This paper extends monotonicity-based inversion methods to the inverse coefficient problem for isotropic nonlocal elliptic equations: where , , and is a bounded open set. The paper establishes monotonicity relationships between the leading coefficient and the (partial) exterior Dirichlet-to-Neumann (DN) map. The main results demonstrate that monotone ordering of coefficients implies corresponding ordering of DN maps. Furthermore, localized potentials for nonlocal equations are constructed, yielding local uniqueness results for the fractional inverse problem.
Consider the exterior value problem:
(-\nabla \cdot \sigma\nabla)^s u = 0 & \text{in } \Omega \\ u = f & \text{in } \Omega^e \end{cases}$$ where $\sigma$ satisfies: $$0 < \lambda \leq \sigma(x) \leq \lambda^{-1} \text{ for } x \in \Omega, \quad \sigma = 1 \text{ in } \Omega^e$$ The DN map is defined as: $$\Lambda_\sigma : \tilde{H}^s(W) \to H^{-s}(W), \quad f \mapsto (-\nabla \cdot \sigma\nabla)^s u^f|_W$$ ### Core Technical Methods #### 1. Caffarelli-Silvestre Extension Transforms the nonlocal problem into an extension problem: $$\begin{cases} \nabla_{x,y} \cdot (y^{1-2s}\tilde{\sigma}\nabla_{x,y}\tilde{u}) = 0 & \text{in } \mathbb{R}^{n+1}_+ \\ \tilde{u}(x,0) = u(x) & \text{on } \mathbb{R}^n \end{cases}$$ where $\tilde{\sigma}(x) = \begin{pmatrix} \sigma(x)I_n & 0 \\ 0 & 1 \end{pmatrix}$ #### 2. Monotonicity Relation Proof Through bilinear form analysis, derives the key inequality: $$\int_{\Omega \times (0,\infty)} y^{1-2s}(\sigma_1 - \sigma_2)|\nabla\tilde{u}^f_2|^2 dx dy \leq d_s \langle(\Lambda_{\sigma_1} - \Lambda_{\sigma_2})f, f\rangle$$ #### 3. Localized Potential Construction Utilizing Runge approximation properties, constructs functions satisfying: - $\nabla v|_D \equiv 0$ (gradient vanishes in specified region) - $\nabla v|_B \not\equiv 0$ (gradient is nonzero in target region) ### Technical Innovations 1. **Extension Problem Method**: Cleverly utilizes Caffarelli-Silvestre extension to transform analysis of nonlocal operators into local analysis in weighted Sobolev spaces 2. **Monotonicity Estimates**: Establishes two different forms of monotonicity inequalities (3.1) and (3.2), providing flexibility for different applications 3. **Localization Construction**: Systematically constructs required localized potentials through duality arguments and truncation function techniques ## Experimental Setup This paper is purely theoretical research with no numerical experiments. The correctness of theoretical results is verified primarily through rigorous mathematical proofs. ### Theoretical Verification Methods 1. **Lemma 3.1**: Establishes monotonicity relations through bilinear forms of the extension problem 2. **Theorem 4.1**: Proves existence of localized potentials using Runge approximation properties 3. **Theorem 1.1**: Combines monotonicity relations and localized potentials to prove local uniqueness ## Main Theoretical Results ### Monotonicity Relations (Lemma 3.1) For coefficients $\sigma_1, \sigma_2$ satisfying the conditions: $$\int_{\Omega \times (0,\infty)} y^{1-2s}(\sigma_1 - \sigma_2)|\nabla\tilde{u}^f_1|^2 dx dy \leq d_s \langle(\Lambda_{\sigma_1} - \Lambda_{\sigma_2})f, f\rangle \leq \int_{\Omega \times (0,\infty)} y^{1-2s}(\sigma_1 - \sigma_2)|\nabla\tilde{u}^f_2|^2 dx dy$$ ### Local Uniqueness (Theorem 1.1) Let $O \subseteq \Omega$ be a connected relatively open subset with $O \cap \partial\Omega \neq \emptyset$. If: - $\sigma_1 \leq \sigma_2$ in $O$, or $\sigma_1 \geq \sigma_2$ in $O$ - $\Lambda_{\sigma_1}f|_W = \Lambda_{\sigma_2}f|_W$ for all $f \in C_c^\infty(W)$ then $\sigma_1 = \sigma_2$ in $O$. ### Runge Approximation (Theorem 4.1) For appropriately chosen measurable sets $B, D \subseteq \Omega$, there exist a function $v(x) \in H^1(B \cup D)$ and $\beta_1(y) \in