Abstract:In this paper, we consider the numerical approximation of time-fractional parabolic problems involving Caputo derivatives in time of order $\alpha$, $0< \alpha<1$. We derive optimal error estimates for semidiscrete Galerkin FE type approximations for problems with smooth and nonsmooth initial data. Our analysis relies on energy arguments and exploits the properties of the inverse of the associated elliptic operator. We present the analysis in a general setting so that it is easily applicable to various spatial approximations such as conforming and nonconforming FEMs, and FEM on nonconvex domains. The finite element approximation in mixed form is also presented and new error estimates are established for smooth and nonsmooth initial data. Finally, an extension of our analysis to a multi-term time-fractional model is discussed.
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
This paper aims to solve the problem of numerical approximation of time - fractional parabolic equations, especially time - fractional differential equations involving Caputo derivatives. Specifically, the paper focuses on the following aspects:
1. **Numerical approximation method**: The paper studies the application of the semi - discrete Galerkin finite element method (FEM) in solving time - fractional parabolic equations. It considers the optimal error estimates in the cases of smooth and non - smooth initial data.
2. **Error estimates**: The author derives the optimal error estimates under the semi - discrete finite element approximation for problems with initial conditions of different smoothness. These estimates are applicable not only to the standard conforming FEM, but also to the non - conforming FEM and the FEM on non - convex domains.
3. **Mixed - form analysis**: The paper also explores the mixed form of problem (1.1) and uses the standard Galerkin mixed finite element method for discretization in space, thereby establishing new error estimates for cases with smooth and non - smooth initial data.
4. **Multi - term time - fractional model**: Finally, the paper discusses the extended application of its analysis method in the multi - term time - fractional model.
### Specific problem description
The specific problem studied in the paper can be represented as the following time - fractional initial - boundary value problem:
\[
\begin{cases}
\partial_t^\alpha u+L u = f & \text{in }\Omega\times(0, T], \\
u(0)=u_0 & \text{in }\Omega, \\
u = 0 & \text{on }\partial\Omega\times(0, T],
\end{cases}
\]
where \(L u=-\text{div}(A(x)\nabla u)+c(x)u\), \(f(x, t)\) is the forcing term, and \(u_0(x)\) is the initial data. Here, \(A(x)=[a_{ij}(x)]_{1\leq i,j\leq2}\) is a symmetric and uniformly positive definite matrix, and \(c(x)\in L^\infty(\Omega)\) is a non - negative function.
### Definition of Caputo derivative
The definition of Caputo derivative used in the paper is:
\[
\partial_t^\alpha\phi(t):=I^{1 - \alpha}\phi'(t):=\int_0^t\omega_{1-\alpha}(t - s)\phi'(s)\,ds,
\]
where \(\omega_\alpha(t):=\frac{t^{\alpha - 1}}{\Gamma(\alpha)}\).
Through the above research, the paper aims to provide a unified method for dealing with the numerical approximation of time - fractional parabolic equations and provides a theoretical basis for practical applications.