Abstract
This paper deals with the computation of exact solutions of a classical NP-hard problem in combinatorial optimization, the \(k\)-cluster problem. This problem consists in finding a heaviest subgraph with \(k\) nodes in an edge weighted graph. We present a branch-and-bound algorithm that applies a novel bounding procedure, based on recent semidefinite programming techniques. We use new semidefinite bounds that are less tight than the standard semidefinite bounds, but cheaper to get. The experiments show that this approach is competitive with the best existing ones.
Similar content being viewed by others
References
Asahiro, Y., Iwama, K., Tamaki, H., Tokuyama, T.: Greedily finding a dense subgraph. J. Algorithms 34(2), 203–221 (2000)
Billionnet, A.: Different formulations for solving the heaviest k-subgraph problem. Inf. Syst. Oper. Res. 43(3), 171–186 (2005)
Billionnet, A., Elloumi, S.: Using a mixed integer quadratic programming solver for the unconstrained quadratic 0–1 problem. Math. Program. 109(1), 55–68 (2006)
Billionnet, A., Elloumi, S., Plateau, M.-C.: Improving the performance of standard solvers for quadratic 0–1 programs by a tight convex reformulation: the QCR method. Discret. Appl. Math. 157(6), 1185–1197 (2009)
Bonnans, J.F., Gilbert, JCh.: Numerical Optimization. Springer, Berlin (2003)
Borchers, B.: CSDP, a C library for semidefinite programming. Optim. Methods Softw. 11(1), 613–623 (1999)
Borsdorf, R., Higham, N.: A preconditionned newton algorithm for the nearest correlation matrix. Submitted (2008)
Burer, S., Monteiro, R.D.C.: A nonlinear programming algorithm for solving semidefinite programs via low-rank factorization. Math. Program. (Series B) 95, 329–357 (2003)
Corneil, D.G., Perl, Y.: Clustering and domination in perfect graphs. Discret. Appl. Math. 9(1), 7–39 (1984)
Erkut, E.: The discrete p-dispersion problem. Eur. J. Oper. Res. 46(1), 48–60 (1990)
Faye, A., Roupin, F.: A cutting planes algorithm based upon a semidefinite relaxation for the quadratic assignment problem. In: ESA 2008, LNCS 3669, pp. 850–861 (2005)
Faye, A., Roupin, F.: Partial lagrangian for general quadratic programming. 4’OR Q. J. Oper. Res. 5(1), 75–88 (2007)
Feige, U., Langberg, M.: Approximation algorithms for maximization problems arising in graph partitioning. J. Algorithms 41(2), 174–211 (2001)
Feige, U., Kortsarz, G., Peleg, D.: The dense k-subgraph problem. Algorithmica 29(3), 410–421 (2001)
Gilbert, JCh., Lemaréchal, C.: Some numerical experiments with variable-storage quasi-Newton algorithms. Math. Program. 45, 407–435 (1989)
Goemans, M.X., Williamson, D.P.: Improved approximation algorithms for maximum cut and satisfiability problems using semidefinite programming. J. ACM 6, 1115–1145 (1995)
Han, Q., Ye, Y., Zhang, J.: An improved rounded method and semidefinite programming relaxation for graph partition. Math. Program. 92(3), 509–535 (2002)
Helmberg, C.: A C++ implementation of the Spectral Bundle Method. http://www-user.tu-chemnitz.de/ helmberg/SBmethod/, Version 1.1.3 (2004)
Helmberg, C., Rendl, F.: A spectral bundle method for semidefinite programming. SIAM J. Optim. 10(3), 673–696 (2000)
Helmberg, C., Rendl, F., Vanderbei, R., Wolkowicz, H.: An interior point method for semidefinite programming. SIAM J. Optim. 6, 342–361 (1996)
Higham, N.: Computing a nearest symmetric correlation matrix-a problem from finance. IMA J. Numer. Anal. 22(3), 329–343 (2002)
Hiriart-Urruty, J.-B., Lemaréchal, C.: Convex Analysis and Minimization Algorithms. Springer, Heidelberg (1993). Two volumes
