R help - Sep 2006 - defining error structure in bivariate mixed models

Hi,
 Using indicator variables I have been able to fit and run the code for
fitting a bivariate mixed model using unstructured covariance matrix

The code is

lme.fit1<- lme(one.var~-1+indic1+indic2+I(indic1*d.time)+I(indic2*d.time),
 random =~ -1+indic1+indic2|m.unit, weights = varIdent(~1|indic1)
,data = new.data)

My variables are
one.var :-  the two response variables stacked one after another
indic1 :- Indicator for variable one
indic2 :- indicator for variable two
d.time :- A covariate
m.unit :- the grouping units.

However
I want to do this with the error structure for the grouping unit as follows
sigma = ( sigma_1      0 )
             (    0    sigma_2)
where
 sigma_1 = sig1 * AR1(rho1) = error relating to variable 1
 sigma_2 = sig2* AR1(rho2) = error relating to variable 2

How can I do this using lme()?

 Any help is greatly appreciated.

Regards

Souvik Banerjee
Junior Research Fellow
Dept of Statistics
University of calcutta

	[[alternative HTML version deleted]]