On Gromov-Hausdorff stability in a boundary rigidity problem (Q538527): Difference between revisions

For a compact Riemannian manifold \(M\) with boundary denote by \(d_{M}(x,y) \) the Riemannian distance between the points \(x,y\in M\). The boundary distance function \(bd_{M}\) is the restriction of \(d_{M}\) to \(\partial M\times\partial M\). Such a manifold \(M\) is called boundary rigid if \(M\) is uniquely determined by \(bd_{M},\) up to an isometry fixing the boundary. \textit{D. Burago} and \textit{S. Ivanov} have proved [Ann. Math. (2) 171, No.~2, 1183--1211 (2010; Zbl 1192.53048)] that, if \(M\) is a region in \(\mathbb{R}^{n}\) with a Riemannian metric which is sufficiently \(C^{2}\)-close to the Euclidean metric \(g_{e},\) then \(M\) is boundary rigid. This result raises naturally the following question: If the boundary distance function of a Riemannian metric \(g\) is close to that of \(g_{e}\) in a suitable topology, is \(g\) necessarily close to \(g_{e}\) in \(C^{r},\) \(r\geq2\) (up to an isometry fixing the boundary)? The results in the present paper are in this spirit. More precisely, the author shows that \(M\) is Gromov-Hausdorff close to a convex Euclidean region \(D\) of the same dimension if the boundary distance function of \(M\) is \(C^{1}\)-close to that of \(D\). More generally, the author proves the same result under the assumptions that the boundary distance function of \(M\) is \(C^{0}\)-close to that of \(D,\) the volumes of \(M\) and \(D\) are almost equal, and the volumes of metric balls in \(M\) have a certain lower bound in terms of the radius.