a = np.array([[[1,2,3],[4,5,6]]])
>>> a
array([[[1, 2, 3],
        [4, 5, 6]]])
>>> a.shape
(1, 2, 3)

I want to reshape it to (3,1,2), but so that the elements along original dim 3 are now along dim 1. I want the result to look like this:

>>> new_a
array([[[1, 4]],
       [[2, 5]],
       [[3, 6]]])

and when I just use reshape, I get the the right shape, but the elements are in the same order, not what I want:

>>> a.reshape((3,1,2))
array([[[1, 2]],
       [[3, 4]],
       [[5, 6]]])

How can I achieve this?

Divakar · Accepted Answer · 2019-05-21 15:04:20Z

3

Simply use np.transpose -

a.transpose(2,0,1)

Sample run -

In [347]: a
Out[347]: 
array([[[1, 2, 3],
        [4, 5, 6]]])

In [348]: a.transpose(2,0,1)
Out[348]: 
array([[[1, 4]],

       [[2, 5]],

       [[3, 6]]])

Alternatively :

With np.moveaxis -

np.moveaxis(a,2,0)

With np.rollaxis -

np.rollaxis(a,2,0)

answered May 21, 2019 at 15:04

Divakar

222k19 gold badges273 silver badges374 bronze badges

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

Comments

Alec · Accepted Answer · 2019-05-21 15:05:10Z

1

There are a few ways, but transpose() is probably the easiest:

array.transpose(2,0,1)

answered May 21, 2019 at 15:05

Alec

9,7338 gold badges44 silver badges71 bronze badges

Comments

Alleo · Accepted Answer · 2019-06-29 06:59:00Z

0

import einops
einops.rearrange(x, 'x y z -> z x y')

And better use some meaningful axes names instead of x, y, z (like width, height, etc.)

answered Jun 29, 2019 at 6:59

Alleo

8,6382 gold badges46 silver badges34 bronze badges

Collectives™ on Stack Overflow

How to reshape numpy array with reorder?

3 Answers 3

Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related