[R-pkgs] Nmix package - Bayesian inference about univariate Gaussian mixtures with an unknown number of components

[Peter Green]

This package is now on CRAN in version 2.0.2, and anyone interested in 
inference on mixture models, or more generally in Bayesian inference for 
"trans-dimensional" problems, is encouraged to try it out. The 
methodology is that presented by Richardson and Green in our 1997 
article in the Journal of the Royal Statistical Society, using 
Reversible jump Markov chain Monte Carlo. The package is based on the 
Fortran code used in the research for that paper, which is available now 
to the R user for the first time. A rich variety of outputs from the 
MCMC sample path is available, optionally, and many features of the 
posterior distribution are presented graphically. Apart from its direct 
use in analysing data, the package might be useful in supporting 
learning about Bayesian computation.

Feedback is most welcome, and I may be open to extending the 
functionality of the package in response to concrete suggestions.

regards,

Peter Green