Fermi Gamma-ray Space Telescope

Inference of Dim Gamma-Ray Point Sources Using Probabilistic Catalogues

S. K. N. Portillo
T. Daylan, D. P. Finkbeiner

Abstract:

Poisson regression of the Fermi-LAT data in the inner Milky Way reveals an extended gamma-ray excess. The anomalous emission falls steeply away from the galactic center and has an energy spectrum that peaks at 1-2 GeV. An important question is whether the signal is coming from a collection of unresolved point sources, possibly recycled pulsars, or constitutes a truly diffuse emission component. Previous analyses have relied on non-Poissonian template fits or wavelet decomposition of the Fermi-LAT data, which find evidence for a population of dim point sources just below the 3FGL flux limit. In order to be able to make conclusions about such a dim population we propose to sample from the catalogue space of point sources in the inner galaxy, where the model dimensionality, i.e., the number of sources, is unknown. Although being a computationally expensive sampling problem, this approach allows us to infer the number, luminosity and radial distribution function of the point source population that is most consistent with the data while providing a Bayesian evidence for the point source hypothesis that is independent of the model indicator. This talk will focus on the method of trans-dimensional sampling using the reversible-jump formalism and its application to the inference of a mock point source population. The talk by Stephen K. N. Portillo will further present the inferred catalogue using the high latitude Fermi-LAT data.