-----maxima-bounces at math.utexas.edu wrote: -----
>I propose that algsys uses poly_reduced_grobner on equations before
>looking for a solution (and that poly_reduced_grobner autoloads).
(1) I think it would be OK to to this.
(2) The function poly_reduced_grobner does not allow non-integer
rational coefficients (try poly_reduced_grobner([x/5 - y/3,x],
[x,y])). So algsys will need to clean up the equations before sending
them to poly_reduced_grobner. OK, no problem. I don't know what
poly_reduced_grobner does to non-integer algebraic coefficients.
(3) If somebody is ambitious, they could write a new solver that is
specialized to solving the output of poly_reduced_grobner.
(4) I'm not sure that the author of grobner is interested in
maintaining his code. Maybe that doesn't matter -- the author of
algsys hasn't fixed any bugs in that code for the past decade or so.
Barton