Speed up indices method in @symtensor
Hi,
This is kind of a minor issue, but I wanted to suggest a change to the indices method in @symtensor that seems to improve its performance considerably. The new version uses the built-in nchoosek function to calculate the indices, and, to calculate the scaling factors, uses a way that is fastes and does not need to calculate the multinomial. I have included the previous version in tests\TestSymTensor
, and added a new test that compares the old version with the new. Followingly I will submit a merge request with these changes.