[R-sig-ME] Specifying pdMat classes in lme

Christoph Scherber Christoph.Scherber at agr.uni-goettingen.de
Wed Oct 5 17:15:04 CEST 2011


Dear all,

I try to re-specify an lme model using the pdMat classes in order to then be able to remove some restrictions on the variance-covariance matrices.

My model has the following random effects structure: ~time | b/c

(where b and c are different split-plot sizes in an agricultural experiment).

What would be an equivalent formulation of such a random effects structure using pdBlocked, pdCompSymm etc.?

I believe it´s something like

random=list(b=pdCompsymm(~c-1)

but how would I include the different slopes for time here?

Many thanks for any help,
Christoph


(using R 2.13.1 in Windows 7 64 Bit, lme version 3.1-102)


More information about the R-sig-mixed-models mailing list