BRPS observations (CPU)

Fair comments. It's a balance I suppose, engaging people to donate resources ( yes, it's quite a lot of real money saved ) without the cart leading the horse though. Largely I see it as a challenge to gain genuine trust from pretty much total strangers.

I'm not sure what S6 data is to be used - it was pretty noisy I think. No-one seems to mention it much, and I haven't asked. It could be a sore point? One thing about good GW data is that you can re-mine it ever more deeply. It's a question of how sure one wants to be, to not miss a given signal type to some level of signal power. This is also true for radio pulsar work except for one important thing - we've not yet ever heard a GW signal.

For ABP/BRP we are just mining known signal types. If you look at pulsar signal profiles they have a similiar general form : such and such a width with two or one peaks, symmetric or pretty much so, with beat-to-beat variability. In fact the profile averaged over many cycles is sort of like a fingerprint of each source system, reminiscent of sonar profiles identifying specific sea vessels.

So one can always return to GW data with a new idea on signal profile. The other thing is the method to reveal a signal at a given profile. It's technically called convolution ( but enacted with the magic of Fourier transforms ), which is akin to stepping along a given signal in the time domain with a template and seeing to what degree a match is obtained. A template would maximally match itself for instance if it was perfectly aligned or in-sync, and less so if the template is slid to either side. You can see that the common area under both curves is at a maximum for match, and then moves to zero with more mismatch. Or you can multiply the template to the signal for a given time offset and integrate over some extended range, it all winds up being much the same calculation ( except it's not actually done that way, but through work in frequency space after Fourier transform, as convolving in the time domain essentially multiplies the respective transforms in the frequency domain ). But I digress.

The signal from a detector is a differential b/w two laser locked arms. This makes a given LIGO setup ( or Virgo or GEO ) fairly omni-directional. In three dimensions the 'antenna' response looks like a bi-lobed peanut shell ( axis of rotational symmetry perpendicular to the plane of the arms which is also the plane of reflection symmetry for the pattern as well ) with nuls or deafness along certain lines away from the corner station ( there are wave modes that make both arms change length identically so there's no differential pickup ). On a per detector basis one can try any sky direction to see what signal 'out there' could have produced a given antenna response here, and you can try to time correlate b/w separated detectors to see if their data supports the idea of commonly responding to some putative 'out there' signal. I've skipped describing GW polarisation patterns, but et cetera ....

The other test of GW data includes varying rates of spin-up or spin-down of the source system so that's another variation to include.

Longer periods of data ( preferably contiguous ) give a greater chance of a true signal rising out of background noise, the signal-to-noise ratio as judged for significance is 20 to 1. Which is pretty strict, but I respect that because our first GW wave detection will be the first ever, and a detection alone has high significance for gravitational theory.

Cheers, Mike.