Relaxation for a class of control systems with unilateral constraints (Q2184832)

The paper is concerned with a nonlinear feedback control system of type \[ \left\{\begin{array}{ll}-x' \in A(x(t))+f(t,x(t))u(t)\\ x(0)=x_0, \quad u(t) \in U(t,x(t))\end{array}\right. \] in a time interval \(T=[0,b]\). Here, \(A:D(A)\subset \mathbb{R}^N \to 2^{\mathbb{R}^N}\) is a maximal monotone mapping and the control constraint multifunction \(U:T\times \mathbb{R}^N \to 2^{\mathbb{R}^N} \setminus \{ \emptyset \}\) has nonconvex values. It is assumed that \(U(t, \cdot )\) is lower semicontinuous for a.a. \(t\in T\). The authors introduce a control relaxed system by \(Q\)-regularization (in the sense of Cesari). Then, they show that every original state is a relaxed state and the set of the original states is dense in the set of the relaxed states, which is closed in \(C(T, \mathbb{R}^N)\).