Three matching tests not closing exponent?

Dubslow · 2012-04-24, 08:28

http://mersenne.org/report_exponent/?exp_lo=25956701

The first two could easily be suspicious (and I've DCd one of my own expos, different hardware), but surely the third test puts the case to rest? I'd rather not test it if I don't have to.

Batalov · 2012-04-24, 08:42

Someone was phishing for the last two hex-digits and didn't get lucky before being blocked by the server?

It would only take 256 tries to successfully "verify" any residue! Why do you assume that these three have the censored digits matched?

aketilander · 2012-04-24, 08:43

Quote:

Originally Posted by Dubslow

http://mersenne.org/report_exponent/?exp_lo=25956701

The first two could easily be suspicious (and I've DCd one of my own expos, different hardware), but surely the third test puts the case to rest? I'd rather not test it if I don't have to.

My guess is that all three tests have the same shift count and therefore the value is only equal to one test. That is, most probably it is not three independant tests, so I would guess that the same user have done all three tests. Maybe his computer broke and he continued from a backup copy without knowing that the result had already been reported and maybe he changed username when he reinstalled Prime95.

Or maybe he installed the program on three different computers starting with a backup-copy with the expo half-way through.

So in order to safely clear the exponent a 4th test is needed.

LaurV · 2012-04-24, 09:21

Al in all, this is quite ok, and makes me feel good, safe, regardless of whatever the reason is. In the sense there are no false-negatives in the database (i.e. missed primes :P). If we are about to eliminate an exponent, then let's be completely sure the associated mersenne number is composite...

Dubslow · 2012-04-26, 19:31

All previous tests had the same matching last two bits.

http://mersenne.org/report_exponent/?exp_lo=25956701
Edit: The chances of them having the same shift value is astronomically small. (That's randomly produced at runtime.)

aketilander · 2012-04-27, 09:45

Quote:

Originally Posted by Dubslow

All previous tests had the same matching last two bits.

http://mersenne.org/report_exponent/?exp_lo=25956701
Edit: The chances of them having the same shift value is astronomically small. (That's randomly produced at runtime.)

Yes you are right, the shift value should, most probably, be different if the three tests were independant, but if the second and third test really started with an intermediary file or backup file at, lets say, iteration 25000000 (on a different computer) then I suppose the shift value would be the same and the server would recognize it. (this is my guess) So my guess is that's the reason why the exponent wasn't cleared after three tests is that they had the same shift count showing that the three test were in fact only 1 complete test. But on the other hand I can be wrong!

I would be happy if someone with better insight in these matters could tell us how it works.

Dubslow · 2012-04-27, 16:58

Quote:

Originally Posted by aketilander

Yes you are right, the shift value should, most probably, be different if the three tests were independant, but if the second and third test really started with an intermediary file or backup file at, lets say, iteration 25000000 (on a different computer) then I suppose the shift value would be the same and the server would recognize it. (this is my guess) So my guess is that's the reason why the exponent wasn't cleared after three tests is that they had the same shift count showing that the three test were in fact only 1 complete test. But on the other hand I can be wrong!

I'd buy that for the first two, since the username is similar, but the third one as well?

I suppose though this makes the most sense out of the running suggestions.

Quote:

Originally Posted by aketilander

I would be happy if someone with better insight in these matters could tell us how it works.

Yes me too please

(Perhaps someone with special database access?

)

bcp19 · 2012-04-27, 17:40

Quote:

Originally Posted by Dubslow

I'd buy that for the first two, since the username is similar, but the third one as well?

I suppose though this makes the most sense out of the running suggestions.
Yes me too please

(Perhaps someone with special database access?

)

If you look through the top producers usernames, there are several S5.... usernames, and a search of Carlo turned up a CarloS56, so coincidence?

Batalov · 2012-04-27, 18:19

These S##### handles were, I believe, bestowed to very early users by the server. Some of them are proudly worn even here, on the forum.

It looks like the savefile was passed on to friends to boost their ratings or something like that. Someone may have a look into other tasks given to the same userids (but then again they may have done only one job, "got bored" and left).

2012-04-24, 08:28	#1
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3·29·83 Posts	Three matching tests not closing exponent? http://mersenne.org/report_exponent/?exp_lo=25956701 The first two could easily be suspicious (and I've DCd one of my own expos, different hardware), but surely the third test puts the case to rest? I'd rather not test it if I don't have to.

2012-04-24, 08:42	#2
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 3²×1,117 Posts	Someone was phishing for the last two hex-digits and didn't get lucky before being blocked by the server? It would only take 256 tries to successfully "verify" any residue! Why do you assume that these three have the censored digits matched? Last fiddled with by Batalov on 2012-04-24 at 08:46

2012-04-24, 09:21	#4
LaurV Romulan Interpreter "name field" Jun 2011 Thailand 2·11·467 Posts	Al in all, this is quite ok, and makes me feel good, safe, regardless of whatever the reason is. In the sense there are no false-negatives in the database (i.e. missed primes :P). If we are about to eliminate an exponent, then let's be completely sure the associated mersenne number is composite... Last fiddled with by LaurV on 2012-04-24 at 09:25

2012-04-26, 19:31	#5
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3·29·83 Posts	All previous tests had the same matching last two bits. http://mersenne.org/report_exponent/?exp_lo=25956701 Edit: The chances of them having the same shift value is astronomically small. (That's randomly produced at runtime.) Last fiddled with by Dubslow on 2012-04-26 at 19:31

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Largest number of LL tests before matching residues achieved?	sdbardwick	Lounge	1	2015-02-03 15:03
Can 1227133513 be the only composite number matching the conditions?	miket	Math	5	2014-08-12 00:41
Residue not matching due to masked bits	patrik	Data	1	2011-09-24 23:44
Planet closing in causing my family's bad luck ?	science_man_88	Astronomy	16	2010-09-21 03:44
Closing Older Threads to Thwart Spammers	Prime Monster	Forum Feedback	0	2004-10-24 23:47

2012-04-27, 18:19	#9
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 3²·1,117 Posts	These S##### handles were, I believe, bestowed to very early users by the server. Some of them are proudly worn even here, on the forum. It looks like the savefile was passed on to friends to boost their ratings or something like that. Someone may have a look into other tasks given to the same userids (but then again they may have done only one job, "got bored" and left).