Efficient parallelization of the FETI-DP algorithm for large-scale electromagnetic simulation

Kedi Zhang, Jianming Jin

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

For decades, the finite-element method (FEM) has been proven to be a versatile approach in modeling complicated materials/systems for its strong adaptability to complex geometries and high numerical accuracy (J.-M. Jin, The Finite Element Method in Electromagnetics. 3rd ed., Hoboken, NJ: Wiley, 2014). Unfortunately, a volumetric discretization, as required by the FEM, would easily yield a linear system with millions or even billions of unknowns for modern engineering applications such as antenna array analysis. The domain decomposition-based FEMs have been therefore developed to enable large-scale electromagnetic (EM) simulations on parallelized computer clusters, among which the dual-primal finite-element tearing and interconnecting (FETI-DP) algorithm for EM analysis is highly powerful because of its numerical stability and scalability (Y.-J. Li and J.-M. Jin, J. Comput. Phys., 228, 3255-3267, 2009).

Original languageEnglish (US)
Title of host publication2015 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium), USNC-URSI 2015 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Number of pages1
ISBN (Electronic)9781479978175
DOIs
StatePublished - Oct 21 2015
EventUSNC-URSI Radio Science Meeting (Joint with AP-S Symposium), USNC-URSI 2015 - Vancouver, Canada
Duration: Jul 19 2015Jul 24 2015

Publication series

Name2015 USNC-URSI Radio Science Meeting (Joint with AP-S Symposium), USNC-URSI 2015 - Proceedings

Other

OtherUSNC-URSI Radio Science Meeting (Joint with AP-S Symposium), USNC-URSI 2015
Country/TerritoryCanada
CityVancouver
Period7/19/157/24/15

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Communication

Fingerprint

Dive into the research topics of 'Efficient parallelization of the FETI-DP algorithm for large-scale electromagnetic simulation'. Together they form a unique fingerprint.

Cite this