A transformed stochastic Euler scheme for multidimensional transmission PDE
Seminar Probabilités & Statistique
10/10/2019 - 14:00 Pierre Etoré (LJK) Salle 106 - Batiment IMAG
In this work we consider multi-dimensional Partial Differential Equations (PDE) of parabolic type in divergence form. The coefficient matrix of the divergence operator is assumed to be discontinuous along some smooth interface. At this interface, the solution of the PDE presents a compatibility transmission condition of its conormal derivatives (multi-dimensional diffraction problem). We prove an existence and uniqueness result for the solution and study its properties. In particular, we provide new estimates for the partial derivatives of the solution in the classical sense. We then establish the link between the PDE of interest and some Hunt process, via the theory of Dirichlet forms. We describe the dynamic of this process in terms of Stochastic Differential Equations (SDE). Unfortunately the obtained SDE is not suitable for numerical purposes. We then choose another approach and construct a low complexity numerical Monte Carlo stochastic Euler scheme to approximate the solution of the PDE of interest. Using the afore mentioned estimates, we prove a convergence rate for our stochastic numerical method when the initial condition belongs to some iterated domain of the divergence form operator. Finally, we compare our results to classical deterministic numerical approximations and illustrate the accuracy of our method. This is a joint work with Miguel Martinez (UMLVPE). Talk will be in English.