GA Strategy

[wossname]

I have a polynomial equation with 7 variables, (V0 to V6)

y = 0.000000000001*(x-V0)*(x-V1)*(x-V2)*(x-V3)*(x-V4)*(x-V5)*(x-V6)

Now I need a way to use a Genetic algorithm to generate a population, (at random to begin with) of equations, each having the 7 variables.

I must find the best 2 out of the population that are closest to the target equation and 'breed' the next generation from them and keep going until a good approximation is found.

I already have a program but it is fundamentally flawed, I am not posting it here because it is in written in PocketC, and i am convinced that I am the only person on this BB that knows it!

Does anyone have an idea how to get this done?

[Arbiter]

Wossname,

I have no idea what you're on about.

I can count to 10 (20 if I remove my socks) if this is of any use to you....

[wossname]

Thanks for the help.

[thinktank]

Originally posted by Arbiter
Wossname,

I have no idea what you're on about.

I can count to 10 (20 if I remove my socks) if this is of any use to you....

hoorah !!! I am much better than arbiter...
I can count up to 20 even with the socks on.

[jim mcnamara]

You once gave me a undergraduate-math-major-bunghole response to some help I foolishly offered.

If you can read, and understand this, you can adapt it. Since you're a brilliant mathmatician.

If you can't - you know what they told Thomas Stearns Eliot.
And don't claim it's horsefeathers because you can't figure it out.

http://www.ececs.uc.edu/~cpurdy/jimhauser/jhauser.htm

And this is the last time I'm offering you anything, Buster. Period. You've got a problem you need to work on. And it ain't GA algorithms.

[jim mcnamara]

Guys - what he's on about is a numerical process (iterative).

You feed the algorithm a guess - even a bad one. It then generates an answer, and 'mutates' the answer-generation
process toward a better solution. You do a it a lot of times, you a decent, but not perfect answer.

There are less obfuscated (means 'in terms that are deliberately hard to understand'- which Wossname seems to favor) examples of doing things like this.

If you've ever seen Winfract or Fractint - it's got Newton's method documented pretty well in it. This is kind of what genetic algorithms do. Except the smarts to improve are built into them.
Newton's method kinda does it on it's own.

I know the fractint algorithms, because I wrote several of the algorithms for it that were part of the whole deal. Back in Arpanet days.

[wossname]

What did I say? Whatever it was that upset you I'm sorry dude.

If you had a grievance against me then why didn't you email me, then I could have apologised directly. I didn't see anything particularly offensive about the my starting post in this thread.

All I can say is sorry, and I am.