mersenneforum.org - Polynomial Request Thread

mersenneforum.org (https://www.mersenneforum.org/index.php)

- Msieve (https://www.mersenneforum.org/forumdisplay.php?f=83)

- - Polynomial Request Thread (https://www.mersenneforum.org/showthread.php?t=18368)

C207 degree 6 CADO polys

Polys with E score > 1.4e-15 that were not listed. The first 3 are unstable in Msieve (it's impossible to get them as Msieve output, even if you input them into Msieve from c207.dat.ms file).
Happy test-sieving!
[code]
Msieve v. 1.53 (SVN unknown)
R0: -4955388528020549250468494969439424
R1: 33855796097504351
A0: -9438740395892936178074542112765263906612538025
A1: 69081937815283108775212714134049651386840
A2: 12867402849083795230924828818574100
A3: -1562388250032818111515101018
A4: -1108427822881037175747
A5: 3644143983202
A6: 12408
skew 14314663.60, size 4.887e-015, alpha -10.254, combined = 1.494e-015 rroots = 4
[/code][code]
Msieve v. 1.53 (SVN unknown)
R0: -4955388528026089108239725703897905
R1: 33855796097504351
A0: 37755290257402036703113988168304087239169560832
A1: -26144711086392789872324688643568787516384
A2: 13456139157296262764671383831301872
A3: -835921001589942020498978930
A4: -1111404314119227362737
A5: 3631961982514
A6: 12408
skew 12043628.98, size 4.875e-015, alpha -10.070, combined = 1.491e-015 rroots = 4
[/code][code]
Msieve v. 1.53 (SVN unknown)
R0: -4955388528034045220322639226382905
R1: 33855796097504351
A0: -23795690899628716404015937931998467096638919168
A1: -13069214324890474716693225916877856996384
A2: 13676728859943426133084249922300672
A3: 210801584097515056973801070
A4: -1115661590971681312737
A5: 3614466702514
A6: 12408
skew 9578476.72, size 4.775e-015, alpha -9.795, combined = 1.469e-015 rroots = 4
[/code][code]
# norm 5.428445e-015 alpha -9.685498 e 1.429e-015 rroots 4
skew: 9771362.65
c0: 7979479956400423863806899180296394659767253495
c1: 10026826390769381753274758639294363746296
c2: 13690422536494438115003183516229084
c3: -60870677396142011674206026
c4: -1114560168912018979867
c5: 3619001106850
c6: 12408
Y0: -4955388528031983165349728528876548
Y1: 33855796097504351
[/code][code]
# norm 5.408138e-015 alpha -10.158255 e 1.422e-015 rroots 4
skew: 16285316.33
c0: -62478946305082331346005606885332447213581443040
c1: -99936963322568214315744550419428615413464
c2: 12834409141784358738327069928183484
c3: 1599473095993982056143452214
c4: -1121253275060862663117
c5: 3591358192210
c6: 12408
Y0: -4955388528044553991719712381924603
Y1: 33855796097504351
[/code][code]
# norm 5.361569e-015 alpha -9.783752 e 1.411e-015 rroots 4
skew: 11325983.55
c0: -58807985983425280615590443815032686489974937225
c1: -26170734171910895751413989153272469329440
c2: 12585633389326022026106327913878200
c3: 1817264709121685189671536142
c4: -1122124477109620883447
c5: 3587744411842
c6: 12408
Y0: -4955388528046197385918081340626494
Y1: 33855796097504351
[/code][code]
# norm 5.310949e-015 alpha -10.165134 e 1.403e-015 rroots 4
skew: 17834893.66
c0: 303089304998442991875512169289457909666560257280
c1: 95550124706188728625916664309702751034896
c2: 12720594838942721263678629505730224
c3: -1695548477210617324344902546
c4: -1107880286267872251617
c5: 3646380475570
c6: 12408
Y0: -4955388528019532188497929841231033
Y1: 33855796097504351
[/code]

Some test-sieving on deg 5 vs 6, run before Max posted the improved deg 6 polys:
[code]deg 6 coeff 12408 score 1.45e-15 tests: #msieve poly, not CADO
alim=rlim=800M, 16e/33LP, mfba/r 96/66 dQ=500, -a side (default)
Q=160M 1.67 sec/rel, 722 rels yield 1.44
Q=320M 1.81 sec/rel, 573 rels yield 1.15

alim=rlim=535M, otherwise same as above:
Q=160M 1.29 sec/rel 710 rels yield 1.42
Q=320M 1.65 sec/rel 562 rels yield 1.12
-----------------------------------------
deg 5 coeff 22320144 score 1.62e-15 tests:
alim=rlim=800M, 16e/33LP, mfba/r 96/66 dQ=500, -a side (default)
Q=160M 1.17 sec/rel, 905 rels yield 1.81
Q=320M 1.31 sec/rel, 691 rels yield 1.38
Q=480M 1.42 sec/rel, 726 rels yield 1.45
Q=640M 1.57 sec/rel, 735 rels yield 1.47
est 580MQ required, say 60M to 640M.
900M relations @ 1.35 sec/rel = 14000 thread-days

alim=rlim=535M, otherwise same as above:
Q=160M 1.08 sec/rel, 871 rels yield 1.74
Q=320M 1.26 sec/rel, 661 rels yield 1.32
Q=480M 1.43 sec/rel, 701 rels yield 1.40
Q=640M 1.51 sec/rel, 708 rels yield 1.42
est 610MQ required, say 60M to 670M.
900M relations @ 1.30 sec.rel = 13500 thread-days[/code]

Tests run on a 6-core haswell i7 at stock 3.3ghz, with 11 other threads running LLR or msieve-LA.

