Numerical approximation of linear parabolic evolution equations revisited

Øyvind Stormark Auestad
2024-10-01
Abstract:We obtain rates of convergence of numerical approximations of linear parabolic evolution equations. Our estimates extend known results like Theorem 3.5 in \cite{thomee} to more general equations and accommodate more advanced numerical approximation techniques. As an example, we consider parabolic equations on surfaces, and surface finite element approximations.
Numerical Analysis
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the convergence rate of numerical approximations for linear parabolic evolution equations. Specifically, the author aims to obtain convergence rate estimates for a broader type of linear parabolic evolution equations and their numerical approximation methods. This includes, but is not limited to: 1. **Extending known results**: The main result in the paper (Theorem 2.2) is a generalization of existing results (such as Theorem 3.5 in reference [13]), and these known results are usually limited to self - adjoint operator equations of the finite element method in Hilbert spaces. However, the results in this paper are applicable to analytic semigroup generators in more general Banach spaces, and the requirements for numerical approximation methods are more relaxed, only needing to satisfy the uniform sector condition (see Assumption 2.1). 2. **Handling more complex situations**: The paper also specifically considers parabolic equations on surfaces and their surface finite - element approximations. Through these extensions, the author is able to handle more complex situations involving elliptic operators, which may not be strictly positive definite. 3. **Optimizing error analysis**: The results in the paper are of great significance for deriving the optimal convergence rate of numerical approximations of stochastic evolution equations. By introducing a parameterized estimation form (such as Theorem 2.2), users can make trade - offs between the regularity of the initial conditions and the singularity or convergence rate at the starting time. 4. **Fully - discrete error estimates**: In addition to semi - discrete error estimates, the paper also combines rational approximations of exponential functions to obtain convergence rate estimates for fully - discrete approximations. In particular, for the backward Euler method, the paper describes in detail the convergence rate of its fully - discrete approximation (see Theorem 2.13 and Theorem 2.23). 5. **Error estimates in different norms**: The paper also explores semi - discrete error estimates in different norms (such as the \(\|(λ + A)^\alpha\cdot\|_X\) norm, where \(\alpha\in[- 1/2,1/2]\)), as well as error estimates for less regular data. In summary, the main objective of this paper is to provide a broader theoretical framework to support and improve numerical solution methods for linear parabolic evolution equations, especially when dealing with complex geometric structures and irregular data.