Title: Relaxation of the distribution function tails for systems described by Fokker-Planck equations

Abstract: We study the formation and the evolution of velocity distribution tails for systems with long-range interactions. In the thermal bath approximation, the evolution of the distribution function of a test particle is governed by a Fokker-Planck equation where the diffusion coefficient depends on the velocity. We extend the theory of Potapenko et al. [Phys. Rev. E, {\bf 56}, 7159 (1997)] developed for power law potentials to the case of an arbitrary potential of interaction. We study how the structure and the progression of the front depend on the behavior of the diffusion coefficient for large velocities. Particular emphasis is given to the case where the velocity dependence of the diffusion coefficient is Gaussian. This situation arises in Fokker-Planck equations associated with one dimensional systems with long-range interactions such as the Hamiltonian Mean Field (HMF) model and in the kinetic theory of two-dimensional point vortices in hydrodynamics. We show that the progression of the front is extremely slow (logarithmic) in that case so that the convergence towards the equilibrium state is peculiar.