Re: random model
From: Darius Blaszijk (dhkblaszyjk_at_zeelandnet.nl)
Date: 03/17/05
- Next message: Robin Chapman: "Re: Factoring integers on a classical computer"
- Previous message: William Elliot: "Re: Nice little total order problem."
- In reply to: Justin Davis: "Re: random model"
- Messages sorted by: [ date ] [ thread ]
Date: Thu, 17 Mar 2005 09:56:58 +0100
Justin,
My GA algorithm is in fact searching for coefficients to fit a dataset. So
what I'm looking a a way to generate functions / models that the GA could
try to fit to.
Lets say I have n-variables. And m-terms that I would like to examine. Then
I should come up with a polynomial that can be any combination of the
"standard" operators (+, -, *, /, ^, sqrt, ln, exp). I dont intend to do
anything dynamic, so no derivatives or anything.
I hope it clarifies things a bit.
Kind regards, Darius Blaszijk
"Justin Davis" <jkd3@duke.edu> schreef in bericht
news:Vo6_d.28933$wl4.848634@twister.southeast.rr.com...
>
> "Darius Blaszijk" <dhkblaszyjk@zeelandnet.nl> wrote in message
> news:77-dnUT_t_-cTqXfRVnyiw@zeelandnet.nl...
> > Hallo,
> >
> > I have a bit of an awkward question for you. I'm toying around with a GA
> > and
> > I would like to test it by applying several models. Unfourtunately the
> > search is rather fast and I have to enter each time a new model by hand.
> > So
> > I was thinking if there are algorithms present that could generate a
> > continuous model automatically, so I can feed it to the GA?
> > The model should incorporate all the standard operators and allow for an
> > undefined amount of parameters. Can this be done?
> > I would appreciate any links or pointers on how to procede from here.
> >
> > Kind regards, Darius Blaszijk
>
> It's not clear, at least to me, what you are modeling. However, if I'm
> guessing correctly, you're asking about an encoding, a model, over which
the
> genetic algorithm works. For instance, one model of human populations
would
> include heights, weights, and skin color; you'd use the GA to come up with
> the best combination of those characteristics. Another model might include
> heights, weights, skin color, and waist measurement. Then again, I'm not
> sure how "standard operators" fit into this, because the operators I'm
> thinking of are inherent in the GA itself, whatever you're modeling; e.g.
> the mutation operator is often a part of the GA itself, not the thing
you're
> modeling.
>
> Could you be more specific about your problem, perhaps giving an example
of
> the models you're using?
>
>
- Next message: Robin Chapman: "Re: Factoring integers on a classical computer"
- Previous message: William Elliot: "Re: Nice little total order problem."
- In reply to: Justin Davis: "Re: random model"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
