Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!ihnp4!ihopa!riccb!jmc
From: jmc@riccb.UUCP (Jeff McQuinn )
Newsgroups: sci.physics
Subject: Re: Re: Analog models of computation
Message-ID: <753@riccb.UUCP>
Date: Tue, 21-Oct-86 05:50:29 EDT
Article-I.D.: riccb.753
Posted: Tue Oct 21 05:50:29 1986
Date-Received: Wed, 22-Oct-86 04:58:46 EDT
References: <8195@watrose.UUCP> <3505@columbia.UUCP>
Distribution: net
Organization: Rockwell Telecommunications, Downers Grove,Il.
Lines: 28

> There are many problems with "analog type solutions". In some cases we are
> interested only in the character of the solution (is the system stable?
> will it oscillate? is the output linear or exponential? etc.). There are
> other applications where "analog" solution gives us required precision.
> Many of these problems are too complex for digital computers and can be quite
> satisfactorily solved on devices known as "analog computers".
> I won't go into details, describing an analog computer, because everybody
> knows or can look at a representative - analog musical synthesiser.
> Similar devices are used in simulations of complex mechanical systems.
> 
> zdenek

One of the most fundemental approaches to engineering is to build a prototype.
The prototype is in effect an analog model for computation.  It becomes
extraordinarily difficult to write software to simulate every nuance of a 
design when your not even sure what all your problems may be.  Let's take
an example of an extremely complex design which would be practically impossible
to simulate on a digital computer.  The Connection Machine tm.  Computer models
are great when it comes to doing many simulations on a well understood problem
where a few variables are changed on each run.  It can save both time and
dollars.  Analog models are unparalleled for finding that unexpected or little
understood problems exist.  The trade off is of course time and money to 
construct the prototype.  Almost no one designs anything anymore without using
both tools.  Even basic research uses both tools extensively.  We are building
ever larger accelerators in an effort to model the conditions during the
theroretical Big Bang.  These are analog models for computation.

					Jeff McQuinn just VAXing around