Determining the directional source of neutrinos

Neutrinos interact very weakly with matter, sometimes traversing cosmic scales before being detected. This makes neutrinos difficult to study, and the data obtained is often noisy, lacking an easy way to reconstruct their paths. Yepez et al. developed a technique for determining the reference angle of a dataset, with applications to neutrinos and beyond.

“Directionality is important because it allows us to connect neutrinos back to their sources, such as the sun, supernovae, or geophysical processes within the Earth,” said author Jeffrey G. Yepez.

Crucial underlying physics is encoded in this information, potentially providing the key to probe matter otherwise inaccessible.

The researchers’ technique bins data into a histogram, then compares it to that of a simulation. In this case, the data is neutron capture locations relative to the beginning of the inverse beta decay process — the reaction that creates a neutron from an antineutrino scattering off a proton. They slowly rotate the measured data, and with each rotation, calculate the difference between the measurement and the reference, called the Frobenius norm of the difference (FND). The rotation angle with the smallest FND is thus the reconstructed reference angle.

The FND and its idealized counterpart, the continuous FND, provide a fast tool for extracting directional information, especially when other techniques are unavailable. This can help complement other directionality techniques or stand on its own for preliminary data analysis or detector diagnosis and design.

Though the scientists were motivated by their own work studying neutrinos, their method is more general and can be used on any type of directional histogram data.

“These techniques may be applied to many other fields where directional or other geometric features appear in two-dimensional data, such as imaging, signal processing, or pattern recognition,” said Yepez. “The broad applicability of our method is one of the most rewarding outcomes of this work.”

Source: “Algorithm to extract direction in 2D discrete distributions and a continuous Frobenius norm,” by Jeffrey G. Yepez, Jackson D. Seligman, Max A. A. Dornfest, Brian C. Crow, John G. Learned, and Viacheslav A. Li, AIP Advances (2026). The article can be accessed at https://doi.org/10.1063/5.0315079 .