Pulsar re-detections in PALFA data
24 Jul 2009 2:00:07 UTC
Topic 194454
(moderation:
I see a thread hasn't started yet, so here it goes. First, I'm really excited to see the results posted that way, including the WU's of the best detections. The raw-raw plots are pretty indecipherable, and I've looked at pulsar search results before...the PRESTO summary is pretty nice and readable. I'd hate to think some poor grad student is poring over hundreds of those raw plots to get a sense of what is the significance of each report.
I'm sure the paper (preprint likely when?) will use these detections to help verify/calibrate the search sensitivity. The solid pickup of a 2.1 ms pulsar at that large a DM (almost 300) gives one hope. The ATNF pulsar cat lists the S1400 as 1.3 +- 0.4 mJy, which is pretty faint. It also shows it to be a binary, just long enough period (95 days) that PRESTO listed it as binary with a negative Pdot.
Good job searchers and crunchers!
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
Pulsar re-detections in PALFA data
)
Martin, with your personal history of this area of expertise, I'm sure we'd all be delighted if you could give us even a quick rundown on how to read any/all aspects of those results and what they mean in laymen's terms [ what is a topo? what is a DM? P-dots, F-dots, polka-dots ... ]
..... nudge, nudge ...... two thumbs up ........... cheesy grins .... frantically waving ..... begging ..... c'mon :-) :-) :-)
Cheers, Mike.
( edit ) And congratulations to all those contributors mentioned as having the digital flows pass through their hardware!! :-)
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
My congratulations for Bernd
)
My congratulations for Bernd Machenschalk!
Maybe some words from the
)
Maybe some words from the
who has done these plots...
Info on the raw detection plots: each point in each plot is one candidate event found by the computers attached to the project.
The position of the point encodes the following information about the candidate event:
- along the left-right (x) axis there's the DM channel
(how much electron gas is between us and the pulsar - can get the distance from that)
- along the front-back (y) axis is the spin frequency of the candidates
(how fast is the pulsar spinning?)
- the up-down (z) axis shows the significance of the candidates
(how likely is it that this is not just noise?)
- the color code shows the strength of Doppler modulation
(how tight is the binary orbit?)
As you can see from the raw detection plots on the re-discovery page each pulsar shows up like a number of more or less cuspy structures in these raw plots.
So looking for cuspy things is one way of finding the pulsars.
To me these plots are well readable but that might be because I made them ;-)
Obviously I generate other plots (like the one with PRESTO) to follow up on promising candidates.
Mike, for more information on the different search parameters have a look at our info pages. They should provide answers to some of your questions.
Cheers, Ben
Einstein@Home Project
Thanks very much Ben.
)
Thanks very much Ben. :-)
Cheers, Mike.
( edit ) Hey Wow Man! That Interactive Binary Pulsar Simulation is Ace!! :-) :-)
( edit ) It shows the Doppler stuff beautifully with progressively earlier arrival of the pulse when it is on the approach vs progressively earlier when receding. Compare with the arrival time when moving entirely across line of sight - the 'back' and 'front' part of the orbit.
( edit ) I am just mesmerized by the geometry of it .... :-)
( edit ) And if you're real naffy you can get a two pulses per rotation, one from each cone at opposite pulsar magnetic poles. Set the pulsar spin axis inclination at about zero, the pulsar magnetic axis inclination at about 90, hold the left mouse button down while over the display area and drag up and down to get the orbital planes near edge on. By transitioning across the 'ecliptic' of the system you can make one pulse more prominent and the other less so ...... now I want both to be pulsars! :-) :-)
( edit ) You zoom in/out by right mouse button drag by the way. It'll also make more sense if you enable 'orbital planes' and 'rotation axes' in the View menu.
( edit ) Hey Hewson! More Decaf ...... :-)
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: Martin, with your
)
I can't resist a "pretty please with sugar on top..."
Benjamin answered some of the questions, and I didn't mean to cast aspersions on his data visualization skills. I know from trying to plot multidimensional data myself that it's tricky and that means the pattern-matching wetware between our ears needs some training to easily discern a "good" event from a "ignore this" event.
As to the parameters in the PRESTO plot, "topo" is short for topocentric, which is jargon for "what we observe here on our telescope sitting on spinning, moving planet Earth." "bary" refers to barycentric, which is the inertial reference frame located at the center of mass of the solar system.
