Dynamic approximation of a random information functional (Q5898327)

Uncovering unknown dynamic regularities of a random process is an actual problem in the theory of dynamic and stochastic systems. Numerous examples from physics show that regularities of deterministic and stochastic systems are formulated and formalized by some extremal principles: a real movement of a physical system represents trajectory-extremals of some functional. This paper has a primary goal of applying the above extremal principle for an approximation of the process' extremals by the most probable objects' trajectories. The model of a controlled random (micro)process (as a solution of a stochastic equation with a class of control functions), the models of the disturbances, and a programmable process are introduced, followed by an extremal probabilistic aproximation of the microprocess with the aid of the applied control. The probabilistic evaluation of the micro- and the macrolevel processes is provided. The probabilities of the processes' proximity, based on the process trajectories' metrical space distances, are applied to evaluate the upper and lower limits of the aproximation. Using the ``functional action'' approach, the extremal problem is expressed by the functionals and the dynamic (macro)trajectories (determined by the parameters of the stochastic equation) and their closeness to the macroprocess' conditional entropy.