MLFMA-FFT Parallel Algorithm for the Solution of Extremely Large Problems in Electromagnetics

Abstract

An efficient parallel implementation of the multilevel fast multipole algorithm-fast Fourier transform (MLFMA-FFT) has been successfully used to solve an electromagnetic problem involving one billion of unknowns, which indeed becomes the largest problem solved with the surface integral-equation approach up to now. In this paper, we present a deep review of this challenging execution, focusing on the details of the parallel implementation step by step, with the aim of describing the different stages of the parallel algorithm and analyzing its overall parallel performance.