Abstract:We present a natural framework for constructing energy-stable time discretization schemes. By leveraging the Onsager principle, we demonstrate its efficacy in formulating partial differential equation models for diverse gradient flow systems. Furthermore, this principle provides a robust basis for developing numerical schemes that uphold crucial physical properties. Within this framework, several widely used schemes emerge naturally, showing its versatility and applicability.
What problem does this paper attempt to address?
The core problem that this paper attempts to solve is to develop an energy - stable time - discretization numerical scheme based on the Onsager principle to simulate and solve physical systems with dissipative properties. Specifically, the authors aim to:
1. **Construct energy - stable numerical methods**: Use the Onsager principle to design time - discretization schemes to ensure that these schemes can maintain the energy - dissipation characteristics of the system. This is crucial for the accurate modeling and simulation of complex systems (such as multiphase flow, diffusion processes, etc.) in many practical applications.
2. **Maintain physical structure**: In addition to energy stability, it is also hoped that the numerical method can maintain other important physical properties, such as mass conservation, positive density, etc. This is very important for ensuring the physical rationality of the numerical solution.
3. **Be applicable to a wide range of dissipative systems**: Through the Onsager principle, the authors show how to derive evolution equations applicable to different types of dissipative systems and further develop time - discretization schemes for non - conservative and conservative unknowns. This includes, but is not limited to, the Allen - Cahn equation, the Cahn - Hilliard equation, the Fokker - Planck equation, the Planck - Nernst - Poisson system, the Maxwell - Stefan diffusion equation, and models of multiphase flow in porous media.
4. **Provide a natural framework**: The paper proposes a natural framework from which a variety of widely - used numerical methods (such as explicit - implicit time - discretization, optimal transport distance, etc.) can be naturally derived, thus demonstrating the universality and applicability of this framework.
### Introduction to the Onsager Principle
The Onsager principle is a fundamental law in thermodynamics, especially suitable for describing the dynamic behavior of dissipative systems near equilibrium. It determines the evolution equation of the system by minimizing the Rayleighian functional. For a given generalized coordinate \(y\), its dynamic evolution is described by the following equation:
\[
A\dot{y}=-\nabla U(y),
\]
where \(A\) is a positive - semi - definite matrix, representing the energy - dissipation characteristics of the system; \(U(y)\) is the potential energy function. According to the Onsager principle, if the system exhibits microscopic reversibility, then \(A = A^{\top}\). Introduce the Rayleighian functional:
\[
R=\Phi+\dot{U},
\]
where:
\[
\Phi=\frac{1}{2}\|z\|_A^2=\frac{1}{2}z\cdot Az,\quad z = \dot{y},
\]
\[
\dot{U}=\nabla U\cdot\dot{y}.
\]
By minimizing the Rayleighian functional, the evolution equation of the system can be obtained. This principle is not only used for modeling but also for deriving numerical methods, especially in ensuring energy stability.
### Time - Discretization Method
Based on the Onsager principle, the authors propose a time - discretization framework. The specific steps are as follows:
1. **Define the discretized Rayleighian functional**:
\[
\Phi(z)+\frac{U(y)-U(y_k)}{\tau},
\]
where \(\tau\) is the time step, and \(y_k\) is the solution at the \(k\)-th step.
2. **Solve the constrained optimization problem**:
\[
y_{k + 1}=\arg\min_y\left\{U(y)+\frac{1}{2\tau}\|y - y_k\|_A^2\right\},
\]
This scheme is called De Giorgi's minimizing movement scheme and has unconditional energy stability.
3. **Apply to specific PDEs**: Apply the above method to various dissipative systems, including systems with non - conservative and conservative unknowns. For non - conservative parameters, time - discretization is similar to the case of ODEs; for conservative parameters, an optimization problem needs to be solved at each step.
Through this way, the authors...