Semismooth Newton method for quadratic programs with bound constraints

被引：3

作者：

Daryina, A. N. ^{[1
]}

Izmailov, A. F. ^{[2
]}

机构：

[1] Russian Acad Sci, Dorodnicyn Comp Ctr, Moscow 119333, Russia

[2] Moscow MV Lomonosov State Univ, Fac Computat Math & Cybernet, Moscow 119992, Russia

来源：

COMPUTATIONAL MATHEMATICS AND MATHEMATICAL PHYSICS | 2009年 / 49卷 / 10期

关键词：

quadratic program; variational inequality; mixed complementarity problem; complementarity function; natural residual; semismooth Newton method; active-set method; projected gradient method; FINITE TERMINATION;

D O I：

10.1134/S0965542509100066

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

Convex quadratic programs with bound constraints are proposed to be solved by applying a semismooth Newton method to the corresponding variational inequality. Computational experiments demonstrate that, for strongly convex problems, this approach can be considerably more efficient than more traditional approaches.

引用

页码：1706 / 1716

页数：11

共 13 条

[1] [Anonymous], COMPUT MATH MATH PHY
[2] [Anonymous], 2000, Numerical Optimization
[3] Billups S. C., 1995, THESIS MADISON
[4] Clarke FH., 1983, OPTIMIZATION NONSMOO
[5] Projected Barzilai-Borwein methods for large-scale box-constrained quadratic programming
Dai, YH
Fletcher, R
[J]. NUMERISCHE MATHEMATIK, 2005, 100 (01) : 21 - 47
[6] Minimizing quadratic functions subject to bound constraints with the rate of convergence and finite termination
Dostál, Z
Schöberl, J
[J]. COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2005, 30 (01) : 23 - 43
[7] On finite termination of an iterative method for linear complementarity problems
Fischer, A
Kanzow, C
[J]. MATHEMATICAL PROGRAMMING, 1996, 74 (03) : 279 - 292
[8] FRIEDLANDER A, 1994, SIAM J OPTIMIZ, V4, P204
[9] On a semi-smooth Newton method and its globalization
Ito, Kazufumi
Kunisch, Karl
[J]. MATHEMATICAL PROGRAMMING, 2009, 118 (02) : 347 - 370
[10] IZMAILOV AF, 2003, NUMERICAL OPTIMIZATI

← 1 2 →