mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2013-03-04, 19:02   #1
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

637910 Posts
Default M947 progress

Code:
n: 2847922034528971677290051549879038840869047309341551394654637054075417375981106395118700814298148813029564177907105065410846683443357438966188671989301877928681552715177521447358804775039379552290813448288911916720697
c6: 1
c0: -2
skew: 1.45
Y1: -1
Y0: 365375409332725729550921208179070754913983135744
lpbr: 33
lpba: 33
mfbr: 96
mfba: 66
alambda: 2.6
rlambda: 3.6
alim: 240000000
rlim: 240000000
Run with 16e -r. 20M - 240M ought to be enough. Runtime is about 3.7 CPU-seconds per Q on my slow CPUs, so my 96MQ should take until about mid-June.

Reservations
Code:
16/03/2013  dubslow      20M-26M  11/04/2013 34522766
07/03/2013  jyb          26M-27M  26-33 finished 20/03/2013 42664400
05/03/2013  jyb          27M-28M  ...
07/03/2013  jyb          28M-30M  ...
11/03/2013  jyb          30M-33M  ...
04/03/2013  batalov      33M-34M  31/03/2013  6069731
*18/03/2013 pinhodecarlos 34M-42M
21/04/2013  bdodson      34M-67M  01/05/2013  197573666
04/03/2013  batalov      67M-68M  09/04/2013  5743011
08/04/2013  mathew       68M-72M  19/05/2013 22766932
04/03/2013  fivemack     72M-120M  72-84  finished 31/03/2013 65998911
                                   84-96  finished 20/04/2013 63790339
                                   96-108 finished 08/05/2013 61184029
                                  108-120 finished 27/05/2013 59701118
31/03/2013  bdodson     120M-168M  21/04/2013 224931506
04/03/2013  batalov     168M-172M  09/04/2013 17715018
20/03/2013  bdodson     172M-230M  21/04/2013 243152170
07/03/2013  jcrombie    230M-231M  14/03/2013  3998161
14/03/2013  jcrombie    231M-235M  10/04/2013 15878724
04/04/2013  bdodson     235M-240M  20/04/2013 19769707
The original version had skew=0.891, which I think was a braino; jcrombie did some tests and found 1.45 gave higher yields. The relations from both runs are entirely compatible

Last fiddled with by fivemack on 2013-05-27 at 07:27
fivemack is offline   Reply With Quote
Old 2013-03-04, 20:02   #2
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

100100001011102 Posts
Default

I'll take my favorite chunks (breaks of powers-of-2). Ooops. None left.

Will take 168-172M for starters.
Batalov is offline   Reply With Quote
Old 2013-03-04, 21:12   #3
jcrombie
 
jcrombie's Avatar
 
"Jonathan"
Jul 2010
In a tangled web...

110101102 Posts
Default non-optimum skew?

I'm getting a MurphyE of 1.172e-14 with skew = 0.891 and 1.190e-14 with skew = 1.45.
Doing a little test sieving around q=120000000 seems to give a slightly better yield.

Now I'll let the experts continue this (because I barely know what I'm talking about here).

Cheers,
Jonathan
jcrombie is offline   Reply With Quote
Old 2013-03-04, 22:33   #4
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

6,379 Posts
Default

Quote:
Originally Posted by Batalov View Post
I'll take my favorite chunks (breaks of powers-of-2). Ooops. None left.

Will take 168-172M for starters.
33-34 and 67-68 are available
fivemack is offline   Reply With Quote
Old 2013-03-04, 22:56   #5
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

2·11·421 Posts
Default

Oh, I'll take them as well. Thanks!

The skew proposal is not without merit (even though needs a more sizable test).
Do my eyes deceive me when they tell me that 3LP is on the other side (compared to M929)?

Last fiddled with by Batalov on 2013-03-04 at 22:58
Batalov is offline   Reply With Quote
Old 2013-03-05, 06:57   #6
jcrombie
 
jcrombie's Avatar
 
"Jonathan"
Jul 2010
In a tangled web...

