An exact penalty approach for mathematical programs with equilibrium constraints.

Abstract

This paper presents an exact penalty approach to solve the mathematical problems with equilibrium constraints (MPECs). This work is based on the smoothing functions introduced in [3] but it does not need any complicate updating rule for the smoothing or penalty parameters. Some numerical academic experiments are carried out to show the efficiency and robustness of this new approach. Two generic applications are also considered: the binary quadratic programs and simple number partitioning problems.