| |
| |
| |
|
Page: 1 2 3 4 5
Comments:
<0> have you done any modeling, Mesh?
<1> i did a little modeling in my prime
<2> heh, tried
<2> and failed :)
<0> get back on that horse
<2> eugh
<2> no thankyou =)
<0> how did the ***ignments go?
<2> well
<2> im thinking of quitting today dude :/
<2> to pursue these business oppurtunities at this compan
<2> company
<0> http://www.80stees.com/images/products/Superman_School_t-shirt_lin.jpg
<3> http://www.fl-tw.com/Infinity/infinity_features.php
<4> Rud0lf :)))
<2> darnit Rud0lf
<2> now I must stay
<2> Superman said so :/
<3> http://edition.cnn.com/2006/TECH/fun.games/02/20/sony.playstation.reut/index.html
<0> http://www.eurogamer.net/article.php?article_id=62956
<2> them controllers look nasty
<0> I don't think that's the final design
<5> y0!
<6> they said they're getting rid of the batarang controller
<6> thank god
<6> it's a shame if they ditch dual shock; that's my favorite controller of all time
<7> y
<7> a
<6> HOLY VERTICAL AGREEMENTS BATMAN
<7> :D
<7> lol
<2> oh man, dead rising for the XBox 360
<2> bbl
<8> Hi all
<5> y0!
<5> what's going Paladin_M ?
<8> not alot atm
<8> you ?
<9> i think you is the masterer bitcher
<5> hah hah
<5> I'm working on improving my dev community website.
<8> is it standard when making a network game
<5> Is what?
<8> to only have player cordinates etc. transmit fairly slowly
<8> and instead have the server etc. predict where the client is going to end up
<8> by therere speed ?
<5> Paladin_M: I don't know. I haven't gotten to there yet. :)
<0> yes, in a few years, when computers are smart enough, they'll be able to predict everything, and we won't need players
<10> My plan for world domination. Write a brute force program generator by writing bytes to a file and running it, recording the result. Eventually I'll have written every possible application.
<10> of course I'll need an infinite number of monkeys to test the apps.
<5> GooRooPR: I had a similar idea.
<10> I feel sorry for the monkey that gets a permutation which wipes out the hard disk
<5> GooRooPR: Instead, it would write byte-code and depending upon how many errors, the programs with less errors would 'reproduce, with mutations'.
<5> GooRooPR: Sooner or later youd get a program with zero errors :)
<10> anyone use PhysX before? I keep getting "Can't compute m*** from shapes: must have at least one non-trigger shape!"
<5> GooRooPR: Those would tend to reproduce and cross bytecode with other successful programs (no errors with them).
<5> GooRooPR: So basically you'd start getting larger and large programs with no compiler errors. Of course they may not do anything useful and may have logic/semantic errors which aren't caught by the byte code verifier.
<0> RUEbaiYaT: the search space is too coarse to produce anything useful
<10> hence the infinite monkeys
<0> the infinite monkeys are too hungry to work
<0> they demand more bananas
<6> just get a whole lot of bnanas
<5> Rud0lf: Well, the objective function is simply 'programs with zero bytecode verification/compiler errors tend to reproduce', so yeah.
<10> on a side note. I'm probably going to die.
<10> it was nice knowing you guys
<0> everyone dies
<10> at the very least I won't be able to reproduce. Which is probably a good thing.
<5> Rud0lf: However, if you introduce 'semantically correct' programs which live for a long time and you start looking at evaluating functions interfaces as well as byte code, then you might have something interesting.
<5> Rud0lf: I would think self contained functions would probably work, I'm not sure about functions that can p*** data in/out.
<0> has genetic programming produced anything useful?
<6> uh. yes
<0> like?
<9> like is like similar
<0> wall following algorithms? :)
<5> Rud0lf: For configuring a large set of variables to fit a specific solution set, sure.
<0> that's genetic algorithms, not genetic programming :)
<10> PitDroid is a genetic mutation gone awry
<5> Rud0lf: oh. I'm not familiar with genetic programming. I would guess it would depend upon what you would consider 'useful'.
<6> RUEbaiYaT: it's just a kind of simulated annealing
<0> anything of scientific or commercial use :)
<0> http://en.wikipedia.org/wiki/Genetic_programming
<11> Hello :)
<9> Why hello there Mr. [Relic]. Perhaps you should've knocked first.
<6> RUEbaiYaT: the idea behind GA is to express a solution set as a "genome" onto which you can transpose genes and simulate ***ual reproduction to combine and divest approaches
<6> oh, i guess i'm back on ignore
<6> huhu
<6> i wish he'd make his mind up
<5> Rud0lf: I would think that the byte-verifier is not enough to produce 'useful' programs, just enough to produce 'correct' programs.
<6> actually
<5> well correct in the sense of compilable.
<0> yeah, that's the problem
<6> thanks to the halting problem
<6> you can't byte verify non-trivial applications
<6> but, ruebiayat likes to talk about things he doesn't understand, so whatever
<0> StoneCyqh: you could set a time limit as well :)
<6> which you'll hit each and every time.
<5> Rud0lf: You'd have to write some sort of system to formalize 'usefulness' and provide that as feedback to the creation of the genetic code.
<6> if that sort of thing was realistic you'd see it deployed as a debugging tool by now.
<0> yeah, but how do you formalize usefulness?
<5> Rud0lf: That's a good question. I would suppose, using a black box. Given Input, if the Output is within this range with an error percentage of x, then it's useful.
<6> buhuhuhuhhu
<6> buhuhuhuhuuhuhuhuhuhuhuhhuhuhuhuhuhu
<5> The code generated within the black box, could only use the parameters specified in the input.
<0> so you'd want to use it to find fast approximations to known functions or known data?
<5> You can specifiy data-ranges of course.
<10> we all know it would take a long time to find anything useful. But hey. What else are you gonna do with your time
<5> GooRooPR: actually, using a quantum computer (if it were built), it would generate the code very quickly.
<5> GooRooPR: Since it would evaluate all sequences of the possible bytecode and test it all in parrallel.
<0> RUEbaiYaT: then you wouldn't need GA :)
<10> programs also need data. You'd have to test all possible input as well.
<0> just generate all solutions
<5> Rud0lf: Your brain is a quantum computer.
<5> Rud0lf: down at the quantum level :)
<5> GooRooPR: you'd have to test given data-ranges.
<0> well, then so is any computer
<5> Rud0lf: Not like your brain :)
<0> both are electric
<12> http://www.criminalplaza.nl/click.php?x=Mescob
Return to
#gamedev
or
Go to some related
logs:
#cisco
meatmembers.com passdump
mssql signed int
#csharp
#nhl
#computers
#beginner
#computers
#beginner
quiet feet mod oblivion
|
|
