mersenneforum.org > Data What about the ",30,4000000" P-1s ?
 User Name Remember Me? Password
 Register FAQ Search Today's Posts Mark Forums Read

 2005-03-13, 14:57 #1 cheesehead     "Richard B. Woods" Aug 2002 Wisconsin USA 22×3×641 Posts What about the ",30,4000000" P-1s ? 679 of the exponents in the latest PMINUS1.TXT file have been P-1 factored to B1,B2 limits of 30,4000000. The exponents are clumped in ranges 16475647-16480001, 25890349-25898239, 30597587-30607201, 40012253-40024763 and 44719573-44721137. (Some exponents in those ranges have had P-1 done to higher limits.) According to a utility I wrote to calculate probabilities of finding factors at given B1,B2 for a given exponent (following the corresponding calculation code in Prime95), those limits (B1=30, B2=4000000) give a probability of less than 10^-50 of finding a factor on any of those 679 exponents. That's less than 0.000000000000000000000000000000000000000000000001%. I suggest that all those 679 exponents be considered to have had practically no P-1 done at all, and that the 58 such exponents in the 16475647-16480001 range be immediately added to the current Marin's Mersenne-aries P-1 project.
 2005-03-13, 15:11 #2 cheesehead     "Richard B. Woods" Aug 2002 Wisconsin USA 22·3·641 Posts Pfactor lines for the 58 exponents in the first range are: Pfactor=16475647,65,1 Pfactor=16475681,65,1 Pfactor=16475699,65,1 Pfactor=16475743,65,1 Pfactor=16475813,65,1 Pfactor=16475959,65,1 Pfactor=16475971,65,1 Pfactor=16475993,65,1 Pfactor=16476017,65,1 Pfactor=16476071,65,1 Pfactor=16476101,65,1 Pfactor=16476107,65,1 Pfactor=16476139,65,1 Pfactor=16476209,65,1 Pfactor=16476259,65,1 Pfactor=16476311,65,1 Pfactor=16476403,65,1 Pfactor=16476491,65,1 Pfactor=16476541,65,1 Pfactor=16476583,65,1 Pfactor=16476619,65,1 Pfactor=16476653,65,1 Pfactor=16476749,65,1 Pfactor=16476773,65,1 Pfactor=16476847,65,1 Pfactor=16476907,65,1 Pfactor=16476913,65,1 Pfactor=16476919,65,1 Pfactor=16476991,65,1 Pfactor=16477003,65,1 Pfactor=16477079,65,1 Pfactor=16477081,65,1 Pfactor=16477103,65,1 Pfactor=16477207,65,1 Pfactor=16477211,65,1 Pfactor=16477217,65,1 Pfactor=16477229,65,1 Pfactor=16477273,65,1 Pfactor=16477291,65,1 Pfactor=16477319,65,1 Pfactor=16477327,65,1 Pfactor=16477333,65,1 Pfactor=16477393,65,1 Pfactor=16477459,65,1 Pfactor=16477567,65,1 Pfactor=16477661,65,1 Pfactor=16477711,65,1 Pfactor=16477789,65,1 Pfactor=16478881,65,1 Pfactor=16478927,65,1 Pfactor=16478939,65,1 Pfactor=16478951,65,1 Pfactor=16478981,65,1 Pfactor=16479013,65,1 Pfactor=16479493,65,1 Pfactor=16479653,65,1 Pfactor=16479721,65,1 Pfactor=16480001,65,1 Last fiddled with by cheesehead on 2005-03-13 at 15:16
 2005-03-14, 12:54 #3 dave_0273     Oct 2003 Australia, Brisbane 47010 Posts Aha.. claimed by me!!!! No, you may not have them.... Thanks for picking these up. I have wanted to go back and do further p-1ing on many exponents that are currently in pminus1.txt. There are a fair few exponents have very little or no stage 2 done. The only reason that I haven't started doing this yet, is that there are still many exponents that have had NO p-1 work done on them yet, so they are (at the moment) more important. However, these are effectively 0, so I will do these. Cheesehead, could you please send me that program that you wrote please?? Please??? The one that works out the probabilty of finding a factor given B1 and B2. I will use it to work out what exponents have such low B1 and B2 values that they should be rep-1ed. (lol... prefix and suffix on "p-1".... who would of thought??). Last fiddled with by dave_0273 on 2005-03-14 at 12:56
 2005-03-14, 13:20 #4 garo     Aug 2002 Termonfeckin, IE 22·691 Posts I'd be interested to see if you find any factors on these. It is quite possible that this is due to an error in reporting. I do not think there is any way Prime95 could have chosen these bounds. It is however possible that someone used a manual script and screwed up.
2005-03-14, 16:12   #5
cheesehead

"Richard B. Woods"
Aug 2002
Wisconsin USA

769210 Posts

Quote:
 Originally Posted by garo I do not think there is any way Prime95 could have chosen these bounds.
Yeah, these were obviously performed via Pminus1=xxxxxxxx,30,4000000 worktodo lines, not Pfactor= lines.

