ABSTRACT
The general approach to non-equilibrium statistical physics consists in reducing the original many-body problem to a (restricted) set of relevant variables. This procedure is all the easier and all the more justified if the timescales can be decoupled. The relevant variables are then associated with slow dynamics while (too!) rapid degrees of freedom are treated in an approximate way, buried in a stochastic force, as in the generic example of Brownian motion. The decoupling of timescales, which allows this elegant reduction of a complex dynamics into a, if not simple, at least ordered picture, is only marginally true in the case of the nuclear collisions we aim to describe. Still, this general decoupling scheme constitutes one of the firmest bases of our understanding of nuclear dynamics in the nucleonic regime, via the key role played by one-body degrees of freedom, as already widely illustrated in chapter 2 for nuclei close to equilibrium.
