Factors that doesn't verify?

[hc_grove]

My sieve clients have found a couple of factors that doesn't verify according to http://www.seventeenorbust.com/sieve/

When I try submitting the following factors:
180567238560433 | 28433*2^18448657+1
180944075565311 | 55459*2^12539398+1
181376155768793 | 55459*2^17709394+1
181776502050749 | 24737*2^13971991+1

I get this result:
0 of 4 verified in 0.04 secs.
0 of the results were new results and saved to the database.

I don't have any other means og verifying these, but there is a problem with either the verification script or the client (proth_sieve_cmov v.40 for Linux).

Note: I have no problems submitting the 5 factors with n<10M the clients have also found during the night, if I submit them seperately from the 4 mentioned, but if I submit any of them along with any of the 4 factors, none of them verify. :confused:

.Henrik

[Nuri]

I have the same problem.

I have a feeling that it has something to do with the Brief server maintenece (http://www.free-dc.org/forum/showthread.php?s=&threadid=5232).

Just a wild guess.

[Joe O]

Congratulations, You've now graduated to



the big time! (http://www.seventeenorbust.com/largesieve/)


Hint:
2^64 is approximately 18446744073709600000

[hc_grove]

Originally posted by Joe O
Congratulations, You've now graduated to

the big time! (http://www.seventeenorbust.com/largesieve/)

Hint:
2^64 is approximately 18446744073709600000
:bs:
Which is approximately 100000 times as much as the factors!

I considered that (and even tried submitting them there - as the finder of the largest factor so far I know that page), no difference. And some of the mentioned factors with n<10M are larger than the smallest of those that doesn't verify.

.Henrik

PS: I know have six factors that I can't submit. :mad:

[Joe O]

OK, so your factors are somewhere between 2^47 and 2^48.

So, my post was just a big red herring!

[MikeH]

I'm 100% certain it's the server not the proth sieve. Try submitting some old factors - one's that verified fine in the past. Now these don't verify.:(

[Nuri]

It seems like we'll accumulate a few more factors until the server issue is resolved.

At the moment, I have 64 factors awaiting submission. Luckily, none of them needs urgent submission, as they are all at least 200k far from the next k/n pair.

[jjjjL]

I've been working on this for about an hour.

Already ruled out a few things but haven't figured out the root cause yet.

So far I've ruled out that the factor table being too large and that either the factor table or the test table queries were failing to be stored.

I've also manually checked my factor verification program. It still works when I run it from the command line.

There is no reasonable explaination for this happening. I'll keep pluging and let you know when it's working.

Cheers,
Louie

[rosebud]

It seems to accept bigger factors...

I just submitted some random factors from result.txt and

961094450858074349|67607*2^4022171+1

was accepted and verified.

Maybe something with the contraints got mixed up...

[jjjjL]

Originally posted by rosebud
It seems to accept bigger factors...

I just submitted some random factors from result.txt and

961094450858074349|67607*2^4022171+1

was accepted and verified.

Maybe something with the contraints got mixed up...

This is an odd wrinkle in the problem. This factor does work. That removes a lot of possible explainations for this.

What I do know:

The factors hc_grove posted ARE valid.
The webscript DOES NOT recognize them as valid.
They are NOT being saved to the database. :(

The factor rosebud posed is valid too.
The webscript recognizes it as valid.



What's worse is that the actual verifier recognizes all them as valid, it just doesn't do it when I use the webpage frontend. It would have made much more sense if it didn't work for all numbers because of the kernel upgrade. I'm completely confused.

Cheers,
Louie

[rosebud]

Is there any chance that you can post the interesting part of the verifier script/webpage?

[Alien88]

trying to figure it out...

[Nuri]

As far as I can see, it has something to do with the n value.

It verifies factors where n is smaller than (or equal to) 8357208. 25608499169039|33661*2^8357208+1 is verified. This is the largest n that I could find that is verified.

Whereas, it does not verify factors where n is larger than (or equal to) 8358045. 30569064815243|27653*2^8358045+1 is not verified. This is the smallest n that I could find that is not verified.

Those n value are very close to 2^23 (=8388608), but it's probably just a conicidence.

Another thing I noticed, if a bunch of factors are submitted, it does not verify factors below the first unverified factor, even if their n is below 8357208. For example;

Submitting

65185621818509 | 10223*2^5293277+1
65187400270849 | 10223*2^7435817+1
65191796767183 | 5359*2^3458982+1
65193566266403 | 5359*2^7593606+1
65196231333869 | 28433*2^10474945+1

results in 4 of 5 verified, wheras submitting (just the rows are changed)

65185621818509 | 10223*2^5293277+1
65187400270849 | 10223*2^7435817+1
65196231333869 | 28433*2^10474945+1
65191796767183 | 5359*2^3458982+1
65193566266403 | 5359*2^7593606+1

results in 2 of 5 verified, or

65196231333869 | 28433*2^10474945+1
65185621818509 | 10223*2^5293277+1
65187400270849 | 10223*2^7435817+1
65191796767183 | 5359*2^3458982+1
65193566266403 | 5359*2^7593606+1

results in 0 of 5 verified, etc.

