FPGA Central

On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
<[email protected]> wrote:
>
>Just Cocky wrote:
>> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>> <spam@s[email protected]> wrote:
>>
>>>DNA mutates. This really does't expalin why some mutations survive and
>>>other s become exinct. The fact that DNA mutate supports an piece of
>>>Evolutionary theory but it doesn't corroborate it in whole.
>>>
>>
>>
>> The other piece, of course, is selection.
>>
>>
>>>I'm an engineer.
>>>
>>
>>
>> Have fun with Genetic Algorithms/Programming then:
>>
>> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>> (Paperback)
>> by Melanie Mitchell
>
>No one has shown that genetic algorithms do better that plain old branch
>and bound.
>

Who said anything about doing better?

>
>IMHO genetic algorithms are a misnomer.
>

You are definitely a lost cause.

>>
>> http://www.amazon.com/gp/product/026...lance&n=283155
>>
>>
>>>I like theories that are predictive in a more concrete way.
>>>
>>
>>
>> Like what?
>
>Do you normally snip in such a disengenous manner?
>
>Mechanics does a pretty good job of predicting the locations of
>kinematic objects. I've read that quantum electrodynamics has been
>verified by experiment to 12 significant figures. Need more?
>

If you were a little bit more informed, you'd know that quantum
mechanics has shattered the notion that location can be predicted.

I'm done with you. Too much arrogance and ignorance at the same time
makes me sick.

On Thu, 22 Dec 2005 11:15:15 -0500, Stan Pawlukiewicz
<[email protected]> wrote:
>
>Just Cocky wrote:
>> On Thu, 22 Dec 2005 09:58:10 -0500, Stan Pawlukiewicz
>> <[email protected]> wrote:
>>
>>>Just Cocky wrote:
>>>
>>>>(snip)
>>>>
>>>>The tests certianly exist. But they require access to very high energy
>>>>levels. The inability to test right now is essentially technological.
>>>
>>>I don't believe that this is correct.
>>>
>>
>>
>> Who cares what you believe?
>>
>>
>>>No one is talking about building
>>>larger particle smashers to verify string theory (as yet).
>>>
>>
>>
>> It's a technological impossibility as of now.
>>
>>
>>>You just took a leap of faith and asserted that a test certainly exists
>>>without any evidence that it may.
>>>
>>
>>
>> String theory makes *PREDICTIONS*. As such, the theory is falsiable
>> because experiments can be devised to test those predictions.
>>
>> For example:
>> http://www.superstringtheory.com/experm/exper4.html
>>
>> Why do you insist on your arrogant posture when it is quite obvious
>> you know nothing about the subject at hand?
>
>This is a quote from your citation:
>
>There is a good chance we could start to see superpartners in future
>particle experiments.
>
>"good chance" and "could" Don't talk to me about arrogance.
>

Total lost cause.

Just Cocky wrote:
> On Thu, 22 Dec 2005 10:43:16 -0500, Stan Pawlukiewicz
> <[email protected]> wrote:
>
>>Just Cocky wrote:
>>
>>>On Thu, 22 Dec 2005 09:44:29 -0500, Stan Pawlukiewicz
>>><[email protected]> wrote:
>>>
>>>
>>>>Just Cocky wrote:
>>>>
>>>>
>>>>>On Wed, 21 Dec 2005 17:22:16 -0500, Stan Pawlukiewicz
>>>>><[email protected]> wrote:
>>>>>
>>>>>
>>>>>
>>>>>>Rune Allnor wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>>Stan Pawlukiewicz wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>Just Cocky wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>>On Wed, 21 Dec 2005 15:48:50 -0500, Stan Pawlukiewicz
>>>>>>>>><[email protected]> wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>>Jerry Avins wrote:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>>Stan Pawlukiewicz wrote:
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>>Jerry Avins wrote:
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>...
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>>>Only theories that can generate testable predictions are properly
>>>>>>>>>>>>>subjects of science.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>How about string theory?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>String theory is now in its infancy. People find it interesting and are
>>>>>>>>>>>exploring its ramifications. If none of that leads ultimately to a
>>>>>>>>>>>testable prediction, it will fade away.
>>>>>>>>>>>
>>>>>>>>>>>Jerry
>>>>>>>>>>
>>>>>>>>>>So it should not be mentioned in any public school science class,
>>>>>>>>>>because it really isn't a science yet. It might be interpreted as the
>>>>>>>>>>establishment of a pasta deity.
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>String theory is a speculative theory. But it's scientific because it
>>>>>>>>>is, in principle, falsifiable.
>>>>>>>>
>>>>>>>>So if it's falsified, then it's a valid scientific theory? Please
>>>>>>>>explain this further. I don't get it.
>>>>>>>
>>>>>>>
>>>>>>>A theory is "scientific" if one can make some prediction based
>>>>>>>on the theory, and then design an experiment to see if the prediction
>>>>>>>holds. Some theories pass this test, others don't. What we know as
>>>>>>>"science" is the set of theories that repeatedly pass this test.
>>>>>>>
>>>>>>>Rune
>>>>>>>
>>>>>>
>>>>>>So by this criteria string theory is, what?
>>>>>>
>>>>>
>>>>>
>>>>>A speculative theory.
>>>>
>>>>I asked Rune the question. The question was if the theory was
>>>>scientific. All theories are speculative to some extent.
>>>>
>>>
>>>
>>>It is scientific (because it's falsiable) and speculative (because
>>>it's virtually impossible to do any experiments to disprove it right
>>>now.)
>>
>>What makes it falsifiable. You said that in principle its falsifiable.
>>What principle?
>>
>
>
> The theory is falsifibale because it makes predictions that can be
> eventually tested and proved to be wrong.

