Metrics for generalized persistence modules

From MaRDI portal
(Redirected from Publication:895700)



Abstract: We consider the question of defining interleaving metrics on generalized persistence modules over arbitrary preordered sets. Our constructions are functorial, which implies a form of stability for these metrics. We describe a large class of examples, inverse-image persistence modules, which occur whenever a topological space is mapped to a metric space. Several standard theories of persistence and their stability can be described in this framework. This includes the classical case of sublevelset persistent homology. We introduce a distinction between `soft' and `hard' stability theorems. While our treatment is direct and elementary, the approach can be explained abstractly in terms of monoidal functors.


Persistent topology is one of the main tools of topological data analysis. Its key idea is to study filtered spaces, classically pairs \((X, f)\) where \(X\) is a topological space and \(f:X\to \mathbb{R}\) is a filtering function, in the sense that the filtration is given by the sublevel sets \(f^{-1}\left( (-\infty, a]\right)\) for \(a\in \mathbb{R}\) [\textit{H. Edelsbrunner} and \textit{J. Harer}, Contemporary Mathematics 453, 257--282 (2008; Zbl 1145.55007)]. In shape recognition or retrieval it is essential to have distances; for pairs \((X, f)\), the first distance introduced was the \textit{natural pseudodistance}, defined already for \(n\)D valued functions \(f\) in [\textit{P. Frosini} and \textit{M. Mulazzani}, Bull. Belg. Math. Soc. Simon Stevin 6, No. 3, 455--464 (1999; Zbl 0937.55010)]. Soon enough, the attention focused on the \textit{persistence module}, a powerful descriptor of such pairs: It is a collection of vector spaces, indexed by the range of \(f\), classically obtained as images of the homology modules under the homorphisms induced by the inclusions of sublevel sets [\textit{A. Zomorodian} and \textit{G. Carlsson}, Discrete Comput. Geom. 33, No. 2, 249--274 (2005; Zbl 1069.55003)]. (But equivalent pairs may fail to be distinguished by their persistence modules: [\textit{F. Cagliari} et al., in: M. Ferri (ed.) et al., Computational topology in image context. 4th international workshop, CTIC 2012, Bertinoro, Italy, May 28--30, 2012. Proceedings. Berlin: Springer. Lecture Notes in Computer Science 7309, 148--156 (2012; Zbl 1357.68262)].) A distance between persistence modules was then introduced in [\textit{F. Chazal} et al., Proc. of the 25th annual symposium on Computational geometry, ACM, 237--246 (2009)] and extended to multidimensional indexing in [\textit{M. Lesnick}, Found. Comput. Math. 15, No. 3, 613--650 (2015; Zbl 1335.55006)]: the \textit{interleaving distance}. Categorification of the process -- started \textit{ante litteram} in [\textit{F. Cagliari} et al., Acta Appl. Math. 67, No.3, 225--235 (2001; Zbl 0995.68098)] -- has been performed systematically in [\textit{P. Bubenik} and \textit{J. A. Scott}, Discrete Comput. Geom. 51, No. 3, 600--627 (2014; Zbl 1295.55005)]. The present article completes the project in two directions: First, it extends the indexing to any preordered set. Second, it develops a full theory of the interleaving distance. Keystone of the paper is the notion of \textit{translation} in a preordered set; the notions of \textit{sublinear projection} and of \textit{superlinear family} provide the connection with the classic interleaving distance. The metric setting is as wide as the one of \textit{Lawvere metrics} (no symmetry required, distinct objects may have zero distance). It might seem that the authors have lost sight of the original object of persistence, plunging in a totally formal setting, but this is not the case: Section 4.3, \textit{Inverse-Image Persistence Theories}, is devoted to recover most flavours of the current topological persistence as particular cases. Stability is also treated in a general way. All in all, this paper gives the (presently?) most general and most powerful setting for dealing with distances of persistence modules.




Cited in
(69)








This page was built for publication: Metrics for generalized persistence modules

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q895700)