This is the package developed for DNN-based energy reconstruction and signal/background separation in the nEXO experiment. The package is built on pytorch platform. It takes the nEXO charge simulation as input (possibly adding photon information in the future), and perform particle identification (plan to add reconstruction in the future) based on convolutional neural networks.

1. config - Configurations of deep learning algorithms.
2. utils - data loader and train/validation scripts.
3. networks - neural network architetures.
4. scripts.

• pytorch(1.3.1+), pandas(0.23.4+), scipy(1.2.1+), numpy(1.16.6+), uproot(3.12) ,h5py(3.7.0)
• SparseConvNet
• ROOT6
• dataset - The nEXO simulation and simulation result is not open.