Jäger, G., Srivastav, A.: Improved approximation algorithms for maximum graph partitioning problems. J. Combin. Optim. 10(2), 133–167 (2005)
Keil, J.M., Brecht, T.B.: The complexity of clustering in planar graphs. J. Comb. Math. Comb. Comput. 9, 155–159 (1991)
Khot, S.: Ruling out ptas for graph min-bisection, dense k-subgraph, and bipartite clique. SIAM J. Comput. 36, 1025–1071 (2005)
Kočvara, M., Stingl, M.: On the solution of large-scale sdp problems by the modified barrier method using iterative solvers. Math. Program. 109(2), 413–444 (2007)
Krarup, J., Pisinger, D., Plastria, F.: Discrete location problems with push-pull objectives. Discret. Appl. Math. 123, 363–378 (2002)
Lemaréchal C., Oustry F. (1999) Semidefinite relaxations and Lagrangian duality with application to combinatorial optimization. Research report 3710, INRIA
Liazi, M., Milis, I., Zissimopoulos, V.: A constant approximation algorithm for the densest k-subgraph problem on chordal graphs. Inf. Process. Lett. 108(1), 29–32 (2008)
Lovász, L.: On the Shannon capacity of a graph. IEEE Trans. Inf. Theory IT 25, 1–7 (1979)
Malick, J.: A dual approach to semidefinite least-squares problems. SIAM J. Matrix Anal. Appl. 26(1), 272–284 (2004)
Malick, J.: Spherical constraint in Boolean quadratic programming. J. Glob. Optim. 39(4), 609–622 (2007)
Malick, J., Roupin, F.: Numerical study of SDP bounds for the \(k\)-cluster problem. E. Notes Discret. Math. 36(1), 399–406 (2010). Proceedings of ISCO 2010
Malick, J., Roupin, F.: On the bridge between combinatorial optimization and nonlinear optimization: a family of semidefinite bounds leading to newton-like methods. To appear in Math Programming B. http://hal.archives-ouvertes.fr/hal-00662367 (2012)
Malick, J., Povh, J., Rendl, F., Wiegele, A.: Regularization methods for semidefinite programming. SIAM J. Optim. 20(1), 336–356 (2009)
Martin, A., Achterberg, T., Koch, T.: Branching rules revisited. Oper. Res. Lett. 33(1), 342–354 (2005)
Nocedal, J., Wright, S.: Numerical Optimization. Springer, NY (1999)
Perl, Y., Shiloach, Y.: Efficient optimization of monotonic functions on trees. SIAM J. Algebr. Discret. Methods 4(4), 512–516 (1983)
Pisinger, D.: Upper bounds and exact algorithms for p-dispersion problems. Comput. Oper. Res. 33, 1380–1398 (2006)
Plateau, M.-C.: Reformulations quadratiques convexes pour la programmation quadratique en variables 0–1. PhD thesis, Conservatoire National des Arts et Métiers (CNAM, Paris) (2006)
Qi, H., Sun, D.: Quadratic convergence and numerical experiments of Newton’s method for computing the nearest correlation matrix. SIAM J. Matrix Anal. Appl. 28, 360–385 (2006)
Rendl, F., Rinaldi, G., Wiegele, A.: Solving max-cut to optimality by intersecting semidefinite and polyedral relaxations. Math. Program. 121, 307–335 (2010)
Roupin, F.: From linear to semidefinite programming: an algorithm to obtain semidefinite relaxations for bivalent quadratic problems. J. Comb. Optim. 8(4), 469–493 (2004)
Todd, M.J.: Semidefinite optimization. Acta Numer. 10, 515–560 (2001)
Tutuncu, R.H., Toh, K.C., Todd, M.J.: Inexact primal-dual path-following algorithms for a special class of convex quadratic SDP and related problems. Pac. J. Optim. 3(1), 135–164 (2006)
Acknowledgments
This work was supported by CNRS (“GdR Recherche Opérationnelle”) and Grenoble University (Université Joseph Fourier, through “Pôle Math-STIC”). We thank Quentin Monnet and Lise-Marie Veillon (Master students of ENSIIE, Evry, France) for their help in developing parts of the solver. We also thank the two anonymous referees for numerous suggestions about a preliminary version of this article.
