Message-Id: <3.0.6.32.19990218103218.008d7180@pop.globalserve.net>
X-Sender: derbyshire AT pop DOT globalserve DOT net
X-Mailer: QUALCOMM Windows Eudora Light Version 3.0.6 (32)
Date: Thu, 18 Feb 1999 10:32:18 -0500
To: djgpp AT delorie DOT com
From: Paul Derbyshire <pderbysh AT usa DOT net>
Subject: Re: An inline assembler RNG for C
In-Reply-To: <20yalvsvea.fsf@Sky.inp.nsk.su>
References: <George Marsaglia's message of "Wed, 17 Feb 1999 13:35:29 -0500">
 <36CB0BF1 DOT 23FF9E2 AT stat DOT fsu DOT edu>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Reply-To: djgpp AT delorie DOT com

At 07:18 PM 2/18/99 +0600, you wrote:
>Can it be proved that there is no situation when x(i+1)=x(i) and
>c(i+1)=c(i), and generated sequence will not degrade to x(i)?

The original poster stated as much...indeed, he stated the repeat time
should be a very large prime number in the same ballpark as 2^32.


-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  |http://surf.to/pgd.net
_____________________ ____|________     Paul Derbyshire     pderbysh AT usa DOT net
Programmer & Humanist|ICQ: 10423848|