Extremal property of the initial value of the adjoint variable in the minimization problem of the energy functional on trajectories of linear systems with constrained control (Q1585398)

The author considers the linear optimal control problem in \(\mathbb R^n\) on a fixed interval of time \([0, T]\): \[ \dot x = Ax+ u,\quad 0 \leq t \leq T,\quad x\in\mathbb R^n,\quad u\in U\subset\mathbb R^n,\quad x(0)=x_0,\quad x(T)=0, \] \[ L(u)=(1/2)\int_0^T\|u(t)\|^2 dt\to\min_{u(\cdot)}. \] Here \(x\) is the phase vector; \(u\) is the control; the control domain \(U\ni 0\) is a convex compact set; \(A\in\mathbb R^{n\times n}\) is a known matrix that does not depend on time \(t\); \(x_0\) is a given initial state of the control object; the positive parameter \(T\) satisfies the inequality \(T>T_{\text{ot}}(x_0),\) \(T_{\text{ot}}(x_0)\) is the optimal time of passage of the considered control system from the initial state \(x_0\) to the origin. Given the initial data \(\{A,x_0,U, T\},\) it is required to find the optimal control \(u_{\text{op}}(t),\) the optimal trajectory \(x_{\text{op}}(t),\) and the optimal value \(L_{\text{op}}\) of the functional \(L(u).\) To solve this problem, the Pontryagin maximum principle is applied. The existence of solution in the class of measurable admissible controls is directly verified by using standard arguments dealing with a minimizing sequence. In consequence of the uniqueness of control, being proved, which is determined from the boundary-value problem of the maximum principle, the maximum principle simultaneously serves as a necessary and sufficient optimality condition in the studied problem. The optimal control is a continuous (and even Lipschitzian) function of time. Examples illustrating applications of the tools and theory developed in this paper are given.