Optical Interferometers for Exoplanet Detection

[redliox]

As I understand it, a traditional interferometer is a pair of telescopes that compare light from the same source and nullify the noise thanks to the slight differences in the 'scopes' signals.  Easier to do with radio and harder to do with optical telescopes since radio is on the scale of meters and beyond versus micrometers or less for nearly everything else.

We're starting to see satellites that coordinate with each other more often with lasers, and these spacecraft, even in LEO, are often hundreds of kilometers apart.  Commercial companies, not just government projects, are beginning to get crazy precise.

I have two questions in light of this precision as well as computer technology, AI or otherwise, that might allow for optical interferometry:

1) Do optical telescopes for interferometry need to be physically linked, or just aligned, to work?
2) With the right coordination, could telescope constellations pull this off? Pairs of satellites in LaGrange points come to mind for example.

[VSECOTSPE]

1) Do optical telescopes for interferometry need to be physically linked, or just aligned, to work?

A nulling interferometer can be made to work either way.  In the infrared spectrum useful for exoplanet characterization, the mission needs to control the positions of its telescopes to something like ~10^-5 meters, and IIRC, the mission needs to know the positions of its telescopes to within an order of magnitude less than that.  There are trade offs on whether that optical bench is on one spacecraft or distributed across several.

2) With the right coordination, could telescope constellations pull this off? Pairs of satellites in LaGrange points come to mind for example.

Yes, see the old Terrestrial Planet Finder Interferometer mission concept/studies.  They included both single spacecraft and multiple spacecraft missions for an infrared nulling interferometer.

https://en.m.wikipedia.org/wiki/Terrestrial_Planet_Finder

https://exoplanets.nasa.gov/exep/files/exep/TPFIswgReport2007.pdf

https://www.astro.princeton.edu/~dns/FRS120/RFI/134_TPF-I_Lawson_EOS.pdf

NGST/JWST cost growth ate the budgets for the other astronomical origins telescopes like TPF and its tech demonstrator SIM.  As a result, I think more progress has been made with coronagraphs over the past couple decades, so I don’t think a return to interferometers is on the horizon.

[edzieba]

As I understand it, a traditional interferometer is a pair of telescopes that compare light from the same source and nullify the noise thanks to the slight differences in the 'scopes' signals.

An interferometer interferes the light gathered from two or more locations via interference (hence the name). It's not a noise-suppression technique or a 'comparison' of two images, but an operation in the time domain that needs sub-wavelength-accurate synchronisation between the signals.
For long-wavelength (low frequency) radio interferometry, it is not unfeasible to use synchronised local atomic clocks to timestamp the recorded signal and then play back the synchronised signals to produce the interfered output. But to do the same in visible-light wavelengths is more than three orders of magnitude harder in terms of timing accuracy.
e.g. the Event Horizon Telescope made observations in the 1.3cm wavelength, or 23 GHz or ~43 picoseconds for a single cycle. An observation at ~500nm (O-III spectral line) is 600 THz or ~1.7 femtoseconds (0.0017 picoseconds) for a single cycle.

[redliox]

For long-wavelength (low frequency) radio interferometry, it is not unfeasible to use synchronised local atomic clocks to timestamp the recorded signal and then play back the synchronised signals to produce the interfered output. But to do the same in visible-light wavelengths is more than three orders of magnitude harder in terms of timing accuracy.

So multiple telescopes with essentially super-accurate-time-stamps monitoring the same star system at the same time could be digitally used as an interferometer?...

[sanman]

For long-wavelength (low frequency) radio interferometry, it is not unfeasible to use synchronised local atomic clocks to timestamp the recorded signal and then play back the synchronised signals to produce the interfered output. But to do the same in visible-light wavelengths is more than three orders of magnitude harder in terms of timing accuracy.

So multiple telescopes with essentially super-accurate-time-stamps monitoring the same star system at the same time could be digitally used as an interferometer?...

Isn't that just known as Very Long Baseline Interferometry?

[edzieba]

For long-wavelength (low frequency) radio interferometry, it is not unfeasible to use synchronised local atomic clocks to timestamp the recorded signal and then play back the synchronised signals to produce the interfered output. But to do the same in visible-light wavelengths is more than three orders of magnitude harder in terms of timing accuracy.

So multiple telescopes with essentially super-accurate-time-stamps monitoring the same star system at the same time could be digitally used as an interferometer?...

Isn't that just known as Very Long Baseline Interferometry?

