To anyone still curious about my genetic algorithm

To anyone still curious about my genetic algorithm - Printable Version

+- The Lurker Lounge Forums (https://www.lurkerlounge.com/forums)
+-- Forum: The Lurker Lounge (https://www.lurkerlounge.com/forums/forum-4.html)
+--- Forum: The Lounge (https://www.lurkerlounge.com/forums/forum-12.html)
+--- Thread: To anyone still curious about my genetic algorithm (/thread-4809.html)

To anyone still curious about my genetic algorithm - GenericKen - 02-21-2006

Thanks to everyone who participated in the original thread here:
http://www.lurkerlounge.com/forums/index.p...wtopic=7273&hl= :whistling:

I actually finished it a month ago, but I just got it online borrowing a friends' webspace.
http://www.schtm.com/kenneth/flash/particlesdilemma.html

Give the population some time to converge to a solution. Lemme know if you hit any bugs. I'm pretty sure most of the theory panned out, but I'm sure Kylearan could tell better than I could.

To anyone still curious about my genetic algorithm - Dako-ta - 02-21-2006

Hi, GenericKen,

GenericKen,Feb 21 2006, 06:48 AM Wrote:Thanks to everyone who participated in the original thread here:
http://www.schtm.com/kenneth/flash/particlesdilemma.html
I actually finished it a month ago, but I just got it online borrowing a friends' webspace.
http://www.schtm.com/kenneth/flash/particlesdilemma.html
Give the population some time to converge to a solution. Lemme know if you hit any bugs. I'm pretty sure most of the theory panned out, but I'm sure Kylearan could tell better than I could.
[right][snapback]102505[/snapback][/right]

I don't think I understand what this is really all about. Even though I have a biology background, math wasn't my strong point.

However, I'm even more confused why you gave the same link twice, implying they're different... :wacko:

Any chance the first link is the wrong one? :P

Regards,

Dako-ta

"You do not really understand something unless you can explain it to your grandmother."

- Albert Einstein

To anyone still curious about my genetic algorithm - ima_nerd - 02-21-2006

I think he meant this one. :whistling:

Pretty amazing stuff. I'm a first year CS major and that is just incomprehensible how you would even start to make something like that.

To anyone still curious about my genetic algorithm - whyBish - 02-22-2006

GenericKen,Feb 21 2006, 07:48 PM Wrote:[right][snapback]102505[/snapback][/right]

Congrats, I'll have to go to work to view it though... no way I want flash installed on my home machine :/

To anyone still curious about my genetic algorithm - Kylearan - 02-22-2006

Hi,

nice work!

One thing, though: You wrote in the code description that you've not seen a GA that encodes mutation rate in its genome. It has been some years since I've studied the theory of GAs, but the idea to include mutation rate into the genome is quite old actually, and has been investigated thoroughly. My memory is a bit fuzzy about it, but I seem to remember that it's no advantage to do so; it doesn't speed up the GA converging in any way, and only makes it more complicated. But then it doesn't slow it down much either as far as I remember, so...

I can't give you the exact details, but that should be easy to find in the literature. If you're really interested, I could try to find the relevant papers again.

-Kylearan

To anyone still curious about my genetic algorithm - GenericKen - 02-22-2006

Kylearan,Feb 22 2006, 07:19 AM Wrote:Hi,

nice work!

One thing, though: You wrote in the code description that you've not seen a GA that encodes mutation rate in its genome. It has been some years since I've studied the theory of GAs, but the idea to include mutation rate into the genome is quite old actually, and has been investigated thoroughly. My memory is a bit fuzzy about it, but I seem to remember that it's no advantage to do so; it doesn't speed up the GA converging in any way, and only makes it more complicated. But then it doesn't slow it down much either as far as I remember, so...

I can't give you the exact details, but that should be easy to find in the literature. If you're really interested, I could try to find the relevant papers again.

-Kylearan
[right][snapback]102626[/snapback][/right]

No advantage to do so over what? Setting the mutation rate by hand? What if I have no idea what an acceptable mutation rate will be for a particular set of parameters? :blush: What if the properties of the game played suddenly changed so that the population would need a different mutation rate?

I wasn't really thinking about speeding up convergances, but rather about avoiding having to hardcode parameters myself. :whistling:

To anyone still curious about my genetic algorithm - Kylearan - 02-22-2006

Hi,

GenericKen,Feb 22 2006, 09:09 AM Wrote:What if I have no idea what an acceptable mutation rate will be for a particular set of parameters?

Then you would be in good company. :lol: I usually set the mutation rate so that roughly one mutation per mating will occur. I do this only because it "feels" good, not beause I know that this is a good choice. :whistling: Have you taken a look at what value the mutation parameter ends up in your scenario? Does it end up at a stable value, or oscillates it between several values?

Quote:I wasn't really thinking about speeding up convergances

Ah, okay. Your original post sounded like you were concerned about speed. :)

-Kylearan