On controllability of linear stochastic systems (Q4485351)

Linear, continuous-time, stochastic control systems are considered. The concepts of complete controllability, approximate controllability and \(S\)-controllability are introduced and discussed. It is generally assumed that the stochastic control system is partially observable and that values of controls are not constrained. Using methods taken from functional analysis and the theory of stochastic differential equations, necessary and sufficient conditions for different types of controllability are formulated and proved. It is shown that the notions of complete controllability and approximate controllability are equivalent. Moreover, they are also equivalent to the controllability of linear stochastic systems controlled by Gaussian processes. Several remarks and comments on stochastic controllability problems are also presented. It should be pointed out that similar stochastic controllability problems have been considered in the paper [\textit{J. Klamka} and \textit{L. Socha}, IEEE Trans. Autom. Control 22, 880-881 (1977; Zbl 0363.93048)].