1: \begin{abstract}

2:   We perform a detailed study of the relaxation towards equilibrium in

3:   the Hamiltonian Mean-Field (HMF) model, a prototype for long-range

4:   interactions in $N$-particle dynamics.

5:   In particular, we point out the role played

6:   by the infinity of stationary states of the

7:   associated $N\to \infty$ Vlasov dynamics.

8:   In this context, we derive a new general criterion

9:   for the stability of any spatially homogeneous distribution,

10:   and compare its analytical predictions

11:   with numerical simulations of the Hamiltonian, finite $N$, dynamics.

12:   We then propose and verify numerically a scenario for

13:   the relaxation process, relying on the Vlasov equation.

14:   When starting from a non stationary or a Vlasov 

15:   unstable stationary initial state,

16:   the system shows initially a rapid convergence

17:   towards a stable stationary state of the Vlasov equation

18:   via non stationary states:

19:   we characterize numerically this dynamical instability

20:   in the finite $N$ system by introducing appropriate indicators.

21:   This first step of the evolution towards Boltzmann-Gibbs equilibrium

22:   is followed by a slow quasi-stationary process,

23:   that proceeds through different stable stationary states of

24:   the Vlasov equation. If the finite $N$ system is initialized

25:   in a Vlasov stable homogenous state,

26:   it remains trapped in a quasi-stationary state

27:   for times that increase with the nontrivial power law $N^{1.7}$.

28:   Single particle momentum distributions in such a quasi-stationary regime

29:   do not have power-law tails, and hence

30:   cannot be fitted by the $q$-exponential

31:   distributions derived from Tsallis statistics.

32: \end{abstract}

33: