mieskolainen / deepefficiency

DeepEfficiency - optimal efficiency inversion in higher dimensions at the LHC [arXiv:1809.06101]

DeepEfficiency

https://arxiv.org/abs/1809.06101

Invited talk at IRN Terascale, Annecy, France, 2019
https://indico.in2p3.fr/event/18701/contributions/72003/

Requirements: Python3 & Tensorflow 1.8+ & ROOT libraries

This is a research level proof-of-principle code. Depending on the physics application, additional algorithms, estimators and regularization techniques may be needed.

Get ascii (.csv) out from ROOT trees

root printascii.c+ -b -q

Train DeepEfficiency networks

train.sh

Obtain efficiency inversion estimates

predict.sh

Plot differential distributions

make && ./deeplot

Reference

If you use this work in your research, please cite the paper:

@article{mieskolainen2018deepefficiency,
    title={DeepEfficiency - optimal efficiency inversion in higher dimensions at the LHC},
    author={Mikael Mieskolainen},
    year={2018},
    journal={arXiv:1809.06101},
    eprint={1809.06101},
    archivePrefix={arXiv},
    primaryClass={physics.data-an}
}