Odd perfect related road blocks II

[chris2be8]

Number: m48_1229_58
N=156445373157917262062818893461378349313744615605072820462305628039469266808566071267141482109280452511431794672038610337311351315792145215121974123020286232070106460452222042095231   ( 180 digits)
SNFS difficulty: 185 digits.
Divisors found:
 r1=101607554056580376683209616614716919080982619801 (pp48)
 r2=1539702186618923356465786988796188728120453669008698397215644969019906236645572992446620399396865641702894694266278730211365745161431 (pp133)
Version: Msieve v. 1.49 (SVN 18sep2011)
Total time: 85.96 hours.
Scaled time: 233.73 units (timescale=2.719).
Factorization parameters were as follows:
#1229 58 156445373157917262062818893461378349313744615605072820462305628039469266808566071267141482109280452511431794672038610337311351315792145215121974123020286232070106460452222042095231 34178
# Record 48 from mostwant.txt
# sigma(1229^58) = 1 * 156445373157917262062818893461378349313744615605072820462305628039469266808566071267141482109280452511431794672038610337311351315792145215121974123020286232070106460452222042095231
# SNFS difficulty = 183  SNFS equivalent = 131  GNFS difficulty = 180
# cofactor =
n: 156445373157917262062818893461378349313744615605072820462305628039469266808566071267141482109280452511431794672038610337311351315792145215121974123020286232070106460452222042095231
type: snfs
name: m48_1229_58
m:  11874698581722514603062932731512446641
c5: 1
c4: 0
c3: 0
c2: 0
c1: 0
c0: -1229
Factor base limits: 8600000/8600000
Large primes per side: 3
Large prime bits: 28/28
Max factor residue bits: 55/55
Sieved rational special-q in [4300000, 11100001)
Relations: 20160429 relations
Pruned matrix : 1753131 x 1753357
Total sieving time: 81.09 hours.
Total relation processing time: 0.23 hours.
Matrix solve time: 4.56 hours.
Time per square root: 0.08 hours.
Prototype def-par.txt line would be:
snfs,185.000,5,0,0,0,0,0,0,0,0,8600000,8600000,28,28,55,55,2.5,2.5,100000
total time: 85.96 hours.
Started Mon Oct 24 20:37:24 2011, ended Wed Oct 26 18:40:46 2011, elapsed 1 days, 22:3:22
 --------- CPU info (if available) ----------
Memory: 2057992k/2096832k available (2050k kernel code, 38452k reserved, 1017k data, 316k init)
Calibrating delay using timer specific routine.. 5604.04 BogoMIPS (lpj=11208093)
Calibrating delay using timer specific routine.. 5600.37 BogoMIPS (lpj=11200740)

Does that count as an ECM miss?

Chris K

[wblipp]

Originally Posted by chris2be8

Does that count as an ECM miss?

Not for SNFS. The rule of thumb for SNFS is ECM to 2/9. 2/9 of 185 digits is 41 digits. I actually ran ECM to 45 digits. Finding a 48 digit factor would have been modestly lucky.

[chris2be8]

Originally Posted by wblipp

The following numbers from the 100 Most Wanted Roadblocks have had ECM to at least 2/9 of the SNFS difficulty, and are ready for SNFS factoring. Only the first one is large enough for RSALS.

163^89-1 C195 weight=163041
853^67-1 C194 weight=38851
1301^59-1 C181 weight=29908
1381^61-1 C189 weight=24140
1361^61-1 C189 weight=24014
1481^61-1 C191 weight=20179
1487^61-1 C191 weight=19585
1489^61-1 C191 weight=19439
2269^53-1 C175 weight=15620

Since no one else seems interested I'll take all except the first (RSALS is welcome to it). I've got a script to use several systems to factor a series of numbers that I need to test.

I'll probably do them in order of size so problems with the script won't lose too much work.

Chris K

[fivemack]

I'll take 163^89-1

[pinhodecarlos]

Originally Posted by chris2be8

I've got a script to use several systems to factor a series of numbers that I need to test.

Chris K

Can you share it?

