From Mary Powers (2002.07.04)
Happy 4th of July! It actually rained here - and the fires are getting
under control.
But my motive for writing is this from Bill:
[From Bill Powers (2002.07.02.2031 MDT)]
Success. If you don't have Dag's PCT CD-ROM, get it. Simcon 4.5 is in
Forssell - 1995 disks and threads/SIMCON
There is a complete writeup including a definition for each kind of
computing block, descriptions of all program directives, and a tutorial on
writing, running, and plotting SIMCON simulations. This is the version in
which a Simcon program consists of a file of ASCII lines that follow a
specific format.
The only thing lacking is the source code, unfortunately. Any programmer
should be able to implement all of the features of the program except one:
the algorithm for automatically figuring the order of precedence in blocks
that are "grouped", an operation that makes all the computations within a
single group execute in one iteration (rather than each block in a series
passing its output to the next during one iteration). Wolfgang Zocher
worked out this algorithm and I don't know how it works. The algorithm
isn't absolutely essential -- the programs will run without it, but there
can be subtle untruths in the results. Nothing to worry about in simple
programs.
I will continue to search for the source code of Simcon 4.5 (it's in ANSI
C), If anyone else has it stashed somewhere on an old disk, now is the time
to resurrect it. The main reason for needing the source is to see how
Wolfgang optimized the grouping. I'm pretty hazy on how he got the thing to
work with GNUPLOT, too.
Bill Williams, could there be a version of Simcon on your laptop in the
directory \TC?
I strongly advise reading simcon.txt, especially the section explaining
grouping, which also explains how Simcon acts as a parallel computer, all
computing blocks executing simultaneously (in effect).
It will become clear after you've read this document that Simcon is a
simulator for low-order systems. Giving it the power to simulate
higher-order control will be a considerable challenge -- in fact you could
say that this task defines the problems for the next century of development
of PCT.
I think we will have fun at the meeting discussing the specifications in
the document and adding whatever new ones seem reasonable for the near
future. One thing that occurs to me is that we'll want to organize any
additions so the existing system can be treated as a lower level, with the
higher systems superimposed. Like the alligator ridden by the horse ridden
by the monkey, although of course with 11 levels or so instead of only 3.
I guess that signatures don't work with the cheap (free) version
of Eudora. So here it is -- I don't plan to go on repeating the quote very
long -- anything you see too many times gets tiresome, and typing it is
even more so.
Best,
Bill P.
"Pay no attention to that man behind the curtain"
The Wizard of Oz
* * *
Now from Mary P.
While working with SIMCOM and other modelling languages will indeed be fun
for some who attend the meeting, there are no doubt others whose expertise
and interests lie elsewhere. So I would just like to say that modelling
meetings will probably take place during "free" afternoons. I think that a
relatively non-technical discussion of what modelling is about would be
appropriate for a regular session. It's nice to look at demos such as
Bill's and Ricks's, but I'm not sure if there has been much discussion
recently about the structure producing the demos - the programming that
converts a digital computer into a hierarchy of control systems. Some sense
of the control systems that underlie models and demos helps to support the
concept that similar arrangements - PCT - underlie the behavior we see of
living systems. That is the point of doing the modelling.