[Help-glpk] Re: unbounded ray

[Andrew Makhorin]

>A patch would be good.
>I'm planning to do Bender's decomposition without a known feasible solution.
>Infeasible solutions generate extreme rays in the dual problem.

I've included the api routine lpx_get_ray_info in glpk 4.3 (released
to-day). This routine being called after lpx_simplex allows determining
which non-basic variable causes primal unoundness (note, however, that
if there are several such variables, only one of them is reported). Its
description can be found in the reference manual.

Andrew Makhorin