2×107 Posts
Default the empircal recapitulates the theorectical

just some extra info:

with skew = 0.891
Code:
total yield: 31234, q=120006041 (0.57243 sec/rel)
5.17 rel/q

with skew = 1.45
Code:
total yield: 31623, q=120006013 (0.56493 sec/rel)
5.26 rel/q

So, apprx. a 1.7% improvement. Nothing to get too much excited about.
jcrombie is offline   Reply With Quote
Old 2013-03-05, 08:34   #7
axn
 
axn's Avatar
 
Jun 2003

10010111011012 Posts
Default

The nominal skew should be (c0/c6)^(1/6) = 1.12 rather than (c6/c0)^(1/6) = 0.891.

After scoring various skews with msieve:
Code:
skew 0.89, size 6.032e-014, alpha 2.482, combined = 1.330e-014 rroots = 2
skew 1.00, size 6.032e-014, alpha 2.482, combined = 1.338e-014 rroots = 2
skew 1.12, size 6.032e-014, alpha 2.482, combined = 1.344e-014 rroots = 2
skew 1.20, size 6.032e-014, alpha 2.482, combined = 1.347e-014 rroots = 2
skew 1.30, size 6.032e-014, alpha 2.482, combined = 1.349e-014 rroots = 2
skew 1.35, size 6.032e-014, alpha 2.482, combined = 1.349e-014 rroots = 2
skew 1.36, size 6.032e-014, alpha 2.482, combined = 1.349e-014 rroots = 2
skew 1.37, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.38, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.40, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.45, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.46, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.47, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.48, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.49, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.50, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.55, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.56, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.57, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.58, size 6.032e-014, alpha 2.482, combined = 1.350e-014 rroots = 2
skew 1.59, size 6.032e-014, alpha 2.482, combined = 1.349e-014 rroots = 2
skew 1.60, size 6.032e-014, alpha 2.482, combined = 1.349e-014 rroots = 2
The optimum appears to be around 1.48, but anything between 1.3-1.6 is indistinguishable. 1.350/1.330= 1.015 or just 1.5% improvement between the best and worst. 1.350/1.344 = 1.0044 or just 0.4% improvement over the nominal one.
axn is online now   Reply With Quote
Old 2013-03-05, 14:00   #8
jasonp
Tribal Bullet
 
jasonp's Avatar
 
Oct 2004

1101110011102 Posts
Default

Note that the constants buried in the E-value calculation above are specific to the sieving area (1e16) and the factor base bound (primes < about 1M; actually the rational bound is less than the algebraic one, which is particularly unrealistic for large SNFS).
jasonp is offline   Reply With Quote
Old 2013-03-05, 14:33   #9
R.D. Silverman
 
R.D. Silverman's Avatar
 
Nov 2003

22·5·373 Posts
Default

Quote:
Originally Posted by jasonp View Post
Note that the constants buried in the E-value calculation above are specific to the sieving area (1e16) and the factor base bound (primes < about 1M; actually the rational bound is less than the algebraic one, which is particularly unrealistic for large SNFS).
Are they really using a factor base bound of only 1M??? This is much too
small! Or do you mean the size of the factor base set is 1M?
R.D. Silverman is offline   Reply With Quote
Old 2013-03-05, 18:06   #10
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

3,361 Posts
Default

Quote:
Originally Posted by R.D. Silverman View Post
Are they really using a factor base bound of only 1M??? This is much too
small! Or do you mean the size of the factor base set is 1M?
I assume he really means factor base bound. But Jason's talking about the constants used in msieve's Murphy-E calculation, which are completely unrelated to the actual job parameters of post #1.

I guess this means that the calculated E values are useful for comparative purposes, but perhaps not 100% accurate in an absolute sense?
bsquared is offline   Reply With Quote
Old 2013-03-05, 20:50   #11
jyb
 
jyb's Avatar
 
Aug 2005
Seattle, WA

110011001102 Posts
Default

I'll take 27M - 28M.
jyb is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Progress bsquared YAFU 20 2014-05-22 16:52
Nice progress! schickel FactorDB 29 2012-07-18 17:03
Progress R.D. Silverman Factoring 0 2012-05-22 14:03
In Progress? R.D. Silverman Cunningham Tables 33 2010-05-07 14:02
Picturing progress ATH Data 1 2006-06-22 23:04

All times are UTC. The time now is 08:54.

Sat Jan 23 08:54:15 UTC 2021 up 51 days, 5:05, 0 users, load averages: 1.67, 1.50, 1.56

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.