[GiNaC-devel] Speeding up simplify_indexed(): 2nd try

Dear all, Here is a new attempt at the patch. The problem with the previous one was that if one symmetrizes only amongst indices of precisely the same type, also renaming of indices should be done taking these types into account. Another advantage of this is that if a user uses for instance mu and nu for indices with a variance, these will no longer end up being used as indices without variance. I think this makes output more readable. Best, Chris