What are we doing? - The science behind BRP5
14 Jun 2013 7:37:50 UTC
Topic 196999
(moderation:
I apologize if this is the wrong place to post this question.
I see that we are re-examining, or "extending" the search in the Perseus Arm data. BRP5
Are we just re-examining old data to have something to do, or did something in the application change the parameters of the search, or is this data we didn't look-at before from the same source?
I can't find much in the way of an explanation. Maybe I don't know where to look.
Any help would be appreciated.
What are we doing? - The science behind BRP5
)
Hi,
I'll use this opportunity to tell you a bit about the BRP5 "Perseus Arm Survey" search.
This search uses data from the Parkes radio telescope in Australia, which also provided data for the BRP3 (Parkes Multi-beam Pulsar Survey = PMPS) search. As you know, BRP3 lead to the discovery of 24 new pulsars, the scientific publication can be found here. We expect it to be accepted for publication in the Astrophysical Journal very soon.
The Perseus Arm Survey (PAS) is an extension of the initial Parkes Multi-beam Pulsar Survey, covering a different region of the sky (the Perseus Arm) than the PMPS. The set-up of the observations is almost identical to that of the PMPS. This makes it easy for us, to adapt our BRP3 search pipeline to this new data set, which has never been searched on Einstein@Home before.
The PAS has been analyzed before, but only with conventional methods which are not very sensitive to pulsar in compact binary systems. The results of this analysis were the discovery of 14 new pulsar, which are published here. Our goal is to find more previously unknown pulsars missed by the first analysis. We're confident that this will work, because we've seen in the PMPS analysis, that there are differences between the various pipelines and that E@H found pulsars missed by earlier analyses.
The main difference with respect to the BRP3 run is that we have extended the orbital parameter space significantly. In BRP3 we had to limit ourself to pulsars which orbits were oriented such that the radius of the binary motion projected along our line of sight is small. Searching for these systems is easier because the Doppler effect is not as strong as for those with large projected radii.
For BRP5 we have now extended this to larger values. This allows us to find more pulsars in binary systems than before, but is computationally more demanding. The number of orbital templates (that is, the different kinds of binary orbits we have to test for) increased by a factor of 20 from roughly 12,000 to roughly 240,000.
This large number of orbital templates means that the computation time increases by about the same factor, actually a bit less due to other savings in the code. In any case, the application would run unacceptably long on CPUs. Therefore, BRP5 is a GPU-only run.
The total projected runtime is of order eight months, although it might be a shorter in the end as I expected an increasing number of powerful GPUs to show up on the project. We already have an extension of that run prepared, which would then complete coverage of the orbital parameter space we're interested in. That second half would run for the same time as the first BRP5 run, underway now.
I you have further questions, let me know. I'll keep an eye on this thread.
Cheers,
Benjamin
P.S.: I moved this thread into the Science forums, after all this is its topic.
Einstein@Home Project
Thank you Benjamin! It's
)
Thank you Benjamin! It's really good to get more info on whats behind the crunching that we do.
Gord
I appreciate your taking the
)
I appreciate your taking the time to explain.
Thank you.
You said, "For BRP5 we have now extended this to larger values."
I have seen mention of a binary orbital period of 937 days.
Can you give a big, broad, general, non-specific, fuzzy maximum orbital period for our current search? I would prefer not to trouble you to look-up a specific answer as it wouldn't matter.
Are we talking decades or centuries? I wouldn't understand or have curiosity beyond a broad, non-specific answer like "decades" or "centuries."
Would this search be able to identify whether or not such a long period orbit were decaying, or is the observation time just too brief? I thought I should explain that I wondered if there would be an orbital almost-fit, and a fit, etc, until you had a "tendency")
I really appreciate your time and trouble to satisfy my curiosity and make the search more interesting and therefore more rewarding.
Hi, RE: You said,
)
Hi,
Yes, that concerns the projected orbital radii. What I had written might've been confusing. Let's take a step back to help you understand the computational problem.
You find pulsars by detecting their clock-like pulsed signals. We do this by using a method called the Fourier transform, which represents the observational data "in frequency domain". That means that if there was a pulsar spinning 45 times a second, there would be a signal at a frequency of 45 Hertz. Spinning twice per second… 2 Hertz. While this regular pulsing might be tricky to see by just looking at the observational data directly, it is really easy to see if you look at the data with the Fourier transform.
Now, this only works if the pulsar spin frequency doesn't change over the time of the observation. If the pulsar is "isolated" (not in orbit with a companion star), then this is the case. All you need to do, is Fourier-transform the data, look at the results, and pick out strong signals. It's a bit more involved, but that's in principle what's happening.
If, however, the pulsar is in a binary system, the orbital motion will result in Doppler effect, which in turn makes the spin frequency go up or down, depending on the exact orientation, size and period of the orbit and the position of the pulsar along the orbit. Then the Fourier transform does not work as well as before, since the signal is smeared over many neighboring frequencies. It's signal-to-noise ratio decreases significantly. When you know the exact orientation, size and period of the orbit and the position of the pulsar along the orbit, you can correct for that and recover the signal. Since for unknown pulsars, this is unknown we have to search over a wide range of parameters.
