ϵ-Pareto Optimality for Nondifferentiable Multiobjective Programming via Penalty Function

AbstractNecessary and sufficient conditions without a constraint qualification for ϵ-Pareto optimality of multiobjective programming are derived. The necessary Kuhn–Tucker condition suggests the establishment of a Wolf-type duality theorem for nondifferentiable, convex, multiobjective minimization problems. The generalized ϵ-saddle point for Pareto optimality of the vector Lagrangian is studied