Investigation of predictability of Chaotic Signals

Abstract

Variables and functions for expressing quality of the prediction of the chaotic signal can serve as tools for simplified but useful description of the chaotic signal. In this paper, the known generation mechanism of chaos is used as a predictor but using the perturbed signal as an input. Prediction performance of the predictor is compared with the estimate of the predictability based on a mutual information between known quantities and a predicted value. mentioned two methods of investigation have different requirements for a priori knowledge of chaos. Certain constraints are given in consequence of an imperfection of numerical calculations and computer simulations.