No label defined (Q3972320)

[For the entire collection see Zbl 0707.00014.] Making use of the theory of nonanticipating operators, the author presents an interesting abstract attempt at reformulating optimal control problems for ordinary and functional-differential equations. \{Reviewer's remarks: (1) Take \(G(y,u)(t)=Ay(t)+bu(t)\) with \(A\in L(\mathbb{R}^ n\)), \(b\in\mathbb{R}^ n\). If \(\text{Re} \lambda(A)<0\), \(y(0)\neq0\) then the function \(x:S=(-\infty,0]\ni t\mapsto e^{tA}y(0)\in\mathbb{R}^ n=Y\) is only locally Lipschitzian. Thus \(x\notin L(S,Y)\) and consequently there are problems with the initial domain \(D(x,Y)\). (2) The main result concerning the existence of optimal controls (Theorem 4, p. 82) seems to be a very weak result since for many typical problems of optimal control either the constraints cannot be interpreted as a compact subset of the space \(M(I,U)\) or \(F\) is no Lipschitzian mapping from \(D\) into \(L(I,\mathbb{R})\). These arguments show that several modifications and improvements are required to make this attempt more adequate to practical problems\}.