Searching for pulsars in PALFA data from Arecibo

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284700169
RAC: 113726

RE: Picture Jocelyn Bell,

Message 88995 in response to message 88994

Quote:
Picture Jocelyn Bell, or soon after the Arecibo operators, poring over a strip chart analog recording looking for periodic bumps in the trace. Now we've got thousands of computers doing it.


Yes! How many meters per second would the paper chart have to run at to be able to get a decent human eyeball resolution of the fastest known pulsar, say ~ 700+ Hz ?

If you want a pulse gap of 1cm, say, then that's seven meters per second. If you wanted the pulse width to be 1cm, so as to discern features ( assume width is 0.1 of the period ) then that's 70 meters/sec **.

Imagine trying to keep such a record coherent over years to get spindowns and what not. A bit like the WW2 code crackers before the 'bombes'. You couldn't even conceive of a project like LIGO without digital computers. It would have to be realtime analog computing at best, or maybe some hope with magnetic tape storage.

Cheers, Mike.

( edit ) I think you can do analog FFT circuits .....

( edit ) ** Hey, that's ~ 250 km/hour !! :-)

( edit ) OK. Getting silly now. Take the Hulse-Taylor pulsar, found in 1974. About 17 Hz. Follow to present day ~ 35 years. Total length of paper tape at 1cm per pulse width = 35 * 365 * 24 * 3600 * 17 * 10 = 187639200 km = 187639200/149598000 ~ one and one quarter of the Earth-Sun distance. About 19 billion rotations of the star. Try not dropping a count/carry during that! :-)

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

Martin Ryba
Martin Ryba
Joined: 9 Apr 09
Posts: 48
Credit: 152722276
RAC: 35422

RE: ( edit ) OK. Getting

Message 88996 in response to message 88995

Quote:
( edit ) OK. Getting silly now. Take the Hulse-Taylor pulsar, found in 1974. About 17 Hz. Follow to present day ~ 35 years. Total length of paper tape at 1cm per pulse width = 35 * 365 * 24 * 3600 * 17 * 10 = 187639200 km = 187639200/149598000 ~ one and one quarter of the Earth-Sun distance. About 19 billion rotations of the star. Try not dropping a count/carry during that! :-)

Less silly, dusting off my thesis, I stated that the observations of PSR B1937+21 (J1939+21) that we had collected from mid-1984 to the time of my thesis in early 1991 spanned exactly 130,476,403,197 rotations of the pulsar. Given that number of significant digits, the TEMPO timing software has to take care to avoid rounding errors from even double precision floating point calculations, since the TOA (pulse phase) measurements are good to better than 1/1000 of a period. There have now been some more significant gaps in the collection of timing data on that pulsar, but I would think we should still be capable of resolving the integer rotation number to present. If so, data spanning 1984 to now (25 years * 3e7 seconds/year * 642 rev/sec) = 506 billion revolutions.

B1937+21's signal was too weak to see on a strip chart (or if you used enough bandwidth the signal was smeared by dispersion). I do remember the excitement of hooking analog output from my coherent dedisperser circuit to an analog signal averager at Arecibo and watching the pulses come out of the noise in a few seconds. Occasional giant pulses would pop above the noise floor though, especially down at 430 MHz. I remember having a spectrum analyzer listening to the live signal from another nearby pulsar (0950?) with a roughly quarter second period and taking the "video out" directly to a speaker: pop, pop, pop, bang!, pop, pop, bang!, pop, pop,...

Too bad the LIGO gang can't have that kind of fun with their instrument. The Arecibo control room is a fun place to hang out at oh-dark-thirty in the morning.

"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)

Bill592
Bill592
Joined: 25 Feb 05
Posts: 786
Credit: 70825065
RAC: 0

RE: B1937+21's signal was

Message 88997 in response to message 88996

Quote:

B1937+21's signal was too weak to see on a strip chart

You would have had fun trying to crunch all of that data on the

486 DX 2’s that were ‘state of the art’ in 1991 :-)

Bill

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: ...If so, data spanning

Message 88998 in response to message 88996

Quote:
...If so, data spanning 1984 to now (25 years * 3e7 seconds/year * 642 rev/sec) = 506 billion revolutions...


Which kind of billion? (I'm just too lazy to count the zeros myself ;-)

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284700169
RAC: 113726

RE: RE: ...If so, data

Message 88999 in response to message 88998

Quote:
Quote:
...If so, data spanning 1984 to now (25 years * 3e7 seconds/year * 642 rev/sec) = 506 billion revolutions...

Which kind of billion? (I'm just too lazy to count the zeros myself ;-)


That'd be the 1000 million type of billion = 10^9

I think the original British billion was a million of millions = 10^12 which we now call a trillion. I think the British did call the 10^9 type of billion a 'milliard' ???

Quote:
Too bad the LIGO gang can't have that kind of fun with their instrument. The Arecibo control room is a fun place to hang out at oh-dark-thirty in the morning.


Actually they do, a few years ago I saw a Hanford ilog entry by Peter Saulson and on a night shift too. He'd hooked up an audio amp/stereo to something derived from the differential arm error signal ( effectively the interferometer dark port output ) and described for us all sorts of creaks, groans and stuff. No gravity waves per se but the movement of the instrument was made audible. Now I wonder if we can con them to stream it live ..... :-)

Cheers, Mike.

( edit ) Livingston iLog actually

Quote:

Topic: general Author: Peter Saulson
Sun Jan 4 16:35:59 2004 UTC
Listening to the ifo output on the loudspeaker in the control room is a lot of fun. The sound reminds me of a dog that is dreaming it is chasing something, breathing dramatically and irregularly as it "runs". It is a high pitched wheezing sound, broad-band high frequency noise modulated on time scales of order a second (just as one sees in graphs of the AS_Q time series, no surprise.)

