Professor Annie Millet, Université Paris 1 - Data Science
Date: - Location: Eurecom
Abstract: We study finite dimensional diffusion processes $(X_t, t\in [0,1])$ such that $X_t$ has a density for every $t>0$. We give a necessary and sufficient condition for the time reversed process $(Y_t=X_{1-t}, t\in [0,1])$ to be a diffusion, and identify its diffusion and drift coefficients. We prove similar results for a diffusion $(X_t^i, i\in Z, t\in [0,1])$ solution to an infinite dimensional system of SDEs. The proofs use stochastic calculus of variations; this is joint work with D. Nualart and M. Sanz-Sol\'e. Bio: Annie Millet is a Full Professor at Université Paris 1 Panthéon Sorbonne. Her current research topic is Nonlinear Stochastic Partial Equations, such as the Navier-Stokes equations and more general hydro dynamical models, the Non Linear Schrödinger equation, ...) subject to an additive or multiplicative stochastic perturbation. She is interested in strong/weak rates of convergence of numerical schemes of approximation, and by the effect of the noise on blow-up phenomenons for the focusing NLS (theoretically and numerically).
