Model-independent Measurement of the Atmospheric Muon Neutrino Energy Spectrum up to 2.5 PeV

TL;DR

This paper presents a model-independent method to measure the atmospheric muon neutrino energy spectrum up to 2.5 PeV using IceCube data, employing machine learning and likelihood unfolding techniques.

Contribution

It introduces a novel, model-independent unfolding approach combined with machine learning to estimate the muon neutrino spectrum over a broad energy range.

Findings

First model-independent muon neutrino spectrum measurement up to 2.5 PeV.

Achieved over 99% purity in muon neutrino sample.

Demonstrated the effectiveness of a decision tree-based binning scheme.

Abstract

The IceCube Observatory at the South Pole allows the measurement of the diffuse neutrino flux. The assumption of specific flux parametrizations limits the range of spectral shapes. Given the increasing statistics of the data recorded, model-independent unfolding approaches can overcome these limitations. In this contribution, a model-independent approach to estimate the muon neutrino flux between 125 GeV and 2.5 PeV is presented. In order to extract muon neutrinos from the data taken by the detector, a machine-learning-based method is employed. This yields an efficient muon neutrino sample with a purity of over 99 %. The spectrum is estimated by a Likelihood-based unfolding technique involving a novel binning scheme using a decision tree on three years of IceCube data. This measurement provides the first model-independent muon neutrino spectrum for multiple years in this energy regime.