Study On Parallelization Of Multilevel Green's Function Interpolation Method

With the fast development of electronic science and technology,the research on electromagnetic (EM) scattering from large scale and complex target has attracts increasing attentions. Although many fast algorithms have been presented to solve some electromagnetic problems, they can not meet the computational demands on a single computer for most of larger-scale complicate EM problems. Parallel computation provides an effective way to solve this problem.In this thesis, the parallelization of a new fast algorithms-multilevel green function interpolation method (MLGFIM) is developed. Parallel schemes are based on different parallel tools on the parallel computers with different kinds of memory structure. Firstly, the MLGFIM is paralleled by using MPI for the parallel computers with distributed memory structure. The parallel of geometry information and the construction of a distributed tree are analyzed and achieved. The strategy of parallelization is presented in detail. Secondly, the MLGFIM is paralleled by using OpenMP for the parallel computers with shared memory structure. The details of the strategy are expatiated. The bistatic radar scattering section (RCS) of electrically large dielectric sphere, plane, car and the resultant parallel accelerated ratios are given to demonstrate the accuracy and efficiency of the proposed parallel algorithms.