Ricci curvature and the manifold learning problem (Q1631004)

Consider the approximate determination of the Ricci curvature of low-dimensional submanifolds $\Sigma^d$ of a certain domain $\Sigma \subset \mathbb{R}^N$ for large $N$. Problems of this type arise, e.g., in topological data analysis, in the measure of the robustness of networks, etc.. The curvature is approximated using the induced metric from a sample of $n$ i.i.d. points ${\xi_1,\dots,\xi_n}$, generated in or in a vicinity of $\Sigma^d$ by realizations of a uniformly distributed random vector in $\Sigma$. Asymptotic results for $n \rightarrow \infty$ are given.