Abstract:The iterative nature of many discretisation methods for continuous dynamical systems has led to the study of the connections between iterative numerical methods in numerical linear algebra and continuous dynamical systems. Certain researchers have used the explicit Euler method to understand this connection, but this method has its limitation. In this study, we present a new connection between successive over-relaxation (SOR)-type methods and gradient systems; this connection is based on discrete gradient methods. The focus of the discussion is the equivalence between SOR-type methods and discrete gradient methods applied to gradient systems. The discussion leads to new interpretations for SOR-type methods. For example, we found a new way to derive these methods; these methods monotonically decrease a certain quadratic function and obtain a new interpretation of the relaxation parameter. We also obtained a new discrete gradient while studying the new connection.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to explore and establish a new connection between the iterative solution methods of linear equations (especially the SOR - type methods) and the discrete gradient methods of gradient systems. Specifically, the author aims to overcome the limitations of traditional connection methods (such as the explicit Euler method) and find a new connection method, so as to gain a deeper understanding of the SOR - type methods and have no difference in convergence conditions.
### Main problems and goals
1. **Limitations of existing methods**:
- When dealing with some important iterative methods (such as the successive over - relaxation method (SOR)), traditional connection methods (such as the explicit Euler method), due to involving complex parameters and matrix inverse operations, are difficult to provide in - depth understanding.
- There are fundamental differences in the convergence conditions between the iterative solution methods of linear equations and ordinary differential equations (ODEs), which limit the effect of understanding iterative methods from the perspective of ODEs.
2. **Goals of the new connection**:
- Find a new connection method that can provide a deeper understanding of the SOR - type methods.
- Ensure that this new connection is consistent with the existing iterative methods in convergence conditions, thereby providing theoretical support for the development of better iterative methods.
### Specific content of the new connection
The author establishes an equivalence relationship between the SOR - type methods and the gradient system by introducing the discrete gradient method. The specific steps are as follows:
- **Define the energy function**: For a given symmetric positive definite matrix \( A\in\mathbb{R}^{n\times n} \) and a vector \( \mathbf{b}\in\mathbb{R}^n \), define the energy function \( f(\mathbf{x})=\frac{1}{2}\mathbf{x}^\top A\mathbf{x}-\mathbf{x}^\top\mathbf{b} \).
- **Discrete gradient method**: Use the discrete gradient method to discretize the gradient system \( \frac{d\mathbf{x}(t)}{dt}=-P\nabla f(\mathbf{x}(t)) \), where \( P \) is a symmetric positive definite matrix.
- **Equivalence relationship**: Prove that the SOR method can be regarded as a special case of a certain discrete gradient scheme, that is:
\[
G_{DG} = G_{SOR}, \quad \mathbf{c}_{DG} = \mathbf{c}_{SOR}
\]
where \( h = \frac{2\omega}{2-\omega} \) is the relationship between the step size and the relaxation parameter.
### Results and significance
Through this new connection, the author not only found a new derivation method of the SOR method, but also revealed the property that these methods monotonically decrease a certain quadratic function, and gave a new interpretation of the relaxation parameter. In addition, a new discrete gradient was obtained during the research process, further expanding the understanding of the SOR - type methods.
In conclusion, this paper establishes a new connection between the SOR - type methods and the gradient system by introducing the discrete gradient method, overcomes the limitations of traditional methods, and provides a new perspective for understanding and improving iterative solution methods.