[ILUG] random number generation

[Brian Foster]

  | From: JMcCarthy at mortgagelink.ie
  | Date: Wed, 21 Jun 2006 13:02:39 +0100
  | 
  |[ What is ] The best way to generate a random number without using a
  | computer (aka I [ ... ] am bored in work & want to do a few math
  | equations).

 measure an unpredictable physical process; e.g.,
 radioactive decay, lava lamps (previously mentioned),
 Brownian motion, &tc.  “measure” can be interpreted
 in different ways (may depend on the process); e.g.,
 time between events, number of events in a second,
 “location” of an event, &tc.

 solving an equation seems unlikely to give a sequence
 of random values.  nominally, you'd get the same values
 each time you solve it.  (I suppose chaotic systems,
 such as 3 body orbits, might be a counter example — but
 that particular counter-example is also an unpredictable
 (over a sufficiently long time frame) physical process.)

 some solutions can be transformed into a random (or at
 least apparently random) sequence; e.g., whilst π is not
 random (it always has the same value), its digit sequence
 _seems_ to be “normal” (this has never been proven); that
 is, each of the ten digits 0-9 occurs about 1/10th of the
 time, each of hundred pairs 00-99 occurs about 1/100th
 of the time, and so on.  (random sequences are usually
 normal.)  computing the digits of π is a routine test of
 supercomputers (and famously broke an early Cray), and
 more than a billion (10^9) digits are known.  (and there
 is a algorithm (“BBP”, IIRC) for computing any binary
 digit of π without having to compute preceeding digits.)

cheers!
	-blf-
-- 
Experienced (20+ yrs) kernel/software Eng: | Brian Foster   Montpellier,
 • Unix, embedded, &tc;  • Linux;  • doc;  | blf at utvinternet.ie   FRANCE
 • IDL, automated testing, process, &tc.   |  Stop E$$o (ExxonMobile)!
Résumé (CV) http://www.blf.utvinternet.ie  |     http://www.stopesso.com