Comparing Leja and Krylov Approximations of Large Scale Matrix Exponentials

  • L. Bergamaschi
  • M. Caliari
  • A. Martínez
  • M. Vianello
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3994)


We have implemented a numerical code (ReLPM, Real Leja Points Method) for polynomial interpolation of the matrix exponential propagators exp (\({\it \Delta}\) tA) v and ϕ(\({\it \Delta}\) tA) v, ϕ(z) = (exp (z) – 1)/z. The ReLPM code is tested and compared with Krylov-based routines, on large scale sparse matrices arising from the spatial discretization of 2D and 3D advection-diffusion equations.


Spatial Discretization Polynomial Interpolation Krylov Subspace Superlinear Convergence Homogeneous Dirichlet Boundary Condition 


  1. 1.
    Baglama, J., Calvetti, D., Reichel, L.: Fast Leja points. Electron. Trans. Numer. Anal. 7, 124–140 (1998)MATHMathSciNetGoogle Scholar
  2. 2.
    Bergamaschi, L., Caliari, M., Vianello, M.: The ReLPM exponential integrator for FE discretizations of advection-diffusion equations. In: Bubak, M., van Albada, G.D., Sloot, P.M.A., Dongarra, J. (eds.) ICCS 2004. LNCS, vol. 3039, pp. 434–442. Springer, Heidelberg (2004)CrossRefGoogle Scholar
  3. 3.
    Caliari, M.: Accurate evaluation of divided differences for polynomial interpolation of exponential propagators. Draft (2005)Google Scholar
  4. 4.
    Caliari, M., Vianello, M., Bergamaschi, L.: Interpolating discrete advection-diffusion propagators at Leja sequences. J. Comput. Appl. Math. 172, 79–99 (2004)MATHCrossRefMathSciNetGoogle Scholar
  5. 5.
    Cox, S.M., Matthews, P.C.: Exponential time differencing for stiff systems. J. Comput. Phys. 176, 430–455 (2002)MATHCrossRefMathSciNetGoogle Scholar
  6. 6.
    Druskin, V.L., Knizhnerman, L.A.: Two polynomial methods for calculating functions of symmetric matrices. U.S.S.R. Comput. Math. and Math. Phys. 29, 112–121 (1989)MATHCrossRefMathSciNetGoogle Scholar
  7. 7.
    Hochbruck, M., Lubich, C., Selhofer, H.: Exponential integrators for large systems of differential equations. SIAM J. Sci. Comput. 19, 1552–1574 (1998)MATHCrossRefMathSciNetGoogle Scholar
  8. 8.
    Hochbruck, M., Ostermann, A.: Exponential Runge-Kutta methods for parabolic problems. Appl. Numer. Math. 53, 323–339 (2005)MATHCrossRefMathSciNetGoogle Scholar
  9. 9.
    McCurdy, A., Ng, C., Parlett, B.N.: Accurate Computation of Divided Differences of the Exponential Function. Math. Comp. 43, 501–528 (1984)MATHCrossRefMathSciNetGoogle Scholar
  10. 10.
    Moret, I., Novati, P.: The computation of functions of matrices by truncated Faber series. Numer. Funct. Anal. Optim. 22, 697–719 (2001)MATHCrossRefMathSciNetGoogle Scholar
  11. 11.
    Novati, P.: A polynomial method based on Feiér points for the computation of functions of unsymmetric matrices. Appl. Numer. Math. 44, 201–224 (2003)MATHCrossRefMathSciNetGoogle Scholar
  12. 12.
    Saad, Y.: SPARSKIT: a basic tool kit for sparse matrix computations. Dept. of Computer Science and Engineering, University of Minnesota. Version 2 (2005)Google Scholar
  13. 13.
    Schaefer, M.J.: A polynomial based iterative method for linear parabolic equations. J. Comput. Appl. Math. 29, 35–50 (1990)MATHCrossRefMathSciNetGoogle Scholar
  14. 14.
    Sidje, R.B.: Expokit. A software package for computing matrix exponentials. ACM Trans. Math. Software 24, 130–156 (1998)MATHCrossRefGoogle Scholar
  15. 15.
    Tal-Ezer, H.: Polynomial approximation of functions of matrices and applications. J. Sci. Comput. 4, 25–60 (1989)MATHCrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • L. Bergamaschi
    • 1
  • M. Caliari
    • 2
  • A. Martínez
    • 2
  • M. Vianello
    • 2
  1. 1.Dept. of Math. Methods and ModelsUniversity of Padova 
  2. 2.Dept. of Pure and Appl. Math.University of Padova 

Personalised recommendations