652 thoughts on “Evolving Wind Turbine Blades”

  1. Frankie: LoL! Was that supposed to be a refutation? I was just stating a fact that seems to be overlooked. You look to be avoiding reality

    I’ll try to be clearer. A model, a simulation such as the exercise in the OP is necessarily a simplification of reality. If a model was good enough to account for all aspects of reality, it would be reality.

  2. The minimal GA simply consists of replication – no mutation, no crossover, no selection. Even that has a behaviour. If you label all starting strings and keep track, one of them will be the ancestor of every genotype in a future population. This is the “M&M’s” minimal drift case. Interestingly, you don’t even need to give your organisms a genotype. It can be an array of strings of length zero. It doesn’t solve any problems, but it does model evolution (drift without mutation). The reason it can do so with strings of length zero is that replication, which would occupy real bits in a real genotype, is an unvarying given (generally), not part of the replicated string.

    Then you can allow mutation, which would have to be able to add bits to get your length zero strings to contain something. Depending on the level of mutation, you would still get a tendency for one of these ‘genotypes’ to fix in the future population. But, if you watched long term, the fixed genotype would keep changing. The population would continually evolve.

    This could hardly be termed a search.

    Now if you add some differential – some reason why one genotype might have a higher chance of staying in the contest than another – then, you get adaptation. The strings adapt to the presence of this selective factor, by concentrating those that get copied more often (produce more offspring) under it.

    You can allow the factor itself to vary.

    You can allow for different parts of the genotype to be evaluated according to different criteria.

    You can add crossover.

    You can add range geometry (an important factor often subsumed in population genetic models) and some kind of geometric or genetic isolation probability.

    You can allow replication itself to be subject to variation (it is usually constant in most GAs, subject to fitness of the extended string).

    The world is your lobster. Write one, have a play. It may be instructive.

  4. Frankie:
    Alan Fox,

    Nice of you to avoid responding to the relevant parts of the post. I know why you did that.

    Relevant part of what post? Quoting or linking would be helpful.

  5. Frankie:

    All GAs do is emulate many people working on the same problem.

    Rich:

    Even those you said were in Lt. Data from Star Trek?

    Oooh. Linky, please?

  6. keiths,

    Holding tank for general chatter about GAs

    “oleg- I get it- you are too stupid to think outside of your simple-minded box

    Look at it this way- Star Trek: Next generation- Lt Data was able to rewire his neuro networks due to the algorithms INSIDE OF HIM

    If today’s programmers could figure out a way to get the GA inside of their organisms, they would- now they are doing the best they can-“

  8. Mung seems to have a comment stuck in the moderation queue. Could a kindly admin release it? Thanks!

  9. OMagain: People disagree sometimes. You seem to see it as a sign of weakness when in fact it’s a sign of strength.

    No, I conclude that I must be very strong.

  13. As a matter of some interest, what do people think I am arguing for and against in this thread? Even if you don’t care to comment, at least try to give it some thought.

    And if you do want to comment and you just aren’t sure what it is I am arguing for or against, how can I help?

  14. Mung: As a matter of some interest, what do people think I am arguing for and against in this thread? Even if you don’t care to comment, at least try to give it some thought.

    You’ve been arguing for something? I did not notice that.

    And if you do want to comment and you just aren’t sure what it is I am arguing for or against, how can I help?

    By being clearer, perhaps.

  15. Richardthughes: So we have a population of 1000
    They perform some task, that is empirically measured.
    Let’s call that Fitness.

    They don’t call it an objective function for nothing, right?

    A fitness function is a particular type of objective function that is used to summarise, as a single figure of merit, how close a given design solution is to achieving the set aims.

    – Fitness Function

    Now what shall we do? Shall we pretend that this is not relevant to genetic algorithms?

    In particular, in the fields of genetic programming and genetic algorithms, each design solution is commonly represented as a string of numbers (referred to as a chromosome). After each round of testing, or simulation, the idea is to delete the ‘n’ worst design solutions, and to breed ‘n’ new ones from the best design solutions. Each design solution, therefore, needs to be awarded a figure of merit, to indicate how close it came to meeting the overall specification, and this is generated by applying the fitness function to the test, or simulation, results obtained from that solution.

    The reason that genetic algorithms cannot be considered to be a lazy way of performing design work is precisely because of the effort involved in designing a workable fitness function. Even though it is no longer the human designer, but the computer, that comes up with the final design, it is the human designer who has to design the fitness function. If this is designed badly, the algorithm will either converge on an inappropriate solution, or will have difficulty converging at all.

    Moreover, the fitness function must not only correlate closely with the designer’s goal, it must also be computed quickly. Speed of execution is very important, as a typical genetic algorithm must be iterated many times in order to produce a usable result for a non-trivial problem.

    Checkmate.

  16. Sadly, some people here are arguing that “Fitness and reproductive success are the same thing.” But as I said, that’s not how a GA works. And now we see that Richardthughes agrees. I said that above, as well.

