[R] Prediction of the lme part of a gamm model estimated with mgcv

[niandra]

Dear useRs, 

I'm using the mgcv package for estimate a model with a smooth effect and a
spatial covarianza matrix.
I use the command predict(model$lme) to obtain a prediction with the fixed
and random part of the model, but i need this kind of prevision also on a
grid with different points than  the ones used to estimate the model.
I read this:
http://r.789695.n4.nabble.com/mgcv-gamm-predict-to-reflect-random-s-effects-td3622738.html
http://r.789695.n4.nabble.com/mgcv-gamm-predict-to-reflect-random-s-effects-td3622738.html 
which explain that the problem is "Basically gamm treats all random effects
as 'part of the noise' in the model specification".
In this discussion seems that there is a way (maybe really difficult) to
obtain a prevision of the lme part of a gamm model, but really i don't
understand how.
Someone can help me.

--
View this message in context: http://r.789695.n4.nabble.com/Prediction-of-the-lme-part-of-a-gamm-model-estimated-with-mgcv-tp4631977.html
Sent from the R help mailing list archive at Nabble.com.