Computational Experience with an Interior Point Cutting Plane Algorithm

There has been a great deal of success in the last twenty years with the use of cutting plane algorithms to solve specialized integer programming problems. Generally, these algorithms work by solving a sequence of linear programming relaxations of the integer programming problem, and they use the simplex algorithm to solve the relaxations. In this paper, we describe experiments using a predictor-corrector interior point method to solve the relaxations. For some problems, the interior point code requires considerably less time than a simplex based cutting plane algorithm.