The quasilinear theory in the approach of long-range systems to quasi-stationary states

Abstract

We develop a quasilinear theory of the Vlasov equation in order to describe the approach of systems with long-range interactions to quasi-stationary states. The quasilinear theory is based on the assumption that, although the initial distribution is not Vlasov stable, nevertheless its evolution towards a Vlasov stable stationary state is such that it is always only slightly inhomogeneous. We derive a diffusion equation governing the evolution of the velocity distribution of the system towards a steady state. This steady state is expected to correspond to the space-averaged quasi-stationary distribution function reached by the Vlasov equation as a result of a collisionless relaxation. We compare the prediction of the quasilinear theory to direct numerical simulations of the Hamiltonian mean field model, starting from an unstable spatially homogeneous distribution, either Gaussian or semi-elliptical. In the Gaussian case, we find that the quasilinear theory works reasonably well for weakly unstable initial conditions (i.e. close to the critical energy ) and that it is able to predict the energy marking the effective out-of-equilibrium phase transition between unmagnetized and magnetized quasi-stationary states found in the numerical simulations. Similarly, the quasilinear theory works well for energies close to the instability threshold of the semi-elliptical case , and it predicts an effective out-of-equilibrium transition at . In both situations, the quasilinear theory works less well at energies lower than the out-of-equilibrium transition, the disagreement with the numerical simulations increasing with decreasing energy. In that case, we observe, in agreement with our previous numerical study (Campa and Chavanis 2013 Eur. Phys. J. B 86 170), that the quasi-stationary states are remarkably well fitted by polytropic distributions (Tsallis distributions) with index n = 2 (Gaussian case) or n = 1 (semi-elliptical case). In particular, these polytropic distributions are able to account for the region of negative specific heats in the out-of-equilibrium caloric curve, unlike the Boltzmann and Lynden-Bell distributions.