Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.ecn.bgu.edu!psuvax1!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!math.ohio-state.edu!jussieu.fr!fdn.fr!uunet!in1.uu.net!hasler!root
From: root@hasler.uucp
Subject: Re: wanted: FAST binary random number gene
Message-ID: <D5su6o.1tA@hasler.uucp>
Organization: Dancing Horse Hill
In-Reply-To: JEThomas@ix.netcom.com's message of 19 Mar 1995 20:00:54 GMT
Date: Tue, 21 Mar 1995 16:37:35 GMT
References: <3kckoe$d7j@Germany.EU.net>
Lines: 22

<3kgchv$bhv$1@mhadf.production.compuserve.com> <3ki2hm$r8q@ixnews2.ix.netcom.com>

Jonah Thomas writes:

> If you just get results from hardware, how do you
> validate them?  You could get a string of 1 & 0 from some piece of
> hardware, and record them, and find out that there were more 1's than
> 0's.  Should you adjust the hardware to get more 0's?  But maybe you
> were just having a run.  You'll be off 1 SD about a 3rd of the time,
> and off 2 SD about 5% of the time, and off 3 SD about 1% of the time
> and so on, if the hardware is set right.  Adjust it and you might be
> chasing a moving target.  Don't adjust it and you might be just
> wrong.  So to be safe you need a mathematical routine that you can

I have to disagree with this a little bit.  I think it is posssible to build a
hardware rng with a provable distribution.  I don't think that there is a market
for it, though.
-- 

John Hasler uunet!hasler!root
Dancing Horse Hill
Elmwood, Wisconsin USA
