ANNDensity

ANNs for density estimation in flavour physics amplitude analyses. Code used for feasibility studies reported in https://arxiv.org/abs/1902.01452

Prerequisites

The code needs TensorFlow environment to run, together with two libraries, AmpliTF (https://github.com/apoluekt/AmpliTF) and TFA2 (https://github.com/apoluekt/TFA2). The latter includes instructions to set up the full environment in conda: https://github.com/apoluekt/TFA2/blob/master/doc/01_installation.md

Content of the package

The two subdirectories contain examples of density estimation for the efficiency shape and background over the square Dalitz plot phase space of the Ds->K+pi+pi- decay.

• Ds2KpipiEfficiency: Estimation of efficiency (Sections 6 and 8.2 of arXiv:1902.01452)
• Ds2KpipiBackground: Estimation of background (Sections 7.2 and 8.3 of arXiv:1902.01452)

Running the code

Each subdirectory contains the file runme.sh which can be executed to perform all the steps (generation ot toy MC samples, ANN training, fitting etc.). It's best to run them on a machine with good GPU. The whole cycle takes a few hours on a machine with Tesla p100 card.