Class "lmeLevel"

Description

An object representing a single level of random effects in a linear mixed-effects model.

Objects from the Class

Objects can be created by calls of the form new("lmeLevel", ...) or by calls to the generic constructor lmeLevel.

Slots

precision:

A precision matrix of class "pdMat"

groups:

The grouping factor (of class "factor") for the level.

columns:

An "integer" vector of the columns in the model matrix that correspond to this level.

parsInd:

An "integer" vector giving the subscripts in the overall parameter vector that apply to this level.

originalRows:

A list of integer vectors of row numbers in the original model matrix that correspond to each of the levels of the grouping factor.

decomposedRows:

A list of integer vectors of row numbers in the decomposed model matrix that correspond to each of the levels of the grouping factor.

storedRows:

A list of integer vectors of row numbers in the stored data matrix that correspond to each of the levels of the grouping factor.

nrow:

integer: the number of rows in the precision matrix.

updateFactor:

Object of class "matrix" - the matrix A used to evaluate the gradient of the log-likelihood and to formulate the ECME step.

hessianArray:

Object of class "array" - the array H used to evaluate the Hessian of the log-likelihood.

nlev:

integer: the number of levels in the grouping factor

Methods

EMupdate<-

signature(x = "lmeLevel", nlev = "missing", value = "matrix"): update the level within the EM or ECME algorithm

LMEgradient

signature(x = "lmeLevel", A = "matrix", nlev = "missing"): calculate the gradient of the log-likelihood with respect to the parameters from the decomposition of the scaled BLUPs and their variances.

coef

signature(object = "lmeLevel"): extract the parameters (from the precision matrix)

coef<-

signature(object = "lmeLevel", value = "numeric"): set the parameters.

Note

In reStruct objects the random-effects terms and the fixed-effects terms and the response are all represented by lmeLevel objects. The precision slot in the fixed-effects and response objects is an uninitialized pdMat objects and the nrow slot is zero.

Author(s)

Saikat DebRoy saikat@stat.wisc.edu and Douglas Bates bates@stat.wisc.edu

See Also

lmeLevel