IMSA SIR Neutrino Event Reconstruction in DUNE NDLAr “2×2” with AI/ML

Internship Description

Neutrinos are tiny, nearly massless particles that rarely interact with matter, making them hard to study. The Deep Underground Neutrino Experiment (DUNE) will use large liquid-argon time projection chambers (LArTPCs) to capture the tracks and energy left by particles created when a neutrino interacts. To make the most precise measurements (including whether neutrinos violate CP symmetry), we need accurate, efficient neutrino event reconstruction performed by identifying charged particle tracks from detector data.

In this project, we will use artificial intelligence and machine learning (AI/ML) to help reconstruct and identify particles in the DUNE Near Detector liquid-argon prototype called NDLAr “2×2” (also known as ProtoDUNEND). We’ll analyze simulated events and early prototype data to measure how well different AI/ML methods can find and identify particles such as muons, electrons/photons (gammas), protons, and pions.

Education and Experience Requirements

What student will learn and do:

• Learn the basics of neutrino physics and how LArTPC detectors work.
• Work in Python and Jupyter notebooks to explore event data and build simple ML models.
• Use data set with help of AI/ML tools to:
• Reconstruct particle “objects” from detector hits.
• Classify particles (e.g., muon vs electron/photon vs proton vs pion).
• Select a specific interaction channel (e.g., CCQE: chargedcurrent quasielastic) for physics studies.
• Create visualizations to explain results and uncertainties.
• Time permitting, compare different models (e.g., a convolutional neural network, CNN, that sees image like views; and a graph neural network, GNN, that operates on point clouds/hits).

Internship Family

Visiting Student High School Research

Internship Category

IMSA Student