[vsni.co.uk] Contact us | Site map
Forum | VSN International Forum Index
Author Message
Post new topic

<  ASReml  ~  matern class

j.hadfield@ed.ac.uk
Posted: Tue Feb 21, 2012 11:15 am Reply with quote
Joined: 13 Aug 2008 Posts: 26
Hi Everyone,

I'm trying to fit a spatial model with exponential correlation function. However, I can't recover simulation values and so I'm wondering whether I have misunderstood the syntax.

As I understand it leaving all mtrn values at their default should result in an isotropic exponential correlation function with scale parameter phi.

I have fitted the model

m.1<-asreml(y~1, random=~units, rcov=~mtrn(lat, long, phi=1), data=dat)

using Vu for the units variance and Vr for the residual variance I would expect the variance of y at position i to be Vu+Vr and the covariance between positions i and j to be Vr*exp(-d*phi) where d is the Euclidian distance between i and j.

However, I'm not recovering anything close to the true values of Vr, Vu and phi (or 1/phi). Any help would be welcome.

Thanks for any help,

Jarrod
View user's profile Send private message
j.hadfield@ed.ac.uk
Posted: Wed Feb 22, 2012 2:06 pm Reply with quote
Joined: 13 Aug 2008 Posts: 26
Hi,

Problem solved. Fitting the spatial model in the G structure:

m.1<-asreml(w.1~1, random=~mtrnv(lat, long, phi=1), rcov=~units, data=dat)

gives what I expect. This leads me to another question - my main question - but I will start a new post .....

Kind Regards,

Jarrod
View user's profile Send private message

Display posts from previous:  

All times are GMT
Page 1 of 1
Post new topic

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum