Stats

[masser]

msieve is a pretty cool little factoring program... it's the first useful computational number theory program I have been able to compile and use on my wife's Powerbook G4.

ltd's factor of 66916*5^125-1 on the Riesel stats page is actually the 46 digit factor below:

Fri Aug 10 07:54:47 2007 prp46 factor: 4759099711946098409535704924808155914387497323

The factor of 289184*5^128-1 has also been truncated at 45 digits; I will work on fixing that one next, unless ltd has a record of the actual factor.

masser

[ltd]

Hi,

due to a hd crash earlier this year i lost my own factors. I have a backup of all the mails with factors i received but forgot to but my own factors to this directory.

So thanks for redoing the factor recalculating.

Cheers,

Lars

[masser]

ltd's 45 digit factor,

401694244324514676277709623403369488558647705 | 289184*5^128-1

should be the 46 digit factor...

4016942443245146762777096234033694885586477057 | 289184*5^128-1

EDIT:

421375084969702712205980311619670526720489021 | 60394*5^132+1

should be

4213750849697027122059803116196705267204890213 | 60394*5^132+1

Edit 2:

A P29 credited to me is actually composite:

39041505082991434800517010213 | 138514*5^750-1, but
79104311811311 | 39041505082991434800517010213

Same thing with a P39 credited to me:

121664461662516457957154127262589496751 | 200062*5^293+1, but
494384300675642389 | 121664461662516457957154127262589496751

[ltd]

Finaly i have imported all the corrections to the factors.

Thanks for your additional work that helped to create a correct picture


Lars

[mdettweiler]

My completed range of 11000-11005 had 71 factors found; however, when I visit the stats page, and add together my factors shown on both the Sierpinski and Riesel pages, the total is only 70.

I know that when I submitted my range to the sieve import page, it said that one line was not valid, but that 71 lines were valid, so I assumed the "not valid" line was the blank line at the end of my factors that had been copied and pasted in from the sr5sieve factors file.

Is it possible to tell from the sieve import logs whether any of my factors weren't received, and if so, which one?

[masser]

It sounds like all of your factors were received. The reason you only see 70 factors in the stats is that one of your factors eliminated a candidate that must have already been eliminated by someone else's sieving.

[mdettweiler]

Originally Posted by masser

It sounds like all of your factors were received. The reason you only see 70 factors in the stats is that one of your factors eliminated a candidate that must have already been eliminated by someone else's sieving.

But if that was the case, wouldn't the sieve import have said that one of the factors wasn't new? If I remember correctly, when it shows the results of a submission, one of the lines is "x already known factors". Or, would the sieve import page not report the factor as "already known" if the other copy of the same factor was submitted by email rather than by the sieve import page?

[masser]

That's a good point. The submission page will tell you if the number has already been sieved out.

Here's another possibility: A prime may have been found for one of the sequences between the time of your reservation and the time of your submission. You reserved that range on 9/11 and submitted on 10/1. By my count, two primes were found in the meantime, one for k=5114 and one for k=155056. Perhaps one of the factors you found eliminated a candidate from one of these sequences?

FYI, in a few hours I will move this discussion to the Stats thread.

regards,
masser

[mdettweiler]

Originally Posted by masser

That's a good point. The submission page will tell you if the number has already been sieved out.

Here's another possibility: A prime may have been found for one of the sequences between the time of your reservation and the time of your submission. You reserved that range on 9/11 and submitted on 10/1. By my count, two primes were found in the meantime, one for k=5114 and one for k=155056. Perhaps one of the factors you found eliminated a candidate from one of these sequences?

FYI, in a few hours I will move this discussion to the Stats thread.

regards,
masser

I know I manually removed the k's for both of those primes from my copy of the sieve file; but yeah, I guess I could have found a factor for one of those k's before the prime was found, i.e. before I removed the k from the sieve file.

Thanks for your help!

Edit: Now that I think about it, would this seeming anomaly be potentially caused by a factor being returned for a number that was already PRP tested? This would seem to be the case, because earlier, I had entered some factors for some smaller numbers that I had factored with msieve, and they didn't show up in the stats--yet their k's were still listed as no prime found yet. (I had grabbed the candidates to test from this thread.) However, according to the aforementioned thread, all of the candidates listed there were already PRP tested--and since none of them showed up in the sieve stats either, it seems as if this anomaly is most likely caused by the number that the factor is for being already PRP tested.

[masser]

Your factor submission will count regardless of whether the candidate removed has been PRP-tested. We keep all candidates in the dat file until a factor has been found for them, even if they have been double-tested.

For the msieve work, I have found several factors recently and submitted them without updating the thread you cited. Perhaps I found a factor for some of the candidates you factored. I should update that thread again with some of my ECM progress... sorry if this is the case. If you do want to try msieve on any of the low n candidates, please post the candidate you want to factor and I will take it out of my ECM list. I know michaf completed the n=148 candidate and I recently factored several candidates in the n=150-160 range.

[mdettweiler]

Originally Posted by masser

Your factor submission will count regardless of whether the candidate removed has been PRP-tested. We keep all candidates in the dat file until a factor has been found for them, even if they have been double-tested.

For the msieve work, I have found several factors recently and submitted them without updating the thread you cited. Perhaps I found a factor for some of the candidates you factored. I should update that thread again with some of my ECM progress... sorry if this is the case. If you do want to try msieve on any of the low n candidates, please post the candidate you want to factor and I will take it out of my ECM list. I know michaf completed the n=148 candidate and I recently factored several candidates in the n=150-160 range.

I just did a few of the smallest candidates listed in that thread (even though some of them had been already been factored with ECM or msieve by others, I figured I'd do it just to try it out). The candidates I factored are:
151026*5^110-1 (from the beginning of the thread, not in the huge lists of candidates later on)
93254*5^111+1
71098*5^112+1
I didn't get any "factor has already been submitted" errors for any of them except the last one, for which one of the factors was already submitted (but not the other one; since msieve finds two factors per candidate, I figured I may as well submit both for good measure even though one will do fine). However, my name didn't even show up in the sieve stats for either Riesel or Sierpinski until I submitted the results for my recent sieve range--and still, the totals do not reflect the earlier factors. Not that it matters--most of the earlier factors, if not all, were already found by others and posted in the ECM factoring thread, so it's not like they should necessarily be shown under my name. If I try to re-submit any of them, though, the sieve import page says that all of them are already in the database, so it appears that they were submitted OK.

BTW, how would I go about factoring a number with ECM? The only programs that I've found that do ECM are Prime95, and GMP-ECM (I can't figure out how to get Prime95 to ECM anything that's not a Mersenne number; and as for GMP-ECM, which I downloaded a precompiled copy of, it didn't include much in the way of hints on how to use it).