Polynomial worst-case iteration complexity of quasi-Newton primal-dual interior point algorithms for linear programming

被引:1
作者
Gondzio, Jacek [1 ,2 ]
Sobral, Francisco N. C. [3 ]
机构
[1] Univ Edinburgh, Sch Math, JCMB,Kings Bldg, Edinburgh EH9 3FD, Scotland
[2] Univ Edinburgh, Maxwell Inst Math Sci, JCMB,Kings Bldg, Edinburgh EH9 3FD, Scotland
[3] Univ Estadual Maringa, Dept Math, Ave Colombo 5790, BR-87020900 Maringa, Parana, Brazil
关键词
Quasi-Newton methods; Broyden update; Primal-dual Interior Point Methods; Polynomial worst-case iteration complexity; PRECONDITIONERS; SYSTEMS;
D O I
10.1007/s10589-024-00584-6
中图分类号
C93 [管理学]; O22 [运筹学];
学科分类号
070105 ; 12 ; 1201 ; 1202 ; 120202 ;
摘要
Quasi-Newton methods are well known techniques for large-scale numerical optimization. They use an approximation of the Hessian in optimization problems or the Jacobian in system of nonlinear equations. In the Interior Point context, quasi-Newton algorithms compute low-rank updates of the matrix associated with the Newton systems, instead of computing it from scratch at every iteration. In this work, we show that a simplified quasi-Newton primal-dual interior point algorithm for linear programming, which alternates between Newton and quasi-Newton iterations, enjoys polynomial worst-case iteration complexity. Feasible and infeasible cases of the algorithm are considered and the most common neighborhoods of the central path are analyzed. To the best of our knowledge, this is the first attempt to deliver polynomial worst-case iteration complexity bounds for these methods. Unsurprisingly, the worst-case complexity results obtained when quasi-Newton directions are used are worse than their counterparts when Newton directions are employed. However, quasi-Newton updates are very attractive for large-scale optimization problems where the cost of factorizing the matrices is much higher than the cost of solving linear systems.
引用
收藏
页码:649 / 681
页数:33
相关论文
共 20 条
[1]   UPDATING CONSTRAINT PRECONDITIONERS FOR KKT SYSTEMS IN QUADRATIC PROGRAMMING VIA LOW-RANK CORRECTIONS [J].
Bellavia, Stefania ;
De Simone, Valentina ;
di Serafino, Daniela ;
Morini, Benedetta .
SIAM JOURNAL ON OPTIMIZATION, 2015, 25 (03) :1787-1808
[2]   BFGS-like updates of constraint preconditioners for sequences of KKT linear systems in quadratic programming [J].
Bergamaschi, L. ;
De Simone, V. ;
di Serafino, D. ;
Martinez, A. .
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS, 2018, 25 (05)
[3]   Further development of multiple centrality correctors for interior point methods [J].
Colombo, Marco ;
Gondzio, Jacek .
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2008, 41 (03) :277-305
[4]   On mutual impact of numerical linear algebra and large-scale optimization with focus on interior point methods [J].
D'Apuzzo, Marco ;
De Simone, Valentina ;
di Serafino, Daniela .
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2010, 45 (02) :283-310
[5]   A VARIABLE-METRIC VARIANT OF THE KARMARKAR ALGORITHM FOR LINEAR-PROGRAMMING [J].
DENNIS, JE ;
MORSHEDI, AM ;
TURNER, K .
MATHEMATICAL PROGRAMMING, 1987, 39 (01) :1-20
[6]   A structured modified Newton approach for solving systems of nonlinear equations arising in interior-point methods for quadratic programming [J].
Ek, David ;
Forsgren, Anders .
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2023, 86 (01) :1-48
[7]   Quasi-Newton approaches to interior point methods for quadratic problems [J].
Gondzio, J. ;
Sobral, F. N. C. .
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS, 2019, 74 (01) :93-120
[8]   HOPDM (VERSION-2.12) - A FAST LP SOLVER BASED ON A PRIMAL-DUAL INTERIOR-POINT METHOD [J].
GONDZIO, J .
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 1995, 85 (01) :221-225
[9]   Interior point methods 25 years later [J].
Gondzio, Jacek .
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH, 2012, 218 (03) :587-601
[10]  
Gondzio J, 2009, COMPUT MANAG SCI, V6, P135, DOI 10.1007/s10287-008-0090-3