Right-hand side decomposition for variational inequalities (Q2250070): Difference between revisions

The author considers a general class of variational inequality problems with functional binding constraints in a finite-dimensional space setting. The cost mapping need not be the gradient of any function, i.e. the problem is not equivalent to an optimization problem. The author suggests to use the so-called right-hand side decomposition technique and to transform it into a collection of small dimensional variational inequalities. The master problem is then a set-valued variational inequality whose solution gives us a solution of the initial problem. A general iterative method, which falls into the combined relaxation framework, is suggested for the master problem. It appears convergent under monotonicity assumptions. Moreover, a primal-dual regularization method for partial problems is also presented. It enables one to utilize single-valued approximations for the cost mapping of the master problem within the same decomposition approach, which leads to essentially simpler solution methods.