The physics of galaxy clustering. I. A model for subhalo populations

We present a semianalytic model for cold dark matter halo substructure that can be used as a framework for studying the physics of galaxy formation and as an ingredient in halo models of galaxy clustering. The model has the following main ingredients: ( 1) extended Press-Schechtermass accretion histories, ( 2) host halo density profiles computed according to the trends observed in cosmological simulations, ( 3) distributions of initial orbital parameters of accreting subhalos measured in a high-resolution simulation of three MilkyWay - size halos, and ( 4) integration of the orbital evolution of subhalos including the effects of dynamical friction and tidal mass loss. We perform a comprehensive comparison of the model calculations to the results of a suite of high-resolution cosmological simulations. The comparisons show that subhalo statistics such as the velocity and mass functions, the radial distributions, and the halo occupation distributions agree well over 3 orders of magnitude in host halo mass and at various redshifts. We find that both in the simulations and in our model the radial distributions of subhalos are significantly shallower than that of the dark matter density. The abundance of subhalos in a host is set by competition between tidal disruption and new accretion. Halos of high mass and halos at high redshift tend to host more subhalos because the subhalos have, on average, been accreted more recently. Similarly, at a fixed mass and epoch, halos that formed more recently host a larger number of subhalos. Observed fossil groups'' may represent an extreme tail of this correlation. We find a related correlation between host halo concentration and satellite abundance at fixed host mass, N-sat proportional to c(vir)(-a), where a changes with redshift and host-to-subhalo mass ratio. Lastly, we use our substructure model to populate host halos in one of the high-resolution cosmological simulations, replacing the actual subhalos resolved in this simulation and using the host mass as the only input for the model calculation. We show that the resulting correlation function of such a hybrid halo ensemble is indistinguishable from that measured directly in the simulation. This supports one of the key tenets of the standard halo model, i.e., the assumption that the halo occupation distribution is statistically independent of the host halo environment.