[R-sig-ME] Fwd: Band - Structure variance and covariance

[Ben Bolker]

   This went to r-sig-mixed-models-owner using r-project.org by accident, 
resending ...


-------- Forwarded Message --------
Subject: Re: Band - Structure variance and covariance
Date: Tue, 31 May 2022 10:38:24 -0400
From: Ben Bolker <bbolker using gmail.com>
To: Breno Gabriel <omatematico.breno using gmail.com>, 
r-sig-mixed-models-owner using r-project.org

   There is no lme package [sic], I'm going to assume that you're using 
the lme() *function* from the nlme package.  In principle you can define 
your own classes of covariance matrix, but in practice this is tricky / 
not very well documented. (If you want to dig in this I would take a 
look at the internal code of the correlation structures defined in the 
'ape' package ...

   Others may have more useful advice ...

On 2022-05-31 10:14 a.m., Breno Gabriel wrote:
> Hi.
> How could I introduce a matrix of *Band* variances and covariances into
> error matrix R?
> I have a mix of mixed effects where repeated measurements are not
> equidistant, but I did not find any computational routine or how to
> introduce this matrix through *R software*.
> I am using the *lme library*.
> 
> The matrix she has this structure (for example 3x3):
> 
> \sigma_1         \sigma_{12}       0
> \sigma_{12}    \sigma_2          \sigma_{23}
> 0                     \sigma_{23}     \sigma_3
> 
> 	[[alternative HTML version deleted]]
> 

-- 
Dr. Benjamin Bolker
Professor, Mathematics & Statistics and Biology, McMaster University
Director, School of Computational Science and Engineering
(Acting) Graduate chair, Mathematics & Statistics