Advertisement

Cutting down on Fill Using Nested Dissection: Provably Good Elimination Orderings

  • Ajit Agrawal
  • Philip Klein
  • R. Ravi
Part of the The IMA Volumes in Mathematics and its Applications book series (IMA, volume 56)

Abstract

In the last two decades, many heuristics have been developed for finding good elimination orderings for sparse Cholesky factorization. These heuristics aim to find elimination orderings with either low fill, low operation count, or low elimination height. Though many heuristics seem to perform well in practice, there has been a marked absence of much theoretical analysis to back these heuristics. Indeed, few heuristics are known to provide any guarantee on the quality of the elimination ordering produced for arbitrary matrices.

In this work, we present the first polynomial-time ordering algorithm that guarantees approximately optimal fill. Our algorithm is a variant of the well-known nested dissection algorithm. Our ordering performs particularly well when the number of elements in each row (and hence each column) of the coefficient matrix is small. Fortunately, many problems in practice, especially those arising from finite-element methods, have such a property due to the physical constraints of the problems being modeled.

Our ordering heuristic guarantees not only low fill, but also approximately optimal operation count, and approximately optimal elimination height. Elimination orderings with small height and low fill are of much interest when performing factorization on parallel machines. No previous ordering heuristic guaranteed even small elimination height.

We will describe our ordering algorithm and prove its performance bounds. We shall also present some experimental results comparing the quality of the orderings produced by our heuristic to those produced by two other well-known heuristics.

