Abstract
We present a scalable branching method for the dual decomposition of stochastic mixed-integer programming. Our new branching method is based on the branching method proposed by Carøe and Schultz that creates branching disjunctions on first-stage variables only. We propose improvements to the process for creating branching disjunctions, including (1) branching on the optimal solutions of the Dantzig–Wolfe reformulation of the restricted master problem and (2) using a more comprehensive (yet simple) measure for the dispersions associated with subproblem solution infeasibility. We prove that the proposed branching process leads to an algorithm that terminates finitely, and we provide conditions under which globally optimal solutions can be identified after termination. We have implemented our new branching method, as well as the Carøe–Schultz method and a branch-and-price method, in the open-source software package DSP. Using SIPLIB test instances, we present extensive numerical results to demonstrate that the proposed branching method significantly reduces the number of node subproblems and solution times.
Similar content being viewed by others
References
Ahmed, S.: A scenario decomposition algorithm for 0–1 stochastic programs. Oper. Res. Lett. 41(6), 565–569 (2013)
Ahmed, S., Tawarmalani, M., Sahinidis, N.: A finite branch-and-bound algorithm for two-stage stochastic integer programs. Math. Program. 100(2), 355–377 (2004). https://doi.org/10.1007/s10107-003-0475-6
Amor, H.B., Desrosiers, J., Frangioni, A.: On the choice of explicit stabilizing terms in column generation. Discret. Appl. Math. 157(6), 1167–1184 (2009)
Atakan, S., Sen, S.: A progressive hedging based branch-and-bound algorithm for mixed-integer stochastic programs. Comput. Manag. Sci. pp. 1–40 (2018)
Berthold, T.: Primal heuristics for mixed integer programs. Ph.D. thesis, Technischen Universität Berlin (2006)
Birge, J.R., Louveaux, F.: Introduction to Stochastic Programming. Springer, Berlin (2011)
Boland, N., Christiansen, J., Dandurand, B., Eberhard, A., Linderoth, J., Luedtke, J., Oliveira, F.: Progressive hedging with a Frank-Wolfe based method for computing stochastic mixed-integer programming Lagrangian dual bounds. SIAM J. Optim. 28(2), 1312–1336 (2018)
Boland, N., Christiansen, J., Dandurand, B., Eberhard, A., Oliveira, F.: A parallelizable augmented Lagrangian method applied to large-scale non-convex-constrained optimization problems. Math. Programm. (2018). https://doi.org/10.1007/s10107-018-1253-9
Carøe, C.C., Schultz, R.: Dual decomposition in stochastic integer programming. Oper. Res. Lett. 24(1), 37–45 (1999)
de Oliveira, W., Sagastizábal, C.: Bundle methods in the XXIst century: a bird’s-eye view. Pesquisa Operacional 34, 647–670 (2014)
Elhedhli, S., Naoum-Sawaya, J.: Improved branching disjunctions for branch-and-bound: an analytic center approach. Eur. J. Oper. Res. 247(1), 37–45 (2015)
Feltenmark, S., Kiwiel, K.: Dual applications of proximal bundle methods, including Lagrangian relaxation of nonconvex problems. SIAM J. Optim. 10(3), 697–721 (2000)
Gamrath, G., Lübbecke, M.E.: Experiments with a generic Dantzig–Wolfe decomposition for integer programs. In: International Symposium on Experimental Algorithms. Springer, pp. 239–252 (2010)
Kim, K., Anitescu, M., Zavala, V.M.: An asynchronous decomposition algorithm for security constrained unit commitment under contingency events. In: 2018 Power Systems Computation Conference (PSCC). IEEE, pp. 1–8 (2018)
Kim, K., Botterud, A., Qiu, F.: Temporal decomposition for improved unit commitment in power system production cost modeling. IEEE Trans. Power Syst. (2018)
Kim, K., Mehrotra, S.: A two-stage stochastic integer programming approach to integrated staffing and scheduling with application to nurse management. Oper. Res. 63(6), 1431–1451 (2015)
Kim, K., Petra, C.G., Zavala, V.M.: An asynchronous bundle-trust-region method for dual decomposition of stochastic mixed-integer programming. SIAM J. Optim. 29(1), 318–342 (2019)
