Re: best fitting
- To: mathgroup at smc.vnet.net
- Subject: [mg86725] Re: best fitting
- From: dh <dh at metrohm.ch>
- Date: Wed, 19 Mar 2008 05:26:58 -0500 (EST)
- References: <fro394$i7k$1@smc.vnet.net>
Hi,
this recursive definition is a hard nut. I do not know a direct way, but
you could try iteration. Assume a K, make your fit, calculate a new K,
make another fit, ..., then hope that the whole procedure converges.
hope this helps, Daniel
umby wrote:
> Hi,
>
>
>
> I have some experimental data in the form of quadruplets {P, W, t, D}.
>
> How can I find the best fit parameters: alfa, A and E1 of the function:
>
>
>
> P = { (W - K^2/3) / t + A [(W - K^2/3) / t]^3} E1 (t^4) / (25^2)
>
>
>
> where K = P / (alfa D^0.5).
>
>
>
> Thank you
>
> -u
>
>
>