tregress function

Truncated student generator for Bayesian regression simulation

Simulates n random vectors $X$ exactly distributed from the d-dimensional Student distribution with df=$\nu$ degrees of freedom, mean zero and scale matrix sigma, conditional on $l<X<u$,

tregress(n, lb, ub, sigma, df)

Arguments

• n: number of observations
• lb: vector of lower truncation limits
• ub: vector of upper truncation limits
• sigma: scale matrix
• df: degrees of freedom

Returns

list with components

• R: n vector of scale
• Z: a d by n matrix

so that $\sqrt(\nu)Z/R$ follows a truncated Student distribution

Examples

d <- 5
tregress(lb =rep(-2, d), ub = rep(2, d), df = 3, n = 10,
  sigma = diag(0.5, d) + matrix(1, d, d))

References

Z. I. Botev and P. L'Ecuyer (2015), Efficient probability estimation and simulation of the truncated multivariate Student-t distribution, Proceedings of the 2015 Winter Simulation Conference, pp. 380-391,

Author(s)

Matlab code by Zdravko Botev, R port by Leo Belzile