This repository contains the code presented in the paper Anomaly Detection in Collider Physics via Factorized Observables. The data and results are stored in Dropbox due to their size.
This directory contains notebooks and helper functions for the analyses presented in the paper. Most notably, FORCE.ipynb contains an example for data loading, model creation, and model application, as well as a signal sensitivity sweep. For the processing.py and observables.py scripts, the events_anomalydetection_v2.h5 can be pulled from the LHCO R&D Zenodo Link.
This directory contains all the code for the toy gaussian analysis presented in the paper.
This directory contains all the figures in the paper. The code that generates these figures are in plots.ipynb, mutual_info.ipynb, and gauss_test.ipynb, while the data to generate these figures are on the Dropbox.