A collection of functions to deal with the truncated univariate and multivariate normal and Student distributions, described in Botev (2017) and Botev and L'Ecuyer (2015).
Main features are
- simulation from multivariate truncated Normal and student distributions using an accept-reject algorithm based on minimax exponential tilting.
- (quasi) Monte-Carlo estimation of the distribution function using separation-of-variables together with exponential tilting for provable performances and theoretical upper bound on the error.
- Cholesky decomposition using the reordering algorithm of Gibson, Glasbey and Elston (1994).
To install the latest version from CRAN, use
install.packages("TruncatedNormal")or else install the latest development from Github via
devtools::install_github("lbelzile/TruncatedNormal")