Combining numpy multi-dimensional arrays

Question

I.m facing a little issue to combine arrays in a certain manner. Let's say we have

a=array([[1,1,1],[2,2,2],[3,3,3]])

b=array([[10,10,10],[20,20,20],[30,30,30]])

I wish to get

c=array([[[1,1,1],[10,10,10]],[[2,2,2],[20,20,20]],[[3,3,3],[30,30,30]]])

The real issue is that my arrays a and b are much longer than 3 coordinates!

The best I achieved using concatenate is:

concatenate((a,b),axis=2)

which results in

array([[ 1, 1, 1, 10, 10, 10], [ 2, 2, 2, 20, 20, 20], [ 3, 3, 3, 30, 30, 30]])

it is pretty good but not have enough depth.

Also, I've tried something from another question to get the desired depth:

d=concatenate((a[...,None],b[...,None]),axis=2)

but results in:

 array([[[ 1, 10],
    [ 1, 10],
    [ 1, 10]],

   [[ 2, 20],
    [ 2, 20],
    [ 2, 20]],

   [[ 3, 30],
    [ 3, 30],
    [ 3, 30]]])

Which still does not works...

Joran Beasley · Accepted Answer · 2013-11-13 16:33:07Z

6

ummm zip(a,b) ?

is not what you want??

>>> a=array([[1,1,1],[2,2,2],[3,3,3]]);b=array([[10,10,10],[20,20,20],[30,30,30]
>>> zip(a,b)
[(array([1, 1, 1]), array([10, 10, 10])), (array([2, 2, 2]), array([20, 20, 20])), (array([3, 3, 3]), array([30, 30, 30]))]

answered Nov 13, 2013 at 16:33

Joran Beasley

114k13 gold badges167 silver badges187 bronze badges

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

3 Comments

user2988577 Over a year ago

Thank you! I was looking for array(zip(a,b))

Jaime Over a year ago

If you want the output to be an array, i.e. if you do np.array(zip(a, b)), this is over x10 slower than @BiRico's approach for larger arrays.

Joran Beasley Over a year ago

yeah thats probably a reasonable statement ... what about if you dont need to convert it back into a np.array? (Im too lazy to check)

Bi Rico · Accepted Answer · 2013-11-13 16:42:50Z

5

It seems like you want to add a new axis between 0 and 1 so put the None in the middle. This will shift axis 1 be axis 2 and create a new dimension at 1. Like so:

a = array([[1,1,1],[2,2,2],[3,3,3]])
b = array([[10,10,10],[20,20,20],[30,30,30]])
c = concatenate((a[:, None, :], b[:, None, :]), axis=1)

>>> c
array([[[ 1,  1,  1],
    [10, 10, 10]],

   [[ 2,  2,  2],
    [20, 20, 20]],

   [[ 3,  3,  3],
    [30, 30, 30]]])

answered Nov 13, 2013 at 16:42

Bi Rico

25.9k3 gold badges57 silver badges75 bronze badges

Comments

Jeremiah O'Neil · Accepted Answer · 2016-02-09 04:08:17Z

You're looking for numpy.stack. It's used for joining arrays along a new axis; in contrast to 'numpy.concatenate', which is for joining arrays along an existing axis. With stack, you specify the axis to join along in terms of which axis it would be after the stacking; so you would specify axis 1.

a = array([[1,1,1],[2,2,2],[3,3,3]])
b = array([[10,10,10],[20,20,20],[30,30,30]])
c = stack((a, b), axis=1)

>>> c
array([[[ 1,  1,  1],
    [10, 10, 10]],

   [[ 2,  2,  2],
    [20, 20, 20]],

   [[ 3,  3,  3],
    [30, 30, 30]]])

Collectives™ on Stack Overflow

Combining numpy multi-dimensional arrays

3 Answers 3

3 Comments

Comments

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

3 Comments

Comments

Comments

Linked

Related