Hidden in plain sight: E@H discovery of an elusive gamma-ray pulsar

We are glad to announce the discovery of a new gamma-ray pulsar hidden in plain sight in data from the Fermi Gamma-ray Space Telescope.
The object has been discovered by the computers of Connor Barry of Lafayette, Colorado, USA and Rich Johnson of Hayward, California, USA -- Congratulations!
The scientific paper has been published in the Astrophysical Journal Letters. An open-access preprint version is found here.

To learn more about it, check out the press release (a German version is also available here).

Thanks to everyone participating in Einstein@Home and enabling such exciting discoveries!

Holger Pletsch, Project Scientist and Research Group Leader

Comments

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,587
Credit: 311,164,408
RAC: 78,936

Hidden in plain sight: E@H discovery of an elusive gamma-ray pul

Yo ! What can I say ? Does E@H just ROCK the cosmos or what !! :-) :-)

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

Wurgl (speak^Wcrunching for Special: Off-Topic)
Wurgl (speak^Wc...
Joined: 11 Feb 05
Posts: 321
Credit: 140,550,008
RAC: 0

Another article in german

AgentB
AgentB
Joined: 17 Mar 12
Posts: 915
Credit: 513,211,304
RAC: 0

RE: Thanks to everyone

Quote:

Thanks to everyone participating in Einstein@Home and enabling such exciting discoveries!

It is exciting to be a part. I find it quite amazing that just a handful of photons, having travelled for thousands of years across the Milky Way, hit a detector in a satellite over a 6 year period and generate a few KB of data hidden in many many GB of noise... and a couple of PCs with some smart software, worked out where and what it was that made them.

... The curious incident of the invisible pulsar.

Holmes?

- Yes Watson.

How do you find a pulsar you cannot see ?

- Elementary my dear Watson, find a pulse.

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,587
Credit: 311,164,408
RAC: 78,936

LOL ! :-) Cheers, Mike.

LOL ! :-)

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

tbret
tbret
Joined: 12 Mar 05
Posts: 2,115
Credit: 4,861,254,633
RAC: 42,503

Congratulations to the

Congratulations to the project and to Connor and Rich !

How cool is it to be involved in a project that doesn't just write papers about the project, but actually finds new things that might not have been discovered without it?

Starleen
Starleen
Joined: 21 Jul 15
Posts: 1
Credit: 52,280
RAC: 0

I've just received this

I've just received this message, congratulation! :)

electricity9013
electricity9013
Joined: 21 Jul 15
Posts: 1
Credit: 6,288,369
RAC: 537

It's my pleasure to process

It's my pleasure to process the data from the other side of the earth.Congratulations!

astro-marwil
astro-marwil
Joined: 28 May 05
Posts: 531
Credit: 632,316,543
RAC: 1,106,785

Hallo! It´s nice to see,

Hallo!
It´s nice to see, that our prolonged effort in E@H does show some nice results.
Congaratulations to the dicoverers.
I´m missing in the paper on arxiv.org a sentence or two about, what theory about the generation of such high energetic only beams of radiation is favored or disfavored. But it might bee, that I didn´t find this, as I didn´t read the whole paper sentence by sentence over the whole length.

Kind reagards and happy crunching
Martin

Yavanius
Yavanius
Joined: 1 Jan 15
Posts: 17
Credit: 5,277,870
RAC: 440

This popped up on my Goggle

This popped up on my Goggle News under my 'distributed computing' category:

http://phys.org/news/2015-08-scientists-elusive-gamma-ray-pulsar.html

Scientists discover elusive gamma-ray pulsar with distributed computing project
August 7, 2015


Scientists discover elusive gamma-ray pulsar with distributed computing project
Fermi-LAT sky map with the celestial neighborhood of the newly discovered pulsar PSR J1906+0722 featuring several other gamma-ray pulsars (not labeled). The color scale shows the gamma-ray intensity. The dashed square at the centre encloses …more

Gamma-ray pulsars are remnants of explosions that end the lives of massive stars. They are highly-magnetized and rapidly rotating compact neutron stars. Like a cosmic lighthouse they emit gamma-ray photons in a characteristic pattern that repeats with every rotation. However, since only very few gamma-ray photons are detected, finding this hidden rhythm in the arrival times of the photons is computationally challenging. Now, an international team led by researchers at the Max Planck Institute for Gravitational Physics (Albert Einstein Institute, AEI) in Hannover, Germany, has discovered a new gamma-ray pulsar hidden in plain sight in data from the Fermi Gamma-ray Space Telescope. The improved, adaptive data analysis methods and the computing power from the distributed volunteer computing project Einstein@Home were key to their success.

Searching for an elusive gamma-ray pulsar

The gamma-ray source in which the discovery was made was thought to be a pulsar from the year 2012 on, based on the energy distribution of the gamma-ray photons observed by the Large Area Telescope (LAT) on board the Fermi Gamma-ray Space Telescope. For years, it was one of the brightest Fermi-LAT catalogue sources without a known association. "Everyone believed that the source now known as PSR J1906+0722 was a pulsar. The tricky part was to show that the gamma-ray photons carry the imprint of the pulsar's rotation and arrive according to that hidden rhythm. Many had tried that before, but to no avail," says Holger Pletsch, leader of an independent research group at the AEI and co-author of the paper that now appeared in The Astrophysical Journal Letters...