From the web site you posted:

"There is a good chance we could start to see superpartners in future
particle experiments."

The terms "good chance" and "could" are known as weasel words. They
might have said "good chance that we will" and saved a weasel word but
instead they used two.

The best you can say TODAY is that string theory might be falsifiable.
If you can't see that, it's your problem.



(big snip)

Just Cocky wrote:
> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
> <[email protected]> wrote:
>
>>Just Cocky wrote:
>>
>>>On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>>><[email protected]> wrote:
>>>
>>>
>>>>DNA mutates. This really does't expalin why some mutations survive and
>>>>other s become exinct. The fact that DNA mutate supports an piece of
>>>>Evolutionary theory but it doesn't corroborate it in whole.
>>>>
>>>
>>>
>>>The other piece, of course, is selection.
>>>
>>>
>>>
>>>>I'm an engineer.
>>>>
>>>
>>>
>>>Have fun with Genetic Algorithms/Programming then:
>>>
>>>An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>>>(Paperback)
>>>by Melanie Mitchell
>>
>>No one has shown that genetic algorithms do better that plain old branch
>>and bound.
>>
>
>
> Who said anything about doing better?
>
>
>>IMHO genetic algorithms are a misnomer.
>>
>
>
> You are definitely a lost cause.
>
>
>>>http://www.amazon.com/gp/product/026...lance&n=283155
>>>
>>>
>>>
>>>>I like theories that are predictive in a more concrete way.
>>>>
>>>
>>>
>>>Like what?
>>
>>Do you normally snip in such a disengenous manner?
>>
>>Mechanics does a pretty good job of predicting the locations of
>>kinematic objects. I've read that quantum electrodynamics has been
>>verified by experiment to 12 significant figures. Need more?
>>
>
>
> If you were a little bit more informed, you'd know that quantum
> mechanics has shattered the notion that location can be predicted.

It's actaully bound by an inequality and it works well enough on
macroscopic objects. Ever fly on an airplane.

>
> I'm done with you. Too much arrogance and ignorance at the same time
> makes me sick.
It doesn't seem to take much.

Bryan Hackney <[email protected]> writes:
> Macro-evolution is what I would call big steps required to get from
> amoeba to aardvark.
>
> If you think there are no big steps, and everything is just one big
> continuum from single cell to human, then I suggest that almost all
> the fossil record is missing.

And?

Since nearly all biologicals degenerate to nearly nothing
at the end of their lifetimes, what would you expect to
find?

As for not all the pieces having been found, following the
trail of evolution could be likened to assembling multiple
jigsaw puzzles that have been scattered by tornados and
then buried. The chances of finding any single piece is
small and determining which piece to which puzzle it is
is even smaller.

Just Cocky wrote:
> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
> <[email protected]> wrote:
> >
> >Just Cocky wrote:
> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> >> <[email protected]> wrote:
> >>
> >>>DNA mutates. This really does't expalin why some mutations survive and
> >>>other s become exinct. The fact that DNA mutate supports an piece of
> >>>Evolutionary theory but it doesn't corroborate it in whole.
> >>>
> >>
> >>
> >> The other piece, of course, is selection.
> >>
> >>
> >>>I'm an engineer.
> >>>
> >>
> >>
> >> Have fun with Genetic Algorithms/Programming then:
> >>
> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> >> (Paperback)
> >> by Melanie Mitchell
> >
> >No one has shown that genetic algorithms do better that plain old branch
> >and bound.
> >
>
> Who said anything about doing better?

Why would anyone use a GA if there were no benefit from using it?

> >
> >IMHO genetic algorithms are a misnomer.
> >
>
> You are definitely a lost cause.

GAs are mainly smoke and mirrors. They are cool to the student, and a
professor
collects plenty "impressor credits" with customers when suggesting
using GAs,
but when all is said and done they are almost always outperfomed by
simpler,
more robust methods.

There are several points that work in disfavour of GAs:
- There is no information of how good a soltion to the optimization
problem is. Unlike standard methods, no estimates for the Jacobian
and/
or Hessian are produced by a GA.
- The GA can run literally forever without finding anything close to
the
optimum solution for the optimization problem. The GA samples the
solution space in a random fashion, and does not exploit any
structure.
- The term "convergence rate" does not make sense in a GA. Hence,
there is no way of telling when a GA has run "sufficiently long." So
one is
left with "gut feeling" as to when to stop the thing.

