Qualification-Free Optimality Conditions for Convex Programs with Separable Inequality Constraints

We show that separable convex functions enjoy ε-subdifferential sum formula as well as the Fenchel duality without a regularity assumption, and establish that for convex programs with separable convex constraints a new partially asymptotic Lagrange multiplier conditions hold without a constraint qualification. Examples are given to illustrate the results