strange factors distribution??

pegaso56 · 2009-03-14, 13:31

Hi

, I was playing a bit with top 100 trial factors producers, and i found something really curious -at least for me-. Relation between exponents tested and successes is 98.60% for Will Eddington, fall to around 36.74% and 29,60% for the next five , and then well...

Has this something to do with range selected to try?
Regards, Carlos

Ghz-dia Intentos Hallados Indice
97 Will 965005 671540 662166 0,9860
31 David 3753200 11525604 4233988 0,3674
88 ATH 1112196 5959340 2173086 0,3647
79 S207114 1266338 1330276 468615 0,3523
72 Team_Bundu 1495327 1495252 504953 0,3377
33 arnaud 3533488 84918 25132 0,2960
2 Sturle 40530447 569926 81821 0,1436
21 monst 4658748 4225938 556604 0,1317
51 George 2212120 114585 11792 0,1029
37 James 3020591 194939 17440 0,0895
15 GrunwalderGIMP 6745535 322462 24777 0,0768
83 KriZp 1184503 1247286 72736 0,0583
64 Team_Inspector 1779833 31011 1518 0,0490
1 ANONYMOUS 127754299 1873574 81309 0,0434
96 Jan 967307 73710 3150 0,0427
99 Oleg 914283 6378 233 0,0365
29 ScottBardwick 4157571 178980 6466 0,0361
22 mcurtis 4615751 1112185 37344 0,0336
68 KYOJI_KAMEI 1657042 32798 1073 0,0327
35 sasaki 3103553 442947 14382 0,0325
82 carlos56 1201473 428949 13818 0,0322
47 Carsten 2444434 155029 4968 0,0320
100 S530729 885316 99496 3026 0,0304
87 andreasbomke 1115926 133629 4016 0,0301
59 oodaira 1933116 216136 6276 0,0290
46 Hebridean_Boy 2468834 18023 480 0,0266
76 Xyzzy 1319354 260524 6644 0,0255
85 BJB 1164195 99243 2520 0,0254
19 Architects 4995780 1499917 37660 0,0251
34 Derek 3198359 541630 13342 0,0246
66 Smok_bmv 1749894 26649 641 0,0241
55 HBendtz 2057308 51566 1179 0,0229
25 Wojciech 4348467 171800 3923 0,0228
20 ComputerraRU 4983458 79197 1795 0,0227
52 glennpat 2085311 23752 458 0,0193
67 Geoff 1735707 9179 171 0,0186
36 Luigi 3038628 39890 734 0,0184
40 WileECoyote 2752807 28986 532 0,0184
42 bayanne 2583312 38833 709 0,0183
71 linded 1567729 19417 341 0,0176
94 fjansson 986008 5225 91 0,0174
28 James 4249174 36028 623 0,0173
84 kevshaw 1165789 14524 250 0,0172
26 EPF 4333358 30522 525 0,0172
23 ANONYMOUS 4526281 19340 331 0,0171
60 kutens 1921113 14554 249 0,0171
86 jkolonko 1125964 9582 163 0,0170
43 Team_Deutschland 2542364 37107 631 0,0170
14 RMAC9.5 6882486 40655 690 0,0170
11 binarydigits 7873968 27877 471 0,0169
5 ITS 14590588 100373 1695 0,0169
77 Helgen 1282853 11969 202 0,0169
10 tsc 10803198 99273 1671 0,0168
38 sannerud.com 2920937 20687 348 0,0168
4 Team 17666007 191038 3206 0,0168
8 JC_Herles 12898613 96071 1610 0,0168
18 dlsilver 5259964 38433 643 0,0167
58 Guido 2000990 21235 355 0,0167
30 PrimeCruncher 3884430 22912 383 0,0167
44 subgenius 2542188 48124 804 0,0167
70 cmarble 1631063 18877 315 0,0167
6 TheJudger 14312103 143825 2400 0,0167
75 Robert 1363845 27458 458 0,0167
90 greensinoz 1074652 13617 227 0,0167
9 DSheets 10829832 136896 2282 0,0167
41 ANONYMOUS 2645770 59220 987 0,0167
24 smintheus 4392263 36602 610 0,0167
54 Net_Force 2061351 433485 7224 0,0167
62 outlnder 1789892 22626 377 0,0167
39 Matt 2760846 27731 462 0,0167
98 Piepen 927592 10805 180 0,0167
45 iPrime 2514811 20231 337 0,0167
48 delong 2438808 57232 953 0,0167
17 curtisc 6140101 93272 1553 0,0167
3 ahmerali 21673638 58079 967 0,0166
93 shy24 988534 12496 208 0,0166
89 Stephan 1075635 21346 355 0,0166
57 Texaport 2034219 18459 306 0,0166
61 S00030 1823924 15185 251 0,0165
53 WorknPlay 2066741 17010 281 0,0165
95 Dk 973290 6561 108 0,0165
7 tmorrow 13075722 73214 1201 0,0164
49 David 2367826 25802 423 0,0164
16 korva 6479371 70330 1148 0,0163
65 raymondf 1752290 14401 235 0,0163
12 Marzbug 7466877 49252 796 0,0162
69 Gordon 1650954 13156 211 0,0160
13 precius1 7061882 39237 628 0,0160
27 S36049 4333257 24159 385 0,0159
63 stevecody 1787639 13114 208 0,0159
32 S485122 3607522 25198 395 0,0157
78 Son 1277908 8027 125 0,0156
50 trif 2232331 15546 241 0,0155
74 BranMuffin 1384275 9063 140 0,0154
91 tdhinton 1044229 8432 130 0,0154
56 1997rj7 2040390 55100 847 0,0154
81 Hazmattom 1237820 9595 146 0,0152
73 ANONYMOUS 1469724 7686 114 0,0148
80 Gymnasium_Raubling 1244776 7914 115 0,0145
92 fes016 1026931 5378 76 0,0141

