Polynomial Request Thread - Page 109

swellman · 2018-04-14, 22:56

A few more with scores a bit worse but hoping at least one can be improved by Max.

Code:

# norm 9.981944e-20 alpha -7.648241 e 3.583e-15 rroots 5
skew: 9656661903.65
c0: 15814593842842758666131181362060227944236903185423215
c1: 92421141918994112071166249062419715541540850
c2: -8700038074230934293283812063185242
c3: -1810279238334858887062394
c4: 87253536396999
c5: 9516
Y0: -1375191354064811838431965872985306732352
Y1: 3156801492645074161


# norm 9.932519e-20 alpha -7.770819 e 3.574e-15 rroots 3
skew: 9949495045.12
c0: 257331074317305990444507785047542611249814764807317776
c1: 72276970282733565226204049235335558244362268
c2: -4881726693420490277339189013487330
c3: -2000412876400873445880614
c4: 55586872228899
c5: 9516
Y0: -1375191354066912834075365376256116019247
Y1: 3156801492645074161


# norm 9.947003e-20 alpha -7.556915 e 3.572e-15 rroots 5
skew: 8882002650.71
c0: 115588237217384955386355049156955042323460036840603931
c1: 74265911089535986784955953917986734855404318
c2: 3472115250429972754111117643691950
c3: -2127776985869815079577554
c4: -7740038754801
c5: 9516
Y0: -1375191354071114399588563062293357128162
Y1: 3156801492645074161


# norm 9.932519e-20 alpha -7.770819 e 3.574e-15 rroots 3
skew: 9949495045.12
c0: 257331074317305990444507785047542611249814764807317776
c1: 72276970282733565226204049235335558244362268
c2: -4881726693420490277339189013487330
c3: -2000412876400873445880614
c4: 55586872228899
c5: 9516
Y0: -1375191354066912834075365376256116019247
Y1: 3156801492645074161

Max - Apologies if these are all just variations on the c=9516 theme that don't help.

VBCurtis · 2018-04-14, 23:36

First msieve hit from C200:

Code:

# norm 1.060548e-19 alpha -8.306292 e 3.677e-15 rroots 5
skew: 697608847.11
c0: -40875461427637243387538732304230400612944885731471
c1: 337351765525998980707353108823320558700912
c2: 4854957201609301971262498131868888
c3: -12505209711201145384822768
c4: -10472604823577361
c5: 1009800
Y0: -541012060454628767650971980401131480162
Y1: 18838355335163598623

First from CADO:

Code:

n: 46802612901958501868841642965751124516436202272587430374287737787818526612306217484595957815607978887434356$
skew: 515294842.026
c0: -18767604301794724568739522102957989342449153502601
c1: 1187430458137265550443270176264298191065767
c2: 252401375725248456245856214549773
c3: -6392702231042674615691663
c4: 1708421116152924
c5: 5027400
Y0: -609023179646324520044401061467999841338
Y1: 232970125802380714109627
# MurphyE (Bf=1.00e+08,Bg=1.30e+08,area=5.37e+16) = 2.15e-12
# f(x) = 5027400*x^5+1708421116152924*x^4-6392702231042674615691663*x^3+252401375725248456245856214549773*x^2+$
# g(x) = 232970125802380714109627*x-609023179646324520044401061467999841338

I didn't run the CADO poly through msieve to see its "real" score.

VBCurtis · 2018-04-14, 23:41

Quote:

Originally Posted by Max0526

@VBCurtis
Did you have a chance to test-sieve C197 polys?
Which one is the winner?

#1 and #3 (7.87 vs 7.57) are very close. Still testing additional Q values to get a more complete picture of the yield/timing curves. The 7.87 is faster, but the 7.57 yields 25% better; avoiding the higher Q values may make up the speed.

I'm also still unsure of best alim/rlim values, and the relative performance of the two polys changes with changes in lim's. My most-detailed effort is at alim=268M, rlim=400M. I'll try 268/268 before I post to the queue-thread.

