stan-dev/rstan

What is RStan?

RStan is the official R package that lets you use Stan, a powerful probabilistic programming language, from within the R statistical environment. Stan performs Bayesian inference via modern Markov‑chain Monte Carlo (MCMC) and variational algorithms, so RStan is essentially a bridge that lets R users write Stan models, compile them, and run inference without leaving R.

Key capabilities

Capability	What it means for you
Write Stan models in R	Define a model in Stan’s own language (a separate `.stan` file) and call it from R with `stan()` or `sampling()`.
Fast inference engines	Uses Stan’s C++‑based Hamiltonian Monte Carlo (NUTS) and variational inference under the hood, giving state‑of‑the‑art Bayesian posterior sampling.
Automatic differentiation	Stan’s autodiff computes gradients of the log‑posterior efficiently, which is essential for HMC.
Diagnostics & post‑processing	Returns an `stanfit` object that integrates with R’s `coda`, `bayesplot`, and other packages for convergence checks, trace plots, and posterior summaries.
Cross‑platform	Works on Windows, macOS, and Linux; binaries are available on CRAN for easy installation.
Extensible	You can add custom C++ functions to a Stan model and expose them to R.

How to get started

Install – install.packages("rstan") (or follow the detailed RStan Getting Started wiki for system‑specific steps).
Write a model – create a file model.stan containing Stan code (e.g., a simple linear regression).

Fit the model – in R:

library(rstan)
fit <- stan(file = "model.stan", data = my_data, iter = 2000, chains = 4)
print(fit)
plot(fit)

Explore results – use summary(fit), extract(fit), or the bayesplot package for visual diagnostics.

The repository’s README points to a wiki page with step‑by‑step installation instructions and translations for French, Japanese, Chinese, and Portuguese.

Where to find more information

Online docs & vignettes – https://mc-stan.org/rstan/
Stan language reference – https://mc-stan.org/users/documentation/
Community help – Stan Discourse forum (https://discourse.mc-stan.org)
Bug reports / feature requests – GitHub Issues (https://github.com/stan-dev/rstan/issues)

Licensing & contribution policy

The R package itself is released under GPL‑3.
The bundled Stan code is under the new BSD license.
Contributions must follow the Stan AI Contribution Policy, ensuring responsible use of AI‑related tooling.

Bottom line: RStan lets R users tap into Stan’s cutting‑edge Bayesian inference engine directly from their familiar R workflow, making sophisticated probabilistic modeling accessible without leaving the R ecosystem.