Vector optimization with variable domination structure: a unifying approach (Q6999380)

A constrained vector optimization problem \(F(x)\rightarrow \min\) s.t. \(x\in \Omega\) with set valued map \(F:X \rightrightarrows Y\), Asplund spaces \(X,Y\) and close set \(\Omega \subset X\) is considered. Two well-known types \(K,Q\) of non pointed cone mappings defining variable domination structures in \(Y\) are used, type I: \(K:X\rightrightarrows Y\) and type II: \(Q:Y\rightrightarrows Y\). The point \((\bar x,\bar y) \in \Omega \times Y \) is a local nondominated point of type I and type II if for all \(x \in U\cap\, \Omega\) for some neighborhood \(U\) of \(x\) it holds \((F(x)-\bar y)\cap (K(-x))\subseteq K(x)\) and \(\forall \, y\in F(x)\setminus \{\bar y \},\; \bar y\notin y+Q(y)\setminus (-Q(y))\cup \{0\}\), respectively. In suitable higher product spaces both types of problems are transformed into each other because of the use of non pointed order cone mappings and thus, necessary optimality conditions of both types are also transformed into each other. For deriving necessary conditions of optimality like subgradient KKT conditions the authors use at the one hand the extremality principle of Mordukhowich and at the other hand the incompatibility of the openness of special constructed sums of set valued maps at non dominated points for both types of problems. Together with the above mentioned transformations, the main results Theorem 4.2 for type I and Theorem 4.6 for type II are proven several times. All in all, the authors present in a very fine manner and with detailed proofs a unified theory to derive necessary conditions of optimality for constrained vector optimization problems with variable domination cone structure.