Abstract
The purpose of this paper is to provide strong reformulations for binary quadratic problems. We propose a first methodological analysis on a family of reformulations combining Dantzig–Wolfe and Quadratic Convex optimization principles. We show that a few reformulations of our family yield continuous relaxations that are strong in terms of dual bounds and computationally efficient to optimize. As a representative case study, we apply them to a cardinality constrained quadratic knapsack problem, providing extensive experimental insights. We report and analyze in depth a particular reformulation providing continuous relaxations whose solutions turn out to be integer optima in all our tests.
Similar content being viewed by others
Data Availability Statement
The full set of instances used in the paper is made available as Ceselli, A., Létocart, L., Traversi, E. “Binary quadratic problem decomposition methods—kQKP dataset”, UNIMI Dataverse (2021), URL https://urldefense.proofpoint.com/v2/url?u=https-3A_doi.org_10.13130_RD-5FUNIMI_3QA23K&d=DwIDaQ&is included as reference [13] in the paper.
References
Adams, W., Forrester, R., Glover, F.: Comparisons and enhancement strategies for linearizing mixed 0–1 quadratic programs. Discret. Optim. 1(2), 99–120 (2004). https://doi.org/10.1016/j.disopt.2004.03.006
Aganagic, M., Mokhtari, S.: Security constrained economic dispatch using nonlinear Dantzig–Wolfe decomposition. IEEE Trans. Power Syst. 12(1), 105–112 (1997)
Ahlatçioglu, A., Bussieck, M., Esen, M., Guignard, M., Jagla, J.H., Meeraus, A.: Combining QCR and CHR for convex quadratic pure 0–1 programming problems with linear constraints. Ann. OR 199(1), 33–49 (2012)
Balas, E., Zemel, E.: An algorithm for large zero-one knapsack problems. Oper. Res. 28, 1130–1154 (1980)
Basso, S., Ceselli, A., Tettamanzi, A.: Random sampling and machine learning to understand good decompositions. Ann. Oper. Res. 284(2), 501–526 (2020). https://doi.org/10.1007/s10479-018-3067-9
Bergner, M., Caprara, A., Ceselli, A., Furini, F., Lübbecke, M., Malaguti, E., Traversi, E.: Automatic Dantzig–Wolfe reformulation of mixed integer programs. Math. Program. 149(1–2), 391–424 (2015)
Billionnet, A., Elloumi, S., Lambert, A.: Extending the QCR method to general mixed-integer programs. Math. Program. 131(1–2), 381–401 (2012)
Billionnet, A., Elloumi, S., Plateau, M.C.: Improving the performance of standard solvers via a tighter convex reformulation of constrained quadratic 0–1 programs: the QCR method. Discret. Appl. Math. 157, 1185–1197 (2009)
Billionnet, A., Soutif, E.: An exact method based on Lagrangean decomposition for the 0–1 quadratic knapsak problem. Eur. J. Oper. Res. 157, 565–575 (2004)
Buchheim, C., Wiegele, A.: Semidefinite relaxations for non-convex quadratic mixed-integer programming. Math. Program. 141(1–2), 435–452 (2013). https://doi.org/10.1007/s10107-012-0534-y
Caprara, A., Pisinger, D., Toth, P.: Exact solution of the quadratic knapsack problem. INFORMS J. Comput. 11(2), 125–137 (1999)
Ceselli, A., Létocart, L., Traversi, E.: Dantzig Wolfe decomposition and objective function convexification for binary quadratic problems: the cardinality constrained quadratic knapsack case. Tech. Rep. 5923, 1–32 (2017)
Ceselli, A., Létocart, L., Traversi, E.: Binary quadratic problem decomposition methods—kQKP dataset—UNIMI dataverse (2021). https://doi.org/10.13130/RD_UNIMI/3QA23K
Dantzig, G., Wolfe, P.: Decomposition principle for linear programs. Oper. Res. 8, 101–111 (1960)
Desaulniers, G., Desrosiers, J., Solomon, M.: Column Generation, vol. 5. Springer Science & Business Media, Berlin (2006)
Fayard, D., Plateau, G.: Algorithm 47: an algorithm for the solution of the 0–1 knapsack problem. Computing 28, 269–287 (1982)
Faye, A., Roupin, F.: Partial Lagrangian relaxation for general quadratic programming. 4OR 5(1), 75–88 (2007)
Fortet, R.: Applications de lalgèbre de boole en recherche opérationnelle. Revue Française de Recherche Opérationnelle 4, 17–26 (1960)
