Dortmund 2021 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 71: Data analysis, Information technology III

T 71.1: Talk

Wednesday, March 17, 2021, 16:00–16:15, Tu

Usage of neural networks in photon identification in ATLAS — •Florian Kirfel and Oleh Kivernyk — Physikalisches Institut der Universität Bonn

Optimal photon identification in ATLAS analyses plays an important role in precise measurements of Higgs boson properties and in the search for new particles.

Currently photons are selected using a set of cuts on calorimeter variables which describe the shape of electromagnetic showers. These cuts were optimized using Monte Carlo simulations of photons and jets. Due to the simulations not being ideal, the selection efficiencies must be corrected to match data. The efficiency measurement in data is not simple and requires assumptions about some shower shape variables, i.e. that they are independent of the isolation of the photon candidate.

Artificial neural networks are employed to improve the current photon identification. Decorrelation of the neural network output from the isolation variable results in an improvement of the efficiency measurement in data.