报告题目：Bayesian computational framework for the high dimensional stochastic inversion problems

报告人：Dr. Wenju Zhao (赵文举博士), Department of Mathematics, Southern University of Science and Technology, Shenzhen, Guangdong

报告时间：2018年11月1日 14:00至15:00

报告地点：山东大学中心校区知新楼B1044

摘要：

This research aims to present an efficient Bayesian computational framework to solve stochastic inverse fluid dynamic model, e.g., recovering the stochastic coefficients.  The difficulties come from sparse observations, noisy measurements, and highly nonlinear, non-Gaussian properties resulting that performing stochastic inversion with a high-dimensional uncertain parameter space is computationally prohibitive even with efficient gradient information. A efficiently objective probabilistic solution is then  constructed by incorporating some techniques related to the PDE constrained optimal control, Bayesian inferences, polynomial chaos,  principle orthogonal decomposition and kernel principle component analysis, etc.  Then, classical Metropolis-Hastings MCMC and gradient-based Langevin Markov Chain Monte Carlo (LMCMC) method are performed and compared that show that the efficiency of methods.

赵文举博士的简介：

Dr. Wenju Zhao is postdoctoral researcher under the supervise of Professor Tao Tang in the Department of Mathematics at Southern University of Science and Technology, Shenzhen. He got his doctorate degree under the supervise of Professor Max Gunzburger in Computational Science at Florida State University, USA.

邀请人：赵卫东