On Parallelizing Dual Decomposition in Stochastic Integer Programming

25 Pages Posted: 24 Oct 2012

See all articles by MIles Lubin

MIles Lubin

Argonne National Laboratory

Kipp Martin

University of Chicago - Booth School of Business

Cosmin Petra

Argonne National Laboratory

Burhaneddin Sandikci

University of Chicago - Booth School of Business

Date Written: October 23, 2012

Abstract

For stochastic mixed-integer programs, we revisit the dual decomposition algorithm of Caroe 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.

Keywords: stochastic programming, mixed-integer programming, column generation, dual decomposition, parallel computing, bundle methods

Suggested Citation

Lubin, MIles and Martin, Kipp and Petra, Cosmin and Sandikci, Burhaneddin, On Parallelizing Dual Decomposition in Stochastic Integer Programming (October 23, 2012). Chicago Booth Research Paper No. 12-50. Available at SSRN: https://ssrn.com/abstract=2165991 or http://dx.doi.org/10.2139/ssrn.2165991

MIles Lubin

Argonne National Laboratory ( email )

9700 S. Cass Avenue
Argonne, IL 60439
United States

Kipp Martin

University of Chicago - Booth School of Business ( email )

5807 South Woodlawn Avenue
Chicago, IL 60637
United States

Cosmin Petra

Argonne National Laboratory ( email )

9700 S. Cass Avenue
Argonne, IL 60439
United States

Burhaneddin Sandikci (Contact Author)

University of Chicago - Booth School of Business ( email )

5807 S. Woodlawn Avenue
Chicago, IL 60637
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
79
Abstract Views
1,222
rank
305,836
PlumX Metrics