A Combination of CMAES-APOP Algorithm and Quasi-Newton Method

Abstract

In this paper, we present an approach for combining the CMAES-APOP with a local search in order to make a hybrid evolutionary algorithm. This combination is based on the information of population size in the evolution process of the CMAES-APOP algorithm while the local search is quasi-Newton line search algorithm. We will give some conditions to efficiently active the local search inside CMAES-APOP. Some numerical experiments on multi-modal optimization problems will show the efficiency of proposed approach.