Machine Learning Application for Λ Hyperon Reconstruction in CBM at FAIR

• Ilya Selyuzhenkov (GSI / MEPhI)
• Viktor Klochkov (Eberhard Karls Universität Tübingen)
• Shahid Khan
• Andrea Dubla
• Oleksii Lubynets

The Compressed Baryonic Matter (CBM) experiment at FAIR will investigate the QCD phase diagram in the region of high net-baryon densities in the collision energy range of $\sqrt{s_{\mathrm{NN}}}$ = 2.7 - 4.9 GeV with high interaction rate, up to 10 MHz, provided by the SIS100 accelerator. Enhanced production of strange baryons can signal transition to a new phase of the QCD matter. Λ hyperons are the most abundantly produced strange baryons. They weakly decay, with a branching ratio of 64%, into a proton and a negatively charged pion ($\pi^{-}$). To reconstruct the $\Lambda \rightarrow \pi^{-} + p$ decay kinematics, Particle-Finder Simple (PFSimple) package is used. PFSimple interfaces the mathematics of the Kalman Filter Particle (KFParticle) package and provides a convenient interface to control the reconstruction parameters. For the reduction of combinatorial background specific selection criteria needs to be applied to the proton and π- tracks and Λ-candidates decay topology. The performance for Λ hyperon reconstruction in CBM with Machine Learning (ML) algorithms such as XGBoost will be presented.

Discussion