Heavy Flavor Jet Tagging with Machine Learning at sPHENIX in $p+p$ Collisions

Heavy flavor jets produced in high-energy particle collisions are a unique probe for testing perturbative quantum chromodynamics and represent one of the major scientific priorities of the new sPHENIX experiment. Measurements of charm and bottom jets rely on advanced vertex and tracking detectors, as well as high-resolution calorimeters, to distinguish their rare signals from substantial backgrounds. The sPHENIX experiment, which consists of high-precision tracking and calorimeter detector subsystems, has collected $13.3~pb^{-1}$ of triggered data within the acceptance of the full detector at a 1.5 mrad crossing angle in 200 GeV $p+p$ collisions from Run 2024. Full heavy flavor jet measurements at mid-rapidity will be performed for the first time at the Relativistic Heavy Ion Collider using the sPHENIX hadronic calorimeters. We will present heavy flavor jet tagging studies using both traditional selection methods and novel Neutral Network Machine Learning (ML) algorithms in sPHENIX 200 GeV $p+p$ simulation. The ML approach is expected to significantly enhance the jet identification performance, particularly for bottom jets. Initial simulation studies indicate that ML-based method can improve the bottom jet tagging efficiency by approximately a factor of two at the same jet purity. We will discuss the ongoing development of jet tagging methods in simulation, as well as current efforts in Run 2024 data quality assurance and calibration related to heavy flavor jet measurements.