2004MNRAS.355.1217H

We study analytically the coarse- and fine-grained distribution function (DF) established by the self-similar infall of collisionless matter. We find this function explicitly for isotropic and spherically symmetric systems in terms of cosmological initial conditions. The coarse-grained function is structureless and steady but the familiar phase-space sheet substructure is recovered in the fine-grained limit. By breaking the self-similarity of the halo infall we are able to argue for a central density flattening. In addition there will be an edge steepening. The best-fitting analytic density function is likely to be provided by a high-order polytrope fit smoothly to an outer power law of index -3 for isolated systems. There may be a transition to a -4 power law in the outer regions of tidally truncated systems.

As we find that the central flattening is progressive in time, dynamically young systems such as galaxy clusters may well possess a Navarro, Frenk and White type density profile, while primordial dwarf galaxies, for example, are expected to have cores. This progressive flattening is expected to end either in the non-singular isothermal sphere, or in the non-singular metastable polytropic cores; as the DFs associated with each of these arise naturally in the bulk halo during the infall. We suggest, based on previous studies of the evolution of de-stabilized polytropes, that a collisionless system may pass through a family of polytropes of increasing order, finally approaching the limit of the non-singular isothermal sphere, if the `violent' collective relaxation is frequently re-excited by `merger' events. Thus central dominant (cD) galaxies, and indeed all bright galaxies that have grown in this fashion, should be in polytropic states. Our results suggest that no physics beyond that of wave-particle scattering is necessary to explain the nature of dark matter density profiles. However, this may be assisted by the scattering of particles from the centre of the system by the infall of dwarf galaxies, galactic nuclei or black holes (e.g. Nakano & Makino), all of which would restart pure dynamical relaxation.