Repeated matrix squaring for the parallel solution of linear systems

  • Bruno Codenotti
  • Mauro Leoncini
  • Giovanni Resta
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 605)


Given a n×n nonsingular linear system Ax=b, we prove that the solution x can be computed in parallel time ranging from Ω(log n) to O(log2n), provided that the condition number, μ(A), of A is bounded by a polynomial in n. In particular, if μ(A) =O(1), a time bound O(log n) is achieved. To obtain this result, we reduce the computation of x to repeated matrix squaring and prove that a number of steps independent of n is sufficient to approximate x up to a relative error 2−d, d=O(1). This algorithm has both theoretical and practical interest, achieving the same bound of previously published parallel solvers, but being far more simple.


Linear System Condition Number Spectral Radius Parallel Solution Arithmetic Circuit 
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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    S. J. Berkowitz. On computing the determinant in small parallel time using a small number of processors. Inform. Process. Lett., 18:147–150, 1984.CrossRefGoogle Scholar
  2. 2.
    B. Codenotti. Parallel solution of linear systems by repeated squaring. Applied Math. Letters, 3:19–20, 1990.CrossRefMathSciNetGoogle Scholar
  3. 3.
    B. Codenotti and F. Flandoli. A monte carlo method for the parallel solution of linear systems. Journal of Complexity, 5:107–117, 1989.CrossRefGoogle Scholar
  4. 4.
    B. Codenotti and M. Leoncini. Parallel Complexity of Linear System Solution. World Scientific Pu. Co., Singapore, 1991.Google Scholar
  5. 5.
    S. A. Cook. A taxonomy of problems with fast parallel algorithms. Inform. and Control, 64:2–22, 1985.CrossRefGoogle Scholar
  6. 6.
    D. Coppersmith and S. Winograd. Matrix multiplication via arithmetic progression. In Proc. 19th Annual ACM Symposium on Theory of Computing, pages 1–6, Berkeley, CA, 1987. Springer-Verlag.Google Scholar
  7. 7.
    M. Cosnard and Y. Robert. Complexity of parallel QR factorization. J. Assoc. Cornput. Mach., 33:712–723, 1986.Google Scholar
  8. 8.
    L. Csanky. Fast parallel matrix inversion algorithms. SIAM J. Comput., 5:618–623, 1976.CrossRefGoogle Scholar
  9. 9.
    J. von zur Gathen. Parallel arithmetic computations: a survey. In Lecture notes in Computer Science, volume 233, pages 93–122. Springer-Verlag, New-York, 1986.Google Scholar
  10. 10.
    J. von zur Gathen. Parallel linear algebra. In J. Reif, editor, Synthesis of Parallel Algorithms. 1991. to appear.Google Scholar
  11. 11.
    G. H. Golub and C. F. van Loan. Matrix Computations. Johns Hopkins University Press, Baltimore, Md, 1989.Google Scholar
  12. 12.
    V. Pan. Complexity of parallel matrix computations. Theoret. Comput. Sci., 54:65–85, 1987.CrossRefGoogle Scholar
  13. 13.
    V. Pan and J. Reif. Efficient parallel solution of linear systems. In Proc. 17th Annual ACM Symposium on Theory of Computing, pages 143–152, 1985.Google Scholar
  14. 14.
    V. Pan and J. Reif. Fast and efficient parallel solution of dense linear systems. Comput. and Math. with Appl., 17:1481–1491, 1989.CrossRefGoogle Scholar
  15. 15.
    F. P. Preparata. Inverting a Vandermonde matrix in minimum parallel time. Inform. Process. Lett., 38:291–294, 1991.CrossRefMathSciNetGoogle Scholar
  16. 16.
    A. H. Sameh and D. J. Kuck. On stable parallel linear systems solvers. J. Assoc. Comput. Mach., 25:81–91, 1978.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1992

Authors and Affiliations

  • Bruno Codenotti
    • 1
  • Mauro Leoncini
    • 2
  • Giovanni Resta
    • 2
  1. 1.Int. Comp. Sci. Instit.Berkeley
  2. 2.Ist. di Elaborazione dell'InformazioneConsiglio Nazionale delle RicerchePisaItaly

Personalised recommendations