Relaxation in non-convex optimal control problems described by first-order evolution equations (Q4948078)

Within the framework of evolution triple of spaces \((V, H, V^*)\), the following optimal control problem is considered \[ \inf \int_T g(t, x(t), u(t)) dt, \;\;T = [0, 1] \tag{P} \] with respect to \((x, u)\) such that \[ \begin{cases}\dot{x}(t) + A(t, x(t)) = f(t, x(t)) u(t) \\ x(0) = x_0, \;\;u(t) \in U(t, x(t)) \end{cases} \] where \(A : T \times V \to V^*\) is a nonlinear operator, \(f : T \times H \to {\mathcal L}(Y, H)\) is a nonlinear map, \(U : T \times H \to Y\) is a closed-valued nonconvex multifunction, \(g : T \times H \times Y \to \mathbb{R}\) is nonconvex with respect to \(u\) and \(Y\) is a control space. The main theorem establishes the relationship between the solutions to (P) and the relaxed problem corresponding to (P). The optimal control Lagrange problem described by a parabolic PDE illustrates the applicability of this result.