LoLX_ML is a machine learning project designed for Light Only Liquid Xenon (LoLX) experiments. It provides tools for position reconstruction and charge prediction using neural networks.

LoLX_ML
├── ml
│   ├── Beta_data_processor.py       # Processes beta simulation data for ML
│   ├── ChannelMap.py                # Maps SiPM IDs to channels and positions
│   ├── ml_Prediction.py             # Main inference class for model predictions
│   ├── ml_trainer.py                # Model training script
│   ├── ml_trainer_kl.py             # Model training script with KL divergence
│   ├── photonBomb_data_processor.py # Processes photon bomb data
│   ├── SiPMid_vs_chans.csv          # SiPM ID to channel mapping
├── utilities
│   ├── posRecon.py                  # Position reconstruction utilities
│   ├── posRecon_ml.py               # ML-based position reconstruction
├── test                             # Directory for test scripts
├── training_data                    # Directory for training data
│   ├── scalers                      # Directory for data scalers
│   ├── trained_model.keras          # Trained forward model
│   ├── inverse_model.keras          # Trained inverse model
├── yaml
│   ├── config.yaml                  # Configuration for data processing
│   ├── configML.yaml                # Configuration for postion to charge distribution
│   ├── configML_inverse.yaml        # Configuration for charge distribution to position
│   ├── configPred.yaml              # Configuration for model inference
│   ├── configInverseValidate.yaml   # Configuration for inverse model validation
├── README.md                        # Project documentation

To set up the project, clone the repository and install the required dependencies:

git clone <repository-url>
cd LoLX_ML

python ml/Beta_data_processor.py --config yaml/config.yaml
python ml/ml_trainer.py --config yaml/configML.yaml
python ml/ml_Prediction.py --config yaml/configPred.yaml

• Use ml/Beta_data_processor.py or ml/photonBomb_data_processor.py to process raw data into a format suitable for training or inference.
• Configuration files for data processing are located in the yaml/ directory.

• Train forward models using ml/ml_trainer.py or ml/ml_trainer_kl.py.
• Train inverse models using ml/ml_trainer.py with yaml/configML_inverse.yaml.
• Update YAML configuration files with appropriate paths and parameters for training.

• Run predictions using ml/ml_Prediction.py.
• Ensure the appropriate config YAML is properly configured with the paths to the model and data.

• Use utilities/posRecon.py and utilities/posRecon_ml.py for position reconstruction.
• Jupyter notebooks demonstrate position reconstruction workflow.
• The MLPositionReconstructor class handles loading models and scalers for fast predictions.

• Visualization scripts are integrated into the data processors and inference scripts to generate plots for analysis.

All configurations are stored in the yaml/ directory:

• config.yaml: For processing raw simulation data from Chroma.
• configML.yaml: For forward model training (position → light pattern).
• configML_inverse.yaml: For inverse model training (light pattern → position).
• configPred.yaml: For validating the forward model.
• configInverseValidate.yaml: For validating the inverse model.