On asymptotic global error estimation and control of finite difference solutions for semilinear parabolic equations
Introduction
We consider semilinear parabolic partial differential equations in space dimensions, where is an elliptic operator, and assume that an appropriate system of boundary conditions and the initial condition are given. The initial boundary value problem is assumed to be well posed and to have a unique continuous solution .
The method of lines is used to solve (1) numerically. We first discretize the PDE in space by means of finite differences of order on a (possibly non-uniform) spatial mesh and solve the resulting system of ODEs using existing time integrators. For simplicity, we shall assume that this system of time-dependent ODEs can be written in the general form with a unique solution vector being a grid function on . Let be the usual restriction operator defined by , where and is the number of all mesh points. Then we take as initial condition .
To simplify the following derivations, we assume that is given by with a finite difference approximation of , and .
To solve the initial value problem (3), we apply a numerical integration method of order at a certain time grid using local control of accuracy. This yields approximations to , which may be calculated for other values of by using a suitable interpolation method provided by the integrator. The global time error is then defined by Numerical experiments in [1] for ODE systems have shown that classical global error estimation based on the first variational equation is remarkably reliable. In addition, having the property of tolerance proportionality, that is, there exists a linear relationship between the global time error and the local accuracy tolerance, can be successfully controlled by a second run with an adjusted local tolerance. Numerous techniques to estimate global errors are described in [2]. A comparison of various adaptive grid methods for partial differential equations and implementation issues are presented in [3], [4].
In order for the method of lines to be used efficiently, it is necessary to take also into account the spatial discretization error. Defining the spatial discretization error by the vector of overall global errors may be written as sum of the global time and spatial error, that is, We assume that is -times continuously differentiable with respect to and -times continuously differentiable with respect to . Then, with maximum step sizes in space and in time it holds for the global space and time error that and , respectively.
Although a posteriori error estimates and adaptive algorithms for the efficient solution of parabolic problems are well established (see e.g. [5], [6] and references therein), the separation of global time and spatial discretization errors is still a challenge. First experiences to estimate and balance the spatial discretization error and the error due to time integration of the ODEs within the method of lines have been made by Schönauer, Schnepf, and Raith [7]. In their control strategy, the spatial mesh is initially chosen and remains fixed. The spatial truncation error is designated to be the level to which the local time error must be adapted. Lawson, Berzins, and Dew [8] proposed to additionally control the local time error with respect to the contribution of the existing error from the previous time steps to the global error at the end of the next time step. The error in time is enabled to vary in relation to the spatial discretization error, ensuring that the method of lines with a fixed spatial mesh is being used efficiently. A successful attempt to assess and to equilibrate the individual discretization errors with respect to a given quantity of interest has been made by Schmich and Vexler [9]. An adjoint linear parabolic problem has to be solved backwards in time to derive useful error bounds, which are used to enhance the resolution in time and space to meet a user-prescribed accuracy tolerance.
It is the purpose of this paper to present a new asymptotic error control strategy for the global errors , based on asymptotic estimates. We will mainly focus on reliability. So our aim is to provide error estimates which are not only asymptotically exact, but also work reliably for moderate tolerances, that is for relatively coarse discretizations. Approximations of the error transport equations for spatial and temporal global errors are derived by using asymptotic estimates that neglect higher order error terms for sufficiently small step sizes in space and time. The approximate global errors are measured in discrete -norms. A priori bounds for the global error in such norms are well known, see e.g. [10], [11]. However, reliable a posteriori error estimation and efficient control of the accuracy of the solution numerically computed to an imposed tolerance level are still challenging. We achieve asymptotic global error control by iteratively improving the temporal and spatial discretizations according to asymptotic estimates of and . The global time error is estimated and controlled along the way fully described in [1]. To estimate the global spatial error, we follow an approach proposed in [12] (see also [8]) and use Richardson extrapolation to set up a linearized error transport equation. Both strategies have to be combined in the right manner in order to make sure that they work reliably. Therefore, we have developed an appropriate control rule for the global spatial error. To control the overall global error more efficiently, we also consider a new fully space–time adaptive approach.
Throughout the paper we will use the terms ‘approximation’ and ‘estimation’ in the sense of asymptotic estimates, i.e., estimates that involve the Landau symbol .
