Categorification of persistent homology (Q2249475)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Categorification of persistent homology
scientific article

    Statements

    Categorification of persistent homology (English)
    0 references
    0 references
    0 references
    1 July 2014
    0 references
    This paper develops persistent topology from a categorical viewpoint. Thus the main objects of interest are \(({\mathbb R}, \leq)\)-indexed diagrams in some target category -- most usually \(\mathbf{Vec}\) the category of finite-dimensional vector spaces. Note that this includes the cases of \(({\mathbb Z}_{+}, \leq)\) and \(({\mathbb Z}, \leq)\). Given a diagram \(F \in \mathbf{Top}^{(\mathbb R, \leq)}\) the \(p\)-persistent \(k\)th homology group of \(F(a) = f^{-1}(-\infty, a]\) is defined to be the image of the map \(H_{k}F(a \leq a+p)\). The authors define a category of \(\epsilon\)-interleavings of \(({\mathbb R}, \leq)\)-indexed diagrams in a target category. They show that if the target category is abelian then so is the interleaving category. If \(F, G \in \mathbf{D}^{(\mathbb R, \leq)}\) where \(\mathbf{D}\) is some category, the authors define the notion of interleaving distance between \(F\) and \(G\). They show that this is an extended pseudo-mteric. Barcodes and persistence diagrams can be defined for finite-type diagrams in \(\mathbf{Vec}^{(\mathbb R, \leq)}\) the notion of bottleneck distance is thus introduced into this categorical framework. The authors show that the set of finite bar codes with the bottleneck distance embeds isometrically into the space \(\mathbf{Vec}^{(\mathbb R, \leq)}\) with the interleaving distance. Finally the authors show how the notion of stability of persistent homology fits into their categorical framework.
    0 references
    0 references
    applied topology
    0 references
    persistent topology
    0 references
    topological persistence
    0 references
    diagrams indexed by the poset of real numbers
    0 references
    interleaving distance
    0 references
    bottleneck distance
    0 references
    barcodes
    0 references
    0 references
    0 references
    0 references