Charged Particle Tracking via Edge-Classifying Interacting Networks

Repo Organization

• graph_construction/

  • build_heptrkx_classic.py: HEP.TrkX construction algorithm to build graphs in the pixel barrel layers, example usage:

    python build_heptrkx_classic.py configs/heptrkx_classic.py
    

  • build_heptrkx_plus.py: HEP.TrkX construction algorithm modified to build graphs in the pixel barrel+endcaps layers, example usage:

    python build_heptrkx_plus.py configs/heptrkx_plus.py
    

  • build_dbgraphs.py: DBScan clustering algorithm used build graphs via clustering in eta-phi space, example usage:

    python build_dbgraphs.py configs/dbgraphs.py
    

• models/

  • interaction_network.py: network architecture
  • dataset.py: custom data loader for multi-threaded graph loading (necessary for large graphs)
  • track_fitter.py: class to fit hits in conformal space

• run_interaction_network.py: IN training script, example usage:

  python run_interaction_network.py --pt=1 --construction=heptrkx_plus --lr=0.005 --gamma=0.9 --save-model