A decomposition-dualization approach for solving constrained convex minimization problems with applications to discretized obstacle problems (Q1113621): Difference between revisions

The author shows that if the constrained convex minimization problem can be transformed into a convex-additively decomposed and almost separable form and the conjugate functionals in the Fenchel-Rockafellar dual problem can be easily computed then the effective techniques of unconstrained optimization can be applied. By this approach it is possible to compute the conjugate functionals and their derivatives simultaneously and the method becomes efficiently applicable for numerical computations on high performance parallel computers. Finally, numerical tests for solving two concrete examples of discretized obstacle problems are given.