Aliqueit Failing to Find Msieve Factors

[EdH]

Hopefully it's not too early to claim success, but I copied the following section of code in factmsieve.py:

      m = re.search('prp', l)
      if m:
        tmp = int(re.sub('.*factor: ', '', l))
        if 1 < len(str(tmp)) < fact_p['dec_digits']:
          fact_p['divisors'].append(tmp)

and pasted it just below, with the highlighted modification:

      m = re.search(' factor: ', l)
      if m:
        tmp = int(re.sub('.*factor: ', '', l))
        if 1 < len(str(tmp)) < fact_p['dec_digits']:
          fact_p['divisors'].append(tmp)

This appears to be working.

[chris2be8]

I think you will have problems with a 3 (or more) way split.

I've dug up some output from various versions of msieve with more that 2 factors:

Thu Aug 25 17:44:55 2016  Msieve v. 1.52 (SVN 956)
Thu Aug 25 17:45:00 2016  found factor: 709721709761207119267114609162209121
Thu Aug 25 17:45:04 2016  found factor: 709721709761207119267114609162209121
Thu Aug 25 17:45:08 2016  prp27 factor: 694252367117125307114583259
Thu Aug 25 17:45:08 2016  prp36 factor: 202952563227217748487822304015976827
Thu Aug 25 17:45:08 2016  prp36 factor: 709721709761207119267114609162209121

Sun Jun  5 06:36:22 2016  Msieve v. 1.53 (SVN 988M)
Sun Jun  5 06:43:25 2016  found factor: 1918765498657738825183679613663123166289011628683
Sun Jun  5 06:45:46 2016  p49 factor: 1918765498657738825183679613663123166289011628683
Sun Jun  5 06:45:46 2016  p68 factor: 14200533162507666436434648509544081144056643030084489018590680193131
Sun Jun  5 06:45:46 2016  p70 factor: 2955008600658532912101064869317444928677769566075973337951057995503471

That's output from grep'ing for "factor:" in the log, + the msieve version.

Your fix will probably match the "found factor:" lines as well as the "pnn factor:" lines (if I understand the Python correctly). Note you only get the "found factor:" lines in a 3 or more way split.

That's why my fix for looks for "pnn factor:". But I don't know how to code that in Python.

Chris

[EdH]

Originally Posted by chris2be8

I think you will have problems with a 3 (or more) way split.

I've dug up some output from various versions of msieve with more that 2 factors:

Thu Aug 25 17:44:55 2016  Msieve v. 1.52 (SVN 956)
Thu Aug 25 17:45:00 2016  found factor: 709721709761207119267114609162209121
Thu Aug 25 17:45:04 2016  found factor: 709721709761207119267114609162209121
Thu Aug 25 17:45:08 2016  prp27 factor: 694252367117125307114583259
Thu Aug 25 17:45:08 2016  prp36 factor: 202952563227217748487822304015976827
Thu Aug 25 17:45:08 2016  prp36 factor: 709721709761207119267114609162209121

Sun Jun  5 06:36:22 2016  Msieve v. 1.53 (SVN 988M)
Sun Jun  5 06:43:25 2016  found factor: 1918765498657738825183679613663123166289011628683
Sun Jun  5 06:45:46 2016  p49 factor: 1918765498657738825183679613663123166289011628683
Sun Jun  5 06:45:46 2016  p68 factor: 14200533162507666436434648509544081144056643030084489018590680193131
Sun Jun  5 06:45:46 2016  p70 factor: 2955008600658532912101064869317444928677769566075973337951057995503471

That's output from grep'ing for "factor:" in the log, + the msieve version.

Your fix will probably match the "found factor:" lines as well as the "pnn factor:" lines (if I understand the Python correctly). Note you only get the "found factor:" lines in a 3 or more way split.

That's why my fix for looks for "pnn factor:". But I don't know how to code that in Python.

Chris

I don't think it will cause a problem in Aliqueit, since it verifies each factor. It might throw a, "WARNING: factor doesn't divide n" message, but we'll have to see. I started a run with your first set on one of the machines. I was going to test the second set, too, but 185 digits might be a bit many for my current machines to work with. I could probably build some more test cases without any difficulty, but I'll see what your first set does first.

As to programming it to search, instead of looking for pnn factor:, I could simply look for "factor:" located at address 16 past the first colon. I'll play with that if my running test fails, or, maybe even if it doesn't...

[EdH]

Looks like it ran fine. The ggnfs.log shows only the three factors, with no duplication:

Number: test
N = 100000000000000000000000040000000000000000000008474000000000000000000001694000000000000000000179353
factor: 694252367117125307114583259
factor: 202952563227217748487822304015976827
factor: 709721709761207119267114609162209121

The test.log (msieve run):

...
Mon Sep  5 16:50:11 2016  found factor: 709721709761207119267114609162209121
7 lines
Mon Sep  5 16:50:35 2016  found factor: 709721709761207119267114609162209121
15 lines
Mon Sep  5 16:51:23 2016  found factor: 709721709761207119267114609162209121
8 lines
Mon Sep  5 16:51:47 2016  p27 factor: 694252367117125307114583259
Mon Sep  5 16:51:47 2016  p36 factor: 202952563227217748487822304015976827
Mon Sep  5 16:51:47 2016  p36 factor: 709721709761207119267114609162209121

Something in the back of my mind says that the python script takes care of duplicates, but a quick scan didn't turn up anything.

Anyway, Aliqueit never saw any duplicates, so everything ran fine.

Thanks for the help.