alpertron · 2009-03-14, 14:00

Will has a very large database of factors of Mersenne numbers that was dumped to GIMPS, so it is clear that almost all his contributions were ok.

cheesehead · 2009-03-14, 15:52

To put it another way: There are a few participants whose statistics reflect not a set of normal GIMPS assignment reports, but a large number of successful factor-findings (from various sources) without the corresponding failure reports.

Perhaps, in time, the PrimeNet reports will be refined to separate those anomalies from the normal cases in the rankings.

alpertron · 2009-03-14, 17:37

The other contributors with more than 10% of success started the new wave of Primenet 5.0 trial factoring in the range 79.3 - 1000M. Since smaller factors have a greater probability of being discovered than larger ones, they found lots (literally millions as you can see on your list) of small factors.

Uncwilly · 2009-03-15, 00:22

Quote:

Originally Posted by alpertron

The other contributors with more than 10% of success started the new wave of Primenet 5.0 trial factoring in the range 79.3 - 1000M. Since smaller factors have a greater probability of being discovered than larger ones, they found lots (literally millions as you can see on your list) of small factors.

I am on the list and am doing TF's in a higher range.

chalsall · 2009-06-28, 19:39

So... I like to hunt for factors of MPrime candidates.

I happened to notice today that a range of candidates became available up in the M120000000 et al range, even though these were previously reported as being tested.

These had been reported as "no factor below 2^63" by GrunwalderGIMP, but suddenly they were available for re-testing from 2^58.

Humm.... Assuming I'd not find any new factors, but always willing to contribute cycles to the effort, I reserved 2500 exponents to take from 58 to 63.

Much to my surprise, after testing only 39 candidates, my systems have already found 5 factors which GrunwalderGIMP reported as having had none.

Humm....

Can anyone please tell us all what gives?

Prime95 · 2009-06-28, 20:49

Quote:

Originally Posted by chalsall

These had been reported as "no factor below 2^63" by GrunwalderGIMP, but suddenly they were available for re-testing from 2^58.

Garo noticed the problem and brought it to my attention. GrunwalderGIMP and garo are retesting all his factoring efforts in the 120M - 121M range. The database was rolled back so that they would get proper CPU credit.

GrunwalderGIMP used Mfactor which may have a bug or he may have been running it on a bad machine.

If you'd like to spot check GrunwalderGIMPs work in earlier ranges, that might be useful.

chalsall · 2009-06-28, 21:33

Quote:

Originally Posted by Prime95

If you'd like to spot check GrunwalderGIMPs work in earlier ranges, that might be useful.

Doing so right now. In the 120M range.

So far I have tested 276 candidates. I have found 20 unreported factors within same.

The remaining 2186 candidates should be completed within 12 hours.

This does raise a question however: should we double check negative factoring results?

Prime95 · 2009-06-28, 21:42

Quote:

Originally Posted by chalsall

Doing so right now. In the 120M range.

I meant checking some of his work in ranges below 120M. Your effort may duplicate the retesting being done by garo and GrunwalderGIMP.

chalsall · 2009-06-28, 21:54

Quote:

Originally Posted by Prime95

I meant checking some of his work in ranges below 120M. Your effort may duplicate the retesting being done by garo and GrunwalderGIMP.