Quote:
 It is however possible that someone used a manual script and screwed up.
Maybe, but perhaps someone just wanted to look for factors that were somewhat-large but easy to find (requiring not much CPU-time -- maybe a way for a slow CPU to contribute), and didn't necessarily think about the consequences of actually reporting such a low B1 bound (that it would put off mainstream efforts). I myself have done some small-bound (not that small) P-1 on my Puttputt75 but knew enough not to report unsuccessful results until I'd extended them higher or was committing to do so. I don't know who did these "30,4000000" P-1s, though I've been tempted several times to ask on this forum and the mailing list.

2005-03-14, 16:32   #6
cheesehead

"Richard B. Woods"
Aug 2002
Wisconsin USA

769210 Posts

Quote:
 Originally Posted by dave_0273 There are a fair few exponents have very little or no stage 2 done. The only reason that I haven't started doing this yet, is that there are still many exponents that have had NO p-1 work done on them yet, so they are (at the moment) more important.
Yes, I've thought of that, too. That was part of my original incentive for writing the probability utility -- to see where the trade-off (of cost of re-doing the previous work in the course of going to higher limits versus the value of extra factors found in terms of L-L time saved) line was. That line turned out to be higher than I'd imagined. It was a good idea for you to work on un-P-1ed exponents instead of extending small limits (assuming, in all this, that one doesn't have a copy of the save file from the end of the small-limit P-1 run). The probability function rises so slowly that it is indeed more productive to use your CPUs as you have.

Quote:
 However, these are effectively 0, so I will do these.
Yeah -- when I wrote "small" in the preceding paragraph, I meant order of B1=10^4. A B1 of 30 is smaller than "small".

Quote:
 Cheesehead, could you please send me that program that you wrote please??
It's written for Microsoft QBASIC (included with Windows 98, but not Windows XP) and comes with no documentation. Still want it?

Last fiddled with by cheesehead on 2005-03-14 at 16:35

2005-03-14, 17:14   #7
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

22·839 Posts

Quote:
 Originally Posted by cheesehead It's written for Microsoft QBASIC (included with Windows 98, but not Windows XP) and comes with no documentation. Still want it?
I don't know about Dave, but I'd like to see it, whatever language you wrote it in, please

2005-03-15, 00:44   #8
cheesehead

"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts

Okay. Without documentation, support, or warranty (but with a couple of test input files and corresponding output), here it is.

Reminder 1: I derived it from code in Prime95.

Reminder 2: It is a utility for calculating quantities associated with Prime95 P-1 factoring such as estimated cost (time) of computation or probability of finding a factor, not a program for actually performing P-1 factoring.
Attached Files
 PM1E.zip (50.6 KB, 161 views)

Last fiddled with by cheesehead on 2005-03-15 at 00:59

 2005-03-15, 20:27 #9 James Heinrich     "James Heinrich" May 2004 ex-Northern Ontario 22·839 Posts I tried running it, without success. If I run it as-is, I get "Bad file name or number" on line 388 (PRINT #7, "Unexpected error ="; ERR). If I comment that out, I get: Code: D:\PM1E\P95SF.DAT D:\PM1E\P95FFT.DAT D:\PM1E\P95FAC.DAT D:\PM1E\PM1IN.DAT Unexpected error = 53 Anyhow, I've tried porting some of the code to PHP, and I've got the time estimates for stage1/stage2 apparently almost working right, except the numbers I get back are double what they should be. If I could get your program working I could try and compare my output to your program's output and see if it's my porting that's erroneous or what.
 2005-03-20, 15:49 #10 James Heinrich     "James Heinrich" May 2004 ex-Northern Ontario 335610 Posts I've ported the B1/B2 probability code to PHP, with reasonable success (although for reasons unknown to me I've had to add a 1.15 "fudge factor" multiplier to get numbers that match Prime95) http://mersenne-aries.sili.net/prob.php
2005-03-21, 11:48   #11
cheesehead

"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts

Quote:
 Originally Posted by James Heinrich (although for reasons unknown to me I've had to add a 1.15 "fudge factor" multiplier to get numbers that match Prime95)
Did you read the "Fudge factor" section near the beginning of PM1E.bas? :-)

 Thread Tools

 Similar Threads Thread Thread Starter Forum Replies Last Post MooMoo2 Other Chess Games 5 2016-10-22 01:55 wildrabbitt Miscellaneous Math 11 2015-03-06 08:17 James Heinrich Software 2 2005-03-19 21:58 nitai1999 Software 7 2004-08-26 18:12

All times are UTC. The time now is 08:09.

Sun May 9 08:09:01 UTC 2021 up 31 days, 2:49, 0 users, load averages: 2.36, 2.62, 2.90

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.