Inverse problems for integro-differential operators
This book explores inverse problems for certain partial differential problems (PDEs) involving nonlocal elliptic operators, with the fractional Laplacian serving as a notable example. This book is organized into nine chapters.\N\N\textbf{Chapter 1.} The authors begin by introducing the concept of inverse problems and then highlight key differences between those involving classical (local) differential operators and integro-differential operators, underscoring the need for fundamentally different approaches.\N\N\textbf{Chapter 2.} The authors list some preliminary materials. They first introduce some fractional Sobolev spaces and Hölder spaces. Then they introduce the fractional Laplacian \((-\Delta)^{s}\) and the fractional elliptic operator \((-\nabla\cdot\sigma\nabla)^{s}\) of order \(0<s<1\), and outline several key properties, including the unique continuation property (also known as the antilocality property).\N\N\textbf{Chapter 3.} Let \(\Omega\subset\mathbb{R}^{n}\) be a bounded Lipschitz domain and let \(q\in L^{\infty}(\Omega)\). The authors first discuss the well-posedness for the fractional Schrödinger equation (provided that 0 is not an eigenvalue):\N\[\N\left\{\begin{aligned} & ((-\Delta)^{s}+q)u=0 && \text{in \(\Omega\),} \\\N& u=f && \text{in \(\Omega_{e}\),} \end{aligned}\right.\N\]\Nwhere \(\Omega_{e}:=\mathbb{R}^{n}\setminus\overline{\Omega}\) denotes the exterior domain of \(\Omega\) . Accordingly, one can define the Dirichlet-to-Neumann map (DN map)\N\[\N\Lambda_{q}[f] := \left.(-\Delta)^{s}u_{f}\right|_{\Omega_{e}},\N\]\Nwhere \(u=u_{f}\) is the unique solution to the above exterior-value problem. The authors then discuss the inverse problems from the following perspectives:\N\begin{itemize}\N\item \textit{Global uniqueness.} The potential \(q\) can be uniquely determined by the DN map \(\Lambda_{q}\).\N\item \textit{Logarithmic stability.} Under some apriori assumptions, one can obtain a stability estimate\N\[\N\|q_{1}-q_{2}\|_{Z^{-s}(\Omega)} \le \omega \left(\|\Lambda_{q_{1}}-\Lambda_{q_{2}}\|_{*}\right)\N\]\Nwhere \(\omega\) is a modulus of continuity satisfying \(\omega(t)\le C|\log t|^{-\mu}\) for all \(0<t\le 1\).\N\item \textit{Global uniqueness with single measurement.} If we additionally assume that \(s\ge\frac{1}{4}\) or \(q\in C^{0}(\overline{\Omega})\), the potential \(q\) can be uniquely determined from a single measurement \(\Lambda_{q}f\) for a fixed \(0\not\equiv f\in C_{c}^{\infty}(\Omega_{e})\).\N\end{itemize}\NThe authors also exhibit an alternative method, called the monotonicity-based inversion, to study the fractional Schrödinger equation.\N\N\textbf{Chapter 4.} With the same notations as in previous chapters, the authors study the well-posedness for fractional Schrödinger equation for \(1/2<s<1\) with a drift term (provided that 0 is not an eigenvalue):\N\[\N\left\{\begin{aligned} & ((-\Delta)^{s}+b\cdot\nabla+c)u=0 && \text{in \(\Omega\),} \\\N& u=f && \text{in \(\Omega_{e}\),} \end{aligned}\right.\N\]\Nand similarly define the DN map\N\[\N\Lambda_{b,c}[f] := \left.(-\Delta)^{s}u_{f}\right|_{\Omega_{e}},\N\]\Nwhere \(u=u_{f}\) is the unique solution to the above exterior-value problem. The authors then discuss the inverse problems from the following perspectives:\N\begin{itemize}\N\item \textit{Global uniqueness.} The coefficients \(b\) and \(c\) can be simultaneously determined by the DN map \(\Lambda_{b,c}\).\N\item \textit{Logarithmic stability.} Under some apriori assumptions, one can obtain a stability estimate\N\[\N\|c_{1}-c_{2}\|_{H^{-s}(\Omega)} + \|b_{1}-b_{2}\|_{H^{-s}(\Omega)} \le \omega \left(\|\Lambda_{b_{1},c_{1}}-\Lambda_{b_{2},c_{2}}\|_{*}\right)\N\]\Nwhere \(\omega\) is a modulus of continuity satisfying \(\omega(t)\le C|\log t|^{-\mu}\) for all \(0<t\le 1\).\N\item \textit{Global uniqueness with \(n+1\) measurements.} There exists exterior datas \(f_{1},\cdots,f_{n+1} \in C_{c}^{\infty}(\Omega_{e})\) such that the coefficients \(b\) and \(c\) can be simultaneously determined from \(n+1\) measurements \(\{\Lambda_{b,c}f_{j}\}_{j=1}^{n+1}\)\N\end{itemize}\N\N\textbf{Chapter 5.} With the same notations as in previous chapters, the authors study the well-posedness for fractional Schrödinger equation (provided that 0 is not an eigenvalue):\N\[\N\left\{\begin{aligned} & ((-\nabla\cdot\sigma\nabla)^{s}+q)u=0 && \text{in \(\Omega\),} \\\N& u=f && \text{in \(\Omega_{e}\),} \end{aligned}\right.\N\]\Nand similarly define the DN map\N\[\N\Lambda_{q}[f] := \left.