I am using PyTorch on Windows 10, with Python 3.6.6 and Cuda 9.0.
I would like to make batch inversion of matrices. Since this is not supported by torch.inverse
, I use the following:
def b_inv(b_mat):
eye = b_mat.new_ones(b_mat.size(-1)).diag().expand_as(b_mat)
b_inv, _ = torch.gesv(eye, b_mat)
return b_inv
But I get the following error:
torch.gesv MAGMA library not found in compilation. Please rebuild with MAGMA.
What does that mean? Does it mean PyTorch has to be built from sources to use MAGMA?