Params are 33/66/2.7 rational, 33/95/3.7 algebraic.

swellman · 2018-04-14, 23:56

@VBCurtis - you’re much more expert in test sieving than I, but in my experience I’d take a 25% increase in yield over a bit more speed all day long. Partly because speed is almost irrelevant in BOINC, but a poor yield can make sieving hard or even impossible if Q gets very large.

Just my $0.02.

VBCurtis · 2018-04-15, 00:59

That's where I lean as well for such a large job. Similar logic leads to giving up 10% of sieve speed to reduce lim's by 30%, as smaller lim's lead to smaller matrices (all else equal). Smaller lim's also mean less memory use; 400/536M made 15e use nearly 1000MB memory. Those large settings should work for the upcoming C200.

Yield with alim=400M, rlim=536M is very close to 2.0. With 268/400M, we're looking at yield of 1.8 and Q=50M-600M sieve range, entirely reasonable. That should produce 900M+ raw relations.

Max0526 · 2018-04-15, 12:55

Quote:

Originally Posted by swellman

Code:

R0: -1375191354069134045183322114551111049796
R1: 3156801492645074161
A0: -338528721008222025614235066235791150075370103170041
A1: 1866015770516966978978053260633526834790530
A2: -527119001454503782276190397930010
A3: -2109749727333229354016618
A4: 22108294382679
A5: 9516
skew 3340598530.89, size 8.520e-20, alpha -7.008, combined = 3.630e-15 rroots = 5

Better score than my last, though skew is still ridiculous . Hoping we can reach a 3.9+.

After spinning:

Code:

Y0: -2750382708133725334613417581253101468804
Y1: 3156801492645074161
A0: -332672675114694145181372303273491852806460469241208920
A1: 340888787128559157309935768355107525561775726
A2: -19908046545539943397396400874113167
A3: -3993710753834215715892388
A4: 56342996697999
A5: 4758
skew: 15629035780.85
# size 8.338e-20, alpha -9.049, combined = 3.704e-15 rroots = 5

Max0526 · 2018-04-15, 14:43

Quote:

Originally Posted by VBCurtis

First msieve hit from C200:

Code:

# norm 1.060548e-19 alpha -8.306292 e 3.677e-15 rroots 5
skew: 697608847.11
c0: -40875461427637243387538732304230400612944885731471
c1: 337351765525998980707353108823320558700912
c2: 4854957201609301971262498131868888
c3: -12505209711201145384822768
c4: -10472604823577361
c5: 1009800
Y0: -541012060454628767650971980401131480162
Y1: 18838355335163598623

First from CADO:

Code:

n: 46802612901958501868841642965751124516436202272587430374287737787818526612306217484595957815607978887434356$
skew: 515294842.026
c0: -18767604301794724568739522102957989342449153502601
c1: 1187430458137265550443270176264298191065767
c2: 252401375725248456245856214549773
c3: -6392702231042674615691663
c4: 1708421116152924
c5: 5027400
Y0: -609023179646324520044401061467999841338
Y1: 232970125802380714109627
# MurphyE (Bf=1.00e+08,Bg=1.30e+08,area=5.37e+16) = 2.15e-12
# f(x) = 5027400*x^5+1708421116152924*x^4-6392702231042674615691663*x^3+252401375725248456245856214549773*x^2+$
# g(x) = 232970125802380714109627*x-609023179646324520044401061467999841338

I didn't run the CADO poly through msieve to see its "real" score.

I can't spin the first poly. The second one with correct score

Code:

skew: 669955760.84960
c0: -18767604301794724568739522102957989342449153502601
c1: 1187430458137265550443270176264298191065767
c2: 252401375725248456245856214549773
c3: -6392702231042674615691663
c4: 1708421116152924
c5: 5027400
Y0: -609023179646324520044401061467999841338
Y1: 232970125802380714109627
# MurphyE = 3.79221133e-15

spins into

Code:

Y0: -609023141860534518999014907067872816202
Y1: 232970125802380714109627
c0: 6597855742130600571092019979486718189877646469415
c1: 811355857686769572713280062435647502578631
c2: -2373973916391152933433647573915523
c3: -3961823123299030107336335
c4: 5785430560968924
c5: 5027400
skew: 615255866.07
# size 9.582e-20, alpha -8.316, combined = 3.999e-15 rroots = 5

So close to a 4-handle!

