Re: NEC-LIST: NEC List Newbie

Quoting Phil Hystad <phystad_at_mac.com>, on Sat 12 Jul 2008 10:19:42 AM PDT:

> NEC-List:
>
> I am not sure if this is the right forum for asking a few questions but
> I will proceed. I have been researching NEC2 with the goal of building
> it from scratch and then experimenting with various problems. So, I
> have the following questions...
>
> 1. I have read through the content made available by the following URLs:
>
> www.nec2.org
> and
> www.si-list.net/swindex.html
>
>
> Is this the best source for NEC2 sources. So far, I have
> downloaded the file
> nec2dx_src.zip.
>
> Also, the information at these web sites appears old (2005), is
> there anything
> more current or is this still current.

Pretty current. Not like Maxwell's equations have changed and the
code needs updating. The actual modeling codes don't change, it's
just the user interfaces, etc.

Bear in mind that NEC2 for people who make their living using these
codes has been supplanted by NEC4 (not free) and other various and
sundry EM codes. NEC4 doesn't have some of the numerical issues that
NEC2 has, and it handles insulation and buried wires, as well.
There's a whole raft of finite element codes out there too.

>
>
> 2. I notice that there are other NEC2 derived sources such as the C
> language version
> which looks like it came from f2c. Has anyone done a NEC2
> implementation or
> conversion to languages: Java or at least Fortran 90?
>
Nope.. why bother (not to be flip...)

1) NEC2 as it sits, works. There's not much to be gained by fighting
through rewriting it. It's basically a big matrix math number
crunching, and FORTRAN works just fine for that. It's not clear that
Java or more modern FORTRAN would buy you much. Where you see the
activity in other languages in in pre/post processors.

2) If you wanted to make changes, you'd really want to start with the
source for NEC4. But, again, it's not clear that the basic processing
in NEC4 "needs" any of the capability of the other languages, and with
the existing code, you have the very considerable advantage of a piece
of software that is fairly complex (mathematically) and has had a LOT
of validation.

3) the output from compiling and linking the output of processing NEC
through f2c doesn't work as well (if at all) compared to just using a
native FORTRAN compiler. You're much better off finding a better
FORTRAN compiler than fooling with trying to get it to work in C.
 From a high performance computation standpoint, the FORTRAN compilers
are at a very high level of refinement and quality. There are good
compilers available for just about any platform you care to name.

So, if one has time/resources to fool with NEC in general, your time
is better repaid in other ways than trying to port it to another
language or compiler.

Jim Lux

-- 
The NEC-List mailing list
NEC-List_at_robomod.net
http://www.robomod.net/mailman/listinfo/nec-list