What is an ECU

CSGnet Archive CSG2004
1

[Martin Taylor 2004.12.1011.00.10]

[From Rick Marken (2004.12.10.1600)]

I'm still wondering what an ECU is. What is an ECU, Martin? Is an ECU
defined by the nature of the output function? Is a proportional controller,
where o = ke, not and ECU?

An ECU is an Elementary Control Unit. It controls a single (real)
valued perceptual signal. The single-valued perceptual signal is the
output of a perceptual input function that may have many inputs. It
is compared with a single-valued reference signal and the
single-valued error signal that results from that comparison provides
the input to an output function whose output influences the world
outside the ECU. That world affects the possibly many inputs to the
perceptual function of the ECU.

In other words, an ECU is what you usually draw when you talk about a
simple control system. The nature of its perceptual function and its
output function don't matter. What matters is that it controls a
perceptual signal that at any moment can be specified by one real
number.

What it is distinct from is a complex control system, such as a
hierarchic structure of ECUs, or a control system for a
multi-dimensional or complex-valued perceptual signal.

The "classic" HPCT structure consists of many (11?) layers of ECUs,
each of which controls a single-valued perceptual signal that is
generated by a perceptual function that may have many inputs from
lower-level ECUs or from sensors.

Does that help?

Martin

2

an ECU is also I believe the French monies of the Middle-Ages.

Sincerely

I enjoy your post, esp. the one on the sped of the hand (some months ago)...
like when addicts say: It was not me but my hand...

please post if you wish

(Dr. Lansky is a friend)

paule

···

----- Original Message -----
From: "Martin Taylor" <mmt-csg@ROGERS.COM>
To: <CSGNET@listserv.uiuc.edu>
Sent: Saturday, December 11, 2004 12:22 AM
Subject: What is an ECU

[Martin Taylor 2004.12.1011.00.10]

[From Rick Marken (2004.12.10.1600)]

I'm still wondering what an ECU is. What is an ECU, Martin? Is an ECU
defined by the nature of the output function? Is a proportional
controller,
where o = ke, not and ECU?

An ECU is an Elementary Control Unit. It controls a single (real)
valued perceptual signal. The single-valued perceptual signal is the
output of a perceptual input function that may have many inputs. It
is compared with a single-valued reference signal and the
single-valued error signal that results from that comparison provides
the input to an output function whose output influences the world
outside the ECU. That world affects the possibly many inputs to the
perceptual function of the ECU.

In other words, an ECU is what you usually draw when you talk about a
simple control system. The nature of its perceptual function and its
output function don't matter. What matters is that it controls a
perceptual signal that at any moment can be specified by one real
number.

What it is distinct from is a complex control system, such as a
hierarchic structure of ECUs, or a control system for a
multi-dimensional or complex-valued perceptual signal.

The "classic" HPCT structure consists of many (11?) layers of ECUs,
each of which controls a single-valued perceptual signal that is
generated by a perceptual function that may have many inputs from
lower-level ECUs or from sensors.

Does that help?

Martin

3

[From Rick Marken (2004.12.11.1220)]

Martin Taylor (2004.12.1011.00.10) --

An ECU is an Elementary Control Unit. It controls a
single (real)valued perceptual signal.

What it is distinct from is a complex control system,
such as a hierarchic structure of ECUs, or a control
system for a multi-dimensional or complex-valued perceptual
signal.

A hierarchical structure of control systems is not really a control system,
is it? It's a hierarchy of control systems. And a control system for
multi-dimensional or complex-valued perceptual signals is kind of like the
unicorn, is it not? You are the only one I know who has actually seen one.
I won't believe it exists until I see it implemented as a set of
differential equations or, better, as a computer program.

I think the notion of an ECU creates the false impression that there is a
distinction between control systems that doesn't really exist.

The "classic" HPCT structure

I think this gives the impression that HPCT a soft drink rather than a model.

Best

Rick

···

---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400

4

[Martin Taylor 2004.12.11.16.52]

[From Rick Marken (2004.12.11.1220)]

Martin Taylor (2004.12.1011.00.10) --

An ECU is an Elementary Control Unit. It controls a
single (real)valued perceptual signal.

What it is distinct from is a complex control system,
such as a hierarchic structure of ECUs, or a control
system for a multi-dimensional or complex-valued perceptual
signal.

