Parallel multilevel fast multipole method for solving large scale electromagnetic problems [radar cross section problems]

The parallel multilevel fast multipole method (MLFMM) is an efficient computational algorithm for solving large-scale radar cross section (RCS) problems. The code is developed and implemented on massively parallel, distributed memory computer systems. The multilevel grouping structure is partitioned carefully to separate computational tasks and reduce data communication. A message passing interface (MPI) is employed to perform the data transfer and support the matrix-vector product in the parallel iterative procedure. The accuracy of the code is verified by comparison with results of a benchmark target. The computational complexity and efficiency of the parallel MLFMM is discussed.