Re: How do you design a circuit?

John Larkin wrote:
On Mon, 05 Jun 2006 17:16:04 GMT, "Kevin Aylward"
<see_website@xxxxxxxxxxxxx> wrote:

John Larkin wrote:
On Sun, 04 Jun 2006 08:33:25 GMT, "Kevin Aylward"
<see_website@xxxxxxxxxxxxx> wrote:


You need to stand back and be honest with yourself to understand
what it is that you do and what you really are. Have a deko at "The
Selfish Gene" by Richard Dawkin.


You need to stand back and explain why I have designed thousands of
original circuits and sold hundreds of millions of dollars worth of
them.

Its trivial to explain "original circuits", see below.

Hint: Darwin was right about microevolution


(which is what you
say your designs are) but wrong about macroevolution

No chance at all. What, are you one of those religious nuts now?

(which many of
mine are.)


With al due respect, John, you design skills are not a substitute
for a lack of knowledge as to what is the cause of your design
skills.

From,

http://www.anasoft.co.uk/replicators/intelligence.html


Illusion of Creativity

Since, the brain is an Intelligent Replicator instructed by its genes
and memes, all artistic, musicianship, inventiveness, etc must be the
result of a random process. That is any new idea (trait), must be
due to a randomly generated trait.

Support for the claim:

There are two options

1 A new idea is derivable from existing ideas - i.e. not
intrinsically new

2 A new idea is not derivable from existing ideas - i.e.
intrinsically new

If "1" is correct for an idea, then by construction it is not a new
idea, it is simply a logical consequence of existing traits.

If "2", the idea is new, that is not derivable from existing traits,
then that idea must therefore be non predictable. However, a non
predictable trait is, essentially, a random trait, by definition.

Thus, those of us that believe that having a "bright" idea is
something to be intrinsically valued, actually value a coin toss as
a "superior" way to go. Indeed, it is being able to derive
unsuspected results from existing knowledge that would appear to
have the greater worth.

So, creativity is nothing more that the brains randam trait generator
feature of its Intelligent Replicator kicking in.


But you're just playing with words and definitions.

No. I am showing what must be happening from basic principles.

What's really
important here is coming up with a model for design that actually
works and that serves as a guide toward practices that actually
produce new and better circuits *and* improve over time [1]. Anything
else is just fooling around with words, which is fun but doesn't help
sell electronics.

Its defacto that anything new is open to lots of potential for things to
go wrong. The de-facto best way to design new products is to design it
based on existing well tried and trusted techniques known to work Sure,
one often has to introduce something a bit new, but engineers that
*really* know their stuff minimise the amount of new ideas. The "not
invented here" scenery is a sure recipe for disaster.


So how's this:

Consider a problem that has a finite number of possible solutions;
encode them all as a list of binary numbers. Suppose that we can
encode all possible solutions to a given problem in, say, 100 bits,
ie, there are 2^100 possible solutions. We can solve the problem if we
a) spin through all possible states of a 100-bit counter and b) we
have a "goodness" mask for recognizing and scoring each proposed
solution.

Rippling a 100-bit counter, one state at a time, might take a while.

A quantum computer could contain a 100-bit counter in the form of,
say, 100 atoms each of which has two states. If the atomic states are
quantum-entangled, the atoms do not assume any one code at an instant,
they can exist in all 2^100 states simultaneously: all possible
solutions are present, and we just need to apply the test mask to
collapse them to an actual solution. Small (as, say, 5-bit) quantum
computers have been constructed from Bose-Einstein states, and they
hold promise for solving problems like finding huge prime numbers.
This is not sci-fi.

I am quite aware of the notion of quantum computing. I have a little
issue with it as I don't see how it quite fits in with the quantum
ensemble
interpretation, which accounts for quantum mechanics without any need
for a objects to physically exist in more than one state at once.

http://www.anasoft.co.uk/quantummechanics/index.html

Apparantly, though, according to Ballentine, QC is still allowed for.


A human brain is probably a quantum computer;

I dont agree that the basic feature of a brain is quantum based. Much
research has been done that identifies that the brain operates by a
Darwinian
Genetic Algorithm. I agree that to generate true randamness, quantum
features must be involved.

I know Roger Penrose thinks that Quantum is significant, but there is no
accounting for pure mathematicians getting their noses stuck into human
behaviour.

properly set up, it can
evaluate not one solution at a time, but an enormous set of them,
perhaps all of them. This is very different from evolutionary
algorithms that would, in effect, start with a mediocre 100-bit
solution and randomly fiddle bits to see if it got any better.

What evidence to you have that the brain operates in this manner? We
actually have extensive evidence that the brain operates by Genetic
Algorithms. we used this knowledge to solve real life problems with such
techniques.

Secondly, the genetic algorithms of the brain, to the best of my
knowledge do operate in parallel.


Quantum computing is probably how the brain achieves incredible speed
using millisecond logic elements.

Again, this is really just speculation. I have no trouble in accepting
that billion neuron circuits can result in consciousness. QC is just
another complication. Its the Occams razor thing.

The brain is not actually particularly fast in many aspects. In fact it
has to re-assemble things to make them seem sequential. e.g. see.

http://www.susanblackmore.co.uk/Articles/jcs02.htm

Autistic savants do incredible math
operations at speeds inconsistant with sequential millisecond-speed
algorithms.

No one is claiming that the brains neurons act purely sequentially.

It took thousands of gigahertz-speed CPUs to play a
masters-level chess game, and chess is just a fairly simple, finite
algorithmic thing. So what's going on inside a chess master's brain?

A parallel genetic algorithm by my book.


This train of thought is of course not directly provable, but it is
testable in that it leads to deliberate approaches to problem solving
(ie, creating new circuits) that not only work impressively, but that
get better with practice. And there are the counter-cases, where
people do the opposite and consistantly design crap.


How did you lean what a cascode is? a Bessel filter? A switching
regulator?
The thermal noise formula? etc... The reality is that 99.9999% of what
we do in electronics, or anything else, is by using knowledge we learnt,
i.e. copied, i.e. replicated. Its daft to think that we are so wonderful
that we can reinvent the wheel for every design.

I'm sorry John, but the vast bulk of what we know is indeed copying from
existing knowledge, i.e. replication. We then modify that knowledge, and
select the best modifications. This is the Darwinian algorithm, and it
is the way the universe works. It that simple really.

Kevin Aylward B.Sc.
431infoEXTRACT@xxxxxxxxxxxxx
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

"There are none more ignorant and useless,than they that seek answers
on their knees, with their eyes closed"



.

Loading