3 ms·

Any idea what the duration of these GW events is, are we single blips talking microseconds or milliseconds, or a series of blips with distinctive patterns over

Any idea what the duration of these GW events is, are we single blips talking microseconds or milliseconds, or a series of blips with distinctive patterns over several seconds/minutes? or something else?

stefco_ 9y ago
BH/BH mergers are in our sensitive frequency band for less than a second, and they get louder and higher-frequency until their peak volume, at which point they ring-down to silence. Neutron stars should spend longer in our sensitive band. There are also probably pulsars and (maybe) other signals out there that are nearly periodic and that we are hearing all the time! But those would be much much quieter (think of a burning flame vs. an explosion; the faster you release your energy, the more noticeable it is). Furthermore, those sources should be pretty close to a pure sine wave. That's hard to detect, because so much of our noise looks similar. We look at compact binary coalescences so closely in part because they are so distinct that we can be fully confident that our detector didn't produce them (we have a very talented and suitably paranoid approach to studying our background noise without prior assumptions of its form). You can't be so sure with these faint, continuous sinusoidal signals UNLESS you do a fully coherent search over long periods of time. Our detectors are not equally sensitive in all directions, and they can pick up direction as a phase-delay between the detectors in our network. As the earth rotates, both detected direction and strain amplitude from a a continuous source should vary. You can imagine doing what is basically a very, very long timescale fourier transform of the signal; the longer you integrate over, the more you filter out sine waves with frequencies nearby your chosen frequency. But this is a computationally expensive approach because longer integration times mean you need to test more frequencies as well as doing the longer integration. There are approaches to finding good compromises re: fully-coherent search windows, and I'm sure there's plenty of other clever voodoo that goes into the continuous waveform searches, but I don't know much more than that.
- redcalx 9y ago
  Thanks very much. From what you're saying, I wonder then in a hypothetical scenario in which we were 100% confident that all local noise (all non-GW based noise) was eliminated, I imagine we'd be left with lots of actual GW based noise, from which a greater number of specific events could be picked out with the type of analysis you're talking about. I wonder then if there's work to understand how much GW noise there is, and work on lowering the local non-GW noise floor further, etc. I guess that just means ever larger detectors - and of course correlating results from detectors around the world. All very cool and interesting work anyways.

stefco_ 9y ago

BH/BH mergers are in our sensitive frequency band for less than a second, and they get louder and higher-frequency until their peak volume, at which point they ring-down to silence. Neutron stars should spend longer in our sensitive band. There are also probably pulsars and (maybe) other signals out there that are nearly periodic and that we are hearing all the time! But those would be much much quieter (think of a burning flame vs. an explosion; the faster you release your energy, the more noticeable it is). Furthermore, those sources should be pretty close to a pure sine wave. That's hard to detect, because so much of our noise looks similar. We look at compact binary coalescences so closely in part because they are so distinct that we can be fully confident that our detector didn't produce them (we have a very talented and suitably paranoid approach to studying our background noise without prior assumptions of its form). You can't be so sure with these faint, continuous sinusoidal signals UNLESS you do a fully coherent search over long periods of time. Our detectors are not equally sensitive in all directions, and they can pick up direction as a phase-delay between the detectors in our network. As the earth rotates, both detected direction and strain amplitude from a a continuous source should vary. You can imagine doing what is basically a very, very long timescale fourier transform of the signal; the longer you integrate over, the more you filter out sine waves with frequencies nearby your chosen frequency. But this is a computationally expensive approach because longer integration times mean you need to test more frequencies as well as doing the longer integration. There are approaches to finding good compromises re: fully-coherent search windows, and I'm sure there's plenty of other clever voodoo that goes into the continuous waveform searches, but I don't know much more than that.

redcalx 9y ago
Thanks very much. From what you're saying, I wonder then in a hypothetical scenario in which we were 100% confident that all local noise (all non-GW based noise) was eliminated, I imagine we'd be left with lots of actual GW based noise, from which a greater number of specific events could be picked out with the type of analysis you're talking about. I wonder then if there's work to understand how much GW noise there is, and work on lowering the local non-GW noise floor further, etc. I guess that just means ever larger detectors - and of course correlating results from detectors around the world. All very cool and interesting work anyways.