This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: multifit_nlin on linear data
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: tlb at tlb dot org
- Cc: gsl-discuss at sources dot redhat dot com
- Date: Sun, 14 Jul 2002 21:34:23 +0100 (BST)
- Subject: Re: multifit_nlin on linear data
- References: <200207122024.g6CKODl4055949@tlb.blackwells.net>
tlb@tlb.org writes:
> I have a multiparameter fitting problem which is nonlinear in general,
> but has linear behavior in some regions where I can exactly calculate
> the derivatives. So sometimes when I use gsl_multifit_fdfsolver it's
> able to jump to the correct solution in a single jump, as you'd expect
> it to.
>
> Should I be using a different termination critereon than
> gsl_multifit_test_delta?
I'd recommend using a different termination criterion. The step size
is not a reliable indicator on its own. ETOLF indicates that the
algorithm can't go any further downhill, so you can usually
take that as signalling a minimum if you know that the behavior
of the function is reasonable.
regards
Brian Gough