That should work fine. I recommend using the new interface
gsl_multifit_nlinear (just released in version 2.2.1), as there are many
new features and methods and the old multifit_fdfsolver will be
deprecated in a future release.
On 09/06/2016 08:39 AM, viktor drobot wrote:
Hi all! I know that GSL nonlinear least squares routines try to rescale
parameters so they should be approximate the same order of magnitude. In my
work typical range of parameter values is 10^-7 -- 10^4 (but oftne only
10^-5 -- 10^2)
Is that the extremal case for nonlinear fitting or NLS routines could
handle it successfully?
