• DocumentCode
    726059
  • Title

    A robust meshless method with QR-decomposed radial basis functions

  • Author

    Shunchuan Yang ; Zhizhang Chen ; Yiqiang Yu ; Ponomarenko, Sergey

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Dalhousie Univ., Halifax, NS, Canada
  • fYear
    2015
  • fDate
    17-22 May 2015
  • Firstpage
    1
  • Lastpage
    3
  • Abstract
    Ill-condition of the interpolation matrix has presented a hurdle for applying a node-based meshless method to practical modeling and simulation of electromagnetic structures. In this paper, a robust meshless method is proposed that does not have the problem. It employs the QR method to decompose the Gaussian radial basis function; as a result, the matrix ill-condition that persists with a mesheless method with a small shape parameter is theoretically removed. In other words, unlike the conventional meshless method, the proposed method is insensitive to the shape parameters and can work well even when the radial basis functions becomes extremely flat.
  • Keywords
    Gaussian processes; computational electromagnetics; interpolation; matrix algebra; radial basis function networks; Gaussian radial basis function; QR-decomposed radial basis functions; electromagnetic structures; interpolation matrix; matrix ill-condition; robust meshless method; Gold; Kernel; Matrix decomposition; Shape; Meshless; QR-Decomposition; Radial basis function (RBF);
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Microwave Symposium (IMS), 2015 IEEE MTT-S International
  • Conference_Location
    Phoenix, AZ
  • Type

    conf

  • DOI
    10.1109/MWSYM.2015.7166880
  • Filename
    7166880