Friday, November 16, 2018 - 10:15
SFB 1294 Seminar, University of Potsdam
Karl-Liebknecht- Str. 24-25, 14476 Potsdam OT Golm, House 9, Room 0.14
When sampling measures on Hilbert spaces one option is to first discretize the space and then apply standard Markov Chain Monte Carlo methods. This works but the finer the discretization is, the slower convergence rate of the chains is. Alternatively one can also first consider a well defined method on the function space directly and afterwards discretize. I will talk about a version of the HMC method on Hilbert spaces that has a diffusion limit converging to an SPDE. This leads to methods that converge equally well independently of the chosen discretization.
submitted by Liv Heinecke (liv.heinecke@uni-potsdam.de, 0331-977-203137)