This thesis concentrates on the parallel techniques and their applications in computational electromagnetics. Based on the different characteristics of memory requirement and CPU time in different levels in the Multi-Level Fast Multipole Algorithm (MLFMA), a new highly efficient parallel approach of MLFMA is proposed and implemented in this paper. Combined this parallel approach of MLFMA with parallel multifrontal solver for sparse matrix, a parallel algorithm for hybrid finite-element/boundary integral/multilevel fast multipole algorithm (FE/BI/MLFMA) has been firstly designed and preliminarily implemented. To check the accuracy and efficiency of the algorithms developed in this paper, the scattering by various complex large objects have been computed, showing their simulation capacities for applications in real world. Last but not least, by using our MLFMA code, an application problem, one of the hot issues in remote sensing, is studied, in which SAR return signals from random surface are simulated and investigated statistically, and some interesting conclusions are obtained.
|