Polynomial Request Thread - Page 79

wombatman · 2017-04-18, 13:05

I had planned on requesting the 16e queue, in large part due to the linear algebra aspect.

As for the trial sieving, I'm certainly amenable to trying out 34/35-bit LPs, though I'll have to figure out how to actually carry out the trial-sieving since I have essentially no experience with CADO in that regard.

VBCurtis · 2017-04-18, 16:41

Do you have the >33-bit enabled ggnfs sievers? Henryzz has sent them around a couple times; I have copies for windows-64 but not for linux. Some time in the past, Henry compiled me a 15e siever capable of 34LP, but that laptop died a while back and I've been lazy about locating the file.

I, too, have not yet tried CADO trial-sieving; this c207 is as good a reason as any for me to learn how; we should make a new thread for discussion of our trials with this number, methinks.

swellman · 2017-04-18, 16:54

Quote:

Originally Posted by VBCurtis

I'm interested in helping with it, with poly select and possibly LA.
At present my best box has 16GB, but I have a planned upgrade for a dual-xeon sandy bridge to either 24GB or 48GB; 24 should be enough.

Do you still have interest in the C186 cofactor of 122!+1? Firejuggler and I have found a few polys near the top of the range (as documented in the best poly thread). Not sure how much longer we need to run poly select, as we are moving into record territory at this point.

wombatman · 2017-04-18, 16:59

Quote:

Originally Posted by VBCurtis

Do you have the >33-bit enabled ggnfs sievers? Henryzz has sent them around a couple times; I have copies for windows-64 but not for linux. Some time in the past, Henry compiled me a 15e siever capable of 34LP, but that laptop died a while back and I've been lazy about locating the file.

I, too, have not yet tried CADO trial-sieving; this c207 is as good a reason as any for me to learn how; we should make a new thread for discussion of our trials with this number, methinks.

Truth be told, I don't know. I compiled the ones I have myself, so I would guess not? What's necessary to increase the LP limit?

VBCurtis · 2017-04-18, 19:38

Quote:

Originally Posted by swellman

Do you still have interest in the C186 cofactor of 122!+1? Firejuggler and I have found a few polys near the top of the range (as documented in the best poly thread). Not sure how much longer we need to run poly select, as we are moving into record territory at this point.

I've run my GPU for one day so far; I run root-opt about once a day, and my first day's hits only produced 3.10e-14 so I didn't report it. I'll give it until the weekend, unless you're happy with trial sieving the two good polys you two have found.

VBCurtis · 2017-04-18, 20:11

Quote:

Originally Posted by wombatman

Truth be told, I don't know. I compiled the ones I have myself, so I would guess not? What's necessary to increase the LP limit?

henryzz is the expert, but my recollection is that the LP bound is a simple line-item in the source, a check that merely needs to be disabled; however, the 96-bit mfbr/mfba limit is *not* so easily removed.

So, we can test 96/33 and 66/33 vs 96/33 and 68/34 vs 96/34 and 68/34 with ggnfs, and then larger mfbr/mfba with CADO to see if the benefit of larger parameters overcomes the slower siever; then again, if Greg accepts this number for 16e queue, CADO won't matter! Of course, neither will 34LP, since the regular 16e queue is 33LP-limited. Henryzz has also explored the 'f' siever, but I don't recall what it opens up for us other than composite special-q.

swellman · 2017-04-18, 20:30

I've been wondering lately if it would be worth starting a new sticky thread called "16e sieving candidates be here" (or something like that - you get the idea). It could serve as a sort of reserve pool of fully vetted potential factorization jobs that can't be run on 15e.

Rules such as it must be fully ECM'd, a good poly provided, and only a very limited number of candidates allowed at one time. Maybe a person could only get one such job on the list per year. Moderation could be a loathsome task though.

Greg could ignore the entire thread or perhaps take on a job if he starts running low on good 16e candidates.

Dunno, just thinking out loud. Greg probably has enough potential work to carry him into the 22nd century without such a thread!

Max0526 · 2017-04-18, 20:46

Omitted Msieve polys with E score > 3e-14. Please consider them for trial-sieving too.

Code:

# norm 3.189670e-018 alpha -7.395072 e 3.100e-014 rroots 5
skew: 384769629.38
c0: 63457246292092765169599270619240407583489369692
c1: 2481910669053098098298266623338206590010
c2: 7143712830295841466252750641276
c3: -66541412714783740631029
c4: -30533221621986
c5: 87576
Y0: -1256249753153425806728935665707366805
Y1: 239461046954332589

Code:

# norm 3.192917e-018 alpha -7.124867 e 3.086e-014 rroots 5
skew: 270873105.09
c0: 567827784825725995309502347857610609314723304
c1: 647614812227635173168820934436712784776
c2: 1608567348218501768986820239748
c3: -69210999928466565572773
c4: -18649348461906
c5: 87576
Y0: -1256249753146926937840689457422250431
Y1: 239461046954332589

Code:

# norm 3.314646e-018 alpha -7.428610 e 3.053e-014 rroots 5
skew: 528073658.25
c0: -179912512721188926517787544022254706226761486641
c1: 7207638637239199738459870224201505477901
c2: 31544793065345878774121719372883
c3: -20039812404175720871533
c4: -105419817845706
c5: 87576
Y0: -1256249753194378631097663080676537196
Y1: 239461046954332589

Code:

