Open singerng opened 5 months ago
Hello, thank you for the report. Indeed our "blackbox" method modulo 2 will not work at all, but still I can reproduce this strange "silent" fail. I'll investigate ...
In the meantime, if you want that kind of rank via Wiedemann's iteration, one way is to ask for a certification as follows:
Method::Blackbox meth; meth.certifyInconsistency=true; rank(rankResult, M, meth);
This will setup an extension field and produce the rank with much higher probability.
When I try to calculate the rank of a large blackbox sparse matrix over GF2, the code is always outputting zero. Here is an MWE:
Is this known behavior because the blackbox solvers are randomized and fail if the field size is small? Or is something else going on? I thought maybe I am supposed to use the
GF2
field class directly but can't seem to manage to make that work either.If the matrix is smaller or if I use SparseElimination instead, everything works fine.