Package 'mvvg'

Calculate Matrix-Variate Variance Gamma Density

Description

Determines density of observations from a Matrix-variate variance gamma (MVVG) distribution, under the identifiability constraint set by [].

Usage

dmvvg(X, M, A, Sigma, Psi, gamma, log = FALSE)

Arguments

X	$p \times q$ observed matrix value
M	$p \times q$ location matrix
A	$p \times q$ skewness matrix
Sigma	$p \times p$ covariance matrix
Psi	$q \times q$ covariance matrix
gamma	scalar mixing parameter
log	returns log-likelihood if TRUE, default is FALSE.

Details

MVVG samples are formulated through the normal variance-mean mixture $M + WA + \sqrt{W}Z$, where $W \sim Gamma(\gamma, \gamma)$.

Gamma must be $>0$. Sigma and Psi must be positive definite covariance matrices.

Value

dmvvg returns the probability density corresponding to the inputted values and parameters.

Author(s)

Samuel Soon

See Also

rmvvg

Examples

M <- cbind(rep(1, 5), c(1, 0, 1, 0, 1))
A <- matrix(c(1,2), 5, 2, byrow = TRUE)
Sigma <- diag(5)
Psi <- matrix(c(4,2,2,3), 2, 2)
gamma <- 3

X <- rmvvg(1, M, A, Sigma, Psi, gamma)[[1]]
dmvvg(X, M, A, Sigma, Psi, gamma)

---

Example Matrix

Description

$5 \times 2$ matrix intended for use as an example in dmvvg.

Usage

example_matrix

Format

An object of class matrix (inherits from array) with 5 rows and 2 columns.

Author(s)

Samuel Soon

---

Generate Matrix-Variate Variance Gamma Samples

Description

Generates random samples from the matrix-variate variance gamma (MVVG) distribution, under the identifiability constraint set by [].

Usage

rmvvg(n, M, A, Sigma, Psi, gamma)

Arguments

n	number of observations
M	$p \times q$ location matrix
A	$p \times q$ skewness matrix
Sigma	$p \times p$ covariance matrix
Psi	$q \times q$ covariance matrix
gamma	scalar mixing parameter

Details

MVVG samples are formulated through the normal variance-mean mixture $M + WA + \sqrt{W}Z$, where $W \sim Gamma(\gamma, \gamma)$.

Gamma must be $>0$. Sigma and Psi must be positive definite covariance matrices.

Value

rmvvg returns a list of random samples.

Author(s)

Samuel Soon

See Also

dmvvg

Examples

M <- cbind(rep(1, 5), c(1, 0, 1, 0, 1))
A <- matrix(c(1,2), 5, 2, byrow = TRUE)
Sigma <- diag(5)
Psi <- matrix(c(4,2,2,3), 2, 2)
gamma <- 3

rmvvg(2, M, A, Sigma, Psi, gamma)

---