# Mathematics Colloquia and Seminars

### A discrete-time approach to data-based stochastic model reduction for chaotic systems

PDE and Applied Math Seminar

 Speaker: Fei Lu, LBNL Location: 2112 MSB Start time: Thu, Nov 10 2016, 4:10PM

The need to deduce reduced computational models from discrete observations of
complex systems arises in many areas of science and engineering, e.g. in
meteorology, climate modeling, and materials science. The challenges come
mainly from non-Markovian effects and nonlinear interactions between
observed and unobserved variables, and from the difficulty in inference from
discrete data.

We address these challenges by developing discrete-time stochastic models. We
discuss the comparative advantages of discrete models and show by example that
they can capture the long-time statistics and can be used to make medium-term
predictions. The examples include the Lorenz 96 model (which is a simplified
model of the atmosphere) and the Kuramoto-Sivashinski model of
spatiotemporally chaotic dynamics.