Estudo de configurações de modelos híbridos de ilhas para obtenção de uma ou mais soluções em otimização via meta-heurísticas

• Autor(a) principal: Magalhães, Thiago Tavares
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do LNCC
• Texto Completo: https://tede.lncc.br/handle/tede/239
• topic: Otimização matemática; Mathematical optimization; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::MATEMATICA DA COMPUTACAO::MODELOS ANALITICOS E DE SIMULACAO

Optimization is a permanent subject for research, aiming to provide solutions that improve the utilization of resources and activities of various kinds. Due to the advances in distributed computational processing architectures, analysis and suggestion of parallelizable techniques for treatment of problems has been increasingly highlighted. Among them, the model known as ``island model'', which proposes the achievement of better optimized results in shorter execution time,and brings in a new set of parameters and possibilities, deserves special attention. Thus, this work presents interconnected sets of experiments and proposals that complement the literature concerning 1) the possibilities of model hybridization and the best strategies for hybridization, 2) the study of the parameter known as ``migration policy'' and its influence on diversification and exploitation, and 3) the applicability of the model and of the different implemented migration policies in the context of optimization aiming at several optima. In this way, this work joins the knowledge from the literature and experiments and suggestions that are not founded in previous researches. From these results it is expected that new implementations or applications of island hybrid models for optimization via metaheuristics, aiming at one or more optima, can be more efficiently developed.