Solving large sparse linear systems in end-to-end accelerator structure simulations

Author

Lee, Lie-Quan ; Ge, Lixin ; Kowalski, Marc ; Li, Zenghai ; Ng, Cho-Kuen ; Schussman, Greg ; Wolf, Michael ; Ko, Kwok

Author_Institution

Stanford Linear Accelerator Center, Menlo Park, CA, USA

fYear

2004

fDate

26-30 April 2004

Firstpage

8

Abstract

Summary form only given. We present a case study of solving very large sparse linear systems in end-to-end accelerator structure simulations. Both direct solvers and iterative solvers are investigated. A parallel multilevel preconditioner based on hierarchical finite element basis functions is considered and has been implemented to accelerate the convergence of iterative solvers. A linear system with matrix size 93,147,736 and with 3,964,961,944 nonzeros from 3D electromagnetic finite element discretization has been solved in less than 8 minutes with 1024 CPUs on the NERSC IBM SP. The resource utilization as well as the application performance for these solvers is discussed.

Keywords

convergence of numerical methods; digital simulation; finite element analysis; high energy physics instrumentation computing; iterative methods; linear accelerators; parallel programming; resource allocation; sparse matrices; 3D electromagnetic finite element discretization; NERSC IBM SP; direct solver; end-to-end accelerator structure simulations; hierarchical finite element basis function; iterative solver; multilevel preconditioner; resource utilization; sparse linear system; Acceleration; Application software; Concurrent computing; Convergence; Electromagnetic modeling; Equations; Finite element methods; Linear accelerators; Linear systems; Sparse matrices;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel and Distributed Processing Symposium, 2004. Proceedings. 18th International

Print_ISBN

0-7695-2132-0

Type

conf

DOI

10.1109/IPDPS.2004.1302909

Filename

1302909