Rune

On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
wrote:
>
>Just Cocky wrote:
>> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
>> <[email protected]> wrote:
>> >
>> >Just Cocky wrote:
>> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>> >> <[email protected]> wrote:
>> >>
>> >>>DNA mutates. This really does't expalin why some mutations survive and
>> >>>other s become exinct. The fact that DNA mutate supports an piece of
>> >>>Evolutionary theory but it doesn't corroborate it in whole.
>> >>>
>> >>
>> >>
>> >> The other piece, of course, is selection.
>> >>
>> >>
>> >>>I'm an engineer.
>> >>>
>> >>
>> >>
>> >> Have fun with Genetic Algorithms/Programming then:
>> >>
>> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>> >> (Paperback)
>> >> by Melanie Mitchell
>> >
>> >No one has shown that genetic algorithms do better that plain old branch
>> >and bound.
>> >
>>
>> Who said anything about doing better?
>
>Why would anyone use a GA if there were no benefit from using it?
>

What is your point, given this thread's context?

Stan Pawlukiewicz wrote:

> 1) The most ardent atheists I personally know, either went to Catholic
> schools and/or had strongly religious parents.

Yes, those Jesuits have much to answer for. ;-)

> 2) I don't ever recall that being skeptical was a winning strategy in
> public school in any science class. The most successful students where
> the ones who regurgitated what was feed them. I can only conclude that
> my science classes were trying to teach me how to be a penguin via metaphor.

Being skeptical was never a winning strategy in any high school (or
lower) course I ever took. Science, regurgitation, English Lit.,
regurgitation, French, regurgitation, Economics, regurgitation. The
closest I ever came to seeing skepticism was an hilarious piss-take of
the teacher during a poetry criticism --- it was an excrutiatingly
accurate regurgitation of precisely what the teacher would have said,
and how that specific teacher would have said it.

> Merry Christmas to all.

And to you and yours, Stan.

Ciao,

Peter K.

Just Cocky wrote:
> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
> wrote:
> >
> >Just Cocky wrote:
> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
> >> <[email protected]> wrote:
> >> >
> >> >Just Cocky wrote:
> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> >> >> <[email protected]> wrote:
> >> >>
> >> >>>DNA mutates. This really does't expalin why some mutations survive and
> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
> >> >>>
> >> >>
> >> >>
> >> >> The other piece, of course, is selection.
> >> >>
> >> >>
> >> >>>I'm an engineer.
> >> >>>
> >> >>
> >> >>
> >> >> Have fun with Genetic Algorithms/Programming then:
> >> >>
> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> >> >> (Paperback)
> >> >> by Melanie Mitchell
> >> >
> >> >No one has shown that genetic algorithms do better that plain old branch
> >> >and bound.
> >> >
> >>
> >> Who said anything about doing better?
> >
> >Why would anyone use a GA if there were no benefit from using it?
> >
>
> What is your point, given this thread's context?

What "context"? Stan presents his views and you bring GAs out of thin
air.
Stan comes up with a completely valid criticism agains GAs, and you
refute it as if the GAs, that you yourself introduced to the thread,
was
of no importance. I responded by explaining in somewhat more detail
why GAs are smoke and mirrors.

I can't se that there is much "context" to deviate from.

Rune

On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
wrote:
>
>Just Cocky wrote:
>> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
>> wrote:
>> >
>> >Just Cocky wrote:
>> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
>> >> <[email protected]> wrote:
>> >> >
>> >> >Just Cocky wrote:
>> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>> >> >> <[email protected]> wrote:
>> >> >>
>> >> >>>DNA mutates. This really does't expalin why some mutations survive and
>> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
>> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
>> >> >>>
>> >> >>
>> >> >>
>> >> >> The other piece, of course, is selection.
>> >> >>
>> >> >>
>> >> >>>I'm an engineer.
>> >> >>>
>> >> >>
>> >> >>
>> >> >> Have fun with Genetic Algorithms/Programming then:
>> >> >>
>> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>> >> >> (Paperback)
>> >> >> by Melanie Mitchell
>> >> >
>> >> >No one has shown that genetic algorithms do better that plain old branch
>> >> >and bound.
>> >> >
>> >>
>> >> Who said anything about doing better?
>> >
>> >Why would anyone use a GA if there were no benefit from using it?
>> >
>>
>> What is your point, given this thread's context?
>
>What "context"?
>

The Theory of Evolution, whose ideas are behind genetic algorithms,
genius!

Just Cocky wrote:
> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
> wrote:

...

>>>What is your point, given this thread's context?
>>
>>What "context"?
>>
>
>
> The Theory of Evolution, whose ideas are behind genetic algorithms,
> genius!

You've been sneering entirely too long. I'm surprised that anyone still
bothers to respond to you.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

On Thu, 22 Dec 2005 16:48:01 -0500, Jerry Avins <[email protected]> wrote:
>
>Just Cocky wrote:
>> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
>> wrote:
>
> ...
>
>>>>What is your point, given this thread's context?
>>>
>>>What "context"?
>>>
>>
>>
>> The Theory of Evolution, whose ideas are behind genetic algorithms,
>> genius!
>
>You've been sneering entirely too long. I'm surprised that anyone still
>bothers to respond to you.
>

Yeah, and what's your point?

