Installation

Installation#

The MBIRJAX package is available through PyPI or from source on GitHub.

Install from PyPI

In order to install from PyPI, use one of the following commands.

For CPU only:

pip install --upgrade mbirjax

For CPU with a CUDA12-enabled GPU:

pip install --upgrade mbirjax[cuda12]

Installing from source

  1. Download the source code

In order to download the python code, move to a directory of your choice and run the following two commands:

git clone https://github.com/cabouman/mbirjax.git
cd mbirjax

  1. Install the conda environment and package

Clean install using mbirjax/dev_scripts - We provide bash scripts that will do a clean install of MBIRJAX in a new conda environment using the following commands:

cd dev_scripts
source clean_install_all.sh

Optional Pixi development environment

For contributors who use Pixi, MBIRJAX also provides an optional reproducible development environment. This does not replace the conda installation workflow above.

For the default CPU environment on Linux or Apple Silicon macOS:

pixi run smoke
pixi run test-fast

For a CUDA-enabled Linux system:

pixi run -e cuda smoke-jax
pixi run -e cuda test-fast

Additional useful tasks include:

pixi run test
pixi run test-data
pixi run docs

Installing on Windows

In order to use MBIRJAX on a windows computer with an NVIDIA GPU, we recommend that you use Windows Subsystem for Linux (WSL). Download this pdf file for details on how to install MBIRJAX with WSL.