Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!mit-eddie!uw-beaver!uw-june!entropy!martin
From: martin@entropy.ms.washington.edu (Don Martin)
Newsgroups: sci.math,sci.physics,sci.crypt,sci.math.stat
Subject: Re: Do you use RANDOM NUMBERS?
Message-ID: <553@entropy.ms.washington.edu>
Date: Wed, 18-Mar-87 23:34:59 EST
Article-I.D.: entropy.553
Posted: Wed Mar 18 23:34:59 1987
Date-Received: Fri, 20-Mar-87 04:51:34 EST
References: <5712@reed.UUCP>
Organization: UW MathStat, Seattle
Lines: 33
Keywords: Random numbers, leprechauns
Summary: Actual uses of random numbers
Xref: mnetor sci.math:753 sci.physics:932 sci.crypt:254 sci.math.stat:70

In article <5712@reed.UUCP>, mdr@reed.UUCP (Mike Rutenberg) writes:
> How do people currently use RANDOM NUMBERS?  Simulations?  Anything else?
> 
> How do you generate them?
> 
> Do you think your results would benefit from using
> truly random numbers rather than pseudo-random stuff?

For most applications real random numbers are undesirable.
Consider debugging a monte-carlo or simulation program
when you cannot repeat the sequence that led to the
problem. Also, the better psuedo random number generators are
quite good for most applications. Thus there is little
motivation to come up with hardware to substitute for a 
short simple and reasonably fast program. ( the time spent
computing random numbers is usually only a small fraction
of the program time. Thus speeding up random number generation
does not usually result in much overall improvement. )

It is very hard to get a really good hardware random number
generator. Very small effects can signifcantly bias the
results. I tried breadboarding a fast counter and using the
time that I held down a switch to get a random number and
found that even this was biassed. Try reading the intro to
1000000 Random Digits ( one of the few fictional works 
published by RAND Corp ;-) . This will give you an idea of 
of the problems.

> Is it of enough interest to justify some time and effort?
My guess is prbably not. There have been one or two attempts
to do this and sell one time pad systems.

Don Martin   martin@entropy