This docker image provides everything you need for Bayesian data analysis in R. It comes with RStudio Server and Quarto on Ubuntu. Binary packages are accessed via bspm using install.packages().

Note, on a Windows host, you may want to increase the number of CPUs available to WSL via %UserProfile%\.wslconfig

• brms (newest on CRAN)
• rstanarm (newest on CRAN)

• cmdstanr (newest on https://stan-dev.r-universe.dev/builds)
• rstan (newest on CRAN)

• data.table (newest on CRAN)
• tidyverse (newest on CRAN)

• bayesplot (newest on CRAN)
• easystats (newest on CRAN)
• ggeffects (newest on CRAN)
• shinystan (newest on CRAN)
• tidybayes (newest on CRAN)

Some users will want to just install Docker Desktop, pull the image, and run it.

docker pull jmgirard/rocker-bayes
docker run -e PASSWORD=pass -p 8787:8787 jmgirard/rocker-bayes

Then navigate to http://localhost:8787 in your web browser and enter "rstudio" and "pass".
Use volumes or bind mounts to grant the container access to persistent storage or host directories.

You could also download the Dockerfile from GitHub and build it yourself.

git clone https://github.com/jmgirard/rocker-bayes.git
cd rocker-bayes
docker-compose up --build -d

Then navigate to http://localhost:8787 in your web browser and enter "rstudio" and "pass".
You can also customize the port and password by editing .env in a text editor.

Note that this small model won't get much benefit from within-chain parallelization.
It's just used to quickly test that everything is working.

library(brms)
fit_serial <- brm(
  count ~ zAge + zBase * Trt + (1|patient),
  data = epilepsy, family = poisson(),
  chains = 4, cores = 4, backend = "cmdstanr"
)
fit_parallel <- update(
  fit_serial, chains = 2, cores = 2,
  backend = "cmdstanr", threads = threading(2)
)