tmvrnorm.ExpT: '"tmvrnorm"' Sampler for '"ExpT"' (Exponential Tilting) S3...

In lineqGPR: Gaussian Process Regression Models with Linear Inequality Constraints

Description

Sampler for truncated multivariate normal distributions via exponential tilting using the package TruncatedNormal (Botev, 2017).

Usage

## S3 method for class 'ExpT'
tmvrnorm(object, nsim, control = NULL, ...)

Arguments

object	an object with "ExpT" S3 class containing: mu (mean vector), Sigma (covariance matrix), lb (lower bound vector), ub (upper bound vector).
nsim	an integer corresponding to the number of simulations.
control	extra parameters required for the MC/MCMC sampler.
...	further arguments passed to or from other methods.

Value

A matrix with the simulated samples. Samples are indexed by columns.

Author(s)

A. F. Lopez-Lopera.

References

Botev, Z. I. (2017), "The normal law under linear restrictions: simulation and estimation via minimax tilting". Journal of the Royal Statistical Society: Series B (Statistical Methodology), 79(1):125-148. [link]

See Also

tmvrnorm.RSM, tmvrnorm.HMC

Examples

n <- 100
x <- seq(0, 1, length = n)
Sigma <- kernCompute(x1 = x, type = "gaussian", par = c(1,0.2))
tmgPar <- list(mu = rep(0,n), Sigma = Sigma + 1e-9*diag(n), lb = rep(-1,n), ub = rep(1,n))
class(tmgPar) <- "ExpT"
y <- tmvrnorm(tmgPar, nsim = 10)
matplot(x, y, type = 'l', ylim = c(-1,1),
        main = "Constrained samples using expontial tilting")
abline(h = c(-1,1), lty = 2)

lineqGPR documentation built on Jan. 11, 2020, 9:23 a.m.