Solving quasi-variational inequalities via their KKT conditions (Q2452385)

The authors propose a totally different approach to the solution of quasi-variational inequalities (QVIs). Assuming that the feasible set mapping \(K(\cdot)\) is described by a finite number of parametric inequalities, the Karush-Kuhn-Tucker (KKT) conditions of the QVI is reformulated then as a system of constrained equations, and then an interior-point method is applied. Main result: The authors propose and analyze a globally convergent algorithm for the solution of a finite-dimensional QVI. The convergence is established for a wide array of different classes of problems including QVIs with the feasible set given by moving sets, linear systems with variable right-hand sides, box constraints with variable bounds, and bilinear constraints. These results surpass by far existing convergence analyses, the latter all having a somewhat limited scope. The results in this paper constitute an important step towards the development of theoretically reliable and numerically efficient methods for the solution of QVIs.