An efficient parallel SAI preconditioner for multilevel fast multipole algorithm for scattering by extremely large complex targets

Author

Xiao-Min Pan ; Xin-Qing Sheng

Author_Institution

Center for Electromagn. Simulation, Beijing Inst. of Technol., Beijing, China

fYear

2009

fDate

3-6 Nov. 2009

Firstpage

407

Lastpage

410

Abstract

A highly efficient parallel SAI preconditioner is presented. This preconditioner is incorporated into our previously well developed parallel MLFMA code, and successfully computes scattering by a sphere with a diameter of 750 wavelengths, involving over 315 millions unknowns. The capability of this parallel SAI preconditioner is further demonstrated on computing scattering by an extremely large tank with a size of over 300-wavelengths containing around 35 millions unknowns, where convergence cannot be achieved without this SAI preconditioner.

Keywords

approximation theory; computational electromagnetics; electromagnetic wave scattering; iterative methods; mathematics computing; matrix inversion; method of moments; parallel algorithms; sparse matrices; electric conducting sphere; electromagnetic wave scattering computation; extremely-large complex target; extremely-large tank; iterative method; method of moments; multilevel fast multipole algorithm; parallel MLFMA code; parallel SAI preconditioner; sparse approximate inverse matrix; Multilevel fast multipole algorithm (MLFMA); Parallel; Preconditioner; Sparse approximate inverse (SAI);

fLanguage

English

Publisher

iet

Conference_Titel

Microwave Technology and Computational Electromagnetics, 2009. ICMTCE. International Conference on

Conference_Location

Beijing

Print_ISBN

978-1-84919-140-1

Type

conf

DOI

10.1049/cp.2009.1355

Filename

5521232