[Lme4-commits] r1427 - pkg/lme4Eigen/man
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Tue Oct 18 22:31:25 CEST 2011
Author: dmbates
Date: 2011-10-18 22:31:24 +0200 (Tue, 18 Oct 2011)
New Revision: 1427
Removed:
pkg/lme4Eigen/man/chmSp.Rd
pkg/lme4Eigen/man/deFeMod.Rd
pkg/lme4Eigen/man/feModule-class.Rd
pkg/lme4Eigen/man/glmFamily.Rd
pkg/lme4Eigen/man/glmerResp.Rd
pkg/lme4Eigen/man/lmerResp-class.Rd
pkg/lme4Eigen/man/lmerResp.Rd
pkg/lme4Eigen/man/reModule-class.Rd
pkg/lme4Eigen/man/reModule.Rd
Log:
Remove documentation of no-longer-used classes from Rcpp modules
Deleted: pkg/lme4Eigen/man/chmSp.Rd
===================================================================
--- pkg/lme4Eigen/man/chmSp.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/chmSp.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,29 +0,0 @@
-\name{chmSp}
-\alias{chmSp}
-\docType{class}
-\title{
- An C++ class for compressed column-oriented sparse matrices.
-}
-\description{
- A C++Class for \code{\linkS4class{dgCMatrix}} objects.
-}
-\section{Constructors}{
- \describe{
- \item{\code{chmSp(Rcpp::S4)}:}{from an
- \code{\linkS4class{dgCMatrix}} object}
- }
-}
-\section{Fields}{
- \describe{
- \item{\code{ncol}}{integer - number of columns (read only)}
- \item{\code{nnz}}{integer - number of non-zeros (read only)}
- \item{\code{nrow}}{integer - number of rows (read only)}
- \item{\code{nzmax}}{integer - maximum number of non-zeros (read
- only)}
- }
-}
-%\details{}
-%\source{}
-%\references{}
-%\examples{}
-\keyword{datasets}
Deleted: pkg/lme4Eigen/man/deFeMod.Rd
===================================================================
--- pkg/lme4Eigen/man/deFeMod.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/deFeMod.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,71 +0,0 @@
-\name{deFeMod}
-\alias{deFeMod}
-\alias{coerce,Rcpp_deFeMod,deFeMod-method}
-\alias{show,Rcpp_deFeMod-method}
-\docType{class}
-\title{
- An C++ class for information related to the fixed-effects parameters.
-}
-\description{
- This C++Class encapsulates the information associated with the fixed
- effects in a mixed-effects model.
-}
-\section{Constructors}{
- \describe{
- \item{\code{deFeMod(Rcpp::S4, int)}:}{from an S4
- \code{\linkS4class{deFeMod}} object and the number of
- observations, \code{n}}
- \item{\code{deFeMod(X, n, q)}}{from the dense model matrix, \code{X}
- and dimensions}
- }
-}
-\section{Read-Only Fields}{
- \describe{
- \item{\code{CcNumer}}{numeric scalar - contribution to numerator of
- the convergence criterion}
- \item{\code{RX}}{\code{\linkS4class{Cholesky}} - dense Cholesky
- factor}
- \item{\code{RZX}}{\code{\linkS4class{dgeMatrix}} - Off-diagonal
- block in large Cholesky factor}
- \item{\code{UtV}}{\code{\linkS4class{dgeMatrix}} - Weighted
- crossproduct}
- \item{\code{V}}{\code{\linkS4class{dgeMatrix}} - Scaled model matrix}
- \item{\code{VtV}}{\code{\linkS4class{dpoMatrix}} - Weighted
- crossproduct}
- \item{\code{Vtr}}{numeric - Weighted crossproduct of model matrix
- and residuals}
- \item{\code{X}}{\code{\linkS4class{dgeMatrix}} - original model
- matrix}
- \item{\code{coef}}{numeric - coefficient vector after increment}
- \item{\code{ldRX2}}{numeric - log of the square of the determinant
- of \code{RX}}
- }
-}
-\section{Read-Write Fields}{
- \describe{
- \item{\code{coef0}}{numeric - base coefficient vector}
- \item{\code{incr}}{numeric - increment for coefficient vector}
- }
-}
-\section{Methods}{
- \describe{
- \item{\code{linPred1(fac)}}{numeric - update \code{coef} to
- \code{coef0 + fac*incr} and return linear predictor contribution}
- \item{\code{reweight(sqrtXwt, wtres)}}{update V and Vtr for new
- weights.}
- \item{\code{solveIncr()}}{update \code{fac} from \code{VtV} and
- solve for \code{incr} from \code{Vtr}}
- \item{\code{updateIncr(cu)}}{update \code{incr} given
- \code{cu}}
