Mild solutions of semilinear evolution inclusions and optimal control (Q1896416)

The paper is devoted to the study of semilinear evolution inclusions \[ \dot x(t)\in A(t) x(t)+ F(t, x(t)),\quad x(0)= x_0, \] where \(x\in X\) with \(X\) being a separable Banach space and \(\{A(t)\}\) \((t\in T)\) is a family of unbounded linear operators which generates an evolution operator \(S(t, s)\). The author investigates the ``mild'' solutions \(x(t)\) of this inclusion, i.e., such that \[ x(t)= S(t, 0) x_0+ \int^t_0 S(t, s) f(s) ds,\quad t\in T, \] with \(f(\cdot)\) being a selector of \(F(\cdot, x(\cdot))\) that belongs to the Lebesgue-Bochner space \(L^1(\Omega, X)\). The main results given here are the followings: (i) The conditions on the data under which the set of mild solutions is nonempty and compact in \(C(T, X)\); (ii) The density (or relaxation) theorem for mild solution of the primary problem and the ``relaxed'' one (with \(F\) replaced by \(\overline {\text{conv}} F\)); (iii) The continuous dependence theorem of mild solutions for the perturbed mapping \(F(t, x, \lambda)\) from \(\lambda\). The results are applied to infinite-dimensional optimal control problems. An example of feedback control problem for the parabolic differential system is considered.