Year Two: Toward Expansion of the Standard Model Using Multi-Hadron Final States and Muon Neutrino Reconstruction
ISEF · 2026 Physics and Astronomy Second Award
Overview
Neutrino oscillation measurements aim to find fundamental physics beyond the Standard Model, but precision is limited by our inability to distinguish genuine signals from interfering nuclear reactions that produce similar signatures. A previously observed deficit in muon-neutrino events at Fermilab is compatible with a theoretical new particle, but could also arise from poorly understood interactions inside argon nuclei. Current methods treat this ambiguity as a roughly ±30% systematic uncertainty. This two-year study developed an end-to-end analysis framework to address this challenge. Year 1 established machine learning-based event reconstruction and selection in the ICARUS detector, achieving kinematic distributions across 6,282 events. Year 2 added a novel hypergraph-based classification system that maps neutrino interactions through resonance production to final topologies, explicitly modeling Final State Interactions where 40-60% of produced pions are absorbed before escaping the nucleus, counteracting effects that scramble oscillation signatures. By constructing energy-dependent response matrices with smooth interpolation across 0.3-3.0 GeV, this project achieved 94% prediction accuracy, an 82% improvement over standard methods. By investigating hadronic decay patterns, we uncovered a significant revision in previous theoretical sterile neutrino scenarios that would require impossible particle production patterns. This methodology is directly applicable to scientists throughout the field of particle physics and the $3 billion DUNE experiment, providing a systematic path to distinguish true signal from detector systematics in the search for new subatomic particles.
Awards (1)
- Second Award of $2,400 $2,400
Competition history
- ISEF 2026
Resources
Related projects
ISEF · 2023
Determining DUNE Convolutional Neural Network Neutrino Identification Biases
AJAS · 2022
Computers can analyze complex subatomic particle data using neural networks
CSEF · 2018
Achieving Improved Accuracy Model for Jet Energy Measurements in the Large Hadron Collider (LHC) Using Machine Learning
ISEF · 2019
Improving Particle Classification in WIMP Dark Matter Detection Experiments Using Neural Networks
ISEF · 2020
Quantum Machine Learning Frameworks for Improved SiD Calorimetry and Higgs Boson Analysis
CSEF · 2026
Accelerating LHC Exotic Physics Discovery via Optimal Transport Flow Matching Calorimeter Simulation
ISEF · 2020
Measuring and Modeling the Energy Spectrum of Cosmic Ray Muons
AJAS · 2017
Examining a New Method for the Discovery of Hypothetical Exotic Particles
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: Regeneron International Science and Engineering Fair