无网格数值模拟的并行算法研究  

Parallel algorithms of meshless numerical simulation

作　　者：卢德唐[1] 曾清红[1] 林春阳[1] 

机构地区：[1]中国科学技术大学力学与机械工程系,安徽合肥230027

出　　处：《中国科学技术大学学报》2006年第12期1299-1307,共9页JUSTC

基　　金：国家自然科学基金(10672159);国家重点基础研究发展(973)计划(2006CB705805)资助

摘　　要：对无网格数值模拟的并行算法进行了详细研究.包括使用并行桶搜索算法进行节点搜索,使用并行几何搜索算法进行样点搜索,并行计算无网格形函数及其导数,边界条件的并行处理,使用并行预处理共轭梯度法求解方程组以及负载平衡等.最后给出了无网格数值模拟并行计算的实施流程和计算实例.计算结果表明,无网格数值模拟具有很高的并行性和很好的并行效率,计算规模越大,并行效率越高.The parallel algorithms of meshless numerical simulation was studied. In the parallel implementation of meshless methods, parallel bucket search was used for node search and parallel geometry search was employed for sample point search. The parallel computation of meshless shape functions and their derivatives was discussed and the parallel treatment of essential boundary conditions and the parallel solution of system of equations were also detailed. For the sake of parallel efficiency, the importance of load balance in parallel computing of meshless methods was emphasized. Finally, the flowchart of parallel meshless computing and a numerical example were given. The example shows that meshless methods possess high parallel potential and parallel efficiency, and that the larger the computation, the higher the parallel efficiency.

关 键 词：无网格方法 数值模拟 并行算法 并行桶搜索 并行几何搜索 并行预处理共轭梯度法 负载平衡 

分 类 号：TP338.6[自动化与计算机技术—计算机系统结构] O246[自动化与计算机技术—计算机科学与技术]