Spectral criteria to stability and observability of mean-field stochastic periodic systems (Q2151865)

Firstly, the paper tackles the problem of stability of the discrete time stochastic system in \(R^n\) with matrix coefficients \( A_t, \bar A_t, C_t, \bar C_t\) depending periodically on time \(t\) with a period~\(\theta\), \begin{align*} x_{t+1} = A_t x_t + \bar A_t Ex_t + (C_t x_t + \bar C_t Ex_t)w_t, \; t=0,1,\ldots \end{align*} with initial data \(x_0=\xi\in R^n\), where \(E\) stands for the expectation, \((w_t)\) is a one-dimensional stationary process in the wide sense, \(Ew_t=0\) and \(\mbox{cov}(w_t w_s) = \delta_{ts}\). The terminology ``mean-field'' here is due to the terms with expectations. Theorem 2.1 states necessary and sufficient conditions of the asymptotic stability of such a system in the square mean sense, as well as of the weak stability in terms of the spectrum of an auxiliary monodromy operator. Theorem 2.2 provides one more criterion for the same properties in terms of the existence of a quadratic form which is a Lyapunov matrix-valued function for some positive definite matrix sequence. Theorem 2.3 gives a criterion in terms of linear inequalities for the stabilization of the controlled system \begin{align*} &x_{t+1} \!=\! A_t x_t \!+\! \bar A_t Ex_t \!+\!B_tu_t \!+\! \bar B_t Eu_t \\ &+\! (C_t x_t \!+\! \bar C_t Ex_t \!+\! D_t u_t \!+\! \bar D_t Eu_t)w_t, \; t\!=\!0,1,\ldots \end{align*} Further, for the system \begin{align*} &x_{t+1} = A_t x_t + \bar A_t Ex_t + (C_t x_t + \bar C_t Ex_t)w_t, \\ &y_t = Q_t x_t + \bar Q_t Ex_t, \quad t=0,1,\ldots, \end{align*} with the input process \(x_t\) and observations \(y_t\), theorem 3.1 states the Popov-Belevitch-Hautus type criterion in spectral terms for the ``exact observability''. Finally, theorem 3.2 establishes a Barbashin-Krasovskii type result which links the properties of the observability and asymptotic mean square stability for the latter system. Quite a few earlier results are cited related to homogeneous stochastic systems of a similar sort, to the LQ stochastic control, et al. Particularly, the monographs by \textit{S. Bittanti} and \textit{P. Colaneri} [Periodic systems. Filtering and control. London: Springer (2009; Zbl 1163.93003)] and by \textit{V. Dragan} et al. [Mathematical methods in robust control of discrete-time linear stochastic systems. Dordrecht: Springer (2010; Zbl 1183.93001)] are important bases for the proofs.