Kim, K., Zavala, V.M.: Algorithmic innovations and software for the dual decomposition method applied to stochastic mixed-integer programs. Math. Programm. Comput. pp. 1–42 (2017)
Kirst, P., Stein, O., Steuermann, P.: Deterministic upper bounds for spatial branch-and-bound methods in global minimization with nonconvex constraints. TOP 23(2), 591–616 (2015)
Kiwiel, K.C.: Proximity control in bundle methods for convex nondifferentiable minimization. Math. Program. 46(1–3), 105–122 (1990)
Kiwiel, K.C.: Approximations in proximal bundle methods and decomposition of convex programs. J. Optim. Theory Appl. 84(3), 529–548 (1995)
Kiwiel, K.C., Lemaréchal, C.: An inexact bundle variant suited to column generation. Math. Program. 118(1), 177–206 (2009)
Li, C., Grossmann, I.: An improved L-shaped method for two-stage convex 0–1 mixed integer nonlinear stochastic programs. Comput. Chem. Eng. 112, 165–179 (2018). https://doi.org/10.1016/j.compchemeng.2018.01.017
Li, C., Grossmann, I.: A finite \(\epsilon \)-convergence algorithm for two-stage stochastic convex nonlinear programs with mixed-binary first and second-stage variables. J. Global Optim. 75(4), 921–947 (2019)
Li, C., Grossmann, I.: A generalized Benders decomposition-based branch and cut algorithm for two-stage stochastic programs with nonconvex constraints and mixed-binary first and second stage variables. J. Global Optim. 75(2), 247–272 (2019)
Lubin, M., Martin, K., Petra, C., Sandıkçı, B.: On parallelizing dual decomposition in stochastic integer programming. Oper. Res. Lett. 41(3), 252–258 (2013)
Lulli, G., Sen, S.: A branch-and-price algorithm for multistage stochastic integer programming with application to stochastic batch-sizing problems. Manag. Sci. 50(6), 786–796 (2004)
Mahajan, A., Ralphs, T.K.: Experiments with branching using general disjunctions. In: Operations Research and Cyber-Infrastructure. Springer, pp. 101–118 (2009)
McMullen, P.: The maximum number of faces of a convex polytope. Mathematika XVII, 179–184 (1970)
McMullen, P., Shephard, G.: Convex Polytopes and the Upper Bound Conjecture. Cambridge University Press, Cambridge (1971)
Munguía, L.M., Oxberry, G., Rajan, D.: PIPS-SBB: a parallel distributed-memory branch-and-bound algorithm for stochastic mixed-integer programs. In: Parallel and Distributed Processing Symposium Workshops, 2016 IEEE International. IEEE, pp. 730–739 (2016)
Nemhauser, G., Wolsey, L.: Integer and Combinatorial Optimization. Wiley-Interscience, New York (1988)
Ntaimo, L.: Decomposition algorithms for stochastic combinatorial optimization: computational experiments and extensions. Ph.D. thesis, The University of Arizona (2004)
Ntaimo, L., Sen, S.: The million-variable “march” for stochastic combinatorial optimization. J. Glob. Optim. 32(3), 385–400 (2005). https://doi.org/10.1007/s10898-004-5910-6
Ogbe, E., Li, X.: A new cross decomposition method for stochastic mixed-integer linear programming. Eur. J. Oper. Res. 256(2), 487–499 (2017)
Oliveira, W., Sagastizábal, C., Scheimberg, S.: Inexact bundle methods for two-stage stochastic programming. SIAM J. Optim. 21(2), 517–544 (2011)
Qi, Y., Sen, S.: The ancestral Benders’ cutting plane algorithm with multi-term disjunctions for mixed-integer recourse decisions in stochastic programming. Math. Program. 161(1), 193–235 (2017)
Ralphs, T.K., Galati, M.V.: Decomposition in integer linear programming. In: Integer Programming. CRC Press, pp. 73–126 (2005)
Romeijnders, W., Schultz, R., van der Vlerk, M., Klein Haneveld, W.: A convex approximation for two-stage mixed-integer recourse models with a uniform error bound. SIAM J. Optim. 26(1), 426–447 (2016)
Ryan, K., Rajan, D., Ahmed, S.: Scenario decomposition for 0-1 stochastic programs: improvements and asynchronous implementation. In: Parallel and Distributed Processing Symposium Workshops, 2016 IEEE International. IEEE, pp. 722–729 (2016)
Sherali, H., Zhu, X.: On solving discrete two-stage stochastic programs having mixed-integer first- and second-stage variables. Math. Program. 108(2), 597–616 (2006)
