A Framework for Genetic Algorithms in Parallel Environments
Department of Life and Medical Sciences, Doshisha University, Kyoto, Japan
Information Processing Society of Japan (IPSJ) SIG Notes 2011-MPS-84(6), 1-6, 2011
@article{tomoyuki2011framework,
title={A Framework for Genetic Algorithms in Parallel Environments},
author={Tomoyuki, H. and Ryosuke, Y. and Masato, Y. and Mitsunori, M.},
journal={Information Processing Society of Japan (IPSJ) SIG Notes},
volume={2011},
number={6},
pages={1–6},
year={2011}
}
In this research, we developed a framework to execute genetic algorithms (GA) in various parallel environments. GA researchers can prepare implementations of GA operators and fitness functions using this framework. We have prepared several types of communication library in various parallel environments. Combining GA implementations and our libraries, GA researchers can benefit from parallel processing without requiring deep knowledge of different parallel architectures. In the proposed framework, the GA model is restricted to a micro-grained model. In this paper, parallel libraries for a Windows cluster environment, multi-core CPU environment, and GPGPU environment are described. A simple GA was implemented with the proposed framework. Computational performance is also discussed through numerical examples. In this research, we developed a framework to execute genetic algorithms (GA) in various parallel environments. GA researchers can prepare implementations of GA operators and fitness functions using this framework. We have prepared several types of communication library in various parallel environments. Combining GA implementations and our libraries, GA researchers can benefit from parallel processing without requiring deep knowledge of different parallel architectures. In the proposed framework, the GA model is restricted to a micro-grained model. In this paper, parallel libraries for a Windows cluster environment, multi-core CPU environment, and GPGPU environment are described. A simple GA was implemented with the proposed framework. Computational performance is also discussed through numerical examples.
December 20, 2011 by hgpu
