Re: NEC-LIST: General comments on MM formulations/ Solutions

From: <dchatterjee_at_email.domain.hidden>
Date: Tue, 03 Dec 1996 20:01:47 -0600 (UTC -06:00)

  Greetings EM-ers !

  I am responding to the above topic again. Thanks are due to Prof.
  D.B.Davidson and Dr. Frans Meyer (Stellenbosch, South Africa) as
  they recently drew my attention to the following two papers:

  [1] U. Jakobus and F. Landstorfer, " Improved PO-MM Hybrid Formulation
       ...... ", IEEE-T-AP, Feb. 1995, pp. 162-169.

  [2] ---------, " Improvement of the PO-MoM Hybrid Method by accounting
       for Effects of Perfectly Conducting Wedges ", IEEE-T-AP, Oct. 1995,
       pp. 1123-1129.

  The authors of the two papers actually develop a current-based model and
  they are using the UTD diffraction coefficients to correct the current
  on the arbitrary surface(s). (See [2, Eqs. (13) to (17)].) While the
  proposal for developing a current-based hybrid model is no doubt very
  attractive computationally, the crux of the question concerning the c.p.u
  time rests on the development of an efficient form of the Green's function
  for the canonical problem. For example, had the dyadic diffraction
  coefficients not been available the development of [2, Eqs. (13)-(17)]
  may have been somewhat more difficult. These UTD dyadic diffraction
  coefficients in [2] are actually uniform asymptotic forms of the exact,
  2-D line-source Green's functions for the perfectly conducting wedge.
  (The usefulness of these dyadic diff. coeffs. is of course well-known.)

  Consequently, a code that uses a suitable (read "asymptotic") Green's
  function for some canonical problem(s), may be expected to produce
  reasonably good results (when compared to exact solutions) at an
  affordable computational cost. The key to the development of efficient
  MoM codes rests on the formulation of the appropriate Green's function
  - where appropriate analytical reductions play a pivotal role. Such
  techniques play a key role in handling electrically large problems.
  
  Some useful insight on this issue maybe gleaned from the following paper:

  [3] E. K. Miller, " A Selective Survey of Computational Electromagnetics ",
      IEEE-T-AP, September 1988.
   
  - cheers,
 =================================================================

  Deb Chatterjee

  Radar Systems and Remote Sensing Laboratory
  Department of Electrical Engineering and Computer Science
  Room 322, Nichols Hall
  2291 Irving Hill Road
  Lawrence, KS 66045-2969

  tel: (913)864-7742
  fax: (913)864-7789
  e-mail: dchatterjee_at_kuhub.cc.ukans.edu
Received on Wed Dec 04 1996 - 09:32:10 EST

This archive was generated by hypermail 2.2.0 : Sat Oct 02 2010 - 00:10:37 EDT