Galactic Center Excess in a New Light: Disentangling the γ-Ray Sky with Bayesian Graph Convolutional Neural Networks

Florian List; Nicholas L Rodd; Geraint F Lewis; Ishaan Bhat

doi:10.1103/PhysRevLett.125.241102

Galactic Center Excess in a New Light: Disentangling the γ-Ray Sky with Bayesian Graph Convolutional Neural Networks

Phys Rev Lett. 2020 Dec 11;125(24):241102. doi: 10.1103/PhysRevLett.125.241102.

Authors

Florian List¹, Nicholas L Rodd^{2

3}, Geraint F Lewis¹, Ishaan Bhat⁴

Affiliations

¹ Sydney Institute for Astronomy, School of Physics, A28, The University of Sydney, New South Wales 2006, Australia.
² Berkeley Center for Theoretical Physics, University of California, Berkeley, California 94720, USA.
³ Theoretical Physics Group, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA.
⁴ UMC Utrecht, Image Sciences Institute, 3508 GA Utrecht, The Netherlands.

PMID: 33412055
DOI: 10.1103/PhysRevLett.125.241102

Abstract

A fundamental question regarding the Galactic Center excess (GCE) is whether the underlying structure is pointlike or smooth, often framed in terms of a millisecond pulsar or annihilating dark matter (DM) origin for the emission. We show that Bayesian neural networks (NNs) have the potential to resolve this debate. In simulated data, the method is able to predict the flux fractions from inner Galaxy emission components to on average ∼0.5%. When applied to the Fermi photon-count map, the NN identifies a smooth GCE in the data, suggestive of the presence of DM, with the estimates for the background templates being consistent with existing results.