Augmented Lagrangian methods under the constant positive linear dependence constraint qualification

被引：135

作者：

Andreani, R.

Birgin, E. G.

Martinez, J. M.

Schuverdt, M. L.

机构：

[1] Univ Sao Paulo, IME, Dept Comp Sci, BR-05508090 Sao Paulo, Brazil

[2] Univ Estadual Campinas, IMECC, Dept Appl Math, BR-13081970 Campinas, SP, Brazil

来源：

MATHEMATICAL PROGRAMMING | 2008年 / 111卷 / 1-2期

基金：

巴西圣保罗研究基金会;

关键词：

nonlinear programming; augmented Lagrangian methods; KKT systems; numerical experiments;

D O I：

10.1007/s10107-006-0077-1

中图分类号：

TP31 [计算机软件];

学科分类号：

081202 ; 0835 ;

摘要：

Two Augmented Lagrangian algorithms for solving KKT systems are introduced. The algorithms differ in the way in which penalty parameters are updated. Possibly infeasible accumulation points are characterized. It is proved that feasible limit points that satisfy the Constant Positive Linear Dependence constraint qualification are KKT solutions. Boundedness of the penalty parameters is proved under suitable assumptions. Numerical experiments are presented.

引用

页码：5 / 32

页数：28

共 34 条

[1] On the relation between constant positive linear dependence condition and quasinormality constraint qualification [J].