Possible bug, converges to very non-optimal solution

[svigerske]

Issue created by migration from Trac.

Original creator: partaniamit

Original creation time: 2016-03-25 18:58:17

Assignee: @tkralphs

Version: 5.6.13

I am using symphony to solve some simple and some complex MILPs. For the attached problem (MPS) the lower bound and upper bound converge to a solution which is far away from optimum. All other solvers (CBC, CPLEX, Gurobi) give the right answer which is roughly 0.004, but Symphony gives 0.5 as answer.

Another possible bug is that sometime for very simple milps, preprocessing results infeasibility when I use callable library, but not in the interactive mode. When I set prep_level to -1 (no preprocessing) it gives the right answer. I am attaching the mps file.

[svigerske]

Attachment Non_optimal.MPS by partaniamit created at 2016-03-25 18:58:46

[svigerske]

Attachment Preprocessing_bug.MPS by partaniamit created at 2016-03-25 18:59:03

[svigerske]

Attachment Non_optimal_scaled.MPS by partaniamit created at 2016-03-25 18:59:18

[svigerske]

Comment by @tkralphs created at 2016-03-26 14:26:01

Changing status from new to assigned.