On Tue, 6 Jun 2000, Eric Hahn wrote:
>
> Second, the type of information I've received concerning the
> statistical data I've received (and is contained in the data),
> is of a statistical probability nature. For example(s):
> M14,989,627 has a 93.42% probability of having a 53-bit factor.
Did you remember to exclude the fact that M14,989,627 HAS a 53-bit factor
from the dataset used to calculate this probability?
> M14,999,953 has a 9.31% probability of having a 53-bit factor.
>
> M101 has a 89.17% probability of having 2 factors.
> M101 has a 73.25% probability of having highly compositable factors.
When getting to this probability, was the datapoint that M101 HAS 2
factors removed, or was it included in the statistical set used to
calculate this probability?
> M113 has a 92.36% probability of having 5 factors.
> M113 has a 64.03% probability of having highly compositable factors.
When getting to this probability, was the datapoint that M113 HAS 5
factors removed, or was it included in the statistical set used to
calculate this probability?
> NOTE: These are just examples that I've come up with to give an
> idea of the nature of the data (which I used previously known
> information to create). I don't even know if the data is stored
> in such a manner or is interpreted to this manner...
What's the algorithm used to come up with these probabilities?
Remember that it's very easy to accidentally fool yourself into seeing
patterns where none are, unless you're very careful not to use preexisting
knowledge for generating the predictions you test agains it.
> Since these are statistical probablilities, I can't even
> guarantee the accuracy of the data. However, in a test run
> that was made to validate the data (not by me, mind you),
Ah, sorry, the questions I ask you is for the people who did the run
then, but they're still relevant.
> approx. 436 exponents that had high probability to have a
> 53- or 54-bit factor was tested and 401 had a 53- or 54-bit
> factor found! There's no telling how many exponents in the
> range tested have a factor of this size which wasn't tested
> due to a lower probability...
>
> One last thing... The data is based on current information
> as of 4-6 weeks ago, some of which may have changed. It
> "predicts" information that isn't known as of that date, such as
> # of factors in M727 and M751, range where M#39 and M#40 is,
> etc. (which was why I had the subject I did).
>
> If anybody has further questions or is still confused, let
> me know and I'll try to clarify some more...
>
> Again, the ideas, suggestions, comments, etc. I'm looking for
> covers these areas and ???:
> 1) Should I post a *tiny* fragment of this information?
> 2) What information should I post, if I do?
> 3) Would it be beneficial to the overall effort?
> 4) Would it divert resources that could be used better otherwise?
> 5) Could it cause problems with regard to that "p" hunting term?
> 6) Anything else??
>
> Eric
>
>
> _________________________________________________________________
> Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
> Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
>
--
Henrik Olsen, Dawn Solutions I/S URL=http://www.iaeste.dk/~henrik/
What is worth doing is worth the trouble of asking somebody to do it.
The Devil's Dictionary, Ambrose Bierce.
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
