David Aristoff's homepage Title: From DtN map to data-driven stochastic parameterization

Abstract: I will discuss the idea of DtN map in the reduction of molecular models. This will be formulated as a boundary integral method. When the initial state is prepared according to an appropriate probability distribution, the boundary integral approach will yield a reduced model driven by colored noise. With the ideas from absorbing boundary conditions (ABCs), we found an hierarchy of stochastic models. Finally, we discuss how to use data to determine the parameters in the stochastic model.