Will I (Re-)Discover a Pulsar?
18 Nov 2010 18:02:55 UTC
Topic 195460
(moderation:
Many members/participants/users/volunteers are asking themself and in the forums:
* Do I have a chance with my small PC against these really big crunchers?
* How high is my chance to take a discovery or rediscovery of a pulsar?
* Is this chance really statistically distributed?
With this contribution I try to answer these questions by analyzing the list of ABPS-Rediscoveries published 11th July 2010.
If you´re interested to read more about this, you will find more here.
I´m very much interrested to get your comments.
Kind regards
Martin
Will I (Re-)Discover a Pulsar?
)
Hi astro-marwil,
wow, how much time did you invest in writing your article? This one is very thorough.
After only reading your post here I thought: The quotient of cobblestones and rediscoveries should be equal. Meaning if a participant with (example!) 1M cobblestones has 10 (re-)discoveries, then a participant with 100K c.'s should have 1 (re-)discovery.
Obviously: http://einsteinathome.org/account/59911
You mean, is it distributed like one of these?
There is a test named Kolmogorov–Smirnov test to check this. (Do not ask how to do it, it is long time ago I had to use it.)
Deduce it from your data?
Just thoughts...
Michael
Team Linux Users Everywhere
RE: Many
)
I think the answer is yes you have a chance and yes it is a real chance too! The only difference between your pc and the guys with a lot of pc's, like me, is that we go thru more units than you do while looking. We go thru the same group of units just at a faster rate. So statistically we should find them faster than you do, but only because we throw more resources at the project. BUT it also means we may be going thru the chaff faster too, leaving the units with the discoveries to you, there is no way to know that due to the pure randomness of the work unit distribution. In the end if you don't crunch you WON'T find anything, and I personally feel the same way. We both have the same chance that the unit with the next discovery, or re-discovery, is in the next unit we crunch! It is kind of like that pretty girl in school, since we never asked her out she never went out with us, but if we had asked her she might have! An old saying I have modified that I like goes 'If you never ask you will never get a yes answer!' I use it in Boinc to say 'if you don't crunch you won't find anything!'
RE: Many
)
Wow! Impressive piece of work! Well done indeed :-)
I'll look over it more thoroughly when time permits, but at a glance it shows that on a per-RAC basis we all have an equal chance of being hit by lightning! [ By this I mean a Poisson distribution well models low probability random events like lightning strikes, lotto wins, rural car accidents .... ]. Put in other ways :
- RAC is a good indicator of activity.
- there's no 'rigging' going on.
- the BOINC scheduling algorithms don't discriminate on RAC ie. if you're at the teller window you'll get served [ well, if the teller is there etc :-) ].
- you've got to 'be in it to win it' ie. buy the lotto ticket, stand in an open field in a thunderstorm, ask the pretty girl ...
Your second question ( How high is my chance to take a discovery or rediscovery of a pulsar? ) I think is the hardest one. We're now doing anti-center searching so the low hanging fruit could well be gone. Having said that, the Colvin's/Gebhardt effort found an unusual pulsar variant, so maybe a different ( from historical 'standard' ) pulsar population will be trawled by enabling more complex searches that DC/BOINC has made available to the data stream. Not forgetting that in the practical sense a re-discovery is easier to nominate as such, simply because we already have the template!
You've pointed out some 'normalisation' problems ( ie. related to the denominator of a probability quotient ) but I don't think that greatly affects your main conclusions, just introduces error into the absolute value of the chances.
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
Greetings everyone, I just
)
Greetings everyone,
I just happened to find this rather new string, read it, and have begun wondering, how a person would be notified IF their PC's "number crunching" had made or contributed to a discovery or rediscovery.
Would the notification be made by e-mail, PM, or in another fashion?
Sorry for straying a slight distance off topic and thanks in advance for an answer to my poser. :-)
To get back to the matter being discussed, It is my opinion that Mike Hewson's partial answer of, "You've got to be in it, to win it.", fairly sums up how most "whatever@home" volunteers view their chances and expectations of "hitting the jack pot".
Further, like myself, I suspect that, many of the volunteers contribute their spare available resource(s), mainly due to an interest in a related field of study or the subject matter alone, while giving them a sense of contribution to the greater whole of the research being done, sans actual expectation of "winning the big one", as it were.
In my humble opinion, the simple act of allowing one's oun personal computer(s) to be utilized just to crunch some research numbers, makes us ALL discoverers and therefore, winners in our oun right. ;-)
So. . . . Good luck to us all! :-)
Best regards,
Wishing you continuing scientific "fun" and, at least, always valid results,
Laters,
Rick "WHOSIT" W.
Hallo Michael, Mikey, Mike
)
Hallo Michael, Mikey, Mike and Whosit !
Thank you for your responses.
Here are some answers on your questions and statements.
