On Information/Entropy Flow in Stochastic Dynamical Systems

摘要

The objective of this paper is to show how some basic informational quality measures (such as entropy and relative entropy / Kullback divergence) of stochastic dynamical systems depend on the system properties and characteristics of the external/internal randomness. First, the Shannon entropy flow in dynamic systems with random initial states is considered with emphasis on the effects of the system properties. Next, we quantify the influence of random external noise as well as the parametric randomness on the entropy and on the Kullback-Leibler relative entropy of the system. The analysis is illustrated by specific dynamical systems for which the entropy change in time is presented graphically.