Re: [Help-glpk] problem with glpsol

[Andrew Makhorin]

> I use glpsol and find the following solution. 

> Is this solution correct (row 49 and row 50).

No, the solution is wrong as indicated by KKT:

> Karush-Kuhn-Tucker optimality conditions:

> KKT.PB: max.abs.err. = 2.52e+001 on row 22
>         max.rel.err. = 6.17e-001 on row 22
>         PRIMAL SOLUTION IS INFEASIBLE

This might happen because the lp is badly scaled, and the simplex
solver missed some constraint(s) due to a tolerance.

Could you write your instance in mps or cplex format, gzip it, and
post it to me? This would help to see what is wrong. Thanks.


Andrew Makhorin