Uniqueness of motion by mean curvature perturbed by stochastic noise. (Q1426662): Difference between revisions

A time dependent hypersurface in \({\mathbb R}^n\) is said to evolve by mean curvature if the outward normal velocity at every point of the surface equals its mean curvature. A motion by mean curvature (MMC) can be defined as a level set of the evolution equation in \({\mathbb R}^n \times {\mathbb R}_+\): \[ {\partial \over \partial t} u = | Du | \left( \text{ div} \left( {Du \over | Du| } \right) + g(t) \right) \] where \(g(t)\) is a time varying function. In the paper under review, the level set evolution equation is perturbed by addition of a Gaussian white noise in the Stratonovich sense: \[ d u = | Du | \left( \text{div} \left( {Du \over | Du| } \right) + g(t) \right) dt + \varepsilon | Du | \circ dW(t). \] Given two particular choices of initial data yielding non-unique deterministic MMCs, the authors study the convergence to deterministic MMCs as the noise perturbation vanishes. In the first example, consisting of two touching balls with \(g(t)=0\), convergence is shown to hold in the almost sure sense while in the second example, where the balls are separated and \(g(t)\) is a decreasing function, the probability law of the limit is determined. The proofs rely on viscosity solution methods, recently put forward by P. L. Lions and the first author for nonlinear second-order stochastic PDEs.