On parallelizing dual decomposition in stochastic integer programming

被引:44
作者
Lubin, Miles [1 ]
Martin, Kipp [2 ]
Petra, Cosmin G. [1 ]
Sandikci, Burhaneddin [2 ]
机构
[1] Argonne Natl Lab, Math & Comp Sci Div, Argonne, IL 60439 USA
[2] Univ Chicago, Booth Sch Business, Chicago, IL 60637 USA
关键词
Stochastic programming; Mixed-integer programming; Column generation; Dual decomposition; Parallel computing; Bundle methods; BUNDLE; ALGORITHMS;
D O I
10.1016/j.orl.2013.02.003
中图分类号
C93 [管理学]; O22 [运筹学];
学科分类号
070105 [运筹学与控制论]; 120117 [社会管理工程];
摘要
For stochastic mixed-integer programs, we revisit the dual decomposition algorithm of Came and Schultz from a computational perspective with the aim of its parallelization. We address an important bottleneck of parallel execution by identifying a formulation that permits the parallel solution of the master program by using structure-exploiting interior-point solvers. Our results demonstrate the potential for parallel speedup and the importance of regularization (stabilization) in the dual optimization. Load imbalance is identified as a remaining barrier to parallel scalability. (C) 2013 Elsevier B.V. All rights reserved.
引用
收藏
页码:252 / 258
页数:7
相关论文
共 25 条
[1]
SCIP: solving constraint integer programs [J].
Achterberg, Tobias .
MATHEMATICAL PROGRAMMING COMPUTATION, 2009, 1 (01) :1-41
[2]
Amdahl G. M., 1967, P APR 18 20 1967 SPR, P483, DOI [10.1145/1465482.1465560, DOI 10.1145/1465482.1465560]
[3]
Birge JR, 2011, SPRINGER SER OPER RE, P3, DOI 10.1007/978-1-4614-0237-4
[4]
Comparison of bundle and classical column generation [J].
Briant, O. ;
Lemarechal, C. ;
Meurdesoif, Ph. ;
Michel, S. ;
Perrot, N. ;
Vanderbeck, F. .
MATHEMATICAL PROGRAMMING, 2008, 113 (02) :299-344
[5]
Dual decomposition in stochastic integer programming [J].
Caroe, CC ;
Schultz, R .
OPERATIONS RESEARCH LETTERS, 1999, 24 (1-2) :37-45
[6]
MA57 - A code for the solution of sparse symmetric definite and indefinite systems [J].
Duff, IS .
ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE, 2004, 30 (02) :118-144
[7]
Solving two-stage stochastic programming problems with level decomposition [J].
Fabian, Csaba I. ;
Szoke, Zoltan .
COMPUTATIONAL MANAGEMENT SCIENCE, 2007, 4 (04) :313-353
[8]
Forrest John., CLP
[9]
About Lagrangian methods in integer optimization [J].
Frangioni, A .
ANNALS OF OPERATIONS RESEARCH, 2005, 139 (01) :163-193
[10]
Object-oriented software for quadratic programming [J].
Gertz, EM ;
Wright, SJ .
ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE, 2003, 29 (01) :58-81