Not only to Michael :
First I meant, are the chances really equally distributed, so that everyone’s chance to take a rediscovery only depends on how much he is crunching at its speed. And secondly, can this become proved by fitting the available data to a well-known statistical law? A much other question is, what the fitting parameters mean for us. What can we learn from these? For example: Probably the Kumaraswamy distribution will fit likely the data in figure 2, as both have a long tail at the lower than at the upper side. But what can we learn from the fitting parameters? And how do you fit fig. 6 and 7, with its saddle like curve and a very sharp fall off at the high end?
Thank you for this hint. I did not check for it before, but will see, what I can learn from it. As said in the first part of the appendix of my study , I have principle doubts that any of these “standard†statistical laws are allowed to apply to this problem. It is absolutely fundamental to every BOINC-project that every participant can determine at every time whether he take part or not at every project and how high its activity is, without any notice to any project at any time. One can check twice a day the activity of every participant, but there is no data pool, to check for it in the past. And that is good so. But so it is impossible to correct the statistics for this varying activity, which would be very elaborated but necessary to get equal preconditions for every participant and the application of any of these “standard†statistical laws. So my doubts are of fundamental, principal nature. From figure and table A1 we can learn, that about 5/6 of the participants make use of this freedom, and lean 2/3 have increased their activity. For this has to be paid attention.
Principally, yes. But be careful. This figure was derived for this relative narrow time window of nearly 6 weeks and the density of stars etc. at the area of sky analyzed at that time. For example, we are now analyzing the anti-center direction, where the density of stars is much less and in the mean they are younger than in the center direction at that time. So the probability to get a (re-)discovery now has reduced drastically. Figure 8 show that your chance to take a (re-)discovery is proportional to your activity and there are no indications for privileging or discriminating some-one.
Not only for Mikey:
You describe very nicely what most of us think and why we are doing this. But did you prove for it that the system behave so? That was my approach and I found, yes you have a real chance and you can quantify this. See figure 8.
Not only for Mike:
Your sentence let me think about this again and I found, it´s not so easy and clear. This is because RAC become built up from two major components, ABPS and S5GC1, and the relative amount of ABPS-files changes over time. But we have just luck. The relative amount of ABPS-files is acceptable constant within our interval of interest. So I completed the appendix of my small study by this aspect.
Not only for Whosit :
That notification isn´t so difficult. Each valid result became inspected by a scientist, whether there is valuable information included. If he finds after careful checks a (re-)discovery, he knows by the workunit-name the ID-numbers of the involved participants. You can test this. If you have running the advanced view of the BOINC-manager, open the tap “jobs†and click on one of them. You will see in the left column a field “webpagesâ€. Click on the 5th button from the bottom named “your resultsâ€. The browser will open and show you a list of your jobs. If you click in the 2nd column from the left named “Workunit ID†into a field, a new page opens, showing you the state and other parameters of this workunit.
We wish you further much fun and always valid results at minimum.
I will be pleased to get your further comments.
Kind regards
Martin
quote
)
quote "astro-marwil"
Greetings all,
Hello and thank you astro-marwil, for your rapid answer in reply to my posting. :-)
However, after following your instructions (and getting to the same place by other means as well), I found only, that the work unit was: awaiting validation, validated, data error, etc., etc..
Therefore, my question stands, as to how one would be notified? Would it say there, "New Discovery" or "Re-discovery", in the box before or after "Validated"?
This is not important at all and I'm having fun with this and "MilkyWay@home" sans the information. I'm just curious about it. I suspect an e-mail would be sent, as they are all viewable by the "upper" administration. Not sure however. ;-)
BTW, I am considering pilfering this, changing it a little, incorporating it in my sig, and using it to close my posts from now on:
"We wish you further much fun and always valid results at minimum."
I believe it to be a very nice sentiment indeed astro-marwil. :-)
So. . . . . . .
Wishing you continuing scientific fun and always, at least, valid results,
Wishing you continuing scientific "fun" and, at least, always valid results,
Laters,
Rick "WHOSIT" W.
RE: quote
)
They would contact you directly BEFORE putting anything on the website! First they would try email, then they would try thru your Team then they would look on the internet to see if your were 'findable'. IOW they would put some effort into it but...keep you email address current!!!!
Hello Whosit
)
Hello Whosit !
Sorry, I misunderstood you. All (re-)discoveries become listed here. You can search therein for your User-ID. But this list is somewhat old - 3 month -. Bernd Machenschalk promissed, that it will become updated relative soon. And Bruce Allen promissed, to inform every "discoverer" at minimum by email. But since then, there was no (re-)discovery published, as far as I know.
I wish you further much fun and always valid results at minimum.
Kind regards
Martin
Greetings
)
Greetings all,
astro-marwil,
Thank you for the VERY fast reply. You have answered my question to the fullest extent possible.
I can now sleep at night again. :-)
Wishing you continuing scientific "fun" and, at least, always valid results,
Laters,
Rick "WHOSIT" W.
RE: Hello Whosit
)
With the re-discovery rate demonstrated on that page throughout the summer, it's remarkable there have been none since the end of August 2010.
Has cruncher participation dropped off that much?
Or did a crontab get waylaid by PAM or something?
Anyway, cheers to a good new year. :)