High-Order and Energy-Stable Implicit-Explicit Relaxation Runge-Kutta Schemes for Gradient Flows

Yuxiu Cheng, Kun Wang, Kai YangarXiv 2025 · OpenReview

TL;DR

This paper introduces a new class of numerical methods (IMEX RRK-SAV schemes) for simulating phase-field gradient flows, which are crucial for understanding material science and fluid dynamics. These methods achieve both high accuracy and guaranteed energy stability, overcoming limitations of existing approaches.

What Problem Does This Paper Address?

Phase-field models are essential for describing how systems evolve towards thermodynamic equilibrium, particularly in phenomena like material phase transitions and fluid mixing. These models are governed by gradient flows, where the system's free energy naturally decreases over time. Accurately simulating these processes over long periods requires numerical schemes that not only capture the complex dynamics with high precision but also strictly preserve this energy dissipation property. Traditional methods often struggle to achieve both high-order accuracy and energy stability simultaneously, leading to either inaccurate results or computational instability. This paper aims to bridge this gap by developing a novel numerical framework that addresses these challenges.

Methods

The authors propose a class of Implicit-Explicit Relaxation Runge-Kutta (IMEX RRK) schemes combined with the Scalar Auxiliary Variable (SAV) method.

  • SAV Method: This technique reformulates the original gradient flow equations by introducing an auxiliary variable. This reformulation simplifies the treatment of nonlinear terms and allows for the definition of a "modified energy" that is easier to control.
  • IMEX RK: This approach splits the equations into "stiff" (implicitly treated) and "non-stiff" (explicitly treated) parts. This hybrid strategy balances the stability of implicit methods with the computational efficiency of explicit methods.
  • Relaxation Technique: To further enhance energy stability, a "relaxation coefficient" is introduced. This coefficient is carefully chosen at each time step to ensure that the modified energy consistently decreases, thereby guaranteeing the energy dissipation law.
  • Diagonally Implicit-Explicit Runge-Kutta (IMEX RK): The specific type of IMEX RK method used is diagonally implicit-explicit, which simplifies the structure of the scheme.

The combination of these techniques allows for high-order accuracy while rigorously proving unconditional energy stability.

Datasets

Not discussed in the paper. The paper focuses on theoretical development and numerical validation using standard test cases for phase-field models.

Key Insights and Ideas

  • Synergy of SAV, IMEX, and Relaxation: The paper demonstrates that combining the SAV method for handling nonlinearities, IMEX RK for balancing accuracy and efficiency, and a relaxation technique for guaranteed energy stability is a powerful strategy for gradient flow simulations.
  • Unconditional Energy Stability: A significant achievement is the rigorous proof that the proposed schemes preserve the energy dissipation law unconditionally, meaning it holds regardless of the time step size. This is crucial for long-time simulations.
  • High-Order Accuracy: The methods are designed to achieve arbitrary high-order accuracy, which is essential for capturing fine details and reducing numerical diffusion in complex phenomena.
  • Two Interpretations of Numerical Solutions: The paper highlights two ways to interpret the numerical solutions at the end of a time step: Incremental Direction Technique (IDT) and Relaxation Technique (RT). The RT approach, which is directly tied to the relaxation coefficient, is shown to achieve the full order of accuracy, while IDT results in one order lower accuracy.

Key Results

  • Theoretical Guarantees: The paper provides rigorous proofs for the energy dissipation law and the order of accuracy for the proposed IMEX RRK-SAV schemes.
  • Numerical Validation: The theoretical findings are validated through numerical experiments on the Allen-Cahn (AC) and Cahn-Hilliard (CH) equations, as well as multi-component gradient flows (vector-valued AC equations).
  • Accuracy: Numerical results confirm that the RT interpretation of the schemes achieves the expected p-th order accuracy, while the IDT interpretation achieves (p-1)-th order accuracy. For example, a 2nd-order RT method achieved 1.9571 order accuracy in the AC equation example.
  • Energy Stability: Figures illustrating the discrete energy evolution (e.g., Figure 3 for AC, Figure 7 for CH) show a clear and monotonic decrease in energy over time, confirming the energy dissipation property.
  • Efficiency: While not explicitly quantified in terms of speedup, the IMEX approach inherently balances computational cost with accuracy, making it more efficient than fully implicit methods for similar accuracy levels.

Limitations and Future Work

  • Small Time Step for Taylor Expansion: The analysis of the relaxation coefficient relies on Taylor expansions, which requires the time step size to be sufficiently small. The paper notes this condition in Remark 3.3.
  • Extension to Other Models: The authors suggest extending their approach to other phase-field models and more complex scenarios in future work.
  • Multi-component Flows: The paper demonstrates the extension to multi-component gradient flows, showcasing its applicability to more complex physical systems.

The paper successfully introduces a robust and accurate numerical framework for a wide range of phase-field gradient flow problems, offering a valuable tool for researchers in various scientific disciplines.

Summary updated

Abstract

In this paper, we propose a class of high-order and energy-stable implicit-explicit relaxation Runge-Kutta (IMEX RRK) schemes for solving the phase-field gradient flow models. By incorporating the scalar auxiliary variable (SAV) method, the original equations are reformulated into equivalent forms, and the modified energy is introduced. Then, based on the reformulated equations, we propose a kind of IMEX RRK methods, which are rigorously proved to preserve the energy dissipation law and achieve high-order accuracy for both Allen-Cahn and Cahn-Hilliard equations. Numerical experiments are conducted to validate the theoretical results, including the accuracy of the approximate solution and the efficiency of the proposed scheme. Furthermore, the schemes are extended to multi-component gradient flows, with the vector-valued Allen-Cahn equations serving as a representative example.
BibTeX
@misc{cheng2025high,
  author = {Yuxiu Cheng and Kun Wang and Kai Yang},
  title = {{High-Order and Energy-Stable Implicit-Explicit Relaxation Runge-Kutta Schemes for Gradient Flows}},
  year = {2025},
  eprint = {2503.18844},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2503.18844},
}
Figure