Convert Numpy array of square arrays to numpy square array, conserving "layout"

Question

First time i ask something here. I am kind of 'blocked'.

I have an array composed of n x n arrays (lets take n=3 for simplification):

[
    [
        [ 0,  1,  2],
        [ 3,  4,  5],
        [ 6,  7,  8]
    ],
   [
        [9, 10, 11],
        [12, 13, 14],
        [15, 16, 17]
    ],
   [
        [18, 19, 20],
        [21, 22, 23],
        [24, 25, 26]
    ]
        ]

(altho my array contains a lot more than 3 3*3 arrays)

i want to achieve a 2D array like this :

[0,1,2,9,10,11,18,19,20]  
[3,4,5,12,13,14,21,22,23]    
[6,7,8,15,16,17,24,25,26]

Is there a trick i haven't thought of because i cant think of any way to accomplish the transformation.

Thanks

pthibault · Accepted Answer · 2018-03-07 21:23:07Z

2

Slightly cleaner than moveaxis maybe:

import numpy as np
a = np.arange(27).reshape(3,3,3)
a.swapaxes(0,1).reshape(3,-1)
array([[ 0,  1,  2,  9, 10, 11, 18, 19, 20],
   [ 3,  4,  5, 12, 13, 14, 21, 22, 23],
   [ 6,  7,  8, 15, 16, 17, 24, 25, 26]])

answered Mar 7, 2018 at 21:23

pthibault

5245 silver badges6 bronze badges

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

Comments

hpaulj · Accepted Answer · 2018-03-07 21:36:11Z

1

Think of this as a list of 3 arrays, that you want to join horizontally:

In [171]: arr = np.arange(27).reshape(3,3,3)
In [172]: np.hstack(arr)
Out[172]: 
array([[ 0,  1,  2,  9, 10, 11, 18, 19, 20],
       [ 3,  4,  5, 12, 13, 14, 21, 22, 23],
       [ 6,  7,  8, 15, 16, 17, 24, 25, 26]])

In [173]: arr
Out[173]: 
array([[[ 0,  1,  2],
        [ 3,  4,  5],
        [ 6,  7,  8]],

       [[ 9, 10, 11],
        [12, 13, 14],
        [15, 16, 17]],

       [[18, 19, 20],
        [21, 22, 23],
        [24, 25, 26]]])

Of I like to test ideas with arrays with differing dimensions. Then the various axis manipulations becomes more obvious.

In [174]: arr = np.arange(24).reshape(2,3,4)
In [175]: arr
Out[175]: 
array([[[ 0,  1,  2,  3],
        [ 4,  5,  6,  7],
        [ 8,  9, 10, 11]],

       [[12, 13, 14, 15],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
In [176]: np.hstack(arr)
Out[176]: 
array([[ 0,  1,  2,  3, 12, 13, 14, 15],
       [ 4,  5,  6,  7, 16, 17, 18, 19],
       [ 8,  9, 10, 11, 20, 21, 22, 23]])

In [177]: np.vstack(arr)
Out[177]: 
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11],
       [12, 13, 14, 15],
       [16, 17, 18, 19],
       [20, 21, 22, 23]])

But there's nothing wrong with the variations on the transpose and reshape answers, if starting from a 3d array (as opposed to a list of arrays):

In [187]: arr.transpose(1,0,2).reshape(3,-1)
Out[187]: 
array([[ 0,  1,  2,  9, 10, 11, 18, 19, 20],
       [ 3,  4,  5, 12, 13, 14, 21, 22, 23],
       [ 6,  7,  8, 15, 16, 17, 24, 25, 26]])

edited Mar 7, 2018 at 21:36

answered Mar 7, 2018 at 21:27

hpaulj

233k14 gold badges260 silver badges392 bronze badges

2 Comments

Yannic Hamann Over a year ago

should be probably np.stack instead

hpaulj Over a year ago

@YannicHamann I initially suggested stack, but in this case it doesn't help. hstack or `concatenate on axis 1 is so much simpler.

Paul Panzer · Accepted Answer · 2018-03-07 21:38:23Z

1

You could use np.block

>>> import numpy as np
>>> X = np.arange(27).reshape(3, 3, 3)
>>> 
>>> np.block(list(X))
array([[ 0,  1,  2,  9, 10, 11, 18, 19, 20],
       [ 3,  4,  5, 12, 13, 14, 21, 22, 23],
       [ 6,  7,  8, 15, 16, 17, 24, 25, 26]])

answered Mar 7, 2018 at 21:38

Paul Panzer

53.3k3 gold badges59 silver badges103 bronze badges

Comments

ascripter · Accepted Answer · 2018-03-07 21:15:35Z

0

A simple reshape doesn't suffice since you have to change the order of the axes first:

import numpy as np
a = np.array([[[0,1,2],[3,4,5],[6,7,8]],[[9,10,11],[12,13,14],[15,16,17]],[[18,19,20],[21,22,23],[24,25,26]]])
np.moveaxis(a, 0, 1).reshape(3,9)

[[ 0  1  2  9 10 11 18 19 20]
 [ 3  4  5 12 13 14 21 22 23]
 [ 6  7  8 15 16 17 24 25 26]]

answered Mar 7, 2018 at 21:15

ascripter

6,31512 gold badges54 silver badges74 bronze badges

1 Comment

Mik K Over a year ago

thank you very much ascripter. I hadnt read about moveaxis. Thanks again

Collectives™ on Stack Overflow

Convert Numpy array of square arrays to numpy square array, conserving "layout"

4 Answers 4

Comments

2 Comments

Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

Comments

2 Comments

Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related