Affine Transform bug in torch.distributions

Hi guys,

I have the distribution of a MxN random matrix, say A, and I want to get the distribution of Av, where v is a fixed N-dimensional vector. Obviously, the output of this should be an M dimensional vector.

But, when I use the command,

TransformedDistribution( A,  AffineTransform(0, v))

I get again a distribution over MxN matrices. It is as if the TransformedDistribution does not work.

Any ideas on why is this happening?

Thank you