Weak convergence of numerical methods for stochastic gradient systems in the t->infinity limit Seminar

Numerical methods for SDEs are frequently analyzed in terms of their weak convergence order, specifically the order of accuracy of averages of suitable test functions determined on finite time intervals. In this talk I will discuss the convergence of certain methods for stochastic gradient systems in the weak sense and the progression to a superconvergence property (i.e. a higher order of accuracy for averages with respect to the invariant measure) in the large time limit. This talk is joint work with Charles Matthews (Edinburgh) and Michael Tretyakov (Nottingham).