Nested finite element approximation of parabolic SPDEs with Whittle-Matérn noise

Øyvind Stormark Auestad

2024-06-17

Abstract:We propose a new type of fully discrete finite element approximation of a class of semilinear stochastic parabolic equations with additive noise. Our discretization differs from the ones typically considered, in that we employ a nested finite element approximation of the noise. This is well suited for dealing with covariance operators defined in terms of (negative powers of) elliptic operators, like that of Whittle-Matérn random fields. We derive strong and pathwise convergence rates for our proposed discretization, and our results are supported by numerical experiments.

Numerical Analysis

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is how to perform fully - discrete finite - element approximations of semilinear stochastic parabolic partial differential equations (SPDEs) with Whittle - Matérn noise effectively and efficiently. Specifically, the author proposes a new nested finite - element method to handle the noise terms in such equations in order to overcome the simulation and computational difficulties of traditional methods.

### Problem Background

Traditional finite - element approximation methods usually adopt two ways when dealing with noise:
1. **L2 Orthogonal Projection**: Project the noise into the finite - element space.
2. **Truncated Karhunen - Loève Expansion**: Approximate the noise through eigenvalue decomposition.

These two methods have certain limitations in practical applications:
- The L2 orthogonal projection requires knowledge of the probability distribution of the projected noise, which is often difficult to obtain in many cases.
- The Karhunen - Loève expansion requires knowledge of the eigenvalues and eigenvectors of the covariance operator, and this information is usually uncomputable in high - dimensional spaces.

### Proposed New Method

To solve the above problems, the author introduces a new nested finite - element approximation method. The core idea of this method is to perform approximation in two steps:
1. **First Step**: First, perform a finite - element approximation of the noise \( A^{-\gamma}_2 W \).
2. **Second Step**: Then, use this approximation result to replace the noise term in the original equation, and then perform a finite - element approximation of the entire equation.

This method is especially suitable for cases where the covariance operator is defined by the negative fractional power of an elliptic operator, such as the Whittle - Matérn random field.

### Main Contributions

1. **New Discretization Method**: A new nested finite - element approximation method is proposed, which can handle SPDEs with noise more effectively.
2. **Convergence Rate Analysis**: The strong convergence and path convergence rates of the proposed method are derived, and these convergence rates are verified through numerical experiments.

### Mathematical Model

Consider the semilinear stochastic parabolic partial differential equation in the following form:
\[ du = -A_1 u \, dt + F(u) \, dt + A^{-\gamma}_2 dW, \quad u(0)=\xi, \]
where:
- \( W \) is a cylindrical Wiener process;
- \( A_j \) is a second - order elliptic operator;
- \( F \) is a Lipschitz nonlinear term;
- \( \gamma \) is a real - valued parameter;
- \( \xi \) is a random variable.

### Conclusion

Through the nested finite - element method, the author not only provides a new discretization strategy but also proves the convergence of this method and demonstrates its effectiveness in numerical simulation. This method provides new ideas and tools for handling SPDEs with complex noise structures.