Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2008

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

Search the Archive

RE: Intermediate Evaluation in FindMinimum

  • To: mathgroup at smc.vnet.net
  • Subject: [mg86657] RE: [mg86617] Intermediate Evaluation in FindMinimum
  • From: "Ausman, Kevin" <ausman at okstate.edu>
  • Date: Sun, 16 Mar 2008 04:40:28 -0500 (EST)
  • References: <200803130933.EAA01644@smc.vnet.net> <200803150810.DAA15236@smc.vnet.net> <2003.98.212.159.172.1205630116.squirrel@webmail.wolfram.com>

> Your chiSquare is having trouble when passed something it cannot fully
> evaluate. Instead try:
>
> chiSquare[xpt : {_?NumericQ ..}, mdl : {_?NumericQ ..}] :=
>  chiSquare[xpt, mdl] = Total[(xpt - mdl)^2]
>
> Daniel Lichtblau
> Wolfram Research

I was afraid that was going to be the suggestion. Unfortunately, when I 
do that I cannot use the Levenberg-Marquardt method (FindMinimum doesn't 
recognize chiSquare as being of the appropriate format), and the other 
methods don't seem to converge well for my data/model combination. I 
tried to find a way to specify the residuals manually to force it to use 
Levenberg-Marquardt, but then I end up with the same problem that I had 
when chiSquare wasn't doing the parameter-checking.

I will go back to my earlier plan of writing up the full problem and 
posting that instead of a simplified version. Hopefully I will have that 
put together (and documented) within a day or so.

Kevin Ausman



  • Prev by Date: Re: help to find position location
  • Next by Date: RE: Re: Pi is not a real number - the News from Mathematica Mathgroup
  • Previous by thread: Intermediate Evaluation in FindMinimum
  • Next by thread: Re: LevenbergMarquardt