Still, C., Westerlund, T.: Solving convex MINLP optimization problems using a sequential cutting plane algorithm. Comput. Optim. Appl. 34(1), 63–83 (2006)
Tavaslioglu, O., Prokopyev, O., Schaefer, A.: Solving stochastic and bilevel mixed-integer programs via a generalized value function. Oper. Res. 67(6), 1659–1677 (2019)
Trespalacios, F., Grossmann, I.E.: Lagrangean relaxation of the hull-reformulation of linear generalized disjunctive programs and its use in disjunctive branch and bound. Eur. J. Oper. Res. 253(2), 314–327 (2016)
Xu, Y., Ralphs, T.K., Ladányi, L., Saltzman, M.J.: Alps: a framework for implementing parallel tree search algorithms. In: The Next Wave in Computing, Optimization, and Decision Technologies. Springer, pp. 319–334 (2005)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
All authors declare no conflict of interest.
Code availability statement
The software code and experiment scripts are available to reproduce the results published in the manuscript at https://doi.org/10.5281/zenodo.1470838.
Data availability statement
The algorithms developed in the paper are also available in the most recent version of the software package DSP available in https://github.com/Argonne-National-Laboratory/DSP.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
We report the numerical results from the different branching methods for solving smkp instances. This set of instances is challenging for the dual decomposition because the instances have a larger number of constraints and variables in the first stage than those in the second stage. This makes each iteration of the dual decomposition take a significant amount of time.
To circumvent the issue, we set a 300-s time limit for each scenario subproblem solution and use a (possibly suboptimal) feasible solution to generate the inequalities (5b) for the Lagrangian dual problem. We emphasize that the cuts generated at a suboptimal feasible solution are valid. Moreover, we use 4-h time limit for each instance (Table 5).
We report the numerical results without any heuristic in Table 6. The BNP and CS+DW methods found the global optimal solutions at the root node for all the instances, because the primal solutions \((\hat{x}_j,\hat{y}_j)\) obtained from the root node subproblem were integer feasible. Note that the differences in total solution time were due mainly to the wall-clock time limit of 300 s for subproblem solutions. For the given time limit, CPLEX may return slightly different solutions even for the same problem. On the other hand, the CS method was not able to find any feasible solution (and thus an upper bound) for all the instances. However, we found that the lower bounds obtained by CS method are the same as the optimal objective values for all the instances. The reason is that the average point \(\bar{x}\) computed by the CS method did not represent valid primal solutions for the instances, which required solving more node subproblems for finding a primal feasible solution. This observation is consistent with results with the CS method for sslp instances, as shown in Table 2.
Table 6 reports the numerical results from using the CS method with the fixing-first heuristic. Note that we do not report the results for the other methods because the global optimal solutions were already obtained at the root node without any heuristic. With the heuristic, the CS method found feasible solutions for 12 instances, of which 8 instances were optimal. However, the CS method still failed to find a feasible solution for the other 8 instances within the 4-h time limit.
Since smkp instances have binary variables only, the simple rounding heuristic does nothing but fixing the average point for checking the feasibility. As a result, the numerical results are same as those without any heuristic and thus not reported.
Rights and permissions
About this article
Cite this article
Kim, K., Dandurand, B. Scalable branching on dual decomposition of stochastic mixed-integer programming problems. Math. Prog. Comp. 14, 1–41 (2022). https://doi.org/10.1007/s12532-021-00212-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12532-021-00212-y