Abstract
A 0-1 quadratic programming model is presented for solving the strategic problem of timing the location of facilities and the assignment of customers to facilities in a multi-period setting. It is assumed that all parameters are known and, on the other hand, the quadratic character of the objective function is due to considering the interaction cost incurred by the joint assignment of customers belonging to different categories to a facility at a period. The plain use of a state-of-the-art MILP engine with capabilities for dealing with quadratic terms does not give any advantage over the matheuristic algorithm proposed in this work. In fact, the MILP engine was frequently running out of memory before reaching optimality for the equivalent mixed 0-1 linear formulation, being its best lower bound at that time instant too far from the incumbent solution for the large-sized instances which we have worked with. As an alternative, a fix-and-relax algorithm is presented. A deep computational comparison between MILP alternatives is performed, such that fix-and-relax provides a solution value very close to (and, frequently, a better than) the one provided by the MILP engine. The time required by fix-and-relax is very affordable, being frequently two times smaller than the time required by the MILP engine.
Similar content being viewed by others
References
Albareda-Sambola M, Alonso-Ayuso A, Escudero LF, Fernández E, Hinojosa Y, Pizarro C (2010) A computational comparison of several formulations for the multi-period location-assignment problem. Top 18:62–80
Albareda-Sambola M, Alonso-Ayuso A, Escudero LF, Fernández E, Pizarro C (2013) On solving the multi-period location-assignment problem under uncertainty. Comput Oper Res 40:2878–2892
Albareda-Sambola M, Fernández E, Hinojosa Y, Puerto J (2009) The multi-period incremental service facility location problem. Comput Oper Res 36:1356–1375
Bliek C, Bonami P, Lodi A (2014) Solving mixed-integer quadratic programming problems with IBM-CPLEX: a progress report. In: Proceedings of the twenty-sixth RAMP symposium. Hosei University, Tokyo
Boland N, Dey SS, Kalinowski T, Molinaro M, Rigterink F (2017) Bounding the gap between the McCormick relaxation and the convex hull for bilinear functions. Math Program Ser A 162:523–535
Burkard RE, Fincke U (1982) On random quadratic bottleneck assignment problems. Math Program 23:227–232
Castro J, Nasini S, Saldanha-da-Gama F (2016) A cutting-plane approach for large-scale capacitated multi-period facility location using a specialized interior-point method. Math Program Ser. A. doi:10.1007/s10107-016-1067-6
Correia I, Melo T (2016) Multi-period capacitated facility location under delayed demand satisfaction. Eur J Oper Res 255:729–746
Dash S, Puget J-F (2015) On quadratic unconstraintred binary optimization problems defined on Chimera graphs. Optima 98:2–6
Dias J, Captivo ME, Climaco J (2007) Dynamic location problems with discrete expansion and reduction sizes of available capacities. Investig Oper 27:107–130
Dillenberger Ch, Escudero LF, Wollensak A, Zhang W (1994) On practical resource allocation for production planning and scheduling with period overlapping setups. Eur J Oper Res 75:275–286
Escudero LF, Salmerón J (2005) On a Fix-and-Relax framework for large-scale resource-constrained project scheduling. Ann Oper Res 140:163–188
Fortet R (1960) Application de l’algebre de boole en recherche operationelle. Revue Francaise de Recherche Operationelle 4:17–26
Glover F (1975) Improved linear integer programming formulations of nonlinear integer problems. Manag Sci 22:455–460
Hammer PL, Rudeanu S (1968) Boolean methods in operations research and related areas. Springer, Berlin
Jena SD, Cordeau JF, Gendron B (2016) Lagrangean heuristics for large-scale dynamic facility location with generalized modular capacities. INFORMS J Comput. doi:10.1287/ijoc.2016.0738
Jena SD, Cordeau JF, Gendron B (2015) Dynamic facility location with generalized modular capacity. Transp Sci 49:484–499
Jena SD, Cordeau JF, Gendron B (2016) Solving a dynamic facility location problems with partial closing and reopening. Comput Oper Res 67:143–154
Krislock N, Malick J, Roupin F (2014) Improved semi-definite bounding procedure for solving max-cut problems to optimality. Math Program 143:61–86
Laporte G, Nickel S, Saldanha da Gama F (eds) (2015) Location science. Springer, Berlin
McCormick GP (1976) Computability of global solutions to factorable nonconvex programs: part I—convex underestimating problems. Math Programm 10:147–175
Nickel S, Saldanha da Gama F (2015) Multi-period facility location. In: Laporte G, Nickel S, Saldanha da Gama F (eds) Location science. Springer, Berlin, pp 289–310
O’Kelly ME (1987) A quadratic integer program for the location of interacting hub facilities. Eur J Oper Res 22:393–404
Sherali H (2007) RLT: a unified approach for discrete and continuous nonconvex optimization. Ann Oper Res 149:185–193
Sherali HD, Adams WP (1994) A hierarchy of relaxations and convex hull characterizations for mixed-integer zero-one programming problems. Discrete Appl Math 53:83–106
Zaied ANH, Shawky LAE (2014) A survey of the quadratic assignment problem. Int J Comput Appl 101:28–35
Acknowledgements
This research has been partially supported by the projects MTM2015-63710-P and MTM2015-65317-P from the Spanish Ministry of Economy and Competitiveness. The authors would like to thank to the two anonymous reviewers for their help on clarifying some concepts presented in the manuscript and strongly improving its presentation.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
The appendix presents in Tables 5 and 6 the sequence of tighter lower bounds \(\underline{Z}_{x\mathrm{FR}}^\ell \) and \(\underline{Z}_{y\mathrm{FR}}^\ell \) on the optimal solution of the original MILP model (31) and the related elapsed times obtained by the models \(\underline{\mathrm{MILP}}_{x\mathrm{FR}}^\ell \) (35) and \(\underline{\mathrm{MILP}}_{y\mathrm{FR}}^\ell \) (36), respectively, for the FR levels, \(\ell =1,\ldots ,5\). It can be observed that the yFR bound for \(\ell =5\) is still computationally affordable while comparing it with the time required by the plain use of CPLEX, at least.
Rights and permissions
About this article
Cite this article
Escudero, L.F., Pizarro Romero, C. On solving a large-scale problem on facility location and customer assignment with interaction costs along a time horizon. TOP 25, 601–622 (2017). https://doi.org/10.1007/s11750-017-0461-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11750-017-0461-4
Keywords
- Location assignment
- Interaction costs
- 0-1 QP model
- Equivalent 0-1 MILP formulation
- Fix-and-relax matheuristic