OMNNSimulator

This project uses machine learning to simulate the response of multi-PMT optical modules. The Neural Network models are trained on Geant4 simulations which are very accurate but slow.

Information about scripts in OMClassifier:

torch_trazing_model.py :

This script converts a model into a format that can be used and deployed in C++.

OMSimC++Test :

Test how to use and deploy a model in C++.

convert_hdf5.py :

Script that converts Geant4 output .txt files into hdf5 format suitable for training.

main.py :

Script that can be used for the training of Neural Network models.

eff_area.py

Script that can be used to calculate optical module's effective areas with the Neural Network models. It is usually an useful check.

src/omsimulator/data/dataloader.py

Dataloader used for training.

src/omsimulator/model/model.py

Neural Network model.

src/omsimulator/model/model_torch.py

Neural Network model used by torch_trazing_model. It does not inherit from torch_ligthning.