Yes, and as I mentioned upthread it works for long wavelengths (RF) but not for short wavelengths (optical) due to the timing requirements. More accurate and more stable clocks would be a prerequisite for optical VLBI, but not the only one (e.g. at the scales of the wavelength of light, the 'solid' rock of the Earth between two widely spaced observatories will be wobbly and spongey).

[Paul451]

It's worth pointing out that the issue isn't just timing. You are "interfering" the properties of the waves to produce the desired output.

Radio receivers can capture that wave-based information, convert that to an electronic signal and interfere the electronic signal to mimic a physical interaction of the input radio.

Optical sensors, otoh, are glorified photon counters. All the information you need for interferometry is lost, so you can't do electronic interferometry.

Hence to do optical interferometry, you need to use the original light itself. Ie, the actual input light coming into each telescope has to be directed into a single shared physical interferometer. That adds a whole extra layer of complex optics on top of the telescopes themselves, and trying to get that to work over any significant distance...

[redliox]

https://theconversation.com/a-new-thin-lensed-telescope-design-could-far-surpass-james-webb-goodbye-mirrors-hello-diffractive-lenses-206055

[geza]

It's worth pointing out that the issue isn't just timing. You are "interfering" the properties of the waves to produce the desired output.

Radio receivers can capture that wave-based information, convert that to an electronic signal and interfere the electronic signal to mimic a physical interaction of the input radio.

Optical sensors, otoh, are glorified photon counters. All the information you need for interferometry is lost, so you can't do electronic interferometry.

Hence to do optical interferometry, you need to use the original light itself. Ie, the actual input light coming into each telescope has to be directed into a single shared physical interferometer. That adds a whole extra layer of complex optics on top of the telescopes themselves, and trying to get that to work over any significant distance...

Saying the same differently. Our electronics can only handle frequencies below a maximal frequency. We call an electromagnetic wave, as radio wave, if we can produce it electronically, i.e., if its frequency is below that maximum. Such wave can be recorded electronically, as an oscillation. We can time-stamp it. Afterward, we can compare the time-stamped oscillations recorded independently in different radio telescopes. The timing (phase) differences provide us the directional information. The distance between the radio telescopes can be arbitrarily large. They need not interact in real time.

We cannot do the same by optical frequencies, because an optical oscillation is just too fast for our electronics. We have no way to observe and record the phase of the incoming light. Locally, we can just measure the amplitude of the oscillation. The only way to detect phase differences is to arrange an interference between the incoming lights. For this the two optical telescopes must be part of a single optical system during the observation. Their relative geometry must be controlled with a precision scaled by the wavelength. This is non-trivial even if they are mechanically connected. Controlling relative positions of free-flying spacecrafts with such precision is extremely hard - but not impossible.

[edzieba]

We can detect and measure optical phase with electronics, it just doesn't help with VLBI.
Optical phase sensors are commodity items (e.g. every Kinect 2 camera contains a 2D array of phase sensors in order to perform phase-offset time-of-flight measurements), the problem is those sensors are 'clocked' relative to a reference signal, so you still need a phase-accurate reference signal that can be transported over useful baseline distances without distortion. If you e.g. use a laser reference over a single continuous optical fibre for transport (and make sure its entire length is protected from expansion, contraction, stress, thermal changes, and other sources of error) you still have the problem that the two ends are many trillions of oscillations apart but you do not know how many (and what the offset non-integer fraction is, which is what you really need to know) and that physical motion of both observing ends will also be changing their phase offset in a manner not reflected by that reference signal.

[Solarsail]

I wonder if it would be desirable to partially approximate interferometry using unusually arranged mirror segments on a physical telescope structure (with in space manufacture).  Say, instead of approximating a circular mirror configuration with mirror segments, how about positioning mirror segments in a straight line?  What kind of image quality could you get with 2 meter diameter parabolic-offset mirrors arranged in a line, 100 meters wide by 2 meters high?

[edzieba]

I wonder if it would be desirable to partially approximate interferometry using unusually arranged mirror segments on a physical telescope structure (with in space manufacture).  Say, instead of approximating a circular mirror configuration with mirror segments, how about positioning mirror segments in a straight line?  What kind of image quality could you get with 2 meter diameter parabolic-offset mirrors arranged in a line, 100 meters wide by 2 meters high?

You would end up with effective resolution being anisotropic.

[sanman]

What about Very Long Baseline Interferometry?

Is it doable for optical wavelengths, or just a pipe dream?

[laszlo]

What about Very Long Baseline Interferometry?

Is it doable for optical wavelengths, or just a pipe dream?

Did you not read all the responses since your last post? The question has been answered.