Dual Bregman proximal methods for large-scale 0-1 problems

We describe an extension of the Bregman proximal method for convex programming, employing B-functions as generalizations of Bregman functions that cover more applications. We allow inexact subproblem solutions, increasing their accuracy successively to retain global convergence. Our framework is applied to Lagrangian relaxations of large-scale set covering problems that arise in airline crew scheduling.