is a list of mersenne numbers i've tested available?

[imo]

I've tested 6 numbers to date. Unfortunately I've only noted the last two down. I've accessed my account report but it merely telsl me things of interest like CPU years and the total number of tests done. Can I access a list of the actual numbers I have tested in the past at all?
Thanks in advance

[1997rj7]

There is no simple way to do this. It can be done, with a little work.

Go to http://www.mersenne.org/status.htm and scroll down to the bottom of the page, where it says "The Mersenne Database".

You will need to download 3 files:
A zip file of double-checked exponents and their residues. (lucas_v.zip)
A zip file of exponents that have not been verified. (hrf3.zip)
A zip file of incorrect LL tests. (bad.zip)

Unzipping these files produces the files lucas_v.txt, hrf3.txt, and bad.txt. Your exponents will be in one of these three files. Tests that have been verified (have 2 or more matching residues) are in lucas_v. Tests that have not been verified are in hrf3. Tests that do not match are in bad.txt.

Note: These files are updated about every 2 weeks. If you have just finished a test, it may not be in the files yet.

Note 2: This method works only for LL tests. It doesn't work for factoring assignments.

[Mini-Geek]

If all of the computer's hard drives are available, all you need to do is open the results.txt file that's in the Prime95 folder and look for the results of the numbers there. If the results.txt files aren't available, 1997rj7's solution would probably be the easiest.

[petrw1]

Originally Posted by 1997rj7

You will need to download 3 files:
A zip file of double-checked exponents and their residues. (lucas_v.zip)
A zip file of exponents that have not been verified. (hrf3.zip)
A zip file of incorrect LL tests. (bad.zip)

In hrf3.zip there are no exponents below 15M. I believe this is due to some cleanup by George .... http://www.mersenneforum.org/showthread.php?t=8015

[petrw1]

So I went through these three lists as you requested and found 67 exponents. My current status report shows 66 exponents (there were none completed in the last few days).

Then, I used the recently learned method for easily computing P90 Years and computed that I should have about 1.6098 points more than is recorded on my stats.

Lo and behold I have one entry from the files you listed that seems to address my discrepancy.
Exponent..Userid..ComputerId..Points
13540147 petrw1 CFB6CA956 1.069738592

Who do I point this out to that can verify, and if necessary correct it?

[cheesehead]

Originally Posted by petrw1

In hrf3.zip there are no exponents below 15M. I believe this is due to some cleanup by George .... http://www.mersenneforum.org/showthread.php?t=8015

No, there George was referring to his actual database, not the hrf3.txt file (which is really a report generated from the database).

The reason there are no exponents below 15M in hrf3.txt is that hrf3.txt contains only exponents that have _not_ yet been doublechecked (successfully, with matching residues), but all exponents below 15M have been doublechecked (successfully, with matching residues) by now.

If someone reports a new LL result for an exponent that is below the nnnnnnnn that's posted on the GIMPS Status page at "All exponents below nnnnnnnn have been tested and double-checked.", that new result will show up in lucas_v.txt (or bad.txt, if the residue doesn't match) rather than hrf3.txt.

Originally Posted by petrw1

Who do I point this out to that can verify, and if necessary correct it?

Wait a few days, then try your status report again to see if that result is still missing. If it still is, then e-mail to woltman@alum.mit.edu.

[petrw1]

Thanks ... that makes sense

PS I'm surprised how many exponents in LUCUS_V.txt have 3, or even 4 entries, often by the same person with all 3 or 4 residues matching (if I am reading it right). Was there a time that 2 tests wasn't considered reliable enough?

[cheesehead]

Originally Posted by petrw1

PS I'm surprised how many exponents in LUCUS_V.txt have 3, or even 4 entries, often by the same person with all 3 or 4 residues matching (if I am reading it right). Was there a time that 2 tests wasn't considered reliable enough?

In a large project with tens of thousands of human contributors using imperfect software and hardware, there are going to be some glitches. Not every oddity is by design.

We occasionally see a single user turn in reports of multiple L-L tests on the same exponent, even though that is not what GIMPS is designed to encourage. Why? (a) Maybe a new user, not fully understanding the instructions, manually runs the same test two or more times. (b) Maybe someone does that deliberately, just to see what happens, or indeed as part of a deliberate test to make sure that two different systems produce the same result. (c) Since we can't tell whether two different systems have been assigned the same ID by a user, perhaps a test of the same exponent is run on two completely separate machines, then reported as from the same computer ID. (d) Maybe a PrimeNet software bug, for a while, caused sending multiple copies of the same assignment to the same user. (e) Sometimes a system re-transmits the same test result report to PrimeNet because an earlier glitch caused it not to erase results after reporting them, or there was a transmission error that might (from the user POV) have caused the first report not to be successful, so Prime95 automatuically saves it to report again. (f) Insert ones favorite other theory here. (g) Etc., etc., ...

BTW, the next-to-last field in a LUCAS_V.txt entry is the pseudorandom bit shift used to circularly permute the data going through FFT calculations. That number should usually be different on each line of otherwise identical user and computer IDs, unless the same report was transmitted multiple times before being reset on the user end.