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

Re: Levenberg-Marquart code

• To: mathgroup at smc.vnet.net
• Subject: [mg19653] Re: Levenberg-Marquart code
• From: "Atul Sharma" <atulksharma at yahoo.com>
• Date: Tue, 7 Sep 1999 00:28:43 -0400
• References: <7qvthp$4n8@smc.vnet.net>
• Sender: owner-wri-mathgroup at wolfram.com

On the web page maintained by Lukas Finschi, you will find code for the L-M
algorithm written in C with an accompanying calling routine via Mathlink.
Not exactly what you were looking for, but it is ready to run from Mathematica 3.0.1
(or S-Plus) and has the advantage of being fast.

The URL is: http://www.ifor.math.ethz.ch/~finschi/Papers/LevMar.html

A.

----------------------------------------------------------------------------
--------------------------
Atul Sharma MD, FRCP(C)
Pediatric Nephrologist,
McGill University/Montreal Children's Hospital


Virgil Stokes wrote in message <7qvthp$4n8 at smc.vnet.net>...
>I have been comparing FindMinimum and NonLinearFit on several data sets.
>The results show that the Levenberg-Marquart algorithm
>nearly always performs better -- at least for the data sets that I have
used.
>
>Unfortunately,  the Levenberg-Marquart algorithm is not an option for
>FindMinimum
>(Mathematica 3.01). Does anyone have Mathematica code for the
Levenberg-Marquart
>
>algorithm?
>
>-- Virgil
>
>