# norm 3.117142e-018 alpha -7.123827 e 3.045e-014 rroots 5
skew: 443683644.30
c0: -209462565500456328030157808882360949652670777442
c1: 506327284748823705289224413228237301449
c2: 11108042236661667312539929015612
c3: -9105667184810192452426
c4: -72216986893688
c5: 12144
Y0: -1865007827235172946991740797177893475
Y1: 390398004681926561

Code:

# norm 3.106528e-018 alpha -7.365982 e 3.030e-014 rroots 3
skew: 338817573.51
c0: -13120848972011797290000158921858441386702083245
c1: -218062070751737125319214847188350637155
c2: 3184359143352621299966211423071
c3: -68591498058627994053901
c4: -21987085696746
c5: 87576
Y0: -1256249753148752227994871528901574808
Y1: 239461046954332589

Code:

# norm 3.069442e-018 alpha -7.073796 e 3.017e-014 rroots 5
skew: 431674254.65
c0: -149473956464465948030892803556739648169066769657
c1: 472913475353882086189839397988031853118
c2: 10899993398554596774465764556541
c3: -11084198477590689145866
c4: -71799895445288
c5: 12144
Y0: -1865007827232491266009770198847361180
Y1: 390398004681926561

Code:

# norm 3.033319e-018 alpha -7.400606 e 3.009e-014 rroots 5
skew: 459393172.72
c0: -61110513473054279510605545472813649230060260933
c1: 4650778249784287066549484541202146006965
c2: 5633387210825279522479882485071
c3: -67409876872274214336589
c4: -27242174655786
c5: 87576
Y0: -1256249753151626049827268240803342320
Y1: 239461046954332589

kurtb · 2017-04-19, 07:54

@vbcurtis/swellman

There is a number that has been msieve-LA-solved with boundary 34 without problems?
And necessary relations: I guess 1.45e9...
Currently our team works with 2 repunit-numbers (GNFS/c197 and c199) with 16e and 33/33 and 66/96
Kurt

firejuggler · 2017-04-20, 12:50

Here is a dozen poly that require CADO optimisation, if any of you can. (score above 3.1e-14)

Max0526 · 2017-04-20, 23:05

Look what Msieve (not CADO) found:

Code:

# norm 3.632953e-018 alpha -7.386782 e 3.363e-014 rroots 5
skew: 179152242.69
c0: -7788531738652397448489317133243454017698208000
c1: 145195916705848862555330399829833045340
c2: 7198906960646480927043798547919
c3: 10342353139172479900710
c4: -235669009706064
c5: 154980
Y0: -1120721953097761668868808682670911997
Y1: 39537117065588599

Code:

# norm 3.497360e-018 alpha -7.287086 e 3.292e-014 rroots 5
skew: 179370177.12
c0: -7034233375759405175224008951632097469963220455
c1: 184741509272840299404516248709208748942
c2: 6143926300830857319382299355155
c3: -26296781257536394092786
c4: -203332362190164
c5: 154980
Y0: -1120721953096111781375784017466079488
Y1: 39537117065588599

2017-04-18, 20:30	#865
swellman Jun 2012 2²×13×59 Posts	16e I've been wondering lately if it would be worth starting a new sticky thread called "16e sieving candidates be here" (or something like that - you get the idea). It could serve as a sort of reserve pool of fully vetted potential factorization jobs that can't be run on 15e. Rules such as it must be fully ECM'd, a good poly provided, and only a very limited number of candidates allowed at one time. Maybe a person could only get one such job on the list per year. Moderation could be a loathsome task though. Greg could ignore the entire thread or perhaps take on a job if he starts running low on good 16e candidates. Dunno, just thinking out loud. Greg probably has enough potential work to carry him into the 22nd century without such a thread! Last fiddled with by swellman on 2017-04-18 at 20:31

2017-04-19, 07:54	#867
kurtb "Beschorner Kurt" Jul 2016 Germany 2²·5 Posts	LPB = 34 @vbcurtis/swellman There is a number that has been msieve-LA-solved with boundary 34 without problems? And necessary relations: I guess 1.45e9... Currently our team works with 2 repunit-numbers (GNFS/c197 and c199) with 16e and 33/33 and 66/96 Kurt

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GIMPS wiki account request thread	ixfd64	mersennewiki	169	2018-09-21 05:43
Polynomial Discriminant is n^k for an n-1 degree polynomial	carpetpool	Miscellaneous Math	14	2017-02-18 19:46
Lost Prime Raider password request thread	cheesehead	Forum Feedback	6	2009-07-28 13:02
Polynomial	R.D. Silverman	NFSNET Discussion	13	2005-09-16 20:07
Deutscher Thread (german thread)	TauCeti	NFSNET Discussion	0	2003-12-11 22:12

2017-04-18, 13:05	#859
wombatman I moo ablest echo power! May 2013 29×61 Posts	I had planned on requesting the 16e queue, in large part due to the linear algebra aspect. As for the trial sieving, I'm certainly amenable to trying out 34/35-bit LPs, though I'll have to figure out how to actually carry out the trial-sieving since I have essentially no experience with CADO in that regard.

2017-04-18, 16:41	#860
VBCurtis "Curtis" Feb 2005 Riverside, CA 4,861 Posts	Do you have the >33-bit enabled ggnfs sievers? Henryzz has sent them around a couple times; I have copies for windows-64 but not for linux. Some time in the past, Henry compiled me a 15e siever capable of 34LP, but that laptop died a while back and I've been lazy about locating the file. I, too, have not yet tried CADO trial-sieving; this c207 is as good a reason as any for me to learn how; we should make a new thread for discussion of our trials with this number, methinks.