Solving combinatorial problems at particle colliders using machine learning

Physical Review D Publication: https://journals.aps.org/prd/pdf/10.1103/PhysRevD.106.016001

Publication Date: July 5, 2022

Github code: https://github.com/badeaa3/cannonball-rpv-stops

Abstract: High-multiplicity signatures at particle colliders can arise in Standard Model processes and beyond. With such signatures, difficulties often arise from the large dimensionality of the kinematic space. For final states containing a single type of particle signature, this results in a combinatorial problem that hides underlying kinematic information. We explore using a neural network that includes a Lorentz Layer to extract high-dimensional correlations. We use the case of squark decays in R-Parity-violating Supersym- metry as a benchmark, comparing the performance to that of classical methods. With this approach, we demonstrate significant improvement over traditional methods.

Zip Data Folder Structure:

- MG5_aMC_v2.7.3.tar.gz
- pp_45j: generation of $ p p > j j j j (j) $

gridpacks: gridpack file and cross section file

gridpackruns: root files from above gridpack

- pp_t1t1j_msu3_scan: generation of $ p p > ~t1 ~t1 (j) $ with scanning over the stop mass 

gridpacks: gridpacks from generation of 300 GeV - 2 TeV stops at parton level with MadGraph5. txt file contains cross sections and run number explaining which file corresponding to a particular mass point.

gridpackruns: root files from above gridpack generations

mg5_pythia_delphes: single production of a 1 TeV stop $ p p > ~t1 ~t1 (j) $ ran through Pythia 8.2 and Delphes 3.5.0

- model_final

msu3_vs_ptsmear_input_shuf0_eval_hnet.root: final results final

ptsmearX.X: weights.npz (final model weights), weights_logger.npz (training logger), run.txt (stdout of training)