From: Mark Holm 
To: Frank Q 
CC: atm{at}shore.net
Reply-To: Mark Holm 




Frank Q wrote:

> Hi Mark
>
> Regarding your statement (see snippet below) stating that ray optics
> is "just plain wrong", could you define what you mean by
"wrong".


Fair enough!  I'll try.

Ray optics does not predict several important phenomena commonly observed
and used by opticians, including ATM's.  When I say 'does not predict"
I don't mean just that the predictions are quantitatively wrong.  Ray
optics doesn't predict their existence at all!  This is a pretty serious
failing for a theory.  You have to be pretty desperate to continue
accepting a theory that fails to predict
common phenomena.  (Even more desperate than physicists are right now with
the problem that quantum mechanics and relativity can't be brought
together.)

What are these common phenomena that ray optics doesn't predict.

A partial list:

1. The Airy disc and ring pattern of an image of a point source. 2. (Not so
commonly used) The apodization techniques, mentioned by Richard Schwartz in
another post on this thread, used to modify the Airy disc and ring pattern.
3. Diffraction spikes from spider vanes, etc. 4. Diffraction effects seen
in the Foucault test and Ronchi test.  (All those extra shadows that make
interpretation difficult.) 5. Antireflection coatings of any type on
lenses, prisms, etc. 6. Multilayer dielectric highly reflective coatings on
mirrors. 7. Multilayer dielectric filters, both broad and narrow band.
(Such as (relatively) inexpensive H-alpha filters.) 8. Newton's fringes and
elaborations thereof used to interference test flats and
other mating optical surfaces.
9. Any other interferometric test or technique including radio interferometry to
make observations of astronomical radio sources that would be impossible without it.

(The Michaelson-Morley experiment that is one of the keystones of relativity was
performed with a Michaelson interferometer.)(Observations of interference
effects with electrons were part of the early experimental impetus for
quantum mechanics.)

10. Spectra produced by diffraction gratings.  (A very large chunk of
astrophysics, a very large chunk of the rest of modern physics, a good bit
of quantum mechanics, a lot of modern chemistry, and more I haven't thought
of depends on the routine use of diffraction gratings to produce spectra.)

11. Iridescence in butterfly wings, bird feathers, etc. (Related to 6 and 7)

12. Colored rings in oil films on water. (related to 5, 8 and 9)

13. LCD displays in portable electronic devices.

14. Polarizing filters including sunglasses. (Gets into a pretty scary part
of wave optics.)


>>From your statement, I get the impression that if I use a raytrace program
> to calculate the path of a bundle of parallel rays off a parabolic reflector
> and these rays are brought to a common point of intersection (ie the focus)
> then this in reality is not what happens. Granted that they are not brought
> to
> an infinitely small point - wave optics provide the proper explanation - but
> for many applications, ray optics provide a very adequate analysis of the
> system.


I have to let this question wait for now.  I need to get to bed tonight.
Suffice for now, that I agree a model does not need to be perfect, or
predict all aspects of a situation to be useful.  The important thing is to
know when one is getting to the limits of the model and is likely to get
invalid predictions.  This discussion started from a question regarding
imperfections in
Newtonian diagonal mirrors.  There isn't much point discussing really bad
diagonals, everybody knows they are useless.  Therefore, one is left to
discuss diagonals that are merely marginal: not really good and not really
bad.  This is
exactly at the (admittedly rather fuzzy) boundary where ray optics ceases
to make good predictions.

>
> Isn't  "just plain wrong" a little strong in this situation ??


Well. I don't think so or I wouldn't have written it ;)

Newtonian dynamics is also just plain wrong, as is wave optics, but
Newtonian dynamics is good enough that most macroscopic engineering can be
faithfully based on it (well, except thermo. but that isn't really ever
macroscopic, and radio wave engineering, but I count that as a subset of
optics) and wave theory is good enough for a really broad swath of
practical optics.

As I mentioned before, even wave optics isn't good enough to explain how a
thin film of aluminum atoms can reflect light, or how refractive index
happens, or how the spectra of atoms and molecules happen.  For that, you
need quantum mechanics.

I figure that if I can do an experiment with less than 5 dollars worth of
materials that produces a result not predicted by a theory, then the theory
is, in some important sense, just plain wrong.  I can do an experiment with
less than one cent worth of oil that will produce an effect not predicted
by ray optics!

BTW, in another vein, my younger son has just been learning Mendelian
genetics in High School Biology.  Now we know that Mendelian genetics is
just plain wrong
too.  If Gregor Mendel had studied corn instead of sweet peas, it is unlikely he
would ever have made any sense out of his observations. (Corn genetics is pretty
crazy.)  Luckily for biology, Mendel happened to study plants in which the
DNA mostly works in pretty simple ways.  We now know that the stuff has all
kinds of
wild behavior Mendel never imagined.  (Of course he didn't know about DNA
at all.  The genes of his theory were not localized to any known physical
structure.)

Now that was OT, but connected as another illustration of a theory that
while basically wrong, still has some useful aspects.

Mark Holm

mdholm{at}telerama.com

--- BBBS/NT v4.00 MP