Now comes the relevant part. The Doppler effect becomes stronger for short orbital periods and large orbital sizes. It is more difficult to search for short orbital periods at relatively large orbital sizes (=projected orbital radii). Our search is tuned to be sensitive to very short orbits. But for the first part of BRP5 we have set an upper limit on the orbital sizes (=projected orbital radii) we are sensitive to. That means we can see systems in very short orbital periods, but not if the radius is too large (= the companion is too massive). For BRP3 that limit was even tighter, which is why I set that we now extended our search to larger values.
The long orbital periods are very easy to detect. If the orbital period is longer than the observation time, the pulsar basically doesn't move at all during the observation. Thus, no Doppler effect happens, and the pulsar appears almost isolated.
For the BRP5 search we are searching for orbital periods larger than 86 minutes, in other words to maximum orbital periods of infinity. Previous searches were sensitive to orbital periods longer than a couple of hours, so we are really extending this limit downward a lot! Our search is set up that we are sensitive to companions as massive as 1.6 times the mass of our Sun. This will allow us to detect pulsars in double neutron star systems, which are one of the most interesting. When both parts of BRP5 are done, we will have tested all possible orbital orientations and sizes in this mass range, so we should have a very clear idea of what is out there.
Correct. That's for one of the pulsars we found in the PMPS data (BRP3 run). That's actually not what we're targeting for, but we can detect it anyways. As I said we search for any orbit longer than 86 minutes, and 937 days is much longer than that lower bound :-)
The survey observed each sky position, and therefore each possible pulsar, only for 35 minutes. This is by far too short to detect any orbital decay. This requires dedicated observations over at least one year.
Einstein@Home Project
RE: ... which are one of
)
I'm still here! ;-) I must have understood.
So, why are those the most interesting?
Explanation for my question:
I have seen discussions where binary neutron star systems are apparently being looked-at as a source for a steadily recurring gravity wave. I say "apparently" because the usual explanation of why we look for pulsars is that we want to use them as nothing more than timing beacons so we can see a wave's effect in one plane.
As clocks, it seems on first blush that isolated pulsars would be better.
The follow-on question is, as massive and compact as neutron stars are, there's a theoretical limit to their mass (I think I recall reading that). If masses orbiting each other should "make waves," it "feels" as though we already know precisely what triangulation of pulsars we would need relative to a known cosmic event in order to directly detect the waves we have pretty-well known are there for about forty years.
I'm not a Physicist. I'm quite sure I don't understand the difficulties.
Is there a book I should read that would explain the problem that doesn't resort to equations? All of the Mathematics I once knew is lost at the end of some neural pathway long overgrown by the weeds of disuse. Any attempt to describe all of this as precisely and completely as the equations can would probably do more damage to my ego than it would improve my understanding.
As clocks in the sky, pulsars
)
As clocks in the sky, pulsars can be very useful. In binaries doubly so!
For instance : the pulsation of one star neutron may, if the system has suitable alignment, be eclipsed by the other star. During those occasions of eclipse the signal has to pass through the gravitational field of that second object. Hence that becomes an examination of that objects field, and is a test of whichever theory you might like to invoke for the effect of gravity upon light propagation.
Also stars in close pairs like these tend to muck about with each other's alignments causing precession and such. Pulsar activity by neutron stars serve as 'clocks' inside the system that aren't available in non-pulsing binaries. Again the detail of such is testing ground for gravitational theory.
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
Hi, RE: RE: ...
)
Hi,
Why do we love binary pulsars? They are one of the most stringent testbeds for Einstein's general theory of relativity (GR). They are very compact and massive objects, sometimes in very close orbits. That means that the acting gravitational forces are enormous, and that we have to use GR to describe these systems. Mike has already mentioned a couple of the interesting effects below:
Shapiro delay: the radio waves from the pulsar pass through the gravitational potential well of the companion. The depth of the well and thus the signal delay depends on the mass of the companion, which in turn can be computed from the observations.
relativistic periastron advance: GR predicts that the elliptic orbits we see celestial bodies move on are actually not closed. The ellipses rotate in space over time. This depends on the total mass (pulsar + companion), and thus the observation of this effect allows the measurement of the total mass. A similar effect is known from our solar system, but there the effect is very small, but was known from observations of Mercury's orbit since 1859.
orbital shrinking: GR also predicts, that the orbital motion should lead to emission of gravitational waves (GWs). The energy radiated away in GWs leads to a shrinking of the orbit. This effects was first observed in the first double neutron star system, the famous "Hulse-Taylor pulsar" PSR B1913+16. Hulse and Taylor received the Nobel prize in 1993 for their discovery, which was the first indirect observation of gravitational waves.
Astronomers can combine multiple of these tests together with other observations such that they have more measurements than unknowns. In that case they can use their observations to formulate a test of GR. So far, there's been no indication that GR might be wrong or incomplete, and the precision to which GR has been proven right is getting better and better. But as usual in science, there's always room for new discoveries.
