Abstract:This paper studies the inverse problem of determination the history for a stochastic diffusion process, by means of the value at the final time $T$. By establishing a new Carleman estimate, the conditional stability of the problem is proven. Based on the idea of Tikhonov method, a regularized solution is proposed. The analysis of the existence and uniqueness of the regularized solution, and proof for error estimate under an a-proior assumption are present. Numerical verification of the regularization, including numerical algorithm and examples are also illustrated.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the inverse problem of stochastic parabolic equations. Specifically, it is the process of determining the history from the terminal value. More precisely, the paper focuses on how to determine the solution of the stochastic diffusion process through the measured values at the final time \(T\). This involves two main aspects:
1. **Conditional stability**: Suppose two solutions \(u\) and \(\hat{u}\) are given, and their values at the final time \(T\) are \(u(T)\) and \(\hat{u}(T)\) respectively. The question is whether a positive constant \(C\) can be found such that the following inequality holds:
\[
|u - \hat{u}| \leq C \| u(T) - \hat{u}(T) \|
\]
where the norms on both sides are appropriate.
2. **Reconstruction**: Whether the solution \(u\) can be reconstructed from the terminal value \(u(T)\).
### Main contributions
- **New Carleman estimate**: In order to prove conditional stability, the author established a new global Carleman estimate. This estimate uses an exponential weight function instead of the previous double - exponential weight function. This improvement enhances the results of conditional stability.
- **Regularized solution**: Based on the Tikhonov regularization method, a regularized solution is proposed, and its existence and uniqueness are analyzed. In addition, the error estimate under a priori assumptions is also proved.
- **Numerical verification**: A numerical algorithm is provided, including combining the conjugate gradient method and the Picard iteration method to solve the regularization problem. The effectiveness of the regularization method is verified through numerical experiments.
### Technical details
- **Carleman estimate**: By introducing a new weight function \(\psi=(t + 1)^\lambda\) and \(\phi = e^\psi\), the following Carleman estimate is established:
\[
E\int_{\epsilon}^T\int_G\lambda^2(t + 1)^{\lambda - 2}v^2\,dt\,dx+E\int_{\epsilon}^T\int_G\lambda(t + 1)^{-1}|\nabla v|^2\,dt\,dx
\leq C\left[\lambda(T + 1)^{\lambda - 1}E\|v(T)\|^2_{L^2(G)}+E\|v(\epsilon)\|^2_{L^2(G)}+E\int_{\epsilon}^T\int_G\phi^2(|\nabla g|^2+g^2+f^2)\,dx\,dt\right]
\]
- **Tikhonov regularization**: A Tikhonov - type functional \(J_\alpha(y_0)\) is defined, and the initial value \(y_0\) is solved by minimizing this functional. The specific form is:
\[
J_\alpha(y_0)=E\int_G\left(y_0+\int_0^T\left[\sum_{i,j = 1}^n(a_{ij}y_i)_j+b_1\cdot\nabla y+b_2y+f\right]ds+\int_0^T(b_3y+g)dW(s)-u_T^\delta\right)^2dx+\alpha E\|y_0\|^2_{H^2(G)}
\]
- **Numerical algorithm**: Combining the conjugate gradient method and the Picard iteration method, a numerical algorithm for solving the forward - backward stochastic parabolic equation is proposed. Specific steps include initial guess, solving the forward problem, solving the adjoint problem, calculating the descent direction and step size, etc.
Through these methods, the paper not only theoretically proves the conditional stability and the feasibility of reconstruction.