The existence problem of optimal control for nonlinear processes (Q801286)

This well structured paper is concerned with the existence of optimal controls for processes governed by general nonlinear systems of ordinary differential equations. The authors extend already known results by \textit{E. B. Lee} and \textit{L. Markus} [Arch. Rat. Mech. Analysis 8, 36-58 (1961; Zbl 0099.087)] and by \textit{L. D. Berkovitz} [Optimal control theory (1974; Zbl 0295.49001)] using more general assumptions on the control system and on the class of admissible controls. Therefore this class may contain functions which are frequently used in engineering, especially bang-bang controls. Following this aim, some classes of controls are carefully investigated: Lipschitz functions, functions of bounded variation, functions which satisfy a Lipschitz condition in the mean, piecewise constant functions with a finite number of switching points. A counterexample is given in order to show that a result on approximated optimal controls - this result is only a theoretical one - given in the above cited paper of Lee and Markus is not true. A correct result is established essentially by taking BV instead of \(C^ 1\) as the class of approximating controls.