dNIW: Density function for Normal-Inverse-Wishart (NIW) distribution.
Description
Get the density of a NIW sample. For a random vector mu, and a random matrix Sigma, the density function is defined as:
$$sqrt(2 pi^p |Sigma/k|)^{-1} exp(-1/2 (mu-m )^T (Sigma/k)^{-1} (mu-m)) (2^{(v p)/2} Gamma_p(v/2) |S|^{-v/2})^{-1} |Sigma|^{(-v-p-1)/2} exp(-1/2 tr(Sigma^{-1} S))$$
Where p is the dimension of mu and Sigma.
Usage
dNIW(mu, Sigma, m, k, v, S, LOG = TRUE)
Arguments
mu
numeric, the Gaussian sample.
Sigma
matrix, a symmetric positive definite matrix, the Inverse-Wishart sample.
m
numeric, mean of mu.
k
numeric, precision of mu.
v
numeric, degree of freedom of Sigma.
S
numeric, a symmetric positive definite scale matrix of Sigma, S is proportional to E(Sigma).
LOG
logical, return log density of LOG=TRUE, default TRUE.
Value
A numeric vector, the probability density of (mu,Sigma).
References
O'Hagan, Anthony, and Jonathan J. Forster. Kendall's advanced theory of statistics, volume 2B: Bayesian inference. Vol. 2. Arnold, 2004.
MARolA, K. V., JT KBNT, and J. M. Bibly. Multivariate analysis. AcadeInic Press, Londres, 1979.
# NOT RUN {S <- crossprod(matrix(rnorm(15),5,3))
Sigma <- crossprod(matrix(rnorm(15),5,3))
mu <- runif(3)
m <- runif(3)
dNIW(mu=mu,Sigma=Sigma,m=m,k=2,v=4,S=S,LOG = TRUE)
# }