基于高阶四面体矢量元的大规模本征值求解  

Efficient Solution of Large-Scale Eigenvalue Problems Based on Higher Order Tetrahedral Vector Elements

作　　者：尹文禄[1] 叶良丰[1] 邓聪[1] 柴舜连[1] 毛钧杰[1] 

机构地区：[1]国防科技大学电子科学与工程学院,长沙410073

出　　处：《微波学报》2010年第1期12-18,共7页Journal of Microwaves

摘　　要：采用H1(curl)四面体矢量元分析大规模电磁本征值问题。基于基函数分类和单元矩阵分块技术,提出了一种新型的针对高阶四面体矢量元编程的全局编码方法,并结合RCM重排序技术压缩整体矩阵带宽。通过分析一个矩形谐振腔,系统研究了基于不同求解技术的ARPACK本征值求解器的性能,并将其用于分析各种复杂腔体的谐振问题。Large-scale electromagnetic eigenvalue problems are analyzed by using H1 （curl） tetrahedral vector elements. A new global coding method is proposed to implement higher order tetrahedral vector elements by using the classified bases and block matrix technique, and the bandwidth of the global matrix is reduced by RCM reordering technology. The results of a numerical experiment that investigates the resonant problem of a rectangular cavity, show the performance of AR- PACK eigensolvers based on various solving techniques, which are also applied to the eigen-solution of complex cavities.

关 键 词：有限元方法 高阶 四面体矢量元 RCM技术 本征值问题 ARPACK软件包 

分 类 号：O441.4[理学—电磁学]