The outline of this paper is as follows: In Section 2, we will linearize the transport equations for the global spatial and the global time error. These contain the residual time error and the spatial truncation error, which are approximated in Sections 3 Approximation of the residual time error, 4 Approximation of the spatial truncation error. In Section 5 we describe the discretization formulas used to approximate the solutions of the error transport equations, as well as the strategies used to adaptively adjust the time step size and the spatial mesh in dependence on the residual time error and the spatial truncation error. Now that we have approximations to the global time and global spatial error, Section 6 suggests strategies to adapt the local tolerances such that in further runs first the global time error and then the global spatial error respect some global tolerances provided by the user. Finally, numerical examples and a summary are given in Sections 7 Numerical illustrations, 8 Summary.
Section snippets
Spatial and time error
By making use of the restriction operator , the spatial truncation error is defined by From (3), (10), it follows that the global spatial error representing the accumulation of the spatial discretization error is the solution of the initial value problem Assuming to be twice continuously differentiable, the mean value theorem for vector functions applied to yields
Approximation of the residual time error
The numerical approximation of the global time error as defined in (15) requires the construction of an appropriate nearby solution which is used in (13) to define the residual time error . The usual way is to construct an interpolatory polynomial from the numerical solutions by using Lagrange or Hermite interpolation. The latter one exploits the fact that with approximations at certain time points also first derivatives are given. In the following
Approximation of the spatial truncation error
An efficient strategy to estimate the spatial truncation error by Richardson extrapolation is proposed in [12]. We will adopt this approach to our setting.
Suppose we are given a second semi-discretization of the PDE system (1), now with doubled local mesh sizes , In practice, one first chooses and constructs then through uniform refinement. We assume that the solution to the discretized PDE on the coarse mesh exists and is unique.
The example discretization formulas
In order to keep the illustration as simple as possible we restrict ourselves to one space dimension. For the spatial discretization of (1) we use standard second-order finite differences. Hence we have . The discrete -norm on a non-uniform mesh for a vector is defined through Here, the components and which are given by the boundary values are not considered.
Adaptive time integration. The example time
The control rules
Like for the ODE case studied in [1] our aim is to provide global error estimates and to control the accuracy of the numerically computed solution to the imposed tolerance level. Let and be the global tolerances. Then we start with the local tolerances , and in the spatially adaptive case also with , and , where the factor ensures that the residual time error is small with respect to the spatial truncation error and therefore the
Numerical illustrations
To illustrate the performance of the global error estimators and the control strategy, we consider three test problems: (i) the highly stable heat equation with nonhomogeneous Neumann boundary conditions [12], (ii) the nonlinear convection-dominated Burgers’ equation [8], [12], and (iii) the Allen–Cahn equation modelling a diffusion–reaction problem [1]. Analytic solutions are known for all three problems. Uniform spatial refinement is studied for all three test cases. For Burgers’ and
Summary
We have developed an error control strategy for finite difference solutions of parabolic equations, involving both temporal and spatial discretization errors. The global time error strategy discussed in [1] appears to provide an excellent starting point for the development of such an algorithm. The classical ODE approach used there and the principle of tolerance proportionality are combined with an efficient estimation of the spatial error and mesh adaptation to control the overall global
References (15)
- et al.
On global error estimation and control for initial value problems
SIAM J. Sci. Comput.
(2007) Thirteen ways to estimate global error
Numer. Math.
(1986)- et al.
Some user-oriented comparisons of adaptive grid methods for partial differential equations in one space dimension
Appl. Numer. Math.
(1998) - et al.
A MATLAB implementation of upwind finite differences and adaptive grids in the method of lines
J. Comput. Appl. Math.
(2005) A fully adaptive MOL-treatment of parabolic 1D-problems with extrapolation techniques
Appl. Numer. Math.
(1996)- et al.
Experiences in designing PDE software with selfadaptive variable step size/order difference methods
Computing
(1984)
Cited by (4)
Unconditional error analysis of linearized BDF2 mixed virtual element method for semilinear parabolic problems on polygonal meshes
2024, Journal of Computational and Applied MathematicsEfficient goal-oriented global error estimators for BDF methods using discrete adjoints
2017, Journal of Computational and Applied MathematicsAdaptive multilevel trust-region methods for timedependent PDE-constrained optimization
2017, Portugaliae Mathematica