[R-sig-ME] Another MCMCglmm question - fixing correlations

[Jarrod Hadfield]

Hi Malcolm,

I'm glad MCMCglmm seems to fit spatial simultaneous autoregressive lag  
models . I had never heard of them before, but looking at some of the  
literature I agree they can be modeled using the sir function.  The  
code is not thoroughly tested (I have only used it on one real data  
set, for which it was intended) and is not efficient yet for large  
models, so I would treat with caution. In particular MCMCglmm is  
constraining rho such that L (I-rho*W) is positive definite and I am  
not sure this is a valid strategy for all SIR models.  Any pointers,  
particularly to the relevant literature would be welcome.

Cheers,

Jarrod




On 5 Dec 2010, at 16:16, Malcolm Fairbrother wrote:

> spatial simultaneous autoregressive lag models


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.