C_c^\infty((0,\infty))$ such that $v(x)\beta_1(y)$ can be approximated by solutions of the extension problem in the $H^1_x(B \cup D, y^{1-2s})$ norm. ## Related Work ### Main Research Directions 1. **Fractional Calderón Problem**: [GSU20] first investigated the Calderón problem for fractional Schrödinger equations 2. **Variable Coefficient Nonlocal Operators**: [GLX17] extended to variable coefficient cases 3. **Monotonicity Methods**: [HL19, HL20] established monotonicity relations for lower-order potentials 4. **Nonlocal-to-Local Reduction**: [CGRU23, LLU23] utilized extension problems for reduction ### Innovations of This Paper Compared to existing work, this paper is the first to: - Establish monotonicity relations for leading coefficients - Systematically construct localized potentials for nonlocal equations - Provide a complete theoretical framework for local uniqueness ## Conclusions and Discussion ### Main Conclusions 1. Successfully establishes monotonicity relationships between leading coefficients and DN maps in fractional elliptic equations 2. Constructs effective localized potentials, realizing Runge approximation properties 3. Proves local uniqueness theorem, providing new theoretical tools for fractional inverse problems ### Limitations 1. **Dimension Restriction**: Requires $n \geq 3$; lower-dimensional cases need special treatment 2. **Coefficient Conditions**: Requires $\sigma$ to be known in $\Omega^e$ and equal to 1 3. **Connectivity Requirements**: Local uniqueness requires the domain to be connected to the boundary ### Future Directions 1. Extension to inverse obstacle problems 2. Establishment of Lipschitz stability estimates 3. Development of practical reconstruction algorithms 4. Investigation of nonlinear cases ## In-Depth Evaluation ### Strengths 1. **Theoretical Rigor**: Complete proofs, clear logic, and refined technical handling 2. **Methodological Innovation**: Skillfully combines extension problem methods with monotonicity techniques 3. **Result Completeness**: Forms a complete theoretical chain from monotonicity relations to local uniqueness 4. **Technical Depth**: Involves fractional Sobolev spaces, weighted spaces, and other advanced theories ### Weaknesses 1. **Application Scope**: Pure theoretical research lacking numerical verification and practical applications 2. **Condition Restrictions**: Multiple technical conditions limit the generality of results 3. **Computational Complexity**: Insufficient discussion of computational complexity in calculating DN maps and constructing localized potentials ### Impact 1. **Theoretical Contribution**: Provides important tools for fractional inverse problem theory 2. **Method Value**: Systematic application of monotonicity methods to nonlocal problems 3. **Subsequent Research**: Provides new technical pathways for related research ### Applicable Scenarios 1. Parameter identification in fractional diffusion equations 2. Physical parameter reconstruction in nonlocal media 3. Mathematical modeling of anomalous diffusion phenomena 4. Inverse problems in medical imaging and geophysical exploration ## References The paper cites abundant relevant literature, primarily including: - [GSU20] Pioneering work by Ghosh, Salo, Uhlmann on the fractional Calderón problem - [CGRU23] Recent advances by Covi et al. on nonlocal-to-local reduction - [HL19, HL20] Systematic research by Harrach, Lin on monotonicity methods - [CS07] Classical theory by Caffarelli, Silvestre on extension problems --- This paper makes important theoretical contributions to the field of inverse problems for fractional elliptic equations, laying a solid foundation for further development in this field. Although it is purely theoretical research, its methods and results have important guiding significance for related application fields.