Non-uniqueness results for the anisotropic Calderón problem with data measured on disjoint sets (Q2421908)

In this article the authors prove non-uniqueness results for the Calderón problem when the Dirichlet and Neumann boundary data are fixed along disjoint subsets of the boundary. The counterexamples are exhibited over certain specific two and three dimensional Riemannian manifolds however admit straightforward generalizations to higher dimensions. More precisely, let $(M,g)$ be a Riemannian manifold with smooth boundary $\partial M$ and let $\Gamma_D$ and $\Gamma_N$ be two open subsets in the boundary. Let $-\Delta_g$ denote the positive (scalar) Laplacian and consider the Dirichlet problem \[ \begin{cases} -\Delta_gu=\lambda^2u & \text{on }M\\ u=\psi & \text{on }\Gamma_D\\ u=0 & \text{on }\partial M\setminus\Gamma_D \end{cases} \] with associated \textit{partial Dirichlet-to-Neumann map $\Lambda_{g,\Gamma_D,\Gamma_N}(\lambda^2)$ at frequency $\lambda^2$} (partial DN map for short) defined by \[ \Lambda_{g,\Gamma_D,\Gamma_N}(\lambda^2)(\psi):= (\partial_\nu u)\vert_{\Gamma_N}. \] The \textit{anisotropic Calderón problem} can be in general formulated as follows: Does the knowledge of $\Lambda_{g,\Gamma_D,\Gamma_N}(\lambda^2)$ determine the metric $g$ uniquely? This is the main question of the article but it is targeted in various further refined forms (see questions Q1, Q2 and Q3 in the Introduction). The main result of the article is that if $\Gamma_D\cap\Gamma_N=\emptyset$ then uniqueness fails i.e. the bulk metric $g$ cannot be recovered from knowing the partial DN map alone when the open subsets of the boundary along which boundary data are known are disjoint (see Theorems 1.1 and 1.2 in the article). The two counterexamples are formulated over $S^1\times [0,1]$ with a conformally product metric and $S^1\times S^1\times [0,1]$ with a diagonal metric, respectively, but the authors remark that their results admit generalizations to more general metrics and higher dimensions as well.