Just Cocky wrote:
> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
> wrote:
> >
> >Just Cocky wrote:
> >> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
> >> wrote:
> >> >
> >> >Just Cocky wrote:
> >> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
> >> >> <[email protected]> wrote:
> >> >> >
> >> >> >Just Cocky wrote:
> >> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> >> >> >> <[email protected]> wrote:
> >> >> >>
> >> >> >>>DNA mutates. This really does't expalin why some mutations survive and
> >> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
> >> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
> >> >> >>>
> >> >> >>
> >> >> >>
> >> >> >> The other piece, of course, is selection.
> >> >> >>
> >> >> >>
> >> >> >>>I'm an engineer.
> >> >> >>>
> >> >> >>
> >> >> >>
> >> >> >> Have fun with Genetic Algorithms/Programming then:
> >> >> >>
> >> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> >> >> >> (Paperback)
> >> >> >> by Melanie Mitchell
> >> >> >
> >> >> >No one has shown that genetic algorithms do better that plain old branch
> >> >> >and bound.
> >> >> >
> >> >>
> >> >> Who said anything about doing better?
> >> >
> >> >Why would anyone use a GA if there were no benefit from using it?
> >> >
> >>
> >> What is your point, given this thread's context?
> >
> >What "context"?
> >
>
> The Theory of Evolution, whose ideas are behind genetic algorithms,
> genius!

Those are two completely different things. Every now and then some
"genius" gets the idea that GAs ought to be used to predict some
higly competitive scenario, say, the stock market. I heard that idea
first 12 years ago when I first started looking into GAs, and every
two or three years since.

Now, the main difference between GAs and "Darwinian" evolution,
is that GAs aim to generate a set of coefficients (a "genome") that
somehow represents a *desired* state. This is the main point. In the
case of the GA the user defines what state he wants to end up with,
and it is up to the GA to find the "genome" that represents this state,

in an otherwise static environment.

In the case of "Darwinian" evolution, there is a population of
individes
that are more or less well adapted to their living conditions. These
living conditions change as time go by, which means that the population

that enjoy the benefit today, may be at a disadvantage tomorrow.

For instance, the native human population at the Masai Mara in eastern
Africa, are very tall, thing and have a very dark complexion. The
benefit
of this, is that the ration of body urface to body volume is very high,

and the skin knducts heat very well. These people can stand the
heat of the grasslands better than probably anyone else. At the other
extreme is the inuites of the arctic, who are short, chubby and of
light complexion. These factors serve to conserve body heat, making
them better suited to stai in the cold without suffering from
hypothermia.

If the climatic factors changed in one of these places, other
population
groups who find these areas too harsh to live, might find these areas
habitable, and might compete with the natives masai or inuites.

This is the essence of "Darwinian" evolution, which is caused by
the dynamic change of living conditions. GAs are only smoke and
mirrors, whichever way they are viewed.

Rune

Stan Pawlukiewicz wrote:
> Jerry Avins wrote:
>
>> Stan Pawlukiewicz wrote:
>>
>>> Just Cocky wrote:
>>
>>
>>
>> ...
>>
>>>> String theory is a speculative theory. But it's scientific because it
>>>> is, in principle, falsifiable.
>>>
>>>
>>>
>>>
>>> So if it's falsified, then it's a valid scientific theory? Please
>>> explain this further. I don't get it.
>>
>>
>>
>> He didn't write "falsified", but "falsifiable". If no circumstance can
>> demonstrate that a statement is false, that statement is not
>> falsifiable. "The sum of the squares on the two sides is equal to the
>> square on the hypotenuse" is falsifiable.
>
>
>
>
> That's just a way to say that
>
>> it's testable while specifying what a test must be able to do.
>
>
> This I can understand. Thank you. So by this, if indeed string theory
> can not in principle be tested, it is not falsifiable.

Correct, for now. It's called a "theory" by courtesy, but I see it as
organized speculation. exploring ramifications of several hypotheses. In
most observational sciences, cosmology and geology chief among them, the
logical consequences of hypotheses are all that can be tested anyway,
but string theory hasn't yet gotten that far, and it may never do.
Theories are never proven, but when enough confirming instances have
been amassed, we have confidence in them. Usually, when contradictions
to well established theories turn up, relatively small changes to the
theory accommodate them. (Newton was wrong. Einstein fixed it.)
Sometimes, as with the very fertile theory of phlogiston and the old
notion of static continents, a new paradigm completely replaces the old.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

On Thu, 22 Dec 2005 11:50:55 -0500, Just Cocky wrote:

> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> <[email protected]> wrote:
>>
>>DNA mutates. This really does't expalin why some mutations survive and
>>other s become exinct. The fact that DNA mutate supports an piece of
>>Evolutionary theory but it doesn't corroborate it in whole.
>>
>
> The other piece, of course, is selection.
>
>>
>>I'm an engineer.
>>
>
> Have fun with Genetic Algorithms/Programming then:
>
> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> (Paperback)
> by Melanie Mitchell

Goldberg's "Genetic Algorithms in Search, Optimization and Machine
Intelligence" is also a really good read, particularly if you like
practical examples with working code.

