Di-Muon cocktail reconstruction using Machine Learning technique in CBM experiment at FAIR

The CBM experiment at FAIR will investigate strongly interacting matter at high baryon density and moderate temperature. One of proposed key observables is the measurement of the low mass vector mesons (LMVMs), which can be detected via their di-lepton decay channel. Di-leptons are clean probes for hot and dense matter, since they only interact electromagnetically, they escape the medium nearly unperturbed, thus allowing unique access to the properties of the medium. The Muon Chamber (MuCh) detector system is being built to identify the muon pairs originating from the direct and Dalitz decay of low mass vector mesons, in a background mostly populated by muons from weak decay of pions and kaons produced in the collisions. In the future, CBM experiment at FAIR, will add to existing experimental data with new results from the intermediate energy range, probing the di-lepton emission from the high net baryon density region. We report, simulation results for the reconstruction of di-muon spectra for 8AGeV , where A is the mass number of the nucleus being accelerated, central AuAu collisions using machine learning (ML) techniques for selection of muon track candidates. Various ML algorithms like Gradient boosted decision trees (BDTG), K-Nearest neighbour (KNN), Multi-layer Preceptron (MLP), HMatrix etc. from the TMVA class have been employed for the present study. The results from different ML models have been compared with the traditional selection manual cuts based method for reconstruction of omega (ω), eta (η), phi (ϕ) mesons and full freeze-out di-muon cocktail spectra and improvements in dimuon performance with ML classifiers are reported. For comparable S/B ratio, the pair reconstruction efficiency and significance is observed to be increased significantly for omega (ω), eta (η), phi (ϕ) mesons using ML techniques.