What you are referring to here is several things at once. Let me try to separate them in my reply. Pulsars (binary or isolated) are great tools to test Einstein's general theory of relativity (see above). One current "hot topic" has to do with measuring gravitational waves emitted by the pulsars themselves, or their orbital motion. The other approach uses the clock-like behavior of pulsars, to detect gravitational waves emitted by other objects. The pulsars become gravitational-wave detectors in the latter case.
Isolated pulsar can emit gravitational waves. This is what we're looking for with the Einstein@Home gravitational-wave searches. If a pulsar has a tiny mountain, it would emit continuous gravitational at twice its spin frequency. So far, the Einstein@Home searches were blind, looking for these signals from any point in the sky and at any frequency. The new Einstein@Home search will target specific points in the sky at which we know neutron stars must exist, but where no radio pulsar has yet been found.
The direct detection of GWs from the orbital motion of compact pulsar systems cannot be achieved with the ground-based detectors in use today (LIGO, Virgo, GEO600). They will only be able to detect frequencies above ~20 Hz, because of the seismic motion of the Earth. If you want to detect gravitational waves in from the orbital motion, we're talking about frequencies in the milli-Hertz range. This would be possible by building a LISA-like gravitational wave detector, which just has been proposed to the European space agency under the name of "eLISA - The Gravitational Universe".
The use of pulsars as beacons to detect GWs from other sources is explored in pulsar timing arrays (PTAs). These observe an ensemble of dozens of pulsars of decades and try to detect tiny deviations in the pulse arrival times that cannot be explained by any known effect in GR or other astrophysics. What remains after subtracting all known effects should contain the signature of very-low frequency GWs. These are emitted for example by supermassive black hole binaries.
I'm afraid I don't know any good book, maybe others have some tips. But I do recommend you have a look at the Einstein Online webpages on gravitational waves. These webpages are a part of the larger Einstein Online webpages, which also feature some recommendations for books.
Cheers,
Benjamin
Einstein@Home Project
Mike and Benjamin, Thank
)
Mike and Benjamin,
Thank you for taking the time to explain.
I'll get right to what I suspect are my final questions:
You said:
Is there any reason for anyone to believe there can be any surface features on a neutron star, or are we just sort-of hoping?
And lastly, are we looking for neutron stars which are specifically not pulsars? (you used both terms)
Thank you, again.
(BTW - Mike, I smile every time I see Pascal's quote. It sounds more literary than you usually think of scientists' writings, and very insightful.)
RE: Is there any reason for
)
I think that's tantamount to asking what a neutron star might be made of ie. can the qualities of the substance ( equation of state ) permit bumps to sit higher than surrounds in the presence of both huge gravitational field ( pulling inwards ) together with the rotational speed ( throwing outwards )? Ben ?
Any and all we can lay our hands on I think. The idea is that all pulsars are neutron stars but not all neutron stars are pulsars. That implies that either/or some neutron stars don't pulse at all in any direction, or if they do then not in our direction. From memory there are plots of aggregated pulsar characteristics ( frequency vs frequency derivative ? ) that has a band across it ( Valley Of Death ? ) which is not populated, implying I guess alot about the mechanism that generates the pulses. All I can imagine is enormous lightning bolts arcing from one part of a neutron star's surface to another, but I really should let Ben answer this stuff .... it's his specialty! :-)
Cheers, Mike.
( I chuckled for a good half hour when I first found that quote. It nicely apologises for my fits of verbosity .... )
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
Hi, Mike already partially
)
Hi,
Mike already partially replied to your questions, but let me add my two cents as well :-)
Nobody really knows what a neutron looks like in detail, so there is quite a bit of uncertainty regarding the exact structure. We assume neutron stars contain a lot of neutrons, some remaining charged matter (electrons, protons, neutron-rich atomic nuclei), but we don't know. The problem with neutron star matter is that it is so completely different than anything we can create in laboratories on Earth.
The only thing we can do is model them mathematically and numerically simulate them and see what the maths and simulations tell us. And from those we know that neutron stars should have a crust (couple hundred meters thick) that is sort of cristalline. And that crystal could build up tensions which could lead to deformations of the neutron stars. There are other theoretical models as well predicting neutron star deformations, but so far we cannot tell, whether one (or any) of them is/are true.
Nonetheless, there's only way one way to find out: we have to look for the gravitational waves from these objects. Exactly that is what the Einstein@Home GW runs are doing.
The radio and the gamma-ray pulsar searches on Einstein@Home can only (by definition) find radio or gamma-ray pulsars, respectively. To be visible as a pulsar (radio or gamma-ray) the emission regions on the neutron stars have to point towards Earth. If they don't, the neutron star is there, but you won't see it as a pulsar, even though the neutron star is actively emitting. In other words: the pulsar might just not be pointing at us.
The radio and gamma-ray emission might not be happening at all, it could be screened at the neutron star by intervening plasma clouds, or the pulsar might've switched off. That happens when they get too old, this is the "pulsar graveyard" mentioned by Mike.
If any of these electromagnetically invisible neutron stars is in fact deformed, there is a good chance that one day we'll see it with the Einstein@Home GW searches. These searches in a sense nicely complement the other efforts to discover new neutron stars.
Cheers,
Benjamin
Einstein@Home Project