regress
(Package: regress) :
Fit a Gaussian Linear Model with Linear Covariance Structure
Fits Gaussian linear models in which the covariance structure can be expressed as a linear combination of known matrices. For example, random effects, block effects models and spatial models that include a nugget effect. Fits model by maximising the log-likelihood of the model. The choice of kernel affects which likelihood and by default it is the REML log likelihood or restricted log likelihood but the ordinary log-likelihood is also possible. The regress algorithm uses a Newton-Raphson algorithm to locate the maximum of the log-likelihood surface. Some computational efficiencies are achieved when all variance components are associated with factors. In such a random effects model the matrix inversion is computed using the Sherman-Morrison-Woodbury identities.