33 lines
1.1 KiB
R
33 lines
1.1 KiB
R
|
# Saddlepoint approximation of the distribution function of the Esscher
|
||
|
# transform of the GAL distribution
|
||
|
#
|
||
|
# Author: Francois Pelletier
|
||
|
#
|
||
|
# LGPL 3.0
|
||
|
###############################################################################
|
||
|
|
||
|
#' Saddlepoint approximation of the distribution function of the Esscher
|
||
|
#' transform of the GAL distribution
|
||
|
#' @param x vector of quantiles
|
||
|
#' @param param Parameter vector
|
||
|
#' @param eval.time Time of the process
|
||
|
#' @param type Choose between "mu" or "kappa" parametrization
|
||
|
#' @param log Logical for log-parameters
|
||
|
|
||
|
#' @return Saddlepoint approximation of the distribution function
|
||
|
psaddleapproxEsscherGAL <- function(x,param,eval.time=1,type="mu",log=FALSE)
|
||
|
{
|
||
|
s <- saddlepointEsscherGAL(x,param,eval.time,type,log)
|
||
|
u <- s * sqrt(diffcgfEsscherGAL(s,2,param,eval.time,type,log))
|
||
|
w <- sign(s)*sqrt(2*(s*x-cgfEsscherGAL(s,param,type,log)))
|
||
|
|
||
|
if(x==mGAL(1,param,type,log))
|
||
|
{
|
||
|
1/2 + diffcgfEsscherGAL(0,3,param,eval.time,type,log)/
|
||
|
(6*sqrt(2*pi)*diffcgfEsscherGAL(0,2,param,eval.time,type,log)^(3/2))
|
||
|
}
|
||
|
else
|
||
|
{
|
||
|
pnorm(w)+dnorm(w)*(1/w-1/u)
|
||
|
}
|
||
|
}
|