(-\Delta)^{s}u_{f}\right|_{\Omega_{e}},\N\]\Nwhere \(u=u_{f}\) is the unique solution to the above exterior-value problem. One can show that the potential \(q\) can be uniquely determined from the knowledge of the DN map \(\Lambda_{q}\). The authors also discuss some inverse obstacle problems inspired by Schiffer's problem (\url{https://www.scilag.net/problem/P-180522.1}). At the end of this chapter, they also discuss about the fractional Calderón problem (with variable coefficient nonlocal operators) and reformulate it into a geometrical setting. Let \((M,g)\) be a smooth closed connected Riemannian manifold (i.e., compact manifold without boundary) of \(\dim(M)\ge 2\). Given \(0<s<1\), consider the source problem in the nonlocal elliptic equation\N\[\N(-\Delta_{g})^{s}u=f \quad \text{in \(M\),}\N\]\Nwhere \(-\Delta_{g}\) is the positive Laplace-Beltrami operator on \(M\). Given any \(f\in C^{\infty}(M)\) with \((f,1)_{L^{2}(M)}=0\), there exists a unique solution \(u=u^{f}\) to the above equation with \((u^{f},1)_{L^{2}(M)}=0\). Given any open set \(\mathcal{O}\subset M\), one can define the source-to-solution map\N\[\NL_{M,g,\mathcal{O}}(f):=u^{f} \quad \text{for all \(f\in C_{c}^{\infty}(\mathcal{O})\) with \((f,1)_{L^{2}(M)}=0\).}\N\]\NThey present a result showing that \((M,g)\) can be uniquely determined (up to diffeomorphism) by the operator \(L_{M,g,\mathcal{O}}\).\N\N\textbf{Chapter 6.} The authors discuss an inverse problem for the fractional wave equation, which was earlier investigated in [SIAM J. Math. Anal. 54, No. 3, 3379--3419 (2022; Zbl 1492.35427)]. Thanks to the remarkable properties of the fractional Laplacian, the inverse problems for the fractional wave equation have similar results as those inverse problems for the fractional Schrödinger equation introduced in Chapter~3. Let \(\Omega\subset\mathbb{R}^{n}\) be a nonempty bounded Lipschitz domain. Given \(T>0\), \(s\in(0,1)\) and \(q=q(x)\in L^{\infty}(\Omega)\), consider the initial exterior value problem for the wave equation with the fractional Laplacian:\N\[\N\left\{\begin{aligned} & (\partial_{t}^{2} + (-\Delta)^{s}+q)u=0 && \text{in \(\Omega_{T}\),} \\\N& u=f && \text{in \((\Omega_{e})_{T}\)}, \\\N& u=\partial_{t}u=0 && \text{in \(\mathbb{R}^{n}\times\{0\}\).} \end{aligned}\right.\N\]\NHere and after, the authors use the notation \(A_{t}:=A\times(0,t)\) for all \(A\subset\mathbb{R}^{n}\) and \(t>0\). The fractional wave equation can be regarded as a special case of the peridynamics, which models the nonlocal elasticity theory, see e.g. the article ``\textit{Introduction to peridynamics} (\url{https://doi.org/10.1201/9781315373331}) in \textit{Handbook of Peridynamic Modeling}. By the well-posedness of the above exterior-value problem, one can similarly define the DN map\N\[\N\Lambda_{q}[f] := \left.(-\Delta)^{s}u_{f}\right|_{(\Omega_{e})_{T}},\N\]\Nwhere \(u=u_{f}\) is the unique solution to the above initial-exterior-value problem. The authors then discuss the inverse problems from the following perspectives:\N\begin{itemize}\N\item \textit{Global uniqueness.} The potential \(q\) can be uniquely determined by the DN map \(\Lambda_{q}\).\N\item \textit{Logarithmic stability.} Under some apriori assumptions, one can obtain a stability estimate\N\[\N\|q_{1}-q_{2}\|_{Z^{-s-\gamma}(\Omega;t)} \le \omega \left(\|\Lambda_{q_{1}}-\Lambda_{q_{2}}\|_{*}\right)\N\]\Nwhere \(\omega\) is a modulus of continuity satisfying \(\omega(t)\le C|\log t|^{-\mu}\) for all \(0<t\le 1\).\N\item \textit{Optimality of the logarithmic stability.} The authors also discuss about the exponential instability of this inverse problem, which shows that the logarithmic stability estimate is almost optimal.\N\end{itemize}\N\N\textbf{Chapters 7 and 8.} In Chapter~7, the authors consider the fractional semilinear equation of the form\N\[\N\left\{\begin{aligned} & (-\Delta)^{s}u + a(x,u)=0 && \text{in \(\Omega\),} \\\N&u=f && \text{in \(\Omega_{e}\).} \end{aligned}\right.\N\]\Nand demonstrate the use of linearization techniques in addressing these types of problems. In Chapter~8, the authors extend the monotonicity method in Section~3.5 to the case of fractional semilinear elliptic equations with power-type linearities, that is, \(a(x,u)=q(x)u^{m}\) for integer \(m\ge 2\).\N\N\textbf{Chapter 9.} Finally, the authors list some unsolved problems in this chapter.\N\NAs the title suggests, this book provides a summary and review of recent advances in inverse problems involving integro-differential operators. It is also well-suited for graduate students and early-career researchers interested in this active area of study.
- The fractional anisotropic Calderón problem for a nonlocal parabolic equation on closed Riemannian manifolds
- Optimal Runge approximation for nonlocal wave equations and unique determination of polyhomogeneous nonlinearities
- The Calderón problem for the logarithmic Schrödinger equation
- The Calderón problem for fractional space-time parabolic operators with drift
- Recovery of a general analytic Hamiltonian and running costs in mean field games
This page was built for publication: Inverse problems for integro-differential operators
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7009483)
