Getting diagonal matrix stripe automatically in numpy or pytorch

Question

I need to get a diagonal stripe of the matrix (not sure about the terminology here, diagonal matrix stripe seems to describe it best).

Say, I have a matrix of size K

Answer 1

Extending Paul's answer. You can do the same in PyTorch using diag multiple times (I do not think there is any direct function to do strides in PyTorch)

In [1]: import torch

In [2]: def stripe(a):
   ...:     i, j = a.size()
   ...:     assert(i>=j)
   ...:     out = torch.zeros((i-j+1, j))
   ...:     for diag in range(0, i-j+1):
   ...:         out[diag] = torch.diag(a, -diag)
   ...:     return out
   ...: 

In [3]: a = torch.randn((6, 3))

In [4]: a
Out[4]: 

 0.7669  0.6808 -0.6102
-1.0624 -1.2016 -0.7308
 1.4054 -1.0621  0.2618
-0.9505 -0.9322 -0.4321
-0.0134 -1.3684  0.1883
-0.8499  0.2533 -0.3976
[torch.FloatTensor of size 6x3]

In [5]: stripe(a)
Out[5]: 

 0.7669 -1.2016  0.2618
-1.0624 -1.0621 -0.4321
 1.4054 -0.9322  0.1883
-0.9505 -1.3684 -0.3976
[torch.FloatTensor of size 4x3]