But I have reserved through the V5 server factoring from 58 to 63 of (currently) 2118 candidates (and counting down...) (I have AIDs, and everything!)

(I'm now up to finding 25 factors of candidates which were previously reported as being factor-free.)

I will consider testing some other previously reported "no-factor-found-below" results in the future.

But my question stands -- perhaps expanded: should we double check negative factoring results from everyone -- not just those of GrunwalderGIMP?

Prime95 · 2009-06-28, 22:18

Quote:

Originally Posted by chalsall

But my question stands -- perhaps expanded: should we double check negative factoring results from everyone -- not just those of GrunwalderGIMP?

That isn't cost effective at this time. This event seems to be a rarity. Also, a decent percentage of missed factors will be found by P-1.

2009-03-14, 13:31	#1
pegaso56 Oct 2006 Rosario, Argentina 2B₁₆ Posts	strange factors distribution?? Hi, I was playing a bit with top 100 trial factors producers, and i found something really curious -at least for me-. Relation between exponents tested and successes is 98.60% for Will Eddington, fall to around 36.74% and 29,60% for the next five , and then well... Has this something to do with range selected to try? Regards, Carlos Ghz-dia Intentos Hallados Indice 97 Will 965005 671540 662166 0,9860 31 David 3753200 11525604 4233988 0,3674 88 ATH 1112196 5959340 2173086 0,3647 79 S207114 1266338 1330276 468615 0,3523 72 Team_Bundu 1495327 1495252 504953 0,3377 33 arnaud 3533488 84918 25132 0,2960 2 Sturle 40530447 569926 81821 0,1436 21 monst 4658748 4225938 556604 0,1317 51 George 2212120 114585 11792 0,1029 37 James 3020591 194939 17440 0,0895 15 GrunwalderGIMP 6745535 322462 24777 0,0768 83 KriZp 1184503 1247286 72736 0,0583 64 Team_Inspector 1779833 31011 1518 0,0490 1 ANONYMOUS 127754299 1873574 81309 0,0434 96 Jan 967307 73710 3150 0,0427 99 Oleg 914283 6378 233 0,0365 29 ScottBardwick 4157571 178980 6466 0,0361 22 mcurtis 4615751 1112185 37344 0,0336 68 KYOJI_KAMEI 1657042 32798 1073 0,0327 35 sasaki 3103553 442947 14382 0,0325 82 carlos56 1201473 428949 13818 0,0322 47 Carsten 2444434 155029 4968 0,0320 100 S530729 885316 99496 3026 0,0304 87 andreasbomke 1115926 133629 4016 0,0301 59 oodaira 1933116 216136 6276 0,0290 46 Hebridean_Boy 2468834 18023 480 0,0266 76 Xyzzy 1319354 260524 6644 0,0255 85 BJB 1164195 99243 2520 0,0254 19 Architects 4995780 1499917 37660 0,0251 34 Derek 3198359 541630 13342 0,0246 66 Smok_bmv 1749894 26649 641 0,0241 55 HBendtz 2057308 51566 1179 0,0229 25 Wojciech 4348467 171800 3923 0,0228 20 ComputerraRU 4983458 79197 1795 0,0227 52 glennpat 2085311 23752 458 0,0193 67 Geoff 1735707 9179 171 0,0186 36 Luigi 3038628 39890 734 0,0184 40 WileECoyote 2752807 28986 532 0,0184 42 bayanne 2583312 38833 709 0,0183 71 linded 1567729 19417 341 0,0176 94 fjansson 986008 5225 91 0,0174 28 James 4249174 36028 623 0,0173 84 kevshaw 1165789 14524 250 0,0172 26 EPF 4333358 30522 525 0,0172 23 ANONYMOUS 4526281 19340 331 0,0171 60 kutens 1921113 14554 249 0,0171 86 jkolonko 1125964 9582 163 0,0170 43 Team_Deutschland 2542364 37107 631 0,0170 14 RMAC9.5 6882486 40655 690 0,0170 11 binarydigits 7873968 27877 471 0,0169 5 ITS 14590588 100373 1695 0,0169 77 Helgen 1282853 11969 202 0,0169 10 tsc 10803198 99273 1671 0,0168 38 sannerud.com 2920937 20687 348 0,0168 4 Team 17666007 191038 3206 0,0168 8 JC_Herles 12898613 96071 1610 0,0168 18 dlsilver 5259964 38433 643 0,0167 58 Guido 2000990 21235 355 0,0167 30 PrimeCruncher 3884430 22912 383 0,0167 44 subgenius 2542188 48124 804 0,0167 70 cmarble 1631063 18877 315 0,0167 6 TheJudger 14312103 143825 2400 0,0167 75 Robert 1363845 27458 458 0,0167 90 greensinoz 1074652 13617 227 0,0167 9 DSheets 10829832 136896 2282 0,0167 41 ANONYMOUS 2645770 59220 987 0,0167 24 smintheus 4392263 36602 610 0,0167 54 Net_Force 2061351 433485 7224 0,0167 62 outlnder 1789892 22626 377 0,0167 39 Matt 2760846 27731 462 0,0167 98 Piepen 927592 10805 180 0,0167 45 iPrime 2514811 20231 337 0,0167 48 delong 2438808 57232 953 0,0167 17 curtisc 6140101 93272 1553 0,0167 3 ahmerali 21673638 58079 967 0,0166 93 shy24 988534 12496 208 0,0166 89 Stephan 1075635 21346 355 0,0166 57 Texaport 2034219 18459 306 0,0166 61 S00030 1823924 15185 251 0,0165 53 WorknPlay 2066741 17010 281 0,0165 95 Dk 973290 6561 108 0,0165 7 tmorrow 13075722 73214 1201 0,0164 49 David 2367826 25802 423 0,0164 16 korva 6479371 70330 1148 0,0163 65 raymondf 1752290 14401 235 0,0163 12 Marzbug 7466877 49252 796 0,0162 69 Gordon 1650954 13156 211 0,0160 13 precius1 7061882 39237 628 0,0160 27 S36049 4333257 24159 385 0,0159 63 stevecody 1787639 13114 208 0,0159 32 S485122 3607522 25198 395 0,0157 78 Son 1277908 8027 125 0,0156 50 trif 2232331 15546 241 0,0155 74 BranMuffin 1384275 9063 140 0,0154 91 tdhinton 1044229 8432 130 0,0154 56 1997rj7 2040390 55100 847 0,0154 81 Hazmattom 1237820 9595 146 0,0152 73 ANONYMOUS 1469724 7686 114 0,0148 80 Gymnasium_Raubling 1244776 7914 115 0,0145 92 fes016 1026931 5378 76 0,0141

