Factor found that should have been found by P-1

[tha]

Originally Posted by VictordeHolland

Considering the amount of LL/DC tests with bad residues, I find 295 P-1 factor misses not that much.

The chance that a wrong LL residue is found is eventually 100%, the chance that a wrong P-1 result is smoked out is only a few percent at most. Worse, a wrong dedicated P-1 machine can produce a lot of wrong results in little time.

[tha]

Originally Posted by James Heinrich

So last night's 1160 exponents is now down to 295.

That is probably down too much since not all exponents posted in this thread made it to the list. Also, I don't know how often the listed is updated, but a new exponent I found today and added to the mersenne v5 server did not yet pop up in the list after a few hours. That exponent is http://www.mersenne.ca/exponent/12526721.

[James Heinrich]

The list is updated once per day, roughly 11pm Eastern time.

If you notice a particular exponent that should be on the list but isn't, and the factor you found was reported to mersenne.org more than 24h ago, please let me know.

[alpertron]

Notice that the biggest factor in http://www.mersenne.ca/exponent/48152443 is the product of the first three factors.

[TheMawn]

I think you're looking at the composite factors, which are just all combinations of any number of prime factors. One such combination is all three prime factors.

[Dubslow]

Originally Posted by TheMawn

I think you're looking at the composite factors, which are just all combinations of any number of prime factors. One such combination is all three prime factors.

No he's not. The page says 4 prime factors, but the fourth listed is indeed as he claims.

[James Heinrich]

Originally Posted by alpertron

Notice that the biggest factor in http://www.mersenne.ca/exponent/48152443 is the product of the first three factors.

Thanks. In my quest to include the missing missed P-1 factors I may have inadvertently imported some composite factors. I'm trying to carefully weed out what I may have imported, so far I've purged the one you mentioned, as well as one on M8,724,319 and M77,999,869.

[ric]

Originally Posted by James Heinrich

If you notice a particular exponent that should be on the list but isn't, and the factor you found was reported to mersenne.org more than 24h ago, please let me know.

I have found two in December/January, namely this (1) and this (2) (whose former B1/B2 limits were 395000,10171250). The third and last one so far comes instead from mid-march and is this (3) (and again its former B1/B2 were 205000,973750). I also noticed that a more recent one in the 18M range made its way to the list directly.

Please consider that in cases (2) and (3) the original lower B1/B2's appear to have been overwritten with new larger values. Just in case, I've some saved data from mid-December, to corroborate this.

HTH
ric

[James Heinrich]

Originally Posted by ric

Please consider that in cases (2) and (3) the original lower B1/B2's appear to have been overwritten with new larger values. Just in case, I've some saved data from mid-December, to corroborate this.

It all relates to old P-1s which were done in a time when PrimeNet didn't keep a record of who did the P-1 or when, just that it was done to certain bounds. I've created pseudo P-1 entries on mersenne.ca for those that were still in that factorless state as of 15-Mar-2015, but the previous bounds would have been clobbered by any new P-1 submissions prior to that date.

I'm not terribly concerned about the missing data. If George wants to dig up the ancient v4 database and pass the submission logs my way I could in theory parse it all and extract the original P-1 submissions, but beyond that it's just a fuzzy memory that (in my opinion) can stay that way.

[ric]

Originally Posted by James Heinrich

I'm not terribly concerned about the missing data.

Neither am I, given their zero contribution to overall workload reduction - just wanted to draw attention to the fact that those 3 numbers were treated in three different ways by mersenne.ca update scripts:

1. found in December (before the 3/15 pseudo-entries update you mentioned) - original B1/B2 limits maintained in row named Actual in the "Last-known PrimeNet details" section;
2. found in January (still before the 3/15 update) - original B1/B2 limits overwritten in the same "Actual" row;
3. found on March 19th (after the 3/15 update) - original B1/B2 limits still overwritten in the same "Actual" row.

Sure, these 3 might be red herrings - naturally they can stay this way, provided it's an informed choice

[James Heinrich]

Originally Posted by ric

1. found in December (before the 3/15 pseudo-entries update you mentioned) - original B1/B2 limits maintained in row named Actual in the "Last-known PrimeNet details" section;

This has been fixed now. I somehow missed 195 P-1 entries that have now been added.