Multilevel fast multipole acceleration in the Nystrm discretization of surface electromagnetic integral equations for composite objects

Mei Song Tong, Weng Cho Chew

Research output: Contribution to journalArticlepeer-review

Abstract

The multilevel fast multipole algorithm (MLFMA) based on the Nystrm discretization of surface integral equations (SIEs) is developed for solving electromagnetic (EM) scattering by large composite objects. Traditionally, the MLFMA is based on the method of moments (MoM) discretization for the SIEs and it usually works well when the robust Rao-Wilton-Glisson (RWG) basis function is enough to represent unknown currents. However, the RWG basis function may not represent both the electric and magnetic current in solving the electric field integral equation (EFIE) and magnetic field integral equation (MFIE) for penetrable objects, and how one represents another current could be a problem in the MoM. In this work, we use the Nystrm method as a tool to discretize the SIEs and incorporate the MLFMA to accelerate the solutions for electrically large problems. The advantages of the Nystrm discretization include the simple mechanism of implementation, lower requirements on mesh quality, and no use of basis and testing functions. These benefits are particularly desired in the MLFMA because the solved problems are very large and complex in general. Numerical examples are presented to demonstrate the effectiveness of the proposed scheme.

Original languageEnglish (US)
Article number5498961
Pages (from-to)3411-3416
Number of pages6
JournalIEEE Transactions on Antennas and Propagation
Volume58
Issue number10
DOIs
StatePublished - Oct 2010

Keywords

  • Electromagnetic scattering
  • Nystrm discretization
  • multilevel fast multipole algorithm
  • surface integral equations

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Multilevel fast multipole acceleration in the Nystrm discretization of surface electromagnetic integral equations for composite objects'. Together they form a unique fingerprint.

Cite this