Optimization Based Methods for Partially Observed Chaotic Systems

DOI10.1007/S10208-018-9388-XMaRDI QIDQ6282937arXivWikidataFDO

Authors Daniel Paulin, Ajay Jasra, Dan Crisan, Alexandros Beskos

Publication date 8 February 2017

Published in arXiv

Full work available at URL https://doi.org/10.1007/S10208-018-9388-X

Mathematics Subject Classification ID

Calculus of variations and optimal control; optimization (49-XX) Numerical analysis (65-XX) Systems theory; control (93-XX)

arXiv classification

math.DSmath.OCstat.ME

Abstract: In this paper we consider filtering and smoothing of partially observed chaotic dynamical systems that are discretely observed, with an additive Gaussian noise in the observation. These models are found in a wide variety of real applications and include the Lorenz 96' model. In the context of a fixed observation interval $T$, observation time step $h$ and Gaussian observation variance $sigm{a}_Z^2$, we show under assumptions that the filter and smoother are well approximated by a Gaussian with high probability when $h$ and $sigm{a}_Z^{2}h$ are sufficiently small. Based on this result we show that the Maximum-a-posteriori (MAP) estimators are asymptotically optimal in mean square error as $sigm{a}_Z^{2}h$ tends to $0$. Given these results, we provide a batch algorithm for the smoother and filter, based on Newton's method, to obtain the MAP. In particular, we show that if the initial point is close enough to the MAP, then Newton's method converges to it at a fast rate. We also provide a method for computing such an initial point. These results contribute to the theoretical understanding of widely used 4D-Var data assimilation method. Our approach is illustrated numerically on the Lorenz 96' model with state vector up to 1 million dimensions, with code running in the order of minutes. To our knowledge the results in this paper are the first of their type for this class of models.

This page was built for publication: Optimization Based Methods for Partially Observed Chaotic Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6282937)