Another look at recovering local homology from samples of stratified sets (Q2240089)

A large part of Topological Data Analysis has to do with deducing the topology of an object out of a sampling of it; this is, e.g., the case with persistent homology [\textit{G. Carlsson}, Bull. Am. Math. Soc., New Ser. 46, No. 2, 255--308 (2009; Zbl 1172.62002)]. Much is known about the influence of the sample density on the possibility of computing the homology of the sampled object [\textit{P. Niyogi} et al., Discrete Comput. Geom. 39, No. 1--3, 419--441 (2008; Zbl 1148.68048)]. It may be important to obtain not just the global homology of the object of interest (a subset \(K\) of a metric space), but also local homology: the relative homology \(H(K, K- \{x\})\) for all points \(x\in K\). This may reveal, e.g., the presence and nature of singular points. The introduction of the present paper quotes and discusses previous research on this theme [\textit{P. Bendich} et al., ``Inferring local homology from sampled stratified spaces'', in: 48th Annual IEEE Symposium on Foundations of Computer Science (FOCS'07), 536--546 (2007); \textit{P. Skraba} and \textit{B. Wang}, in: Proceedings of the 25th annual ACM-SIAM symposium on discrete algorithms, SODA 2014, Portland, OR, USA, January 5--7, 2014. Philadelphia, PA: Society for Industrial and Applied Mathematics (SIAM); New York, NY: Association for Computing Machinery (ACM). 174--192 (2014; Zbl 1422.68258); \textit{T. K. Dey} et al., ``Dimension detection with local homology'', in: Proc. 26th Canadian Conference on Computational Geometry, 273--279 (2014)]. The paper itself is aimed at progress on that research line for a large class of stratified sets, in particular subsets \(K\) of \(\mathbb{R}^n\) that are compact neighborhood retracts and possess a Whitney stratification [\textit{M. J. Pflaum}, Analytic and geometric study of stratified spaces. Berlin: Springer (2001; Zbl 0988.58003)]. Key ideas are a smart interleaving of spaces and a particular property (\textit{seemliness}), declined in various ways, which enables the computation of local homology and is recognized in relevant settings (Thms. 3 to 5). The proofs are detailed and exhaustive, the examples are simple but meaningful.