Have a look at the partition numbers - Page 3

smh · 2007-05-22, 10:29

Quote:

Originally Posted by fivemack

That's quite impressive: 45 factorisations of about eight hours each in three days, so at least five full-time CPUs. Do you have a patch for the "lambda-comp" bug during polynomial selection that has thwarted my attempts at running completely automated ggnfs?

The c104's actually take a bit longer on the servers i'm running them (some are dual cpu). The 8 hours is alos exluding poly search as that is not recorded in the log file.

I don't know what you mean with a "lambda-comp" patch. I'm just creating a batchfile where i call factlat.pl with the relevant .n file.

In between factorizations i run a cleanup script to remove all files exept the results file.

It's really fire and forget. ATM i don't have time to create poly's etc, thats why i'm running these numbers. It's just copy/past from a file i grep from the website

fivemack · 2007-05-22, 11:34

Ah. I find that about one factlat.pl run in three terminates half-way through polynomial selection with a message 'lambda-comp' from pol51opt.

This isn't a problem if your batchfile notices that the run has failed and launches it again - it's generally found a half-decent polynomial by that point and just carries on using that one - but that makes the run-script a little harder to write. It's a little more annoying if you launch four jobs manually, go to bed, and find that two CPUs have idled overnight.

smh · 2007-05-22, 13:23

Quote:

Originally Posted by fivemack

Ah. I find that about one factlat.pl run in three terminates half-way through polynomial selection with a message 'lambda-comp' from pol51opt.

This isn't a problem if your batchfile notices that the run has failed and launches it again - it's generally found a half-decent polynomial by that point and just carries on using that one - but that makes the run-script a little harder to write. It's a little more annoying if you launch four jobs manually, go to bed, and find that two CPUs have idled overnight.

I can't find it in my own script (are you using the latest from sourceforge?) but i remember long time ago, commenting out one or two lines in the script (think a line with 'die')

My batch file looks something like:

factlat.pl 1
call cleanup.bat
factlat.pl 2
call cleanup.bat
....

Where i put the numbers in 1.n, 2.n, etc

For composites below 110 digits i don't treally care if the poly is optimal or not. An hour more or less is not worth the time babysitting my computers.

smh · 2007-05-24, 21:36

Quote:

Originally Posted by smh

I'll continue with the 105 digit composites.

Done.

I'll continue with the 106 digit composites.

Is anyone working on the smallest composites (101 and 102 digits)?
Although GNFS might be faster, they are nice candidates for msieve.

nuggetprime · 2007-05-28, 17:25

reserving 28314 c101 for GNFS

Andi47 · 2007-05-30, 05:03

Quote:

Originally Posted by Andi47

Reserving 29328 c101

Tue May 29 11:42:42 2007 prp49 factor: 3519998070322949609923151104348917738657885664351
Tue May 29 11:42:42 2007 prp53 factor: 17003238992013333150854317713525675502414489255329857

Quote:

Originally Posted by Andi47

Edit: Reserving 29130 c103

Mon May 28 05:40:56 2007 prp46 factor: 7701007183832465344772923932080967488831333947
Mon May 28 05:40:56 2007 prp57 factor: 572015095894199635535366612427087111835980569963266301783

Andi47 · 2007-05-30, 17:09

Reserving 20647 c101 and 20891 c102

Andi47 · 2007-05-31, 17:40

Quote:

Originally Posted by Andi47

Reserving 20647 c101 and 20891 c102

20647 c101:
Thu May 31 08:56:49 2007 prp36 factor: 552308749347187909031142322734203299
Thu May 31 08:56:49 2007 prp65 factor: 24927766151680063520982636974281103909592483730458610592774550481

20891 c102:
Thu May 31 11:24:36 2007 prp48 factor: 348576993433347561253167275618901474537488577701
Thu May 31 11:24:36 2007 prp55 factor: 1110662031911332571745059263860379031583132441968015533

smh · 2007-05-31, 18:40

Quote:

Originally Posted by smh

I'll continue with the 106 digit composites.

Done, including the first 4 c107's.

nuggetprime · 2007-06-01, 12:34

Quote:

Originally Posted by nuggetprime

reserving 28314 c101 for GNFS

I didn't factor this one with gnfs because I didn't know how to use ggnfs. but now I know and with my next reservation, 29328, I will try to factor, with factLat.pl, my first number using gnfs.

Factors of 28314:

Code:

p47 factor: 11293260017991808632377462474444514902170890901
p55 factor: 6800209839747876175751542155675097921696821774445715093

(factored with msieve/QS)

nuggetprime

smh · 2007-06-03, 07:48

Andi,

Did you report the factors you found?

All numbers for which you showed factors are still marked composite on the site.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Fun with partition function	Batalov	And now for something completely different	24	2018-02-27 17:03
Carmichael numbers and Devaraj numbers	devarajkandadai	Number Theory Discussion Group	0	2017-07-09 05:07
Partition number congruences	fivemack	Math	0	2007-05-18 19:39
Linux/SUSE noob trouble - Resize partition	OmbooHankvald	Linux	19	2005-11-18 10:39
LLT numbers, linkd with Mersenne and Fermat numbers	T.Rex	Math	4	2005-05-07 08:25

2007-05-22, 11:34	#24
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 1100100010111₂ Posts	Ah. I find that about one factlat.pl run in three terminates half-way through polynomial selection with a message 'lambda-comp' from pol51opt. This isn't a problem if your batchfile notices that the run has failed and launches it again - it's generally found a half-decent polynomial by that point and just carries on using that one - but that makes the run-script a little harder to write. It's a little more annoying if you launch four jobs manually, go to bed, and find that two CPUs have idled overnight.

2007-05-28, 17:25	#27
nuggetprime Mar 2007 Austria 100101110₂ Posts	reserving 28314 c101 for GNFS

2007-05-30, 17:09	#29
Andi47 Oct 2004 Austria 2·17·73 Posts	Reserving 20647 c101 and 20891 c102

2007-06-03, 07:48	#33
smh "Sander" Oct 2002 52.345322,5.52471 29·41 Posts	Andi, Did you report the factors you found? All numbers for which you showed factors are still marked composite on the site.