计算机与现代化 ›› 2011, Vol. 1 ›› Issue (1): 49-3.doi: 10.3969/j.issn.1006-2475.2011.01.014

• 算法分析与设计 • 上一篇    下一篇

并行差分进化算法

丛婷婷,祝永志   

  1. 曲阜师范大学计算机科学学院,山东 日照 276826
  • 收稿日期:2010-09-01 修回日期:1900-01-01 出版日期:2011-01-20 发布日期:2011-01-20

Parallel Differential Evolution Algorithm

ONG Ting-ting, ZHU Yong-zhi   

  1. Computer Science College, Qufu Normal University, Rizhao 276826, China
  • Received:2010-09-01 Revised:1900-01-01 Online:2011-01-20 Published:2011-01-20

摘要:

并行处理已经成为现代计算技术的一项关键技术。近几年软件的发展使得异构计算机集合可以作为并行计算资源使用,本文探讨如何在环形网络拓扑结构中将差分进化算法并行,以提高该方法的速度和性能。实验结果表明,子群中交流信息分配到不同节点的范围对该算法的性能有重大影响。此外,并不是差分进化中所有的突变策略都同样对此参数的值敏感。

关键词: 并行处理, 环形网络拓扑, 差分进化算法

Abstract:

Parallel processing has become a key technology in modern computing technology. In recent years, the development of software leads to a collection of heterogeneous computers to use parallel computing resources. This paper explores how works of the structure in the ring network topology parallel differential evolution algorithm to improve the method’s speed and performance. The results show that the exchange of information subpopulations in the range assigned to different nodes of the performance of the algorithm has a significant impact. In addition, not all of the mutations differential evolution strategies are equally sensitive to this parameter.

Key words: parallel processing, ring network topology, differential evolution algorithm