GREAT/MRAT bug, was: alt_eigen more info



On 2012-12-09, Barton Willis <willisb at unk.edu> wrote:

> (%i1) m : matrix([0,-(2*%i*(%i+1))/(2*%i+1),0,-(2*%i*(%i+1))/(2*%i+1)],[(2*%i*(%i+1))/(2*%i+1),0,(2*%i*(%i+1))/(2*%i+1),0],[0,-(2*%i*(%i+1))/(2*%i+1),0,-(2*%i*(%i+1))/(2*%i+1)],[(2*%i*(%i+1))/(2*%i+1),0,(2*%i*(%i+1))/(2*%i+1),0]);

> (%i6) nullspace(m), ratmx : true;
> Maxima encountered a Lisp error:
>  value #:%I4173 is not of the expected type LIST.

A stack trace shows it's in GREAT (actually one of its callees) when it
stumbles. The argument to GREAT is an expression which contains MRAT.
I'm pretty sure I've seen GREAT having trouble with MRAT before (some
kind of a simplification bug, don't remember how to reproduce it now).
I'll look for a simpler example to reproduce the GREAT bug.

best

Robert Dodier