• DocumentCode
    2156486
  • Title

    A fast direct finite element solver for large-scale 3-D electromagnetic analysis

  • Author

    Zhou, Bangda ; Jiao, Dan

  • Author_Institution
    Sch. of Electr. & Comput. Eng., Purdue Univ., West Lafayette, IN, USA
  • fYear
    2012
  • fDate
    8-14 July 2012
  • Firstpage
    1
  • Lastpage
    2
  • Abstract
    A fast direct solver is developed for the vector finite element based analysis of general 3-D electromagnetic problems. In this solver, we fully take advantage of the sparsity in the original matrix of size N and the reduced fill-ins due to nested dissection ordering. We also organize the overall factorization into a sequence of partial factorizations of dense smaller matrices by the multifrontal algorithm. We then accelerate the computation of all the intermediate dense matrices by ℌ-matrix based fast arithmetic. The computational complexity of the proposed solver is theoretically analyzed and shown to be O(N) in both time and storage for problems with a relatively small electric size. The solver´s controlled accuracy is also proved. Numerical experiments demonstrate that the proposed solver outperforms both the state-of-the-art multifrontal solver and the ℌ-based direct solver for 3-D electromagnetic analysis.
  • Keywords
    computational complexity; electromagnetic fields; finite element analysis; matrix decomposition; ℌ-matrix based fast arithmetic; computational complexity; dense smaller matrices partial factorizations; fast direct finite element solver; intermediate dense matrices; large-scale 3D electromagnetic analysis; multifrontal solver algorithm; nested dissection ordering; vector finite element based analysis; Acceleration; Accuracy; Complexity theory; Electromagnetic analysis; Finite element methods; Particle separators; Sparse matrices;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Antennas and Propagation Society International Symposium (APSURSI), 2012 IEEE
  • Conference_Location
    Chicago, IL
  • ISSN
    1522-3965
  • Print_ISBN
    978-1-4673-0461-0
  • Type

    conf

  • DOI
    10.1109/APS.2012.6349135
  • Filename
    6349135