mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2007-11-27, 13:37   #1
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

143608 Posts
Default Bizarre problem with gnfs-lasieve4I15e

I'm trying to work out optimal parameters for a rather large SNFS job, and was led, out of a desire to pull lots of relations out of each special-Q so that I wouldn't have to run special-Q well into the area of diminishing returns, to build gnfs-lasieve4I15e by replacing

Code:
all : $(BINDIR)/gnfs-lasieve4I12e $(BINDIR)/gnfs-lasieve4I13e \
      $(BINDIR)/gnfs-lasieve4I14e
with
Code:
all : $(BINDIR)/gnfs-lasieve4I12e $(BINDIR)/gnfs-lasieve4I13e \
      $(BINDIR)/gnfs-lasieve4I14e $(BINDIR)/gnfs-lasieve4I15e
in ggnfs/src/lasieve4/Makefile

But I found that this gave me fewer relations when I used a small-prime bound of 80 million than when I used one of 70 million.

The input file is below; I just ran 'gnfs-lasieve4I15e -a job'

Code:
n: 27312187167948796049898887842782596405655381943670578373120353781997447000328700858938482409418918448758778194857586271196003004729048486053340769161697167767956404785889687728602979284674332324623675660860406658739006337124213538943374972682241
type: snfs
skew: 1
c6: 2
c0: -1
Y1: -1
Y0: 1393796574908163946345982392040522594123776 
alim: 70 or 80 million
rlim: 70 or 80 million
lpba: 30
lpbr: 30
mfba: 62
mfbr: 62
alambda: 2.6
rlambda: 2.6
q0: 100000000
qintsize: 10000
For example, the relation
Code:
-77588747,213413649:41fe9f3,4eeceeb,642B,9991,1D98D,32E883,90F4A3,1D69,D,89:d809cef,16d10281,4E541,5BFC27,B92861,A7,ACF,5F5E1DF
appears in the output from sieving to 70M and not when I sieve to 80M; diffing the output files indicates that going to 80M lost me some relations and gained me none. If I do the same from the output files from running the same input file under gnfs-lasieve4I14e, I find (as expected) that I only gain relations.

The algebraic-side factors (spot the special-Q) are
Code:
[167 1]
[2767 1]
[320833 1]
[6028327 1]
[12134497 1]
[100000223 1] *
[226532591 1]
[382796417 1]
and the rational-side factors are
Code:
[13 1]
[137 1]
[7529 1]
[25643 1]
[39313 1]
[121229 1]
[3336323 1]
[9499811 1]
[69200371 1]
[82759403 1]
so there are no factors on either side in the 70M .. 80M range!
fivemack is offline   Reply With Quote
Old 2007-11-27, 17:12   #2
jasonp
Tribal Bullet
 
jasonp's Avatar
 
Oct 2004

2×29×61 Posts
Default

Quote:
Originally Posted by fivemack View Post
But I found that this gave me fewer relations when I used a small-prime bound of 80 million than when I used one of 70 million.
Grepping through the lattice siever source, most of the machine-specific versions of lasieve/*/siever_config.h define L1_BITS to be 14, so I suspect the lattice size cannot exceed this number. Maybe you can increment L1_BITS and recompile?
jasonp is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
2 holes in bizarre theorem about composite Mersenne Numbers wildrabbitt Math 120 2016-09-29 21:52
3,697+ (GNFS 220.9) pinhodecarlos NFS@Home 0 2014-12-24 19:13
3,766+ (GNFS 215.5) pinhodecarlos NFS@Home 34 2014-04-01 21:27
Looking for windows-compiled gnfs-lasieve4I15e Andi47 Factoring 15 2008-02-21 19:15
Bizarre!!!! Alert!!!! R.D. Silverman Programming 5 2005-07-27 16:21

All times are UTC. The time now is 04:41.

Mon May 17 04:41:01 UTC 2021 up 38 days, 23:21, 0 users, load averages: 2.93, 2.38, 2.54

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.