Message-ID:	<3517F669.53B2@pobox.oleane.com>
Date:	Tue, 24 Mar 1998 19:07:37 +0100
From:	Francois Charton <deef AT pobox DOT oleane DOT com>
Organization:	CCMSA
MIME-Version:	1.0
To:	"Salvador Eduardo Tropea (SET)" <salvador AT inti DOT gov DOT ar>
CC:	djgpp AT delorie DOT com
Subject:	Re: real random numbers
References:	<m0yHU7T-000S2iC AT inti DOT gov DOT ar>

Salvador Eduardo Tropea (SET) wrote:
> 
> I wrote:
> 
> > I really doubt that such an AD system would produce a desirably random
> > series of numbers, ie a series which pass classical tests for randomness.
> > Has anyone tries it?
> 
> Are you kidding! The noise generated in a diode junction IS NOISE and is
> generated by the combination of electrones, if you can predict it ... man
> you'll gain the next 100 Nobel prices ;-))). 


Probably. But unpredictable does not mean random. Many phenomena known as 
"noisy" or "chaotic", though absolutely unpredictable, still can show 
intrinsic correlations which disqualify them as random sources. If I 
remember correctly, Knuth even quotes the example of time between blips 
on a Geiger counter (quantic noise, right?) as one of a *bad* RNG... In 
mathematics, the series of primes is a well known example of non 
predictable series (ie there is no formula giving the value of the next 
one from that of the previous), however, it is known to possess many 
statistical characteristics which do not make it predictable, but ruins 
it as a "random" series.

Randomness is a pure mathematical concept. The only way to know whether a 
series is random or not is to test it. Hence my question: has anyone 
tried to test such a "natural noise series" to classical tests for random 
numbers (Knuth's spectral test for instance)? It would be a good way to 
know whether such natural noise sources make good RNG or not (but again, 
I suspect the latter).

Regards,
Francois

- Raw text -