Polynomial Request Thread

[swellman]

Originally Posted by swellman

As to the proposed poly search, I will start on c202 2,2822M with CADO using the following:

P 4500000
admin 5e6
admax 6e6
adrange 840
incr 210
nq 15625
nrkeep 200
roeffort 60
soeffort 40

Any advice on my choice of poly search parameters welcome. This isn’t my first CADO search but I’m still struggling to find a balance between a thorough search and the speed to produce useful results faster than say a month for a c193 split between two machines. Other than to buy more/better machines of course.

Finally got this started yesterday, results expected late Sunday or Monday.

Is anyone else working on this poly search?

[Gimarel]

I'm on it, but it will take some time.

[swellman]

Got two polys out of CADO which I shamefacedly post here in the vain hope they can be spun upwards.

n: 2878977728559287815052575151319519593747737572669438174523671884800458948881238032864742745954457543352584602789170909263013709933513493218010306049288727816920075222096439894902906049990057855640036213
skew: 113638472.681
c0: -1076258529482989349678924135954854168678397306052
c1: -8131766920278814357742170744967839426631
c2: -731724322102606332691393883816814
c3: 10747261484637535888072439
c4: 60976377380058078
c5: -35994420
Y0: -863437652044104943500678714387082524062
Y1: 633325481744161748199683
# MurphyE (Bf=8.590e+09,Bg=4.295e+09,area=2.684e+16) = 4.641e-09
# f(x) = -35994420*x^5+60976377380058078*x^4+10747261484637535888072439*x^3-731724322102606332691393883816814*x^2-8131766920278814357742170744967839426631*x-1076258529482989349678924135954854168678397306052
# g(x) = 633325481744161748199683*x-863437652044104943500678714387082524062
# cownoise: 182087472.51279 2.92265817e-15

n: 2878977728559287815052575151319519593747737572669438174523671884800458948881238032864742745954457543352584602789170909263013709933513493218010306049288727816920075222096439894902906049990057855640036213
skew: 40306979.172
c0: 9354836733742272841540818538414728477182035797
c1: -504223254377930141727081071839880226633
c2: -41074003689424568986623413230827
c3: -46503943631410368946398
c4: 21107273014217411
c5: 5076330
Y0: -892759410134832746791100588114750634826
Y1: 8046610406777517872873239
# MurphyE (Bf=8.590e+09,Bg=4.295e+09,area=2.684e+16) = 4.662e-09
# f(x) = 5076330*x^5+21107273014217411*x^4-46503943631410368946398*x^3-41074003689424568986623413230827*x^2-504223254377930141727081071839880226633*x+9354836733742272841540818538414728477182035797
# g(x) = 8046610406777517872873239*x-892759410134832746791100588114750634826
# cownoise: 59955041.23606 2.90944824e-15

I am going to be on holiday for a bit so I must stop this poly search. Sorry I couldn't find better!

[Gimarel]

My current best, but my search is not finished.

# norm 1.038278e-19 alpha -8.614545 e 3.762e-15 rroots 5
skew: 99260735.36
c0: 705623580768249433029321195277854018509065721700
c1: 41406014162288721104716122831187983965235
c2: -518753744311153928017660090000412
c3: -17138344397586532214827361
c4: 72141077340095190
c5: 349272000
Y0: -439967813196311857499704300466274550538
Y1: 1951902326436078822209

[Gimarel]

Got a better poly.

# norm 1.121850e-19 alpha -6.665556 e 3.978e-15 rroots 5
skew: 112346767.83
c0: 50746711744685344924176387670904741515092442286
c1: 1044154952713500053950758366352183787663
c2: -65038740724359413257318356678622
c3: -598222092821290881886597
c4: 6298279902327162
c5: 5292000
Y0: -1017020465728756419365218426382865599528
Y1: 6517470457980264948383

If this is considered good enough, I'll continue with 2,2694M.

[swellman]

Originally Posted by Gimarel

Got a better poly.

# norm 1.121850e-19 alpha -6.665556 e 3.978e-15 rroots 5
skew: 112346767.83
c0: 50746711744685344924176387670904741515092442286
c1: 1044154952713500053950758366352183787663
c2: -65038740724359413257318356678622
c3: -598222092821290881886597
c4: 6298279902327162
c5: 5292000
Y0: -1017020465728756419365218426382865599528
Y1: 6517470457980264948383

If this is considered good enough, I'll continue with 2,2694M.

I say this poly is more than sufficient to proceed into GNFS. Anyone disagree?

[charybdis]

Poly looks good enough to me. As I suggested this number I'm happy to test-sieve it, as long as someone points me to the correct version of the 16e binary that's used by NFS@Home.

[swellman]

Originally Posted by charybdis

Poly looks https://www.mersenneforum.org/showpo...8&postcount=83good enough to me. As I suggested this number I'm happy to test-sieve it, as long as someone points me to the correct version of the 16e binary that's used by NFS@Home.

I believe this version (Windows) is close to what Greg uses but obviously is not the latest. Be sure to use the -d flag!

I’ll reach out to Greg about the latest Linux version.

[frmky]

Originally Posted by charybdis

Poly looks good enough to me. As I suggested this number I'm happy to test-sieve it, as long as someone points me to the correct version of the 16e binary that's used by NFS@Home.

It's a mostly standard lasieve 16e V5. You can compile it from https://github.com/gchilders/lasieve...e/tree/windows if you wish. The "windows" branch compiles fine in Linux and has the code to support septics and octics.

[charybdis]

Thanks guys! Managed to compile v5-16e, now test-sieving.

[swellman]

Originally Posted by charybdis

The Cunningham "Smaller-but-Needed" list currently consists of the following three composites, all from the recently added base-2 extension. They are the only remaining Cunningham composites below 211 digits.

202   2,2822M  2878977728559287815052575151319519593747737572669438174523671884800458948881238032864742745954457543352584602789170909263013709933513493218010306049288727816920075222096439894902906049990057855640036213
204   2,2694M  139509877596603485375566104232711132933505090002104860193106340622036083417378999487597616101577321202438219965668565622839544621355235792839908741220494879663844102714143032594102986041297412959583752893
204   2,2694L  577090071913159205243125819618792325643496686739633326605837693143901266730717227684238738802007560969316182854058876472912134267393693448001500880131336068115619176328037303470386467463341202235674056013

These would be good targets for the 16e-small queue over the next couple of months. Don't have spare cycles myself right now but I hope the usual suspects are willing to help find some polynomials. I'm assuming that GNFS will outperform the difficulty-271 octics for 2,2694L/M, though it might be a good idea for someone to make sure by test-sieving against another 204-digit GNFS (3,748+ springs to mind).

[@swellman: don't know how long the yoyo queue backlog is, but a few thousand curves at 850M on each number wouldn't do any harm]

Ryan has been hitting these with ECM (along with a set of HCNs and some XYYXF composites). Progress to date:

13950987759660348537556610423271113293350509000210...:
  B1=850e6: 16980 curves (21.51%)

28789777285592878150525751513195195937477375726694...:
  B1=850e6: 17424 curves (22.08%)

57709007191315920524312581961879232564349668673963...:
  B1=850e6: 17248 curves (21.85%)

Seeing that 287897772… is now scheduled for GNFS on 16e_small, it would seem prudent to cease all further ECM on that number.