Schemes of involving dual variables in inverse barrier functions for problems of linear and convex programming

被引:1
|
作者
Popov, L. D. [1 ]
机构
[1] Russian Acad Sci, Inst Math & Mech, Ural Div, Ekaterinburg 620219, Russia
关键词
mathematical programming; interior penalty methods; barrier functions; Lagrange multipliers; numerical methods;
D O I
10.1134/S0081543809060169
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
A new scheme of the method of inverse barrier functions is proposed for problems of linear and convex programming. The scheme is based on the idea of a parametric shifting of the constraints of the original problem, similarly to what was done in the method of a modified Lagrange function for the usual quadratic penalty function. The description of the method, the proof of its convergence, and the results of numerical experiments are presented.
引用
收藏
页码:S205 / S217
页数:13
相关论文
共 50 条