2009-06-28, 19:39	#6
chalsall If I May "Chris Halsall" Sep 2002 Barbados 3⁵·47 Posts	Should GrunwalderGIMP's negatives be questioned? So... I like to hunt for factors of MPrime candidates. I happened to notice today that a range of candidates became available up in the M120000000 et al range, even though these were previously reported as being tested. These had been reported as "no factor below 2^63" by GrunwalderGIMP, but suddenly they were available for re-testing from 2^58. Humm.... Assuming I'd not find any new factors, but always willing to contribute cycles to the effort, I reserved 2500 exponents to take from 58 to 63. Much to my surprise, after testing only 39 candidates, my systems have already found 5 factors which GrunwalderGIMP reported as having had none. Humm.... Can anyone please tell us all what gives?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Distribution of Mersenne Factors	tapion64	Miscellaneous Math	21	2014-04-18 21:02
Known factors distribution graphs	James Heinrich	Data	21	2013-09-26 19:54
Distribution of Mersenne prime factors mod 6	alpertron	Math	0	2006-06-23 20:07
A strange new (?) fact about Mersenne factors	ChriS	Math	14	2006-04-12 17:36
Silverman & Wagstaff on Joint Distribution of Ultimate and Penultimate Prime Factors	wblipp	Math	12	2006-04-02 18:40

2009-03-14, 14:00	#2
alpertron Aug 2002 Buenos Aires, Argentina 3×7×73 Posts	Will has a very large database of factors of Mersenne numbers that was dumped to GIMPS, so it is clear that almost all his contributions were ok.

2009-03-14, 15:52	#3
cheesehead "Richard B. Woods" Aug 2002 Wisconsin USA 2²×3×641 Posts	To put it another way: There are a few participants whose statistics reflect not a set of normal GIMPS assignment reports, but a large number of successful factor-findings (from various sources) without the corresponding failure reports. Perhaps, in time, the PrimeNet reports will be refined to separate those anomalies from the normal cases in the rankings.

2009-03-14, 17:37	#4
alpertron Aug 2002 Buenos Aires, Argentina 3×7×73 Posts	The other contributors with more than 10% of success started the new wave of Primenet 5.0 trial factoring in the range 79.3 - 1000M. Since smaller factors have a greater probability of being discovered than larger ones, they found lots (literally millions as you can see on your list) of small factors.