Mind the gap: The discrepancy between simulation and reality drives interpretations of the Galactic Center Excess

The origin of the so-called Galactic Center Excess in GeV gamma rays has been debated for more than 10 years. What makes this excess so interesting is the possibility of interpreting it as additional radiation consistent with that expected from dark matter annihilation. Alternatively, the excess could come from undetected point sources. In this work, we examine the following questions: Since the majority of the previously reported interpretations of this excess are highly dependent on the simulation, how does the model used for the simulation affect the interpretations? Are such uncertainties taken into account? When different models lead to different conclusions, there may be a general gap between these simulations and reality that influences our conclusions. To investigate these questions, we build an ultra-fast and powerful inference pipeline based on convolutional deep ensemble networks and test the interpretations with a wide range of different models to simulate the excess. We find that our conclusions (dark matter or not) strongly depend on the type of simulation and that this is not revealed by systematic uncertainties. Furthermore, we measure whether reality lies in the simulation parameter space and conclude that there is a gap to reality in all simulated models. Our approach offers a means to assess the severity of the reality gap in future works. Our work questions the validity of conclusions (dark matter) drawn about the GCE in other works: Has the reality gap been closed and at the same time is the model correct?