Loading ...
Sorry, an error occurred while loading the content.
 

Procrels problem

Expand Messages
  • sashamkrt
    Last 2 runs of procrels shows strange info, i.e. GGNFS-0.77.1-20060513-pentium4 : procrels There were 29501/2637722 duplicates. RelProcTime: 1905.8
    Message 1 of 9 , Jun 1, 2006
      Last 2 runs of procrels shows strange info, i.e.
      GGNFS-0.77.1-20060513-pentium4 : procrels
      There were 29501/2637722 duplicates.
      RelProcTime: 1905.8
      largePrimes: 44863010 , relations: 39744321

      Before last run of procrels number of relations was 39737644
      After procrels it must be 39737644+2608221{2637722-29501}=42345865
      but it shows only 39744321.
    • Anton Korobeynikov
      Hello, sashamkrt. You wrote Thursday, June 1, 2006, 4:05:21 PM: s Before last run of procrels number of relations was 39737644 s After procrels it must be
      Message 2 of 9 , Jun 1, 2006
        Hello, sashamkrt.

        You wrote Thursday, June 1, 2006, 4:05:21 PM:

        s> Before last run of procrels number of relations was 39737644
        s> After procrels it must be 39737644+2608221{2637722-29501}=42345865
        s> but it shows only 39744321.
        Something went wrong with your input file. Have you changed factor base
        (e.g. large prime bounds) in the middle of your computation?
        You might try to remove #ifdef at procrels.c:949 and see, whether
        there are any error messages.

        --
        With best regards,
        Anton mailto:asl@...

        Thursday, June 1, 2006 4:22:18 PM

        Faculty of Mathematics & Mechanics, Saint-Petersburg State University
      • sashamkrt
        ... Why procrels did not accept relations when special-q ~500M? type: gnfs rlim: 18000000 alim: 18000000 lpbr: 29 lpba: 29 mfbr: 58 mfba: 58 rlambda: 2.8
        Message 3 of 9 , Jun 1, 2006
          > Before last run of procrels number of relations was 39737644
          > After procrels it must be 39737644+2608221{2637722-29501}=42345865
          > but it shows only 39744321.

          Why procrels did not accept relations when special-q >~500M?

          type: gnfs
          rlim: 18000000
          alim: 18000000
          lpbr: 29
          lpba: 29
          mfbr: 58
          mfba: 58
          rlambda: 2.8
          alambda: 2.8
        • sashamkrt
          ... base ... Thanks for info. Input file hasn t changed type: gnfs rlim: 18000000 alim: 18000000 lpbr: 29 lpba: 29 mfbr: 58 mfba: 58 rlambda: 2.8 alambda: 2.8
          Message 4 of 9 , Jun 1, 2006
            > Something went wrong with your input file. Have you changed factor
            base
            > (e.g. large prime bounds) in the middle of your computation?
            > You might try to remove #ifdef at procrels.c:949 and see, whether
            > there are any error messages.
            Thanks for info.
            Input file hasn't changed
            type: gnfs
            rlim: 18000000
            alim: 18000000
            lpbr: 29
            lpba: 29
            mfbr: 58
            mfba: 58
            rlambda: 2.8
            alambda: 2.8

            I think procrels stop accept relation when special-q become > ~500M.
          • Anton Korobeynikov
            Hello, sashamkrt. You wrote Thursday, June 1, 2006, 4:46:01 PM: s I think procrels stop accept relation when special-q become ~500M. Procrels don t know
            Message 5 of 9 , Jun 1, 2006
              Hello, sashamkrt.

              You wrote Thursday, June 1, 2006, 4:46:01 PM:

              s> I think procrels stop accept relation when special-q become > ~500M.
              Procrels don't know anything about special-q's :) It's a term only for
              lattice-siever. Maybe you've found another upper bound of ggnfs. Which
              side you're sieving? Could you show me siever output (single
              relation), which isn't accepted by ggnfs?


              --
              With best regards,
              Anton mailto:asl@...

              Thursday, June 1, 2006 5:06:57 PM

              Faculty of Mathematics & Mechanics, Saint-Petersburg State University
            • sashamkrt
              ... Already sieved special-q: [9M;570M). special-q: 520M-... accepted: - 123464148112,67575:1bc9dc9,15edf6a9,6823CF,9B6BD3,13,89,4CF,2:85bc371,12
              Message 6 of 9 , Jun 1, 2006
                > Maybe you've found another upper bound of ggnfs. Which
                > side you're sieving? Could you show me siever output (single
                > relation), which isn't accepted by ggnfs?

