Solving Generalized Semi-Infinite Programming Problems With A Trust Region Method

Abstract

In this paper, a trust region method for generalized semi-infinite programming problems is presented. The method is based on [O. Yi-gui, "A filter trust region method for solving semi-infinite programming problems", J. Appl. Math. Comput. 29, 311 (2009)]. We transformed the method from standard to generalized semi-infinite programming problems. The semismooth reformulation of the Karush-Kuhn-Tucker conditions using nonlinear complementarity functions is used. Under some standard regularity condition from semi-infinite programming, the method is convergent globally and superlinearly. Numerical examples from generalized semi-infinite programming illustrate the performance of the proposed method.