Properties of the simulated agn activity function in isolated galaxy mergers

Abstract

Numerical modeling of active galactic nuclei (AGN) poses many challenges, from uncertainties about the underlying physics to dynamic range issues. We present a study of simulated activity functions (AF; the differential of the amount of time spent by a black hole above a given Eddington ratio) in simulations of mergers of Milky-Way like galaxy models using seven different BH feedback algorithms, accretion algorithms, and initial conditions. When considered over the entire simulation the simulated AFs are more dominant at high Eddington ratios than observationally (Schechter-type) inferred AFs. However, during passive evolutionary stages there is considerably closer agreement with observational results. We also demonstrate that two separate algorithms produce AFs which are approximately mass invariant, in agreement with observations. Lastly, we show that numerical uncertainties in AFs and other properties of the remnant such as black hole mass, star formation rates, and accretion rates, are well below a factor of two.