A Framework for Constraint Programming Based Column Generation

Column generation is a state-of-the-art method for optimally solving di cult large-scale optimization problems such as airline crew assignment. We show how to apply column generation even if those problems have complex constraints that are beyond the scope of pure OR methods. We achieve this by formulating the subproblem as a constraint satisfaction problem (CSP). We also show how to e ciently treat the special case of shortest path problems by introducing an e cient path constraint that exploits dual values from the master problem to exclude nodes that will not lead to paths with negative reduced costs. We demonstrate that this propagation signi cantly reduces the time needed to solve crew assignment problems.

• Type: Conference
• Year: 1999
• Where: CP
• Authors: Ulrich Junker, Stefan E. Karisch, Niklas Kohl, Bo Vaaben, Torsten Fahle, Meinolf Sellmann
• Keywords: Artificial Intelligence, Column Generation, CP 1999, Crew Assignment, Large-scale Optimization Problems