Approximating persistent homology in Euclidean space through collapses (Q2352514)

In studying the persistent homology of a data set one derives a Čech complex by choosing input points from the set. Because the number of simplices in this Čech complex grows exponentially with the number of input points chosen this method poses serious computational problems. In this paper the authors describe a method of collapsing in a simplicial complex that has two computational benefits. One is that by using a simplex tree that is a \textit{trie} (also called a prefix tree) one obtains an efficient description of the simplex. The second uses the collapses at points in the algorithm for computing the persistent homology to reduce the size of the complex. Furthermore the authors show the accuracy of the approximations obtained by these methods. The paper describes experimental results that test the feasibility of the authors' implementation.