ParaGA:一个并行遗传算法的C++类库

PARAGA: A C++ CLASS LIBRARY FOR PARALLEL GENETIC ALGORITHM

介绍了一个基于ＰＶＭ并行环境的并行遗传算法的Ｃ＋＋类库ＰａｒａＧＡ的设计和实现．ＰａｒａＧＡ以使用方便和灵活为主要目标，提供了透明的并行机制，使不具有并行程序设计经验的用户可以方便地编写并行遗传算法的程序．高级用户也可通过类库提供的若干方法来获得优化的并行性能．类库采用粗粒度模型，支持并行遗传算法的３种迁移模式及ＳＰＭＤ和Ｍａｓｔｅｒ／Ｓｌａｖｅ两种编程模式．ＰａｒａＧＡ也提供了实现负载平衡分配及利用ＰＶＭ快速通信机制的方法．

A PVM based C++ class library for parallel genetic algorithm programming, ParaGA, is designed and implemented. ParaGA takes convenience and flexibility as its main objectives. By offering transparent parallel implementation for genetic algorithm, the library allows users without parallel programming experience to write parallel GA programs. However, experienced programmers can use various methods supplied by ParaGA to get optimal parallel performance. Based on coarse granularity model, ParaGA supports to write both SPMD and Master/Slave parallel programs with three optional migration methods. ParaGA also provides methods for programs to achieve balance load distribution and make use of the fast communication mechanism of PVM.