Hi, Most questions about finite fields matmul have been answered, in short: - one could quickly fix the pb that sage does not use linbox by default (I'm surprised by this fact) - in the current state, operands are converted from ints to double, hence a overhead in both time and memory. - Burcin & I started to address this by rewriting the matrix-modn- dense class using floating point representation: a prototype code was working but we wen't into troubles when polishing it and making it work with the coercion system. Unfortunately I never found time to get back to it since then. - Further timing improvements are expected by using floats instead of doubles for tiny fields (eg roughly GF(p<100)) with no pain (already available in LinBox) - Non prime fields are supported in LinBox, as long as their cardinality remains small (LinBox implementation has to be improved following http://hal.archives-ouvertes.fr/hal-00315772 Theorem 2) but the wrapping in sage is missing. So I think it might probably be simpler to wrapp it rather than creating a new spkg with MeatAxe.
Clément PS: weird, my reply by email has been rejected and I'm now posting via the web interface. On Jul 13, 3:11 pm, Simon King <simon.k...@uni-jena.de> wrote: > Hi Martin, > > On 13 Jul., 15:03, Martin Albrecht <martinralbre...@googlemail.com> > wrote: > > > > BTW, Sage's current implementation needs 906.69 s for a > > > single multiplication. > > > We suck at this it seems :) > > Totally. > > According to the Boothby-Bradshaw article you were referring to, their > ideas were supposed to be implemented in Sage. Is that done, yet? I > wouldn't like to double the work... > > Best regards, > Simon -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