  17. Mung,

    So you’re still arguing that there is design going on with a GA model. I agree. I even agree that one could describe selection as a design process in biological evolution. I even have a candidate for the evolutionary designer. It’s the environment.

  18. What I think about ID can be summed up in two point/counterpoints at Sandwalk, between Torley and Moran:

    Regarding epigenetics, I don’t pretend to have any in-depth knowledge of the subject.

    That’s not true. Any outside observer would assume that you knew what you were talking about. Otherwise, you would be asking questions to help you understand and not posing challenges that question naturalism and evolution.

    I’m afraid your technical comments about chromatin and histones went a little over my head, ..

    Then why are you presuming to understand these issues well enough to challenge Jerry Coyne?

  19. Alan Fox: You’ve been arguing for something? I did not notice that.

    Yes, well, you already know what I think about the claim that admins can only send posts to Guano. It’s just not possible that they could make up an entirely new heading such as “The Whine Cellar” and actually move posts to it. Right, Alan?

  21. Alan Fox: You’ve been arguing for something? I did not notice that.

    That was then.

    Alan Fox: So you’re still arguing that there is design going on with a GA model. I agree.

    And this is now.

    What made you change your mind there in those 12 minutes?

  23. Richardthughes: In my example (of a GA) Fitness was correlated to but not the same as reproductive success.

    Well, what I was trying to do was come up with a GA where they were not just correlated, but the same. Can you do that?

    After all, “Fitness and reproductive success are the same thing,” and “It’s reproductive success full stop.”

    So far you’re just supporting my claim. The challenge, should you choose to accept it, is to design a GA where they are the same thing.

    My claim is that GA’s are not like that. That they do not work that way. They work along the way that you describe. If not, why not?

  24. Mung,

    A=B. that is your claim. Entailment, the correlation of A and B is 1.

    Would you like to retract your claim?

    You can have B is informed by A, if you like.

  25. And still trying to get the small ball in the hole, if an organism has a fitness of 2.3, how does it have 2.3 offspring, oh sage of GAs?

  27. Let me see if I can clarify a bit. I have a lot of experience with genetic simulations. In them, when a genotype has a “fitness”, it is a number which gives the expected number of offspring a newborn individual has. The probabilities of survival and reproduction are arranged so as to have that as the expectation. (“Expected” is used here in the mathematical sense, not in the sense of human wishes.)

    But in many GAs, EAs, etc, there is a quantity computed that is called a “fitness” but which ends up being used, say, to rank individuals, in a process of choosing survivors, or who mates. Those “fitnesses” are not fitness in the population genetic sense. The people calling them “fitness” do not have much knowledge of the population genetic theory.

    I wonder whether some of disagreements here spring from this ambiguous useage.

  29. Joe Felsenstein: But in many GAs, EAs, etc, there is a quantity computed that is called a “fitness” but which ends up being used, say, to rank individuals, in a process of choosing survivors, or who mates. Those “fitnesses” are not fitness in the population genetic sense.

    Do you think calling what goes on in a GA “natural selection” contributes to that confusion or helps people avoid that ambiguity?

  30. Alan Fox: About what?

    About your contradictory claims, what else? What happened in those 12 minutes to change your mind?

    Alan Fox: You’ve been arguing for something? I did not notice that.

    Alan Fox: So you’re still arguing that there is design going on with a GA model. I agree.

  31. Alan, my fitness function maximizes diversity. Do you have some reason to believe that diversity is anti-thetical to evolution? It’s also directional. The best reproducing are always represented in greater numbers in the next generation.

  32. Mung: Do you think calling what goes on in a GA “natural selection” contributes to that confusion or helps people avoid that ambiguity?

    What goes on in a GA can model natural selection, if properly formulated.

    Whether it is natural selection, I leave to others to fuss about.

  34. DNA_Jock: Let’s suppose that, in addition to optimizing for efficiency, he also optimized for Exxon-logoness; it’s a fairly safe bet that the resulting co-optimized turbines would have near-neighbors who were slightly more wind efficient and less Exxon-like. However, it could simultaneously be the case that the co-optimized turbines would be more wind efficient than the ones found by optimizing for efficiency alone.

    Oh is that right? So its possible that by adding in the selection criteria of looking like an Exxon logo, the computer would come up with a MORE efficient model than without that selection criteria in place? So without that selection criteria, the computer wouldn’t have known that Exxon is a very useful shape for wind efficiency?

    Were you taking drugs when you wrote that? If that were the case DNA Jock, then adding in ANY other selection criteria MIGHT make the computer better at finding a solution. So you can add in a selection feature which selects ones that look like Mary Poppins. Also it will simultaneously select ones that will make the windturbine useful to be used as a hammer as well as a windturbine.

