No label defined (Q4501647)

This paper deals with the existence of solutions for optimal control problems of the form \[ \text{Minimize}\quad J(x,u)=\varphi(x(T)), \] where the dynamics of the system is \(x'=f(t,x,u)\), \(x(0)=x_0\), and the control \(u\) satisfies \(u(t)\in U\), for almost every \(t\in [0,T]\). After giving a review of results based on convexification-deconvexification procedures, the author presents results based on optimality conditions. More precisely, introducing the argmaximum mapping \(M(t,x,p)=\text{argmax}_U\langle p,f(t,p,\cdot) \rangle\) and the set \(\overline V(t)=f(t,\overline x(t),\) \(M(t,\overline x(t), \overline p(t)))\), for an optimal relaxed trajectory \(\overline x\), and the corresponding adjoint state \(\overline p\), it is proved that if \(\overline V(t)\) is convex a.e., then the original optimal control problem admits solutions. Other results involving second order conditions are also stated. Many nontrivial examples are given, showing the interest of the results. Most of these results may be found in the book by the author: ``Approximation methods in problems of optimization and control'' (1988; Zbl 0643.49001). For the entire collection see [Zbl 0921.00020].