This should get some good discussion / denial!
652 thoughts on “Evolving Wind Turbine Blades”
Leave a Reply
You must be logged in to post a comment.
The Skeptical Zone
"I beseech you, in the bowels of Christ, think it possible that you may be mistaken."
NO NO NO WRONG. Correlation is an empirical, mathematical function. You don’t understand math.
http://www.statisticshowto.com/what-is-the-correlation-coefficient-formula/
Perhaps it’s* the aspect of biological evolution that lends itself least well to modelling in a GA.
*I’m thinking sex, meiosis, gene shuffling and crossover etc.
dazz,
Why is this GA only making wind turbine blades? Because that is what it was designed to do. Why are the final blades more efficient than the starting blades? Because the program actively changed the starting blades and directed those products towards its pre-specification.
Alan Fox,
And how does NS explain meiosis? Do tell.
I don’t recall saying anything about smuggling information in via the fitness filter or in any other way. Do you have a quote?
This is obviously wrong. In the OP example, one starts with a primitive candidate and generates variations, the better performing ones of which are used as raw material for further variation. The shapes are varied randomly, the design aspect is in the selection, set by the environment, the wind tunnel, and the candidates performance. There is no ultimate solution.
Beats me you can say this, which is fine, and continue to claim the process is directed.
Depends what you think “it” is. If you mean “it” as the selecting of the candidates depending on the performance in some test, of course.
Does “may be thinking” suggest a specific quote? I’m merely speculating how your argument might go, if you end up making one.
Joe Felsenstein,
I disagree. Most on the ‘evo’ side have been sticking consistently with mean reproductive output, integrating both survival and fecundity.
How is fitness quantified differently in GA’s than in nature?
As long as that property is not called fitness. What is that property called in a GA?
Fitness can be different in 2 GAs, Mung. Have you ever worked with / coded them?
Alan Fox,
I think that misses Prof. Felsenstein’s point that fitness includes fecundity as well as survival. But sex as described is well represented in typical GAs (except perhaps for gene shuffling and the non-crossover recombinational mechanism of gene conversion in diploids). It has interesting, and powerful, effects on ‘search’ times and explorability.
Frankie, GA’s paths are contingent to random variation, just like evolution. It can’t be both “contingent” and directed.
So to sum it up:
1. You claim that GA’s target a solution, but can’t explain why they follow different paths each run
2. You claim each time it seems to get stuck, it’s reached “the” solution, even if it’s different each run, but you can never know if the next mutation may produce a fitter candidate, hence you’re wrong in saying you know it’s reached a solution.
3. You contradict yourself when you claim GA’s are directed and contingent.
4. You are incapable of considering you might be wrong and refuse to test your claims, ignoring my offer to test what would happen to a GA if the random variation was removed.
In conclusion, you’ve been proven wrong, and proven intellectually dishonest by refusing to put your claims to the test.
Back to ignoring you now, you’re not worth anyone’s time
Mung,
Fitness. The ‘fitness function’ is the thing that causes different genotypes to have different numbers of offspring – to have different fitnesses.
Alan Fox,
Alan, The GA is designing the blades so obviously my comment is correct. The blades would never change if the GA didn’t tell them to. A more efficient blade would never arise if the GA didn’t direct it to.
Umm engineering trial and error is directed, Alan
The GA, Alan, we are talking about the GA.
Look we are done here. You will never change your mind as long as you can hide behind moderation and not address the points made. Take this to an honest and open forum with qualified moderators and you would be toast
🙂
You see the word design and you start salivating, LOL. So lame
Bad news for you is that “design” was produced by recombination, random variation, and fitness selection, just like in evolution.
Tell us what you think that “pre specification” was, Frankie. Explicitly what did it look like?
Bye!
Recommendations? How about ‘After the Bar Closes’ where anything goes?
Meet him in the parking lot with Cash + Donuts!
Perhaps I should have been bolder and disagreed with his remark about us overlooking fecundity as a key aspect of gene spread and survival as you just did. 🙁
OK. I’m interested in learning more. I’d appreciate some info if not too much trouble.
Models of what?
So?
dazz,
ID is not anti-evolution. The GA is contingent on the form that came before.
