Quantitative simplification of filtered simplicial complexes (Q2225664)

Filtered simplicial complexes arise in many areas of algebraic topology, and are particularly prevalent in topological data analysis, where they give information about complex data sets by means of the persistent homology associated to a filtered complex, which is algorithmically generated from the data. A major issue in topological data analysis is clearly computability of persistent homology of large filtered simplicial complexes. This paper introduces a new invariant of filtered simplicial complexes called codensity, which controls the effect of removing a vertex and all simplices containing it from a filtered simplicial complex on persistent homology. The paper also introduces an interleaving type pseudo-distance function \(d_I^F\) for filtered simplicial complexes. The authors prove a stability theorem for this pseudo-metric, and use it to define vertex codensity which allows them to obtain an upper bound on the distance between a filtered complex \(X^*\) and the same complex with a vertex removed. Finally, the paper shows for any filtered finite simplicial complex, there exists a full subcomplex \(C^*\) that is minimal in the appropriate sense and is unique up to isomorphism, such that \(d_I^F(X^*, C^*)=0\).