Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1894260 | Chaos, Solitons & Fractals | 2006 | 9 Pages |
Abstract
We address the problem of estimating parameters of chaotic dynamical systems from a time series in a situation when some of state variables are not observed and/or the data are very noisy. Using specially developed quantitative criteria, we compare performance of the original multiple shooting approach (Bock’s algorithm) and its modified version. The latter is shown to be significantly superior for long chaotic time series. In particular, it allows to obtain accurate estimates for much worse starting guesses for the estimated parameters.
Related Topics
Physical Sciences and Engineering
Physics and Astronomy
Statistical and Nonlinear Physics
Authors
Boris P. Bezruchko, Dmitry A. Smirnov, Ilya V. Sysoev,