Generalized projection method for non-Lipschitz multivalued monotone variational inequalities (Q2390720)

A multivalued variational inequality whose cost operator \(F\) (a strongly monotone operator) is not assumed to be Lipschitz, is analysed. For solving of the proposed problem the authors generalize the projection method to the case of a strongly (not necessarily Lipschitz) variational inequality. At each iteration, at most one projection onto the feasible domain is needed and the search direction can be determined from any point in the image of the current iterate. It is shown how to embed the new method (the proposed algorithm) in a polyhedral outer approximation procedure. The projections by solving strongly convex quadratic programs with linear constraints are obtained. This is helpful (form a computation view point) when the convex set is not simple. Finally, the application of the proposed method in the framework of the proximal point method is discussed. The authors give the approximation rule and show how to use it in the presented algorithm to implement inexact proximal point methods for (not necessarily strongly) monotone multivalued variational inequalities whose cost operator is not assumed to be Lipschitz.