Strong Convergence Rates for a Full Discretization of Stochastic Wave Equation with Nonlinear Damping

Xiaojie Wang (王小捷), Meng Cai, David Cohen, Xiaojie Wang

TL;DR

This paper develops and analyzes a new numerical method for solving a complex type of partial differential equation (the stochastic wave equation with nonlinear damping). The authors prove that their method accurately approximates the true solution, achieving specific convergence rates in both space and time, which is crucial for reliable simulations of phenomena like vibrating strings or DNA strands.

What Problem Does This Paper Address?

The paper tackles the challenge of accurately simulating stochastic wave equations (SWEs) that include a nonlinear damping term. These equations are important for modeling physical systems where energy dissipation is significant and depends non-linearly on the system's velocity. While numerical methods for simpler versions of SWEs exist, the presence of the super-linearly growing damping term introduces significant mathematical hurdles, making error analysis and proving convergence rates difficult. Existing methods often struggle with these nonlinearities, especially when the noise is not globally Lipschitz.

Methods

The authors propose a "full discretization" approach, meaning they discretize both space and time.

  • Spatial Discretization: They employ a spectral Galerkin method. This method uses a basis of eigenfunctions of the Laplace operator to represent the spatial solution. This is a standard technique for problems with smooth solutions and regular domains, offering good accuracy.
  • Temporal Discretization: They introduce a "modified implicit exponential Euler scheme." This is a novel time-stepping method designed to handle the nonlinear damping term. The "implicit" nature means that the solution at the next time step depends on itself, requiring iterative solutions, while the "exponential Euler" part suggests a specific way of handling the linear terms to improve stability and accuracy.

The core of their analysis involves first establishing upper mean-square error bounds and then deriving mean-square convergence rates. A key innovation is achieving these results without needing overly strict assumptions on the moments of the numerical solution.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and numerical experiments to validate the theory, rather than using external datasets.

Key Insights and Ideas

  • Handling Super-linear Damping: The primary challenge is the nonlinear damping term, which grows faster than linearly with the velocity. This non-globally Lipschitz behavior complicates standard error analysis techniques. The paper's success in deriving convergence rates under these conditions is a significant contribution.
  • Dimension-Specific Analysis: The authors highlight that the analysis for the 2D case is more intricate than for 1D. This is due to the failure of certain Sobolev embedding inequalities in higher dimensions, necessitating a more subtle approach that leads to a slightly reduced convergence rate (an infinitesimal factor).
  • No A Priori High-Order Moment Bounds: A crucial aspect of their method is that it doesn't require proving that high-order moments of the numerical solution are bounded beforehand. This simplifies the analysis and makes the method more broadly applicable.
  • Spectral Galerkin Advantage: The paper notes that the spectral Galerkin method is preferred over finite element methods for this problem because the projection operator in the spectral method preserves certain desirable properties of the nonlinearity, which is not the case for finite element projections.

Key Results

The paper establishes strong convergence rates for the proposed fully discrete scheme:

  • Dimension 1: The scheme achieves a convergence rate of $O(N^{-1/2})$ in space and $O(\tau)$ in time, where $N$ is related to the spatial discretization parameter and $\tau$ is the time step size.
  • Dimension 2: The scheme achieves a convergence rate of $O(N^{-1/2 + \epsilon})$ in space and $O(\tau^{1-\epsilon})$ in time, where $\epsilon$ is an arbitrarily small positive parameter. This reflects the increased difficulty in higher dimensions.

Numerical experiments are presented and confirm these theoretical findings, showing that the errors decrease as predicted by the derived rates.

Limitations and Future Work

  • Higher Spatial Dimensions (d ≥ 3): The error analysis for dimensions three and higher is noted as non-trivial and requires further investigation.
  • Multiplicative Noise: The paper only considers additive noise. Extending the analysis to multiplicative noise, especially with non-globally Lipschitz nonlinearities, is identified as a challenging but important future direction.
  • Complex Domains: The current spectral Galerkin method relies on finding eigenvalues and eigenfunctions of the Laplace operator, which is difficult for complex domains. Exploring finite element methods (FEM) for such cases is suggested, though their error analysis is more involved.
  • Computational Cost: The proposed scheme involves computations that might be costly, prompting research into alternative schemes or optimizations.
  • Taming Techniques: The authors suggest exploring "taming techniques" for nonlinearities, which have shown promise in other SPDE contexts, to potentially develop more efficient numerical schemes.

Summary updated

Abstract

The paper establishes the strong convergence rates of a spatio-temporal full discretization of the stochastic wave equation with nonlinear damping in dimension one and two. We discretize the SPDE by applying a spectral Galerkin method in space and a modified implicit exponential Euler scheme in time. The presence of the super-linearly growing damping in the underlying model brings challenges into the error analysis. To address these difficulties, we first achieve upper mean-square error bounds, and then obtain mean-square convergence rates of the considered numerical solution. This is done without requiring the moment bounds of the full approximations. The main result shows that, in dimension one, the scheme admits a convergence rate of order $\tfrac12$ in space and order $1$ in time. In dimension two, the error analysis is more subtle and can be done at the expense of an order reduction due to an infinitesimal factor. Numerical experiments are performed and confirm our theoretical findings.
BibTeX
@article{2025strong,
  author = {Xiaojie Wang (王小捷) and Meng Cai and David Cohen and Xiaojie Wang},
  title = {{Strong Convergence Rates for a Full Discretization of Stochastic Wave Equation with Nonlinear Damping}},
  journal = {J. Sci. Comput},
  year = {2025},
  doi = {10.1007/s10915-024-02758-0},
}
Figure