mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Five or Bust - The Dual Sierpinski Problem

Reply
 
Thread Tools
Old 2009-11-24, 01:48   #12
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Quote:
Originally Posted by henryzz View Post
i am not sure we know enough about prime numbers to work out accurate predictions
on this project the predictions keep on being way too high for instance
First of all, huge congratulations to engracio on an enormous PRP!

Secondly, I was thinking the same thing as henryzz, that possibly there's something we don't yet know that's causing our predictions to be way too high. I wonder if there's something specific to the Dual Sierpinski conjecture that makes it more "prime-dense" than others?
mdettweiler is offline   Reply With Quote
Old 2009-11-24, 01:57   #13
engracio
 
engracio's Avatar
 
May 2007

112 Posts
Default

Quote:
Originally Posted by philmoore View Post
I don't have any problem notifying everyone right away, but I would probably prefer to do it via email or PM than post it on the Forum. I wrote Jeff just a couple of hours after I received your email, telling him to pull all the 41693 numbers from his 5.15-5.16M reservation until we knew for sure, but I also asked him if he would be willing to finish the 5.09-5.10M reservation so that we knew whether your prp was truly the smallest. But if he hadn't wanted to do that, I was willing to finish that set myself. I wrote Ben the day afterwards, telling him the situation and asking if he was willing to finish the 41693 numbers in his 4.97-5.00M reservation, and I also let Kent, who was sieving, know fairly early on. The reason I wanted to finish the smaller numbers is to complete the Sloane sequence A067760 up to (78557-1)/2:
http://www.research.att.com/~njas/sequences/index.html?q=A067760&language=english
But if anyone wants to jettison the rest of their tests next time, no problem, I am willing to finish them. The work stoppage may be more than a day or two if the next prp is quite a bit larger, but the prospect of a false pseudo-prime is pretty remote.

Of course, this assumes that the project stays small. If it grows a great deal, it would probably make more sense to do as you suggest.
Thanks Phil that makes sense to me, just to make sure that we have all the blocks checked. I did not realize that was your reasoning. I believe I completed all other wu prior to the 4th PRP, if not it would be just a couple.

e
engracio is offline   Reply With Quote
Old 2009-11-24, 02:04   #14
engracio
 
engracio's Avatar
 
May 2007

11110012 Posts
Default

Quote:
Originally Posted by Jeff Gilchrist View Post
Congrats everyone, great work!

Now I'm starting to wonder if paleseptember = engracio as a second account because how is it possible for anyone besides him to find a PRP in this project?!?!?!

Just kidding of course, looks like I just missed that reservation. Maybe next time.
I hear you man. Unless I am mistaken he is down there and we is up here, so no we be not cousins.

On the other hand, to up our chances for the last sequence, maybe we need for mister prime of FOB to have him help us with the last one. Last one is the charm to paraphrase a common saying, yea right....
engracio is offline   Reply With Quote
Old 2009-11-24, 03:31   #15
philmoore
 
philmoore's Avatar
 
"Phil"
Sep 2002
Tracktown, U.S.A.

21358 Posts
Default

Mini-Geek calls it the butterfly effect: because he took a single work file 4.80M, that caused this latest prime to be discovered by Engracio rather than Jeff. Jeff, you have my permission to find the next one, but let me say in advance, I will be pleased no matter who finds it!

These primes certainly seem to be consistently showing up early - I just hope this trend continues a little longer. Maybe mdettweiler and henryzz are right, that there is something we don't know that is making the problem a little easier than predicted.
philmoore is offline   Reply With Quote
Old 2009-11-24, 04:13   #16
engracio
 
engracio's Avatar
 
May 2007

112 Posts
Default

Quote:
Originally Posted by philmoore View Post

These primes certainly seem to be consistently showing up early - I just hope this trend continues a little longer. Maybe mdettweiler and henryzz are right, that there is something we don't know that is making the problem a little easier than predicted.
Sounds like someone with a little bit of time and lots of know how, they can bite into and be busy for a while.
engracio is offline   Reply With Quote
Old 2009-11-25, 02:51   #17
geoff
 
geoff's Avatar
 
Mar 2003
New Zealand

13×89 Posts
Default

Congratulations engracio! What a breakthrough for this project.
geoff is offline   Reply With Quote
Old 2009-11-25, 03:13   #18
Kman1293
 
Feb 2008

138 Posts
Default

Congrats engracio!
Kman1293 is offline   Reply With Quote
Old 2009-11-27, 12:13   #19
Zuzu
 
Sep 2009

1110 Posts
Default

Congrats to Engracio and all the group, we're lucky indeed !!!
Regarding the stats for the last prime: The 63M figure for 50% prob looks correct, but I don't understand why "We shouldn't expect 1 prime until 191.63M" according to Mini-Geek...
BTW much congrats and luck, I didn't expect a PRP so early, the 50% prob was around 17M indeed...
(sorry for being spartan in the form but I'm abroad, having not much time...)

