Neutrinos are also incredibly small and light. They have some mass, but not much and they rarely interact with other matter. They come in three types or "flavors": electron, muon, and tau.

Cosmic rays travel through space then crash into the earth's atmosphere and produce  air showers that Include neutrinos and many other types of particles. When neutrinos are produced and start traveling, they can change from one flavor to another. The atmospheric neutrinos are then detected by DeepCore, a denser array of sensors in the center of the IceCube detector at the South Pole.This process is called neutrino oscillation and the IceCube Detector, a massive neutrino detector buried deep in the ice at the South Pole, has a special area called DeepCore that can detect lower-energy neutrinos.

Scientists at the IceCube Neutrino Observatory in Antarctica have made a breakthrough in measuring neutrinos. Using advanced computer techniques, they've achieved the most precise measurements to date of how these particles change as they travel through space, helping us understand fundamental properties of the universe that could lead to new discoveries in physics.

Shiqi Yu

Shiqi Yu, a research assistant professor in the Department of Physics & Astronomy at the University of Utah and others who published their findings recently in Physical Review Letters analyzed data from over 150,000 neutrino events collected over nine years (2012-2021). They used advanced computer programs called convolutional neural networks (CNNs) to process this data. The team made the most precise measurements ever of two important properties related to neutrino oscillation: Delta m²₃₂ and sin²(θ₂₃). These numbers help describe how neutrinos change as they travel.

“We also carefully studied the systematic uncertainties that arise from our imperfect knowledge of our models and chose some to use as free nuisance parameters that fit together with the physics parameters for our data,” says Yu.

Using CNNs, which use three-dimensional data for image classification, Yu and co-lead of the study Jessie Micallef first developed use cases for the CNNs to focus on the DeepCore region and trained them to reconstruct different properties of particle interactions in the detector. They then used the CNN reconstructions to select qualified neutrino interactions that happened in or near the DeepCore region to produce a neutrino-dominated dataset with well-reconstructed energies and zenith angles.

Jessie Micallef

Yu notes that the CNN-reconstructed analysis-level dataset is already being used for other neutrino oscillation analyses, such as determining the neutrino mass ordering and non-standard neutrino interactions and for atmospheric tau neutrino appearance analyses.

“The atmospheric neutrino dataset from DeepCore exhibits relatively high energies in the oscillation analyses, which is unique compared to existing accelerator-based experiments,” says Yu. “Given our dataset and independent analysis, it is interesting to see agreement and consistency in physics parameter measurements.”

This research helps confirm and refine our understanding of how neutrinos — fundamental particles that can tell us a lot about the universe — behave. The techniques developed here, animated by machine learning, can be used in future studies to learn even more about neutrinos and the universe. Those future studies will be informed by IceCube which is planning an upgrade in 2025-2026 that will allow for even more detailed measurements of neutrinos.

By studying neutrino detection and the phenomenon of neutrino oscillation, scientists like Shiqi Yu hope to answer big questions about the nature of matter, energy and the cosmos.