Exponential Rosenbrock methods of order five - construction, analysis and numerical comparisons

Authors:
Vu Thai Luan;Alexander Ostermann
Affiliations:
-;-
Venue:
Journal of Computational and Applied Mathematics
Year:
2014

Citing 13
Cited 0

An iterative solution method for solving f(A)x=b, using Krylov subspace information obtained for the symmetric positive definite matrix A

Journal of Computational and Applied Mathematics
On Krylov Subspace Approximations to the Matrix Exponential Operator

SIAM Journal on Numerical Analysis
Exponential Integrators for Large Systems of Differential Equations

SIAM Journal on Scientific Computing
Exponential time differencing for stiff systems

Journal of Computational Physics
Fourth-Order Time-Stepping for Stiff PDEs

SIAM Journal on Scientific Computing
Generalized integrating factor methods for stiff PDEs

Journal of Computational Physics
Explicit Exponential Runge-Kutta Methods for Semilinear Parabolic Problems

SIAM Journal on Numerical Analysis
Efficient integration of large stiff systems of ODEs with exponential propagation iterative (EPI) methods

Journal of Computational Physics
Implementation of exponential Rosenbrock-type integrators

Applied Numerical Mathematics
Exponential Runge--Kutta methods for parabolic problems

Applied Numerical Mathematics
Exponential Rosenbrock-Type Methods

SIAM Journal on Numerical Analysis
A new class of exponential propagation iterative methods of Runge-Kutta type (EPIRK)

Journal of Computational Physics
Computing the Action of the Matrix Exponential, with an Application to Exponential Integrators

SIAM Journal on Scientific Computing

Quantified Score

Hi-index	7.29

Visualization

Abstract

The aim of this paper is to construct high-order exponential Rosenbrock methods and to analyze their convergence properties for the time discretization of large-scale systems of stiff differential equations. We present a new and simple approach for deriving the stiff order conditions. These conditions allow us to construct new pairs of embedded methods of high order. As an example, we present a fifth-order method with five stages. For particular problems the order conditions can be simplified. It is then even possible to construct a method of order 5 with three stages only. The error analysis is performed in an abstract framework of strongly continuous semigroups that allows us to treat semilinear evolution equations in Banach spaces. Convergence results are proved for variable step size implementations. To demonstrate the efficiency of the new integrators, we give some numerical experiments in MATLAB. In particular, numerical comparisons for semilinear parabolic PDEs in one and two space dimensions are presented.