A hierarchical structure of control systems is not really a control system,
is it? It's a hierarchy of control systems.

A question of terminology. You say that a structure of Elementary
Control Systems doesn't control, because it's controlling a lot of
perceptions at once. I call that a control system, but not an
elementary one.

And a control system for
multi-dimensional or complex-valued perceptual signals is kind of like the
unicorn, is it not? You are the only one I know who has actually seen one.

You might benefit by expanding your circle of acquaintances.

Here's one described in equations (all values are complex, meaning,
for example, P = p +iq, where i = sqrt(-1)):

P = S
E = R - P
O = G*E
S = D + O

Of course, the same equations work, and are more descriptive, if the
symbols represent Laplace Transforms. But anyway, now you and I are
two people who have seen a control system that controls a
complex-valued perception.

I leave it as an elementary exercise for the reader to duplicate
these equations with different symbols (making two ECUs) and imagine
them operating concurrently. That is a control system that controls a
two-dimensional variable.

I think the notion of an ECU creates the false impression that there is a
distinction between control systems that doesn't really exist.

Whatever you care to name them, there is a difference between a
structure that controls a single-valued real variable and an
interconnected network or hierarchy of such elemental units. It just
happens to be very convenient to call the simple one that serves as
the element in the structure an Elementary Control System, and the
complex structure a complex control system.

> The "classic" HPCT structure

I think this gives the impression that HPCT a soft drink rather than a model.

Well, it isn't really a model, is it, in the sense that you and Bill
and several others use the term? It's a proposed structure that could
be the basis for a model, and that has been instantiated in several
models, not least by yourself.

Martin

5

[From Rick Marken (2004.12.11.1650)]

Martin Taylor (2004.12.11.16.52) --

Rick Marken (2004.12.11.1220)--

And a control system for
multi-dimensional or complex-valued perceptual
signals is kind of like the unicorn, is it not?
You are the only one I know who has actually seen one.

You might benefit by expanding your circle of acquaintances.

Here's one described in equations (all values are complex,
meaning,for example, P = p +iq, where i = sqrt(-1)):

P = S
E = R - P
O = G*E
S = D + O

Of course, the same equations work, and are more descriptive,
if the symbols represent Laplace Transforms. But anyway, now
you and I are two people who have seen a control system that
controls a complex-valued perception.

I don't believe that two different ways of representing the variables in a
control loop (real vs complex, for example) means that you have two
different kinds of control system. For example, the variables in a simple AC
circuit (an AC source with a single resistor) can be represented as real or
complex variables. But the circuit is the same in both cases. It is not an
elementary circuit when described in terms of real variables and a complex
circuit when described in terms of complex variables. I think the same is
true for a control loop.

I leave it as an elementary exercise for the reader to duplicate
these equations with different symbols (making two ECUs) and
imagine them operating concurrently. That is a control system
that controls a two-dimensional variable.

It sounds like two control systems, each controlling a single variable. A
single control system that controls a two dimensional variable can be easily
described in terms of real or complex variables. In such a control system p
= f( x1, x2), when x1 and x2 can be real variables _or_ complex conjugates.
The two dimensions of variation are in the input to the perceptual function
(x1 and x2) not in the output, which is a scalar variable that can also be
represented by a real or a complex number.

I think the notion of an ECU creates the false impression that there is a
distinction between control systems that doesn't really exist.

Whatever you care to name them, there is a difference
between a structure that controls a single-valued real
variable and an interconnected network or hierarchy of
such elemental units.

I think it is just confusing to talk about a structure that controls an
interconnected network or hierarchy elemental control units (systems). Each
control system in a hierarchy controls a single-valued variable (whether
that variable is represented by real or complex numbers) which is the
perceptual variable. The perceptual variable in any system may be (and
typically is) a function of many inputs.

Well, [HPCT-RM] isn't really a model, is it, in the sense
that you and Bill and several others use the term? It's a
proposed structure that could be the basis for a model, and
that has been instantiated in severalmodels, not least by
yourself.

I'll buy that! I'll have a nice, tall (11 level) HPCT classic on ice (it was
85 degrees out here LA today; thank you, President Bush, for this lovely,
global warming day;-))

Best

Rick

···

---
Richard S. Marken
marken@mindreadings.com
Home 310 474-0313
Cell 310 729-1400