A duality theory for non-convex problems in the calculus of variations (Q724314)

Announced in a recent short note of the authors, a novel duality theory for nonconvex minimization variational problems with possibly mixed Dirichlet and Neumann boundary conditions is proposed, aiming to become a suitable alternative for the established duality theory in convex analysis. The dual problem is formulated as a maximization linear programming problem, however in its development the Fenchel conjugate function plays an important role. The main result of the paper shows that there is no duality gap for the considered primal-dual pair of variational problems and from here necessary and sufficient optimality conditions for them are derived. For numerical implementations a reformulation as a min-max result of the main duality principle is provided. An application to a celebrated free boundary problem is presented as well, while some proofs left for the final section close the paper.