Approximating coarse Ricci curvature on submanifolds of Euclidean space (Q2127861)

The authors define coarse Ricci curvature derived from iterating Laplacians on smooth metric measure spaces in [the authors, Result. Math. 72, No. 4, 1823--1837 (2017; Zbl 1385.53030)]. \textit{M. Belkin} and \textit{P. Niyogi} shows that one can approximate the Laplacian operator using heat kernels for a closed embedded submanifold of Euclidean space under certain conditions in [J. Comput. Syst. Sci. 74, No. 8, 1289--1308 (2008; Zbl 1157.68056)]. Building on these two works, the authors provide a detailed analysis of approximating the Ricci curvature by a 1-parameter family of operators. To be specific, the authors use a kernel to approximate the Ricci curvature of a closed submanifold of Euclidean space. The kernel is computed in terms of an integrand that employs only ambient space information, even though the integrand will eventually be restricted to the submanifold for the purpose of defining a convolution operator on the submanifold. This approach is similar to the approximation of the Laplace-Beltrami operator on submanifolds of Euclidean space. The inequalities of estimates can be found in Theorems 2.6, 2.8, and 2.9 in their paper. In particular, the authors give a detailed proof of Proposition 3.2 in [the authors, Adv. Math. 342, 14--66 (2019; Zbl 1415.93297)], which is instrumental in the author's manifold learning program for estimating Ricci curvature.