>>>>> "Bob" == Bob McElrath <bob+maxima@mcelrath.org> writes:
Bob> equal to the machine precision. Algorithms like multiply and
Bob> determinant should be able to operate on them using octave/lapack
Bob> algorithms.
Although I haven't tried it, I suspect f2cl could convert lapack. The
result won't be as fast as an FFI to lapack, but might be good enough
for maxima. I do know that some BLAS routines and maybe one LAPACK
routine have been converted successfully.
FWIW, I'm not actually proposing that maxima use an f2cl'd version of
lapack. But perhaps slow is better than not at all.
Ray