Stanford 50: State of the Art and Future Directions of Computational Mathematics and Numerical Computing


  • March 30, 2007
  • 4:00 pm - 4:25 pm

MCMC in infinite dimensions

Andrew Stuart (University of Warwick)

In many application areas it is of interest to sample a probability measure on a space of functions: an infinite dimensional sampling problem. Applications include molecular dynamics, signal processing, econometrics and data assimilation. For this reason it is important to be able to develop efficient algorithms to perform sampling for such problems. Markov Chain Monte Carlo (MCMC) has proved an effective tool in a wide variety of applications and it is natural to develop an understand of its computational complexity in the context of sampling function space.

In this talk I will illustrate the applications of interest; describe their common mathematical structure; and overview the theoretical understanding that has been developed for the sampling of problems with this mathematical structure.

Stanford University Home Page