NEC-LIST: Memory Requirements for GNEC

>From the desk of Keith Lysiak

The following note includes some things I have found through trial and
error. They involve setting the parameter MAXMAT in NEC4. In
particular they were derived using GNEC with NEC4 from
Nittany-Scientific. I have not had the chance to verify them on the
original NEC41 code (if someone has the chance to try them, please let
me know). These settings may be common knowledge but they were not
clear to me until I tried it. These also do not pertain to most
applications but become particularly useful when running multiple
problems using NT4.

There seem to be some special settings needed when dealing with
symmetry and/or Numerical Green Function (NGF). For example, you need
to set MAXMAT to the square root of two (1.414) times the number of
segments in the problem if you are doing NGF. On the other hand, when
using symmetry, MAXMAT needs to be set to the square root of the
number of symmetrical sections x the number of original segments.

Let me try to clarify with an example. If you have a 1000 segment
problem and you want to use NGF, you need square root of 2 x 1000 or
MAXMAT = 1415. If you have a 1000 segment problem and you use three
fold symmetry to create a total of 3000 segments, you need square root
of 3 x 1000 or MAXMAT=1733 (not 1000 or 3000).

Now, what if you have 1000 segments, three fold symmetry and you want
to use NGF. What should MAXMAT be set to? As it turns out,
MAXMAT=2445. This is square root 2 x square root 3 x 1000.

Although MAXMAT needs to be set to these limits to prevent disk
swapping, it does not appear that you need to have this much physical
memory to run this size of problem. At first glance it would appear
that you can not run as big of problem if you use NGF. The NGF seems
to set aside more memory that it really needs and will not begin to
swap to disk if MAXMAT is set higher than what you can normally run.

I hope this makes some sense and is helpful to others trying to do
larger problems.

End of Message
Keith A. Lysiak, Ph.D 6220 Culebra Road
Southwest Research Institute P.O.Drawer 28510
Senior Research Engineer San Antonio, TX 78228-0510
Radiolocation Systems Section Voice: (210) 522-5776
Signal Exploitation and Geolocation Division FAX: (210) 522-2709
Received on Thu Oct 22 1998 - 18:19:23 EDT