CODA (Convergence Diagnostics and Output Analysis) is a suite of S functions that provide an object-based infrastructure for analysing the output of Markov Chain Monte Carlo (MCMC) simulations. 

Getting hold of the software

CODA for R is available as an R package from the Comprehensive R Archive Network (CRAN). If you are a unix user, you should install the package from source using the instructions in the R FAQ. If you are a Windows user, you should install the prebuilt binary, also available from CRAN.

CODA for S-PLUS is distributed as an S-PLUS package from the Comprehensive S-PLUS Archive Network (CSAN).

Citing CODA

If you use CODA, please cite the paper published in R News. Here is a bibtex entry, including links to the online paper.
AUTHOR = {Martyn Plummer and Nicky Best and Kate Cowles and Karen Vines},
TITLE = {{CODA}: Convergence Diagnosis and Output Analysis for {MCMC}},
JOURNAL = {R News},
YEAR = 2006,
PAGES = {7--11},
MONTH = {March},
URL = {},
PDF = {}

Background Reading

The main emphasis of CODA is on convergence diagnostics.   There are two useful reviews of MCMC convergence diagnostics. The first one is by the original author of CODA.

  1. Cowles, MK and Carlin, BP (1995) Markov Chain Monte Carlo diagnostics: A comparative review, J Amer Stat Soc, 91, 883-904.
  2. Brooks, SP and Roberts, GO (1998) Assessing convergence of Markov Chain Monte Carlo algorithms, Statistics and Computing. 8, 319-335.

This page is maintained by Martyn Plummer (plummer at iarc dot fr).