- \item{\code{updateRzxpRxp(Lambdap, Lp)}}{update the
- triangular factor sections given external pointers to
- \code{Lambda} and \code{L}}
- \item{\code{updateUtV(Utp)}}{update \code{UtV} given an external
- pointer to \code{Ut}}
- }
-}
-%\details{}
-%\source{}
-%\references{}
-%\examples{}
-\keyword{datasets}
Deleted: pkg/lme4Eigen/man/feModule-class.Rd
===================================================================
--- pkg/lme4Eigen/man/feModule-class.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/feModule-class.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,41 +0,0 @@
-\name{feModule-class}
-\Rdversion{1.1}
-\docType{class}
-\alias{feModule-class}
-\alias{deFeMod-class}
-\alias{spFeMod-class}
-\title{Class "deFeMod"}
-\description{
- The \code{"deFeMod"} and \code{"spFeMod"} classes are response module
- classes for the fixed-effects terms in a mixed model. They inherit
- from the \code{"\linkS4class{predModule}"} classes
- \code{"\linkS4class{dPredModule}"} and
- \code{"\linkS4class{sPredModule}"}, respectively and contain
- additional slots related to the random-effects module. These
- additional slots are inherited from the
- \code{"\linkS4class{feModule}"} class.
-}
-\section{Objects from the Class}{
- Objects from these classes are usually created as part of
- \code{\linkS4class{merMod}} object returned by functions
- \code{\link{lmer}}, \code{\link{glmer}} or \code{\link{nlmer}}.
-}
-\section{Slots}{
- \describe{
- \item{\code{RZX}:}{The off-diagonal part of the Cholesky
- decomposition for the penalized least squares problem - of class
- \code{"\linkS4class{dgeMatrix}"} for \code{"deFeMod"} objects or
- \code{"\linkS4class{dgCMatrix}"} for \code{"spFeMod"} objects.}
- \item{\code{coef}, \code{X}, \code{fac}:}{Inherited from the
- appropriate \code{"\linkS4class{predModule}"} class.}
- }
-}
-\section{Methods}{
-No methods defined with class "feModule" in the signature.
-}
-\seealso{\code{\linkS4class{merMod}}}
-\examples{
-showClass("deFeMod")
-showClass("spFeMod")
-}
-\keyword{classes}
Deleted: pkg/lme4Eigen/man/glmFamily.Rd
===================================================================
--- pkg/lme4Eigen/man/glmFamily.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/glmFamily.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,66 +0,0 @@
-\name{glmFamily}
-\alias{glmFamily}
-\docType{class}
-\title{
- An C++ class for glm family objects
-}
-\description{
- An Rcpp module for a reference class representing a glm family
- object. The constructor takes a \code{\link{family}} object. The
- instance provides methods for functions in the family, making these
- available in \R{} and in \code{C++}.
-}
-\section{Constructors}{
- \describe{
- \item{\code{glmFamily(Rcpp::List)}:}{from an (S3)
- \code{\link{family}} object}
- }
-}
-\section{Fields}{
- \describe{
- \item{\code{family}}{character - name of the family}
- \item{\code{link}}{character - name of the link}
- }
-}
-\section{Methods}{
- \describe{
- \item{\code{devResid(mu, weights, y)}}{numeric - evalute the vector
- of deviance residuals}
- \item{\code{linkFun(mu)}}{numeric - evaluate \code{eta}}
- \item{\code{linkInv(eta)}}{numeric - evaluate \code{mu}}
- \item{\code{muEta(eta)}}{numeric - evaluate the derivative of
- \code{mu} with respect to \code{eta}}
- \item{\code{variance(mu)}}{numeric - evaluate the variance}
- }
-}
-\details{
- A glm \code{\link{family}} object is a list of functions including the
- link function (\code{linkfun}), the inverse link function
- (\code{linkinv}), the \code{variance} function, the \code{mu.eta}
- function that calculates the derivative of the mean with respect to
- the linear predictor, and a function that evaluates the (squared)
- deviance residuals.
-
- A \code{Rcpp_glmFamily} object allows these functions to be accessed
- within \code{C++} code or within \R{} code.