[pinhodecarlos]

What's up with this square root phase? Even so will I get at the end the factors?
"Running 6781_59_minus1 for RSALS."

Mon Oct 31 05:56:48 2011  commencing square root phase
Mon Oct 31 05:56:48 2011  reading relations for dependency 1
Mon Oct 31 05:56:50 2011  read 3278726 cycles
Mon Oct 31 05:56:55 2011  cycles contain 8856148 unique relations
Mon Oct 31 06:00:54 2011  read 8856148 relations
Mon Oct 31 06:01:47 2011  multiplying 8856148 relations
Mon Oct 31 06:11:37 2011  multiply complete, coefficients have about 225.90 million bits
Mon Oct 31 06:11:40 2011  initial square root is modulo 127877671
Mon Oct 31 06:23:46 2011  GCD is 1, no factor found
Mon Oct 31 06:23:46 2011  reading relations for dependency 2
Mon Oct 31 06:23:47 2011  read 3280438 cycles
Mon Oct 31 06:23:53 2011  cycles contain 8860514 unique relations
Mon Oct 31 06:27:50 2011  read 8860514 relations
Mon Oct 31 06:28:43 2011  multiplying 8860514 relations
Mon Oct 31 06:38:33 2011  multiply complete, coefficients have about 226.02 million bits
Mon Oct 31 06:38:35 2011  initial square root is modulo 129136351
Mon Oct 31 06:50:44 2011  GCD is N, no factor found
Mon Oct 31 06:50:44 2011  reading relations for dependency 3
Mon Oct 31 06:51:48 2011  read 3282367 cycles
Mon Oct 31 06:51:53 2011  cycles contain 8864594 unique relations
Mon Oct 31 07:45:03 2011  read 8864594 relations
Mon Oct 31 07:45:58 2011  multiplying 8864594 relations
Mon Oct 31 07:55:46 2011  multiply complete, coefficients have about 226.12 million bits
Mon Oct 31 07:55:48 2011  initial square root is modulo 130235647
Mon Oct 31 08:08:15 2011  GCD is 1, no factor found
Mon Oct 31 08:08:15 2011  reading relations for dependency 4
Mon Oct 31 08:08:24 2011  read 3280111 cycles
Mon Oct 31 08:08:29 2011  cycles contain 8859168 unique relations
Mon Oct 31 08:12:29 2011  read 8859168 relations

[henryzz]

You should eventually get the factors. Occassional factorizations need over 10 sqrts so be patient. I don't think we have had more than 1-2 factorizations just run out of dependencies without a bug.

[debrouxl]

Yeah, this square root phase is proceeding normally, you're seeing messages added in recent versions of msieve

Yesterday evening (European time), on the XYYXF ML, Bob Backstrom reported that the square root phase found the complete factorization only on dependency 33, ~43 hours after the beginning of the square root phase (!!).

[pinhodecarlos]

Originally Posted by debrouxl

Yeah, this square root phase is proceeding normally, you're seeing messages added in recent versions of msieve

Yesterday evening (European time), on the XYYXF ML, Bob Backstrom reported that the square root phase found the complete factorization only on dependency 33, ~43 hours after the beginning of the square root phase (!!).

I didn't know about these new messages so I was getting worried. Anyway, post-processing is done.

[jasonp]

In the early days you would sometimes run into factorizations that would only find 3 dependencies, because the matrix had duplicate columns that destroyed most of the nullspace. At first I thought 33 failed dependencies is too many to attribute to chance, but to even get to the final GCD you need two different numbers whose squares mod the input are equal, and unless there's an error message then the square root must have worked correctly but been unlucky.

[pinhodecarlos]

Originally Posted by jasonp

In the early days you would sometimes run into factorizations that would only find 3 dependencies, because the matrix had duplicate columns that destroyed most of the nullspace. At first I thought 33 failed dependencies is too many to attribute to chance, but to even get to the final GCD you need two different numbers whose squares mod the input are equal, and unless there's an error message then the square root must have worked correctly but been unlucky.

At the time I was not thinking clearly because I wanted to start a second factorization in parallel while the first one was running the square root phase before going to work! I always did that to get advantage of all cores.