首页 | 官方网站   微博 | 高级检索  
     

函数优化异步并行演化算法
引用本文:康立山,刘溥,陈毓屏.函数优化异步并行演化算法[J].计算机研究与发展,2001,38(11):1381-1386.
作者姓名:康立山  刘溥  陈毓屏
作者单位:国防科学技术大学并行与分布处理国家重点实验室,长沙,410073;武汉大学软件工程国家重点实验室,武汉,430072
基金项目:国家自然科学基金 ( 60 0 730 43,70 0 710 42 ),并行与分布处理国家重点实验室基金资助
摘    要:提出了一种新型、高效的函数优化异步并行演化算法,利用这个算法,在巨型并行计算机上解决了一些高难度的大型优化问题,其中包括一个超高维的非线性规划问题-BUMP问题。由于BUMP问题的强非线性和超我峰特性,目前还未见有超过50维的BUMP问题的结果发表。而在此不仅仅得到了从2维到50维迄今最好的解,而且一直计算到了1000000维,并得到了满意的结果。数值实验表明,新算法是鲁棒和高效的。

关 键 词:函数优化  演化算法  异步并行算法  计算机

ASYNCHRONOUS PARALLEL EVOLUTIONARY ALGORITHM FOR FUNCTION OPTIMIZATION
KANG Li-Shan,LIU Pu,and CHEN Yu-Ping.ASYNCHRONOUS PARALLEL EVOLUTIONARY ALGORITHM FOR FUNCTION OPTIMIZATION[J].Journal of Computer Research and Development,2001,38(11):1381-1386.
Authors:KANG Li-Shan  LIU Pu  and CHEN Yu-Ping
Abstract:A new efficient asynchronous parallel evolutionary algorithm for function optimization is proposed in this paper. Using this algorithm, some hard optimization problems including a very high dimensional BUMP problem are solved. Because of the super nonlinear and super multimodal characteristics of BUMP problem, there are no results with dimension greater than 50 ever published until now. In spite of these, not only a series of best solutions from 2 to 50 dimensions but also satisfactory results up to 1000000 dimensional BUMP problem have been obtained. The numerical results show that the new asynchronous parallel evolutionary algorithm is robust, effective, and efficient.
Keywords:function optimization  evolutionary algorithm  asynchronous parallel algorithm
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号