Retaining black holes with very large recoil velocities

Recent numerical simulations of binary black hole mergers show the possibility of producing very large recoil velocities (> 3000 km s(-1)). Kicks of this magnitude should be sufficient to eject the final black hole from virtually any galactic potential. This result has been seen as a potential contradiction to observations of supermassive black holes residing in the centers of most galaxies in the local universe. Using an extremely simplified merger tree model, we show that, even in the limit of very large ejection probability, after a small number of merger generations there should still be an appreciable fraction ( > 50%) of galaxies with supermassive black holes today. We go on to argue that the inclusion of more realistic physics ingredients in the merger model should systematically increase this retention fraction, helping to resolve a potential conflict between theory and observation. Finally, we develop a more realistic Monte Carlo model to confirm the qualitative arguments and to estimate occupation fractions as a function of the central galactic velocity dispersion.