Keywords

SIAM Journal Gaussian Elimination Input Graph Chordal Graph Performance Guarantee 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    A. Agrawal, “Network Design and Network Cut Dualities: Approximation Algorithms and Applications,” Ph.D. thesis, Technical Report CS-91–60, Brown University (1991).Google Scholar
  2. [2]
    H. L. Bodlaender, J. R. Gilbert, H. Hafsteinsson and T. Kloks, “Approximating treewidth, pathwidth, and minimum elimination tree height,” Technical Report CSL-90–01, Xerox Corporation, Palo Alto Research Center (1990).Google Scholar
  3. [3]
    E. Cuthill, and J. McKee, “Reducing the bandwidth of sparse symmetric matrices,” Proceedings of the 24th National Conference of the ACM (1969), pp. 157–172.Google Scholar
  4. [4]
    I. S. Duff, A. M. Erisman, and J. K. Reid, “On George’s nested dissection method,” SIAM Journal on Numerical Analysis, vol. 13 (1976), pp. 686–695.MathSciNetMATHCrossRefGoogle Scholar
  5. [5]
    I. Duff, N. Gould, M. Lescrenier, and J. K. Reid, “The multifrontal method in a parallel environment,” in Advances in Numerical Computation, M. Cox and S. Hammarling, eds., Oxford University Press (1990).Google Scholar
  6. [6]
    I. Duff, R. Grimes, and J. G. Lewis, “Users’ guide for the Harwell-Boeing sparse matrix collection,” Manuscript (1988).Google Scholar
  7. [7]
    I. Duff, R. Grimes, and J. G. Lewis, “Sparse matrix test problems,” ACM Transactions on Mathematical Software, vol. 15 (1989), pp. 1–14.MATHCrossRefGoogle Scholar
  8. [8]
    I. Duff, and J. K. Reid, “The multifrontal solution of indefinite sparse symmetric linear equations,” ACM Transactions on Mathematical Software, vol. 9 (1983), pp. 302–325.MathSciNetMATHCrossRefGoogle Scholar
  9. [9]
    I. Duff, and J. K. Reid, Direct Methods for Sparse Matrices, Oxford University Press (1986).Google Scholar
  10. [10]
    K. A. Gallivan et al. Parallel Algorithms for Matrix Computations, SIAM (1990).Google Scholar
  11. [11]
    M. R. Garey and D. S. Johnson, Computers and Intractability: A guide to the theory of NP-completeness, W. H. Freeman, San Francisco (1979).Google Scholar
  12. [12]
    George, J. A., “Computer implementation of a finite element method,” Tech. Report STANCS-208, Stanford University (1971).Google Scholar
  13. [13]
    George, J. A., “Block elimination of finite element system of equations,” in Sparse Matrices and Their Applications, D. J. Rose and R. A. Willoughby, eds., Plenum Press (1972).Google Scholar
  14. [14]
    George, J. A., “Nested Dissection of a regular finite element mesh,” SIAM Journal on Numerical Analysis 10 (1973), pp. 345–367.MathSciNetMATHCrossRefGoogle Scholar
  15. [15]
    George, J. A., “An automatic one-way dissection algorithm for irregular finite-element problems,” SIAM Journal on Numerical Analysis, vol. 17 (1980), pp. 740–751.MathSciNetMATHCrossRefGoogle Scholar
  16. [16]
    George, J. A., and J. W. Liu, “An automatic nested dissection algorithm for irregular finite-element problems,” SIAM Journal on Numerical Analysis, vol. 15 (1978), pp. 1053–1069.MathSciNetMATHCrossRefGoogle Scholar
  17. [17]
    George, J. A., and J. W. Liu, Computer Solution of Large Sparse Positive Definite Systems, Prentice-Hall Inc. (1981).Google Scholar
  18. [18]
    George, J. A., and J. W. Liu, “The evolution of the minimum degree ordering algorithm,” SIAM Review, vol. 31 (1989), pp. 1–19.MathSciNetMATHCrossRefGoogle Scholar
  19. [19]
    George, J. A., J. W. Liu, and E. G. Ng, “User’s guide for SPARSPAK: Waterloo sparse linear equations package,” Tech. Rep. CS78–30 (revised), Dept. of Computer Science, Univ. of Waterloo, Waterloo, Ontario, Canada (1980).Google Scholar
  20. [20]
    N. E. Gibbs, W. G. Poole Jr., and P. K. Stockmeyer, “An algorithm for reducing the bandwidth and profile of a sparse matrix,” SIAM Journal on Numerical Analysis, vol. 13 (1976), pp. 236–250.MathSciNetMATHCrossRefGoogle Scholar
  21. [21]
    J. R. Gilbert, “Some Nested Dissection Order is Nearly Optimal,” Information Processing Letters 26 (1987/88), pp. 325–328.Google Scholar
  22. [22]
    J. R. Gilbert, personal communication (1989).Google Scholar
  23. [23]
    J. R. Gilbert and H. Hafsteinsson, “Approximating treewidth, minimum front size, and minimum elimination tree height,” manuscript, 1989.Google Scholar
  24. [24]
    J. R. Gilbert, D. J. Rose and A. Edenbrandt, “A separator theorem for chordal graphs,” SIAM J. Alg. Disc. Meth. 5 (1984), pp. 306–313.MathSciNetMATHCrossRefGoogle Scholar
  25. [25]
    J. R. Gilbert, and R. Schreiber, “Hightly parallel sparse Cholesky factorization,” Tech. Report CSL-90–7, Xerox Palo Alto Research Center, 1990.Google Scholar
  26. [26]
    J. R. Gilbert, and R. E. Tarjan, “The analysis of a nested dissection algorithm,” Numerische Mathematik, vol. 50 (1987), pp. 377–404.MathSciNetMATHCrossRefGoogle Scholar
  27. [27]
    J. R. Gilbert, and E. Zmijewski, “A parallel graph partitioning algorithm for a message-passing multiprocessor,” International Journal of Parallel Programming, vol. 16 (1987), pp. 427–449.MathSciNetMATHCrossRefGoogle Scholar
  28. [28]
    M. C. Golumbic, Algorithmic Graph Theory and Perfect Graphs, Academic Press, New York (1980).MATHGoogle Scholar
  29. [29]
    A. J. Hoffman, M. S. Martin, and D. J. Rose, “Complexity bounds for regular finite difference and finite element grids,” SIAM Journal on Numerical Analysis, vol. 10 (1973), pp. 364–369.MathSciNetMATHCrossRefGoogle Scholar
  30. [30]
    J. Jess, and H. Kees, “A data structure for parallel L/U decomposition,” IEEE Transactions on Computers, vol. 31 (1982), pp. 231–239.MathSciNetMATHCrossRefGoogle Scholar
  31. [31]