That’s not an argument. Dawkins weasel program was both contingent and directed. You lose
Mung loves evolution, Mwah mwah mwah 🙂
So if it starts at a certain point, and the next is contingent on that first point, and the next to the second point and so on, it SHOULD ALLWAY FOLLOW THE SAME PATH.
It doesn’t, because it’s contingent on RANDOM VARIATION, and if you remove the random variation, IT FOLLOWS THE SAME PATH EVERY TIME
Hence, you’re wrong and pathetically stubborn
Comment moved to guano. It is against the rules to accuse fellow commenters of dishonesty.
Describe how it is both,
Weasel was never anything more than a simple demo of the power of selection. Yes, the target was predetermined and fixed. You’re mixing up maps and territories again.
Indeed. The weasel code also had a specific target before it began.
You’re partially right on that one, the specific paths in that algo were contingent on random variation, but it was directed towards a solution because the fitness function was defined in the same terms as the solution.
But the weasel algo ALWAYS converges into the same solution because it’s directed, while GA’s don’t
In the weasel algo, if you repeat over and over again, it will always be going in the same direction: towards the target. True GA’s don’t do that
dazz,
Weasels are weasely identified. Stoats are stoatally different.
There’s TWO TARGETS! It CAN’T be a search!
Can’t assail that logic!
You have the guts to post a snarky comment like that, after embarrassing yourself by saying that you knew a GA had found a solution at a local maximum, and you knew it had reached a local maximum by applying a different algo?
You guys never cease to entertain
Can you? I bet not
LoL! The weasel program is a GA. Yes it has the exact phrase as a target as opposed to some pre-specified optimum, but that is a distinction without a difference.
The GA in the OP can NEVER produce anything but wind turbine blades because that is what it was designed to do. And the way it does that is the same as having a team of humans do it but with fewer resources and much less time.
If you guys were right some solutions to the OP GA would be something other than wind turbine blades.
And again I am more than willing to put my money where my mouth is and have an honest and knowledgeable person moderate our debate.
LOL!
Alan Fox,
Natural selection is eliminative, it doesn’t select. And there isn’t any difference between having a hard target and having to meet some pre-specification.
You are confusing your lack of understanding with an actual argument
That’s another reason GA’s aren’t like evolution. Thank you.
Mung,
Mung’s crushing logic: “Local maximums are solutions!!!11!!1!1one”
So the algo finds (or so he says) a local maximum at “10”
The next run it finds another one at “20”, passing through 11, 12, 13, 14,… 19
BUT according to our local genius, 10 is a solution and 11 to 19 are not.
Priceless
Evolution works exactly like that. it would never repeat the same route and it’s almost trivial to prove because organisms are part of the environment, and since mutations are random, and the environment changes with changing populations, the environment is subject to some random variation too.
Wait till he discovers the travelling salesman problem.
petrushka,
I kinda wished he’d continued with writing a GA. He would have answered his own questions.
It’s a simple computer program that Richard Dawkins wrote nearly forty years ago as an illustration of how powerful cumulative selection is. I doubt it qualifies as a GA. This is a much better example, even though there was a specification, a fixed target, for the algorithm to work towards.
No indeed. Set an algorithm to find a particular target and you won’t find it looking for solutions to another target.
Indeed. Trial and error describes the process quite well.
Sharks are very successful at surviving in a marine environment but not so successful in a desert.
But these candidates are eliminated precisely because the GA is selecting for performance as a wind turbine blade.
Again, debates don’t answer scientific issues. Testing, experimenting, observing and analysing is what works.
Because the initial starting population was randomly generated, unlike evolution.
But you must understand that in the world of religion, such concepts as testing, observation, and experiment are meaningless – they have no religious referent. Conversely, debates are exactly how religious disputes are approached, and the winner of the debate is the one who persuades more converts.
If nobody knows how the starting population was generated (and nobody does), how can you make this claim?
FAIL. WRONG.
Alan Fox,
Dawkins weasel is a GA, regardless of what you think. And this debate has nothing to do with science. Also evolutionism cannot be tested and there aren’t any experiments that support it
Flint,
LoL! So that is why testing, observations and experiments are an anathema to evolutionism