Um Algoritmo Híbrido Paralelo Cooperativo Baseado em DE, PSO e AG: Uma Avaliação em Computadores Multicore

Abstract

This paper presents a new parallel hybrid algorithm combining Particle Swarm Optimization (PSO), Differential Evolution (ED) and Genetic Algorithms (GA) for optimizing unconstrained numerical functions. Basically, PSO and ED evolves independently in our proposal, then they cooperate between them exchanging their best individual, which undergo GA operators locally in order to search different areas in the search space. The results are evaluated concerning the quality of the solution and speedup against six benchmarks functions. The outcomes are compared with GA, PSO and ED in the serial version and with the classical island model of PSO and DE, considering two and four threads. A comparison with a hybrid algorithm is done as well.

A Meta-Heuristic Based on Genetic Algorithm for Selecting Bailiffs by Districts

“…The fitness function must represent the quality of each individual in a particular problem to be solved [4], [14]. However, as previously mentioned, our approach introduces two more fitness functions: the population and gene fitnesses.…”

“…In other words, there are no applications related to our problem. Thereby, our proposal is to solve the problem using a Genetic Algorithm (GA) [4,13,14]. Indeed, we used a specific GA called selection-mutation GA [1].…”