-
- For several common families the transformation functions:
- \code{linkFun}, \code{linkInv}, \code{muEta} and \code{variance}, are
- evaluated in compiled code. }
-
-\note{
- This class is primarily used for testing. It is not used directly.
-}
-\seealso{\code{\link{glmerResp}}, \code{\link{glmer}}}
-%\source{}
-%\references{}
-\examples{
-bFam <- new(glmFamily, binomial())
-bFam$family
-bFam$link
-set.seed(1)
-mu <- sort(runif(10)) # mean vector with elements in (0,1)
-eta <- bFam$linkFun(mu) # linear predictor
-cbind(mu, eta, muEta=bFam$muEta(eta), variance=bFam$variance(mu))
-}
-\keyword{methods}
Deleted: pkg/lme4Eigen/man/glmerResp.Rd
===================================================================
--- pkg/lme4Eigen/man/glmerResp.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/glmerResp.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,122 +0,0 @@
-\name{glmerResp}
-\alias{glmerResp}
-\alias{show,Rcpp_glmerResp-method}
-\docType{class}
-\title{
- An C++ class for \code{glmRespModule} objects
-}
-\description{
- The \code{"glmRespModule"} S4 class represents the information
- associated with the response in a \code{\link{glm}} or
- \code{\link{glmer}} model. This C++Class encapsulates the
- information and provides methods accessible in \R{} or in \code{C++}.
-}
-\section{Constructors}{
- \describe{
- \item{\code{glmerResp(Rcpp::S4)}:}{from a
- \code{\linkS4class{glmRespMod}} object}
- \item{\code{glmerResp(Rcpp::List, Rcpp::NumericVector)}}{from a glm
- \code{\link{family}} and the response \code{y}}
- \item{\code{glmerResp(Rcpp::List, Rcpp::NumericVector, Rcpp::NumericVector)}}{from a glm
- \code{\link{family}}, the response \code{y} and the \code{weights}}
- \item{\code{glmerResp(Rcpp::List, Rcpp::NumericVector,
- Rcpp::NumericVector, Rcpp::NumericVector)}}{from a glm
- \code{\link{family}}, the response \code{y}, the \code{weights} and
- the \code{offset}}
- \item{\code{glmerResp(Rcpp::List, Rcpp::NumericVector,
- Rcpp::NumericVector, Rcpp::NumericVector, Rcpp::NumericVector)}}{from a glm
- \code{\link{family}}, the response \code{y}, the \code{weights},
- the \code{offset} and the number of replicates, \code{n} (only
- applies to the binomial family).}
- }
-}
-\section{Read-Only Fields}{
- \describe{
- \item{\code{devResid}}{numeric - deviance residuals}
- \item{\code{eta}}{numeric - linear predictor}
- \item{\code{family}}{character - name of glm family}
- \item{\code{link}}{character - name of link function}
- \item{\code{mu}}{numeric - mean vector}
- \item{\code{muEta}}{numeric - derivative of \code{mu} with respect
- to \code{eta}}
- \item{\code{residDeviance}}{numeric scalar - sum of squared deviance
- residuals}
- \item{\code{sqrtWrkWt}}{numeric - square root of the weights applied
- to the working residuals or working response}
- \item{\code{sqrtXwt}}{numeric - square root of the weights applied
- to the model matrix}
- \item{\code{sqrtrwt}}{numeric - square root of the weights applied
- to the residuals}
- \item{\code{variance}}{numeric - vector of unscaled variances}
- \item{\code{weights}}{numeric - prior weights}
- \item{\code{wrkResids}}{numeric - working residuals, on the scale of
- the linear predictor}
- \item{\code{wrkResp}}{numeric - working response, on the scale of
- the linear predictor}
- \item{\code{wrss}}{numeric scalar - weighted sum of squared
- residuals}
- \item{\code{wtres}}{numeric - weighted residuals}
- \item{\code{y}}{numeric - response vector}
- }
-}
-\section{Read-Write Fields}{
- \describe{
- \item{\code{offset}}{numeric - the offset vector, always present
- even if zero}
- \item{\code{pwrss}}{numeric scalar - penalized, weighted sum of
- squared residuals}
- }
-}
-\section{Methods}{
- \describe{
- \item{\code{Laplace(ldL2, ldRX2, sqrLenU)}}{Return the Laplace
- approximation to the profiled deviance. The \code{ldRX2} argument
- is for compatibility; it is not used in the calculation.}
- \item{\code{updateMu(eta)}}{Install a new value of \code{eta} and
- update \code{mu} and derived quantities. Note: the
- \code{sqrtrwt} vector is not updated but the \code{sqrtXwt} vector
- is.}
- \item{\code{updateWts()}}{Update \code{sqrtrwt} and \code{sqrtXwt}
- from the new value of the \code{variance}.}
- }
-}
-
-\details{
- After an object of this class has been constructed, it can be modified
- in two ways: the \code{updateMu} method takes a new value of the
- linear predictor, \code{eta}, and updates the mean, \code{mu}, and the
- weighted residuals, \code{wtres}, returning the weighted residual sum
- of squares. The \code{updateWts} method, which has no arguments,
- updates the variance and the residual and X weights (\code{sqrtrwt}
- and \code{sqrtXwt}.
