mersenneforum.org fond of a factor? Turn yourself in to become inane
 Register FAQ Search Today's Posts Mark Forums Read

2020-09-30, 13:26   #1673
charybdis

Apr 2020

7316 Posts

Quote:
 Originally Posted by James Heinrich If you want to provide me a sublist of factors from that list that cannot possibly be ECM, or perhaps some criteria that I could walk through the database and weed out the non-ECM ones (as I did with the below-TF-limit ones) I'd be happy to trim the database.
These were all done by NFS, per the Cunningham tables site:

1061
1129
991
1193
1151
863
823
1153
853
827
1019
857 (both factors)
1031
887
877
941
859
821
929

I think this is all of them, though there's a small chance that I could have missed a GNFS factorization of a slightly larger exponent.

2020-09-30, 13:54   #1674
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

319010 Posts

Quote:
 Originally Posted by charybdis These were all done by NFS, per the Cunningham tables site: I think this is all of them, though there's a small chance that I could have missed a GNFS factorization of a slightly larger exponent.
Thanks, I have removed those from the data.

2020-09-30, 17:20   #1675
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

2×5×11×29 Posts

Ryan strikes again, 4 6 10 new factors for very small exponents within the space of an hour a few hours:
Quote:
 M4157 has a 144.884-bit (44-digit) factor: 41161769340018076868666055145622172588108913 (ECM,B1=110000000,B2=900514153782) M4259 has a 138.475-bit (42-digit) factor: 484399053828087857927777245183883458323463 (ECM,B1=110000000,B2=900514153782) M4297 has a 136.147-bit (41-digit) factor: 96437429855240462114489441553704488059887 (ECM,B1=110000000,B2=900514153782) M4327 has a 128.869-bit (39-digit) factor: 621316165862254523582603721415095832031 (ECM,B1=110000000,B2=900514153782) M4349 has a 135.127-bit (41-digit) factor: 47571370031120172757491816499652518228991 (ECM,B1=110000000,B2=900514153782) ** now fully factored M4027 has a 159.856-bit (49-digit) factor: 1322603807888784213814883095222492416081030114751 (ECM,B1=110000000,B2=900514153782) #123 largest M4447 has a 168.677-bit (51-digit) factor: 598222550497242756375642930033091976261520001812697 (ECM,B1=110000000,B2=900514153782) #89 largest M4273 has a 173.500-bit (53-digit) factor: 16937472485366180815044482979818421310092932459932943 (ECM,B1=110000000,B2=900514153782) #77 largest M4339 has a 184.369-bit (56-digit) factor: 31656104144718898310053878501271041700569421852927688143 (ECM,B1=110000000,B2=900514153782) #56 largest M4397 has a 197.525-bit (60-digit) factor: 289041852087298746712043757073410343587428123771191294135823 (ECM,B1=110000000,B2=900514153782) #27 largest
edit: 5-10th ones a couple hours later. The 10th through 7th are now #27,56,77,89 on ECM biggest factors list.
And M4349 is now fully-factored.

Last fiddled with by James Heinrich on 2020-10-01 at 18:26

2020-10-01, 04:45   #1676
LaurV
Romulan Interpreter

Jun 2011
Thailand

7·1,279 Posts

Quote:
 Originally Posted by charybdis These were all done by NFS
Thanks, you saved me a lot of time to look for those.
Now the numbers in James' tables look more realistic (going down from 73 digits, and not from 150 ). That's more like "ECM range". Some of the anon results may still be NFS, but I won't bother. One silly idea would be to parse the PrimeNet DB for "sigma" info, as per George, the info is recorded on the server (but not shown by the "beautified" print routine), and I think all newer ECM result (like in the last yy years) should have stored b1/b2/sigma if they are trully ECM results. The lower side is not really interesting, so, eliminating TF/P-1 possible factors or letting them in, won't hurt much either way.

2020-10-01, 11:05   #1677
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

61668 Posts

Quote:
 Originally Posted by LaurV I think all newer ECM result (like in the last yy years) should have stored b1/b2/sigma if they are trully ECM results
If the server parsed and stored sigma values from the effort then it would also have correctly recorded the factoring method. It's the "ancient" results that are questionable because they were basically parsed for "Mx has a factor: y", ignoring all other data, and then guessed if that factor was TF/P-1/ECM by the number of bits. So there's no corroborating data to confirm/deny ECM'ness of a factor.

2020-10-01, 18:36   #1678
mathwiz

Mar 2019

12710 Posts

Quote:
 Originally Posted by James Heinrich Thanks, I have removed those from the data.
https://en.wikipedia.org/wiki/Intege...a_special_form says: "All unfactored parts of the numbers 2n − 1 with n between 1000 and 1200 were factored by a multiple-number-sieve approach in which much of the sieving step could be done simultaneously for multiple numbers, by a group including T. Kleinjung, J. Bos and A. K. Lenstra, starting in 2010."

So presumably we should exclude all $1000 <= n <= 1200$?

