The stiffness of systems of ODEs that arise from spatial discretization of PDEs causes difficulties for both explicit and implicit time-stepping methods. Krylov Subspace Spectral (KSS) methods present a balance between the efficiency of explicit methods and the stability of implicit methods by computing each Fourier coefficient from an individualized approximation of the solution operator of the PDE. While KSS methods are explicit methods that exhibit a high order of accuracy and stability similar to that of implicit methods, their efficiency needs to be improved. A previous asymptotic study of block Lanczos iteration yielded estimates of extremal block Gaussian quadrature nodes for each Fourier component and led to an improvement in efficiency. Here, a more detailed asymptotic study is performed in order to rapidly estimate all nodes, thus drastically reducing computational expense without sacrificing accuracy.

Exponential propagation iterative (EPI) methods provide an efficient approach to the solution of large stiff nonlinear systems of ODE, compared to standard integrators. However, the bulk of the computational effort in these methods is due to products of matrix functions and vectors, which can become very costly at high resolution due to an increase in the number of Krylov projection steps needed to maintain accuracy. Here, it is proposed to modify EPI methods by using KSS methods, instead of standard Krylov projection methods, to compute products of matrix functions and vectors. Numerical experiments demonstrate that this modification causes the number of Krylov projection steps to become bounded independently of the grid size, thus dramtically improving efficiency and scalability.

1.
Atkinson
,
K.
:
An Introduction to Numerical Analysis
, 2nd Ed.
Wiley
(
1989
)
2.
Cibotarica
,
A.
,
Lambers
,
J. V.
,
Palchak
,
E.
:
High-Order Time-Stepping for Nonlinear PDE through Rapid Estimation of Block Gaussian Quadrature Nodes
.
Submitted
.
3.
Golub
,
G. H.
,
Meurant
,
G.
:
Matrices, Moments and Quadrature
.
Proceedings of the 15th Dundee Conference
,
June-July 1993
,
Griffiths
,
D. F.
,
Watson
,
G. A.
(eds.),
Longman Scientific & Technical
(
1994
)
4.
Golub
,
G. H.
,
Underwood
,
R.
: The block Lanczos method for computing eigenvalues.
Mathematical Software III
,
J.
Rice
Ed., (
1977
)
361
377
.
5.
Guidotti
,
P.
,
Kim
,
Y.
,
Lambers
,
J. V.
:
Image Restoration with a New Class of Forward-Backward Diffusion Equations of Perona-Malik Type
.
SIAM Journal on Imaging Sciences
6
(
3
) (
2013
)
1416
1444
.
6.
Hochbruck
,
M.
,
Lubich
,
C.
:
A Gautschi-type method for oscillatory second-order differential equations
,
Numer. Math.
83
(
1999
)
403
426
.
7.
Hochbruck
,
M.
,
Lubich
,
C.
:
On Krylov Subspace Approximations to the Matrix Exponential Operator
.
SIAM J. Numer. Anal.
34
(
1996
)
1911
1925
.
8.
Hochbruck
,
M.
,
Lubich
,
C.
,
Selhofer
,
H.
:
Exponential Integrators for Large Systems of Differential Equations
.
SIAM J. Sci. Comput.
19
(
1998
)
1552
1574
.
9.
Lambers
,
J. V.
:
Enhancement of Krylov Subspace Spectral Methods by Block Lanczos Iteration
.
Electron. T. Numer. Ana.
31
(
2008
)
86
109
.
10.
Lambers
,
J. V.
:
An Explicit, Stable, High-Order Spectral Method for the Wave Equation Based on Block Gaussian Quadrature
.
IAENG Journal of Applied Mathematics
38
(
2008
)
333
348
.
11.
Lambers
,
J. V.
:
Implicitly Defined High-Order Operator Splittings for Parabolic and Hyperbolic Variable-Coefficient PDE Using Modified Moments
.
International Journal of Computational Science
2
(
2008
)
376
401
.
12.
Loffeld
,
J.
,
Tokman
,
M.
:
Comparative Performance of Exponential, Implicit and Explicit Integrators for Stiff Systems of ODEs
.
Submitted
.
13.
Palchak
,
E. M.
,
Cibotarica
,
A.
,
Lambers
,
J. V.
:
Solution of Time-Dependent PDE Through Rapid Estimation of Block Gaussian Quadrature Nodes
.
Submitted
.
14.
Tokman
,
M.
:
Efficient integration of large stiff systems of ODEs with exponential propagation iterative (EPI) methods
.
J. Comp. Phys.
213
(
2006
)
748
776
.
This content is only available via PDF.
You do not currently have access to this content.