On the merit and penalty functions for the D.C. optimization

Abstract

This paper addresses a rather general problem of nonlinear optimization with the inequality constraints and the goal function defined by the (d.c.) functions represented by the difference of two convex functions. In order to reduce the constrained optimization problem to an unconstrained one, we investigate three auxiliary problems with the max-merit, Lagrange and penalty goal functions. Further, their relations to the original problem are estimated by means of the new Global Optimality Conditions and classical Optimization Theory as well as by examples.