mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > YAFU

Reply
 
Thread Tools
Old 2015-02-13, 21:52   #1200
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

59×157 Posts
Default

Quote:
Originally Posted by Dubslow View Post
... (0.23 is rounded up form 2/9ths, but even the 2/9ths was reduced by a fair margin.)
What was the number in question?
It is hard to debug, when you are just saying "This was on a C212."

Not all C212s are of equally unquestionable pedigree. Some of them, after removing an algebraic cofactor, become easier, and that is what I suspect yafu did internally.
37.92/.23 = 164.9 (not a C212)
Batalov is offline   Reply With Quote
Old 2015-02-13, 21:59   #1201
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

3,361 Posts
Default

Quote:
Originally Posted by Batalov View Post
What was the number in question?
It is hard to debug, when you are just saying "This was on a C212."
agreed, but...

Quote:
Originally Posted by Batalov View Post
Not all C212s are of equally unquestionable pedigree. Some of them, after removing an algebraic cofactor, become easier, and that is what I suspect yafu did internally.
37.92/.23 = 164.9 (not a C212)
... in this case yafu (prior to SVN 337 and 338) was doubly reducing the amount of ECM effort. I.e., it noticed that the input was SNFSable and therefore scheduled less ECM, but *did not* notice that the user had already reduced the level of ecm effort via a "-plan custom" line.

Now, it will check if the user has changed the ECM plan away from default and not fiddle with the ECM schedule if so.
bsquared is offline   Reply With Quote
Old 2015-02-13, 22:06   #1202
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

59·157 Posts
Default

Agreed, double corrections are dangerous.

Also, if you have time, you may want to look into those reducible polynomials. I've also found that FactorDB is also at fault on this issue - it actually suggests polynomials like these (in the "Auto-generated SNFS-Polynominal available!" section: http://factordb.com/snfs.php?id=1100000000527588343 for example).
Batalov is offline   Reply With Quote
Old 2015-02-14, 00:04   #1203
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

3,361 Posts
Default

Quote:
Originally Posted by Batalov View Post
Agreed, double corrections are dangerous.

Also, if you have time, you may want to look into those reducible polynomials. I've also found that FactorDB is also at fault on this issue - it actually suggests polynomials like these (in the "Auto-generated SNFS-Polynominal available!" section: http://factordb.com/snfs.php?id=1100000000527588343 for example).
I can't, unfortunately. Bug fixes and minor stuff, ok, but this would be too much.
bsquared is offline   Reply With Quote
Old 2015-02-14, 03:29   #1204
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

59·157 Posts
Smile

You don't have to make the solution too general.

E.g. just a few ad hoc rules: if the input parses as "(num)^5-1", then you replace the poly with "(num)^4+(num)^3+(num)^2+(num)+1".
If it parses as "(num)^5+1", then you replace the poly with "(num)^4-(num)^3+(num)^2-(num)+1".

Similar with "(num)^7-1" and "(num)^7+1".

If the input parses as "(num)^4-1", dump (num)-1, (num)+1, and (num)^2+1 at the user and quit with some error message. Let him worry about that (there is an old anecdote with this punch line).

If the input parses as "(num)^6-1", dump (num)-1, (num)+1, and (num)^4+(num)^2+1 at the user and quit with error message.

If the input parses as "(num)^6+1", dump (num)^2+1 and (num)^4-(num)^2+1 at the user and quit with error message.

Something like that. Or else: the siever will barf a ton of "BAD SCHED" messages and the factorization will either not finish or will take enormous time. Do you need that hassle? I think not. Simply nip the errors in the bud.
Batalov is offline   Reply With Quote
Old 2015-02-14, 05:34   #1205
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

3,361 Posts
Default

Quote:
Originally Posted by Batalov View Post
You don't have to make the solution too general.
But... but... that's the fun! Sometimes I can't help but to start to frame what a general solution would look like and then at some point step back and say, woah, that might be a lot of work. I really don't want yafu to turn into some sort of CA package.

What you suggest is probably not much work. I will have to see. Or maybe Dubslow could come back out of retirement... or someone else that feels adventurous.
bsquared is offline   Reply With Quote
Old 2015-02-14, 09:39   #1206
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3×29×83 Posts
Default

Quote:
Originally Posted by bsquared View Post
But... but... that's the fun! Sometimes I can't help but to start to frame what a general solution would look like and then at some point step back and say, woah, that might be a lot of work. I really don't want yafu to turn into some sort of CA package.

What you suggest is probably not much work. I will have to see. Or maybe Dubslow could come back out of retirement... or someone else that feels adventurous.
I've considered it, but Subversion is a pain in the ass, I have no desire to use it ever again :P
Dubslow is offline   Reply With Quote
Old 2015-02-14, 17:03   #1207
chris2be8
 
chris2be8's Avatar
 
Sep 2009

2·3·331 Posts
Default

Quote:
Originally Posted by Batalov View Post
Or else: the siever will barf a ton of &quot;BAD SCHED&quot; messages and the factorization will either not finish or will take enormous time. Do you need that hassle? I think not. Simply nip the errors in the bud.
I've updated my copy of factMsieve.pl to check how many relations it gets from each range and exit if it got zero. That stops it wasting too much time if this happens. You could make yafu do that as a backup.

Chris
chris2be8 is offline   Reply With Quote
Old 2015-02-15, 18:47   #1208
debrouxl
 
debrouxl's Avatar
 
Sep 2009

977 Posts
Default

Quote:
Originally Posted by Dubslow View Post
I've considered it, but Subversion is a pain in the ass, I have no desire to use it ever again :P
Subversion is at least decent. CVS, VSS and older stuff really are PITA.
debrouxl is offline   Reply With Quote
Old 2015-02-15, 20:19   #1209
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

D2116 Posts
Default

Quote:
Originally Posted by debrouxl View Post
Subversion is at least decent. CVS, VSS and older stuff really are PITA.
Agreed, I don't mind SVN. Dubslow maybe you just to try different clients. I use tortoise and really like it. Although I think you are a Ubuntu user... don't have any suggestions there.
bsquared is offline   Reply With Quote
Old 2015-02-16, 01:08   #1210
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

59×157 Posts
Default

When I hear, "if only they've provided the right shell...", "of course if it was coded it python...", "if not for Ubuntu...", "if not for SVN, I would have coded the hell out of it", it rather reminds me of a (rude) Russian proverb.
Batalov is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Running YAFU via Aliqueit doesn't find yafu.ini EdH YAFU 8 2018-03-14 17:22
YAFU-1.34 bsquared YAFU 119 2015-11-05 16:24
Yafu bug. storflyt32 YAFU 2 2015-06-29 05:19
yafu-1.33 bsquared YAFU 12 2012-11-08 04:12
yafu-1.32.1 bsquared YAFU 21 2012-09-04 19:44

All times are UTC. The time now is 00:48.

Sun Jan 24 00:48:50 UTC 2021 up 51 days, 21 hrs, 0 users, load averages: 2.53, 2.38, 2.15

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.