RichD · 2018-04-15, 15:20

Another small update.

Code:

N: 46802612901958501868841642965751124516436202272587430374287737787818526612306217484595957815607978887434356766596560265669642422106894901815250576734687838908431237216708427946009827065503443998015551
R0: -408314216303435100435458262540789233919
R1: 24460427873149114483
A0: 37468162331679089171608618363581985765182102490960
A1: -375757402625087525587441184814759871150764
A2: -2914294123964357660509801097133188
A3: 11370155166060367158990339
A4: 11371636548964438
A5: 4123800
skew 469722874.58, size 1.011e-19, alpha -8.739, combined = 4.088e-15 rroots = 3

VBCurtis · 2018-04-15, 18:49

Quote:

Originally Posted by Max0526

@VBCurtis
Did you have a chance to test-sieve C197 polys?
Which one is the winner?

Test-sieve results at http://mersenneforum.org/showpost.ph...postcount=1400

RichD · 2018-04-15, 19:07

4.0-4.25M yields nothing of importance. Not even close to the lower end of the expected range.

VBCurtis · 2018-04-15, 20:10

Two more from CADO for the C200:

Code:

n: 468026129019585018688416429657511245164362022725874303742877377878185266123062174845$
Y0: -563511441158041414351093946606736994835
Y1: 29915033353895765503452659
c0: 1021235896704812157082357616365625321869130166816
c1: -64007732066564918703662059235131605120594
c2: 310517199851716872059230176613607
c3: 14734054907549261050441789
c4: 23085508512199572
c5: -252046080
skew: 100405224.736
# # lognorm 64.56, E 55.74, alpha -8.83 (proj -2.60), 5 real roots
# # MurphyE(Bf=1.00e+08,Bg=1.30e+08,area=5.37e+16)=2.27e-12

n: 46802612901958501868841642965751124516436202272587430374287737787818526612306217484595$
skew: 146521018.669
c0: -1872442660276739919980290111336570069874058707196
c1: -102616412513426017372250882501958944741603
c2: 619869176564077326486899317186512
c3: 11587385283031159867231217
c4: -19570630283788782
c5: -57126960
Y0: -557949089283439773749209831628722353318
Y1: 210639254161220778464543
# MurphyE (Bf=1.00e+08,Bg=1.30e+08,area=5.37e+16) = 2.3e-12
# f(x) = -57126960*x^5-19570630283788782*x^4+11587385283031159867231217*x^3+6198691765640$
# g(x) = 210639254161220778464543*x-557949089283439773749209831628722353318

These score 6-7% better than the previous I posted, which converted to msieve score as 3.79; hopefully that means they're right around 4.0 before polishing!

2018-04-15, 19:07	#1198
RichD Sep 2008 Kansas 3,389 Posts	C205 poly 4.0-4.25M yields nothing of importance. Not even close to the lower end of the expected range.

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

2018-04-14, 23:56	#1192
swellman Jun 2012 110000010011₂ Posts	@VBCurtis - you’re much more expert in test sieving than I, but in my experience I’d take a 25% increase in yield over a bit more speed all day long. Partly because speed is almost irrelevant in BOINC, but a poor yield can make sieving hard or even impossible if Q gets very large. Just my $0.02.

2018-04-15, 00:59	#1193
VBCurtis "Curtis" Feb 2005 Riverside, CA 1001100001001₂ Posts	That's where I lean as well for such a large job. Similar logic leads to giving up 10% of sieve speed to reduce lim's by 30%, as smaller lim's lead to smaller matrices (all else equal). Smaller lim's also mean less memory use; 400/536M made 15e use nearly 1000MB memory. Those large settings should work for the upcoming C200. Yield with alim=400M, rlim=536M is very close to 2.0. With 268/400M, we're looking at yield of 1.8 and Q=50M-600M sieve range, entirely reasonable. That should produce 900M+ raw relations.