    So then you run your tests. And it sees some designs, and one is extremely efficient, but it looks nothing like Mary Poppins nor Exxon, and would be horrible as a hammer, no we don’t really need that one, lets keep this one, it looks a LOT like Mary Poppins, almost identical in fact, and I can kind of hammer with him. Now, next test. Ok, this one is a great hammer, not much like Mary Poppins, a little like Exxon if you squint, still not much efficiency, but hey, sometimes you have to compromise. I mean, its not like a better solution would be choosing the ONE WHICH IS MOST EFFICIENT!

    It reminds me of the MM discussion DNA JOCK. You think a generation is ten generations. Where do you buy your reefer? Is it medical marijuana?

  35. Joe Felsenstein: Those “fitnesses” are not fitness in the population genetic sense. The people calling them “fitness” do not have much knowledge of the population genetic theory.

    I wonder whether some of disagreements here spring from this ambiguous useage.

    I wonder whether some of the disagreements here spring from the fact that your definition of fitness is a circle that says what survives best, survives best.

    I think you and DNA Jock should have a discussion about what a generation means.

  36. phoodoo: So its possible that by adding in the selection criteria of looking like an Exxon logo, the computer would come up with a MORE efficient model than without that selection criteria in place?

    Yes, yes it is.
    🙂

  38. DNA_Jock: Yes, yes it is.

    Sure makes your explanation of why the guy who made the GA didn’t try to add more variables kind of curious then doesn’t it Jock?

    I think your explanation was because he was trying to find the most wind efficient. Now you are saying by selecting for things that have nothing whatsoever to do with wind efficiency you might just find the more wind efficient. Yea right.Good luck with that.

    No wonder you struggle with the MM problem.

  39. Please remember new readers, Phoodoo and Mung are REAL people not parody accounts. Check uncommon descent for example.

  40. Please remember, Richardhughes is a parody account.

    He is here to troll and tweak noses.

  41. Richardthughes:
    Please remember new readers, Phoodoo and Mung are REAL people not parody accounts. Check uncommon descent for example.

    “I’m not really a Creationist dimbulb, I just play one on the web!” 😀

  42. Mung: Alan, my fitness function maximizes diversity. Do you have some reason to believe that diversity is anti-thetical to evolution? It’s also directional. The best reproducing are always represented in greater numbers in the next generation.

    You are so smart Mung! Why isn’t evolution unfiltered, unselected randomness that has no chance of building anything worthwhile? Evolution sure could learn a lot from your posts, such as how not to learn or carry the good forward as a greater percentage of the whole.

  43. phoodoo:
    Please remember, Richardhughes is a parody account.

    He is here to troll and tweak noses.

    ‘Troll’ you made up and repeat because you have honesty issues. I do tweak noses, and its harder than you think given you lead with your chin.

  45. phoodoo: I wonder whether some of the disagreements here spring from the fact that your definition of fitness is a circle that says what survives best, survives best.

    Shows how you misunderstand. Fitness is a measure of how well a genotype, or a phenotype class, survive and reproduce. Nothing circular about that, you can measure it by experiments with real organisms.

  46. Joe Felsenstein: Shows how you misunderstand.Fitness is a measure of how well a genotype, or a phenotype class, survive and reproduce.Nothing circular about that, you can measure it by experiments with real organisms.

    The implicit circularity is echoed in the evolutionist catchphrase “survival of the fittest”. Would you take an explicit stand against the phrase? How would you formulate your stand?

  47. Erik: The implicit circularity is echoed in the evolutionist catchphrase “survival of the fittest”. Would you take an explicit stand against the phrase? How would you formulate your stand?

    Depends, sometimes circular, sometimes not.

    But in any case the issue of the validity of this phrase is not what we were talking about. phoodoo was declaring that fitness itself was a circular concept.

    phoodoo was wrong. Period.

  48. Joe Felsenstein,

    What are the properties of the fittest organisms Joe?

    Let me answer for you, because I have seen you evolutionists obfuscate this question time and time again (Um, Allan..). Those that survive and reproduce.

    Saying you are wrong doesn’t win Joe, especially when you say nothing to show you are right.

  50. Joe Felsenstein: But in any case the issue of the validity of this phrase is not what we were talking about. phoodoo was declaring that fitness itself was a circular concept.

    phoodoo was wrong. Period.

    Of course he was wrong, but for a silly reason. He was wrong in narrow scientific sense, because science has no concept of circularity. Since this is so, the argument between you two should have been much shorter. But it’s longer, so probably you guys are not arguing over the narrow scientific sense. You just might be arguing whether the concept of fitness has any broader philosophical implications, so that we should take it into account in our life as human beings. This brings the concept into the realm of general logic, where it’s submitted to tests such as circularity.

    Anyway, it’s all cool when you say that fitness is only a measure of survival of organisms without implying any actual value or meaning (philosophical, logical, moral, vital, common-sense) to fitness and/or survival. In which case all talk about survival and fitness is pointless babble, of interest only to specialized geeks of genetics, but that be science.

    Modern scientists often make glaring philosophical errors when translating their scientific facts into real-life truths. Good to see that you are not one of those.

Leave a Reply