Vector entropy monitoring and control of gaussian stochastic systems

The article deals with the vector approach of entropy monitoring and control. It is the representation of differential entropy of a multidimensional stochastic system as a two-dimensional vector, the components of which are the entropies of randomness and self-organization. The system state is evaluated simultaneously with these two components. Vector control enables efficient entropy change as a two-dimensional vector, the components of which are the entropies of randomness and self-organization. We have formulated an optimization extremum problem for the important case of Gaussian stochastic systems. This problem can be solved by penalty function methods. It is shown that in a number of cases vector entropy control has advantages in comparison with scalar control. We give the examples of entropy monitoring and control for real stochastic systems.