A Study on the Preprocessing of MLFMA Parallel Computation

For the efficient solution of large-scale problems in computational electromagnetics, the hierarchical parallelization of the multilevel fast multipole algorithm (MLFMA) is considered based on the data and the angular spectrum domain decomposition algorithms for each level. The decomposition and computation of the correlative data structure which includes the distributed tree, the interpolation or anterpolation matrix, the translation matrix, the near-neighbor impedance matrix and so on should coincide with the domain decomposition algorithms. Through the experimental analysis and the complexity estimates, the preprocessing method of the above data structure is proposed, which solves the computation problem and the distribution problem of those data structure on distributed memory computers using the message passing interface (MPI).