On the continuous resonant equation for NLS. I: Deterministic analysis. (Q897836)

In this paper, the authors continue their study of the so-called continuous resonant equation in \(2+1\) dimensions: \[ i \partial_t u = \mathcal T(u, u, u), \] where \(\mathcal T\) is a certain trilinear integral operator. This equation was introduced in the paper [``The weakly nonlinear large box limit of the 2D cubic nonlinear Schrödinger equation'', Preprint, \url{arXiv:1308.6267}] by \textit{E. Faou}, \textit{P. Germain} and \textit{Z. Hani}, as a small data approximation for the cubic nonlinear Schrödinger equation in the large-box limit. This is a non-dispersive Hamiltonian equation, similar from this point of view to the Szegö equation. It is well-posed in \(L^2\). In the current paper, the authors show that the continuous resonant equation can also serve as a small data approximation to the 2-dimensional cubic NLS equation with harmonic trapping \[ i \partial_t u - \Delta u + |x|^2 u = |u|^2 u. \] The authors also study the properties of stationary solutions. They prove that stationary solutions with epsilon extra regularity and smoothness must in fact be Schwartz functions. A certain class of \(L^2\) irrotational stationary solutions are shown to be analytic and to decay exponentially. The authors then focus on particular classes of stationary solutions, given by Hermite functions, showing orbital stability (using a variational approach) or instability in certain cases. The authors also make some further conjectures.