Uy, Wayne Isaac: Travel Grant - 2017-2018

College of Engineering
Project Title: 
Enhanced Sampling in Bayesian Inverse Problems
Project Abstract: 
This project is concerned with the development of novel methods that accelerate the convergence of sampling of the posterior density required to solve Bayesian inverse problems. As the commonly used Markov Chain Monte Carlo sampling becomes impractical in high dimensions, we investigate the use of Mix & Match Hamiltonian Monte Carlo sampling developed. This sampling approach uses gradient and Hessian information of the posterior density and for this reason, efficient computational methods for obtaining these quantities need to be developed. The performance of this enhanced sampling approach will be analyzed in an engineering application in which information about the Earth's subsurface will be inferred from measurements of electromagnetic resistivity on the Earth's surface.