Sparsification of the Reduced Matrix of the CBFM for a Memory Efficient Solution of Electrically Large EM Scattering Problems

Abstract

In this paper a sparsification approach is applied to the compressed matrix resulting from the Characteristic Basis Function Method (CBFM) process in order to significantly reduce the memory cost of this direct solver-based largely used numerical technique. Many efforts have been made in recent years to efficiently calculate this matrix but, all of them have focused on the time cost and have not dealt with the memory resources needed to store it. With the proposed sparsification approach, the present work aims to reduce the computational cost associated to the compressed matrix both in terms of CPU time and memory consumption.