Epscor research fellows: NSF: machine learning for faster simulation of neutrino detectors

The National Science Foundation Office of Integrative Activities awarded the University of South Dakota $257,733 on January 1, 2027, under the Integrative Activities program (CFDA 47.083) to develop machine-learning methods that accelerate computer simulations of neutrino detectors used in particle physics research.

The project, led by an assistant professor at USD in collaboration with Dr. Seon-Hee Seo at Fermi National Accelerator Laboratory, will produce a validated, machine-learning-enabled fast simulation framework for liquid-argon time-projection chamber (LARTPC) detectors, benchmarked against the Deep Underground Neutrino Experiment (DUNE). The work will construct empirical detector-response models that predict efficiencies, particle-identification performance, and reconstructed energy, angle, timing, and position directly from particle information. These models will be incorporated into a standalone fast-simulation package capable of generating analysis-level observables without running computationally expensive full-detector simulations. The project includes validation benchmarks, software testing procedures, and documentation to support reusable software for the high-energy physics community. A graduate student will receive training in artificial intelligence and scientific computing through the fellowship structure.

The period of performance runs through December 31, 2028, with place of performance at Batavia, Illinois (Fermilab's location).

Abstract

This research infrastructure improvement (RII) epscor research fellows project provides a fellowship to an assistant professor and training for a graduate student at the university of south dakota. The work will be conducted in collaboration with DR. Seon-HEE SEO at fermi national accelerator laboratory (fermilab). Through the fellowship, the principal investigator (PI) will develop new machine-learning (ML) methods that enable much faster computer simulations of particle detectors used in neutrino physics. These tools will help researchers evaluate detector performance and analyze large experimental datasets more efficiently while maintaining the accuracy needed for scientific studies. The project combines particle physics, computer science, and artificial intelligence (AI) to improve computational research infrastructure for next-generation neutrino experiments. The fellowship will strengthen research collaborations between the university of south dakota and fermilab, provide advanced training for the PI and students in AI and scientific computing, and produce broadly reusable software that will benefit the high-energy physics community. This project will develop a validated, ML-enabled fast simulation framework for liquid-argon time-projection chamber (lartpc) detectors, using the deep underground neutrino experiment (dune) as the primary benchmark. The project will construct empirical detector-response models that predict efficiencies, particle-identification performance, and reconstructed energy, angle, timing, and position directly from truth-level particle information. These response models will be incorporated into a standalone, fast-simulation package capable of producing analysis-level observables without executing computationally expensive full-detector simulations. The project will establish validation benchmarks, software testing procedures, and documentation to facilitate adoption by the broader research community. The fellowship will strengthen computational particle physics research at the university of south dakota through national laboratory collaboration, graduate student training, and sustainable software infrastructure that can be extended to future liquid-argon detector experiments. This project is supported by the epscor research infrastructure improvement program: epscor research fellows (ERF), which supports early- and mid-career investigators in eligible jurisdictions to develop collaborations at the nation?s private, government, or academic research institutions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.