Weakly nonlinear Schrödinger equation with random initial data (Q625491): Difference between revisions

The authors approximate a weakly nonlinear wave equation through a kinetic transport equation, controlling the validity of the kinetic limit for a suitable choice of random initial data. More precisely, as wave equation they consider the nonlinear Schrödinger equation (NLS) discretized on a hyperbolic lattice. The NLS is a Hamiltonian system, and the Cauchy problem for NLS has a unique global solution. Here, the authors are interested in random initial data. From a statistical physics point of view, a very natural choice is to take the initial \(\psi\)-field (NLS) to be distributed according to a Gibbs measure for the Hamiltonian function and \(\lambda_{2}\) norm, which physically means that the wave field is in thermal equilibrium. The solution of the NLS equation yields then a stochastic process stationary in \(x\in \mathbb Z^{d}\) and \(t\in \mathbb R\). Next, the authors consider the continuum NLS equation. In this case, the Gibbs measure at finite volume is a perturbed Gaussian measure which is singular at short distances, and the most basic quantity is the stationary covariance with respect to the Gibbs probability measure. If \(\lambda\) denotes the strength of nonlinearity in the NLS equation, the authors prove that the space-time covariance of NLS solutions has a limit as \(\lambda\to 0\) for time scale \(t={\lambda}^{-2}\tau\), with \(\tau\) fixed and \(|\tau |\) sufficiently small. The limit agrees with predictions from kinetic theory. The authors derive a suitable time-dependent perturbation expansion and develop a graphical language to describe a large, but finite, number of terms in the expansion. They use the method of oscillatory integrals.