Re: [Help-gsl] polynomial fitting using nonlinear least-squares fitting

[Brian Gough]

Lucas Grijander wrote:
> Hi all!
>
> I have to model some 3D data by using multidimensional polynomials. I'm 
> trying to use the nonlinear least-squares fitting functions from GSL. 
> I'm not sure if it is the best option, but I'm having problemes of 
> convergence. The polynomial function should be like this:
>
> Z = f(X,Y)
> Z = c0 + c1X + c2Y + c3X2 + c4Y2 + .....
>
> How should I define the function to minimize in the code? the order of 
> the polynomial is defined by the user, and it can be different for X and Y.

Hello,
You can use the linear least-squares for this (the function is linear in 
the coefficients c which is what matters).  There is a quadratic example 
in the linear least-squares chapter of the manual.

--
Brian Gough
(GSL Maintainer)

Network Theory Ltd,
Publishing the GSL Manual - http://www.network-theory.co.uk/gsl/manual/