Parallel Controlled Random Search algorithms for shape optimization

被引：2

作者：

Hu, YF ^{[1
]}

Maguire, KCF ^{[1
]}

Cokljat, D ^{[1
]}

Blake, RJ ^{[1
]}

机构：

[1] SERC, Daresbury Lab, Warrington WA4 4AD, Cheshire, England

来源：

PARALLEL COMPUTATIONAL FLUID DYNAMICS: RECENT DEVELOPMENTS AND ADVANCES USING PARALLEL COMPUTERS | 1998年

关键词：

D O I：

10.1016/B978-044482849-1/50041-5

中图分类号：

TP39 [计算机的应用];

学科分类号：

081203 ; 0835 ;

摘要：

In this paper parallel Controlled Random Search (CRS) algorithms are first suggested. Unlike the standard CRS algorithms, the parallel algorithms (PCRS) generate more than one new point in parallel at each iteration. The parallel algorithms are tested on a Gray T3D parallel computer, and speedups of up to 47 are achieved. Some comparison is also made with genetic algorithms. Initial results on a design optimization problem are given.

引用

页码：345 / 352

页数：8