Author 
Message 
ktoSend message
Joined: 13 Jan 09 Posts: 27 ID: 34164 Credit: 41,469,365 RAC: 0

I recently put in ~30k WU's to Sophie Germain (around 220k credit)  and did not find any primes.
I've also been watching the subproject status for SG  and like me, it also appears no one else is finding any SG primes... for quite some time
I'd like to understand the SG subproject status page. Why is it for n=666666 there are so many more tasks being run? Based on the same page, other values of n are finding more primes with far fewer tasks.
For example, looking at the SG subproject status page, the probability of finding a prime for n=666666 is 0.0015%, where for other values of n it averages around 0.015%  an order of magnitude better!
I guess it would help if someone could explain:
* how the k and n values are chosen
* why n=666666 is running with a much bigger k range than other values of n
* SG primes are of the form p and 2p+1. What do k and n mean in the p and 2p+1 equation?
* when we SG move to a different n? Seems like n=666666 is pretty dry of primes.
Any thoughts here? Thanks. 



Ok, it's really only n=666666, but even k are included. So something_even*2^6666661 becomes half_of_something_even*2^6666671 and so on. That's why half of the n's are 666,666 (odd k). 1/4 of the n are 666,667 (k multiply of 2), 1/8 are 666,668 (k multiple of 4), and so on.
That should answer Q1 and Q2.
Q3: If p=k*2^n1 then 2*p+1=2*(k*2^n1)+1=2k*2^n2+1= k*2^(n+1)1
So if we find 2 consecutive n values with the same k prime. We found the Sophie Germain.
Q4: I don't know, why 666,666 has the fewest primes although having the most tests. Could be a coincidence or is for a reason the next person will share with us ;) 



but who picked n=666666 anyway.
according to the main thread and wikipedia (and other places)
there are already a large number of SG primes knowns.
ie from wikipedia
The first few Sophie Germain primes are:
2, 3, 5, 11, 23, 29, 41, 53, 83, 89, 113, 131, 173, 179, 191, 233, …... (sequence A005384 in OEIS).
The largest known Sophie Germain prime as of March 2010 is 183027 × 2265440−1.
so who picked n=666666 and why? is there a special reason?. or just cos it looked good.
why not n=123456 or n=999999
just curious mainly.. as to if i think its a worthwhile goal. 


ktoSend message
Joined: 13 Jan 09 Posts: 27 ID: 34164 Credit: 41,469,365 RAC: 0

It took me a couple minutes, but I understand what thommy3 is saying. Thanks.
While the mystery has been reduced, some mystery remains.
Clearly half of the the n=666666 SG's have odd k's. So why are the odd k's producing so few primes? Clearly the even k's did much better.
Maybe this is a behavior specific so SG primes?
Is there a prime distribution function for SG primes that can validate Primegrid's findings for n=666666?
Just wondering.
And to shoelace's point, how are k and n values chosen? Are they chosen by Primegrid admin's? Or is PG deterministically working through k's and n's that have yet to be tested?
When will a new n be chosen? When all k's are exhausted for the chosen k? Because, oddly, referring to SG subproject status, the "k max in progress" is larger than the "k max complete". Does that mean the k's have exceeded the range chosen? If so, maybe it's time for a new n.




I was noticing the lack of SGS primes as well...not sure why it's been so quiet. As for the choice of n=666666, my guess is that the folks in charge wanted the n value to be high enough to maintain top 5000 finds for some time and also be low enough to process a significant amount of WUs to try and give the project a better shot at finding the SG prime, and so just decided to double up the value used for the old TPS search (which was n=333333). I don't think there was anything else particularly significant to the choice, but I could be wrong...
____________




so if we accept that logic.. that it is twice 333333 for TPS.
why did tps pick 333333? 



so if we accept that logic.. that it is twice 333333 for TPS.
why did tps pick 333333?
Not sure...that was long before I starting crunching for Primegrid. I managed to find these discussions on mersenneforum.org about choosing n values:
http://www.mersenneforum.org/showthread.php?t=12056
http://www.mersenneforum.org/showthread.php?t=6974
Don't know if that satisfies your curiosity, but perhaps digging around further on those forums will provide more insight.
____________




I really would be interested to know why it's been so long since SGS discovered a prime. They used to pop up quite regularly, but now we've been stuck on #244 for quite some time.
____________



