Coefficient conditions for existence of an optimal control for systems of differential equations (Q467673)

The considered system is \[ x'(t) = f_1(t, x(t)) + f_2(t, x(t))u(t) \, , \quad x(0) = x_0 \] in a domain \(D\) of the \(n\)-dimensional space. Two optimal control problems are considered. In the first one, the cost functional \[ J(u) = \int_0^\tau L(t, x(t), u(t))dt \] is minimized; \(\tau\) is the exit time of the trajectory, that is, the smallest time \(t\) for which \(x(t)\) hits the boundary. In the second problem, the trajectory stays in \(D\) and the infinite horizon functional \[ J(u) = \int_0^\infty g(t)L(t, x(t), u(t))dt \] is minimized. The authors prove the existence of optimal controls for both problems under assumptions that include smoothness and growth conditions for \(f_1(t, x),\) \(f_2(t, x),\) \(L(t, x, u),\) \(L_x(t, x, u)\) and \(L_u(t, x, u),\) plus convexity of \(L(t, x, u)\) with respect to \(u.\) The control values \(u(t)\) belong to a closed, convex control set \(U.\)