This repository contains code to build a bunch of lab-cosmo Python projects each day, and make the pre-built wheels directly installable with pip.
Library to generate representations for atomic-scale learning, https://github.com/lab-cosmo/librascal
# install rascal dependencies first
pip install numpy scipy ase
# finally install librascal itself (this can not be a single step since there is
# another package called rascal on PyPI)
pip install --index-url https://luthaf.fr/nightly-wheels/ rascal
Data storage and manipulation for atomistic machine learning, https://github.com/lab-cosmo/metatensor
# install the full project
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ metatensor
# install the full project, including the torch interface
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ metatensor[torch]
# install sub-packages (not required if you installed the full project)
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ metatensor-core
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ metatensor-torch
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ metatensor-operations
Computing representations for atomistic machine learning, https://github.com/luthaf/rascaline
# install the full project
pip install --extra-index-url https://luthaf.fr/nightly-wheels/ rascaline- copy one of the workflow in
.github/workflowsto<new-project.yml>and edit it to build the wheels in thebuild-wheelsjob - open a PR with these changes, make sure CI passes and merge it
- add the new project folder to the
index.htmlfile on the gh-pages branch