Notes: -r side is terrible for either degree. 3LP on both sides also bad.

CADO deg 6 highest-scoring:
[code]deg 6 coeff 12408 Max-CADO 1.494e-15:
alim=rlim=535M, 16e/33LP, mfba/r 66/96 dQ=500, -a side
Q=160M 1.24 sec/rel 655 rels yield 1.31
Q=320M 1.67 sec/rel 531 rels yield 1.06
Q=480M 1.93 sec/rel 501 rels yield 1.00[/code]

best i have is
[code]
R0: -2594969779164331152939346445221486
R1: 995648353094895065
A0: -7220800612857021441066164799889995859638469605
A1: -48706107644923334374031972029143808275744
A2: 18386892254939326170999356947890689
A3: 4246446412631814214268177691
A4: -1764058314391010739984
A5: -42416080899155
A6: 601692
skew 7316236.72, size 4.761e-015, alpha -10.108, combined = 1.487e-015 rroots = 4
[/code]
Found with msieve , so might get better with CADO

@firejuggler
Could you please send me your entire c207.dat.ms and c207.dat.p files (or portions of them with A6: 601692) and also your Msieve.exe file (the actual file, not the version number)? Zip/rar and drop them somewhere.
My Msieve has trouble recreating your poly, I need to see what was input/output.
I'll run CADO optimization in the evening.
Thank you in advance.

@firejuggler
Thank you so much. It's very helpful. I'll do my best in CADO.

please note that the msieve in the zip is the non-gpu version, but still produce the same result.

C207 -- 3 new CADO polys

Some new polys that CADO discovered (I will have to look more carefully later to find all the new ones).
Happy test-sieving!
[code]
# norm 5.988245e-015 alpha -10.074812 e 1.568e-015 rroots 6
skew: 6087156.55
c0: -27818207317903532217108003314322004809855567765
c1: -14008456245406788589866531861575554317624
c2: 21085361375389709400077567281343009
c3: 2269491655645290867255498651
c4: -1821830432093170512984
c5: -41420978601875
c6: 601692
Y0: -2594969778889890640892269569504886
Y1: 995648353094895065
[/code][code]
# norm 5.362182e-015 alpha -9.655138 e 1.434e-015 rroots 6
skew: 5818595.33
c0: -3878250778708132924744305406750581315380053641
c1: -12703948280251591233916246988901318901152
c2: 21284594865307341427610482920876563
c3: 2044751941204374966770800995
c4: -1828197809274592669254
c5: -41309836462403
c6: 601692
Y0: -2594969778859238610693890130033796
Y1: 995648353094895065
[/code][code]
# norm 5.315572e-015 alpha -9.634911 e 1.425e-015 rroots 6
skew: 5897594.44
c0: -7209794739338846715424087959841456110054407600
c1: -16287470923988211701593848986534771748020
c2: 21902084499014002897471419376969024
c3: 1072656988257140453485044589
c4: -1855295038939371103139
c5: -40833473295851
c6: 601692
Y0: -2594969778727861814854665631311981
Y1: 995648353094895065
[/code]

[QUOTE=firejuggler;456573]please note that the msieve in the zip is the non-gpu version, but still produce the same result.[/QUOTE]
Thank you. My problem was not Msieve. It was the absence of an entry point poly that can produce E score = 1.487e-015 by Msieve's -npr (root optimization). If you feed
[code]
R0: -2594969779164331152939346445221486
R1: 995648353094895065
A0: -7220800612857021441066164799889995859638469605
A1: -48706107644923334374031972029143808275744
A2: 18386892254939326170999356947890689
A3: 4246446412631814214268177691
A4: -1764058314391010739984
A5: -42416080899155
A6: 601692
skew 7316236.72, size 4.761e-015, alpha -10.108, combined = 1.487e-015 rroots = 4
[/code]into Msieve for -npr, the output will be lower than 1.487e-015. It means that the high E score poly is a side step that doesn't cycle back to itself in 200 standard steps, no matter which version of Msieve you use.
In practice, CADO usually can find a poly with higher E score than Msieve, given that a lot of entry point polys are provided at the beginning.
That's why I would rather have too many different Msieve entry points than not enough.
Alternatively, I can rerun Msieve polysearch from scratch for c6=601692 only, but, as far as I understand, my output can be different from yours because e.g. you could have set norms manually and I didn't, or because my GPU selected a different segment to run through than yours.

122 appears to be the smallest remaining value of n for the series n!+1 which is not fully factored. It may have been factored in the past, but it's not in factordb nor any [url=http://www.leyland.vispa.com/numth/factorization/factors/factorial+.txt]other list[/url] I could find. [url=http://www.uow.edu.au/~ajw01/ecm/]Andrew Walker's site[/url] with such a list appears to be gone, and Kamada's site does not seem to have it either.

Any interest in factoring this [url=http://www.factordb.com/index.php?showid=1100000000007808861]C186[/url]? Does anyone know if it has had enough ECM? I suspect it has survived a lot of ECM over the years but I have no data to support that belief.

[code]
274009141614480910126375195233419466792752179681960790486654456812768607704446242479114109752053246310214719268957764041441399297254302108847740343679614402988482108192973377638165257983
[/code]