A semi-discrete scheme for the stochastic Landau-Lifshitz equation (Q487675)

In this paper, the authors consider the numerical discretization of the Landau-Lifshitz-Gilbert equation for the magnetization of ferromagnetic materials. Taking into account thermal fluctuations, a multiplicative stochastic noise term is considered, which is modelled by the derivative of a trace-class Wiener process. The authors propose a semi-discrete approximation scheme in time which is a \(\theta\)-scheme applied to a variational formulation. Only the linear part is treated implicitly here, while the nonlinear terms are discretized explicitly and one does not require the resolution of a nonlinear problem at each time step. The main result of the paper is the convergence in a weak sense of a subsequence of the scheme to a martingale solution of the stochastic Landau-Lifshitz-Gilbert equation. The proof is based on Prokhorov's theorem for tightness and Skorokhod's representation theorem for an almost surely convergent sequence on a changed probability space. The key technical tools are uniform a-priori estimates, which are only valid in expectation, but these are sufficient to establish the tightness of the laws of the approximations.