Hi everyone, > We do have A._multiply_linbox(B), but we never > use it by default, since when we first wrapped it sadly turned out > to be slower than using our own multi-modular implementation. > This is the sort of thing that may change someday, I hope... Yep, that should change pretty soon!
I was actually planning to spend some time looking at this pb of small matrices with huge ints. Strassen-Winograd is fine for dimensions 2^k but if with odd dimension, it can be better to use other algorithms (3x3 splitting for example, Pan algorithms, Winograd PreStrassen algorithm, and some personnal ideas too). Therefore it would be nice to have a kind of database of the best algorithm (in term of # of mul) for each small dimension (2,3,4,5,6,7...) and have a preprocessing phase that would choose the good combination of these algorithms, for any given n. Clement > --- > > It could make a lot of sense to make _matrix_times_matrix > use strassen when the heights of the input matrices are > large, instead of using classical O(n^3) multiplication. > > Willliam > > > > --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
