Abstract:In this paper, we propose a novel, computationally efficient reduced order method to solve linear parabolic inverse source problems. Our approach provides accurate numerical solutions without relying on specific training data. The forward solution is constructed using a Krylov sequence, while the source term is recovered via the conjugate gradient (CG) method. Under a weak regularity assumption on the solution of the parabolic partial differential equations (PDEs), we establish convergence of the forward solution and provide a rigorous error estimate for our method. Numerical results demonstrate that our approach offers substantial computational savings compared to the traditional finite element method (FEM) and retains equivalent accuracy.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the efficient solution method for linear parabolic inverse source problems in the field of numerical analysis. Specifically, the goal of such problems is to determine the unknown source term from the observations of time - dependent partial differential equations (PDEs) at the final time. Although the traditional finite element method (FEM) can achieve high precision, it has high computational cost and large memory requirements. To solve these problems, this paper proposes a new reduced - order method (ROM), aiming to provide a more computationally efficient and less data - variation - sensitive method to solve linear parabolic inverse source problems.
### Main contributions of the paper
1. **New reduced - order method**:
- A new method is proposed to construct the forward solution based on the Krylov sequence and recover the source term using the conjugate gradient method (CG).
- Under the weak regularity assumption, this method proves the convergence of the forward solution and provides a strict error estimate.
2. **Computational efficiency**:
- Numerical experiments show that, compared with the traditional finite element method, the new method significantly reduces the computational time while maintaining the same precision.
3. **Theoretical analysis**:
- The convergence of the forward solution is established, and a random error estimate is provided to ensure the reliability and stability of the method.
### Specific problem description
The linear parabolic inverse source problem can be formalized as the following partial differential equation system:
\[
\begin{cases}
u_t-\Delta u = f, & \text{in }\Omega\times(0, T] \\
u(\cdot, t) = 0, & \text{on }\partial\Omega\times(0, T] \\
u(\cdot, 0) = 0, & \text{in }\Omega
\end{cases}
\]
where $\Omega\subset\mathbb{R}^d$ is a bounded domain, $T > 0$ is the termination time, $u$ is the solution of the state equation, and $f$ is the time - independent source term. The goal of the inverse problem is to reconstruct the unknown source term $f$ according to the observation at the final time $u(\cdot, T)$.
### Method overview
1. **Forward process**:
- Use the Krylov sequence to construct the forward solution.
- Discretize space and time by the finite element method (FEM) to obtain semi - discrete and fully - discrete equations.
2. **Inverse process**:
- Use the conjugate gradient method (CG) to optimize the objective function to minimize the mean - square error between the predicted solution and the observation.
- Introduce a regularization term to improve the stability of the solution.
3. **Reduced - order method**:
- Use the Krylov subspace to generate a low - dimensional basis, thereby reducing the computational complexity.
- Project the full - order model (FOM) into the reduced - order subspace through the projection matrix to further improve the computational efficiency.
### Numerical results
The paper verifies the effectiveness of the proposed method through multiple test cases, including the reconstruction of single - letter, non - letter - shaped, and multi - letter source terms. The results show that the new method significantly reduces the computational time while maintaining high precision, especially when the spatial grid and time step are finer, the efficiency improvement is particularly obvious.
### Conclusion
The reduced - order method proposed in this paper provides an efficient and stable solution scheme for linear parabolic inverse source problems, especially suitable for application scenarios that require a large amount of computing resources. Future work can further expand the application range of this method and explore its applicability in other types of inverse problems.