Existence of the kernel of a convex set in linear discrete control systems (Q1916605)

The author introduces the concept of a weakly positively invariant set with respect to a linear discrete control system \[ x_{n + 1} = Ax_n - u_n, \quad x_i = \overline x. \tag{*} \] Here \(x,u \in \mathbb{R}^d\), \(A\) is a constant \(d \times d\) matrix and \(u_n\) is a control sequence with values from a compact convex set \(P \subset \mathbb{R}^d\). Weak positive invariance of a set \(Y\) with respect to the system (*) means that for each \(\overline x \in Y\) there exists a control sequence \(u_n\) for which the whole relevant trajectory lies inside \(Y\). The maximal subset of a set \(Y'\) which is weakly positively invariant is called the kernel of \(Y'\). For continuous systems, the property of weak positive invariance is also known under the name of viability. Some properties of weakly invariant subsets, kernels and their recession cones (in the paper, a recession cone is called asymptotic) are investigated. For example, Theorem 4 claims that if all real eigenvalues of the matrix \(A\) are less than one, then a set of \(G\) has a nonempty kernel if and only if \(G\) has a nonempty intersection with the set \[ \{x \in \mathbb{R}^d : x \in Ax - P\}. \] Unfortunately, in the English version several terms are translated in a wrong way.