(Numpy) How to stack 4 arrays into one?-CodePudding

i have a list of 1000 numpy arrays (32x32) each, so

array.shape = (1000, 32, 32). How do i best go about stacking the entire list into a 64x64 array for example, so that the result is (250, 64, 64)? Hereby, the first 4 (32, 32) arrays are concatenated like so:

1   |   2
----------
3   |   4

This would mean that I end up with 250, 64x 64 arrays. Can anyone help out?

CodePudding user response：

You can also go over the colon notation here. This should have the same output as if you iterate over the arrays:

import numpy as np
large_array = np.zeros((1000, 32, 32))
output_array = np.zeros((250, 64, 64))

# Extract each block 1, 2, 3 and 4 and write to output array
# '::4' select only every fourth array
output_array[:, 0:32,  0:32] = large_array[0::4, ...]  # Blocks: '1'
output_array[:, 32:64, 0:32] = large_array[1::4, ...]  # Blocks: '2'
output_array[:, 0:32,  32:64] = large_array[2::4, ...] # Blocks: '3'
output_array[:, 32:64, 32:64] = large_array[3::4, ...] # Blocks: '4'

CodePudding user response：

This is a method that still uses a loop but does not create any new array beside the final one that gets filled iteratively

yourlist.shape == (1000,32,32)
new = np.empty((250, 64, 64))
s1 = slice(0, 32) # you can substitute it below with the standard :32 in the arrays below
s2 = slice(32,64)
for i in range(0, 1000, 4):
   a, b, c, d = yourlist[i:i 4]
   new[i // 4, s1, s1] = a
   new[i // 4, s1, s2] = b
   new[i // 4, s2, s1] = c
   new[i // 4, s2, s2] = d