Re: NEC-LIST: routine for finding complex roots wanted

From: Gerald Burke <burke2_at_email.domain.hidden>
Date: Mon, 25 Sep 2000 09:42:57 -0700

>Does anybody know, please, where may I find a fortran (or C) routine
>for complex roots seeking, numerically, of a generic complex function?
>(that is a MoM matrix determinant) Websites, or books...
>
>Thanks in advance,
>
>Riccardo Leone

Riccardo,

I am no expert on that, but I would try Numerical Recipes for a
routine to find the zeros of a numerically defined function. However,
it may be very slow when the MoM matrix is large.

People at AFWL, Kirtland AFB were trying to do that back in the '80s
when SEM was of great interest. They had us set up a version of NEC
that allowed complex frequencies. Then they integrated around closed
contours in frequency to test if a pole was enclosed. A better way
would probably be the "Frequency Domain Prony" method (J.N.
Brittingham, Proc. IEEE, pp. 263-273, 1980). That only needs real
frequencies, but it will have problems with noisy data. The ZPLOT
program distributed with NEC includes routines for rational-function
interpolation on a sliding window. Then all you need to do is find
the zeros of the denominator polynomial.

Jerry Burke
LLNL
Received on Mon Sep 25 2000 - 20:04:23 EDT

This archive was generated by hypermail 2.2.0 : Sat Oct 02 2010 - 00:10:40 EDT