Author |
Message |
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
We have released an updated Top Prime Finders page, which now counts primes found in all subprojects. The prime numbers are assigned a score which differs according to prime number size, so a giant Woodall prime is "worth" more than a smaller TPS prime."
Top Prime Finders - also located in the left menu
Legend
Primary check = primes found
Double Check = primes double checked
Total = Primary check + Double check
Score = number equivalent to 100,000 digit prime (i.e. a score of 19 represents 19 100000 digit primes)
BOINC Credit = BOINC credit
Scoring Information
The scoring formula is similar to the one used at The Largest Known Primes Database. However, PrimeGrid awards 1/2 points to double checkers. Also, we only display the "normalized" score based on a 100,000 digit prime instead of the 5000th prime which is ever changing.
An indiviudal's total score is the sum of all their primary finds and double checks. Double checks receive 1/2 the score of the primary.
Generally speaking, the score is based on the "difficulty" of finding the prime. Clearly a 1,000,000 digit prime is much harder to find than a 100,000 digit prime. Based on our minor changes, here's the formula for score:
Given: k*b^n+/-1
log_n = ln(k) + n*ln(b)
score = (log_n)^3 * ln(log_n) / 100,000 digit score
For example, using 7668057255*2^666667-1, leaving out the 100,000 digit score section for now, we get the following:
log_n = ln(k) + n*ln(b)
log_n = ln(7668057255) + 666667*ln(2)
log_n = 22.7603 + 462098.3514
log_n = 462121.1117
score = (log_n)^3 * ln(log_n)
score = (462121.1117)^3 * ln(462121.1117)
score = 9.86887E+16 * 13.0435
score = 1.28725E+18
As mentioned earlier, we "normalize" this to a 100,000 digit prime. Applying that, we get the following:
score of a 100,000 digit prime is 1.50733E+17
score = prime score / 100,000 digit score
score = 1.28725E+18 / 1.50733E+17
score = 8.5400
Therefore, the prime 7668057255*2^666667-1 is equivalent to finding 8.54 100,000 digit primes. The primary finder would be awared 8.54 points and the double checker would be awarded 4.27 points since the double checker gets 1/2 credit.
A typical TPS score at n=333,333 is approximately 1.01 points with double check getting 0.51 points.
____________
|
|
|
|
Score = number equivalent to 100,000 digit prime (i.e. a score of 19 represents 19 100000 digit primes)
Given: k*b^n+/-1
log_n = ln(k) + n*ln(b)
score = (log_n)^3 * ln(log_n) / 100,000 digit score
Just out of curiousity, if the scores are based on the number of decimal digits, why are you using natural logarithms?
|
|
|
|
Thanks so much John for implementing this great new option :)
As always you have my respect.
____________
John M. Johnson "Novex" |
|
|
|
Just a question - Are there some prime finds that are being disallowed or otherwise skipped on this page? For example, I show up with 23 primes found and 12 doublechecked, for a total of 35, but my account page and the certificate of computation show 45 TPS primes found (and so did the old version of the page).
Is it just that the new page is somehow behind the times in what it is counting or is it related to a problem in reporting - since I haven't gotten any notices to report any of the new primes since February?
____________
|
|
|
pan2000Volunteer tester Send message
Joined: 23 Jan 07 Posts: 26 ID: 5226 Credit: 328,291,395 RAC: 0
                    
|
Same problem here. Missing 19 primes. |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2649 ID: 1 Credit: 28,208,848 RAC: 122,605
                     
