The curious case of 721003

[Uncwilly]

I was looking through the recently cleared list and saw a very small exponent with a small reported factor. Primenet reports that it was found on Nov 22. Mersenne.ca says so to. They report it as a P-1. But the full report for 721003 shows PRP for the co-factor in 2017. The factor is trivially small. I reran the TF. Prime95 finds the factor, but during a manual coms, it did not report it. It must be an old factor, why did the most recent P-1 result update the discovery?

[tuckerkao]

How did this user had an inaccurate P-1 result reported which was why the server couldn't identify it at first. Mersenne.ca clearly shows that the factor is located well inside the defined bounds.

I applied the same P-1 bounds as the user below, I got F-PM1 on stage 1, see screenshot.

2016-03-17 alpertron NF-PM1 B1=500000, B2=15000000, E=12

[techn1ciaN]

Originally Posted by tuckerkao

How did this user had an inaccurate P-1 result which was why the server couldn't identify at first. Mersenne.ca clearly shows that the factor in well inside the defined bounds.

2016-03-17 alpertron NF-PM1 B1=500000, B2=15000000, E=12

P-1 work lines (of the Pminus1 format) have an optional known_factors parameter for doing deeper factoring without sending useless F-PM1 results by "finding" factor(s) that has/have already been reported. For example, I found a factor of M107230943 in stage 1 with Prime95's automatic B1, then re-ran P-1 to higher bounds with that factor as part of the work line. Nothing else was found, so that returned NF-PM1.

[tuckerkao]

Originally Posted by techn1ciaN

I found a factor of M107230943 in stage 1 with Prime95's automatic B1, then re-ran P-1 to higher bounds with that factor as part of the work line. Nothing else was found, so that returned NF-PM1.

But you had the F-PM1 result reported earlier than the NF-PM1 on M107230943.

There were no F-PM1 or F(trial factoring) results posted prior to the NF-PM1 for M721003

Originally Posted by techn1ciaN

Nothing else was found, so that returned NF-PM1.

PRP-Cofactor will show the known factor on the result line. It's the best that P-1 results include the same info too.

[slandrum]

Many small factors are listed for numbers w/o any TF or PM1 report. All that matters is that the factor is known, not how or even if it was reported.

[techn1ciaN]

Originally Posted by tuckerkao

There were no F-PM1 or F(trial factoring) results posted prior to the NF-PM1 for M721003

The fact that there's no LL recorded on the exponent (if nothing else) tells us that the one current factor of M721003 was already in the database when alpertron ran his P-1. (This forms part of what Uncwilly's original post is asking about.)

PrimeNet did not keep detailed records for much of its life (go back far enough and "PrimeNet" just consisted of emailing George Woltman). This means that many exponents with small factors do not have the "who / what / when" of discoverer, method, TF range (if applicable), and discovery time that we record today. There's therefore nothing to put in an "F" line on the exponent report page and we don't get one.

One can easily confirm this by going to the Recent Results page and checking anything from the PRP-CF wavefront. Take M11876701 for one example.

Originally Posted by tuckerkao

PRP-Cofactor will show the known factor on the result line. It's the best that P-1 results include the same info too.

You make a good point. I don't know what changes would be required and where; currently, it's either reported but not displayed, or not reported. I unfortunately didn't examine the result line for that P-1 re-run I did before it was sent in.

[Viliam Furik]

Originally Posted by Uncwilly

I was looking through the recently cleared list and saw a very small exponent with a small reported factor. Primenet reports that it was found on Nov 22. Mersenne.ca says so to. They report it as a P-1. But the full report for 721003 shows PRP for the co-factor in 2017. The factor is trivially small. I reran the TF. Prime95 finds the factor, but during a manual coms, it did not report it. It must be an old factor, why did the most recent P-1 result update the discovery?

I know from my own experience, that when there is a known anonymous factor (without a record of finding, no TF or P-1 or such), and it's found by P-1 and reported, it gets assigned to the user who reported it. Weirdly, this doesn't happen for TF, only for P-1 AFAIK.

[alpertron]

Originally Posted by tuckerkao

How did this user had an inaccurate P-1 result reported which was why the server couldn't identify it at first. Mersenne.ca clearly shows that the factor is located well inside the defined bounds.

I applied the same P-1 bounds as the user below, I got F-PM1 on stage 1, see screenshot.

2016-03-17 alpertron NF-PM1 B1=500000, B2=15000000, E=12

Prime95 has not found the small factor when I ran P-1 with the bounds above because I added the known factors in the file worktodo.txt

[Dobri]

Originally Posted by Uncwilly

It must be an old factor, why did the most recent P-1 result update the discovery?

Same here, but with an ECM test, see M30047 and its smallest factor 5588743.
The composite factor found by the ECM test is 6086814009550243889 = 5588743 x 1089120399623.
It seems that if no one is listed in the exponent's history for finding a particular factor, the next user who eventually finds said factor is credited for the 'discovery' by the server.