An improved algorithm for combinatorial multi-parametric quadratic programming

Abstract

The goal of multi-parametric quadratic programming (mpQP) is to compute analytic solutions to parameter-dependent constrained optimization problems, e.g., in the context of explicit linear MPC. We propose an improved combinatorial mpQP algorithm that is based on implicit enumeration of all possible optimal active sets and a simple saturation matrix pruning criterion which uses geometric properties of the constraint polyhedron for excluding infeasible candidate active sets. In addition, techniques are presented that allow to reduce the complexity of the discussed algorithm in the presence of symmetric problem constraints. Performance improvements are discussed for two example problems from the area of explicit linear MPC.