Last fiddled with by Zuzu on 2009-11-27 at 12:19 Reason: Pbs with emoticons
Zuzu is offline   Reply With Quote
Old 2009-11-27, 12:46   #20
Mini-Geek
Account Deleted
 
Mini-Geek's Avatar
 
"Tim Sorbera"
Aug 2006
San Antonio, TX USA

102538 Posts
Default

Quote:
Originally Posted by Zuzu View Post
Congrats to Engracio and all the group, we're lucky indeed !!!
Regarding the stats for the last prime: The 63M figure for 50% prob looks correct, but I don't understand why "We shouldn't expect 1 prime until 191.63M" according to Mini-Geek...
BTW much congrats and luck, I didn't expect a PRP so early, the 50% prob was around 17M indeed...
(sorry for being spartan in the form but I'm abroad, having not much time...)
See the equation I had for calculating a 50% chance:
Quote:
Originally Posted by Mini-Geek View Post
63.17M=5.15M*2^(X/.19166) with X such that e^(-X)=0.5
e^(-x)=y is an equation meaning that when you expect x random occurrences, the chances of it not happening are y. We're looking for the chances of it happening (i.e. finding a prime), so we really want 1-e^(-x). (note that I left that out up there because for y=0.5, a 50% chance, they're the same) For the one I showed there, I wanted to know the expected number of primes before we had a 50% chance of finding one. It's about 0.693. The 5.15M*2^(X/.19166) part lets me smooth out calculating between doublings, where I know we expect .19166 primes. This means I can see the size we should expect to get to for any expected number of primes.
By the same idea, 191.63M=5.15M*2^(1/.19166), and 1-e^(-1)=0.632...

You can see this is probably about right by noting that .19166*5 is a little under 1, and if you double 5.15M five times you get 164.8M, which is a little smaller than the number I calculated.

Last fiddled with by Mini-Geek on 2009-11-27 at 12:48
Mini-Geek is offline   Reply With Quote
Old 2009-11-28, 15:32   #21
philmoore
 
philmoore's Avatar
 
"Phil"
Sep 2002
Tracktown, U.S.A.

1,117 Posts
Default

Our new prime is now listed on the Lifchitz site:
http://www.primenumbers.net/prptop/prptop.php.

philmoore is offline   Reply With Quote
Old 2009-12-02, 09:08   #22
Zuzu
 
Sep 2009

11 Posts
Default

OK I understand your calculation: 191M is the exponent n for which the probability of finding no PRP between 1 and n is equal to exp(-1) = 0.368, provided that no PRP has been found between 1 and 5.15M (thus neglecting the possibility of finding a PRP by double-checking).
IMO the reference probability is more based on 50% but as you have written, for each p in ]0,1[ one can similarily calculate the exponent n(p) for which the probability of finding no PRP between 1 and n is equal to p: n(p)=5.15M*2^(-ln(p)/0.19166))

For low values of p, n(p) increases dramatically: 7.54M for 0.9; 14.58M for 0.75; 63.17M for 0.5 but 775M for 0.25; 21G for 0.1; 261G for 0.05 and 88T for 0.01. There is a reasonable probability of finding the last PRP before say 50M (even before 10M if lucky as we were) but not finding any before say 1G is not to be ruled out.
OTOH for the SoB project, the mean number of primes per doubling being 1.04, the range of outcomes is less extended for the next prime. Provided that no prime has been found before 17.2M and neglecting double-check, n(p) would be equal to 18.5M for 0.9; 20.8M for 0.75; 27.3M for 0.5; 43.4M for 0.25; 80.1M for 0.1; 127M for 0.05 and 373M for 0.01.

Last fiddled with by Zuzu on 2009-12-02 at 09:21 Reason: Error corrected
Zuzu is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
probable largest prime. sudaprime Miscellaneous Math 11 2018-02-05 08:10
Gigantic Probable Prime Triplet found Cybertronic Twin Prime Search 18 2011-08-20 13:36
Megadigit probable prime found, our third! philmoore Five or Bust - The Dual Sierpinski Problem 25 2009-09-09 06:48
Another record probable prime found! philmoore Five or Bust - The Dual Sierpinski Problem 15 2009-02-08 19:43
Record probable prime found! philmoore Five or Bust - The Dual Sierpinski Problem 18 2009-01-28 19:47

All times are UTC. The time now is 00:04.

Tue Sep 29 00:04:11 UTC 2020 up 18 days, 21:15, 0 users, load averages: 1.94, 1.63, 1.56

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.