Speaker
Description
Detector simulations serve a critical role in Nuclear and Particle Physics for design and analysis, but these expensive simulation programs also act as computational bottlenecks. This work leverages the PARNASSUS framework to provide an end-to-end generative AI based fast simulation for the CLAS12 detector and reconstruction pipeline at Jefferson Lab. By using Continuous Normalizing Flows with Conditional Flow Matching, this model replicates the full simulation with high fidelity and with inference speeds orders of magnitude faster. This approach naturally handles complex detector responses that depend on the full initial and final state by mapping the generator-level events directly to reconstruction-level events in a set-to-set generation approach. As this work demonstrates, the PARNASSUS framework provides a flexible and accurate tool for fast simulations that aims to alleviate the current computation bottleneck and enable new simulation based techniques for analysis and design.