摘要
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.
出处
《计算机学报》
EI
CSCD
北大核心
1999年第6期591-595,共5页
Chinese Journal of Computers