-
- The implementation of penalized iteratively reweighted least squares
- (PIRLS) in \code{\link{glmer}} uses a Gauss-Newton algorithm to update
- the random-effects and optionally the fixed-effects parameters. In
- this algorithm the weighted residuals, \code{wtres}, and the X
- weights, \code{sqrtXwts}, are all that is needed by the fixed-effects
- and random-effects modules to create a new linear predictor. However,
- to determine an initial value for these coefficients we use one Fisher
- scoring iteration which requires the working response, \code{wrkResp},
- and the working weights, \code{sqrtWrkWt}.
-}
-%\source{}
-%\references{}
-\examples{
-bb <- binomial()
-set.seed(1)
-eta0 <- sort(rnorm(10))
-mu0 <- bb$linkinv(eta0)
-y <- rbinom(length(mu0), 1, mu0)
-gg <- new(glmerResp, bb, y)
-gg$family
-gg$link
-gg$updateMu(eta0) # returns weighted residual sum of squares
-gg$updateWts() # returns weighted residual sum of squares
-comp <- c("y", "eta", "mu", "weights", "offset", "variance", "sqrtrwt",
- "wtres", "wrkResids", "wrkResp", "sqrtWrkWt")
-names(comp) <- comp
-sapply(comp, function(nm) gg[[nm]])
-gg$residDeviance
-}
-\keyword{datasets}
Deleted: pkg/lme4Eigen/man/lmerResp-class.Rd
===================================================================
--- pkg/lme4Eigen/man/lmerResp-class.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/lmerResp-class.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,32 +0,0 @@
-\name{lmerResp-class}
-\Rdversion{1.1}
-\docType{class}
-\alias{lmerResp-class}
-\title{"lmerResp" class}
-\description{
- The \code{"lmerResp"} class is a trivial extension of the
- \code{\linkS4class{respModule}} class% from the \pkg{Matrix} package.
- It adds an integer scalar slot, \code{REML}, which is either
- zero, for maximum likelihood estimation, or \code{p}, the number of
- columns in the fixed-effects model matrix, for REML estimation.
-}
-\section{Objects from the Class}{
- Objects from these classes are usually created as part of
- \code{\linkS4class{merMod}} object returned by \code{\link{lmer}}.
-}
-\section{Slots}{
- \describe{
- \item{\code{REML}:}{Integer indicator of whether to estimate
- parameters according to the REML criterion - \code{lmerResp}
- objects only.}
- }
- All other slots are inherited from the \code{\linkS4class{respModule}}
- class of the \pkg{Matrix} package.
-}
-\section{Methods}{
-No methods defined with class "merResp" in the signature.
-}
-\examples{
-showClass("lmerResp")
-}
-\keyword{classes}
Deleted: pkg/lme4Eigen/man/lmerResp.Rd
===================================================================
--- pkg/lme4Eigen/man/lmerResp.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/lmerResp.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,78 +0,0 @@
-\name{lmerResp}
-\alias{lmerResp}
-\alias{coerce,Rcpp_lmerResp,lmerResp-method}
-\alias{show,Rcpp_lmerResp-method}
-\docType{class}
-\title{
- An C++ class for \code{lmerResp} objects
-}
-\description{
- The \code{\linkS4class{lmerResp}} S4 class represents the information
- associated with the response in an \code{\link{lm}} or
- \code{\link{lmer}} model. This C++Class encapsulates the information
- and provides methods accessible in \R{} or in \code{C++}.