Frankly, once you see how DNA works, and it's power as an optimization
technique (and you count generations, rather than just absolute time), the
whole natural selection/evolution argument becomes almost a no-brainer.

However (in my mind) it leaves open the big question: how did something
so monumentally useful come to exist in the first place? Luckily for
the theory of evolution, I don't believe that it attempts to go there...

--
Andrew

In article <doegmv$813$[email protected]>,
Stan Pawlukiewicz <[email protected]> wrote:
>
>I'm an engineer.

Hmm... I might have guessed. Another data point for the Salem
Hypothesis.

http://en.wikipedia.org/wiki/Salem_hypothesis

-A

Andrew Reilly wrote:
> On Thu, 22 Dec 2005 11:50:55 -0500, Just Cocky wrote:
>
> > On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> > <[email protected]> wrote:
> >>
> >>DNA mutates. This really does't expalin why some mutations survive and
> >>other s become exinct. The fact that DNA mutate supports an piece of
> >>Evolutionary theory but it doesn't corroborate it in whole.
> >>
> >
> > The other piece, of course, is selection.
> >
> >>
> >>I'm an engineer.
> >>
> >
> > Have fun with Genetic Algorithms/Programming then:
> >
> > An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> > (Paperback)
> > by Melanie Mitchell
>
> Goldberg's "Genetic Algorithms in Search, Optimization and Machine
> Intelligence" is also a really good read, particularly if you like
> practical examples with working code.

Yep, that's the text I used to get aquainted with GAs. I ordered it
through
my university department and checked it in with the main library when
I got it. The result was that I spent the next three months fighting
off lending requests from all over the country. It was copy number 2
ever checked in with the national academic library register.

I never again checked in any book I bought, except with the local
secretary.

> Frankly, once you see how DNA works, and it's power as an optimization
> technique (and you count generations, rather than just absolute time), the
> whole natural selection/evolution argument becomes almost a no-brainer.

Well, yes. Goldberg's text does describe how evolution works. The
problem
is that "Darwinian" evolution and evolution in the GA sense differ on
certain key points, as I attempted to comment on in a different post.

Yes, "Darwinian" evolution is a no-brainer once you have read a text
like Goldberg.

No, I don't agree that GAs is a particularly useful optmization
technique.

> However (in my mind) it leaves open the big question: how did something
> so monumentally useful come to exist in the first place? Luckily for
> the theory of evolution, I don't believe that it attempts to go there...

No we are entering philosophical territory. I have some vague
recollection
of somebody mentioning they tried to reproduce the "organic soup"
likely to be present on the young planet Earth. They introduced some
water, carbon, oxygen and nitrogen (possibly some other fairly simple
chemical compounds as well) in a sealed glass container and shot
high voltage sparks through the mixture, to simulate lightning.

After a surprisingly short amount of time, there were several amino
acids present in the mixture. Imagine something like that going on
on a global scale for two billion years, and the formation of DNA is
not
all that impossible.

I'm sorry I can't dig up a precise reference; I heard the story told as

an anecdote several years ago.

Rune

On 22 Dec 2005 14:01:56 -0800, "Rune Allnor" <[email protected]>
wrote:
>
>Just Cocky wrote:
>> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
>> wrote:
>> >
>> >Just Cocky wrote:
>> >> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
>> >> wrote:
>> >> >
>> >> >Just Cocky wrote:
>> >> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
>> >> >> <[email protected]> wrote:
>> >> >> >
>> >> >> >Just Cocky wrote:
>> >> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>> >> >> >> <[email protected]> wrote:
>> >> >> >>
>> >> >> >>>DNA mutates. This really does't expalin why some mutations survive and
>> >> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
>> >> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
>> >> >> >>>
>> >> >> >>
>> >> >> >>
>> >> >> >> The other piece, of course, is selection.
>> >> >> >>
>> >> >> >>
>> >> >> >>>I'm an engineer.
>> >> >> >>>
>> >> >> >>
>> >> >> >>
>> >> >> >> Have fun with Genetic Algorithms/Programming then:
>> >> >> >>
>> >> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>> >> >> >> (Paperback)
>> >> >> >> by Melanie Mitchell
>> >> >> >
>> >> >> >No one has shown that genetic algorithms do better that plain old branch
>> >> >> >and bound.
>> >> >> >
>> >> >>
>> >> >> Who said anything about doing better?
>> >> >
>> >> >Why would anyone use a GA if there were no benefit from using it?
>> >> >
>> >>
>> >> What is your point, given this thread's context?
>> >
>> >What "context"?
>> >
>>
>> The Theory of Evolution, whose ideas are behind genetic algorithms,
>> genius!
>
>Those are two completely different things.
>

I disagree.

>
>Now, the main difference between GAs and "Darwinian" evolution,
>is that GAs aim to generate a set of coefficients (a "genome") that
>somehow represents a *desired* state.
>

Only if you set them up that way. It is certainly possible to
create/use GA's with open ended "final" states.

