October 28, 2008 at 14h30 in the Salle des Conférences, IMB

Given an optimization problem, many mathematical programming formulations can describe its solution set. Moreover, the performance of many solution algorithms depend on the formulation more than on the problem itself. Reformulations are used to transform mathematical programs so that they conserve the mathematical properties of the original formulation, yet they perform in some sense “better” than the original formulation. We shall provide a general theoretical framework for reformulations, including definitions and main properties; we shall then illlustrate an application of reformulations to the problem of identifying and removing symmetries in mathematical programs.