Using R, Python, and CmdStan

This workstation supports scientific computing with R, Python, and CmdStan.

Software is mainly managed using Miniforge, Conda, and Mamba.

1 Main idea

The workstation separates software into different levels:

Level	Example	Who manages it?
System software	Slurm, Quarto, LaTeX, system libraries	Administrator
Shared conda environments	`R-4.4`, `py311`	Administrator/RSE team
Project R packages	`renv` project library	Project users
Personal packages	user home directory	Individual user

For R projects, the recommended approach is:

shared conda R version + project-specific renv library

This means everyone may use the same R version from conda, but each serious project can keep its own R package versions using renv.

which conda

If conda is available, this should show a path.

conda env list

or:

mamba env list

Example for R:

conda activate R-4.4

Example for Python:

conda activate py311

The exact environment names depend on what has been installed.

conda deactivate

Activate the R environment:

conda activate R-4.4

Start R:

Run an R script:

Rscript analysis.R

Check the R package library paths:

.libPaths()

For most users:

quick personal tests can use the personal R library
reproducible project work should use renv
shared packages for all users should be installed by the administrator in the shared conda environment

Do not assume that installing an R package in your own account makes it available to other users.

See R Package Management for the full policy.

Activate a Python environment:

conda activate py311

Start Python:

python

Run a Python script:

python analysis.py

Install Python packages using conda or mamba when possible:

mamba install numpy pandas scipy matplotlib

For project-specific work, consider creating a project-specific environment.

CmdStan is useful for users who prefer CmdStan or CmdStanR rather than RStan.

Check whether CmdStan is available:

ls /data/software

If using CmdStanR in R:

library(cmdstanr)
cmdstanr::cmdstan_path()

Set CmdStan path if needed:

cmdstanr::set_cmdstan_path("/data/software/cmdstan")

cd /data/projects/my-stan-project
conda activate R-4.4
R

Inside R:

install.packages("renv")
renv::init()
install.packages("cmdstanr")
cmdstanr::set_cmdstan_path("/data/software/cmdstan")
renv::snapshot()

For future work in the same project:

cd /data/projects/my-stan-project
conda activate R-4.4
R

Then inside R:

renv::restore()

For reproducible projects:

Create a file called environment-notes.txt in your project folder:

Date: 2026-06-18
Workstation: mod-linux-ws01
Environment: R-4.4
R package management: renv
Main packages: tidyverse, cmdstanr, posterior