Abstract:We consider the numerical approximation of Gaussian random fields on closed surfaces defined as the solution to a fractional stochastic partial differential equation (SPDE) with additive white noise. The SPDE involves two parameters controlling the smoothness and the correlation length of the Gaussian random field. The proposed numerical method relies on the Balakrishnan integral representation of the solution and does not require the approximation of eigenpairs. Rather, it consists of a sinc quadrature coupled with a standard surface finite element method. We provide a complete error analysis of the method and illustrate its performances by several numerical experiments.
What problem does this paper attempt to address?
### Problems the paper attempts to solve
This paper aims to solve the problem of numerical approximation of Gaussian Random Fields (GRFs) on closed surfaces. Specifically, the author considers the numerical approximation methods of GRFs defined on closed surfaces and described by fractional - order Stochastic Partial Differential Equations (SPDEs). These SPDEs contain additive white noise and control the smoothness and correlation length of GRFs through two parameters.
#### Main problems
1. **Numerical approximation**: How to efficiently perform numerical approximation of GRFs on closed surfaces?
2. **Error analysis**: How to ensure that the proposed numerical method has reliable error estimates?
3. **Algorithm design**: Propose a numerical method that does not require approximate eigenpairs to simplify the calculation and improve efficiency.
#### Specific problem descriptions
- **SPDE form**: The form of the studied SPDE is:
\[
(\kappa^2 I - \Delta_\gamma)^s u = w,
\]
where $\kappa > 0$, $\Delta_\gamma$ is the Laplace - Beltrami operator, $w$ is Gaussian white noise with unit variance, $s > \frac{n - 1}{4}$, and $\gamma$ is a closed hypersurface in $\mathbb{R}^n$ ($n = 2,3$).
- **Numerical method**: The paper proposes a method based on the Balakrishnan integral representation, which combines sinc quadrature and the standard surface finite element method. This method avoids the approximation of eigenpairs, thus simplifying the calculation process.
- **Error analysis**: The author provides a complete error analysis, including strong convergence and mean - square error estimates. In particular, they prove the exponential convergence rate of sinc quadrature, and this convergence rate is independent of the finite - element mesh size.
#### Innovation points
- **Improved sinc quadrature convergence**: Compared with previous work, the author improves the convergence of sinc quadrature and derives an exponential convergence rate independent of the mesh size.
- **Geometric error handling**: Consider the influence of geometric errors on eigenvalue approximation, especially in the case of general $C^3$ surfaces.
- **Mean - square error estimate**: Through the analysis of the asymptotic behavior of the eigenvalues of the Laplace - Beltrami operator, the optimal mean - square error estimate is obtained.
In conclusion, this paper solves the problem of efficient numerical approximation of GRFs on closed surfaces and provides reliable error analysis, providing new ideas and methods for research in related fields.