Resolution of mathematical program with equilibrium constraints using inexact restauration

Author(s)

João Luiz Chela

• Document type: Doctoral Thesis
• Press: Campinas, SP.
• Institution: Universidade Estadual de Campinas (UNICAMP). Instituto de Matemática, Estatística e Computação Científica
• Defense date: 2006-08-03

Examining board members

Ana Friedlander; Susana Scheimberg de Makler; Nelso Maculan Filho; Alfredo Iusem; Sandra Augusta Santos

• Advisor: Roberto Andreani; Ana Friedlander

Abstract

A Mathematical Program with Equilibrium Constraints (MPEC) is an optimization problem, where part of the variables are constrained to be solutions of a variational inequality problem parameterized by the other variables. The reformulation of a MPEC, as a classical optimizatlon problem, replacing the variational inequality problem by corresponding the K.K. T system, is also called MPEC. In this context the variational inequality problem is also called the second leveI problem. MPEC problems are harder to solve than classical optimization problems due to their two-level structure. These problems are non-convex, and the feasible region can even be a disconnected one. The objective function of the first level is in general non-differentiable, even in the case where the second level solutions can be expressed as a function of the parameters. In this work, to solve Mathematical Programming Problems we use an Algorithm of Inexact Restoration based in the work of Martínez in [50]. This approach allows to treat the second leveI problem design without reformulation and we do not need any special algorithm designed for non-differentiable optimization. We present theoretical results and numerical experiments, including an application in urban traffic problems (AU)