Hausdorff continuity of approximate solution maps to parametric primal and dual equilibrium problems (Q286211): Difference between revisions

For parametric equilibrium problems \[ \text{(EQ) \quad find} \;\overline{x}\in K(\lambda) \; \text{such that} \; f(\overline{x},y,\mu)\geq 0 \; \text{ for all} \; y\in K(\lambda) \] and their dual ones \[ \text{(dEQ) \quad find} \; \overline{x}\in K(\lambda) \; \text{such that} \; f(y,\overline{x},\mu)\geq 0 \; \text{ for all} \; y\in K(\lambda), \] where \(K:\Lambda\Rightarrow A\) is a multifunction with nonempty convex values, \(f: A \times A \times M \rightarrow \mathbb R\), \(X\) is a locally convex Hausdorff topological vector space, \(Y,Z\) are vector topological spaces, \(A \subseteq X, \Lambda \subseteq Y, M \subseteq Z \) are nonempty subsets, and \((\lambda, \mu) \in \Lambda \times M\), the authors define for \(\epsilon > 0\) \(\epsilon\)-approximate solution sets for (EQ) and (dEQ) as follows (and assume that they are nonempty): \[ \Pi(\epsilon, \lambda,\mu) = \{x\in K(\lambda): f(x,y,\mu)+ \epsilon \geq 0 \;\text{for all}\; y\in K(\lambda)\}, \] \[ \Pi^d(\epsilon, \lambda,\mu) = \{x\in K(\lambda): f(y,x,\mu)- \epsilon \leq 0 \;\text{for all}\; y\in K(\lambda)\}. \] Then, after having mentioned the relevance for practical problems and numerical methods, they prove theorems about Hausdorff continuity of \(\Pi\) and \(\Pi^d\) at \((\epsilon_0,\lambda_0,\mu_0) \in \mathbb {R}_+ \times\Lambda \times M,\) discuss carefully (with examples) the assumptions (esp., there are not only continuity ones but also concavity or (for (dEQ)) convexity conditions for \(f\) with respect to the variable \(x\)) and compare the proved theorems with existing ones. Finally, using the well-known paper of \textit{E. Blum} and \textit{W. Oettli} [Math. Stud. 63, No.1--4, 123--145 (1994; Zbl 0888.49007)], they apply the new Hausdorff continuity theorems to optimization problems, variational inequalities (also the Minty variational inequality appears) and fixed-point problems.