Read more at: http://phys.org/news/2015-08-scientists-elusive-gamma-ray-pulsar.html#jCp[img][/img]

Colin Clark
Colin Clark
Joined: 5 Apr 11
Posts: 2
Credit: 1,316,043
RAC: 0

Hello everyone! Thank you

Hello everyone!

Thank you so much for all of your messages of congratulations! It's really great to hear your interest in our project, and we're all so grateful for your volunteering efforts!

Martin, the paper had to be written to quite a tight word limit, so unfortunately we had to leave out a few things. In answer to your question, the most likely explanation is not that radio-quiet pulsars do not emit at radio wavelengths, but simply that we cannot see the radio emission from them. For example, due to the angle at which we view the pulsar, the radio beam might never cross our line-of-sight. Gamma rays are thought to be emitted from higher above the pulsar's surface, leading to a much wider beam than the radio emission, making it likely that in some cases we can only see high energy emission.

Also, while gamma rays can travel through the interstellar medium almost unimpeded, radio emission suffers from dispersion and absorption as it travels from the pulsar to us. In some cases this can make it very difficult to detect radio pulsations even when the gamma-ray pulsations are bright.

Hope this helps!

Cheers,
Colin

trekmbikes
trekmbikes
Joined: 3 Jan 13
Posts: 1
Credit: 109,140,156
RAC: 0

Super excited to contribute

Super excited to contribute to this! It's fun hijacking my friends' and family members' computer idle time for maniacal purposes.

Credit fully belongs to the scientists who built algorithms which can find patterns in these tiny glitches among (I'm sure) lots of noise, and to those geeks who optimized them to run on modern instruction sets, GPUs, etc. What an awesome and tangible use of the Internet.

-Connor

Michael Hoffmann
Michael Hoffmann
Joined: 31 Oct 10
Posts: 32
Credit: 31,031,260
RAC: 0

I'm more than happy to

I'm more than happy to contribute to the projects, but I got a question concerning FGRP: Isn't there a better approach than this crappy SSE-2 algorithm? It's so awkwardly slow and I'm sure with a different approach you'd have more throughput.

Is there a scientific or technical reason for this? I'd be happy to broaden my knowledge.

Om mani padme hum.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,522
Credit: 709,710,213
RAC: 797,472

RE: I'm more than happy to

Quote:

I'm more than happy to contribute to the projects, but I got a question concerning FGRP: Isn't there a better approach than this crappy SSE-2 algorithm? It's so awkwardly slow and I'm sure with a different approach you'd have more throughput.

Is there a scientific or technical reason for this? I'd be happy to broaden my knowledge.

What makes you think that the current implementation is 'awkwardly slow' or 'crappy'?

We are looking into a GPU version, but at the moment, it's hard to beat the Einstein@Home Binary Radio Pulsar (BRP) search implementation in terms of efficiency, so as we can spend the cycles on the volunteers' GPUs only once, running BRP on the GPUs is currently making the best use of the GPU resources.

Then there would be the possibility of using AVX. Because it will speed up only parts of the code and the speedup in that parts are at most 2x (sometimes much less, see below), the overall gain will be < 2x .

If you are specifically looking at the performance of your AMD Bulldozer generation CPU to others, the performance can in fact be disappointing, but this is just demonstrating that AMD's Bulldozer architecture isn't very strong wrt. floating point performance.

See this article for some benchmarks and explanations http://www.extremetech.com/computing/100583-analyzing-bulldozers-scaling-single-thread-performance which also discusses the AVX implementation of AMD Bulldozer. These chips, like some others, are actually splitting the 256 bit wide AVX vector operations into two 128 bit wide ops internally, so while technically yes they support AVX, they are not making the best use of these instructions. So I'm afraid you would not see that much of a speed-up from AVX on your particular hardware.

HBE

Michael Hoffmann
Michael Hoffmann
Joined: 31 Oct 10
Posts: 32
Credit: 31,031,260
RAC: 0

Thank you for the clarifying

Thank you for the clarifying words. I did not mean to offend. With "algorithm" I did not mean your scientific algorithms but the CPU instruction set. Sorry for that.

With crappy I basically referred to the SSE2 instruction set, which in my opinion was not so efficient. On the other hand it can handle floating point operations with better accuracy and therefore I was wondering if that was maybe needed for FGRP.

Your remark on AMD bulldozer architecture really was a help to better understand the whole thing. So it's basically the CPU, not the instruction set that's causing the issues on my end.

Again my knowledge and understanding increased :)

Om mani padme hum.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,522
Credit: 709,710,213
RAC: 797,472

You're welcome!

You're welcome!

Quote:
With crappy I basically referred to the SSE2 instruction set, which in my opinion was not so efficient. On the other hand it can handle floating point operations with better accuracy and therefore I was wondering if that was maybe needed for FGRP.

Not really. SSE2 can't handle floating point operations with better accuracy than AVX. The one thing from SSE2 instructions that AVX didn't include initially was integer vector arithmetic (came in AVX2), but we don't use that in FGRP.

HB

adrianxw
adrianxw
Joined: 21 Feb 05
Posts: 242
Credit: 322,654,862
RAC: 0

Thanks Colin, that was my

Thanks Colin, that was my question also, answered before I could ask it.

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.