Locating {\gamma}-Ray Sources on the Celestial Sphere via Modal Clustering

TL;DR

This paper introduces a non-parametric clustering algorithm using von Mises-Fisher kernels and mean-shift to identify gamma-ray sources on the celestial sphere, addressing projection issues and enabling automatic detection and significance assessment.

Contribution

It presents a novel directional clustering method tailored for gamma-ray source detection, combining adaptive smoothing, hypothesis testing, and classification techniques.

Findings

Successfully calibrated on simulated Fermi LAT data

Demonstrated on real Fermi LAT data with promising results

Provides an automatic, projection-free source detection approach

Abstract

Searching for as yet undetected gamma-ray sources is a major target of the Fermi LAT Collaboration. We present an algorithm capable of identifying such type of sources by non-parametrically clustering the directions of arrival of the high-energy photons detected by the telescope onboard the Fermi spacecraft. n particular, the sources will be identified using a von Mises-Fisher kernel estimate of the photon count density on the unit sphere via an adjustment of the mean-shift algorithm to account for the directional nature of data. This choice entails a number of desirable benefits. It allows us to by-pass the difficulties inherent on the borders of any projection of the photon directions onto a 2-dimensional plane, while guaranteeing high flexibility. The smoothing parameter will be chosen adaptively, by combining scientific input with optimal selection guidelines, as known from the literature. Using statistical tools from hypothesis testing and classification, we furthermore present an automatic way to skim off sound candidate sources from the gamma-ray emitting diffuse background and to quantify their significance. The algorithm was calibrated on simulated data provided by the Fermi LAT Collaboration and will be illustrated on a real Fermi LAT case-study.