We should make it as easy as possible for user to compute common convergence diagnostics (for example estimated effective sample sizes, Gelman-Rubin $\hat{R}$ diagnostic, Monte Carlo standard errors) and summary statistics for sampled chains. Ideally this will just require either having the sampled chains data structure be such that it can directly used with one of the existing R packages for computing such diagnostics and summaries (see some suggestions below), or providing helper functions to allow massaging the output to the required format:
We should make it as easy as possible for user to compute common convergence diagnostics (for example estimated effective sample sizes, Gelman-Rubin $\hat{R}$ diagnostic, Monte Carlo standard errors) and summary statistics for sampled chains. Ideally this will just require either having the sampled chains data structure be such that it can directly used with one of the existing R packages for computing such diagnostics and summaries (see some suggestions below), or providing helper functions to allow massaging the output to the required format:
posterior: tools for working with posterior distributionscoda: output analysis and diagnostics for MCMCmcmcse: Monte Carlo standard errors for MCMC