I would like to reshape an 2D numpy array in a 3D array consisting of 2x2 (or in general PxP) blocks, respecting the spatial positions.
x = np.arange(100).reshape(10,10)
x
array([[ 0, 1, 2, ..., 7, 8, 9],
[10, 11, 12, ..., 17, 18, 19],
[20, 21, 22, ..., 27, 28, 29],
...,
[70, 71, 72, ..., 77, 78, 79],
[80, 81, 82, ..., 87, 88, 89],
[90, 91, 92, ..., 97, 98, 99]])
y = some_reshape_and_transpose(x)
y[0]
array([[ 0, 1],
[10, 11]])
y[1]
array([[ 2, 3],
[12, 13]])
y.shape
(25, 2, 2)
The precise order of the blocks along the first dimension of the 3D tensor is not important. The reverse command would be appreciated as well !
