Open eschnett opened 12 years ago
Yes. The fix would be to extend MatrixInverse to handle the cases that it does not currently handle. I've never been happy with MatrixInverse, as it is a bit confusing. We should think of a better way of computing the inverse of a matrix in a way which maintains the index positions and doesn't look "wrong".
It is not possible to calculate the inverse of J[ua,lb]; Kranc expects all indices to be lower when calling MatrixInverse.
A work-around is to define a shorthand with all lower indices and to contract with KD to "lower" the index.