>
>In the case of "Darwinian" evolution, there is a population of
>individes
>that are more or less well adapted to their living conditions. These
>living conditions change as time go by, which means that the population
>that enjoy the benefit today, may be at a disadvantage tomorrow.
>

That can certainly be simulated using GA's. First, one doesn't need to
have a static fitness function. Second, one doesn't even need to have
an explicit fitness function. For example, if the ability for an
individual to reproduce is proportional to some "energy" and if that
"energy" results from actions that depend on the specific genome, then
those genomes with the better adapted actions will leave more
offspring. And if "energy" is scarce, the individuals whose genomes
lead to less adapted actions will dye off eventually. These ideas can
be used to evolve a population adapted to playing tic-tac-toe, for
example.

>
>This is the essence of "Darwinian" evolution, which is caused by
>the dynamic change of living conditions.
>

Yes, natrual selection is the essence of Evolution. But GA's
incorporate selection as well. How are they not related conceptually?

>
>GAs are only smoke and mirrors, whichever way they are viewed.
>

This is a totally unfounded conclusion. It's not difficult to Google
for real world applications of GA's.

On Thu, 22 Dec 2005 22:40:55 +0000 (UTC), [email protected] (axlq)
wrote:
>
>In article <doegmv$813$[email protected]>,
>Stan Pawlukiewicz <[email protected]> wrote:
>>
>>I'm an engineer.
>
>Hmm... I might have guessed. Another data point for the Salem
>Hypothesis.
>
>http://en.wikipedia.org/wiki/Salem_hypothesis
>

That's an interesting hypothessis. Although I'm an EE, my passion has
always been Physics. As much as I like to build things, nothing gives
me a bigger hard on than actually understanding how the Universe
works.

On Thu, 22 Dec 2005 15:45:56 -0800, Rune Allnor wrote:
> No, I don't agree that GAs is a particularly useful optmization
> technique.

I admit that my only experience with it was eleven or twelve years ago,
when I used a variation on one of Goldberg's algorithms to explore a space
of speech recognition systems for my thesis. Seemed like a good idea at
the time, but the arguments against that have been presented here
(i.e., not necessarily any better than other techniques, and not able to
tell you when it's done) are indeed consistent with that experience.

Cheers,

--
Andrew

Andrew Reilly wrote:
> On Thu, 22 Dec 2005 15:45:56 -0800, Rune Allnor wrote:
> > No, I don't agree that GAs is a particularly useful optmization
> > technique.
>
> I admit that my only experience with it was eleven or twelve years ago,
> when I used a variation on one of Goldberg's algorithms to explore a space
> of speech recognition systems for my thesis. Seemed like a good idea at
> the time, but the arguments against that have been presented here
> (i.e., not necessarily any better than other techniques, and not able to
> tell you when it's done) are indeed consistent with that experience.

Sounds about right. GAs were hot in the mid 90s, when a student of mine

tested them on an inversion problem we already knew how to solve.
The computer crunched the numbers for a full week using the GA, and
came up with an answer worse than we were able to find in five minutes
using a Levenberg-Marquart scheme. Of course, the GA did not require
the significant amount of training the student needed in order to
initialize the Levenberg-Marquart scheme. To some, the "GA hype" +
no need for a trained human operator outweighs the excessive run-time
of GAs.

Interestingly, I have met very few people who ask for the relevance of
the end result. One of the oponents at my PhD dissertation asked me
how I would like to proceed if I was given carte blanche to follow up
on my thesis, which was based on an inversion of geophysical data.
I said I would like to go out and do ground truth measurements and see
if they corroborated what my numerical analyses indicated. "That's
not an option!" was the very firm response. Of course, if the end
result
doesn't matter, it is irrelevant what inversion scheme is used.

Oh well. I don't know what to make of it.

Rune

Just Cocky wrote:
> On 22 Dec 2005 14:01:56 -0800, "Rune Allnor" <[email protected]>
> wrote:
> >
> >Just Cocky wrote:
> >> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
> >> wrote:
> >> >
> >> >Just Cocky wrote:
> >> >> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
> >> >> wrote:
> >> >> >
> >> >> >Just Cocky wrote:
> >> >> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
> >> >> >> <[email protected]> wrote:
> >> >> >> >
> >> >> >> >Just Cocky wrote:
> >> >> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
> >> >> >> >> <[email protected]> wrote:
> >> >> >> >>
> >> >> >> >>>DNA mutates. This really does't expalin why some mutations survive and
> >> >> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
> >> >> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
> >> >> >> >>>
> >> >> >> >>
> >> >> >> >>
> >> >> >> >> The other piece, of course, is selection.
> >> >> >> >>
> >> >> >> >>
> >> >> >> >>>I'm an engineer.
> >> >> >> >>>
> >> >> >> >>
> >> >> >> >>
> >> >> >> >> Have fun with Genetic Algorithms/Programming then:
> >> >> >> >>
> >> >> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
> >> >> >> >> (Paperback)
> >> >> >> >> by Melanie Mitchell
> >> >> >> >
> >> >> >> >No one has shown that genetic algorithms do better that plain old branch
> >> >> >> >and bound.
> >> >> >> >
> >> >> >>
> >> >> >> Who said anything about doing better?
> >> >> >
> >> >> >Why would anyone use a GA if there were no benefit from using it?
> >> >> >
> >> >>
> >> >> What is your point, given this thread's context?
> >> >
> >> >What "context"?
> >> >
> >>
> >> The Theory of Evolution, whose ideas are behind genetic algorithms,
> >> genius!
> >
> >Those are two completely different things.
> >
>
> I disagree.
>
> >
> >Now, the main difference between GAs and "Darwinian" evolution,
> >is that GAs aim to generate a set of coefficients (a "genome") that
> >somehow represents a *desired* state.
> >
>
> Only if you set them up that way. It is certainly possible to
> create/use GA's with open ended "final" states.