Author information
Authors and Affiliations
Corresponding author
Appendix: Formulation of the \(k\)-cluster as a quadratic \(\{-1,1\}\)-problem
Appendix: Formulation of the \(k\)-cluster as a quadratic \(\{-1,1\}\)-problem
The spherical constraint appears more easily on purely quadratic \(\{-1,1\}\)-optimization problems (see [32]). The second reformulation in Sect. 2 considers the transformation of the \(k\)-cluster problem, from the natural modeling as a quadratic \(\{0,1\}\)-problem (with linear and quadratic constraints), to a purely quadratic \(\{-1,1\}\)-problem. We specify here this transformation, that uses standard techniques.
Lemma 3
With the notation of this section, the \(k\)-cluster problem (1) is equivalent to the quadratic problem in dimension \(n+ 1\)
where the symmetric \((n+1)\times (n+1)\)-matrices \(Q\) and \(Q_j\ (j\in \{0,\ldots ,n\})\) are defined by (5). More precisely, this equivalence means that the optimal values of (1) and (17) are the same and that the optimal solutions coincide as follows:
-
if \(\bar{z}\) is a solution of (1) then \(\bar{x} = (1, 2\bar{z}-e)\) is a solution of (1);
-
if \(\bar{x}\) is a solution of (17), then \(\bar{z} = ((\bar{x}_0\bar{x}_1,\ldots , \bar{x}_0\bar{x}_{n}) +e)/2\) is a solution of (1).
Proof
Operate first the change of variable \(x=2z-e\), and express the objective and the constraints with respect to \(x=(x_1,\ldots ,x_n)\in \mathbb R ^n\). Just develop the objective
and similarly transform \({e}^{\top }z = k\) as \({e}^{\top }x = 2k-n\), and for all \(j=1,\ldots ,n\)
the last equivalence coming from \({e}^{\top }C_je= 2n\) and \(\tilde{e}_j = C_je - 2ke_j\). So we get quadratic problem in \(x\in \{-1,1\}^n\)
The formulation of (3) is equivalent to (18) in the sense that the optimal values are the same and the solutions are in one-to-one correspondance with \(x=2z-e\).
We consider now that the following purely quadratic problem with the additional variable \(x_0\in \{-1,1\}\)
We observe that this problem is equivalent to (17) in view of the definitions of the matrices in (5). So we just have to establish that (18) is equivalent to (19); we do so in two steps. If \((x_1,\ldots ,x_n)\) is feasible in (18), then obviously \((1,x)\) and \((-1,-x)\) are both feasible in (19), with same objective value. It follows that \(\mathrm{val}(19) \le \mathrm{val}(18)\). Conversely if \((x_0,\ldots ,x_n)\) is feasible in (19), then \(x_0=\pm 1\) and \((x_0x_1,\ldots ,x_0x_n)\) is feasible in (18) with the same objective value. It follows that \(\mathrm{val}(18)\le (19)\), so that we have the equality in fact. The relation between the argmins then becomes clear.
\(\square \)
Rights and permissions
About this article
Cite this article
Malick, J., Roupin, F. Solving \(k\)-cluster problems to optimality with semidefinite programming. Math. Program. 136, 279–300 (2012). https://doi.org/10.1007/s10107-012-0604-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10107-012-0604-1
Keywords
- Combinatorial optimization
- \(k\)-cluster
- Exact resolution
- Semidefinite programming
- Lagrangian relaxation
- Branch-and-bound