“…To derive necessary conditions for an optimal control, we would like to differentiate the map (ϕ, ψ) → T (ϕ, ψ). Since the map (ϕ, ψ) → T (ϕ, ψ) is not directly differentiable (see [28]), the idea here consists in approximating the map T (ϕ, ψ) by a family of maps T δ (ϕ, ψ) and replacing the obstacle problem given by (10) and (11) by the following smooth semilinear equation (see [12] and [27]):…”