What is obvious as one does this is that the loud moments correspond to times when the output port is momentarily bright. The AS_DC-by-eye and the sound also have a flutter (is it 10/sec?) What causes that?

What is completely non-obvious is why the histogram of AS_Q should grow broad wings when the AS_DC excursions grow large. (Why shouldn't the width of the histogram just be modulated?) The correlation of glitches with times when AS_DC was large was the pattern that we saw in S2 burst analysis; evidently, we'll find the same pattern in S3. It would be nice to understand (and even better to fix) the mechanism.
- Peter Saulson


Quote:

Topic: SciMon Author: Peter Saulson
Mon Jan 2 22:05:06 2006 UTC
Day Shift SciMon summary
Operators: Harry Overmeir until noon, then Danny Sellers
SciMon: Peter Saulson

A shift during which it was a struggle to get locked and to stay in lock. The cause was high seismic noise in the band below the microseism, caused by winds of up to 15 mph, along with surprisingly high noise in the anthropogenic bands (given that there was no construction work going on today.) The result was only a few sub-hour science mode segments. While in lock, one could see the beam spots bouncing around, and a fair amount of non-stationary noise in the sub-100 Hz band. A few times, there were audible "thumps" in the loudspeaker playing the ifo output; those corresponded with rather gigantic amounts of excess noise, and to minutes where SenseMon showed much lower range. (Two of these occurred in the last several minutes of the science segment that included 1400 Central = 2000 UTC.)


Quote:

Topic: SciMon Author: Peter Saulson
Sun Feb 6 06:00:40 2005 UTC
Evening Shift Summary
SciMons: Peter Saulson (expert), Andy Rodriguez (trainee)
Operators: Tom Evans until 10 p.m., Gary Traylor thereafter

A quiet shift of the finest kind. The ifo has been in lock for over 10 hours by now, with low noise and otherwise good behavior, too.
......
Tom's Operations Log entry mentions loud thumps that we heard over the loudspeaker near 2304 and 2305 UT. It would be nice to find them with some of the Burst Search Event Trigger Generators. Gary's log will mention another at 0459 UT. Let's find this one, too!
......

- Peter Saulson

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

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

RE: I think the British

Message 89000 in response to message 88999

Quote:


I think the British did call the 10^9 type of billion a 'milliard' ???

Cheers, Mike.


We call it a miliardo, though many call it a bilione, by the usual Italian habit of mistranslating English words.
Tullio

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284700169
RAC: 113726

RE: Less silly, dusting off

Message 89001 in response to message 88996

Quote:
Less silly, dusting off my thesis, I stated that the observations of PSR B1937+21 (J1939+21) that we had collected from mid-1984 to the time of my thesis in early 1991 spanned exactly 130,476,403,197 rotations of the pulsar. Given that number of significant digits, the TEMPO timing software has to take care to avoid rounding errors from even double precision floating point calculations, since the TOA (pulse phase) measurements are good to better than 1/1000 of a period. There have now been some more significant gaps in the collection of timing data on that pulsar, but I would think we should still be capable of resolving the integer rotation number to present. If so, data spanning 1984 to now (25 years * 3e7 seconds/year * 642 rev/sec) = 506 billion revolutions.


Wow, good job! My silliness retracted post-haste. That's a bit like asking if Avagadro's number is a prime! Or if you've counted all the rocks in Wales ..... :-)

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

Ver Greeneyes
Ver Greeneyes
Joined: 26 Mar 09
Posts: 140
Credit: 9562235
RAC: 0

RE: I think the original

Message 89002 in response to message 88999

Quote:
I think the original British billion was a million of millions = 10^12 which we now call a trillion. I think the British did call the 10^9 type of billion a 'milliard' ???


Correct, however the word has fallen in disuse. I think they just say 'a thousand million/billion/trillion' instead nowadays, and a billion is still 10^12. The words are still in use in Dutch, however. The discrepancy between the British and the American definitions is very aggravating, because you can never trust anything you read unless someone states up front what they mean.

Martin Ryba
Martin Ryba
Joined: 9 Apr 09
Posts: 48
Credit: 152722276
RAC: 35422

RE: RE: B1937+21's

Message 89003 in response to message 88997

Quote:
Quote:

B1937+21's signal was too weak to see on a strip chart

You would have had fun trying to crunch all of that data on the

486 DX 2’s that were ‘state of the art’ in 1991 :-)

Bill

Ah, now I'm getting into nostalgia. Our "Mark III" system had a couple custom-built ISA cards (full length *wire wrap* boards!) with various MSI and LSI logic chips on them (e.g., separate multipliers and adders). The PC that did the crunching and real time control was a 486 DX2, running DOS with some high memory support. With the FORTRAN77 compiler we had (Lahey), we did have to keep our executable under 600 kB including all arrays (we put COMMAND.COM in high memory). Remember Bill Gates' classic comment that 640 kB should be all anyone would need?

One time, when one of us (not me!) was trying to debug a board, he stuck a voltmeter probe into the ISA connector on the the mother board to make sure the 12V supply was good. Unfortunately, he shorted that 12V over to an address line on the other side of the socket. Pop! went one of the glue chips on the motherboard. There was a crater in the middle where the silicon used to be! So much for that PC!

We did our pulsar survey processing on the "big iron" at a couple of the NSCA supercomputers.

"Better is the enemy of the good." - Voltaire (should be memorized by every requirements lead)

hoarfrost
hoarfrost
Joined: 9 Feb 05
Posts: 207
Credit: 92347753
RAC: 107683

Hello! On page

Hello!

On page Einstein@Home Arecibo Binary Radio Pulsar Search (Re-)Detections we can see information about known binary pulsars. But whether there is data about unknown objects, for which check is necessary?

Thank you!

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.