Random Geometric Complexes

We study the expected topological properties of Čech and Vietoris-Rips complexes built on randomly sampled points in R. These are, in some cases, analogues of known results for connectivity and component counts for random geometric graphs. However, an important difference in this setting is that homology is not monotone in the underlying parameter. In the sparse range, we compute the expectation and variance of the Betti numbers, and establish Central Limit Theorems and concentration of measure. In the dense range, we introduce Morse theoretic arguments to bound the expectation of the Betti numbers, which is the main technical contribution of this article. These results provide a detailed probabilistic picture to compare with the topological statistics of point cloud data.