As Benjamin said, DM is Dispersion Measure. Dispersion is the smearing of the pulses across a radio band (higher frequencies arrive first), and it is caused by the diffuse plasma in the galaxy between the pulsar and us. It's units are cm^-3*parsec (which only an astrophysicist could love). Given that the average electron density in the galaxy is about 0.03 cm^-3, you can take the DM and multiply it by 30 (1/0.03) and get the rough distance in parsecs (or divide by 30 to get it in kiloparsecs). Note that means in 30 milliliters (an ounce for us Americans) of galactic space, you are likely to get *one* free electron.
P is period, Pdot is the rate of change of (spin) period. F is frequency, which is 1/P. Pulsar timing generally uses P, Pdot, etc. but in searching it's often convenient to work in the inverse (F) because that's how an FFT bins things.
Did I miss anything?
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
RE: I can't resist a
)
If I bounce up and down too much like some little yappy dog just whack me with a cushion. :-)
Nup, that's great! :-)
I had thought 'dispersion' was a pure statistical term/comment aka standard deviation or somesuch. Well, initially I thought DM was Doppler Modulation. But I now see it's physical as in 'speed of propagation dependence upon the frequency' as per properties of the medium. Thus derives/becomes a range comment.
'dot' as in time derivative notation I should have guessed. I'd learnt to use d/dt, d/dx etc ....
'topo' vs 'bary'. Ah well, E@H does that for gravitational signals too. I'll have to go away and have a think about (a) Why 'bary' is inertial and (b) Why inertial is a good thing.
[ .... and I also think it's great for the search program to give feedback to the contributors in this way. I can't wait for a similiar gravity wave detection listing. Roll on AdLIGO .... ]
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: 'topo' vs 'bary'. Ah
)
The reason we care about barycentric is that is what stays constant, and therefore is the physical quantity that goes into things like the precise timing model my advisor Joe Taylor used to time the spin and orbit of the first NS-NS binary pulsar to prove gravitational radiation carries away energy as derived from Einstein's general theory. If you didn't convert the pulse arrival times to barycentric values, you wouldn't be able to see the nice consistency that makes pulsars (and especially millisecond pulsars) such excellent clocks. Fortunately our model of the solar system (ephemeris) is good enough that the errors that crop into that conversion are in the fraction of a microsecond range. That's one of the error sources I studied in my thesis. In one case I experimented by tweaking Jupiter's mass within the stated uncertainty and found it did reduce my timing residuals somewhat.
The equations for conversion from topocentric to barycentric are sensitive to the location and distance of the pulsar, so one benefit of earth's motion is that we can very precisely locate the pulsar and try to examine what else is there (VLBI radio sources, supernova remnants, optical sources when the binary partner is not a neutron star, etc.). For a couple of the nearest and/or most precisely timed pulsars we can measure timing parallax (like angular parallax for stars) and use that to determine distance. That helps calibrate the DM-to-distance relation we usually rely on.
(Oh no...I'm rambling now...)
"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)
Hi Marty, RE: I'm
)
Hi Marty,
Thanks for starting this thread!
The truth is that right now it is a little bit too early in the game for us to start talking about publication plans. Although that wlll change fast if we find new systems and especially short-period binaries.
Our main focus now is to try and speed up the application. E@H is currently processing about 18 minutes of Arecibo data per day. We think that, via various tricks, improvements, optimizations, and GPU-enablings, we can increase this processing speed by almost an order of magnitude, to about 2 hours/day. This would enable us to catch up on the PALFA data set (which on average is about an hour of data per day). Hopefully we can get many of these improvements working this summer and fall.
Once we are making progress towards catching up with the full data set, we'll get back to publication plans, sensitivity calibration, etc. Fair enough?
Cheers,
Bruce
Director, Einstein@Home
Just a curiosity; how are
)
Just a curiosity; how are Arecibo data transmitted to Hannover (or Madison?)? I know SETI has problems sending hard disks by ordinary mail (once were tapes).
Tullio
RE: The reason we care
)
Ah yes. I see. Barycentre is by definition the point where nett gravitational force ( from solar system bodies ) will be zero. So it will be, to pretty good approximation, much like that traditional text book example of a spaceship drifting in space unaffected by external forces etc. A 'clean' reference frame to accumulate and compare results within. JPL tracks that avidly I think. I suppose one could assume a pulsar model/data and work back to trim the solar system one too. Down to some level anyway, with the more/better pulsar data. Timing parallax .... ooh, I'll have to have a sit and a think on that one also. Please ramble as you please, it is most informative. :-)
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