2020-10-01, 18:42   #1679
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

2·5·11·29 Posts

Quote:
 Originally Posted by mathwiz So presumably we should exclude all $1000 <= n <= 1200$?
Whether we should or not is beyond my knowing, I'll let others weigh in on that. The data involved:
Code:
| exponent | date_found          | factorbits | factor                                                                    |
+----------+---------------------+------------+---------------------------------------------------------------------------+
|     1013 | 2010-03-04 11:38:00 |    194.712 | 41120912566813018675472321435609728349473493582225344661873               |
|     1051 | 2010-08-08 10:46:00 |    207.069 | 215738012818441827932337543036174144558274385301234576636299249           |
|     1051 | 2013-08-09 11:24:00 |      227.5 | 305017906063256842921494808558019733856326299412534951989303214657199     |
|     1069 | 2013-08-02 15:16:00 |    231.687 | 5557036167944892502666285821951871600803581019193074182942021552512721    |
|     1087 | 2010-02-21 09:25:00 |     200.73 | 2664797814058212286560533454960446792210016180875809243599817             |
|     1163 | 2010-04-18 20:06:00 |    239.239 | 1042816042941845750042952206680089794415014668329850393031910483526456487 |
|     1181 | 2010-03-07 16:11:00 |    240.034 | 1808422353177349564546512035512530001279481259854248860454348989451026887 |
|     1187 | 2010-01-30 14:50:00 |    206.576 | 153327833285998453874202767942570343649971393640068204571694369           |

2020-10-01, 21:00   #1680
charybdis

Apr 2020

5·23 Posts

Quote:
 Originally Posted by mathwiz https://en.wikipedia.org/wiki/Intege...a_special_form says: "All unfactored parts of the numbers 2n − 1 with n between 1000 and 1200 were factored by a multiple-number-sieve approach in which much of the sieving step could be done simultaneously for multiple numbers, by a group including T. Kleinjung, J. Bos and A. K. Lenstra, starting in 2010." So presumably we should exclude all $1000 <= n <= 1200$?
The important word is "unfactored". Lots of these numbers had already been completely factored.

Quote:
 Originally Posted by James Heinrich Whether we should or not is beyond my knowing, I'll let others weigh in on that. The data involved: Code: | exponent | date_found | factorbits | factor | +----------+---------------------+------------+---------------------------------------------------------------------------+ | 1013 | 2010-03-04 11:38:00 | 194.712 | 41120912566813018675472321435609728349473493582225344661873 | | 1051 | 2010-08-08 10:46:00 | 207.069 | 215738012818441827932337543036174144558274385301234576636299249 | | 1051 | 2013-08-09 11:24:00 | 227.5 | 305017906063256842921494808558019733856326299412534951989303214657199 | | 1069 | 2013-08-02 15:16:00 | 231.687 | 5557036167944892502666285821951871600803581019193074182942021552512721 | | 1087 | 2010-02-21 09:25:00 | 200.73 | 2664797814058212286560533454960446792210016180875809243599817 | | 1163 | 2010-04-18 20:06:00 | 239.239 | 1042816042941845750042952206680089794415014668329850393031910483526456487 | | 1181 | 2010-03-07 16:11:00 | 240.034 | 1808422353177349564546512035512530001279481259854248860454348989451026887 | | 1187 | 2010-01-30 14:50:00 | 206.576 | 153327833285998453874202767942570343649971393640068204571694369 |
These are all genuine ECM factors. The ones from 2010 were found by Bos, Kleinjung et al, presumably in preparation for their SNFS factorizations. The other two, despite being listed as "ANONYMOUS", were in fact found by Ryan Propper.

2020-10-01, 21:09   #1681
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

1100011101102 Posts

Quote:
 Originally Posted by charybdis These are all genuine ECM factors. The ones from 2010 were found by Bos, Kleinjung et al, presumably in preparation for their SNFS factorizations. The other two, despite being listed as "ANONYMOUS", were in fact found by Ryan Propper.
I don't know if any of Bos,Kleinjung,etal have Primenet usernames. I have updated M1051, M1069 to belong to Ryan.

 2020-10-01, 23:28 #1682 firejuggler     Apr 2010 Over the rainbow 2·17·73 Posts M13113773 has a 70.491-bit (22-digit) factor: 1659317092853794607729 (P-1,B1=1000000) Interestingly, it should have been found by the earlier P-1 Last fiddled with by firejuggler on 2020-10-01 at 23:30
2020-10-01, 23:45   #1683
James Heinrich

"James Heinrich"
May 2004
ex-Northern Ontario

2·5·11·29 Posts

Quote:
 Originally Posted by firejuggler Interestingly, it should have been found by the earlier P-1
Unfortuantely that's not at all uncommon, due in no small part to a buggy P-1 implementation in early versions of Prime95.

 Similar Threads Thread Thread Starter Forum Replies Last Post ewmayer Programming 3 2016-09-30 07:15 jasong jasong 18 2013-11-15 22:54 Rafael Information & Answers 12 2012-01-02 19:38 rogue Lounge 10 2008-11-21 05:25 nngs Hardware 0 2005-05-20 01:31

All times are UTC. The time now is 15:39.

Tue Dec 1 15:39:43 UTC 2020 up 82 days, 12:50, 3 users, load averages: 1.77, 1.92, 1.86