Invariant set stability in discrete nonlinear systems (Q1369932)

Lyapunov's first method of analysis of the stability of a dynamic system is based on the linearization of the right-hand side near the stationary point. The objective of the paper is to apply this idea for the analysis of the stability of a discrete inclusion. It is assumed that the right-hand side of the inclusion is an upper semi-continuous compact-valued mapping generated by a single-valued function with bounded uncertainties. Stability of a closed set \(X\) with respect to a discrete inclusion means that for any \(\varepsilon\)-neighbourhood of \(X\) there exists a \(\delta\)-neighbourhood of \(X\) such that every trajectory starting in the \(\delta\)-neighbourhood remains in the \(\varepsilon\)-neighbourhood. Asymptotic stability additionally means that all trajectories converge to \(X\) at infinity. The weak stability concept involves only the trajectories corresponding to constant values of the uncertain parameter, whereas strong stability involves all solutions of the inclusion. For an inclusion the concept of a stationary point is replaced by the concept of a stationary set in a natural way. The authors prove several theorems with sufficient conditions of asymptotic stability using different kinds of linearization of multi-valued mappings. For example, in Theorem 3.1 it is obtained that if \(X_0\) is a stationary set of a discrete inclusion with the mapping \(F(\cdot)\) in the right-hand side, then \(X_0\) is strongly asymptotically stable under the condition that \[ \limsup_{\delta\to+0} \delta^{-1}h(F(X_0), F(X_0+\delta S))<1, \] where \(S\) denotes the closed unit sphere in \(R^n\) and \(h(\cdot,\cdot)\) is the Hausdorff distance.