Abstract
This paper provides for the first time some computable smoothing functions for variational inequality problems with general constraints. This paper proposes also a new version of the smoothing Newton method and establishes its global and superlinear (quadratic) convergence under conditions weaker than those previously used in the literature. These are achieved by introducing a general definition for smoothing functions, which include almost all the existing smoothing functions as special cases.
| Original language | English |
|---|---|
| Pages (from-to) | 121-147 |
| Number of pages | 27 |
| Journal | Journal of Optimization Theory and Applications |
| Volume | 113 |
| Issue number | 1 |
| DOIs | |
| Publication status | Published - 1 Apr 2002 |
Keywords
- computable smoothing functions
- quadratic convergence
- smoothing Newton methods
- Variational inequality problems
ASJC Scopus subject areas
- Control and Optimization
- Management Science and Operations Research
- Applied Mathematics