-}
-\section{Constructors}{
- \describe{
- \item{\code{lmerResp(Rcpp::S4)}:}{from a
- \code{\linkS4class{lmerResp}} object}
- \item{\code{lmerResp("integer", "numeric")}}{from the REML indicator
- and the response \code{y}}
- \item{\code{lmerResp("integer", "numeric", "numeric")}}{from the REML
- indicator, the response \code{y} and the \code{weights}}
- \item{\code{lmerResp("integer", "numeric", "numeric",
- "numeric")}}{from the REML indicator, the response \code{y}, the
- \code{weights} and the \code{offset}}
- }
-}
-\section{Read-Only Fields}{
- \describe{
- \item{\code{REML}}{scalar integer - 0L for ML estimation and p for
- REML}
- \item{\code{mu}}{numeric - mean vector (read only)}
- \item{\code{offset}}{numeric - the offset vector, always present
- even if zero (read/write)}
- \item{\code{sqrtXwt}}{numeric - square root of the weights applied
- to the model matrix (read only)}
- \item{\code{sqrtrwt}}{numeric - square root of the weights applied
- to the residuals (read only)}
- \item{\code{weights}}{numeric - prior weights (read/write)}
- \item{\code{wrss}}{numeric scalar - weighted sum of squared
- residuals (read only)}
- \item{\code{wtres}}{numeric - weighted residuals (read only)}
- \item{\code{y}}{numeric - response vector (read only)}
- }
-}
-\section{Methods}{
- \describe{
- \item{\code{Laplace(ldL2, ldRX2, sqrLenU)}}{Return the profiled REML
- criterion or the profiled deviance, according to the setting of
- the REML field}
- \item{\code{updateMu(eta)}}{Update \code{mu} and derived quantities
- given a value of the linear predictor, \code{eta}.}
- \item{\code{updateWts()}}{Update \code{sqrtrwt} and
- \code{sqrtXwt}. (Not generally used for this class.)}
- }
-}
-\details{
- The methods in this class parallel those in the
- \code{"\link{glmerResp}"} class but typically \code{updateWts} is
- not called. The \code{updateMu} method is called after every change
- in the variance component parameter, providing an updated value of the
- residual sum of squares.
-
- The first argument to the constructors (other than from an S4 object)
- is the REML indicator, which is an integer value of 0, for
- ML estimates, or \code{p}, the rank of the fixed-effects model matrix,
- for REML estimates.
-}
-%\source{}
-%\references{}
-\examples{
-ll <- with(sleepstudy, new(lmerResp, 0L, Reaction))
-comp <- c("y", "mu", "weights", "offset", "sqrtrwt", "wtres")
-names(comp) <- comp
-head(sapply(comp, function(nm) ll[[nm]]), n = 10)
-ll$wrss
-}
-\keyword{methods}
Deleted: pkg/lme4Eigen/man/reModule-class.Rd
===================================================================
--- pkg/lme4Eigen/man/reModule-class.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/reModule-class.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,61 +0,0 @@
-\name{reModule-class}
-\title{Class "reModule" of Random Effect Modules}
-\Rdversion{1.1}
-\docType{class}
-\alias{reModule-class}
-\alias{reTrms-class}
-\alias{rcond,reTrms,character-method}
-\description{
- The \code{"reModule"} class incorporates information about the
- random effects in a mixed models. The \code{"reTrms"} subclass
- includes information about random-effects terms from a model
- formula.
-}
-\section{Objects from the Class}{
- Objects from these classes are usually created as part of
- \code{\linkS4class{merMod}} object returned by functions
- \code{\link{lmer}}, \code{\link{glmer}} or \code{\link{nlmer}}.
