alea is a flexible statistical inference framework. The Python package is designed for constructing, handling, and fitting statistical models, computing confidence intervals and conducting sensitivity studies. It is primarily developed for the XENONnT dark matter experiment, but can be used for any statistical inference problem.
Alea aims to model the statistical behaviour of an experiment, which again depends on your knowledge of the underlying physics-- this can range from the very simple, such as measuring a gaussian-distributed random variable, to complex likelihoods where each model component is created by physics simulations (GEANT4), fast detector simulations (for example appletree for XENONnT) or a data-driven method.
If you use alea in your research, please consider citing the software published on zenodo.
You can install alea from PyPI using pip but beware that it is listed there as alea-inference! Thus, you need to run
pip install alea-inference
For the latest version, you can install directly from the GitHub repository by cloning the repository and running
cd alea
pip install .
You are now ready to use alea!
The best way to get started is to check out the documentation and have a look at our tutorial notebooks. To explore the notebooks interactively, you can use Binder.
alea is a public package inherited the spirits of previously private XENON likelihood definition and inference construction code binference that based on the blueice repo https://github.com/JelleAalbers/blueice.
Binference was developed for XENON1T WIMP searches by Knut Dundas Morå, and for the first XENONnT results by Robert Hammann, Knut Dundas Morå and Tim Wolf.