Gamrath, G., Lübbecke, M.: Experiments with a generic Dantzig–Wolfe decomposition for integer programs. In: Festa, P. (ed.) Experimental Algorithms, pp. 239–252. Springer, Berlin, Heidelberg (2010)
Glover, F.: Improved linear integer programming formulations of nonlinear integer problems. Manag. Sci. 22(4), 455–460 (1975)
Gueye, S., Michelon, P.: Miniaturized linearizations for quadratic 0/1 problems. Ann. Oper. Res. 140(1), 235–261 (2005). https://doi.org/10.1007/s10479-005-3973-5
Gurobi Optimization: Gurobi (2018). http://www.gurobi.com/. Version 8.0, http://www.gurobi.com/
Holmberg, K.: MINLP: generalized cross decomposition. In: Encyclopedia of Optimization, pp. 2148–2155. Springer (2009)
IBM: Cplex (2018). Version 12.8.0. https://www-01.ibm.com/software/commerce/optimization/cplex-optimizer/
Jünger, M., Liebling, T., Naddef, D., Nemhauser, G.L., Pulleyblank, W.R., Reinelt, G., Rinaldi, G., Wolsey, L.A. (eds.): 50 Years of Integer Programming 1958–2008. Springer, Berlin (2009)
Krislock, N., Malick, J., Roupin, F.: Biqcrunch: a semidefinite branch-and-bound method for solving binary quadratic problems. ACM Trans. Math. Softw. 43(4), 32:1-32:23 (2017). https://doi.org/10.1145/3005345
Lawphongpanich, S.: Simplicial with truncated Dantzig–Wolfe decomposition for nonlinear multicommodity network flow problems with side constraints. Oper. Res. Lett. 26(1), 33–41 (2000)
Lemaréchal, C., Oustry, F.: Semidefinite relaxations and Lagrangian duality with application to combinatorial optimization. Ph.D. thesis, INRIA (1999)
Lemaréchal, C., Renaud, A.: A geometric study of duality gaps, with applications. Math. Program. 90(3), 399–427 (2001)
Létocart, L., Plateau, M.C., Plateau, G.: An efficient hybrid heuristic method for the 0–1 exact \(k\)-item quadratic knapsack problem. Pesquisa Operacional 34(1), 49–72 (2014)
Létocart, L., Wiegele, A.: Exact solution methods for the k-item quadratic knapsack problem. In: Lecture Notes in Computer Science, Combinatorial Optimization, vol. 9849, pp. 166–176. Springer (2016)
Liberti, L.: Compact linearization for binary quadratic problems. 4OR 5(3), 231–245 (2007). https://doi.org/10.1007/s10288-006-0015-3
Misener, R., Floudas, C.: GloMIQO: global mixed-integer quadratic optimizer. J. Global Optim. 57, 3–50 (2013)
Pisinger, D.: The quadratic knapsack problem: a survey. Discret. Appl. Math. 155, 623–648 (2007)
Pisinger, D., Rasmussen, A.B., Sandvik, R.: Solution of large quadratic knapsack problems through aggressive reduction. INFORMS J. Comput. 19(2), 280–290 (2007)
Plateau, M.C.: Quadratic convex reformulations for quadratic 0–1 programming. 4OR 6, 187–190 (2008)
Rendl, F., Rinaldi, G., Wiegele, A.: Solving max-cut to optimality by intersecting semidefinite and polyhedral relaxations. Math. Program. 121(2), 307 (2010)
SCIP: Scip (2018). http://scip.zib.de/. Version 6.0.0, http://scip.zib.de/
Sherali, H., Adams, W.: A tight linearization and an algorithm for 0–1 quadratic programming problems. Manag. Sci. 32(10), 1274–1290 (1986)
Vanderbeck, F., Savelsbergh, M.: A generic view of Dantzig–Wolfe decomposition in mixed integer programming. Oper. Res. Lett. 34(3), 296–306 (2006)
Wolfe, P.: A duality theorem for non-linear programming. Quart. Appl. Math. 19(3), 239–244 (1961)
Acknowledgements
The authors wish to thank the associate editor and three anonymous reviewers, whose insightful comments and careful proof-checks helped to improve the paper.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Funding
The work has been partially funded by Regione Lombardia - Fondazione Cariplo, grant no. 2015–0717, project “REDNEAT”, and partially undertook when A. Ceselli was visiting LIPN - Université Sorbonne Paris Nord.
Conflict of interest
The authors declare that they have no conflict of interest.
Availability of data and materials
All data analyzed during this study are publicly available. URLs are included in this published article.