JohnHonorary cruncher
Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0

First, thank you to everyone for your wonderful patience. Being on summer hours, it took time to coordinate the investigation into the concerns raised here. We have some bad news and some good news.
The bad news: the validator had a problem with the large k in SGS primes. Any k>12 digits was not being marked as prime. :( This problem only affected the SGS project.
The good news: ALL SGS data will be rechecked. Preliminary review shows at least 295 primes. Unfortunately, no twins. SG testing is completed server side. It will take time to process all the primes. Therefore, please expect primes to start trickling in the next several days.
Thank you kto for inquiring about this and to the rest for continuing the concern. Once all the primes have been processed, it should help resolve the prime average concern. Now let's see if I can answer the other questions. :)
why n=333333 and n=666666
After the record twin at n=195000 was found, a larger n was chosen that allowed single primes to be reported to the Top 5000 Primes list. n=333333 was chosen because it was roughly 100K digits and had an entrance rank of 3092.
As for n=666666, Alan is mostly correct in his response. Also, n=666666 is roughly 200K digits, twice the size of n=333333. On a side note, n=666666 is a quad sieve which allows for the opportunity to find both SG and twin primes.
The SG Prime Search search field and probabilities are listed below. A new n will be chosen before all the k's are exhausted so that there's time to sieve.
k range: 1<k<41T
n=666666 (actual 666666666688)
sieve depth: p=200T
candidates remaining: 34,190,344
Probability of one or more significant pair = 80.1%
Probability of one or more SG = 66.7%
Probability of one or more Twin = 42.3%
____________




This is great news! I was worried that the primes were starting to dry up in this project...
Roughly 50 new primes to come out in the next few days, nice:) I crunched quite a bit of SGS during the dry spell so I like my odds for one!
Cheers!
Alan
____________



JohnHonorary cruncher
Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0

The good news: ALL SGS data will be rechecked. Preliminary review shows at least 295 primes. Unfortunately, no twins. SG testing is completed server side. It will take time to process all the primes. Therefore, please expect primes to start trickling in the next several days.
WOW! That took much longer than expected. A month later and finally, all SGS primes have been accounted for. Thank you again for your patience. Hopefully everyone has enjoyed seeing all these "extra" primes.
Unfortunately, as a side effect, the rash of new primes to the Top 5000 database has put the PPS project in jeopardy of dropping off the list. Therefore, we are calling all hands on deck from now until the Calendula Challenge to please add some resources to the PPS (LLR) project.
Please select Proth Prime Search (LLR) in your PrimeGrid preferences.
____________



mackerelVolunteer tester
Send message
Joined: 2 Oct 08 Posts: 2509 ID: 29980 Credit: 461,464,280 RAC: 310,142

Yay! Now that's complete, can we get an indication of the recent units per prime rate for SGS?
PPS LLR enabled :) I wanted to get pink in it as a long term goal after gold SGS, which can take a little longer. Will have another 6 overclocked (stable!) cores switching over as soon as I get silver in TRPsieve shortly. So many projects, so little CPU time...
While I'm here, anonymous host 114811 seems to be putting out a fair proportion of invalid SGS units. 



Looks like I may have just done the deed  or perhaps I am a confirmer  or maybe there's another explanation. No notifications have been received, but my count of primes found for Sophie Germain has inexplicably gone from zero to one today......
Dang, the suspense is killing me!!!!
____________




First, congrats on getting a hit. Second, I was just informed the SG confirmations are just a tad behind (see post on missing SGS primes). And third, yes not knowing does add a lot of suspense especially if it's your first one.
Regardless congrats again and hopefully you are the prime finding owner :)
____________
@AggieThePew




Looks like I may have just done the deed!
Congrats! Me jealous :) 



First, congrats on getting a hit. Second, I was just informed the SG confirmations are just a tad behind (see post on missing SGS primes). And third, yes not knowing does add a lot of suspense especially if it's your first one.
Regardless congrats again and hopefully you are the prime finding owner :)
Well if I'm not  somebody else will be !!!!
____________




Just read that thread  wow, quite a backlog: 90 SGS)!!!!
Still, what will happen will happen; I can't change what has gone before. Irrespective of what my participation turns out to be, I'll just carry on chugging away, which after all is my function in all of this!
____________




Nice to hear that primes are being found on SGS. Like you I am hanging in my suspenders waiting to find out if I am the finder or confirmer to 2 primes that popped up unexpectedly about 4 days ago.
I am so excited as they are my first that I am checking to see what the result is almost every hour Lol. :)
____________
The art of flying is throwing yourself at the ground and missing.