Sure, it is possible. But why would anybody do it?

> >In the case of "Darwinian" evolution, there is a population of
> >individes
> >that are more or less well adapted to their living conditions. These
> >living conditions change as time go by, which means that the population
> >that enjoy the benefit today, may be at a disadvantage tomorrow.
> >
>
> That can certainly be simulated using GA's. First, one doesn't need to
> have a static fitness function. Second, one doesn't even need to have
> an explicit fitness function. For example, if the ability for an
> individual to reproduce is proportional to some "energy" and if that
> "energy" results from actions that depend on the specific genome, then
> those genomes with the better adapted actions will leave more
> offspring. And if "energy" is scarce, the individuals whose genomes
> lead to less adapted actions will dye off eventually. These ideas can
> be used to evolve a population adapted to playing tic-tac-toe, for
> example.

I can see that these ideas can be used to simulate scenarios where
several species/populations compete for the same habitats, i.e. to
simulate
Darwinian evolution. I can not see how they can be used in other
contexts.
What use will a GA have with an open end state? Who optimizes a system
with respect to some undetermined factor? What use is a dynamic
optimization system if the boundary conditions change in a random
manner?

> >This is the essence of "Darwinian" evolution, which is caused by
> >the dynamic change of living conditions.
> >
>
> Yes, natrual selection is the essence of Evolution. But GA's
> incorporate selection as well. How are they not related conceptually?

The difference is that GAs aim for a desired state in a static (or
deterministically changing) scenario, whereas natural selection
responds on the fly to the random dynamics of a real-world habitate.

> >
> >GAs are only smoke and mirrors, whichever way they are viewed.
> >
>
> This is a totally unfounded conclusion. It's not difficult to Google
> for real world applications of GA's.

What does that have to do with anything? I can think of one - 1 -
application
I have heard of over the last decade where GAs were used, that could
not
be solved at least as well with a "standard" optimization method.

Rune

Hi DSP Gurus, (Free Book)

Proponent of evolution who are dying to prove its validity should read
this book based on solid scientific evidence to disprove darwinism. It
has also colorful pictures and easy to read.

BOOK NAME: The Collapse of the Theory of Evolution in 20 Questions
Free To Download: http://www.harunyahya.com/20questions01.php

Excerpts from the book:

The theory of evolution was put forward as an imaginary hypothesis in
the context of the primitive scientific understanding of the nineteenth
century, and to this day it has not been backed up by any scientific
discovery or experiment. On the contrary, all the methods employed to
confirm the theory have merely proven its invalidity.

However, even today many people think that the theory is a proven fact,
like the force of gravity or the law of buoyancy. Because, as stated at
the beginning, the true nature of the theory of evolution is very
different from what is usually supposed. For this reason, some people
do not know what rotten foundations
this theory has, how it is disproved by science at every turn, and how
evolutionists are trying to keep it alive in its
death throes. Evolutionists have no other support than unconfirmed
hypotheses, biased and unrealistic observations, and imaginary
drawings, methods of psychological suggestion, countless falsehoods,
and sleight-of-hand techniques.

It has actually been proved that it is impossible for the first living
cell, or even just one of the millions of protein molecules in that
cell, to have come about by chance. This has been demonstrated not only
by experiments and observations, but also by mathematical calculations
of probability. In other words, evolution collapses at the very first
step: that of explaining the emergence of the first living cell.

Another terrible dilemma from the point of view of evolution is the DNA
molecule in the nucleus of the living cell, a coding system with 3.5
billion units containing all the details of life. DNA was first
discovered using X-ray crystallography in the late 1940s and early
1950s, and is a giant molecule with a superb plan and design. For many
years, Francis Crick, a Nobel-prize laureate, believed in the theory of
molecular evolution, but eventually even he had to admit to himself
that such a complex molecule could not have emerged spontaneously
by chance, as the result of an evolutionary process:
An honest man, armed with all the knowledge available to us now, could
only state that, in some sense, the origin of life appears at the
moment to be almost a miracle.

