Plans for near future of E@H ?
1 Nov 2012 11:24:38 UTC
Topic 196604
(moderation:
Hallo!
FGRP1 is out of data and S6LV1 will end close to the end of this year.
So I had a look into LSC-Virgo data analysis white paper - many thanks to joe areeda for posting this((:-))) -. But the only I found directly related to E@H was on page 131, in chaper 7 "LSC Computing and Software" the last block: "The LSC also developed the Einstein@Home project to leverage an alternative distributed computing paradigm for its most formidable computing challenge, the search for gravitational waves from isolated pulsars. The pulsar analysis puts reduced demand on quick turn-around and has low data flow, but requires PFlops of computing power. The ....". (We just reached 0.7PFLOPs.)
On A@H are no following up projects for test now.
For fresh data form the Advanced LIGO detectors we will have to wait for another year at minimum, according to the official time schedule.
So what will be the near future of E@H? Will we crunch BRP1 tasks on our CPUs, which is much less effective?
For a short answer we will be pleased.
Kind regards and happy crunching
Martin
P.S.: I found the "LSC-Virgo data analysis white paper" still very much interesting.
Plans for near future of E@H ?
)
Correct. But there may still something to be found in the existing data. For "continuous waves" searches the computing power is still what limits the sensitivity of a search, which roughly translates to the depth we could dig into space.
We are already preparing the next GW search, which will focus fewer promising sky positions, allowing us to reach out further than ever before.
You think? In contrast to the GW search the Radio-Pulsar search lead to a couple of really interesting new discoveries. In my very personal view this is more effective than the GW searches we did so far.
We still haven't caught up with Arecibo data production, AFAIK we haven't yet processed any data from 2012. There are still a lot of Radio Pulsars out there waiting to be discovered!
And finally the Gamma-Ray Pulsar search is suspended because we ran out of manpower, not of data. I hope we can resume that work soon.
Short answer to a long question?
BM
BM
Hallo BM! Thank you for quick
)
Hallo BM!
Thank you for quick answering.
I believe so. I remember, in early days of EaH we where happy reaching 100GFLOPS, and one task lasted much more than a year to finish. Now we have 7000 times of this crunching power, which gives more freedome for more complex project tasks and longer averaging times, resulting in higher sensitivity.
Sorry, my sentence was incomplete. I compared crunching on CPU / GPU. If I remmber correctly the BRP tasks where running for about 10h on CPU instead of some 10min on GPU. Comparing the requiered electrical power to crunch 1 BRP task, I suggest the GPU is also in forehand. Did someone test this? The figure is highly dependend on the used hardware.
We´re happy to heare this and do wait.
Kind regards and happy crunching
Martin
RE: You think? In contrast
)
So, Einstein@Home can profit from the results of pulsar hunting?
I always thought that this secondary project only diverts the resources.
RE: RE: You think? In
)
The ultimate goal of Einstein@Home is to contribute to a better understanding of our universe by studying neutron stars, objects with such extreme conditions that they drive our theories of matter and forces to the limits (and possibly beyond), allowing to test and refine these.
The only way we can study these objects is through their emissions we could detect on (or near) earth - electromagnetic waves (like radio or gamma) or gravitational waves.
In that sense every result, in particular every new discovery of a pulsar, helps a bit to understand - what is possible, and what not, which theories and models do match best what we observe. So all three searches currently running on Einstein@Home do help the general goal.
As for the more narrow goal of detecting gravitational waves, the radio pulsar discoveries of Einstein@Home do help to form pulsar timing arrays that will help us to measure gravitational waves of a different wavelength than what the LIGO detectors and the GW search on Einstein@Home is sensitive for.
Finally the radio pulsar search allows us to keep and even increase our user base and thus computing power, even during periods when there is no new data from the GW detectors available - like now.
BM
BM
RE: RE: In contrast to
)
So you meant 'efficient' instead of 'effective'.
BM
BM
I remember the radio pulsar
)
I remember the radio pulsar search as being initiated to solve several problems : an unavoidable hiatus in new data from the IFO's ie. keep the punters interested while we await GW detector upgrades, to help the PALFA consortium's need for processing their data, and to orthogonally study the same class of stars/systems as GW's will anyway. The gamma search is along the same lines.
If that understanding is correct then : the radio and gamma work have done brilliantly. As for whether it is 'correct' for the E@H project to have solved such issues is a higher question. In any case the value of GW work has not been lost, indeed upper limits have now been placed on certain signal types ( of interest to observers and theorists alike ) not to mention an enormous slab of valuable experience in running the analysis 'pipeline' that E@H is a part of. For the first, or for that matter subsequent, GW detections then validation and confidence in the entire enterprise will be crucial.
As a 'leading edge' activity I think it is quite reasonable & healthy to assess directions from time to time, to reflect upon progress or otherwise and try to make good guesses for the future. Else it wouldn't be called 'research' would it ?? :-)
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
RE: We are already
)
Actually we will first extend the current S6 LineVeto search a bit to higher frequencies.
A number of recent publications and own results of pulsar searches (in radio and gamma-ray) lead us to think that there is a larger population of 'younger' (i.e. faster spinning) pulsars than we expected when we initially set up the search.
Note that this is another example of interaction between the different pulsar searches on Einstein@Home.
BM
BM
RE: RE: We are already
)
Bernd what frequencies are we going out to ? Do we still have the WU count per frequency going like quadratic to frequency ?
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
RE: Bernd what frequencies
)
This is still being discussed. Most likely an additional 50Hz or so. The higher we get, the more we would need to include a possible 2nd order frequency derivative. Our current code should be able to do that, but that hasn't been (sufficiently) validated yet.
Yes, as long as we stick to only first frequency derivative ("spindown").
We are currently working on the Application. It will be only slightly different, returning two result files, one containing the older "2F" statistics, the other containing the newer "LV". In principle we are extending both previous runs (S6LV1 and S6Bucket) at once.
BM
BM
Ah I see. Not only faster
)
Ah I see. Not only faster spinning, but faster evolving too.
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