Variable Step-Size BDF3 Method for Allen-Cahn Equation

Minghua Chen, Fan Yu, Qingdong Zhang, Zhimin ZhangJournal of Computational Mathematics 2023 · DOI

TL;DR

This paper presents the first rigorous theoretical analysis of a variable step-size BDF3 method for solving the Allen-Cahn equation. By developing a novel spectral norm inequality, the authors prove unconditional stability and convergence for this method, overcoming challenges posed by non-positive discrete orthogonal convolution kernels and extending the permissible step-size ratio to 1.405.

What Problem Does This Paper Address?

The Allen-Cahn equation is a fundamental model in materials science and phase-field theory, describing phenomena like phase separation and domain coarsening. Solving it numerically often requires adaptive time-stepping to efficiently capture rapid changes in solution behavior. While variable step-size methods for the related BDF2 scheme are well-understood, extending this analysis to the BDF3 method has been a significant challenge. This difficulty arises from the BDF3 method's increased complexity, particularly the non-positive nature of its discrete orthogonal convolution (DOC) kernels, which complicates standard stability and convergence proofs. This paper aims to bridge this gap by providing a rigorous analysis for the variable step-size BDF3 method applied to the Allen-Cahn equation.

Methods

The core of this work lies in the development of a novel spectral norm inequality. This inequality is crucial for overcoming the analytical difficulties associated with the non-positive DOC kernels of the BDF3 method. The authors also leverage existing techniques, such as the Grenander-Szegö theorem and Sylvester criterion, to analyze the properties of matrices derived from the BDF3 scheme. They establish a new, improved step-size ratio restriction of $r_k \le 1.405$ for unconditional stability and convergence. The analysis involves proving unique solvability and a discrete energy dissipation law, which are key components for establishing the method's robustness.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and numerical verification using a standard Allen-Cahn equation with a specific initial value and forcing term.

Key Insights and Ideas

  • Overcoming Non-Positivity: The paper's main breakthrough is the development of a novel spectral norm inequality that effectively handles the non-positive DOC kernels of the BDF3 method, a significant hurdle in previous analyses.
  • Improved Step-Size Ratio: The analysis leads to a substantially improved step-size ratio restriction ($r_k \le 1.405$) for unconditional stability and convergence compared to previous results for BDF3 methods on parabolic problems ($r_k \le 1.199$).
  • First Rigorous Analysis: This work represents the first theoretical analysis of a variable step-size BDF3 method for the Allen-Cahn equation, filling a notable gap in the literature.
  • Energy Stability: The paper demonstrates that the proposed method preserves a discrete energy dissipation law, which is a critical property for the stability of numerical schemes for such equations.

Key Results

The paper rigorously proves the unconditional stability and convergence of the variable step-size BDF3 method for the Allen-Cahn equation under a new step-size ratio restriction of $r_k \le 1.405$. Numerical experiments confirm the theoretical findings, showing third-order accuracy and demonstrating the method's effectiveness. The analysis also establishes the unique solvability and discrete energy dissipation law for the scheme.

Limitations and Future Work

The authors acknowledge that the analysis is primarily theoretical and focused on the Allen-Cahn equation. While the improved step-size ratio is significant, further research could explore even larger ratios or investigate the method's performance on other types of PDEs. The paper also mentions that the DOC kernels for BDF3 are not always positive, which implies that the analytical techniques used here might not be directly applicable to higher-order BDF methods without further development. Future work could extend this analysis to other phase-field models or explore adaptive strategies for selecting the step-size ratios more dynamically.

Summary updated

Abstract

In this work, we analyze the three-step backward differentiation formula (BDF3) method for solving the Allen-Cahn equation on variable grids. For BDF2 method, the discrete orthogonal convolution (DOC) kernels are positive, the stability and convergence analysis are well established in [Liao and Zhang, Math. Comp., 90 (2021), 1207-1226] and [Chen, Yu, and Zhang, arXiv:2108.02910, 2021]. However, the numerical analysis for BDF3 method with variable steps seems to be highly nontrivial due to the additional degrees of freedom and the non-positivity of DOC kernels. By developing a novel spectral norm inequality, the unconditional stability and convergence are rigorously proved under the updated step ratio restriction r k := k / k-1 1.405 for BDF3 method. Finally, numerical experiments are performed to illustrate the theoretical results. To the best of our knowledge, this is the first theoretical analysis of variable steps BDF3 method for the Allen-Cahn equation.
BibTeX
@article{chen2023variable,
  author = {Minghua Chen and Fan Yu and Qingdong Zhang and Zhimin Zhang},
  title = {{Variable Step-Size BDF3 Method for Allen-Cahn Equation}},
  journal = {Journal of Computational Mathematics},
  year = {2023},
  doi = {10.4208/jcm.2304-m2022-0140},
}
Figure