Evidence for Unresolved $\gamma$-Ray Point Sources in the Inner Galaxy

We present a new method to characterize unresolved point sources (PSs) generalizing traditional template fits to account for non-Poissonian photon statistics. We apply this method to Fermi Large Area Telescope $\gamma$-ray data to characterize PS populations at high latitudes and in the Inner Galaxy. We find that PSs (resolved and unresolved) account for $\sim 50\%$ of the total extragalactic $\gamma$-ray background in the energy range $\sim 1.9$ to 11.9 GeV. Within 10° of the Galactic Center with $|b|\ge 2°$, we find that $\sim 5\%–10\%$ of the flux can be accounted for by a population of unresolved PSs distributed consistently with the observed $\sim \mathrm{GeV}$ $\gamma$-ray excess in this region. The excess is fully absorbed by such a population, in preference to dark-matter annihilation. The inferred source population is dominated by near-threshold sources, which may be detectable in future searches.

Figure 1

The source-count function for high-latitude point sources derived from applying non-Poissonian template fits to data with third Fermi-LAT catalog (3FGL) sources [42] unmasked (green band) and masked (orange band). The colored bands indicate 68% confidence intervals, which are computed pointwise in $F$ from the posteriors for the source-count-function parameters, while the solid and dashed black lines show the median source-count functions. The black points show the source-count function of the detected point sources in the 3FGL. The vertical error bars indicate 68% confidence intervals; the horizontal bars denote bins in $F$.

Figure 2

(Left) Best-fit source-count functions within 10° of the GC and $|b|\ge 2°$, with the 3FGL sources unmasked. The median and 68% confidence intervals are shown for each of the following PS components: NFW (dashed, orange), thin disk (solid, blue), and isotropic (dotted, green). The number of observed 3FGL sources in each bin is indicated. The normalization for the diffuse emission in the fit is consistent with that at high latitudes, as desired. (Right) Posteriors for the flux fraction within 10° of the GC with $|b|\ge 2°$ arising from the separate PS components, with 3FGL sources unmasked. The inset shows the result of removing the NFW PS template from the fit. Dashed vertical lines indicate the 16th, 50th, and 84th percentiles.

Figure 3

Same as Fig. 2, except with 3FGL sources masked.

Figure 4

Results obtained by applying the NPTF to simulated data. (Left column) The source-count functions for the PS templates in the fit when NFW PSs are included in the simulated data (top) or not (bottom). Note that when NFW PSs are not simulated, a NFW DM component is instead. (Right column) The associated posteriors for the fraction of flux absorbed by the different templates in the fit. The inset plots show the results of analyzing the simulated data without a NFW PS template in the fit. All plots are relative to the region within 10° of the GC with $|b|\ge 2°$ and 3FGL sources unmasked. For the flux-fraction plots, the fractions are computed relative to the total number of counts observed in the real data.