[Slatz]

does this small bug mean we should hold off on submitting factors for the time being???

Slatz

[Mystwalker]

Interesting...
I just tested some idea and stumbled over another strange occurence (which is unrelated to my idea...):

25608499169039 | 33661*2^8357208+1
does work.


25608499169039 | 33661*2^8357208+1
138171575414911 | 10223*2^6240857+1
doesn't work at all (not even the first one)!
:confused:

I have to lower the n of the first line to 8357200 in order to get the checker to verify the second factor...

At least the problem seems not to be dependent on k and p...

Louie:
I don't know how the checker works. It is possible to output the values the receiving part of the script hands over to the calculating part? Maybe the bug can be found there.

[rosebud]

It verifies factors where n is smaller than (or equal to) 8357208.
25608499169039|33661*2^8357208+1 is verified. This is the largest n that I could find that is verified.

Whereas, it does not verify factors where n is larger than (or equal to) 8358045. 30569064815243|27653*2^8358045+1 is not verified. This is the smallest n that I could find that is not verified.

Those n value are very close to 2^23 (=8388608), but it's probably just a conicidence.



That really seems to be a coincidence since

23488298830411|24737*2^8388607+1
65185621818509|10223*2^5293277+1

isn't verified.

The first one isn't really a factor, i made that one up, but k and n exist and if 2^23 would be the the first n which creates this bug, at least the second factor should verify properly. But it doesn't...

But I agree that it seems to be dependent on the n and the barrier lies somewhere near your numbers.

[Death]

170009600969081 | 28433*2^1366033+1
170014030938307 | 4847*2^6416463+1
170019651224303 | 33661*2^8532072+1
170023819345189 | 10223*2^18514745+1
170040595076329 | 33661*2^16280568+1
170041258862309 | 33661*2^4862880+1
170048083532443 | 27653*2^1608333+1
170051177520043 | 27653*2^19499361+1
170057139804883 | 33661*2^16206792+1
170070828577943 | 19249*2^18405086+1
170078431188647 | 21181*2^10269092+1
170083370346977 | 4847*2^17225487+1
170097113252531 | 4847*2^9899847+1
170126735275979 | 24737*2^18357967+1
170139052918969 | 10223*2^14254277+1
170140875122293 | 55459*2^5611798+1
170147984382401 | 27653*2^16107225+1
170175793040329 | 55459*2^1915330+1
170176629964163 | 67607*2^14305451+1
171008580929471 | 10223*2^1832009+1

that was already submitted factors. all of them was previously verified.

Verification Results
170009600969081 28433 1366033 verified. 170014030938307 4847 6416463 verified.

Factor table setup returned 1
Test table setup returned 1

2 of 20 verified in 0.19 secs.
0 of the results were new results and saved to the database

:confused:

[Alien88]

well, both louie and i are stumped.. dave looked at it briefly last night and it confused him too. it seems to be an issue with perl, and we're going to work on it some more later tonight..

[cedricvonck]

Here are my results (low n sieving)

38800 38860

Range done & following factors found:
I get an error on the submit page so I post them here.

38803298336753 | 33661*2^883224+1
38804021783123 | 5359*2^1952046+1
38804736780263 | 27653*2^1905081+1
38806930563887 | 33661*2^1536936+1
38825329408037 | 67607*2^1584171+1
38831155566569 | 33661*2^2187504+1

Factors sent in by email. /ceselb

[MikeH]

Any plans to roll back the server (assuming it was the update that caused the problem) if a solution is not forthcomming in a reasonable timeframe, say 10 days?

[kugano]

Originally posted by hc_grove
When I try submitting the following factors:
180567238560433 | 28433*2^18448657+1
180944075565311 | 55459*2^12539398+1
181376155768793 | 55459*2^17709394+1
181776502050749 | 24737*2^13971991+1I used your first factor (28433 / 18448657) to test the submission script with the new Perl. I updated the database to give you credit for that first factor instead of me. You'll need to submit the other three yourself. Just didn't want you to see that the first one was already in the database and be worried about not getting credit ... it's yours =)

[hc_grove]

Originally posted by kugano
I used your first factor (28433 / 18448657) to test the submission script with the new Perl. I updated the database to give you credit for that first factor instead of me. You'll need to submit the other three yourself. Just didn't want you to see that the first one was already in the database and be worried about not getting credit ... it's yours =)

Thank you.