|
2006-10-15, 18:52
|
#12 |
Sep 2004
2·5·283 Posts |
Lars,
If you don't mind I will stay here for a while, not on PSP. There you have Bruce and company. I will also try to bring IronBits to the game, he helped to bring Free-DC to second spot at RieselSieve Project. Now it's easier to bring more people because there's a linux client.... Carlos |
|
|
|
2006-10-15, 21:16
|
#13 |
Jul 2003
wear a mask
2×829 Posts |
Question for ltd and axn1;
How are the sieve scores calculated? Are they based on the size of n or the size of the factor found? I was experimenting with P-1 factoring and found some 15-20 digit factors that seem to have artificially boosted my sieve scores.
I think the scores should be based on the size of the exponent removed, not the size of the factor found... regards |
|
|
|
|
2006-10-15, 21:29
|
#14 |
Apr 2003
22×193 Posts |
Scores are based on the factor found and the n removed.
To limit the impact of p-1 factors the maximum points awarded per factor is 6000. The reason why the size of a factor is used within the calculation is to compensate for the factor density change when it comes to higher sieving ranges. The ideal solution would be to have seperate stats for p-1,ecm and normal sieving. This is not possible at the moment as my database design does not support it and i do not have the time to change it. Cheers, Lars |
|
|
|
|
2006-10-31, 01:26
|
#15 |
|
Jul 2003
wear a mask
31728 Posts |
ltd,
Could you maybe give us the formula used to calculate the score of each found factor? I'm curious. My score per number of factors ratio is very high in the Sierpinski stats... regards, masser |
|
|
|
|
2006-11-02, 18:59
|
#16 | |
Jun 2003
116758 Posts |
Quote:
We need to go back and rescore the P-1 factors. ltd is well aware of the problem. Unfortunately there are no easy solutions so far 
|
|
|
|
|
|
2006-11-02, 19:49
|
#17 |
|
Jul 2003
wear a mask
2×829 Posts |
Ok... that explains it...
I think the formula should be more like: score = C*log(factor)*n or, for even less dependence on the size of the factor: score = C*log(log(factor))*n |
|
|
|
|
2006-11-02, 20:33
|
#18 | |
|
Jun 2003
31×163 Posts |
Quote:
But when a factor is found using P-1 or ECM, the formula is not applicable. Maybe the alternate formula you proposed could be used. Another option would be to score these as a PRP test. |
|
|
|
|
|
2006-11-03, 19:29
|
#19 | |
|
Jul 2003
wear a mask
2·829 Posts |
Quote:
|
|
|
|
|
|
2006-11-04, 08:10
|
#20 |
|
Apr 2003
22·193 Posts |
At the moment there is no easy way to handle p-1 or ECM factors different then "normal" sieving factors.
To solve this in a good way there would be a redesign of the DB structures needed. I see no chance to do this in the near future. The maximum points per factor is limited to 200 points at the moment. Lars |
|
|
|
|
2006-11-05, 16:31
|
#21 |
Oct 2006
1000000112 Posts |
another stat question:
for external k test (not LLRnet based), when i send negative prp results (non-prime), are they incorporated in the llrnet stats or not ? in fact i ask this 'cause i send about 870 negative results for my k (341552) and i see that the maximum tested value is equal to my max tested value. but there was no increase in "prp test done" column. Laurent |
|
|
|
|
2006-11-05, 19:06
|
#22 |
|
Apr 2003
22×193 Posts |
Yes all PRP results are put into the same DB and also into the same stats as the llrnet results. If the tests done does not increase that means that there is an error in one off my scripts.
I will check this out. But again after the next weekend. Lars |
|
|
|
 |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Overall Stats | Fred | PrimeNet | 7 | 2016-02-10 14:48 |
| CSVs for stats available + New combined stats | opyrt | Prime Sierpinski Project | 3 | 2010-05-31 08:13 |
| Stats question | em99010pepe | NFSNET Discussion | 8 | 2005-07-05 03:22 |
| stats on 15k | jocelynl | 15k Search | 5 | 2004-01-13 15:45 |
| P4 On Stats | HiddenWarrior | Hardware | 2 | 2003-08-13 14:39 |
