Re: FindFit & restricting fitting parameter
- To: mathgroup at smc.vnet.net
- Subject: [mg55474] Re: [mg55450] FindFit & restricting fitting parameter
- From: "Janos D. Pinter" <jdpinter at hfx.eastlink.ca>
- Date: Fri, 25 Mar 2005 05:48:15 -0500 (EST)
- References: <200503240841.DAA28052@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
Laurent,
FindFit and all other similar built-in model-fitting routines apply
unconstrained local optimization methods, and hence these may miss the best
fit even w/o constraints in nonlinear models.
You could try instead an optimization based approach: then you can flexibly
specifiy your goodness-of-fit criterion, and also have the option to add
arbitrary constraints regarding the parameters. The function NMinimize or
our third party packages MathOptimizer and MathOptimizer Professional can
be put to use in this regard.
Regards,
Janos
___________________________
Janos D. Pinter, PhD, DSc
PCS Inc.
E-mail: jdpinter at hfx.eastlink.ca
Web: www.pinterconsulting.com
At 04:41 AM 3/24/2005, Laurent Feuz wrote:
>Hi
>
>If I want to fit some data using:
>
>output = FindFit[data,
> Izero*(1 + 1/(xiPEG^2*q^2)), {{Izero, 0.12}, {xiPEG, 10}}, q]
>
>I can assign a starting value to my parameters (like 0.12 for Izero in
>the above case), but can I also restrict its value to a certain range
>(like 0.10<Izero<0.14)?
>
>Thanks for any hints!
>regards
>Laurent
>
>
>
>
>--
>No virus found in this incoming message.
>Checked by AVG Anti-Virus.
>Version: 7.0.308 / Virus Database: 266.8.0 - Release Date: 3/21/2005
--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.308 / Virus Database: 266.8.0 - Release Date: 3/21/2005
- References:
- FindFit & restricting fitting parameter
- From: Laurent Feuz <laurent.feuz@mat.ethz.ch>
- FindFit & restricting fitting parameter