On Lagrange multipliers of the KKT system in Hilbert spaces

Abstract: In this paper we develop a new decomposition framework to deal with Lagrange multipliers of the Karush-Kuhn-Tucker (KKT) system of constrained optimization problems and variational inequalities in Hilbert spaces. It is different from existing frameworks based on separation theorems. We introduce the essential Lagrange multiplier and establish the basic theory of this new multiplier. The essential Lagrange multiplier poses essentially different existence results in finite and infinite-dimensional spaces. It can also be used to give an essential characterization of the convergence of multipliers generated by the classical augmented Lagrangian method. Our analysis reveals that the essential Lagrange multiplier is at the core of both theories and applications of Lagrange multipliers.