SBT Advanced veto

Framework for ML architectures for the SBT veto trained with pytorch.

Models currently include:

• neural network
• message passing GNN

Usage

Ship Veto

The script nnveto_implementation.py provides an example of running the neural network veto within SHIP. The necessary python packages in the Installation section below are first required using pip install --user

The similar example for the gnn will be provided soon.

Inference

Run example script apply_nn.py and apply_gnn.py to apply trained neural network and GNN to nuDIS events.

Training

Currently performed in notebooks GNN_global.ipynb and NN.ipynb however code wiil be factorized in the near future.

Installation:

cpu:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
pip install scikit-learn
pip install dm-tree
pip install torch_geometric
pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.4.0+cpu.html

note for gpu: gpu:

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.4.0+cu124.html

Will add a requirements file later.