Large mirrors can't achieve theoretical resolution, due to surface flaws?

"Kai-Martin Knaak" wrote in message
news
On Mon, 03 Aug 2009 22:07:53 -0400, Neil B. wrote:

I still find it incredible, that wave optics
can ever make the image better than ray-trace in any but a minor way.

Just accept, that ray-tracing ignores each and every effect of the
wavelength. Wave optics by contrast involves no such approximation. So
ray-tracing can at best describe reality as good as wave-optics, but
never better.

Bottom line: If you get better resolution with ray-tracing than with
wave
optics and both methods have been applied correctly, then the outcome
of
wave optics is closer to reality. Whether, or not this is a minor way,
depends on the parameters you are asking for.

---(kaimartin)---

I've agreed many times, if ray-trace resolution is *better* than wave
optics, the WO answer is closer to reality and therefore the image is
worse than the RT prediction. But the question motivating this entire
thread is: what if the RT resolution is substantially *worse* than the
ideal WO image for the aperture, then can WO "cure" the geometric spread
and make the image better than from the RT prediction? Jim Black says it
can, but in a roundabout way that I find hard to appreciate how much the
improvement would be (assuming he's right.)

In article ,
"Neil B." wrote:

How about patchy mirror irregularity about one meter size, with
ray-trace circle of confusion larger than the ideal diffraction spot.

IIRC it makes no difference. The strehl ratio depends only upon the rms
deviation from perfect figure. Remember, however, that rms means
averagingh across the entire aperture. All the light diverted by a patch
reduces the strehl. For small rms deviation, a simple mathematical
approximation can give the answer. In the end,however, the strehl will
be a function of the rms error even for fairly large error. What this
does not tell you is how the off-peak light is distributed. That does
depend on more than just the rms error.

Bill

--
Private Profit; Public Poop! Avoid collateral windfall!

In article ,
"Neil B." writes:
That may be a fine answer among experts, but to me it's jargon.

Sorry about the complex answer. "Convolution" is a mathematical
operation on two functions to produce a third function. In general,
the convolution of a broad function and a narrow one is a slightly
blurred version of the broader function. If the two initial
functions are Gaussians, the convolution is also a Gaussian with
width equal to the quadrature sum of the widths of the input
Gaussians, i.e., sqrt(w1^2+w2^2).

Can wave optics make the image even better (ie, smaller image of a
distant point source) than the ray-traced circle of confusion?

I can't think of a case where that would happen, but I can't prove
it's impossible. It certainly doesn't occur for typical telescopes
and instruments. Telescopes and instruments to work in visible light
are designed using ray tracing (and largely ignoring diffraction) in
most cases. Exceptions would be instruments that explicitly take
advantage of wave optics (e.g., interferometers, grating
spectrographs) or where diffraction effects are crucial (e.g.,
coronagraphs).

For instruments working in the infrared and at longer wavelengths,
wave optics may be important. Designers have to use some hybrid of
ray tracing and wave optics to predict performance. Despite that, I
only know of one company that provides full wave-optics calculations,
which are quite cumbersome. Typically, the instrument design is done
using ray tracing plus analytical estimates of diffraction, then
checked using the full diffraction calculation.

The bottom line is that ray tracing is the standard tool and works
pretty well, but one has to be aware of when diffraction will matter.

--
Help keep our newsgroup healthy; please don't feed the trolls.
Steve Willner Phone 617-495-7123
Cambridge, MA 02138 USA

"Neil B." wrote in message
...

"Kai-Martin Knaak" wrote in message
news
On Mon, 03 Aug 2009 22:07:53 -0400, Neil B. wrote:

I still find it incredible, that wave optics
can ever make the image better than ray-trace in any but a minor
way.

Just accept, that ray-tracing ignores each and every effect of the
wavelength. Wave optics by contrast involves no such approximation.
So
ray-tracing can at best describe reality as good as wave-optics, but
never better.

Bottom line: If you get better resolution with ray-tracing than with
wave
optics and both methods have been applied correctly, then the outcome
of
wave optics is closer to reality. Whether, or not this is a minor
way,
depends on the parameters you are asking for.

---(kaimartin)---

I've agreed many times, if ray-trace resolution is *better* than wave
optics, the WO answer is closer to reality and therefore the image is
worse than the RT prediction. But the question motivating this entire
thread is: what if the RT resolution is substantially *worse* than the
ideal WO image for the aperture, then can WO "cure" the geometric
spread and make the image better than from the RT prediction? Jim
Black says it can, but in a roundabout way that I find hard to
appreciate how much the improvement would be (assuming he's right.)
Well, looks like there's some disagreement here. In a reply 8/5,
"Steve Willner" wrote in message
...

Can wave optics make the image even better (ie, smaller image of a
distant point source) than the ray-traced circle of confusion?

I can't think of a case where that would happen, but I can't prove
it's impossible. It certainly doesn't occur for typical telescopes
and instruments. Telescopes and instruments to work in visible light
are designed using ray tracing (and largely ignoring diffraction) in
most cases. Exceptions would be instruments that explicitly take
advantage of wave optics (e.g., interferometers, grating
spectrographs) or where diffraction effects are crucial (e.g.,
coronagraphs).

For instruments working in the infrared and at longer wavelengths,
wave optics may be important. Designers have to use some hybrid of
ray tracing and wave optics to predict performance. Despite that, I
only know of one company that provides full wave-optics calculations,
which are quite cumbersome. Typically, the instrument design is done
using ray tracing plus analytical estimates of diffraction, then
checked using the full diffraction calculation.

The bottom line is that ray tracing is the standard tool and works
pretty well, but one has to be aware of when diffraction will matter.

So one apparently cognizant person says, he doesn't think resolution
could be better than ray-trace, i.e. that wave effects can "cure" the
spread from a lumpy mirror [appreciably, forget little differences since
we're wondering if e.g. to get focused into spot maybe 1/10 of ray-trace
circle of confusion.]