Optimal control for unstructured nonlinear differential-algebraic equations of arbitrary index (Q956591)

The paper concerns an optimal control problem of which constraint is given by a differential-algebraic equation \(F(t,x(t),u(t),\dot{x}(t))=0\). Here, \(F\), \(x\), and \(u\) take values in finite dimensional spaces. Moreover, \(F\) is sufficiently smooth so that, through a suitable splitting \(x=(x_1,x_2)\), the original constraint reduces to one made up of a differential equation \(\dot{x}_1(t)={\mathcal L}(t,x_1(t),u(t),\dot{u}(t))\) and an algebraic equation \(x_2(t)={\mathcal R}(t,x_1(t),u(t))\) [see \textit{P.~Kunkel} and \textit{V.~Mehrmann}, Differential-algebraic equations. Analysis and numerical solutions. Zürich: EMS Publishing House (2006; Zbl 1095.34004)]. The structure of the reduced constraint often implies that actually \(\dot{u}\) is not present. This is the case in which there are derived necessary optimality conditions as well as a Pontryagin maximum principle [cf.~\textit{T.~Roubiček} and \textit{M.~Valášek}, J. Math. Anal. Appl. 269, 616--641 (2002; Zbl 1020.49001)]. It is mentioned also the virtual impossibility to formulate the results in terms of the original problem. A numerical example is presented.