Numerical experiments with a population shrinking strategy within an electromagnetism-like algorithm

Abstract

This paper extends our previous work done with a modified Electromagnetism-like (EM) algorithm to a benchmark global optimization collection of test problems. The EM algorithm is a population-based stochastic method that uses an attractionrepulsion mechanism to move sample points towards optimality. The modifications include a local search based on the original pattern search method of Hooke and Jeeves and a shrinking strategy that aims to reduce the population size as the iterative process progresses. Performance profiles are used to compare the proposed modifications with the original EM algorithm considering the average number of function evaluations and the best function value.