首页> 外文会议>International Conference on the Computation of Electromagnetic Fields >Hybrid h- and p-Type Multiplicative Schwarz (h-p-MUS) Preconditioned Algorithm of Higher-Order FE-BI-MLFMA for 3D Scattering
【24h】

Hybrid h- and p-Type Multiplicative Schwarz (h-p-MUS) Preconditioned Algorithm of Higher-Order FE-BI-MLFMA for 3D Scattering

机译:用于3D散射的高阶Fe-Bi-MLFMA的杂种H-和P型乘法施瓦茨(H-P-MUS)预处理算法

获取原文

摘要

The key to design algorithms of higher-order finite element-boundary integral-multilevel fast multipole algorithm (FE-BI-MLFMA) is how to construct preconditioners for the FEM matrix part in the FE-BI matrix equation. A hybrid h- and p-Type multiplicative Schwarz (h-p-MUS) preconditioner is proposed for 3D scattering in this paper. First, the hierarchical higher-order FEM matrix is rewritten by classifying unknowns into different groups corresponding to different FEM orders. Then a preconditioner is constructed by the p-type multiplicative Schwarz method (p-MUS). To further improve efficiency in the p-MUS construction, an h-type multiplicative Schwarz preconditioner is employed for different submatrices. Numerical experiments show that this preconditioning scheme can offer a better efficiency both in CPU time and memory requirement than previous algorithms. This h-p-MUS preconditioner for the FEM matrices is applied to different algorithms of FE-BI-MLMFA. Analysis and Numerical results show that the h-p-MUS preconditioned conventional algorithm (h-p-MUS-CA) has a better efficiency over other h-p-MUS preconditioned algorithms. A variety of numerical experiments are performed for complex objects in this paper, demonstrating that this h-p-MUS-CA exhibits superior efficiency in the CPU time and memory, and greatly improves the capability of the higher-order FE-BI-MLFMA.
机译:高阶有限元 - 边界积分 - 多电平快速多极算法(Fe-Bi-MLFMA)设计算法的关键是如何构建FE-BI矩阵方程中的FEM矩阵部分的预处理器。提出了一种杂交H-和P型乘法施瓦茨(H-P-MUS)预处理器,用于本文的3D散射。首先,通过将未知数分类为与不同的FEM订单相对应的不同组来重写分层高阶UP矩阵。然后由p型乘法施瓦茨方法(p-mus)构建一个预处理器。为了进一步提高P-MUS构建的效率,将采用H型乘法施瓦茨预处理器用于不同的子群。数值实验表明,该预处理方案可以在CPU时间和内存要求中提供比以前的算法更好的效率。该用于有限元矩阵的H-P-MUS预处理器应用于Fe-Bi-MLMFA的不同算法。分析和数值结果表明,H-P-MUS预处理的常规算法(H-P-MUS-CA)在其他H-P-MUS预处理算法上具有更好的效率。本文对复杂物体进行了各种数值实验,表明该H-P-Mus-CA在CPU时间和记忆中表现出优异的效率,并且大大提高了高阶FE-BI-MLFMA的能力。

著录项

相似文献

  • 外文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号