摘要
对整体的并行优化算法,如:并行变量分块算法(PVD)、并行梯度分块算法(PGD)、并行变量转换算法(PVT)等进行了分析。这些算法将原最优化问题分解为一系列规模较小的且相互独立的子问题,从而用多台处理机同时对这些子问题求解,减少了工作量、缩短了计算时间。
This paper discussed the parallel algorithms of global optimization, such as parallel variable distribution algorithm, parallel gradient distribution algorithm and parallel variable transformation algorithm. The main characters of these algorithms are that the primal optimization problems are decomposed into smaller ones, which can be solved by many processors in same time. It can reduce the work load and shorten the computing time.
出处
《山东科技大学学报(自然科学版)》
CAS
2006年第2期106-108,共3页
Journal of Shandong University of Science and Technology(Natural Science)
基金
国家自然科学基金(10571109)
关键词
并行梯度分块
并行变量分块
并行变量转换
无约束最优化
并行算法
parallel gradient distribution
parallel variable distribution
parallel variable transformation
unconstrained optimization
parallel algorithm