Learning Solution of Stochastic Partial Differential Equations (A Continuation)

The talk will be firstly introduced J. Warner and S. Koutsourelakis's recent paper of Learning Solution to Multiscale Elliptic Problems with Gaussian Process Models. The basic ideas of prediction using function observations derivation observations for Gaussian process regression will be discussed in the talk. Next we will present the methods of Gaussian process regression for multiscale elliptic PDEs. In addition, we will introduce some new ideas of our recent report of Learning Solution of Stochastic Partial Differential Equations. We wish to apply the meshfree methods and the kernel methods into the numerical approach of SPDEs.

Event Topic:

Computational Mathematics & Statistics