IceCube Neutrino Clustering and Source Modeling

Overview

Abstract

I developed and evaluated a sequence of Bayesian mixture models to classify high-energy neutrino events as astrophysical or atmospheric, combining spatial and energetic information. Models included joint spatial-energetic frameworks, spatial-weighted implementations, and energy-only variants with posterior spatial reweighting. Severe spectral degeneracy limited discrimination, but the workflow demonstrated a coherent Bayesian approach for large-scale event classification.

Insight

Key Takeaway

Energy information alone proved insufficient due to spectral degeneracy; spatial proximity to candidate sources dominated classification. Posterior predictive checks confirm a stable MCMC workflow, while results highlight the need for more flexible models, additional observables, and hierarchical source-level modeling.

Figures

Visuals

Resources

Links