I have tried to use GLPK to solve some large MILP problems, but without
much success. The glpk-20.lp file at
when solved give a final solution that is not feasible, although one
exists. Is there some way to coax some parameters of GLPK or the problem
itself to get an optimal feasible solution of such large MILP problem?
Is the size of such a problem beyond the capability of the MILP solver
Even after doing some modifications such as reducing the coefficients of
the objective function, or replacing the value 30000 to 300 in the
constraints, no feasible solution could be obtained. I do not see any
scaling that would make the GLPK solver find an optimal feasible solution.
It appears that GLPK MILP searching strategy concedes too early to a
solution that does not satisfy the constraints although reaching the
maximum possible value of the objective function.
-- Mario Latendresse
Bug-glpk mailing list