Matrix diagonal part

1

What might be a possible analog for https://www.tensorflow.org/api_docs/python/tf/linalg/diag_part in pytorch?

I didnot manage to find something in docs :frowning:

2

I don’t think that there is an exact equivalent. But here is a workaround:

PS: this example is the same exposed in the link you provided.

import torch
x = torch.Tensor([
                  [[1, 0, 0, 0],
                   [0, 2, 0, 0],
                   [0, 0, 3, 0],
                   [0, 0, 0, 4]],
                  [[5, 0, 0, 0],
                   [0, 6, 0, 0],
                   [0, 0, 7, 0],
                   [0, 0, 0, 8]]
		         ])
y = torch.stack(tuple(t.diag() for t in torch.unbind(x,0)))
3

I think now torch.diagonal serves for same purpose.

import torch
a = np.array([[[1, 2, 3, 4],  # Input shape: (2, 3, 4)
               [5, 6, 7, 8],
               [9, 8, 7, 6]],
              [[5, 4, 3, 2],
               [1, 2, 3, 4],
               [5, 6, 7, 8]]])

a = torch.from_numpy(a)
# a.size() 
torch.diagonal(a, offset=0, dim1=-2, dim2=-1)
# torch.diagonal(b_tensor, offset=0, dim1=1, dim2=2)  # this return same results, set dim

return:

tensor([[1, 6, 7],
        [5, 2, 7]])

More details can be found in: torch — PyTorch master documentation