Conditional Wasserstein Generative Adversarial Networks for Fast Detector Simulation

Detector simulation in high energy physics experiments is a key yet computationally expensive step the event process. There has been much recent interest using deep generative models as faster alternative to full Monte Carlo process situations which utmost accuracy not necessary. In this work we investigate use of conditional Wasserstein Generative Adversarial Networks simulate both hadronization and detector response jets. Our model takes 4-momenta jets formed from partons post-showering pre-hadronization inputs predicts corresponding reconstructed jet. trained on fully simulated tt events publicly available GEANT-based CMS Collaboration. We demonstrate that produces accurate jet transverse momentum ( p T ) distributions over wide range for input parton only fraction time necessary conventional methods, running CPU less than millisecond per event.