Vector optimization. Set-valued and variational analysis. (Q2484094): Difference between revisions

This book presents a study of vector optimization with a strong emphasis on set-valued and variational analysis. Relationships between vector and set-valued optimization and variational analysis (vector variational inequalities and vector variational principles) are the major topics covered. In Chapter 1 some basic definitions, notations and results needed for the main parts of the book are presented. These concern vector optimization (ordering relations, convex cones, minimal points, and minimal solutions), set-valued analysis (selection and fixed point theorems, properties of set-valued functions) and convex as well as generalized convex functions. The nonlinear scalarization function \(\xi_{ea}(y)=\min\{t \in \mathbb R : y \in a + te-C\}\) is presented and extended for a variable domination structure. The topic of Chapter 2 are three types of vector optimization problems, namely with fixed and variable domination structure and with set-valued functions. Scalarization results and characterizations of solutions with Lagrange multipliers are proved for a fixed domination structure. Some of these results are extended to the case of variable domination structure. Then characterizations of Benson proper minimal solutions by scalarization, Lagrange multipliers, saddle points and duality are derived. Next, the authors investigate the continuity of solutions if constraints \(g(x,t) \leq b(t)\) depend on \(t \in T\). For the analysis of set-valued vector optimization problems the contingent epi-derivative is introduced. The most general results are necessary and sufficient conditions for weakly nondominated-like minimal solution pairs for these problems. The final sections of this chapter cover augmented Lagrange duality and penalty function approaches for real-valued vector optimization problems of the form \(\inf_{x \in \mathbb R^n} f(x)\). Using a nonlinear Lagrange approach weak and strong duality without convexity are shown. The relationship between the vector optimization problem and its dual are investigated in some detail. Exact penalization results using a nonlinear penalty function conclude the chapter. Chapter 3 is concerned with vector variational inequalities (VVI). The presentation starts with the study of relationships between VVI and vector optimization problems (VOP): Weakly minimal solutions of VOP solve weak VVIs and the converse is true under \(C\)-convexity. A characterization of a minimal solution of \(\min_{x \in K} f(x)\) as solution of a so-called Minty VVI is given and it is shown that a VVI is a necessary optimality condition for Geoffrion proper minimal solutions. Existence of solutions for weak VVIs is investigated. Then inverse VVIs are defined and their equivalence with other VVIs proved. The gap function is defined and a general VVI and its gap function are introduced. Since the minimization of the gap function is an approach to solve VIs, the convexity and differentiability of gap functions is investigated. Solutions for set-valued VVIs are developed using the concept of selection of a set-valued function. Further on it is established that under suitable conditions the solution set of generalized set-valued quasi variational inequalities is closed and upper semi-continuous. The existence of solutions for two other more general variational inequalities is proved. Following these results vector equilibrium problems are introduced, with certain vector optimization problems, vector variational inequalities, and vector complementarity problems as special cases. The existence of solutions of two types of vector equilibrium problems is proved. The final sections in this chapter are devoted to vector complementarity problems and existence results as well as results on relationships with VOPs and VVIs are shown. Chapter 4 on vector variational principles commences with the statement of Ekeland's variational principle. Vector variational principles by Nemeth, Tammer, and Isac are presented and a unified variational principle for vector-valued functions that includes them is demonstrated. The concept of approximate optimal solution for set-valued functions is introduced and conditions for their existence are proved. This leads to the statement of a variational principle for set-valued functions. In the next part of the chapter vector variants of the \textit{Drop Theorem}, \textit{Petal Theorem}, and \textit{Caristi-Kirk Fixed Point Theorem} are derived and their equivalence to vector variational principles shown. The next section extends these result to the set-valued case. In the last two sections the well-posedness of vector and set-valued optimization problems is studied. Characterizations and criteria for well-posedness are proved and vector variational principles are applied to obtain sufficient conditions. In Chapter 5 various minimax theorems for set-valued and vector-valued functions are proved. Finally, Chapter 6 is about vector network equilibrium models. Results relating (weak) vector equilibrium flows to parametric equilibrium flows are shown. The network vector optimization problem is introduced and necessary and sufficient conditions are given in terms of vector variational inequalities. This analysis is extended to dynamic vector network equilibrium problems.