|
I believe that the new database is showing correct numbers. The issue with inflated counts in your account page is the same as it was with factors being counted more than once when validator was having issues. The current table has all primes that you found, and the stats page simply counts them.
In any case, I will look into it. I've squeezed it into my schedule for Thursday evening.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Score = number equivalent to 100,000 digit prime (i.e. a score of 19 represents 19 100000 digit primes)
Given: k*b^n+/-1
log_n = ln(k) + n*ln(b)
score = (log_n)^3 * ln(log_n) / 100,000 digit score
Just out of curiousity, if the scores are based on the number of decimal digits, why are you using natural logarithms?
Just note that we are "borrowing" this formula from The Largest Known Primes Database. However, to answer your question,
it is my understanding that natural log is a more commonly used expression in mathematics. Wikipedia has a nice explanation as to why this may be the case:
Initially, it might seem that since our numbering system is base 10, this base would be more “natural” than base e. But mathematically, the number 10 is not particularly significant. Its use culturally—as the basis for many societies’ numbering systems—likely arises from humans’ typical number of fingers.[2] And other cultures have based their counting systems on such choices as 5, 20, and 60.[3][4][5]
Log e is a “natural” log because it automatically springs from, and appears so often, in mathematics.
If you wish, you can use log base 10 instead of log base e (ln) in the expression. It will produce similar results.
Thanks so much John for implementing this great new option :)
As always you have my respect.
Thanks! However, I'll have to transfer this to Rytis as he is the one who did all the technical work. :)
____________
|
|
|
|
Thanks so much John for implementing this great new option :)
As always you have my respect.
Thanks! However, I'll have to transfer this to Rytis as he is the one who did all the technical work. :)[/quote]
Ah I actually have been leaving him to his semester finals which I know first hand can be sometimes crazy. But I really do appreciate the effort in which this great community helps one another. I humbly respect you great mods and this wonderful community :)
____________
John M. Johnson "Novex" |
|
|
|
Just note that we are "borrowing" this formula from The Largest Known Primes Database. However, to answer your question,
it is my understanding that natural log is a more commonly used expression in mathematics. Wikipedia has a nice explanation as to why this may be the case:
Thanks for the reply. I know it really makes no difference what base logarithm you use for determining the score. It was just idle curiousity as all mentions are of decimal digits. :) |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2649 ID: 1 Credit: 28,208,848 RAC: 122,605
                     
|
I believe that the new database is showing correct numbers. The issue with inflated counts in your account page is the same as it was with factors being counted more than once when validator was having issues. The current table has all primes that you found, and the stats page simply counts them.
In any case, I will look into it. I've squeezed it into my schedule for Thursday evening.
Thanks to BobW, I think I found some missing primes. Do you think it's OK now?
____________
|
|
|
|
Unfortunately mines still not showing any of the double checks.
____________
John M. Johnson "Novex" |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2649 ID: 1 Credit: 28,208,848 RAC: 122,605
                     
|
Unfortunately mines still not showing any of the double checks.
Because the initial finder still has not reported them to the prime pages.
____________
|
|
|
|
Ah, that would explain it then. Thanks a ton Rytis, as always you have my respect.
____________
John M. Johnson "Novex" |
|
|
|
My 'your account' page says I have one tps prime, but I cannot find myself on the list, and I don't see any way to find the work unit it's talking about
____________
|
|
|
|
It's the one with 2.90 credits instead of 1.45.
28582520
I'm not sure why you get the strange message but this is a prime number.
Warning: mysql_fetch_object(): supplied argument is not a valid MySQL result resource in /home/boinc/proj/html/project/project.inc on line 274
*2^-1 is not prime. |
|
|
|
I looked, it must have only recently been validated. I'm still not on the page http://www.primegrid.com/primes/?section=topusers&by=score though?
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
I looked, it must have only recently been validated. I'm still not on the page http://www.primegrid.com/primes/?section=topusers&by=score though?
You were the double checker...and the primary checker has not reported it to the Prime Pages yet. It will show up after they report it.
____________
|
|
|
|
Ah, OK. This system is very confusing. I like the old days of BOINC when they didn't let us users get into the technical stuff :)
____________
|
|
|
|
Are found primes reported automatically to The Largest Known Primes-List and if so when/how often? Or do i/we have to report them ourselves?
I am somewhat puzzled here... ;) |
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 663 ID: 845 Credit: 2,241,803,723 RAC: 1,221,404
                         
|
Are found primes reported automatically to The Largest Known Primes-List and if so when/how often? Or do i/we have to report them ourselves?
I am somewhat puzzled here... ;)
They are reported automatically, if you give the permission in your preferences. Else you should get an email with the reporting instructions.
At the moment, only primes found in Cullen, Woodall, 321 and PSP LLR subprojects are big enough for the list, TPS and PPS LLR are too small (we have to reach n~350k for big enough numbers ;)).
____________
|
|
|
|
A-OK! Thanks for clarification. |
|
|