It would be good to implement this directly in the Matrix_gf2e_dense class rather than do anything in the generic implementation. However, you have to be a little careful for the case when it is a 0x0 matrix.
Best, Travis On Thursday, January 21, 2021 at 5:48:40 AM UTC+10 dmo...@deductivepress.ca wrote: > Speeding up by a factor of 4500 in a common use case sounds like a good > idea. Please open a ticket. (Or do you want me to do it?) > > On Tuesday, January 19, 2021 at 11:05:58 PM UTC-7 a.simpl...@gmail.com > wrote: > >> I am running Sagemath 9.1 on macOS 10.15.17 >> but I think this issue is more about math. >> >> Consider the following code >> >> G = GL(2^8, GF(2^8)) >> A = G.random_element() >> B = Matrix(A) >> %time B.is_invertible() >> >> C = Matrix(A) >> %time C.nrows() == C.ncols() == C.rank() >> >> The first timer returns 9s while the second timer returns 2ms. >> I believe that this is because >> >> - is_invertible() computes charpoly and read off the det. >> This is a detour when the base ring is a field. >> - rank() uses m4rie, which uses asymptotically fast algorithm for >> rref. >> >> Perhaps we can add one more if-brach in is_invertible() that exploits >> fast rref? >> > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/3d0ca63b-78de-41ae-8b9f-eeaa8c81eb30n%40googlegroups.com.
