Generative models applied to High Energy Physics problems

A source code for my Bachelors thesis at Higher School of Economics. A corresponding thesis text can be found at resources folder.

Generation Examples

Summary

• Getting Started
• Installing
• Config
• Data Preparation
• Train
• Coding style tests
• Authors
• License
• Acknowledgments

Getting Started

To install the package go to installing notes. All requirements can be found at setup.py

Installing

To install the package run the following command

pip install -e .

Config

Before running the code adapt the config file for your setup. Initially it supposes that you have raw root files at ROOT_FILES_DIR directory.

Additionally you can tune processing parameters, that include signal dimension. There are separate config files for training each model, SIGNALS_TRAINING_CONFIG can be taken for a reference.

Data Preparation

First we need to extract raw data

cd tools/data_preparation
python extract_raw_data.py

Second we prepare dataframes for different events

cd tools/data_preparation
python prepare_dfs.py

Third we prepare signals data

cd tools/data_preparation
python prepare_signals.py

Lastly we postprocess our signals

cd tools/data_preparation
python postprocess_signal.py

Train

To run training modify config file accordingly and run

cd tools/train
python train.py -m MODEL_NAME

Coding style tests

YAPF is used for coding style validation

python3 -m yapf -rp . --in-place

License

This project is licensed under the MIT License