dcondmvtnorm: Conditional density of Gaussian or Student subvectors

In lbelzile/mgp: Multivariate generalized Pareto models

Conditional density of Gaussian or Student subvectors

Description

The function computes the conditional density of (truncated) Gaussian or Student components corresponding to indices ind, given the values at the remaining index. The location vector mu and the scale matrix sigma are those of the d+p vector. The routine relies on the CDF approximation based on minimax exponential tilting implemented in the TruncatedNormal package.

Usage

dcondmvtnorm(x, ind, lbound = rep(-Inf, length(ind)), ubound = rep(Inf,
  length(ind)), mu, sigma, df = NULL, model = c("norm", "stud"),
  n = 500, log = FALSE)

Arguments

x	a p vector with the values of process at the remaining coordinates
ind	a d vector of integer indices in \{1, \ldots, d+p\} for which to compute the conditional density
lbound	d vector of lower bounds
ubound	d vector of upper bounds for truncated
mu	d+p vector of location parameters
sigma	d+p by d+p scale matrix
df	degrees of freedom of the d+p dimensional Student process
model	string indicating family, either norm for Gaussian or stud for Student-t
n	sample size for simulations. Default to 500.
log	logical; should log probability be returned? Default to FALSE.

Value

the conditional (log) density of the vector x[ind]

lbelzile/mgp documentation built on Aug. 5, 2023, 2:34 a.m.