Abstract: We introduce a data-based approach to estimating key quantities which arise in the study of nonlinear control and random dynamical systems. Our approach hinges on the observation that much of the existing linear theory may be readily extended to nonlinear systems -with a reasonable expectation of success - once the nonlinear system has been mapped into a high or infinite dimensional Reproducing Kernel Hilbert Space. In particular, we develop computable, non-parametric estimators approximating controllability and observability energy/Lyapunov functions for nonlinear systems, and study the ellipsoids they induce. It is then shown that the controllability energy estimator provides a key means for approximating the invariant measure of an ergodic, stochastically forced nonlinear system. We also apply this approach to the problem of model reduction of nonlinear control systems.
In all cases the relevant quantities are estimated from simulated or observed data. These results collectively argue that there is a reasonable passage from linear dynamical systems theory to a data-based nonlinear dynamical systems theory through reproducing kernel Hilbert spaces. This is a joint work with J. Bouvrie (MIT).
in computational cost. A reduction of computational cost would allow
weather and climate simulations at higher resolution. In the first part
of this talk, I will introduce the concept of inexact hardware and
provide results that show the great potential for the use of inexact
hardware in weather and climate simulations. In the second part of this
talk, I will discuss how rounding errors can be assessed if the forecast
uncertainty and the chaotic behaviour of the atmosphere is acknowledged.
In the last part, I will argue that rounding errors do not necessarily
degrade numerical models, they can actually be beneficial. This
conclusion will be based on simulations with a model of the
one-dimensional Burgers' equation.
In this talk we aim to filter the Majda-McLaughlin-Tabak(MMT) model, which is a one-dimensional prototypical turbulence system. Due to its inherent high dimensionality, we first try to find a low dimensional dynamical system whose statistical property is similar to the original complexity system. This dimensional reduction, called stochastic parametrization, is clearly well-known method but the value of current work lies in the derivation of an analytic closure for the parameters. We then discuss the necessity of the accurate filtering algorithm for this effective dynamics, and introduce the particle filter using the cubature on Wiener space and the recombination skill.