Well, good and bad news for me:
Dear Primefinder,
Congratulations! Our records indicate that a computer registered by you has
found a unique prime number. This computer is running BOINC, is attached to
the PrimeGrid project, and is assigned to the Sophie Germain Prime Search. What makes
this prime unique is that it's large enough to enter the Top 5000 List in Chris Caldwell's
The Largest Known Primes Database.
Since you have autoreporting selected, the following prime was submitted on your
behalf:
Added 97033 : 6130676192025*2^6666671 (200700 digits)
If you have any questions or concerns, please contact me and we will surely resolve any
problem.
Once again, congratulations on your find. Thank you for participating in PrimeGrid.
John Blazek of PrimeGrid
The bad news is that my PSU went West on Friday night!!! Haven't been crunching since then, as its taken that long to get a replacement delivered. Shall be fitting it this morning. Unfortunately, I have had to watch my entire 600 task cache time out waiting for the replacement!
____________




Congrats on your prime.. now for a few anyway your name will be in lights on the main page. I'm still waiting to hear about my one remaining hit.
____________
@AggieThePew




Congrats on your prime.. now for a few anyway your name will be in lights on the main page. I'm still waiting to hear about my one remaining hit.
.ahhh, my moment of fame has almost gone  last in the list atm. I take it you have no news yet?
____________




Grats on being on the front page Area 51
Still waiting to find out if I am primary finder or validation for the 2 that have shown up on my efforts. :)) Hoping that I am primary finder for both
____________
The art of flying is throwing yourself at the ground and missing.




Congrats on your prime.. now for a few anyway your name will be in lights on the main page. I'm still waiting to hear about my one remaining hit.
.ahhh, my moment of fame has almost gone  last in the list atm. I take it you have no news yet?
No news yet and it's been almost 5 weeks... I think they are behind on getting the SGS confirmation in the top 5000 db. Anyway, I know I have 2 hits but going back to find them would be an exercise in going blind :) way too many to review.
____________
@AggieThePew




A fair number of the 'Newly reported primes' on the front page are Sophies, so perhaps they are starting to clear the backlog. However in the end  if you found it you found it  that won't change however long it takes.
I considered going through tasks and trying to find the requisite one, but decided it really wasn't worth the time and although I'm pretty new to this prime game, I absolutely can understand the 'wanting to know' feeling! I even considered writing a PERL script to page scrape my tasks and find the blighter  but then thought that's not fair on the servers!!!!!
Perhaps Santa will come early for you (like tomorrow)!
____________




A fair number of the 'Newly reported primes' on the front page are Sophies, so perhaps they are starting to clear the backlog. However in the end  if you found it you found it  that won't change however long it takes.
I considered going through tasks and trying to find the requisite one, but decided it really wasn't worth the time and although I'm pretty new to this prime game, I absolutely can understand the 'wanting to know' feeling! I even considered writing a PERL script to page scrape my tasks and find the blighter  but then thought that's not fair on the servers!!!!!
Perhaps Santa will come early for you (like tomorrow)!
LOL  one of my wishes on here was in fact a list of work done that could be exported so we could as users browse each of our tasks. Maybe one of these days they can do that.
Since it's been sooo long since the hit occurred I'm just hoping it's like some major unknown SGS and they can't figure out how to validate it :)
Now there's a xmas present!
Cheers (as Tim says)
____________
@AggieThePew




Yippe One is confirmed as finder and the other double checker :)
Entrance number in the top 5000 DB 2582
Dear Primefinder,
Congratulations! Our records indicate that a computer registered by you has
found a unique prime number. This computer is running BOINC, is attached to
the PrimeGrid project, and is assigned to the Sophie Germain Prime Search. What makes
this prime unique is that it's large enough to enter the Top 5000 List in Chris Caldwell's
The Largest Known Primes Database.
Since you have autoreporting selected, the following prime was submitted on your
behalf:
Added 97098 : 6081793616325*2^6666671 (200700 digits)
If you have any questions or concerns, please contact me and we will surely resolve any
problem.
Once again, congratulations on your find. Thank you for participating in PrimeGrid.
John Blazek of PrimeGrid
____________
The art of flying is throwing yourself at the ground and missing.




Yippe One is confirmed as finder and the other double checker :)
Entrance number in the top 5000 DB 2582
Dear Primefinder,
Congratulations! Our records indicate that a computer registered by you has
found a unique prime number. This computer is running BOINC, is attached to
the PrimeGrid project, and is assigned to the Sophie Germain Prime Search. What makes
this prime unique is that it's large enough to enter the Top 5000 List in Chris Caldwell's
The Largest Known Primes Database.
Since you have autoreporting selected, the following prime was submitted on your
behalf:
Added 97098 : 6081793616325*2^6666671 (200700 digits)
If you have any questions or concerns, please contact me and we will surely resolve any
problem.
Once again, congratulations on your find. Thank you for participating in PrimeGrid.
John Blazek of PrimeGrid
Congrats  onto the next one then!
____________




Just wondering if validation for SG primes are still backlogged...can anyone cofirm or is it just a longer process with SGP's? I've had 8 hits with only one confirm...two are fairly recent but the other 5 popped up a while back...probably more than a month and a half ago. 



Ignore that previous post...it was answered in another thread. I also found the edit button, which was staring me in the face. :)
Cheers... 

