MathGroup Archive 2007

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

Search the Archive

Re: Bounding Parameters used in FindFit[]

  • To: mathgroup at smc.vnet.net
  • Subject: [mg75242] Re: Bounding Parameters used in FindFit[]
  • From: Bill Rowe <readnewsciv at sbcglobal.net>
  • Date: Sat, 21 Apr 2007 23:15:24 -0400 (EDT)

On 4/20/07 at 12:30 AM, jlaforge at gmail.com (Joshua LaForge) wrote:

>Is there a method to limit (bound) the search for a least-squares
>fit with FindFit[] in Mathematica 5.2?

>For example I am using FindFit[] to fit a decaying exponential of
>the form P1 * Exp[-t/P2] + P3.  I know from the experiment that P3
>is smaller than the value Mathematica continues to find for that
>parameter.  It would be great if there was a way to bound the search
>for P3 to help Mathematica properly optimize P1 and P2.

Once you require an additional constraint such as P3<P1 && P3<P2
you no longer have a least squares problem and cannot use
=46indFit. You will need to create a function that computes the
sum of squares and use NMinimize on it. NMimimize performs a
constrained optimization.
--
To reply via email subtract one hundred and four


  • Prev by Date: RootSum 2
  • Next by Date: Re: equation solving
  • Previous by thread: Re: Bounding Parameters used in FindFit[]
  • Next by thread: Re: Help in analytical integration 2