idea for statistical correlation telescope with no optics

The trouble with this sort of design is that as the intensity of the light source drops, the amount of time required to integrate goes up, way up! So while in theory this might work for the brighter stars, anything remotely dim may well take more time than we have left before the night sky is blotted out by our star going supernova.

Certainly though it can be used to calculate the position of at least one star... ours.

Hi,

I guess there is a formula for that to relate the information gathering rate of the system via the detector to the information change/decay of the system. If the information gathering rate is not high enough then as you say the system will change before the useful data can be correlated.

If the system was static enough, you could basically run a random number generator and just correlate the output of it and detect properties of the surrounding system that the random number generator exists in, since I think all random number generators output are determined by the system they exist in. With enough time you could theoretically model the whole static universe just monitoring and correlating the random number generator output maybe! :D

cheers, Jamie

In principle, yes, but it requires a coherent detection system (a laser LO beam plus phase sensitive processing). Back around 2001, when I was starting out my optical antenna work, I proposed doing this using a variant of Fienup's phase retrieval algorithm, using a whole bunch of phase-insensitive back ends and a few phase sensitive ones (to seed the phase retrieval process and help it converge rapidly). You should be able to make a 2 pi steradian camera with almost no optical system. (Of course there are huge SNR issues that you have to get over before it becomes practical.)

Cheers

Phil Hobbs

Hmmm... there are some contradictions in what you just posted. First, when you talk about "running" a random number generator that would imply a computer program which is not random by definition. All such computer programs are pseudo-random meaning they have some properties of random but are not truly in that they can be predicted... just as the computer is doing.

Or you could be talking about some "random" process in the real world. But that also contains a contradiction in that again, the definition of random means it is not correlated to the rest of the real world or - again - it could be predicted. In reality, random is just a concept. There are no true random processes other than possibly radioactive decay. However, monitoring anything that is truly random will not tell you anything about the rest of the universe no matter how long you observe it.

Hi,

I meant running a random number generator that is directly coupled to the environment. If you are using a software pseudo random number generator, then the only input from the external world will be unexpected errors in the output. This is effectively a very low data rate (even zero theoretically perhaps) for correlating external information.

cheers, Jamie

Hi,

That would be cool to have a low noise space based telescope using that type of detector maybe..

I just saw this today I guess it may be somewhat related as it can break down the properties of detected light:

cheers, Jamie

This is making less and less sense. Take measurements of something and then add random noise so that you can average out the noise... :?

Hi,

The pseudo random number generator is an extreme case of low information rate about the surrounding environment, since it is just like any other isolated abstract software. The only time it gives information outside of the software is if it does an unexpected calculation/error. A "true" random number generator (ie. one based on measuring radioactive decay) is directly coupled to the physical surroundings via the decay source (which itself is coupled to the rest of the physical surroundings).

If you correlate a large enough portion of the output of a pseudo random number generator that has no errors then you will be able to derive the pseudo random number generator formula, but if somehow there are errors in the output, then the correlation will also show properties of the environment the pseudo random number generator is in.

cheers, Jamie

It has already been done quite effectively in two different variants:

Hanbury-Brown & Twiss Intensity Interferometer used photon temporal correlation function across a pair of large mirror based scopes over a variable baseline to first measure the diameters of the brightest stars.

There is a book about their endeavours first at Jodrell Bank and later at Narrabi - free access copy of their paper from MNRAS at:

Various HEP types in the crossover with astronomy have instrumented remote deserts and even deep ocean with a grid of photon detectors to measure extreme energy cosmic rays and solar neutrinos respectively.

It was used to measure the diameters of all the brightest stars.

Hanbury Brown is one of my technical heroes. With some formatting help from various folks here, I put up a PDF of his book, "The Intensity Interferometer" at

(It's been out of print for over 30 years.)

Besides the SNR problems, the intensity interferometer can only measure the autocorrelation of the scene, because all the phase information in the transform is lost. That's why my scheme used some phase-sensitive pixels and some crystal-radio pixels--to constrain the phase retrieval process and make sure you actually get the right answer.

It was always a long shot, and I never had a good enough tunnel junction process to make umpty-ump thousands of working junctions on a wafer.

Cheers

Phil Hobbs

Just a side note - that will never happen. Our star is insufficiently massive to become a supernova. Though it will become a red giant, which would have much the same effect on the measurements.

Sylvia.

Hi,

That is a different concept requiring two detectors.

cheers, Jamie

Oops, I guess it is the same concept, ie with a single detector as the earth rotates it can function as the second detector too.

cheers, Jamie

Nope. Gotta look at the instantaneous correlations. Just one detector won't work.

Cheers

Phil Hobbs

Hi,

I think you can disregard the temporal offset if there are enough samples to correlate together, however it is a lot more data required than a synchronized 2 detector sample.

cheers, Jamie

Nope, sorry. Read the book--I posted a link to a softcopy.

Cheers

Phil Hobbs

Hi,

Thanks I downloaded it, I think maybe the interferometer converts the phase difference between the two detectors into a light intensity proportional to the diameter of the star basically? (sorry I need to read it more I think)

My idea of using a single detector and correlations has no phase information that can be used to detect a star diameter, but it can still theoretically detect a star diameter as the earth rotates, the star will enter the field of view of the detector at an edge and the light intensity will increase as the full diameter of the star crosses into the detector field of view. This is a statistical below noise measurement but I think theoretically possible to detect a stellar diameter statistically just using intensity measurement, and statistically many star diameters could be detected at the same time with a single pixel detector this way *theoretically* :)

cheers, Jamie

There is a book about their endeavours first at Jodrell Bank and

Various HEP types in the crossover with astronomy have instrumented

Nope. You have to have two time series to correlate them, and if you're looking at the classical fluctuations of the electromagnetic field (which is what HB & Co. were doing), you have to measure them together. Otherwise the correlation will always be very very nearly zero, and you won't get any data.

Well, you'd have to actually do the theory. You don't really even have an arm-waving argument so far.

Cheers

Phil Hobbs

Hi,

For the intensity interferometer, I am following the classical wave description on the wikipedia page:

It mentions a phase difference at the two detectors as what causes the intensity difference. I am having a hard time seeing how this can be used to determine the apparent angular diameter of a far off star. Are multiple measurements required from the intensity interferometer to determine the angular diameter and do the optics have to be pointed at more than a single coordinate, ie. do a sweep across the star's angular size?

For my own single detector idea, I guess the theory of how it could possibly measure star diameter is just an extension to the hand waving argument of how it could detect stars to begin with, in my first post in this thread. If you view the whole sky as a pixel array, and use the single intensity detector that is doing sky sweeps while earth rotates/orbits the sun, then over time statistical correlations will be able to fill in pixels in the sky as likely light sources. Over time theoretically this pixel image could become higher resolution until the light sources themselves can be pixellized into dimensioned objects.

(hand waving mode off!)

cheers, Jamie