Anisotropic Calderón Problem for a Non-Local Second Order Elliptic Operator

Susovan Pramanik · 2025

Read Anisotropic Calderón Problem for a Non-Local Second Order Elliptic Operator by Susovan Pramanik (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-12255.txt.

Introduction

Susovan Pramanik, in a solo paper from 2025, studies an inverse problem for a non-local elliptic operator on curved spaces. The question, a version of the Calderon problem, asks whether measurements made only through boundary-style data can reveal the hidden geometry inside.

The operator here spreads its influence across the space rather than acting only locally, and it carries direction-dependent coefficients. Pramanik works on closed curved spaces, where the data consist of pairs of inputs and responses known as the Cauchy data set. He proves that this data determines the geometry of the space up to the standard freedom of relabeling points. Results of this kind matter because they certify what can and cannot be recovered in imaging and probing problems. The same ideas underlie electrical tomography in medicine and geophysics.

Catalog record

CREATORSusovan Pramanik
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-12255

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Anisotropic Calderón Problem for a Non-Local Second Order Elliptic Operator Susovan Pramanik Abstract . This paper investigates the anisotropic Calderón problem for a non-local elliptic operator of order 2, on closed Riemannian manifolds. We demonstrate that using the Cauchy data set, we can recover the geometry of a closed Riemannian manifold up to standard gauge. 1. Introduction The anisotropic Calderón problem is geometric in nature (cf. [ 39 , 47 ] ). Let us state the well-known open problem in the context of a closed, connected Riemannian manifold (M,g) with \dim(M)>2 . Let \mathcal{O}\subset M be a non-empty open set. The Calderón problem examines whether knowledge of the Cauchy data set \displaystyle\mathcal{C}^{\mathcal{O}}_{M,g}=\{(u|_{\mathcal{O}},(-\Delta_{g})u|_{\mathcal{O}})\mid u\in C^{\infty}(M),\,-\Delta_{g}u=0\,\text{in}\,\,M\setminus\overline{\mathcal{O}}\} determines the isometry class of the manifold (M,g) . (-\Delta_{g}) represents the positive Laplace-Beltrami operator on (M,g) . When \dim(M)=2 , the problem is solved using an additional gauge due to conformal invariance of the Laplace-Beltrami operator. See some of the references [ 42 , 4 , 7 , 29 ] . The problem for real-analytic manifolds in dimensions three and above has been addressed (see the references [ 39 , 38 , 37 ] ), but it is still open for smooth manifolds. Positive results have been observed in transversally anisotropic geometries, see [ 16 , 17 ] . Starting from these celebrated works [ 48 , 35 ] , for current research on the Calderón problem, we refer to the survey article [ 51 ] . The purpose of this work is to explore the anisotropic Calderón problem for a second order elliptic psedodifferential operator ((-\Delta_{g})^{2}+m^{2}\mathbb{I})^{\frac{1}{2}} for some fixed m\in\mathbb{R}\setminus\{0\} , on smooth, closed, connected Riemannian manifolds. Let (M,g) be a closed, connected Riemannian manifold with dimension \dim(M)\geq 2 . We define -\Delta_{g} as the positive definite, self-adjoint Laplace-Beltrami operator on L^{2}(M) with domain \mathcal{D}(-\Delta_{g})=H^{2}(M); [ 50 , pp. 25] . The spectrum of (-\Delta_{g}) over (M,g) consists of a discrete sequence of eigenvalues 0=\lambda_{0}<\lambda_{1}\leq\lambda_{2}\leq\cdots\to+\infty, where each eigenvalue \lambda_{k} has a finite multiplicity d_{k} . Define \pi_{k}:L^{2}(M)\to\ker(-\Delta_{g}-\lambda_{k}) as