scientific article; zbMATH DE number 3979888

From MaRDI portal
Publication:3744786





[For the entire collection see Zbl 0583.00018.] In classical optimal control problems, for any control variable v the state equation admits a unique solution \(y(v)\). In the paper the following situation is considered: the state equation either admits an infinite number of solutions or admits a solution only for control variables v of a very special nature and the solution \(y(v)\) does not depend continuously on v. In such cases, the optimal control problem is formulated as follows: find \((u,y)\in {\mathcal U}_{ad}\times Y\) such that \(J(u,y)\leq J(v,z)\) for all \(v\in {\mathcal U}_{ad}\), (v,z) subject to the state equation. The aim of the paper is, in a number of situations, to give necessary optimality conditions. The following situations are studied: 1. Elliptic systems with an infinite number of solutions (Multi-state systems). 2. Non-well-set problems: i) Cauchy problem for the operator \(-\Delta\) with boundary conditions \(z=v_ 0\) and \(\partial z/\partial \nu =v_ 1\) on a portion of the boundary. ii) backward heat equation. 3. Problems with partial (or incomplete) information: i) parabolic case in which the initial condition is substituted by the information \(z(t_ 1)\in K.\) ii) elliptic case with partial information on z. iii) hyperbolic case \((z-\Delta z=0)\) with partial information on \(z\) \((z(b,t)\in K\), \(K\) closed convex subset of \(L^ 2(0,T))\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3744786)