This paper develops a parallel fast multipole method specialized for uniform brick elements to achieve large-scale micromagnetic simulations of a practical perpendicular recording head within acceptable computation time. In highly parallel computation, it is indispensable to equally distribute the computing load among processes so as to obtain the good parallel speedup. The load balancing technique which can be regarded as the domain decomposition method based on octree structure in the fast multipole method is discussed. The scalability of the parallel fast multipole accelerated micromagnetic simulation is investigated with 100 or more MPI processes.
ASJC Scopus subject areas