Learning and Propagating Lagrangian Variable Bounds for Mixed-Integer Nonlinear Programming

  • Ambros M. Gleixner
  • Stefan Weltge
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7874)

Abstract

Optimization-based bound tightening (OBBT) is a domain reduction technique commonly used in nonconvex mixed-integer nonlinear programming that solves a sequence of auxiliary linear programs. Each variable is minimized and maximized to obtain the tightest bounds valid for a global linear relaxation. This paper shows how the dual solutions of the auxiliary linear programs can be used to learn what we call Lagrangian variable bound constraints. These are linear inequalities that explain OBBT’s domain reductions in terms of the bounds on other variables and the objective value of the incumbent solution. Within a spatial branch-and-bound algorithm, they can be learnt a priori (during OBBT at the root node) and propagated within the search tree at very low computational cost. Experiments with an implementation inside the MINLP solver SCIP show that this reduces the number of branch-andbound nodes and speeds up solution times.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Tawarmalani, M., Sahinidis, N.V.: Convexification and Global Optimization in Continuous and Mixed-Integer Nonlinear Programming: Theory, Algorithms, Software, and Applications. Kluwer Academic Publishers, Dordrecht Boston London (2002)MATHGoogle Scholar
  2. 2.
    Quesada, I., Grossmann, I.E.: Global optimization algorithm for heat exchanger networks. Industrial & Engineering Chemistry Research 32(3), 487–499 (1993), doi:10.1021/ie00015a012CrossRefGoogle Scholar
  3. 3.
    Quesada, I., Grossmann, I.E.: A global optimization algorithm for linear fractional and bilinear programs. Journal of Global Optimization 6, 39–76 (1995), doi:10.1007/BF01106605MathSciNetMATHCrossRefGoogle Scholar
  4. 4.
    Maranas, C.D., Floudas, C.A.: Global optimization in generalized geometric programming. Computers & Chemical Engineering 21(4), 351–369 (1997), doi:10.1016/S0098-1354(96)00282-7MathSciNetCrossRefGoogle Scholar
  5. 5.
    Smith, E.M., Pantelides, C.C.: A symbolic reformulation/spatial branch-and-bound algorithm for the global optimisation of nonconvex MINLPs. Computers & Chemical Engineering 23, 457–478 (1999), doi:10.1016/S0098-1354(98)00286-5CrossRefGoogle Scholar
  6. 6.
    Zamora, J.M., Grossmann, I.E.: A branch and contract algorithm for problems with concave univariate, bilinear and linear fractional terms. Journal of Global Optimization 14, 217–249 (1999), doi:10.1023/A:1008312714792MathSciNetMATHCrossRefGoogle Scholar
  7. 7.
    Adjiman, C.S., Androulakis, I.P., Floudas, C.A.: A global optimization method, αBB, for general twice-differentiable constrained NLPs—II. Implementation and computational results. Computers & Chemical Engineering 22(9), 1159–1179 (1998), doi:10.1016/S0098-1354(98)00218-XCrossRefGoogle Scholar
  8. 8.
    Adjiman, C.S., Androulakis, I.P., Floudas, C.A.: Global optimization of mixed-integer nonlinear problems. AIChE Journal 46(9), 1769–1797 (2000), doi:10.1002/aic.690460908CrossRefGoogle Scholar
  9. 9.
    Belotti, P., Lee, J., Liberti, L., Margot, F., Wächter, A.: Branching and bounds tightening techniques for non-convex MINLP. Optimization Methods & Software 24, 597–634 (2009), doi:10.1080/10556780903087124MathSciNetMATHCrossRefGoogle Scholar
  10. 10.
    COIN-OR: Couenne. Convex Over and Under ENvelopes for Nonlinear Estimation, http://www.coin-or.org/Couenne
  11. 11.
    Misener, R., Floudas, C.A.: GloMIQO: Global mixed-integer quadratic optimizer. Journal of Global Optimization, 1–48 (2012), doi:10.1007/s10898-012-9874-7Google Scholar
  12. 12.
    Computer-Aided Systems Laboratory, Princeton University: GloMIQO. Global Mixed-Integer Quadratic Optimizer, http://helios.princeton.edu/GloMIQO
  13. 13.
    Nowak, I., Vigerske, S.: LaGO: a (heuristic) branch and cut algorithm for nonconvex MINLPs. Central European Journal of Operations Research 16(2), 127–138 (2008), doi:10.1007/s10100-007-0051-xMathSciNetMATHCrossRefGoogle Scholar
  14. 14.
    COIN-OR: LaGO Lagrangian Global Optimizer, http://www.coin-or.org/LaGO
  15. 15.
    Achterberg, T.: Constraint Integer Programming. PhD thesis, TU Berlin (2007), http://opus4.kobv.de/opus4-zib/frontdoor/index/index/docId/1018
  16. 16.
    Vigerske, S.: Decomposition in Multistage Stochastic Programming and a Constraint Integer Programming Approach to MINLP. PhD thesis, HU Berlin (2012)Google Scholar
  17. 17.
    Zuse Institute Berlin, Department of Optimization: SCIP. Solving Constraint Integer Programs, http://scip.zib.de
  18. 18.
    Caprara, A., Locatelli, M.: Global optimization problems and domain reduction strategies. Mathematical Programming 125, 123–137 (2010), doi:10.1007/s10107-008-0263-4MathSciNetMATHCrossRefGoogle Scholar
  19. 19.
    Tawarmalani, M., Sahinidis, N.V.: Global optimization of mixed-integer nonlinear programs: A theoretical and computational study. Mathematical Programming 99, 563–591 (2004), doi:10.1007/s10107-003-0467-6MathSciNetMATHCrossRefGoogle Scholar
  20. 20.
    Berthold, T., Heinz, S., Vigerske, S.: Extending a CIP framework to solve MIQCPs. In: Lee, J., Leyffer, S. (eds.) Mixed41-Integer Nonlinear Optimization. The IMA volumes in Mathematics and its Applications, vol. 154, pp. 427–444. Springer (2012), doi:10.1007/978-1-4614-1927-3_15Google Scholar
  21. 21.
    Bussieck, M., Drud, A., Meeraus, A.: MINLPLib – a collection of test models for mixed-integer nonlinear programming. INFORMS J. on Comput. 15(1), 114–119 (2003)MathSciNetMATHCrossRefGoogle Scholar
  22. 22.
  23. 23.
    COIN-OR: CppAD. A Package for Differentiation of C++ Algorithms. http://www.coin-or.org/CppAD
  24. 24.
    Wächter, A., Biegler, L.T.: On the implementation of a primal-dual interior point filter line search algorithm for large-scale nonlinear programming. Math. Prog. 106(1), 25–57 (2006)MATHCrossRefGoogle Scholar
  25. 25.
    COIN-OR: Ipopt. Interior point optimizer, http://www.coin-or.org/Ipopt
  26. 26.
    Gleixner, A.M., Weltge, S.: Learning and propagating Lagrangian variable bounds for mixed-integer nonlinear programming. ZIB-Report 13-04, Zuse Institute Berlin (2013), http://opus4.kobv.de/opus4-zib/frontdoor/index/index/docId/1763

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  • Ambros M. Gleixner
    • 1
  • Stefan Weltge
    • 2
  1. 1.Zuse Institute BerlinBerlinGermany
  2. 2.Otto-von-Guericke-Universität MagdeburgMagdeburgGermany

Personalised recommendations