Abstract:Solutions to differential equations, which are used to model physical systems, are computed numerically by solving a set of discretized equations. This set of discretized equations is reduced to a large linear system, whose solution is typically found using an iterative solver. We start with an initial guess, $x_0$, and iterate the algorithm to obtain a sequence of solution vectors, $x_k$, which are approximations to the exact solution of the linear system, $x$. The iterative algorithm is said to converge to $x$, in the field of reals, if and only if $x_k$ converges to $x$ in the limit of $k \to \infty$. In this paper, we formally prove the asymptotic convergence of a particular class of iterative methods called the stationary iterative methods, in the Coq theorem prover. We formalize the necessary and sufficient conditions required for the iterative convergence, and extend this result to two classical iterative methods: the Gauss--Seidel method and the Jacobi method. For the Gauss--Seidel method, we also formalize a set of easily testable conditions for iterative convergence, called the Reich theorem, for a particular matrix structure, and apply this on a model problem of the one-dimensional heat equation. We also apply the main theorem of iterative convergence to prove convergence of the Jacobi method on the model problem.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to formally prove the asymptotic convergence of a class of numerical algorithms called **Stationary Iterative Methods** when solving the linear system $ \mathbf{Ax} = \mathbf{b} $. Specifically, the author uses the Coq theorem prover to formally prove the necessary and sufficient conditions of this class of methods and applies them to two classical iterative methods: the Gauss - Seidel method and the Jacobi method. ### Main Problems and Contributions 1. **Formalizing Asymptotic Convergence**: - The main objective of the paper is to formally prove the asymptotic convergence of stationary iterative methods. This includes defining and proving the condition that the spectral radius of the iteration matrix $ \mathbf{S} = -\mathbf{M}^{-1}\mathbf{N} $ is less than 1, which is a necessary and sufficient condition for the convergence of the iterative method. 2. **Convergence of the Gauss - Seidel Method**: - The author formalizes an easily testable condition, namely the Reich theorem, for verifying the convergence of the Gauss - Seidel method. The Reich theorem provides a sufficient condition that if the coefficient matrix $ \mathbf{A} $ is real symmetric and all elements on the main diagonal are positive, then the spectral radius of the Gauss - Seidel method is less than 1. 3. **Convergence of the Jacobi Method**: - The author also applies the formalized asymptotic convergence theorem to the Jacobi method and proves its convergence on a specific model problem. ### Specific Contents - **Overview of Stationary Iterative Methods**: - Stationary iterative methods construct an iterative process $ \mathbf{Mx}_{k + 1}+\mathbf{N}\mathbf{x}_k=\mathbf{b} $ by decomposing the coefficient matrix $ \mathbf{A} $ into $ \mathbf{A}=\mathbf{M}+\mathbf{N} $, where $ \mathbf{M} $ is easy to invert. - In the Gauss - Seidel method, $ \mathbf{M} $ is the lower triangular part of $ \mathbf{A} $ plus the diagonal part, and $ \mathbf{N} $ is the strictly upper triangular part. - In the Jacobi method, $ \mathbf{M} $ is the diagonal part of $ \mathbf{A} $, and $ \mathbf{N} $ is the sum of the strictly lower triangular and upper triangular parts. - **Formalizing Asymptotic Convergence**: - The author formalizes the properties of matrix and vector norms in Coq, especially the $ \ell_2 $ norm. - By formalizing the definition and related properties of the spectral radius, the author proves that the spectral radius being less than 1 is a necessary and sufficient condition for the convergence of the iterative method. - **Formalizing the Reich Theorem**: - The author formalizes the Reich theorem, providing a sufficient condition for the convergence of the Gauss - Seidel method. - The Reich theorem states that if the coefficient matrix $ \mathbf{A} $ is real symmetric and all elements on the main diagonal are positive, then the spectral radius of the Gauss - Seidel method is less than 1. - **Proof of the Convergence of the Jacobi Method**: - The author applies the formalized asymptotic convergence theorem to the Jacobi method and proves its convergence on a specific model problem. ### Conclusion This paper, through formal proof, provides a strict theoretical basis for the asymptotic convergence of stationary iterative methods and provides easily verifiable conditions for the convergence of the Gauss - Seidel method and the Jacobi method. These results are of great significance for the fields of numerical analysis and scientific computing.

Formalization of Asymptotic Convergence for Stationary Iterative Methods

On the Convergence of the Accelerated Riccati Iteration Method.

The Stationary Iterations Revisited

Convergence of General Nonstationary Iterative Methods for Solving Singular Linear Equations

Convergence Criterion of Newton's Method for Singular Systems with Constant Rank Derivatives

Convergence of a Family of Methods with Symmetric, Antisymmetric Parameters and Weight Functions

On the Convergence of General Stationary Iterative Methods for Range‐hermitian Singular Linear Systems

A formal proof of the Lax equivalence theorem for finite difference schemes

On the Convergence of the Variational Iteration Method for Klein-Gordon Problems with Variable Coefficients II

Convergence and Quotient Convergence of Iterative Methods for Solving Singular Linear Equations with Index One

Formalization of a Stochastic Approximation Theorem

Convergence properties of the Runge-Kutta-Chebyshev method

Necessary and sufficient condition for mean-square convergence of linear distributed iterative stochastic processes

Convergence Analysis of Asymptotical Regularization and Runge-Kutta Integrators for Linear Inverse Problems under Variational Source Conditions

Convergence Properties of an Iterative Method for Solving Symmetric Non-linear Equations

Larger convergence regions for an efficient two-step iterative method

An extended study of two multistep higher order convergent methods for solving nonlinear equations

On The Convergence of the Variational Iteration Method Applied to Variable Coefficient Klein-Gordon Problems

A One-Parameter Family of Methods with a Higher Order of Convergence for Equations in a Banach Space

Stationary Behavior of Constant Stepsize SGD Type Algorithms: An Asymptotic Characterization

The mathematical foundations of the asymptotic iteration method