Code availability
The full code was made available for review. We remark that a set of packages were used in this study, that were either open source or available for academic use. Specific references are included in this published article.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
A Details on the QCR method
1.1 A.1 Semidefinite model providing \(\delta ^*\) and \(\rho ^*\)
Let us consider (\(\hbox {BQ}_{\delta ,\rho }\)) (the convexified formulation of a Binary Quadratic Problem, introduced in Sect. 4):
The SDP used to obtain the optimal parameters \(\delta ^*,\rho ^*\) (see [8, 36]) is the following:
Namely, the optimal values \(\delta ^*\) and \(\rho ^*\) of problem (\(\hbox {BQ}_{\delta ,\rho }\)) are simply given by the optimal values of the dual variables of (\(\text {SDP}_{\text {BQ}_{\delta ,\rho }}\)).
1.2 A.2 Semidefinite model providing \(\delta ^*\), \(\rho ^*\) and \(\varGamma ^*\)
Let us consider (\(\hbox {BQ}_{\delta ,\rho ,\varGamma }\)):
The SDP used to obtain the optimal parameters \(\delta ^*,\rho ^*,\varGamma ^*\) (see [7]) is the following:
Namely, the optimal values \(\delta ^*\), \(\rho ^*\) and \(\varGamma ^*\) of problem (\(\hbox {BQ}_{\delta ,\rho ,\varGamma }\)) are simply given by the optimal values of the dual variables of (\(\text {SDP}_{\text {BQ}_{\delta ,\rho ,\varGamma }}\)), with \({\varGamma }^* = {\varGamma ^{+}}^* + {\varGamma ^{-}}^*\).
B Dual of a nonlinear model reformulated with DWR
Let us consider the following formulation
(we recall that the constraints \(x \in [0,1]^n\) are included in the constraints \(g_i(x) \le 0\)).
The formulation of its Wolfe dual is
with
which leads to
If we substituting the definition of \(\nu _p= \pi ^\top x_p - \pi _0 \) in \(L(x, y, \mu , \pi , \pi _0, \nu )\) we have:
leading to the following dual:
C Reinterpretation of the new reformulations for (BQP) with Lagrangian duality
Let us consider the following reformulation of (BQP):
where for each constraint we indicate in squared brakets the corresponding Lagrangian multiplier.
Sections 3 and 4 show that the continuous relaxations of the reformulations showed in Table 1 can be viewed as a specific Lagrangian Dual of (\(\hbox {Q}'\)) where some of the constraints are relaxed in the Lagrangian Function, some are kept in the Lagrangian subproblem and other are dropped. For simplicitly, each constraint is represented by its Lagrangian multiplier. For example, (\(\hbox {Q}_{\delta ,\rho }\)-QM) is obtained by relaxing constraints \([\phi ]\),\([\mu ]\),\([\delta ]\),\([\rho ]\), keeping constraints \([\psi ]\),\([\delta ]\) and dropping constraints \([\varXi ]\),\([\varUpsilon ]\),\([\varGamma ]\):
with \(L(x,w,\phi ,\mu ,\delta ,\rho ) = w^\top Q w + L^\top w + \phi ^\top (x- w) + \mu ^\top (G w - g) + \delta ^\top (x^2 - x) + \rho (A_= x -b_=)^2\).
D Comparing (K\(_{\delta ^*,\rho ^*}\)-QP-kn) with CPLEX
In Table 5 we compare the time needed for computing the root node of (K\(_{\delta ^*,\rho ^*}\)-QP-kn) (that in each test corresponds to the time needed to solve the problem to global integer optimality) with the overall time needed by CPLEX to solve the original formulation with general purpose techniques (BK) and (\(\hbox {BK}_{\delta ^*,\rho ^*}\)). For (BK) the number of instances yielding timeout is also reported: the average computing time does not include them.
The results of Table 5 provide the following observation:
Experimental Observation 13
With (K\(_{\delta ^*,\rho ^*}\)-QP-kn) we are able to close the whole integrality gap faster than the time needed by CPLEX to solve to optimality the same instance. The time needed by (K\(_{\delta ^*,\rho ^*}\)-QP-kn) is similar to that of (\(\hbox {BK}_{\delta ^*,\rho ^*}\)), that is CPLEX after applying an ad-hoc convexification of the objective function.
We remark that none of these techniques uses problem specific algorithms. In this regard, (K\(_{\delta ^*,\rho ^*}\)-QP-kn) has some additional potential, as the column generation scheme can strongly benefit from the use of ad-hoc pricing algorithms.
Rights and permissions
About this article
Cite this article
Ceselli, A., Létocart, L. & Traversi, E. Dantzig–Wolfe reformulations for binary quadratic problems. Math. Prog. Comp. 14, 85–120 (2022). https://doi.org/10.1007/s12532-021-00206-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12532-021-00206-w