Asymptotic behaviour of optimal solutions of control problems governed by inclusions (Q1576799)

The author considers sequences of abstract optimal control problems of the form \[ \min\{J_h(y,u): A_h(y)\in B_h(u), y\in Y, u\in U\}, \] where \(Y\) is the space of states, \(U\) is the space of controls, \(J_h\) are the cost functionals, and the relations \(A_h(y)\in B_h(u)\) represent the state equations (or more generally inclusions). The operators \(A_h\) are assumed to \(G\)-converge to an operator \(A\), which is a rather weak assumption. On the multivalued mappings \(B_h\) two kinds of assumptions are made: i) a first case considers a quite strong convergence of \(B_h\) to \(B\), the so-called ``\(K\)-continuous convergence''; ii) in a second case the convergence of \(B_h\) to \(B\) is much weaker but the multivalued mappings \(B_h\) are assumed to be of local type and the cost functionals of integral type. In both cases the form of the limit optimal control problem is characterized. The paper generalizes to the case of multimappings a previous result obtained by \textit{G. Buttazzo} and \textit{L. Freddi} [Discrete Contin. Dyn. Syst. 1, No. 3, 401-420 (1995; Zbl 0886.49015)].