Intersection theorems for weak \(KKM\) set-valued mappings in the finite dimensional setting (Q2319622)

It is well known that set-valued KKM-mappings \(S:X \rightrightarrows Y\) between sets of vector spaces are used to obtain intersection theorems, which offer broad application possibilities in optimization and nonlinear analysis. ``Weak-KKM'' in the title indicates that there is a second mapping \(T\) in the definition of KKM-mappings, which widens considerably the range of applicability with respect to solution conditions, assumptions for mappings and spaces. For existing results, see, e.g., [\textit{R. P. Agarwal} et al., J. Optim. Theory Appl. 179, No. 3, 761--777 (2018; Zbl 1440.47045)]. In the present paper, one finds new results (intersection theorems) in the finite dimensional setting (\(X\) is compact and convex \(\in \mathbb{R}^n\)), replacing the closedness conditions in the two intersection theorems in [loc. cit.] by open ones. For example, the set \((x,u)\in X \times X: T(x)\cap S(u) \not= \emptyset \) is now assumed to be open in \(X \times X\) and \(S\) has now especially an open graph. For the proofs, one uses the Kakutani fixed point problem for the first case, and to \(S\) having an open graph belongs that \(T\) is lower semicontinuous. Chapter~4 of the paper deals with applications of the new intersection theorems. To get an impression of the results, let us consider Theorem 4.2. There we have \(X\) a nonempty, compact and convex set in \(\mathbb{R}^n\), \(Y\) nonempty and convex in a topological vector space, \(T: X \rightrightarrows Y\) an lsc set-valued mapping with nonempty convex values, \(f:X \times Y \)a real lsc function, such that for all \(y \in Y\), \(f(\cdot,y)\) is quasiconvex and for all \(x \in X,\) \(f(x,\cdot)\) is quasiconcave. Then \[ \min_{x\in X} \sup_{y\in T(x)}f(x,y) \leq \sup_{y \in T(X)}\min_{x\in X} f(x,y). \] In the proof, a mapping \(S\) is defined having open graph, convex values, convex cofibers and is weak-KKM with respect to \(T\). Such a mapping is also constructed in Chapter 5, which gives applications to (three) set-valued equilibrium problems of the form: Find \(x_0 \in X\) and \(y_0 \in P(x_0)\) such that \(F(x_0,y_0,u) \subseteq \operatorname{int}C\) for all \(u \in X\), where \(C\) is a closed cone with nonempty interior. An interesting point in the proof is that \textit{D. T. Luc}'s method [J. Optim. Theory Appl. 138, No. 1, 65--76 (2008; Zbl 1148.49009)] is used to study first the existence of a solution for a variational relation problem.