Adapting distributed evolutionary algorithms to heterogeneous hardware

Distributed computing environments are nowadays composed of many heterogeneous computers able to work cooperatively. Despite this, the most of the work in parallel metaheuristics assumes a homogeneous hardware as the underlying platform. In this work we provide a methodology that enables a distribut...

Descripción completa

Detalles Bibliográficos
Autores: Salto, Carolina, Alba, Enrique
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2015
País:Argentina
Institución:Consejo Nacional de Investigaciones Científicas y Técnicas
Repositorio:CONICET Digital (CONICET)
Idioma:inglés
OAI Identifier:oai:ri.conicet.gov.ar:11336/114217
Acceso en línea:http://hdl.handle.net/11336/114217
Access Level:acceso abierto
Palabra clave:distributed computing
heterogeneity
parallel algorithms
metaheuristics
https://purl.org/becyt/ford/1.2
https://purl.org/becyt/ford/1
Descripción
Sumario:Distributed computing environments are nowadays composed of many heterogeneous computers able to work cooperatively. Despite this, the most of the work in parallel metaheuristics assumes a homogeneous hardware as the underlying platform. In this work we provide a methodology that enables a distributed genetic algorithm to be customized for higher efficiency on any available hardware resources with different computing power, all of them collaborating to solve the same problem. We analyze the impact of heterogeneity in the resulting performance of a parallel metaheuristic and also its efficiency in time. Our conclusion is that the solution quality is comparable to that achieved by using a theoretically faster homogeneous platform, the traditional environment to execute this kind of algorithms, but an interesting finding is that those solutions are found with a lower numerical effort and even in lower running times in some cases.