Directional convexity and finite optimality conditions

For Λ ε{lunate} R^d, we say that a set A⊆R^d is Λ-convex if the segment pq is contained in A whenever p, q ε{lunate} A and p - q ε{lunate} Λ. For the control system x ̇(t)=f(x(t), u(t)), x(0)=0εR^d with u(t)εΩ⊆R^m for tε[0, T], results are provided which establish the Λ-convexity of the reachable set R(T). They rely on a uniqueness assumption for solutions of Pontrjagin's equations and are proven by means of a generalized Mountain Pass Theorem. When the reachable set is Λ-convex, trajectories reaching the boundary of R(T) satisfy some additional necessary conditions. A stronger version of the Pontrjagin Maximum Principle can thus be proven.