Abstract
Primal, dual and saddle-point characterizations of optimality are given for convex programming in the general case (nondifferentiable functions and no constraint qualification).
Research partly supported by NSF.
Research partly supported by NSF.
Notes
Research partly supported by NSF.