Stochastic Analysis, Stochastic Partial Differential Equations and Applications to Fluid Dynamics and Particle Systems
timetable:
Tue 4 Apr, 9:50 - 10:30, Aula Dini
<< Go back
seminar: Stochastic PDEs for Sampling Conditioned Diffusions I
speaker: Andrew Stuart (Warwick University, Mathematics Department)abstract: In many contexts it is of interest to sample paths of diffusion processes, conditional on observations of various kinds. We introduce an SPDE based approach to this problem, generalizing the Langevin equation to this path space setting. The Langevin equation is used to construct a Metropolis-Hastings algorithm which is shown to have desirable sampling properties. Various applications are described, including signal processing, data assimilation and rare event sampling in molecular dynamics. Joint work with Martin Hairer and Jochen Voss.
timetable:
Tue 4 Apr, 9:50 - 10:30, Aula Dini
<< Go back