-}
-\section{Slots}{
- \describe{
- \item{\code{L}:}{Sparse Cholesky factor of class \code{"CHMfactor"}}
- \item{\code{Lambda}:}{Sparse matrix (class \code{"dgCMatrix"})
- representation of the relative covariance factor of the random
- effects}
- \item{\code{Lind}:}{integer vector of indices by which \code{theta}
- generates \code{Lambda}. Must have
- \code{length(Lind)==length(Lambda at x)} and \code{all(Lind \%in\%
- 1:length(theta))}. The update operation is of the form
- \code{Lambda at x[] = theta[Lind]}.}
- \item{\code{Zt}:}{Sparse model matrix, \code{"dgCMatrix"}, for the
- random effects.}
- \item{\code{lower}:}{Lower bounds on the elements of \code{theta}}
- \item{\code{theta}:}{Covariance parameter vector.}
- \item{\code{u}:}{Orthogonal random effects vector defined so that
- \code{b = Lambda \%*\% u}.}
- \item{\code{flist}:}{list of grouping factors from the
- random-effects terms (\code{"reTrms"} only). Its \code{assign}
- attribute matches terms with grouping factors.}
-%%-nL \item{\code{nLevs}:}{number of levels of each grouping factor. Is
-%%-nL identical to \code{sapply(flist, function(f) length(levels(f)))},
-%%-nL saved, as often used.}
- \item{\code{cnms}:}{list of column names from random-effects terms
- (\code{"reTrms"} only).}
- }
-}
-\section{Methods}{
- \describe{
- \item{rcond}{\code{signature(x = "reTrms", norm = "character")}:
- Compute the reciprocal condition numbers (see \code{\link{rcond}})
- for each unique lower triangular block of \eqn{\Lambda}. If one
- is small, say smaller than \code{1e-6}, the variance-covariance
- matrix of the corresponding random effect is close to singular.}
- }
-}
-\seealso{\code{\link{getME}}
-}
-\examples{
-showClass("reModule")
-showClass("reTrms")
-}
-\keyword{classes}
Deleted: pkg/lme4Eigen/man/reModule.Rd
===================================================================
--- pkg/lme4Eigen/man/reModule.Rd 2011-10-18 17:48:09 UTC (rev 1426)
+++ pkg/lme4Eigen/man/reModule.Rd 2011-10-18 20:31:24 UTC (rev 1427)
@@ -1,73 +0,0 @@
-\name{reModule}
-\alias{reModule}
-\alias{coerce,Rcpp_reModule,reModule-method}
-\alias{show,Rcpp_reModule-method}
-\docType{class}
-\title{
- An C++ class for information related to random effects
-}
-\description{
- This C++Class encapsulates the information associated with the random
- effects in a mixed-effects model.
-}
-\section{Constructors}{
- \describe{
- \item{\code{reModule(Rcpp::S4)}:}{from an S4
- \code{\linkS4class{reModule}} object}
- \item{\code{reModule(Zt, Lambda, L, Lind, lower)}}{from components}
- }
-}
-\section{Read-Only Fields}{
- \describe{
- \item{\code{CcNumer}}{numeric scalar - contribution to numerator of
- the convergence criterion}
- \item{\code{L}}{\code{\linkS4class{CHMfactor}} - sparse Cholesky
- factor}
- \item{\code{Lambda}}{\code{\linkS4class{dgCMatrix}} - relative
- covariance factor}
- \item{\code{Lambdap}}{external pointer to \code{Lambda}}
- \item{\code{Lind}}{integer - 1-based index vector into theta for
- \code{Lambda at x}}
- \item{\code{Lp}}{external pointer to \code{L}}
- \item{\code{Ut}}{external pointer to weighted, orthogonal design
- matrix}
- \item{\code{Zt}}{\code{\linkS4class{dgCMatrix}} - transpose of the
- model matrix for the random effects}
- \item{\code{b}}{numeric - random effects on original scale}
- \item{\code{cu}}{numeric - intermediate solution for \code{u}}
- \item{\code{ldL2}}{numeric - logarithm of the square of the
- determinant of \code{L}}
- \item{\code{linPred}}{numeric - linear predictor contribution based
- on current \code{u}}
- \item{\code{lower}}{numeric - lower bounds on components of
- \code{theta}}
- \item{\code{sqrLenU}}{numeric - squared length of u, the orthogonal
- random effects}
- \item{\code{u}}{numeric - orthogonal random effects vector}
- }
-}
-\section{Read-Write Fields}{
- \describe{
- \item{\code{incr}}{numeric - increment for \code{u}}
- \item{\code{theta}}{numeric - variance component parameters}
- \item{\code{u0}}{numeric - base orthogonal random effects vector}
- }
-}
-\section{Methods}{
- \describe{
- \item{\code{installU0()}}{install the current value of \code{u} as \code{u0}}
- \item{\code{linPred1(fac)}}{update \code{u} as \code{u0 + fac*incr}
- and return linear predictor contribution}
- \item{\code{reweight(sqrtXwt, wtres, useU0)}}{update L, Ut and cu
- for new weights. The value of \code{u} is taken from \code{u0} or
- from \code{u0} according to whether or not \code{useU0} is TRUE}
- \item{\code{solveIncr()}}{solve for \code{incr} only. Returns squared length of \code{c1}}
- \item{\code{updateIncr(cu)}}{solve for \code{incr} given the updated
- \code{cu} from the feModule's \code{updateBeta} method}
- }
-}
-%\details{}
-%\source{}
-%\references{}
-%\examples{}
-\keyword{datasets}
More information about the Lme4-commits
mailing list