numpy appending a 2D array to a 3D array

Question

i’d appreciate guidance on appending a 2D zeros array to a 3D array. so as to add multiple layers to a 2×2 matrix.

#so, 
n = np.zeros((1,3,3,), np.ubyte)
n[0][0][1] = n[0][1][0] = 1
#return prints:
[[[0, 1, 0],
  [1, 0, 0],
  [0, 0, 0]]]

then how do you add another 2D array of zero elements to, in pseudocode:

a = np.zeros((3,3,), np.ubyte)
n = np.append(n, a)
#return prints:
 [[[0 1 0]
   [1 0 0]
   [0 0 0]]

  [[0 0 0]
   [0 0 0]
   [0 0 0]]]

and so on with the latter two lines to add any number of layers of the 3×3 2D matrix.

thank you in advance, lucas

Asked By: lucas

||

Answer 1

Generally, the arrays need to have the same shape. To join along an existing axis, you can use np.concatenate:

np.concatenate([n, a[None,...]], axis=0)

Note, this adds an extra dimension to a, or equivalently, we can extend along a new axis using np.stack:

np.stack([n[0], a], axis=0)

Note, in the above case, we removed a dimension from n.

As noted in the comments though, this is inefficient, this will scale very inefficiently (quadratic time on the total size), so keep that in mind.

Note, bot arr[0] and arr[None, ...] are "cheap" operations (constant time) and don’t copy the underlying buffer, they create views.

Question: