Proton-proton collisions and single pions simulated with COCOA

This record contains samples of simulated proton-proton collisions and single charged pions used in [1] to investigate particle flow algorithms based on deep learning. The collisions were generated and showered with Pythia8 at √s = 13 TeV producing three different processes: dijet, ttbar, and Z→Z(nunu)H(bb). The six pion samples contain single π+ with transverse momenta of 2, 5, 10, 20, 40, and 80 GeV. All samples were simulated using GEANT4 in the nearly-hermetic 6-layer calorimeter described in [2]. Information about the branches in the TTree format can be found at [3]. Training and validation were performed with dijet_train.root and dijet_val.root, respectively, while all other files were used for performance testing. [1] HGPflow: Extending Hypergraph Particle Flow to Collider Event Reconstruction[2] Configurable Calorimeter Simulation for AI (COCOA)[3] Output description