[Gretl-devel] Re: Another extra.gfn candidate?
On Mon, 20 May 2019, Allin Cottrell wrote:
> I was thinking that perhaps we may want to change the signature
of the
> function a little to make it more flexible: how about modelling it after
> GSSmax (with the obvious adjustments)? That way, one could easily solve for a
> function with parameters.
That's a bigger job. I'll leave it undocumented for now and we can
revisit it for 2019c.
Fine by me.
Meanwhile I've fixed a bug in the simple version: the
"conv" condition
could be met without finding a zero.
I must admit I din't test it very thoroughly. Besides, when we go for the
"real thing" we may want to switch to a more sophisticated algorithm (but
then, we'll have to find an acceptable compromise between performance and
reliability).
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------