The bottom line should be is EZNEC accurate? Has the programming been
held within the confines provided by the original provider'

What you are referring to is called "validation" of the modeling code.
Since EZNEC uses the NEC2 engine (or the NEC4 engine) underneath, the
inherent modeling accuracy is that of those engines.



the U.S. government. Who overseas the content of this so called
program.
The same as is true of virtually every other RF and EM modeling program
out there.. The manufacturer of the program oversees it. You, as a
customer, get to run whatever validation suites make you happy. If you
don't like the results, don't use the program. If you find a bug, you
report it to the mfr, and usually they roll out a new version sooner or
later that fixes the problem. This is the advantage of using a provider
that has been in business a while.. they've gone through the release
cycle more than once, and they've got lots of eyes looking at the product.

While looking at source code sounds nice, I suspect that very, very few
of the people who use modeling codes would be willing to take the time
needed to go through them and understand how it works, much less try to
find bugs.

But it does happen. Every few years, you see Jerry Burke at LLNL
announce some minor change in the NEC code base to address some bizarre
corner case.

If it has a patent then all would or should be revealed
in the patent disclosure. Has anybody taken this for his own use for
the advancement of science which is the reason for patents?

Or, even better, they could reveal all in a peer reviewed paper in the
technical literature or in a technical report. Gosh... isn't that what
the NEC folks did. There's hundreds of pages of documentation
explaining the theoretical underpinnings of NEC, how it was validated,
etc..

If you're interested in the optimizers....well, there's a raft of papers
and books on the virtues and problems with various optimizers, both in
general and in combination with NEC or other modeling codes. Yes, in
most cases, you have to wonder if the implementation of the algorithm
was properly done, e.g. Excel uses the Nelder-Mead algorithm in it's
"solver".. and MS doesn't tell you how they exactly did it.. so it's up
to you to find some good test cases and see if it works like you think
it should. Or, do reasonableness checks on the output.. but that's
something you should do with an optimizer in any case.


Has anybody upgraded the assigned patent for the sake of science
or has something not been disclosed to prevent true examination
and as such invalidates the patent? Does the government have the
option of review of all algorithms or are they in the same position
the country is with voting machines? Basically the purchaser is really
in the position of caveat emptor especially since all programs provide
different results!

You betcha... You gets to pay yer money and you gets to take yer chances.

But in practice, the reputable mfrs of modeling codes tend to provide
validation examples, if only because that's how they do their own
internal testing. For example, everybody runs the examples in the NEC
manual, because those have been extensively validated, so if your code
gives the same result as the NEC output (subject to the limitations of
NEC and your code), then you say, "yes indeedy, my program works ok!"

Jim