A branch-and-cut algorithm based on semidefinite programming for the minimum k-partition problem

The minimum k-partition (MkP) problem is the problem of partitioning the set of vertices of a graph into k disjoint subsets so as to minimize the total weight of the edges joining vertices in the same partition. The main contribution of this paper is the design and implementation of a branch-and-cut algorithm based on semidefinite programming (SBC) for the MkP problem. The two key ingredients for this algorithm are: the combination of semidefinite programming with polyhedral results; and a novel iterative clustering heuristic (ICH) that finds feasible solutions for the MkP problem. We compare ICH to the hyperplane rounding techniques of Goemans and Williamson and of Frieze and Jerrum, and the computational results support the conclusion that ICH consistently provides better feasible solutions for the MkP problem. ICH is used in our SBC algorithm to provide feasible solutions at each node of the branch-and-bound tree. The SBC algorithm computes globally optimal solutions for dense graphs with up to 60 vertices, for grid graphs with up to 100 vertices, and for different values of k, providing a fast exact approach for k≥3.

[1]  Franz Rendl,et al.  Graph partitioning using linear and semidefinite programming , 2003, Math. Program..

[2]  Martin Grötschel,et al.  An Application of Combinatorial Optimization to Statistical Physics and Circuit Layout Design , 1988, Oper. Res..

[3]  David R. Karger,et al.  Approximate graph coloring by semidefinite programming , 1998, JACM.

[4]  Vipin Kumar,et al.  Multilevel Graph Partitioning Schemes , 1995, ICPP.

[5]  Jos F. Sturm,et al.  A Matlab toolbox for optimization over symmetric cones , 1999 .

[6]  Angelika Wiegele.  Nonlinear optimization techniques applied to combinatorial optimization problems / Angelika Wiegele , 2006 .

[7]  Josep Domingo-Ferrer,et al.  Practical Data-Oriented Microaggregation for Statistical Disclosure Control , 2002, IEEE Trans. Knowl. Data Eng..

[8]  Miguel F. Anjos,et al.  New Convex Relaxations for the Maximum Cut and VLSI Layout Problems , 2001 .

[9]  Franz Rendl,et al.  A Branch and Bound Algorithm for Max-Cut Based on Combining Semidefinite and Polyhedral Relaxations , 2007, IPCO.

[10]  Ali Ridha Mahjoub,et al.  On the cut polytope , 1986, Math. Program..

[11]  Michel Deza,et al.  Geometry of cuts and metrics , 2009, Algorithms and combinatorics.

[12]  B. Borchers.  CSDP, A C library for semidefinite programming , 1999 .

[13]  M. Mézard,et al.  Spin Glass Theory and Beyond , 1987 .

[14]  F. Guerra.  Spin Glasses , 2005, cond-mat/0507581.

[15]  Renato D. C. Monteiro,et al.  A nonlinear programming algorithm for solving semidefinite programs via low-rank factorization , 2003, Math. Program..

[16]  D. V. Pasechnik,et al.  On approximate graph colouring and MAX-k-CUT algorithms based on the theta-function , 2002 .

[17]  Helmut G. Katzgraber,et al.  Critical behavior of the three- and ten-state short-range Potts glass: A Monte Carlo study , 2006 .

[18]  Robert J. Vanderbei,et al.  An Interior-Point Method for Semidefinite Programming , 1996, SIAM J. Optim..

[19]  P. Gritzmann,et al.  Applied geometry and discrete mathematics : the Victor Klee festschrift , 1991 .

[20]  David P. Williamson,et al.  New 3/4-Approximation Algorithms for the Maximum Satisfiability Problem , 1994, SIAM J. Discret. Math..

[21]  Marc E. Pfetsch,et al.  Orbitopal Fixing , 2007, IPCO.

[22]  László Lovász,et al.  On the Shannon capacity of a graph , 1979, IEEE Trans. Inf. Theory.

[23]  Andreas Eisenblätter,et al.  The Semidefinite Relaxation of the k -Partition Polytope Is Strong , 2002, IPCO.

[24]  J. Mitchell.  Realignment in the National Football League: Did they do it right? , 2003 .

[25]  M. R. Rao,et al.  Facets of the K-partition Polytope , 1995, Discret. Appl. Math..

[26]  M. R. Rao,et al.  The partition problem , 1993, Math. Program..

[27]  Takao Asano,et al.  Approximation Algorithms for the Maximum Satisfiability Problem , 1996, Nord. J. Comput..

[28]  Martin Grötschel,et al.  Complete Descriptions of Small Multicut Polytopes , 1990, Applied Geometry And Discrete Mathematics.

[29]  David P. Williamson,et al.  Approximation algorithms for MAX-3-CUT and other problems via complex semidefinite programming , 2001, STOC '01.

[30]  Michael Stingl,et al.  PENNON: A code for convex nonlinear and semidefinite programming , 2003, Optim. Methods Softw..

[31]  Martin Grötschel,et al.  Clique-Web Facets for Multicut Polytopes , 1992, Math. Oper. Res..

[32]  H. Rieger,et al.  New Optimization Algorithms in Physics , 2004 .

[33]  Kim-Chuan Toh,et al.  SDPT3 -- A Matlab Software Package for Semidefinite Programming , 1996 .

[34]  G. Reinelt,et al.  2 Computing Exact Ground Statesof Hard Ising Spin Glass Problemsby Branch-and-Cut , 2005 .

[35]  Jean B. Lasserre,et al.  An Explicit Equivalent Positive Semidefinite Program for Nonlinear 0-1 Programs , 2002, SIAM J. Optim..

[36]  Frauke Liers.  Contributions to Determining Exact Ground-States of Ising Spin-Glasses and to their Physics , 2004 .

[37]  Alan M. Frieze,et al.  Improved Approximation Algorithms for MAX k-CUT and MAX BISECTION , 1995, IPCO.

[38]  A. Eisenblätter.  Frequency Assignment in GSM Networks: Models, Heuristics, and Lower Bounds , 2001 .

[39]  Franz Rendl,et al.  A Boundary Point Method to Solve Semidefinite Programs , 2006, Computing.

[40]  Endre Boros,et al.  The max-cut problem and quadratic 0–1 optimization; polyhedral aspects, relaxations and bounds , 1991, Ann. Oper. Res..

[41]  Michael Jünger,et al.  Experiments in quadratic 0–1 programming , 1989, Math. Program..

[42]  Michael Jünger,et al.  Minimizing breaks by maximizing cuts , 2003, Oper. Res. Lett..

[43]  J. P. Warners,et al.  On Approximate Graph Colouring and MAX-k-CUT Algorithms Based on the θ-Function , 2004, J. Comb. Optim..

[44]  Kim-Chuan Toh,et al.  The Chebyshev Polynomials of a Matrix , 1999, SIAM J. Matrix Anal. Appl..

[45]  C. Helmberg,et al.  Solving quadratic (0,1)-problems by semidefinite programs and cutting planes , 1998 .