Stanisław - very large Stan

Stan is an amazing Bayesian computation framework capable of fitting models with millions of parameters and beyond. This package is (intended to be) a collection of tools to help cmdstanr users on Linux process such massive models fast while conserving RAM.

The key functionality is the stansummary function to obtain posterior summaries of parameter subsets using the stansummary command-line utility bundled with CmdStan. On very large models, this approach is dramatically faster than cmdstanr::summary() and requires less RAM.

Installation

Install directly from github with devtools like so:

devtools::install_github("https://github.com/huffyhenry/Stanislaw")