Normally and positive distributed pseudorandom numbers

Simulation of positive Gaussian variables.

This function generate random variables distributed according to a truncated normal distribution (or, by a translation, to a normal distribution with positive support). This kind of problem is especially interesting for generating variables with MCMC methods.

We use a mixed accept-reject algorithm, i.e. a classical accept-reject algorithm using several proposal distributions, each one being adapted to the different values of the distribution parameters. Then, with respect to these parameter values, the proposal distribution which gives the highest average probability of acceptance is used to simulate a variable.

The ZIP file contains the function rpnorm and an example.

http://lsiit-miv.u-strasbg.fr/lsiit/perso/mazet/rpnorm-en.htm