Re: sine regression

From: Fabian Mohn (fabianmohn_at_gmx.de)
Date: 12/28/04 

Date: Tue, 28 Dec 2004 17:16:13 +0100

Thanks for your offer, but I am writing a program for analyzing the data
from an interferometer during the etching of a surface and the data differs
strongly in different measurements.
Thanks anyway, Fabian

Paul Victor Birke <nonlinear@rogers.com> schrieb in im Newsbeitrag:
41D17A4D.8080805@rogers.com...
> Hallo Fabian!
>
> If you could release the data to my email maybe I could try a fit later
> today. It would be my pleasure!!
>
> Paul
>
> Fabian Mohn wrote:
> > Hello everybody,
> > I've got the following problem: I need to fit data (from an
interferometer)
> > to a function of the form
> > y(t) = a[1] * ( sin( a[2] * t + a[3] ) )^2 + a[4]
> > where a[i] are the unknown parameters. I am not an expert in numerics so
I
> > just tried to use the Levenberg-Marquardt-algorithm for nonlinear
regression
> > from the book "Numerical Recipes". The problem is that the algorithm
works
> > fine for any polynomial function but when I try to use it on the sine
> > function (or exponential functios) it does not find the minimum of the
> > chi^2 - function unless the initial guesses for the parameters are very
> > close to the actual values.
> > Does anybody have any suggestions what I could do about that or what
> > algorithm might be better? I already thought of expanding the sine
function
> > to a polynomial series but this did not work out too well either. Any
help
> > would be appreciated.
> >
> > Fabian
> >
> > -----------------------------------
> > Fabian Mohn
> > Universität Konstanz, Germany
> >
> >

Relevant Pages

  • Re: sine regression
    ... Fabian Mohn wrote: ... > fine for any polynomial function but when I try to use it on the sine ... I already thought of expanding the sine ... use ordinary linear least squares regression ...
    (sci.math.num-analysis)
  • Re: sine regression
    ... If you could release the data to my email maybe I could try a fit later ... Fabian Mohn wrote: ... > fine for any polynomial function but when I try to use it on the sine ... I already thought of expanding the sine function ...
    (sci.math.num-analysis)