    U. Kjwrulff, “Triangulation of graphs — Algorithms giving small total state space,” R 9009, Institute for Electronic Systems, Department of Mathematics and Computer Science, University of Aalborg (1990).Google Scholar
  32. [32]
    P. N. Klein, “A parallel randomized approximation scheme for shortest paths,” Technical Report CS-91–56, Brown University (1991).Google Scholar
  33. [33]
    P. N. Klein, A. Agrawal, R. Ravi and S. Rao, “Approximation through multicommodity flow,” Proceedings of the 31st Annual IEEE Conference on Foundations of Computer Science, (1990), pp. 726–737.Google Scholar
  34. [34]
    P. N. Klein, and S. Kang, “Approximating concurrent flow with uniform demands and capacities: an implementation,” Technical Report CS-91–58, Brown University (1991).Google Scholar
  35. [35]
    P. Klein, C. Stein and E. Tardos, “Leighton-Rao might be practical: faster approximation algorlthme for concurrent flow with uniform capacities,” Proceedings of the 22nd ACM Symposium on Theory of Computing (1990), pp. 310–321.Google Scholar
  36. [36]
    F. T. Leighton and S. Rao, “An approximate max-flow min-cut theorem for uniform multicommodity flow problems with application to approximation algorithms,” Proceedings of the 29th Annual IEEE Conference on Foundations of Computer Science (1988), pp. 422–431.Google Scholar
  37. [37]
    F. T. Leighton, F. Makedon and S. Tragoudas, personal communication, 1990Google Scholar
  38. [38]
    C. Leiserson, and J. Lewis, “Orderings for parallel sparse symmetric factorization,” in Parallel Processing for Scientific Computing,G. Rodrigue, ed., Philadelphia, PA, 1987, SIAM, pp. 27–32.Google Scholar
  39. [39]
    M. Leuze, “Independent set orderings for parallel matrix factorization by Gaussian elimination,” Parallel Computing, vol. 10 (1989), pp. 177–191.MathSciNetMATHCrossRefGoogle Scholar
  40. [40]
    J. Lewis, B. Peyton, and A. Pothen, “A fast algorithm for reordering sparse matrices for parallel factorization,” SIAM Journal on Scientific and Statistical Computing, vol. 10 (1989), pp. 1156–1173.MathSciNetCrossRefGoogle Scholar
  41. [41]
    R. J. Lipton, D. J. Rose and R. E. Tarjan, “Generalized nested dissection,” SIAM Journal on Numerical Analysis 16 (1979), pp. 346–358.MathSciNetMATHCrossRefGoogle Scholar
  42. [42]
    R. J. Lipton and R. E. Tarjan, “Applications of a planar separator theorem,” SIAM Journal on Computing 9 (1980), pp. 615–627.MathSciNetMATHCrossRefGoogle Scholar
  43. [43]
    J. W. Liu, “Modification of the minimum degree algorithm by multiple elimination,” ACM Transactions on Mathematical Software, vol. 12 (1985), pp. 141–153.CrossRefGoogle Scholar
  44. [44]
    J. W. Liu, “Reordering sparse matrices for parallel elimination,” Parallel Computing, vol. 11 (1989), pp. 73–91.MathSciNetMATHCrossRefGoogle Scholar
  45. [45]
    J. W. Liu, “The minimum degree ordering with constraints,” SIAM Journal on Scientific and Statistical Computing, vol. 10 (1989), pp. 1136–1145.MATHCrossRefGoogle Scholar
  46. [46]
    J. W. Liu, “A graph partitioning algorithm by node separators,” ACM Transactions on Mathematical Software, vol. 15 (1989), pp. 198–219.MATHCrossRefGoogle Scholar
  47. [47]
    J. W. Liu, “The role of elimination trees in sparse factorization,” SIAM Journal on Matrix Analysis and Applications, vol. 11 (1990), pp. 134–172.MathSciNetMATHCrossRefGoogle Scholar
  48. [48]
    J. W. Liu, and A. Mirzaian, “A linear reordering algorithm for parallel pivoting of chordal graphs,” SIAM Journal on Discrete Mathematics, vol. 2 (1989), pp. 100–107.MathSciNetMATHCrossRefGoogle Scholar
  49. [49]
    J. W. Liu, and A. H. Sherman, “Comparative analysis of the Cuthill-McKee and the reverse Cuthill-McKee ordering algorithms for sparse matrices,” SIAM Journal on Numerical Analysis, vol. 13 (1976), pp. 198–213.MathSciNetMATHCrossRefGoogle Scholar
  50. [50]
    F. Makedon, and S. Tragoudas, “Approximating the minimum net expansion: near optimal solutions to circuit partitioning problems,” Manuscript (1991).Google Scholar
  51. [51]
    S. Parter, “The use of linear graphs in Gaussian elimination,” SIAM Review, vol. 3 (1961), pp. 364–369.MathSciNetCrossRefGoogle Scholar
  52. [52]
    F. Peters, “Parallel pivoting algorithms for sparse symmetric matrices,” Parallel Computing, vol. 1 (1984), pp. 99–110.MATHCrossRefGoogle Scholar
  53. [53]
    A. Pothen, “The complexity of optimal elimination trees,” Tech. Report CS-88–16, Department of Computer Science, The Pennsylvania State University, University Park, PA, 1988.Google Scholar
  54. [54]
    D. J. Rose, “Triangulated graphs and the elimination process,” Journal of Math. Anal. Appl. 32 (1970), p. 597–609.MATHCrossRefGoogle Scholar
  55. [55]
    D. J. Rose, “A graph-theoretic study of the numerical solution of sparse positive definite systems of linear equations,” in Graph Theory and Computing, R. C. Read, ed., Academic Press (1972), pp. 183–217.Google Scholar
  56. [56]
    D. J. Rose, R. E. Tarjan and G. S. Lueker, “Algorithmic aspects of vertex elimination on graphs,” SIAM J. Comp. 5 (1976), pp. 266–283.MathSciNetMATHCrossRefGoogle Scholar
  57. [57]
    R. Schreiber, “A new implementation of sparse Gaussian elimination,” ACM Trans. on Mathematical Software 8: 3 (1982), pp. 256–276.MathSciNetMATHCrossRefGoogle Scholar
  58. [58]
    M. Yannakakis, “Computing the minimum fill-in is NP-complete,” SIAM J. Algebraic and Discrete Methods 2 (1981), pp. 77–79.MathSciNetMATHCrossRefGoogle Scholar

Copyright information

© Springer-Verlag New York, Inc. 1993

Authors and Affiliations

  • Ajit Agrawal
    • 1
  • Philip Klein
  • R. Ravi
    • 2
  1. 1.Digital Equipment Corp.Massively Parallel Systems GroupMaynardUSA
  2. 2.Brown UniversityProvidenceUSA

Personalised recommendations