Install beam

Python

beam is not on PyPI yet; it installs from a clone into a virtual environment:

git clone https://github.com/imallona/beam.git
python -m venv .venv
source .venv/bin/activate
pip install ./beam

Python 3.12 or newer.

R

The R package is rbeam, installed from the checkout:

devtools::install_local("beam/r/beam")
rbeam::install_beam_python()

rbeam::install_beam_python() installs the Python beam into the active reticulate environment. To use an existing environment instead, point reticulate at it with reticulate::use_python() or RETICULATE_PYTHON. Do not pass dependencies = TRUE to install_local: PlackettLuce depends on CVXR and clarabel, which need a Rust toolchain.

Heterogeneity models

The R-backed models (mixed-effects, Bradley-Terry trees, Plackett-Luce, variance decomposition, network meta-analysis) need their own R packages, installed either from R:

install_beam_heterogeneity_deps()

or from the conda env, which has them as binaries:

conda env create -f beam/envs/heterogeneity.yml