Epsilon-subgradient optimization techniques in convex programming and Lagrangian duality
The paper studies two algorithms for approximately solving convex programs when function values can be computed only approximately, but an \(\epsilon\)-subgradient is available. Such type of problems arise for instance if integer programming problems are handled by formulating Lagrangian dual problems. The second section gives a review about integer programming problems, in which \(\epsilon\)-optimal solutions of relaxed Lagrangian dual problems were be applied successfully in heuristic methods. The third and fourth section contain the theoretical basis of the algorithms to determine Lagrangian \(\epsilon\)-subgradients. In the first algorithm an inexact solution of a relaxed Lagrangian dual problem has to be determined, whereas the second algorithm applies cutting planes.
- Using the ?-subgradient method to solve the dual and the primal mathematical programming problems
- A Proximal Bundle Method with Approximate Subgradient Linearizations
- Lagrangian relaxations on networks by \(\varepsilon \)-subgradient methods
- Ergodic, primal convergence in dual subgradient schemes for convex programming
- scientific article; zbMATH DE number 4197761
This page was built for publication: Epsilon-subgradient optimization techniques in convex programming and Lagrangian duality
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1079495)