Speakers
Description
The objective of electron identification (eID) in the sPHENIX experiment is to accurately distinguish electrons from quarkonium (such as J/psi and Upsilon) di-electron decay events while effectively suppressing hadron background, thereby enhancing the signal-to-noise ratio crucial for studying Upsilon and J/psi suppression within the Quark-Gluon Plasma. We compare traditional cut-based methods with machine learning-based Multi-Variable Analysis (MVA) techniques, showing significant improvements in eID accuracy and hadron rejection through Monte Carlo simulations. We have demonstrated through sPHENIX simulation data studies that machine learning algorithms, such as Boosted Decision Trees (BDT) and Deep Neural Networks (DNN), improve hadron rejection, thus enabling more precise electron selection for di-electron reconstruction. Building on these results, the next phase involves applying machine learning-based eID to sPHENIX experimental data, with the goal of selecting suitable electrons for Upsilon and J/psi reconstruction. This approach will enable better isolation of Upsilon and J/psi signals from background events, providing clearer insights into Upsilon and J/psi production and suppression in heavy-ion collisions.
| Category | Experiment |
|---|---|
| Collaboration (if applicable) | sPHENIX |
