Repairing Misbehaving Mathematical Programming Models: Concepts and a GAMS-Based Approach
- Additional Document Info
- View All
Mathematical-programming formulations can yield faulty answers. Models can be unbounded, infeasible, or optimal with unrealistic answers. I develop techniques for screening mathematical-programming formulations for structural problems pre-and postsolution. The presolution approaches identify problems within single variables and constraints. The post-solution techniques may require model augmentation and rely on theory-based examination of primal and dual solutions. I demonstrate these approaches in the context of linear programming and have computerized them in association with GAMS. They are freely distributed through a web page.
author list (cited authors)