Re: [Help-gsl] Nonlinear least-square-fitting parameter domain

[Patrick Alken]

On Tue, Feb 20, 2007 at 01:44:09AM -0300, Ricardo Zapata wrote:
> Hi,
> 
> Is there any way to give the parameters' range to the solver for 
> nonlinear square-fitting? I need my parameters to be > 0. I used 
> GSL_EDOM in my functions (f and df), but this makes the program to abort 
> instead of trying a new parameter value.

One way I've used in the past is to add a large value to the error
if your params are > 0.