Isolated efficiency in nonsmooth semi-infinite multi-objective programming

ABSTRACT

In this paper, isolated efficient solutions of a given nonsmooth Multi-Objective Semi-Infinite Programming problem (MOSIP) are studied. Two new Data Qualifications (DQs) are introduced and it is shown that these DQs are, to a large extent, weaker than already known Constraint Qualifications (CQs). The relationships between isolated efficiency and some relevant notions existing in the literature, including robustness, are established. Various necessary and sufficient conditions for characterizing isolated efficient solutions of a general problem are derived. It is done invoking the tangent cones, the normal cones, the generalized directional derivatives, and some gap functions. Using these characterizations, the (strongly) perturbed Karush-Kuhn-Tucker (KKT) optimality conditions for MOSIP are analyzed. Furthermore, it is shown that each isolated efficient solution is a Geoffrion properly efficient solution under appropriate assumptions. Moreover, Kuhn-Tucker (KT) and Klinger properly efficient solutions for a nonsmooth MOSIP are defined and it is proved that each isolated efficient solution is a KT properly efficient solution in general, and a Klinger properly efficient solution under a DQ. Finally, in the last section, the largest isolated efficiency constant for a given isolated efficient solution is determined.

Keywords:

• Multi-objective programming
• semi-infinite programming
• nonsmooth optimization
• isolated efficient solution
• robustness
• gap function

Acknowledgements

The authors would like to express their gratitude to the editor in chief of Optimization, handling editor, and two anonymous referees for their helpful comments on the earlier versions of the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was in part supported by a grant from the Iran National Science Foundation (INSF) (No. 96005247).