Support matrix power for lazy tensor

[huguesva]

Hello,

Do you know if it is possible to compute the matrix power of a pykeops torch lazytensor?
I tried via matrix multiplication @ and get the following error:

import torch 
from torchdr import pairwise_distances

X = torch.randn(100, 10)
Y = torch.randn(100, 10)

dist, _ = pairwise_distances(X, Y, backend="keops")    # Compute Euclidean dist as lazy tensor
print(dist @ dist)

gives the error:

AttributeError: 'LazyTensor' object has no attribute 'view'

Thanks a lot!