Sieve Check - stats

[jasong]

I redid 5150-5200, like I was asked. When I turned in the factors online, it said that they were already known, every single one of them. No big deal about the wasted time, just a heads-up. :)

[masser]

hmmm.... I will look into this further. My sincerest apologies about the wasted time and effort.

[geoff]

I've noticed a few times that the factors submission form says all of my factors are already known, when I'm pretty sure I haven't submitted them before.

[masser]

Here is what I suspect happened.

1. jasong originally submitted the factors for the range 5150-5200.

2. before the dat file was updated to reflect jasong's submission, I downloaded it and did some p-1 factoring on some low n values. I found a factor in the 5150-5200 range.

3. I submitted the found factor to the factor submission page - it was recognized as a new factor - maybe incorrectly?

4. I looked through the files at base5 DOT factors AT yahoo DOT com and couldn't find jasong's factors, so I emailed him; he had lost the factor files due to hardware issues, so he agreed to redo the range.

5. His resubmission resulted in no new factors - so they must have been there all along.


Three things seem to have contributed to this muck-up.

1. I wasn't patient/careful enough when squaring p-1 factoring with old sieving results.

2. Some issues with the stats submission pages - the resulting messages are not always consistent with the sieve stats page; I'm not sure this can be fixed.

3. jasong's unfortunate hardware issues.

I will be more careful in the future....

[ltd]

2. Some issues with the stats submission pages - the resulting messages are not always consistent with the sieve stats page; I'm not sure this can be fixed.

So far all factors reported as duplicate from my own tests ( most done on the PSP submission) were real duplicates.

I will implement some more internal reporting next weekend. So if you have an unexpected duplicate report starting from monday on please send me the factor file by mail so i can check what happened. ( So far i can not see how such an error could occur as the check for duplicates searches for an exact match for k,n and complete factor)

I don't know if this can also happen with srsieve but on jjsieve and proth_sieve for example it is possible to stop the client between two savepoint which can lead to a factor found twice.

Cheers,

Lars

[mdettweiler]

30,000-30,005 complete, 23 factors found (expected 22.87). Factors submitted to sieve import.

One thing I found odd, though: When I submitted my factors, I got the following message:
Input contained 23 total lines.
Input contained 23 valid lines.
Input contained 23 valid lines from SR5 k values.
Input contained 0 invalid lines.
Input contained 0 lines with unknown K.
Input contained 1 lines with already known factor.
Input contained 0 wrong factors.

Where would that "already known factor" have come from?

[Xentar]

Maybe a factor, that was found by another user, but is still in the sieve file?

[mdettweiler]

Originally Posted by Xentar

Maybe a factor, that was found by another user, but is still in the sieve file?

But doesn't it only give that message if the exact same factor was found? I think someone said on here once that if a second, but different, factor is found for a candidate, then the sieve import page won't make any noise about it, but you won't get any points awarded.

[masser]

Anon,

Could you post the list of factors here?

[mdettweiler]

Originally Posted by masser

Anon,

Could you post the list of factors here?

Sorry, I deleted the factors file as soon as they were inserted successfully.

Would you be able to see them in a log somewhere?

[thommy]

Probably someone tested sievespeed or something in that range, found a factor. And submitted it.