Article ID: | iaor2012329 |
Volume: | 6 |
Issue: | 1 |
Start Page Number: | 127 |
End Page Number: | 139 |
Publication Date: | Jan 2012 |
Journal: | Optimization Letters |
Authors: | Sherali Hanif, Smith J |
Keywords: | optimization, programming: branch and bound |
In this paper, we consider the generation of disjunctive cuts for 0‐1 mixed‐integer programs by conducting a partial exploration of the branch‐and‐bound tree using quick Lagrangian primal and dual updates. We analyze alternative cut generation strategies based on formulating different disjunctions and adopting various choices of normalization techniques, and indicate how these inequalities can also be generated using a projection from a related high‐order lifted formulation obtained via the Reformulation‐Linearization Technique of Sherali and Adams. We conclude by presenting a brief computational study that motivates the potential benefits of this procedure.