                Already sieved special-q: [9M;570M).
                special-q: 520M-...
                accepted:
                -
                123464148112,67575:1bc9dc9,15edf6a9,6823CF,9B6BD3,13,89,4CF,2:85bc371,12
                325799,F2BF,1CE8B,6F209,812E1,E8F90D,3,3,5,5,7,B,1D,C7,1EFE9209

                -
                69999591071,82896:22b15cd,21CE5,5F833,1DB427,3CE3C9,C5B,B,B:12e0a11,53f3
                bbb,281F,3CDF,3DF3,29861,13AF41,40EA3D,3,3,5,11,119,ACF,2,2,2,1EFE9209

                47480219377,137719:3278317,A5CB,1685A1,16F501,17BF,29,6D,6D,293:230f4d7,
                283397f,7B1D,13C25,24B7E5,6E0803,F53,3,3,3,3,3,5,5,7,D,17,43F,8AD,2,2,2,
                2,2,1EFE9215
                --------
                special-q: 550M-...
                not accepted:
                -
                99204295713,20830:6447137,110da201,4351D,607EE3,6B,161,2E7:2da7059,6641,
                AEC61,D7BB7,117031,2C5487,6B3EDF,1AA7,3,3,5,7,B,2,2,2,2,2,2,2,2,2,2,2,20
                C855E1

                -
                188307166573,157855:4733243,9330BF,1E89,3,6D,259,31D,5CB,79F,1115:16903b
                2b,818a1cd,EC5D,26C461,2DDB71,3913A3,5,7,B,11,17,49,97,125,6C5,2,2,2,20C
                855E1

                142996903174,55513:1b8e9823,d370399,BACB,81979F,B3C41D,2,2,2:581206f,1b9
                321d9,42CB,1D413,EF057,39E933,A44661,5,5,11,1543,3,3,3,B,20C855E7
              • Anton Korobeynikov
                Hello, sashamkrt. You wrote Thursday, June 1, 2006, 5:45:54 PM: s special-q: 550M-... s not accepted: Well. I don t see anything wrong with relations. Will
                Message 7 of 9 , Jun 1, 2006
                  Hello, sashamkrt.

                  You wrote Thursday, June 1, 2006, 5:45:54 PM:

                  s> special-q: 550M-...
                  s> not accepted:
                  Well. I don't see anything wrong with relations. Will you please
                  uncomment _DEBUG define in procrels.c and look for output.

                  --
                  With best regards,
                  Anton mailto:asl@...

                  Thursday, June 1, 2006 5:54:13 PM

                  Faculty of Mathematics & Mechanics, Saint-Petersburg State University
                • sashamkrt
                  ... procerelc accept relations when I m changing lpba from 29 to 30 in factor base file. Maybe this is a limit for special-q in the lattice siever: up to
                  Message 8 of 9 , Jun 1, 2006
                    > Well. I don't see anything wrong with relations. Will you please
                    > uncomment _DEBUG define in procrels.c and look for output.

                    procerelc accept relations when I'm changing lpba from 29 to 30 in
                    factor base file.
                    Maybe this is a limit for special-q in the lattice siever: up to 2^lpba.

                    What esle could I do to get more relations? Classical siever are very
                    slow. Maybe increase the factor base (i.e. from 18M to 20M and sieve in
                    [20M;200M) again)?
                  • Anton Korobeynikov
                    Hello, sashamkrt. You wrote Thursday, June 1, 2006, 8:52:23 PM: s procerelc accept relations when I m changing lpba from 29 to 30 in s factor base file. s
                    Message 9 of 9 , Jun 1, 2006
                      Hello, sashamkrt.

                      You wrote Thursday, June 1, 2006, 8:52:23 PM:

                      s> procerelc accept relations when I'm changing lpba from 29 to 30 in
                      s> factor base file.
                      s> Maybe this is a limit for special-q in the lattice siever: up to 2^lpba.
                      No. This is actually limit in the algorithm. You cannot get any more
                      relations. If you're running perl script, you should get a warning
                      here.

                      s> What esle could I do to get more relations? Classical siever are very
                      s> slow. Maybe increase the factor base (i.e. from 18M to 20M and sieve in
                      s> [20M;200M) again)?
                      Ah. This definitely means, that your parameters was selected bad. 500M
                      special-q's are TOO much. Really. However, nobody run GGNFS in this
                      area before - I expect much troubles.

                      --
                      With best regards,
                      Anton mailto:asl@...

                      Thursday, June 1, 2006 9:33:40 PM

                      Faculty of Mathematics & Mechanics, Saint-Petersburg State University
                    Your message has been successfully submitted and would be delivered to recipients shortly.