Mike Kaliski wrote:
. . .
NEC based programs follow well proven principles, but are not the holy
grail in being able to predict the performance of all antenna types or
configurations. They can accurately predict 'established' antenna design
performance in most circumstances and give a useful insight into what
properties a new and previously untried design might have. . .

They are used daily by a very large number of people to predict, with
excellent accuracy, how new and untried antenna designs will work.

The programs
are, of necessity, constrained by the accuracy of the physical
measurements and formulae used in writing the original program.

That's not quite correct. The "formulae" used are very fundamental
electromagnetic equations. The "physical measurements" used are
fundamental constants such as the speed of light and permittivity and
permeability of free space, all of which are known to much greater
accuracy than antenna performance measurements are capable of
delivering. So neither constitutes any real constraint on program
accuracy. The main constraints are the inability to model all physical
structures. For example, NEC has no provision for including a slab of
dielectric, making microstrip antennas impossible to model accurately
(except the rare ones with air dielectric). There are many such
constraints, of which the user has to be acutely aware. The programs are
very accurate if, and only if, the model is indeed representative of the
physical antenna.

I do not consider the challenge to Newton's Laws to be valid. Just
because there is no true vacuum anywhere in the universe, as far as can
be established, then all objects will eventually come to rest due to
friction. This does not invalidate the premis that in the absence of
friction or any other external influence, an object would continue in
motion along a straight path forever.

This would be funny if it weren't sad. This newsgroup is one of the few
places I can think of where the silly statement I posted about moving
bodies and friction would be taken seriously. But it's really no
surprise, since it's much less unreasonable than the imaginative
alternative theories which are seriously presented, and just as
seriously argued, here daily.

My postulate about objects in motion was a parody of Cecil's rejection
of theoretical cases on the basis that they can't exist in practice, my
intent being to show how such a rejection leads to incorrect results.
But I see it's drawing the same serious response as Cecil's and Art's
postings. All that's missing is one of Richard's quotes from Terman and
support from Derek.

Ultimately whatever predictions are made by computer modelling programs
or theory, the only measure of success is by physical measurement in the
real world.

That's almost true. It's important also to know that even moderately
accurate antenna measurements are extremely difficult and exacting to
make. Consequently, there are often cases where model results are a
better indicator of an antenna's performance than measured ones.
Antennas and models follow the same physical laws. So to the extent that
the model imitates reality, the results are the same.

This philosophy isn't restricted to antenna modeling. Circuit designers
use programs to model circuits which can't be measured. I've designed
circuitry which operates from a few GHz to several tens of GHz, which is
still in production. It's simply impossible to measure voltages and
currents in such circuits, so modeling is heavily used. Often, every PC
board or hybrid circuit pad and trace has to be included as a
transmission line model; I've even had to build a transmission line
model of a capacitor which was 80 mils long. That the overall
performance matched the models' predictions is evidence that the models
matched the physical circuits. IC designers rely entirely on models to
predict the performance of circuits which can be measured only after
huge financial investments to commit the model results to physical
reality. The principles are exactly the same for antenna modeling as for
circuit modeling.

Roy Lewallen, W7EL