MCMSki 2014

mcmski-logI am off to the MCMSki 2014 meeting in Chamonix. I am talking twice. The first talk is in  a session on Computational and Methodological Challenges in evidence synthesis and multi-step (modular models).  The title is a bit of a mouthful, but I will be talking about cuts. There is a demand for a cut function in JAGS but it is incredibly hard to implement it properly.  In fact the cut function, as currently implemented in OpenBUGS does not work correctly. I have been saying this for years, without making any real progress or providing a  solution in JAGS. I agreed to give the talk in order to explain myself to an expert audience.

The second talk is in a round-table discussion on Recent Developments in Software for MCMC with other software developers Andrew Thomas (BUGS), Bob Carpenter (Stan) and Adrien Todeschini (BiiPS). BiiPS is a fork of JAGS that uses the same front end, but sequential Monte Carlo as a back end instead of MCMC.

I will report back after the meeting. In the mean time, you should be able to find a conference diary by Christian Robert at Xi’an’s Og.

Stan

Andrew Gelman has announced the release of Stan version 1.0.0 and its R interface RStan.  Stan – named after Stanislaw Ulam, the inventor of the Monte Carlo method – is a new MCMC program that represents a major technological leap forward. It works flawlessly on my Linux desktop and is very, very fast. (Note that I have nothing to do with the Stan project: I am just posting this here because it is clearly of interest to JAGS users). Continue reading