The SkyLLH framework is an open-source Python-based package licensed under the GPLv3 license. It provides a modular framework for implementing custom likelihood functions and executing log-likelihood ratio hypothesis tests. The idea is to provide a class structure tied to the mathematical objects of the likelihood functions, rather than to entire abstract likelihood models.
Python >= 3.11 is required.
The latest skyllh release can be installed from
PyPI repository:
pip install skyllhOptional dependency groups can be installed with extras:
pip install "skyllh[extras]" # iminuit, pyarrow
pip install "skyllh[dev]" # pre-commit, pytest
pip install "skyllh[docs]" # sphinx and doc-build toolsThe current development version can be installed using pip:
pip install git+https://github.com/icecube/skyllh.gitOptionally, a specific reference can be installed by:
pip install git+https://github.com/icecube/skyllh.git@[ref]where [ref] is a commit hash, branch name, or tag.
conda install -c conda-forge skyllhSeveral publications about the SkyLLH software are available:
- IceCube Collaboration, C. Bellenghi, M. Karl, M. Wolf, et al. PoS ICRC2023 (2023) 1061 DOI
- IceCube Collaboration, T. Kontrimas, M. Wolf, et al. PoS ICRC2021 (2022) 1073 DOI
- IceCube Collaboration, M. Wolf, et al. PoS ICRC2019 (2020) 1035 DOI
These guidelines should help new developers of SkyLLH to join the development process easily.
This repository uses pre-commit to run ruff for linting and formatting on every commit.
Install pre-commit and set up the hooks:
pip install pre-commit
pre-commit installThe hooks will now run automatically on git commit. To run them manually against all files:
pre-commit run --all-files-
When implementing a new feature / change, first an issue must be created describing the new feature / change. Then a branch must be created referring to this issue. We recommend the branch name
fix<ISSUE_NUMBER>, where<ISSUE_NUMBER>is the number of the created issue for this feature / change. -
In cases when SkyLLH needs to be updated because of a change in the i3skyllh package (see below), we recommend the branch name
i3skyllh_<ISSUE_NUMBER>, where<ISSUE_NUMBER>is the number of the issue created in the i3skyllh repository. That way the analysis unit tests workflow will be able to find the correct skyllh branch corresponding to the i3skyllh change automatically.
-
Release version numbers follow the format
v<YY>.<MAJOR>.<MINOR>, where<YY>is the current year,<MAJOR>and<MINOR>are the major and minor version numbers of type integer. Example:v23.2.0. -
Release candidates follow the same format as releases, but have the additional suffix
.rc<NUMBER>, where<NUMBER>is an integer starting with 1. Example:v23.2.0.rc1 -
Before creating the release on github, the version number needs to be updated in the Sphinx documentation:
doc/sphinx/conf.py.
The i3skyllh package provides
complementary pre-defined common analyses and datasets for the
IceCube Neutrino Observatory detector in a private
repository.