R/Dbvn.R

In kkholst/mets: Analysis of Multivariate Event Times

##' Derivatives of the bivariate normal cumulative distribution function
##'
##' @title Derivatives of the bivariate normal cumulative distribution function
##' @param p Parameter vector
##' @param design Design function with defines mean, derivative of mean, variance,
##' and derivative of variance with respect to the parameter p
##' @param Y column vector where the CDF is evaluated
##' @author Klaus K. Holst
##' @usage
##' Dbvn(p,design=function(p,...) {
##'      return(list(mu=cbind(p[1],p[1]),
##'                dmu=cbind(1,1),
##'                S=matrix(c(p[2],p[3],p[3],p[4]),ncol=2),
##'                dS=rbind(c(1,0,0,0),c(0,1,1,0),c(0,0,0,1)))  )},                 
##'      Y=cbind(0,0))
##' @export
Dbvn <- function(p,design=function(p,...) {
                 return(list(mu=cbind(p[1],p[1]),
                             dmu=cbind(1,1),
                             S=matrix(c(p[2],p[3],p[3],p[4]),ncol=2),
                             dS=rbind(c(1,0,0,0),c(0,1,1,0),c(0,0,0,1)))
                        )},                 
                        Y=cbind(0,0)) {
  mS <- design(p)
  U0 <- with(mS,.Call("biprobit0",
                     mu,
                     S,dS,Y,dmu,NULL,FALSE,FALSE,
                     PACKAGE="mets"));
  return(c(U0,mS))
}