Reconstruction and interpolation of manifolds. I: The geometric Whitney problem (Q2216248)

The authors investigate how a Riemannian manifold \((M, g)\) may approach a metric space \((X, dx )\). To approximate a point cloud in \(\mathbb{R}^m\), a smooth \(n\)-dimensional submanifold \(S\subset \mathbb{R}^m\), \(m > n\), must be created. These issues arise in differential geometry, machine learning, and many inverse problems. The authors show that a metric space may be approximated in the Gromov-Hausdorff sense by a Riemannian manifold of fixed dimension, sectional curvature, and injectivity radius. They use the findings to characterize Alexandrov spaces with two-sided curvature bounds. Furthermore, they define in the Hausdorff metric subsets of Euclidean spaces that have fixed dimension, bound main curvatures, and normal injectivity radius. In this paper, the authors propose algorithmic methods to solve the geometric Whitney issue in metric space and the manifold reconstruction problem in Euclidean space.