MathGroup Archive 1998

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

Search the Archive

RE: Q about NonlinearFit




Sami wrote:
 ----------
|I need to use NonlinearFit for a set of data. The model is valid when
|each of the 2 parameters (that I need to determine) is not equal to 1
|(indeterminate when parameters = 1). What is the best way to go about
|evaluating the data ?
|

Does model[para1,para2] approach a value in the limit as (para1->1, 
para2->1).
If this is the case use the value you get in the limit for the case
where  (para1=1) or (para2=1).
Even if the limit is different from different directions you can pick a 
reasonable value for the case where (para1=1) or (para2=1).

A one variable case where the limit exists: Your model is Sin[x]/x.
In this case use    model[x_]:= If[x!=0, Sin[x]/x, 1]. Sin[0]/0 is
Indeterminate, but in the limit it's (1), so it works just fine.

A one variable case where the limit doesn't exist: Your model is (x^x).
You could use     model[x_]:= If[x!=0,  x^x,  1]. This way you get
(model[0]:=1).
(0^0)-is Indeterminate, but using (model[0]->1) is as good as anything
else.

Your example involves two parameters, but the above idea might get you 
started.

Ted Ersek




  • Prev by Date: Re: Two questions
  • Next by Date: Re: ImplicitPlot Problem
  • Prev by thread: Q about NonlinearFit
  • Next by thread: ImplicitPlot Problem