• DocumentCode
    1801193
  • Title

    Approximate boundary integral equations for time-harmonic rough surface scattering

  • Author

    Saillard, M. ; Soriano, G.

  • Author_Institution
    Inst. Fresnel, Marseille, France
  • Volume
    1
  • fYear
    2003
  • fDate
    22-27 June 2003
  • Firstpage
    569
  • Abstract
    The boundary integral formalism, combined with fast numerical solvers, is a very efficient way to deal rigorously with the time-harmonic scattering from a rough surface separating two semi-infinite homogeneous media. However, applying a MoM to an integral equation leads to a linear system with full complex matrices. If the exact matrix elements are used, the number of operations for solving such a system is proportional to N/sup 2/, where N is the number of unknowns. In addition, storing all the matrix elements in RAM is not possible. This is why several numerical methods have been developed to avoid full matrix storage and to make the computation time scale as N logN. This may be achieved by representing the matrix as a superposition of matrices with specific properties, in order to speed up the matrix-vector products that occur in iterative solvers. As a result, a slight approximation of the kernel of the integral equation leads to a drastic gain in terms of computation time and memory requirements. We show that additional approximations can speed up the computation further, and we compare them with existing approximate methods.
  • Keywords
    approximation theory; boundary integral equations; computational complexity; computational electromagnetics; electromagnetic wave scattering; iterative methods; matrix multiplication; method of moments; rough surfaces; MoM; approximate boundary integral equations; complex matrices; computation time; iterative solvers; matrix superposition; matrix-vector products; memory requirements; semi-infinite homogeneous media; surface currents; time-harmonic rough surface scattering; Frequency; Integral equations; Linear systems; Matrix decomposition; Quantum computing; Rough surfaces; Scattering; Sparse matrices; Surface roughness; Transmission line matrix methods;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Antennas and Propagation Society International Symposium, 2003. IEEE
  • Conference_Location
    Columbus, OH, USA
  • Print_ISBN
    0-7803-7846-6
  • Type

    conf

  • DOI
    10.1109/APS.2003.1217523
  • Filename
    1217523