Ask a Physicist

In terms of interferometry, is the baseline between LIGO and the German detector long enough to be able to pinpoint the source of the waves? The Einstein@home screensaver displays the location on the sky of where it is searching, so how is this directionality achieved? This seems to imply that it is an active system rather than just a passive one of detecting waves and then trying to pinpoint them. Or have I misunderstood?

Physically, LIGO is passive. It sits where it is, and that's it. The “pointing” is done in the processing of the data.

How that's done is different for different types of sources. You might guess that with two LIGO sites, something like triangulation could be used to pick up a direction. That's basically true for short-lived signals, although the directionality even then is not too great. LIGO is more like an ear than an eye, since the wavelengths are long compared the size of the detector, and if you try with your eyes closed you'll find that you can't localize sounds nearly as well as sights.

But long-lived periodic signals - the ones Einstein@Home is searching for - are another matter. Even if you start with something (like a bump on a rotating neutron star) that gives off a perfect sine wave signal, it won't stay that way by the time it gets into the data stream. The detectors are stuck to the Earth, which spins in little circles every day and big circles every year. That motion changes (Doppler shifts) the frequency of the signal in a complicated way that is a function of time and also of position on the sky. For example, a source located over the North Pole will not be Doppler shifted by the daily rotation but will be affected by the Earth's orbital motion. And this doesn't depend on having multiple detectors with a long baseline between them, although that's good for other things; it just depends on the Earth moving.

Those complicated Doppler shifts are where the angular resolution comes from. The data analysis has to compensate for the Doppler shift to make any signal as sinusoidal as possible, which helps pull it out of the noise (with something based on a Fourier transform). It has to do one Doppler shift for one sky location, then Fourier transform to see if there's something there; another Doppler shift for another sky location, then Fourier transform; and so on. For an in-depth search, even a small change in sky location makes the Doppler shift different enough to completely wash out any signal if done wrong. The net result is a lot of sky positions to search.

So while the raw data contains (presumably) signals from all over the sky, the processing code picks a single point on the sphere, corrects for the Doppler effect at that point, and looks for periodic signals. The way the correction is done washes out anything that is not coming from very near that point, so then the code repeats for another nearby point, and another, and another....

There's the rub: A more sensitive search needs more sky positions, which means more CPU cycles. Thus Einstein@Home.