Obviously, the formation of central black holes in AGN should be closely connected with the issues of galaxy formation and evolution. The onset of AGN activity may be delayed with respect to galaxy formation, since a deep enough potential well is required. 111 The duration of the lag depends on the precursor history, ranging from ~ 109 years in the case of a relativistic cluster of stellar compact objects 112 to about ~ 10M8-1 years for a supermassive radiation pressure supported star, radiating at the Eddington limit. 113
The complexities of the black hole formation process must in general be investigated numerically. The outcome depends on a variety of factors, including the amount of rotation, the fuel supply, the gas and dust distribution, the stellar density and velocity dispersion (which determine the collision rate), and so on. It seems more likely that the early evolution leads to a single stellar black hole, surrounded by predominantly normal stars and gas, rather than to the relativistic cluster of ~ 108 compact objects within a volume ~ 10 rg across, assumed as an initial condition by Shapiro and Teukolsky. 112 Ozernoy114 found (analytically) that central black holes of various sizes were obtainable, provided the initial velocity dispersion and core density were high enough. Other numerical simulations showed that the evolution from a single seed black hole of a mass M ~ 103 M should lead to a supermassive one. These simulations usually offer no explanation as to the origin of such a seed hole. However, Quinlan and Shapiro 115 showed that formation may arise rather easily from stellar mergers in a dense, but otherwise normal galactic nucleus.