DPRG List  

[DPRG] The state of the discussion

Subject: [DPRG] The state of the discussion
From: Randy M. Dumse rmd at newmicros.com
Date: Mon Apr 17 16:57:14 CDT 2006

> To be clear, I was using the formal (control theory)
> definition of "state":
> The state of a system is the information which is
> required to predict
> future outputs given future inputs.

Yes, I can concede that definition.

> More loosely, state is the
> information contained within a system.  This neatly
> encompasses the
> formal automata theory definition of state: both
> definitions represent
> the same thing if the system actually is a finite
> state machine or
> other automaton.

I don't like this loose definition as well, but I am not
strongly opposed to it either.

However, the nature of the information, is not the same in the
case of the FSM as compared to other systems. In FSM's the state
information (a single value, number or pointer) IS the system
information. There is NO other information in the FSM. In other
systems you can have charges on a cap, stretch of a string, and
a whole host of other analog values. In a FSM there is no
analgous array. Therefore the FSM is more fundamental of the
systems compared.

> What about something like a push-down automaton,
> which is strictly
> more powerful than a finite state machine?

I had to look them up. I don't think it is correct to call a
push-down automaton as strictly more powerful than a FSM. Would
you call a general purpose computer more powerful than a Turing
machine? Seems to me that would be nonsensical. Anything a
general purpose computer can do a Turing machine can do, which
is an accepted fact. So unless you have a problem a FSM cannot
do a a push-down automaton can, then I don't see the point in
saying either is more powerful.

> As I recall, that is not completely correct.  In a Moore
> representation of a state machine, the outputs are
> solely determined
> by the current state, but there is no requirement
> that each state
> produce different outputs.  Thus, the current output alone is
> insufficient to determine the current state of the machine.

Well, working from memory, I'd say that is quite wrong. Without
the time to research where I've read it, perhaps we'll agree to
disagree until better references can be found.

> ... The formal
> ideas that you are using are the basis of automata
> theory, which was
> designed to model the behavior of certain systems.
> Other tools to
> model systems include fuzzy logic, neural nets,
> kinematics, and
> newtonian and quantum physics.

Yes, but each model has areas of best use, and areas where they
match poorly. As you undoubtedly know, some areas of some models
(Quantum and Relativistic) have mutually exclusive properties,
and both cannot be right. But the broader the area where models
are applicable, very likely, the closer they are to being

> I would rather not assume that any system had an
> "intent," mostly
> because I don't want to have to figure out what
> "intent" is.

Great, that leaves millions more for me, or whoever cracks the
"intent compiler" in the future. I have no doubt someday there
will be a whole science of extracting intent, if for recoding
for efficiency if nothing else. In fact, if you think of the
progress made at efficiency in compilers, the conversion has
already begun.

> A system's behavior is independent of
> what tools we use to describe it, and describing it
> in one way does
> not invalidate others.

I quite disagree. If the system is compiled from the tools we
descibe it with, the system does not exist without the tools.
Then you get into the same problem as the observer in quantum
physics. The observer actually participates in the experiment,
and cannot be separated from the results. The same is true of
the tools. If the tools are used to make the system, the system
invariable depends on the tools used to create it. When they
become part of the loop, the system is no longer an independent
process, free of control.

> I don't believe state machines actually exist.  I
> believe they are a
> useful abstraction of the behavior of some systems.
> Thus, there's no
> state machine in there somewhere, rather it is in
> your mind (not that
> that's a bad thing).

Utter Sacrilege!

Now how do I support my ire?

Since this reply is about a month later than your posting, I've
had some time to think of it. I still say the validity of the
state machine approach models nature at some very deep levels.
At the heart of many physics phenomina lies quiesence, in the
sense, of a quiet state that exists over time. The Bohr atom
model touches this. Electrons take up energy with rules, they
cannot release that energy without rules. When they do, they
change dramatically. There is no analog continuous nature to it.

Half lifes of nuclear decay, absorpsion or emission of
electromagnetic waves/particles depend on these state
transitions. Many many Nobel prizes have been awarded to the
experiments that show these quantum effect, starting just about
100 years ago, including Einstein's (not for relativity as some
think) and others. To my mind FSM's reflect the reality that
some processes, the deeper ones all the rest of reality is based
upon, have discontinuous natures, and change abruptly, and only
when certain conditions are met. To me a FSM is a model of this
fundamental reality, and much less an artifical construct we
made up to do control circuits. Fuzzy logic? Yeah, that's very
artifical. Neural Nets ... well not so artifical, because the
model is a model of what nature herself did in one instance to
make brains. But to me, FSM's are models of the deepest physics
we know, and therefore much more fundamental to how reality at
large works.

To me I look at a previously living creature, and see in death,
a loss of state information. No matter what "shortage" (water,
air, etc.) has caused the death, replacement of the missing
material will not start the life again. Yet nothing is missing,
all the components are still present the moment after death.
What has left is the state of all the various processes, now no
longer in alignment, no longer in control.

The living soul may well be state information and not else.


More information about the DPRG mailing list