geoR

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

geoR

Anders Malmberg
Hi,

A quick question on the likfit routine in the geoR package.
It is not possible to bound both tausq.rel and sigmasq. If both of them
are unknown the optimization is carried out wrt tausq.rel
and sigmasq is not considered in the optimization. Correct?

bon weekend.

best regards,
Anders Malmberg



Reply | Threaded
Open this post in threaded view
|

geoR

Ole F. Christensen
Dear Anders

* The maximum likelihood estimate of the partial sill parameter sigmasq
can be expressed explictly as function of the other covariance parameters.
Therefore parameter estimate for sigmasq is NOT found by numerical
maximisation, and as consequence in likfit you cannot specify that that
sigmasq should be within certain limits.

* The parameter estimate for the relative nugget tausq.rel is found by
numerical maximisation. Therefore you can specify that this parameter
should be within certain limits.

Best

Ole

Anders Malmberg wrote:

> Hi,
>
> A quick question on the likfit routine in the geoR package.
> It is not possible to bound both tausq.rel and sigmasq. If both of them
> are unknown the optimization is carried out wrt tausq.rel
> and sigmasq is not considered in the optimization. Correct?
>
> bon weekend.
>
> best regards,
> Anders Malmberg
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>

--
Ole F. Christensen
BiRC - Bioinformatics Research Center
University of Aarhus