Bayesian optimal experimental design

Nested integral estimation using median randomized quasi-Monte Carlo

Arved Bartuska, Postdoctoral Research Fellow, Stochastic Numerics Research Group

Oct 13, 13:00 - 14:00

Bayesian optimal experimental design Quasi-Monte Carlo

Abstract: Nested integrals of the form $\int f\left(\int g(\bs{y},\bs{x})\di{}\bs{x}\right)\di{}\bs{y}$, where $f$ is a nonlinear function, arise in various fields such as Bayesian experimental design, medical decision making, and computational finance. We develop a novel double-loop median randomized quasi-Monte Carlo (RQMC) estimator to address such integrals. It was recently demonstrated that the median RQMC method achieves super-polynomial convergence rates under certain smoothness assumptions. We show that super-polynomial convergence rates are also possible in the nested setting for a

Darcy flow in random porous medium

AMCS 301 Numerical methods for random partial differential equations: hierarchical approximation and machine learning approaches

Random PDEs stochastic algorithms Monte carlo methods Quasi-Monte Carlo Hierarchical regression Multilevel Monte Carlo Stochastic collocation Multi-index Low-rank approximation hierarchical and sparse approximation Bayesian Inversion Bayesian optimal experimental design

A course on modern numerical methods for random partial differential equations