Project Description
DUNE, the Deep Underground Neutrino Experiment, is an international, multi-institutional collaboration with major facilities and participating institutions across multiple U.S. states and countries. It seeks to better understand the properties of neutrinos, helping scientists answer fundamental questions about the nature of matter and the evolution of the universe. One of the key parts of the experiment is simulating the DUNE detectors before they are built and collecting real-world data.
This project provides participants with hands-on experience using a real GPU-based simulation workflow built around DUNE’s larnd-sim application in an HPC environment. This includes setting up the environment, running jobs, profiling, parameter scans, validation, and reproducibility tracking. Through this process, participants gain practical experience in reproducible workflow design, GPU job execution, output validation, and performance analysis in a real scientific computing environment.
No particle physics background is required.
Learning Objectives
- Set up and run a GPU-based physics simulation on an HPC system.
- Validate and interpret simulation outputs using quality-assurance metrics and plots.
- Design and execute controlled parameter sweeps to explore simulation variation.
- Track provenance and reproducibility across multiple simulation runs.
- Profile GPU workflow performance and measure the impact of code changes.
- Explain and apply key HPC concepts, including job scheduling, GPU computing, and batch workflows, by implementing them in a real scientific simulation pipeline.
- Collaborate effectively in a team-based research environment by contributing to shared code repositories, participating in peer reviews, and communicating findings clearly during project meetings.
