There is a publication from the S4 run here: http://www.ligo.org/results/. It's a 37 page PDF, not the nice simple explanation for crunchers that they eventually came up with for S3. Still, at least it shows that something is going on.
There was also a meeting of the LIGO group on 4-5 November, nothing published from it yet.
My condensed version of the 37 pages:
------------------------------------------------
The S4 run was about 10 times better at sniffing out gravitational waves than S3, yet still nothing has been found so far. The LIGO experiment is now getting sensitive enough to actually help guide theoretical cosmology, including ideas coming from string theory and pre-bing-bang models, so let's cheer up anyway ...
------------------------------------------------
:-)
Please correct me if I misunderstood - much of the stuff went above my head.
I'm back ... we've all had a series of interruptions, like Bruce moving and the shippers deciding to keep his stuff, plus the servers all dying in the middle of it. Sorry about that, and thanks for hanging in there.
The S4 post-processing finally got back on track and is well along. I was hoping to present results at the American Physical Society meeting in a couple of weeks. It looks like we won't make that deadline, but shouldn't be too far behind. Then we'll have to have the results vetted by the LIGO Science Collaboration before making them public. That took longer than we thought last time, but next time should go quicker since we learned some lessons.
I have to confess, part of the S4 slowness seems to be my fault. If you look closely you'll see that the search grid in terms of sky positions is different. It's not a simple grid like S3 was, but changes in funny ways across the sky. That's something I put in, which cuts down on the CPU cost by taking advantage of information theory to tell you which directions you need to cover with a lot of points and which are OK to use a few. It helps us get more science per CPU cycle and is the mathematically "proper" thing to do.
But it played hell with the validator. That had to be completely rewritten for the S4 analysis because the grid is different for different stretches of data and you can't just compare two work units at the same sky point. The grid changes because the Earth's velocity vector moves around with time, and the information in the signal is based on that. And now that we're on the post-processing, we're finding that is also more complicated. So it saved CPU time, which thanks to you folks is plentiful, but wound up costing more scientist time that we thought - and that is a scarce resource.
So the improved S5 search is going back to a simple grid. Actually, the interaction of the grid with the hierarchical search would be even more complicated, which gives us more incentive to keep it simple. One of these days we may go back to the fancier, more CPU-efficient grid when we've got other things under control, but for the next while we're trying to keep it simple and keep things moving.
I'm back ... I have to confess, part of the S4 slowness seems to be my fault .... But it played hell with the validator ..... trying to keep it simple and keep things moving.
Now there's some subtle points, eh? I for one, will forgive you .... :-) :-) :-)
Such is the nature of this new beast that this sort of thing is likely to happen and is part & parcel of the process of discovery. It certainly is not evident that the effect on the servers could have been predicted. Now we know .... how lovely is hindsight. :-)
Thanks for the insights!
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
2 weeks later and no link
)
2 weeks later and no link yet? I know they are extrememly busy and have other things to do, but I'm impatient :p
There is a publication from
)
There is a publication from the S4 run here: http://www.ligo.org/results/. It's a 37 page PDF, not the nice simple explanation for crunchers that they eventually came up with for S3. Still, at least it shows that something is going on.
There was also a meeting of the LIGO group on 4-5 November, nothing published from it yet.
RE: There is a publication
)
My condensed version of the 37 pages:
------------------------------------------------
The S4 run was about 10 times better at sniffing out gravitational waves than S3, yet still nothing has been found so far. The LIGO experiment is now getting sensitive enough to actually help guide theoretical cosmology, including ideas coming from string theory and pre-bing-bang models, so let's cheer up anyway ...
------------------------------------------------
:-)
Please correct me if I misunderstood - much of the stuff went above my head.
Greetings, Mr Ragnar Schroder
This might be helpful: LIGO:
)
This might be helpful: LIGO: Status of operation at design sensitivity and future prospects (Sep. 12, '06)
Hi folks, I'm back ...
)
Hi folks,
I'm back ... we've all had a series of interruptions, like Bruce moving and the shippers deciding to keep his stuff, plus the servers all dying in the middle of it. Sorry about that, and thanks for hanging in there.
The S4 post-processing finally got back on track and is well along. I was hoping to present results at the American Physical Society meeting in a couple of weeks. It looks like we won't make that deadline, but shouldn't be too far behind. Then we'll have to have the results vetted by the LIGO Science Collaboration before making them public. That took longer than we thought last time, but next time should go quicker since we learned some lessons.
I have to confess, part of the S4 slowness seems to be my fault. If you look closely you'll see that the search grid in terms of sky positions is different. It's not a simple grid like S3 was, but changes in funny ways across the sky. That's something I put in, which cuts down on the CPU cost by taking advantage of information theory to tell you which directions you need to cover with a lot of points and which are OK to use a few. It helps us get more science per CPU cycle and is the mathematically "proper" thing to do.
But it played hell with the validator. That had to be completely rewritten for the S4 analysis because the grid is different for different stretches of data and you can't just compare two work units at the same sky point. The grid changes because the Earth's velocity vector moves around with time, and the information in the signal is based on that. And now that we're on the post-processing, we're finding that is also more complicated. So it saved CPU time, which thanks to you folks is plentiful, but wound up costing more scientist time that we thought - and that is a scarce resource.
So the improved S5 search is going back to a simple grid. Actually, the interaction of the grid with the hierarchical search would be even more complicated, which gives us more incentive to keep it simple. One of these days we may go back to the fancier, more CPU-efficient grid when we've got other things under control, but for the next while we're trying to keep it simple and keep things moving.
Hope this helps,
Ben
RE: Hi folks, I'm back ...
)
Now there's some subtle points, eh? I for one, will forgive you .... :-) :-) :-)
Such is the nature of this new beast that this sort of thing is likely to happen and is part & parcel of the process of discovery. It certainly is not evident that the effect on the servers could have been predicted. Now we know .... how lovely is hindsight. :-)
Thanks for the insights!
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal