R/ass.dp.r

In GJRM: Generalised Joint Regression Modelling

ass.dp <- function(ass.s, BivD, scc, sccn, nCa){

eps <- sqrt(.Machine$double.eps) # this looks fine here and it is not that dangerous 


if(BivD %in% scc)  ass.s <-  abs(ass.s)   
if(BivD %in% sccn) ass.s <- -abs(ass.s) 

GALs <- c("GAL0","GAL180","GAL90","GAL270","GAL0GAL90","GAL0GAL270","GAL180GAL90","GAL180GAL270")

if(!(BivD %in% c("AMH","FGM","PL","HO",GALs)))  i.rho <- BiCopTau2Par(family = nCa, tau = ass.s)
                                       
if(  BivD %in% c("AMH","FGM") )                 i.rho <- BiCopTau2Par(family = 1,   tau = ass.s)

if(  BivD %in% c("PL") )                        i.rho <- as.numeric(iTau(plackettCopula(), ass.s))



#if(  BivD %in% GALs    )                        i.rho <- as.numeric(iTau(galambosCopula(), ass.s))


if( BivD %in% GALs && BivD %in% scc )        i.rho <-  as.numeric(iTau(galambosCopula(),  ass.s))
if( BivD %in% GALs && BivD %in% sccn )       i.rho <- -as.numeric(iTau(galambosCopula(), -ass.s))



if(  BivD %in% c("HO") )                        i.rho <- ass.s - 1

if(BivD %in% c("N","AMH","FGM","T"))         i.rho <- atanh( i.rho ) 
if(BivD == "F")                              i.rho <- ifelse( abs(i.rho) < eps, eps, i.rho ) 
if(!(BivD %in% c("N","AMH","FGM","F","T")))  i.rho <- abs(i.rho)

if(BivD %in% c("C0","C180","C90","C270",
               "C0C90","C0C270","C180C90","C180C270",GALs))  i.rho <-  log(i.rho)   
               
if(BivD %in% c("J0","J180","J90","J270",
               "G0","G180","G90","G270",
               "J0J90","J0J270","J180J90","J180J270",
               "G0G90","G0G270","G180G90","G180G270"))       i.rho <-  log(i.rho - 1)   

names(i.rho) <- "theta.star"   

i.rho

}