We present a new strategy using artificial intelligence (AI) to build the first AI-based Monte Carlo event generator (MCEG) capable of faithfully generating final state particle phase space in lepton-hadron scattering. We show a blueprint for integrating machine learning strategies with calibrated detector simulations to build a vertex-level, AI-based MCEG, free of theoretical assumptions about femtometer scale physics. As the first steps towards this goal, we present a case study for inclusive electron-proton scattering using synthetic data from the PYTHIA MCEG for testing and validation purposes. Our quantitative results validate our proof of concept and demonstrate the predictive power of the trained models. The work suggests new venues for data preservation to enable future QCD studies of hadrons structure, and the developed technology can boost the science output of physics programs at facilities such as Jefferson Lab and the future Electron-Ion Collider.
Download