[Date Index] [Thread Index] [Author Index]

Re: Levenberg-Marquardt ?

• To: mathgroup at smc.vnet.net
• Subject: [mg23780] Re: Levenberg-Marquardt ?
• From: Rob Knapp <rknapp at wolfram.com>
• Date: Sat, 10 Jun 2000 02:59:25 -0400 (EDT)
• Organization: Wolfram Research, Inc.
• References: <8hfddu$hgp@smc.vnet.net>
• Sender: owner-wri-mathgroup at wolfram.com

Alastair McLean wrote:
> 
> Has the Levenberg-Marquard method of non-linear least
> squares minimisation been implemented as a Mathematica
> function or package yet ?
> 

This is implemented in FindMinimum.

If the objective function is structurally a sum of squares, it is
invoked automatically instead of the usual default method.  You can also
explicitly use the option Method->LevenbergMarquardt

One common use of this is through the package Statistics`NonlinearFit`
which uses this method to do a nonlinear data fit.

Rob Knapp
Wolfram Research