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

HK: Fachverband Physik der Hadronen und Kerne

HK 24: Heavy-Ion Collisions and QCD Phases V

HK 24.2: Talk

Tuesday, March 29, 2022, 16:30–16:45, HK-H1

J/ψ identification in ALICE with XGBoost — •Lasse Bassermann for the ALICE collaboration — Physikalisches Institut der Universität Heidelberg

In ALICE (A Large Ion Collider Experiment), J/ψ meson production is analyzed at midrapidity via the decay to an electron-positron-pair. Until now this identification was done by hand using a cut-based-method, which assumes that electrons, if they match certain parameters, originate from the decay of a J/ψ meson. Another method to identify J/ψ mesons could be through machine learning algorithms, such as XGBoost. XGBoost is an open-source software that provides machine learning algorithms using a gradient boosting framework, where an ensemble of weak prediction models is used.

In this poster the first steps of implementing an XGBoost algorithm for identifying J/ψ mesons are presented. This includes the first adaptions of the algorithm to the data used as well as the selection of the data. First comparisons with the cut-based method are discussed.