Convergence analysis of a projection algorithm for variational inequality problems
- 63 Downloads
In this paper, we propose a projection based Newton-type algorithm for solving the variational inequality problems. A comprehensive study is conducted to analyze both global and local convergence properties of the algorithm. In particular, the algorithm is shown to be of superlinear convergence when the solution is a regular point. In addition, when the Jacobian matrix of the underlying function is positive definite at the solution or the solution is a non-degenerate point, the algorithm still possesses its superlinear convergence. Compared to the relevant projection algorithms in literature, the proposed algorithm is of remarkable advantages in terms of its generalization and favorable convergence properties under relaxed assumptions.
KeywordsVariational inequality problem Global convergence Local convergence rate Regular solution
Mathematics Subject Classification90C33 65K15 49M15
The authors thank the anonymous reviewers for the insightful and constructive comments and suggestions, which helped to improve the presentation of the paper.
- 18.Solodov, M.V.: A class of globally convergent algorithms for pseudomonotone variational inequalities, complementarity: applications, algorithms and extension. In: Ferrs, M. C., Mangasarian, O. L., Pang, J. S. (eds.) Applied Optimization, pp. 297–315. 50, Kluwer Academic Publishers, Dordrecht (2001)CrossRefGoogle Scholar