Results 1 to 22 of 22

Thread: k=22699 low count??

  1. #1
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331

    k=22699 low count??

    Just curious if anyone else noticed a trend with 22699... the # of tests currently pending is low.

    According to the number of tests pending for each k
    10223 69174 tests 261 tests
    19249 25275 tests 98 tests
    21181 58932 tests 222 tests
    22699 24402 tests 91 tests
    24737 63336 tests 287 tests
    33661 58917 tests 268 tests
    55459 83672 tests 383 tests
    67607 20975 tests 94 tests

    Compared the total number of tests per k 22699 should be somewhere between 67607 and 19249
    http://www.aooq73.dsl.pipex.com/scores_p.htm

    So, k=22699 is short a few tests this morning, maybe 5% could be artifical... but worth keeping an eye on if the number of tests continues to decrease. Anyone have 22699 handed out today? Perhaps tommorrow...

  2. #2
    I just tried to test the hypothesis that 22699-tests are still handed out. I waited until the min n value from the secret stats page passed 10231678, which is an n-value for this very k I extracted from the sob.dat (If I did no mistake).
    The number or pending tests from the stats page (which is slightly delayed to the secret stats page) did not decrease, meaning that under the hypothesis, exactly one test-result did come back to the server. What is the p-value of this ?
    We need to know how many values came back in these (about) 15 minutes. In one day, there are 220 values coming back (=are handed out). If they are uniformly spread over the day, it gives us an estimate of roughly 2.2 tests in fifteen minutes. They are of course not uniform, and as its just day in the states, let it be 4 tests handed out in these 15 minutes. 91/1600=1/17.5=0,056875 of the tests are tests for this k (with low weight, indeed).
    The probability for 1 for this k out of 4 tests is 4*0,056875*(1-0,056875)^3=0,1908486852044677734375, mening that we cannot recect the hypothesis yet.

    But we can wait for the next n yet to come which should be
    10232110
    10233694
    10234702
    If the number of tests for 22699 is not increasing in these very moments, it becomes obvious that no more tests are being handed out.

    (Did anybody follow?)

    Yours H.
    PS. I'm probably too lazy to watch another time clicking all the time on the refresh button...
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

  3. #3
    vjs, you were right, THE NEXT PRIME IS FOR k=22699,
    (if it was no bogus residue).

    (I stayed until the next n-passage).

    (At least what I can say from a statistical standpoint: it is highly unlikely that there are still tests for that k handed out.)

    What a pity that it is just the k with the smallest weight; and the P-1-runs I did for this k and N=35M for the largest prime queue go out of the window, too.

    But hey, it should more look like this (at least); but that's for the moment.

    And for vjs and his all-penetrating eye.
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

  4. #4
    So, um, can I take this to mean it has been confirmed as a prime for sure? I'm asking because I need to get this information to Kirk Pearson as accurately and quickly as possible so it can be put up on the DC site. What about the other 2 numbers that are showing below 100 tests left? I will be shocked if 3 numbers fall at once.

  5. #5
    Quote Originally Posted by Jwb52z
    So, um, can I take this to mean it has been confirmed as a prime for sure? I'm asking because I need to get this information to Kirk Pearson as accurately and quickly as possible so it can be put up on the DC site.
    As far as I know, Just because a proth test identifies a prime, it still has to be tested with a separate (lengthier) method before we can claim it's prime for sure.

  6. #6
    Quote Originally Posted by umccullough
    As far as I know, Just because a proth test identifies a prime, it still has to be tested with a separate (lengthier) method before we can claim it's prime for sure.
    You're right, of course. I just thought that it possibly may have already been done.

  7. #7
    Quote Originally Posted by Jwb52z
    You're right, of course. I just thought that it possibly may have already been done.
    We probably won't hear about it until the project admins confirm it...

  8. #8
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331
    I don't want to start anything just yet guys lets just wait and see, until there is offical word from the project leaders no prime exists.

    We will certainly keep our eyes on the stats over the next few days for sure, yes I do follow.

    90 tests and counting down...

    Keen observation on the
    10232110
    10233694
    10234702

    There are also secondpass tests to consider, it's also faster to use Mike_H's p-1 worktodo script to find n's for a particular k.

  9. #9
    10233694 has ben assigned to me so i guess the tests are still going out. Perhaps someone noticed wewere on to something and put it in the queue however.

  10. #10
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331
    Humm... 89 test pending

  11. #11
    88
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

  12. #12
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331
    87

  13. #13
    funny that when the test was assigned to me the pending tests didn't go up?

  14. #14
    86
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

  15. #15
    Bad news people:

    22699 24414 tests 87 tests 9631629 13467862


  16. #16
    Former QueueMaster Ken_g6[TA]'s Avatar
    Join Date
    Nov 2002
    Location
    Colorado
    Posts
    184
    Hmmm - it's only one test. Maybe they put the PRP back in the pool to confirm it?

    (Think positive! )
    Proud member of the friendliest team around, Team Anandtech!
    The Queue is dead! (Or not needed.) Long Live George Woltman!

  17. #17
    To put a squash on any rumors, especially since some people seem very convinced we are hiding a prime, we aren't. There have been no new primes.

  18. #18
    Junior Member
    Join Date
    Jun 2003
    Location
    Ottawa, CAN
    Posts
    16
    Quote Originally Posted by Alien88
    To put a squash on any rumors, especially since some people seem very convinced we are hiding a prime, we aren't. There have been no new primes.
    Aww poop

  19. #19
    Ooops my fault.
    But you know, t'is this hope-thing.
    And as you get always better in hiding new primes, we are keen on the smallest sign.
    Thank you alien for the announcement; and the search continues, I guess.

    As for me, I am not disappointed, though. Proof: .
    Yours H.
    ___________________________________________________________________
    Sievers of all projects unite! You have nothing to lose but some PRP-residues.

  20. #20
    Moderator vjs's Avatar
    Join Date
    Apr 2004
    Location
    ARS DC forum
    Posts
    1,331
    Absolutely a hope thing,

    Reducing the dat size, eliminating a k, faster sieve, searching for large prime n's its what keeps us going.

    If there is no prime then k=22699 having the lowest number of pending test is sort of an issue don't you think? Not sure if I can explain why...

    Is there some possibility that n>10M is not being counted in the total. Or is it just some rare instance where those tests have for some reason, been over factored or returned quicker than average.... perhaps both.

    If it's not prime it's certainly an odd occurance...

    Are people still getting new k=22699 tests for n>10M?

  21. #21

  22. #22
    I love 67607
    Join Date
    Dec 2002
    Location
    Istanbul
    Posts
    752
    I guess this is simply a script error on the stats page. It only counts the tests below 10m. In fact, this is how I calculated that we had more than 50% of our tests were above 10m a couple of days ago. The same figure (i.e. proportion of tests below 10m to total number of pending tests) is roughly 30% as of now (simply add up the pending test numbers for each k and divide by the total pending tests figure, which is roughly 1441/4771).

    BTW, as the figures reflect the lagging tests only, it seems totally normal for a slightly disproportunate number of lagged pending tests for some of the ks.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •