On the derivation of the homogeneous kinetic wave equation

Abstract: The kinetic wave equation arises in many physical situations: the description of small random surface waves, or out of equilibria dynamics for large quantum systems for example. In this talk we are interested in its derivation as an effective equation from the nonlinear Schrodinger equation (NLS) for the microscopic description of a system. More precisely, we will consider (NLS) in a weakly nonlinear regime on the torus in any dimension greater than two, and for highly oscillatory random Gaussian fields as initial data. A conjecture in statistical physics is that there exists a kinetic time scale on which, statistically, the Fourier modes evolve according to the kinetic wave equation. We prove this conjecture up to an arbitrarily small polynomial loss in a particular regime, and obtain a more restricted time scale in other regimes. The main difficulty, that I will comment on, is that one needs to identify the leading order statistically observable nonlinear effects. This means understanding correlation between Fourier modes, and relating randomness with stability and local well-posedness. The key idea of the analysis is the use of Feynman interaction diagrams to understand the solution as colliding linear waves. We use this framework to construct an approximate solution as a truncated series expansion, and use in addition random matrices tools to obtain its nonlinear stability in Bourgain spaces. This is joint work with P. Germain.

analysis of PDEs

Audience: researchers in the topic

PDE seminar via Zoom