Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:12 ADT
Bill Stewart (bill.stewart@pobox.com)
Tue, 07 Jul 1998 19:06:53 -0700
At 01:23 PM 7/7/98 -0400, Perry E. Metzger wrote:
>
>"William H. Geiger III" writes:
>> I would have to see the math and some test data before I would feel
>> comfortable with this. There seems to be a big difference between using
>> the time between to hit to generate X bits and taking the 2 deltas between
>> 3 hits and then generating one bit of data from that.
>
>My big question is this: are there tools for taking a set of random
>numbers dispersed according to a non-uniform distribution, like a
>poisson or normal distribution, and turning them into a set of random
>numbers over a uniform distribution? Given such tools, timing
>intervals between the geiger counter ticks is probably safe --
>otherwise, it may skew the results subtly.
It wouldn't surprise me if there's a friendly toolkit somewhere,
but it's certainly not hard to do. Depending on the probability
distribution of the non-uniform source and the amount of precision
you want, it's easy to build either table lookups or sometimes formulas
(formulas are easy for inverting Poisson/Exponential, and rather
more annoying for inverting Normal distributions.)
The problem is that the easy solutions require knowing the
parameters of the distribution - so you either need to know
the real values for mean, variance, etc., or else you have
to estimate them from samples, which requires burning initial
samples and has wider error distributions since the parameter
estimates will be less accurate. There's lots of statistical
analysis software available, since most real-world processes
don't have a priori known parameters, and often even the
distributions are only guesses, and often the parameters are
what you really want to estimate (e.g. variance in the size of
a widget that's supposed to meet given specs.)
Thanks!
Bill
Bill Stewart, bill.stewart@pobox.com
PGP Fingerprint D454 E202 CBC8 40BF 3C85 B884 0ABE 4639
The following archive was created by hippie-mail 7.98617-22 on Fri Aug 21 1998 - 17:20:12 ADT