the orthogonal projection onto the eigenspace corresponding to \lambda_{k} . Then, for any function f\in L^{2}(M) , \pi_{k}(f)=\sum_{j=1}^{d_{k}}\langle f,\phi_{k_{j}}\rangle_{L^{2}(M)}\phi_{k_{j}}, where, \{\phi_{k_{j}}\}_{j=1}^{d_{k}} forms an orthonormal basis for \ker(-\Delta_{g}-\lambda_{k}) . Set m\in\mathbb{R}\setminus\{0\} . We consider the positively perturbed bi-Laplace operator on (M,g) as: \mathcal{L}_{g}:=(-\Delta_{g})^{2}+m^{2}\mathbb{I}. This is an unbounded operator on L^{2}(M) , with the domain of definition \mathcal{D}(\mathcal{L}_{g})=H^{4}(M) , see [ 49 ] . We now define the operator of our interest, the \frac{1}{2} -fractional power of \mathcal{L}_{g} , through its action over its domain of definition \mathcal{D}(\mathcal{L}_{g}^{\frac{1}{2}})=H^{2}(M) (cf. [ 49 ] ), as (1) \displaystyle\mathcal{L}_{g}^{\frac{1}{2}}u \displaystyle=\left((-\Delta_{g})^{2}+m^{2}\right)^{\frac{1}{2}}u,\quad u\in H^{2}(M) \displaystyle=\sum_{k=0}^{\infty}(\lambda_{k}^{2}+m^{2})^{\frac{1}{2}}\pi_{k}(u). We note that the operator \mathcal{L}_{g}^{\frac{1}{2}} is an elliptic pseudodifferential with order 2 . For related discussion, see Section 2.2 . Because all eigenvalues of \mathcal{L}_{g} are strictly positive, the operator \mathcal{L}_{g}^{\frac{1}{2}} is invertible over the range space \mathcal{R}(\mathcal{L}_{g})=L^{2}(M) , and its inverse is given by (2) \displaystyle\mathcal{L}_{g}^{-\frac{1}{2}}f \displaystyle=\left((-\Delta_{g})^{2}+m^{2}\right)^{-\frac{1}{2}}f,\quad f\in L^{2}(M) \displaystyle=\sum_{k=0}^{\infty}(\lambda_{k}^{2}+m^{2})^{-\frac{1}{2}}\pi_{k}(f). Let \mathcal{O}\subset M be a non-empty open subset of M . For each f\in C_{0}^{\infty}(\mathcal{O}) , the equation (3) \mathcal{L}^{\frac{1}{2}}_{g}u=f\quad\mbox{in }M has a unique solution u=u_{f}\in C^{\infty}(M)\cap L^{2}(M) . See [ 20 ] . The solution is formally expressed as u=\mathcal{L}_{g}^{-\frac{1}{2}}f (cf. ( 2 )). Now we state our main result. Theorem 1.1 . Let (M_{j},g_{j}) be two smooth, closed, connected Riemannian manifolds, with M_{1}\cap M_{2}\neq\emptyset . Consider \mathcal{O}_{j}\subset M_{j},\,j=1,2 to be two non-empty open sets such that (\mathcal{O}_{1},g_{1})=(\mathcal{O}_{2},g_{2})=:(\mathcal{O},g) . Assume that \mathcal{L}_{g_{1}}^{-\frac{1}{2}}f|_{\mathcal{O}}=\mathcal{L}_{g_{2}}^{-\frac{1}{2}}f|_{\mathcal{O}},\quad\forall f\in C^{\infty}_{0}(\mathcal{O}). Then, there exists a diffeomorphism \Phi:M_{1}\to M_{2} such that \Phi^{\ast}g_{2}=g_{1} . Remark 1.2 . In recent years, the research of non-local inverse problems has been quite dynamic. Starting from the article [ 25 ] , without being exhaustive we mention some references [ 23 , 45 , 46 , 24 , 8 , 12 , 40 , 31 , 5 , 36 , 13 , 22 , 53 , 44 , 30 , 15 ] . However, the operator’s order (cf. Section 2.2 ) in consideration falls within (0,2) . This study extends that limit to 2 using a straightforward example of the square root of a positively perturbed bi-Laplace operator. The fractional power of the perturbed Laplace operator is known as the relativistic Schrödinger operator ((-\Delta_{g})+m^{2}\mathbb{I})^{\alpha} , \alpha\in(0,1) , and has its due significance in mathematical physics. We refer to the work of V. Ambrosio [ 1 , 2 ] on this account. Remark 1.3 . Let m\in(0,\infty) , and we introduce another operator of interest, \mathcal{A}_{g}:=\mathcal{L}_{g}^{\frac{1}{2}}-m\mathbb{I}. The definition is: \mathcal{A}_{g}u=\sum_{k=0}^{\infty}\left((\lambda_{k}^{2}+m^{2})^{\frac{1}{2}}-m\right)\pi_{k}u, where u\in\mathcal{D}(\mathcal{A}_{g})=H^{2}(M) , and \{(\lambda_{k},\pi_{k})\}_{k=0}^{\infty} as previously. Let u\in C^{\infty}(M)\cap L^{2}(M) be the