A priori and a posteriori error estimates of a π’ž 0 \mathcal{C}^{0} -in-time method for the wave equation in second order formulation

Z. Dong, L. Mascotto, Z. Wang Β· 2024

Read A priori and a posteriori error estimates of a π’ž 0 \mathcal{C}^{0} -in-time method for the wave equation in second order formulation by Z. Dong, L. Mascotto, Z. Wang (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2411-03264.txt.

Description

Abstract We establish fully-discrete a priori and semi-discrete in time a posteriori error estimates for a discontinuous-continuous Galerkin discretization of the wave equation in second order formulation; the resulting method is a Petrov-Galerkin scheme based on piecewise polynomial test functions and continuous piecewise polynomial trial functions in time, respectively.

Catalog record

CREATORZ. Dong, L. Mascotto, Z. Wang
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE125 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2411-03264

Browse the full library Β· About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A priori and a posteriori error estimates of a π’ž 0 \mathcal{C}^{0} -in-time method for the wave equation in second order formulation Z. Dong, L. Mascotto, Z. Wang Abstract We establish fully-discrete a priori and semi-discrete in time a posteriori error estimates for a discontinuous-continuous Galerkin discretization of the wave equation in second order formulation; the resulting method is a Petrov-Galerkin scheme based on piecewise polynomial test functions and continuous piecewise polynomial trial functions in time, respectively. Crucial tools in the a priori analysis for the fully-discrete formulation are the design of suitable projection and interpolation operators extending those used in the parabolic setting, and stability estimates based on a nonstandard choice of the test function; a priori estimates are shown, which are measured in L ∞ L^{\infty} -type norms in time. For the semi-discrete in time formulation, we exhibit reliable a posteriori error estimates for the error measured in the L ∞ ​ ( L 2 ) L^{\infty}(L^{2}) norm with fully explicit constants; to this aim, we design a reconstruction operator into π’ž 1 \mathcal{C}^{1} piecewise polynomials over the time grid with optimal approximation properties in terms of the polynomial degree distribution and the time steps. Numerical examples illustrate the theoretical findings. AMS subject classification : 65M50; 65M60; 65J10. Keywords : wave equation; discontinuous Galerkin method; adaptive algorithm; h ​ p hp -analysis; a posteriori error analysis. 1 Introduction We establish fully-discrete a priori and semi-discrete in time a posteriori error estimates for a \mathcal{C}^{0} -in-time method, see, e.g., [ 37 ] , approximating solutions to the wave equation in second order formulation, which are explicit in the spatial mesh size, the time steps, and the polynomial degrees. Formulation, a priori estimates, and minimal literature. The \mathcal{C}^{0} -in-time method we are interested in is based on a second order formulation of the wave equation. Compared to several references where first order systems are considered [ 13 , 23 , 14 , 25 ] , the dimension of the corresponding discrete spaces is smaller for fixed time steps. The method lies in between a fully DG and a \mathcal{C}^{1} schemes in time: it employs piecewise polynomial test functions and continuous piecewise polynomial trial functions in time, respectively; upwind terms involving first time derivatives in time of the trial functions are included in the formulation. The polynomial degree in time of the trial functions is larger by 1 than that for the test functions, leading to square systems for each time interval, which are solved sequentially as a time marching scheme. A key tool in deriving stability and a priori error estimates is the choice of an appropriate test function. For instance, in [ 37 ] , a higher order fully-discrete version of the test function in [ 3 ] is used, leading to stability estimates in the L^{\infty} -type norms in time. Nonlinear problems are also discretized with this approach [ 18 ] . A posteriori estimates and minimal literature. A posteriori error estimates are well established for elliptic problems and a considerable amount of work is available for parabolic problems as well; on the other hand, hyperbolic (and in particular wave) problems are less explored. A posteriori error estimates for wave problems in second order formulation are studied in [ 1 ] and later rigorously analysed in [ 4 ] ; first order systems are instead the topic of [ 27 , 25 , 33 ] ; a posteriori error estimates measured in the L^{\infty}(L^{2}) norm are investigated for different time stepping schemes in [ 16 , 15 , 19 ] . A posteriori error estimates that are reliable and efficient in the same norm are instead investigated in [ 6 , 7 ] . Features of the \mathcal{C}^{0} -in-time method. Compared to fully DG schemes in time, the \mathcal{C}^{0} -in-time method involves fewer unknowns; in view of the a posteriori error analysis for the semi-discrete in time version, one can employ tools from the parabolic setting and deduce a posteriori bounds that are fully explicit with respect to the polynomial degree. On the other hand, a modification of the scheme seems to be suited for dynamic mesh changes in space [ 26 , 9 ] . First main contribution of the manuscript. We modify the analysis in [ 37 ] so as to have explicit dependence on the polynomial degree for the a priori analysis of the fully-discrete scheme. We consider uniform polynomial degree in the spatial discretization and possibly nonuniform polynomial degree in the time discretization. Static, locally quasi-uniform meshes in space are considered, allowing for small elements that in explicit schemes would impact on the CFL condition [ 21 ] . Our analysis hinges upon deriving stability estimates for the scheme, which are explicit in the polynomial degrees; see Section 2.1 . Based on such stability estimates and the properties of an integrated ThomΓ©e operator discussed in Section 2.2 , a priori error estimates are proven in Section 2.3 . Error estimates are obtained in Section 2.4 , which are explicit in the spatial mesh size, the time steps, and the polynomial degrees. Optimal rates for the error measured with respect to L^{\infty} -type norms in time is shown for sufficiently regular solutions and data for fixed polynomial degrees in time and space. Second main contribution of the manuscript. For the semi-discrete in time method, we design an error estimator satisfying a posteriori error estimates, with explicit dependence on the polynomial degree distribution in time. The crucial tool in a posteriori error estimates for time-dependent problems is the derivation of a reconstruction operator into smoother spaces. The original idea in the context of parabolic problems traces back to Makridakis and Nochetto [ 27 ] ; the corresponding p -version analysis is detailed in [ 31 ] and later in [ 22 ] . We design a related operator in the wave equation setting and derive p -optimal approximation estimates in several norms in Section 3.2 ; such an operator is instrumental for designing an error estimator that is reliable for the error measured in the L^{\infty}(L^{2}) norm; see Section 3.3 ; in the recent work [ 9 ] , a posteriori error estimates in that norm are derived, which are valid also for dynamic mesh changes. The upper bound is explicit in the polynomial degree distribution and the time steps, without unknown constants. Since the test and trial test functions have different polynomial degrees in time, the a posteriori error bounds involve extra oscillation terms compared to the parabolic setting. While most references focus on explicit time-stepping, we focus on implicit approaches; they are more appropriate for the analysis of coupled wave and parabolic problems, such as fluid-structure interactions, which we plan to study in future work, and allow for elements of different sizes for different regimes without influencing the local time-steps. To the best of our knowledge, we provide for the first time in the literature fully explicit, reliable a posteriori error estimates for a semi-discrete in time method for the approximation of solutions to the wave equation in second order formulation, which are explicit in the polynomial degree distribution in time and optimal in the time steps. The proposed analysis does not hinge upon

Read the complete text