Nonmonotone hybrid tabu search for Inequalities and equalities: an experimental study

Abstract

The main goal of this paper is to analyze the behavior of nonmonotone hybrid tabu search approaches when solving systems of nonlinear inequalities and equalities through the global optimization of an appropriate merit function. The algorithm combines global and local searches and uses a nonmonotone reduction of the merit function to choose the local search. Relaxing the condition aims to call the local search more often and reduces the overall computational effort. Two variants of a perturbed pattern search method are implemented as local search. An experimental study involving a variety of problems available in the literature is presented.