unique solution of the equation \mathcal{A}_{g}u=0\mbox{ in }M with satisfying \langle u,1\rangle_{L^{2}(M)}=0 for f\in C_{0}^{\infty}(\mathcal{O}) . See [ 21 ] . Next, we define the corresponding Cauchy data set: \mathcal{C}^{\mathcal{O}}_{M,g}=\{(u|_{\mathcal{O}},\mathcal{A}_{g}u|_{\mathcal{O}})\mid u\in C^{\infty}(M),\,\,\,\mathcal{A}_{g}u=0\,\text{in}\,M\setminus\overline{\mathcal{O}}\}. This dataset contains solutions to the equation \mathcal{A}_{g}u=0 and serves as the basis on the anisotropic Calderón problem for the operator \mathcal{A}_{g} . Theorem 1.4 . Let (M_{i},g_{i}),\,i=1,2 be two closed, connected Riemannian manifolds with \dim(M_{i})\geq 2 . Assume that \mathcal{O}\subset M_{1}\cap M_{2} is a non-empty open subset where the metrics agree, i.e., (\mathcal{O},g_{1})=(\mathcal{O},g_{2}):=(\mathcal{O},g) . The equality of the Cauchy data set over (\mathcal{O},g) , i.e. \mathcal{C}^{\mathcal{O}}_{M_{1},g_{1}}=\mathcal{C}^{\mathcal{O}}_{M_{2},g_{2}} , implies the existence of a diffeomorphism \Phi:M_{1}\to M_{2} such that \Phi^{\ast}g_{2}=g_{1} . The proof is quite similar to Theorem 1.1 . This paper’s results stem from the work [ 26 ] to solve the nonlocal Calderón’s problem of recovering anisotropic medium or the metric. Subsequent development adds [ 14 , 18 , 10 , 20 , 21 , 19 , 9 ] to further complete the picture in this nonlocal regime. In particular, we now know that, the knowledge of the Cauchy data set \displaystyle\mathcal{C}^{\alpha,\mathcal{O}}_{M,g}=\{(u|_{\mathcal{O}},(-\Delta_{g})^{\alpha}u|_{\mathcal{O}})\mid u\in C^{\infty}(M),\,(-\Delta_{g})^{\alpha}u=0\,\text{in}\,\,M\setminus\overline{\mathcal{O}},\,\,\alpha\in(0,1)\} determines the manifold (M,g) uniquely up to a diffeomorphism. This applies to both closed and open manifolds, see [ 20 , 19 ] . This is based on reducing the problem into the hyperbolic inverse problem, and thanks to the work of [ 33 ] for instance, we have a definite conclusion in the nonlocal analogue of the anisotropic Calderón problem, which is still to be discovered for the local anisotropic Calderón problem. In general, these nonlocal problems are motivated by many models, such as diffusion process [ 3 ] , image processing [ 27 ] , stochastic process [ 6 ] and so on. The paper is organized as follows. In Section 2 , we introduce the fractional power of the elliptic operator using the semigroup approach. Further, we emphasize the symbol and order of these non-local operators. Section 3 is devoted to the proof of our main result. 2. Preliminary understandings. 2.1. Heat semigroup approach in defining the fractional power of \mathcal{L}_{g} Here, we define the fractional power of \mathcal{L}_{g}=((-\Delta_{g})^{2}+m^{2}\mathbb{I}) using the semigroup technique, which is equivalent to the prior definition. Let e^{-t\mathcal{L}_{g}} denote the associated heat semigroup on L^{2}(M) . Let v\in L^{2}(M) , we define (4) e^{-t\mathcal{L}_{g}}v(x)=\int_{M}\mathcal{K}_{\mathcal{L}_{g}}(t,x,y)v(y)\,dV_{g}(y), \mathcal{K}_{\mathcal{L}_{g}}(t,x,y) is the heat kernel for the heat semigroup e^{-t\mathcal{L}_{g}} , which is defined as (5) \displaystyle\mathcal{K}_{\mathcal{L}_{g}}(t,x,y)=\sum_{k=0}^{\infty}e^{-t(\lambda_{k}^{2}+m^{2})}\,\phi_{k}(x)\,\phi_{k}(y) \displaystyle=e^{-m^{2}t}\sum_{k=0}^{\infty}e^{-t\lambda_{k}^{2}}\,\phi_{k}(x)\,\phi_{k}(y) \displaystyle=e^{-m^{2}t}\mathcal{K}_{\Delta_{g}^{2}}(t,x,y), where, K_{\Delta^{2}_{g}}(t,x,y) be the heat kernel for the heat semigroup e^{-t\Delta^{2}_{g}} . Fei He [ 32 ] provides a pointwise estimate of the heat kernel \mathcal{K}_{\Delta_{g}^{2}}(t,x,y) . Theorem 2.1 ( [ 32 ] ) . Let x,y be two

Read the complete text