Mathematics & Theoretical Foundations
Parastoo Heiatian Naeini; Ahmad Rezaee
Abstract
We consider a finite-dimensional multiobjective optimization problem with locally Lipschitz objective functions and finitely many locally Lipschitz inequality constraints. We formulate an active-set Slater test in which σxˆ is selected from a structurally restricted admissible class or selection ...
Read More
We consider a finite-dimensional multiobjective optimization problem with locally Lipschitz objective functions and finitely many locally Lipschitz inequality constraints. We formulate an active-set Slater test in which σxˆ is selected from a structurally restricted admissible class or selection rule Σ(ψ, xˆ) prescribed before testing the qualification, rather than constructed from an MFCQ direction. Pointwise (σxˆ, rj)-invex inequalities with rj ≥ 0 imply the Clarke–Mangasarian–Fromovitz constraint qualification (ClarkeMFCQ). Standard maximum-function and normal-cone arguments then yield normalized Clarke–KKT multipliers at locally weakly efficient solutions, positive objective multipliers for positive weighted-sum solutions, and a perturbed-KKT inclusion at locally isolated efficient solutions. If Σ(ψ, xˆ) contains every locally Lipschitz map, the test is equivalent to Clarke-MFCQ. A nonsmooth nonconvex example illustrates the assumptions and multiplier inclusions.
Control Theory & Systems
Ali Asghar Hojatifard; Nader Kanzi; Shahriar Farahmand Rad
Abstract
This paper aims to establish first-order necessary optimality conditions for non-smooth multi-objective generalized semi-infinite programming problems. These problems involve inequality constraints whose index set depends on the decision vector, and all emerging functions are assumed ...
Read More
This paper aims to establish first-order necessary optimality conditions for non-smooth multi-objective generalized semi-infinite programming problems. These problems involve inequality constraints whose index set depends on the decision vector, and all emerging functions are assumed to be locally Lipschitz. We introduce a new constraint qualification for these problems. Building upon this qualification, we derive an upper estimate for the Clarke sub-differential of the value function of the problem. Furthermore, we demonstrate the necessary optimality conditions for properly efficient solutions to the problem.