Enjoy
Gold

On 22 Dec 2005 18:26:28 -0800, "Rune Allnor" <[email protected]>
wrote:
>
>Just Cocky wrote:
>> On 22 Dec 2005 14:01:56 -0800, "Rune Allnor" <[email protected]>
>> wrote:
>> >
>> >Just Cocky wrote:
>> >> On 22 Dec 2005 12:25:55 -0800, "Rune Allnor" <[email protected]>
>> >> wrote:
>> >> >
>> >> >Just Cocky wrote:
>> >> >> On 22 Dec 2005 10:31:08 -0800, "Rune Allnor" <[email protected]>
>> >> >> wrote:
>> >> >> >
>> >> >> >Just Cocky wrote:
>> >> >> >> On Thu, 22 Dec 2005 11:05:00 -0500, Stan Pawlukiewicz
>> >> >> >> <[email protected]> wrote:
>> >> >> >> >
>> >> >> >> >Just Cocky wrote:
>> >> >> >> >> On Thu, 22 Dec 2005 10:28:31 -0500, Stan Pawlukiewicz
>> >> >> >> >> <[email protected]> wrote:
>> >> >> >> >>
>> >> >> >> >>>DNA mutates. This really does't expalin why some mutations survive and
>> >> >> >> >>>other s become exinct. The fact that DNA mutate supports an piece of
>> >> >> >> >>>Evolutionary theory but it doesn't corroborate it in whole.
>> >> >> >> >>>
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >> The other piece, of course, is selection.
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >>>I'm an engineer.
>> >> >> >> >>>
>> >> >> >> >>
>> >> >> >> >>
>> >> >> >> >> Have fun with Genetic Algorithms/Programming then:
>> >> >> >> >>
>> >> >> >> >> An Introduction to Genetic Algorithms (Complex Adaptive Systems)
>> >> >> >> >> (Paperback)
>> >> >> >> >> by Melanie Mitchell
>> >> >> >> >
>> >> >> >> >No one has shown that genetic algorithms do better that plain old branch
>> >> >> >> >and bound.
>> >> >> >> >
>> >> >> >>
>> >> >> >> Who said anything about doing better?
>> >> >> >
>> >> >> >Why would anyone use a GA if there were no benefit from using it?
>> >> >> >
>> >> >>
>> >> >> What is your point, given this thread's context?
>> >> >
>> >> >What "context"?
>> >> >
>> >>
>> >> The Theory of Evolution, whose ideas are behind genetic algorithms,
>> >> genius!
>> >
>> >Those are two completely different things.
>> >
>>
>> I disagree.
>>
>> >
>> >Now, the main difference between GAs and "Darwinian" evolution,
>> >is that GAs aim to generate a set of coefficients (a "genome") that
>> >somehow represents a *desired* state.
>> >
>>
>> Only if you set them up that way. It is certainly possible to
>> create/use GA's with open ended "final" states.
>
>Sure, it is possible. But why would anybody do it?
>

To gain insight into the mechanics of the Theory of Evolution! That is
the reason I mentioned in the first place, that they can be used in
this fashion without trying to solve any particular problem.

>
>I can see that these ideas can be used to simulate scenarios where
>several species/populations compete for the same habitats, i.e. to
>simulate Darwinian evolution.
>

Exactly!

>
>What use will a GA have with an open end state? Who optimizes a system
>with respect to some undetermined factor? What use is a dynamic
>optimization system if the boundary conditions change in a random
>manner?
>

Pedagogic. To see what happens. Gain insights into the mechanics of
mutation, crossover and selection. This was the context under which I
brought them up, as a tool to understand Evolution.

>
>> >This is the essence of "Darwinian" evolution, which is caused by
>> >the dynamic change of living conditions.
>> >
>>
>> Yes, natrual selection is the essence of Evolution. But GA's
>> incorporate selection as well. How are they not related conceptually?
>
>The difference is that GAs aim for a desired state in a static (or
>deterministically changing) scenario, whereas natural selection
>responds on the fly to the random dynamics of a real-world habitate.
>

Ultimately, there's no reason not to use GA's in that fashion. For
example, I find this a very interesting application of GA's:
http://www.genarts.com/karl/evolved-...creatures.html
The movie is particularly cool. For example, if someone decides to
send a swarm of robotic explorers to Mars, one could try to evolve the
best creatures for the job.

>> >
>> >GAs are only smoke and mirrors, whichever way they are viewed.
>> >
>>
>> This is a totally unfounded conclusion. It's not difficult to Google
>> for real world applications of GA's.
>
>What does that have to do with anything? I can think of one - 1 -
>application
>I have heard of over the last decade where GAs were used, that could
>not
>be solved at least as well with a "standard" optimization method.
>

"Standard" optimization methods aren't particularly good at finding
acceptable solutions to problems subject to multiple and oftentimes
conflicting requirements. In these cases, GA's might provide the best
(only!) reasonable approach. Timetabling comes to mind.

On 22 Dec 2005 18:30:31 -0800, "gold" <[email protected]> wrote:
>
>Proponent of evolution who are dying to prove its validity should read
>this book based on solid scientific evidence to disprove darwinism. It
>has also colorful pictures and easy to read.
>
>BOOK NAME: The Collapse of the Theory of Evolution in 20 Questions
>Free To Download: http://www.harunyahya.com/20questions01.php
>

More fundamentalism crap. This time from the bunch that holds as
sacred a freaking space rock.