CENTER FOR COMPUTATION & TECHNOLOGY

In applications of physics, biology, meteorology and chemistry, researchers are interested in the stationary distribution of a randomly perturbed dynamical system. The distribution density follows the Fokker-Planck PDE which is not easy to solve by classic numerical methods, especially for those dynamical systems of high dimensions. In this work, we introduce a data driven solver for the stationary Fokker-Planck equation, and improve it by its block versions. Analysis of the convergence of the method will also be provided.

Dr. Jiayu Zhai received Ph.D. from Louisiana State University and one part of his Ph.D. thesis research is computations of rare events in stochastic dynamical systems under the supervision of Professor Xiaoliang Wan. Dr. Zhai is currently working as a Visiting Assistant Professor in the Department of Mathematics and Statistics at University of Massachusetts Amherst. His research interests are stochastic analysis and numerical methods.