invGaussconvfactor: Inverse Gaussian convolution factor model

In YafeiXu/CopulaModel: CopulaModel: Dependence Modeling with Copulas

Description

Inverse Gaussian convolution factor model

Usage

rIGconv(n,th0,thvec,ze=1)
pmIGfact(xvec,th0,thvec,zero=0)
pmIGfact.gl(xvec,th0,thvec,gl)
dmIGfact(xvec,th0,thvec,zero=0)
dmIGfact.gl(xvec,th0,thvec,gl)
dbIGfact(x1,x2,th0,th1,th2,zero=0)  
pmIGfcop.gl(uvec,param,gl)
dmIGfcop.gl(uvec,param,gl)

Arguments

n	sample size for simulation
th0	scalar for shape parameter of the shared/common component
thvec	vector of shape parameters of individual components, length d
param	parameter vector with length d+1 with th0,thvec
xvec	vector of length d with positive values
uvec	vector of length d with values in (0,1)
gl	Gauss-Legendre object with components $nodes and $weights
ze	non-convolution parameter zeta, can be set to 1 for copula
zero	tolerance for numerical integration, set as 0.0001 if there are numerial problems
x1	positive value for first variable (bivariate case)
x2	positive value for second variable (bivariate case)
th1	scalar for shape parameter of first variable (bivariate case)
th2	scalar for shape parameter of second variable (bivariate case)

Value

random sample (nxd matrix) for rIGconv

cdf or pdf for remaining functions

See Also

invGauss gammaconvfactor

Examples

n=1000
th0=2
thvec=c(.3,.3)
set.seed(123)
xdat=rIGconv(n,th0,thvec)
cor(xdat)
#plot(xdat)
gl=gausslegendre(25)
pmIGfact(c(1,1.1),th0,c(.4,.4),zero=0)
pmIGfact.gl(c(1,1.1),th0,c(.4,.4),gl)
# check that density is finite on diagonal
dbIGfact(1,1.1,th0,th1=.4,th2=.4,zero=0)
dmIGfact(c(1,1.1),th0,c(.4,.4),zero=0)
dmIGfact.gl(c(1,1.1),th0,c(.4,.4),gl)
dbIGfact(1,1.0001,th0,th1=.4,th2=.7,zero=0)
# copula 
pmIGfcop.gl(c(.5,.6),c(2,.4,.4),gl)
dmIGfcop.gl(c(.5,.6),c(2,.4,.4),gl)

YafeiXu/CopulaModel documentation built on May 9, 2019, 11:07 p.m.