[From Bill Williams 22 June 02 7:40 CST ]
For whatever reason the EXE files don't seem to travel well. Oh well.
Bill Williams
______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/
[From Bill Powers (2002.06.22.2039 MDT)]
[From Bill Williams 22 June 02 7:40 CST ]
For whatever reason the EXE files don't seem to travel well. Oh well.
I believe I have received all the .exe files now. They all appear to run,
through I'm not sure what I'm looking at. I've been out of the house today.
Best,
Bill P.
[ From Bill Williams 23 June 02 3:00 AM CST ]O
[From Bill Powers (2002.06.22.2039 MDT)]
>[From Bill Williams 22 June 02 7:40 CST ]
>
>
>For whatever reason the EXE files don't seem to travel well. Oh well.I believe I have received all the .exe files now. They all appear to run,
through I'm not sure what I'm looking at. I've been out of the house today.Best,
Bill P.
Ok, so the EXE Files did get through. It didn't appear from my email service
that they had.
What the posting was supposed to communicate was the experience I had as a
result of making a mistake in attempting to construct a ring of control loops
which are interconnected. At first (EB.EXE Example Bad )when I made the
mistake of not applying the
disturbance to all the elements in the ring before adding up the effects of the
disturbance I observed a somewhat subtle mistake. The ring that formed on the
screen was somewhat mis-shapened when it was undisturbed. And, when a
disturbance was applied the pulse travelled around the ring in one direction.
I had difficulty understanding what was going on and difficulty identifying
where the code was wrong.
In ( EG.EXE Example Good ) I corrected the problem and the ring that forms is
regular without any lumps where the ends of the ring connect to each other When
a pulse is applied it travels equally in each direction around the ring.
Actually, since it doesn't actually travel, it might be more accurate to
say that as the growing oscilation which the disturbance creates grows equally
in each direction rathr than travels This was more like what I expected to see
at first and was disappointed but curious about the source of the irregular
ring the formed and puzzled by the wave of oscilations that travel in only one
direction around the ring.
Then in BIAS.EXE by changing the gains in the ring so that it is connected more
strongly in one direction around the ring than the other I had a ring that the
regular in form, but for which ( by my intension ) a pulse applied to the ring
travels around the ring in one direction. I still a bit puzzled as to why this
happens. The ring is connected up bothways around the ring. I could more
easily see the pulse travelling more strongly in one direction than the other
but, the way the ring behaves it travels only one way. I'll probably figure it
out, there has to be a perfectly simple explaination, and it may or may not be
worthwhile making the effort to find out.
My main point is that even very simple combinations of control loops can
produce behaviors which are to the niave surprizing and perplexing. And, as an
example of this-- if I'm not mistaken changing the number of units in the ring
has an effect upon the ring's stablity.
I suppose I should point out that contrary to what might be mistakenly
attributed to the behavior of the ring, in a sense it has no material reality.
It's just a collection of control loops without an underlying model of a
physical system that is being controlled. But, regardless of that I find the
behavior of such comparatively simple interactions of loops facinating.
Bill Williams
______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/
[From Bill Powers (2002.06.23.0848 MDT)]
Bill Williams 23 June 02 3:00 AM CST --
>My main point is that even very simple combinations of control loops can
produce behaviors which are to the niave surprizing and perplexing. And, as an
example of this-- if I'm not mistaken changing the number of units in the ring
has an effect upon the ring's stablity.
I agree on the point of getting amazingly complex behavior from what is
really a simple underlying organization. That's a crucial point about PCT
and behavior, and indeed about any working model in relation to the
behaviors it can produce. Stephen Wolfram, in his new book "A new kind of
science," makes the same point, though his model is restricted to cellular
automata (Mike Acree has just sent me a copy to look at, for which thanks).
Behavior is always, I think, far more complex than the structure of the
system that produces it. Much of what we think of as the complexity of
behavior is the sheer multiplicity of the environmental variables that
impinge on behaving systems. Behaving systems are always simpler than the
behavior they create, I would guess.
I suppose I should point out that contrary to what might be mistakenly
attributed to the behavior of the ring, in a sense it has no material reality.
It's just a collection of control loops without an underlying model of a
physical system that is being controlled. But, regardless of that I find the
behavior of such comparatively simple interactions of loops facinating.
I agree that there's a lot to be learned just from studying the behavior of
control system models for their own sake.
One thing I might suggest. When two control systems interact, each one's
output affects the other one's input, and because of that it also affect
the other one's output (which opposes the disturbance), which in turn
affects the first systems input. So there is a feedback loop from one
system's output, through the other system, and back to the first system's
input again. This loop acts in parallel with the direct effects of the
first system's output on its own input.
If you draw such a pair of control systems, you can explore the effects of
various signs of the action of each system on the other. If both systems
are negative feedback systems, the change in each one's output will be the
negative of the disturbance applied to its input. Assuming the two system
are identically organized, this means that the effect of the output of one
system on its own input via the other system will be inverted, so the
feedback effect is positive, tending to counteract the negative feedback
effect of the direct connection between its output and its own input.
This, in turn, tells us something about the stability of a multi-system
organization of identical control systems. There is a limit to the amount
of mutual interaction that can exist without causing a net positive
feedback and instability of the whole system. I don't know what the limit
is for the specific circular organization you have programmed, but the
whole system must be somewhat past that limit, for if I let the "good"
program run indefinitely (without hitting any keys), it gradually changes
form and seems to be heading toward infinity. Also, alternating systems
have opposing effects on the controlled variables (the zig-zags), and the
opposition just gets larger and larger, which implies that there is
positive feedback.
This behavior will be very sensitive to the magnitude of the interaction
constants. If you start with zero interactions, each system will simply
control its own controlled variable and bring it as close to its reference
level as the system gain permits. Then, as you increase the magnitude of
the interactions from zero, there will be more and more mutual effect,
until finally the verge of instability is reached, where I think your model
is working when it is first turned on.
Best,
Bill P.
[From Bill Williams 23 June 02 3:10 CST ]
reply to Powers,
[From Bill Powers (2002.06.23.0848 MDT)]
Bill Williams 23 June 02 3:00 AM CST -
I'm finding Wolfram's book interesting, unfortunate that it wasn't an equally
extended exposition of control theory. Reading Steven Gould's STructure
of evolutionary theory I found about 30 mentions of regulation, homeostatis and
control almost none of which were indexed. Gould's remains worthwhile but
irritating reading despite his not recognizing the phenomena of control as an
important issue.
I think my difficulty with assemblies of loops is the result of my falling back
into an input-output frame of mind when faced with more than one loop.
Bill Williams
______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/
They were received at my end, almost. I receive my email via my employer.
We have a firewall and email security that blocks .exe files. I received
the source code ok.
Steve O
-----Original Message-----
From: William Williams [mailto:w.d.williams@EMAIL.RO]
Sent: Saturday, June 22, 2002 7:39 PM
To: CSGNET@listserv.uiuc.edu
Subject: Re: file transmission
[From Bill Williams 22 June 02 7:40 CST ]
For whatever reason the EXE files don't seem to travel well. Oh well.
Bill Williams
______________________________________________________________________
Do you want a free e-mail for life ? Get it at http://www.email.ro/