Path: news.mathworks.com!newsfeed-00.mathworks.com!NNTP.WPI.EDU!elk.ncren.net!newsflash.concordia.ca!canopus.cc.umanitoba.ca!not-for-mail
From: roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson)
Newsgroups: comp.soft-sys.matlab
Subject: Re: Pseudo random generator
Date: Tue, 1 May 2007 20:21:06 +0000 (UTC)
Organization: National Research Council Canada - Conseil national de rechereches Canada
Lines: 24
Message-ID: <f187fi$1gf$1@canopus.cc.umanitoba.ca>
References: <ef557ed.-1@webcrossing.raydaftYaTP> <ef557ed.3@webcrossing.raydaftYaTP>
NNTP-Posting-Host: origin.ibd.nrc.ca
X-Trace: canopus.cc.umanitoba.ca 1178050866 1551 192.70.172.160 (1 May 2007 20:21:06 GMT)
X-Complaints-To: abuse@cc.umanitoba.ca
NNTP-Posting-Date: Tue, 1 May 2007 20:21:06 +0000 (UTC)
Originator: roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson)
Xref: news.mathworks.com comp.soft-sys.matlab:406295



In article <ef557ed.3@webcrossing.raydaftYaTP>,
Anish Goorah <anish.goorah@gmail.com> wrote:
>All random numbers generated are by definition pseudo random.

If I have a /dev/random system interface that produces true 32 bit
random unsigned long (e.g., it interfaces to a geiger counter or lava
lamp), and I read two values from there and use them to create a
floating point number with 53 bit mantissa, then arguably
I have "generated" the floating point number, but the floating
point number would not be pseudo random.

The matlab randn() algorithm uses pseudo random pairs (or sometimes
additional values are needed) to generate normally distributed
numbers. If the pseudo-random number generation underneath were
replaced with a true random call (e.g., geiger counter, thermal
noise, whatever was available), then although the actual random
values employed might be uniformly distributed, the normal values
output would have to be said to be "generated" -- but they would
still be random, not pseudo random.

The problem here is that "generated" has not been well-defined.

-- 
Programming is what happens while you're busy making other plans.