Classical limit for a system of non-linear random Schrödinger equations (Q2435015)

The paper is concerned with the following random nonlinear Liouville equation for the density operator \(\varrho^\epsilon\) in three dimensions \[ \begin{aligned} i\epsilon\partial_t&\varrho^\epsilon=\left[-\frac{\epsilon^2}{2}\Delta+V^\epsilon+U^\epsilon, \varrho^\epsilon\right], \quad t>0,\quad x\in\mathbb{R}^3\\ {}&\varrho^\epsilon (t=0)=\varrho^\epsilon_0. \end{aligned} \] Here \(\varrho^\epsilon_0\) is a given initial density operator (i.e., a trace class, positive, Hermitian operator), \(V^\epsilon\), \(U^\epsilon\) are multiplication operators, and \([\cdot, \cdot]\) denotes the commutator. The potential \(U^\epsilon\) is the Poisson interaction potential \[ U^\epsilon=\frac{1}{4\pi|x|}* n^\epsilon \] where \(n^\epsilon(t,x)=\rho^\epsilon(t,x,x)\) is the density matrix, \(\rho^\epsilon(t,x,y)\) the integral kernel of \(\varrho^\epsilon\). The real potential \(V^\epsilon\) is random and of the form \[ V^\epsilon (t,x)=\sqrt{\epsilon}\;V\left(\frac{t}{\epsilon},\frac{x}{\epsilon}\right) \] for some random potential \(V\). It is required to be Markovian with a bounded generator. The main result states that the Wigner transform \(W^\epsilon\) of the density matrix converges weakly in \(L^2\) and in probability to the unique solution \(W\) of a Vlasov-Poisson-Boltzmann equation. Moreover, the density and the current are self-averaging. The proof is based on the martingale method for stochastic equations and on compactness techniques for nonlinear partial differential equations in a semi-classical setting.