Strict efficiency in set-valued optimization. (English) Zbl 1201.90179

The authors extend the notion of \(\phi\)-strict minimizer to set-valued maps, in such a way that the concept of a \(\phi\)-strict minimizer, presented in E. M. Bednarczuk [Optimization 53, No. 5–6, 455–474 (2004; Zbl 1153.90529)] for vector functions, as well as the notion of a minimizer of order one introduced for set-valued maps in G. P. Crespi, I. Ginchev and M. Rocca [Math. Methods Oper. Res. 63, No. 1, 87–106 (2006; Zbl 1103.90089)] are generalized in a unified manner.
A structure theorem is proved for a vector-valued function which states that a point is a \(\phi\)-strict minimizer for a vector valued function \(f\) over a set if and only if the point is a \(\phi\)-strict minimizer for a family of scalar functions and sets, each of these functions being the composition of \(f\) with a positive, continuous and linear functional, and the family of sets, a covering of the initial set.
Different kinds of strict minimizers for the set-valued problem are characterized through different kinds of strict minimizers for an associated scalarized problem. Finally, several optimality conditions are established for strict minimizers of order one. A characterization is given for a global minimizer through the radial derivative. Comparisons with other results are made, and some illustrative examples are provide.


90C29 Multi-objective and goal programming
90C46 Optimality conditions and duality in mathematical programming
90C31 Sensitivity, stability, parametric optimization
Full Text: DOI