Abstract
Motivated by nonlinear programming we study a process involving continuous adjustments of both primal and dual variables. Nonsmoothness in problem data or simple restrictions on variables make the velocity of the process possibly nonunique. The process obtained is shown to converge locally under strict monotonicity conditions. Examples are given in terms of augmented Lagrangians. The results are believed to provide a basis for the construction of algorithms.
Original language | English (US) |
---|---|
Pages (from-to) | 264-277 |
Number of pages | 14 |
Journal | Journal of Mathematical Analysis and Applications |
Volume | 141 |
Issue number | 1 |
DOIs | |
State | Published - Jul 1989 |
All Science Journal Classification (ASJC) codes
- Analysis
- Applied Mathematics