A fast algorithm for electrically small composite objects

Abstract

Extensive studies have been conducted by many researchers on low-frequency or electrically small problems using the surface integral equation (SIE) method. However, it does not appear in the literature that problems with a large number of unknowns in this area can be solved efficiently. The paper presents a multi-tree scheme to apply the low-frequency multilevel fast multipole algorithm (LF-MLFMA) to composite objects, so that O(N) CPU time and memory usage is obtained. A new implementation of basis rearrangement is also presented to make the algorithm stable and capable of solving problems with a large number of unknowns.