Computing persistent Stiefel-Whitney classes of line bundles

From MaRDI portal



Abstract: We propose a definition of persistent Stiefel-Whitney classes of vector bundle filtrations. It relies on seeing vector bundles as subsets of some Euclidean spaces. The usual v{C}ech filtration of such a subset can be endowed with a vector bundle structure, that we call a v{C}ech bundle filtration. We show that this construction is stable and consistent. When the dataset is a finite sample of a line bundle, we implement an effective algorithm to compute its persistent Stiefel-Whitney classes. In order to use simplicial approximation techniques in practice, we develop a notion of weak simplicial approximation. As a theoretical example, we give an in-depth study of the normal bundle of the circle, which reduces to understanding the persistent cohomology of the torus knot (1,2). We illustrate our method on several datasets inspired by image analysis.


A standard method of topological data analysis (TDA) typically applied to point cloud data is to consider the Čech complex or the Vietoris-Rips complex associated to the point cloud at varying radii. The resulting sequence of simplicial complexes yields a corresponding sequence of chain complexes whose homology and the corresponding persistence diagrams are the fundamental tools used in the TDA. The current paper proposes an analogous framework for vector bundles where the algebraic invariants in use are Stiefel-Whitney classes. In order to define Stiefel-Whitney classes on a space \(X\) in a way suitable for persistence, the author considers (real) vector bundles as continous maps \(p\colon X\to\mathcal{G}_d(\mathbb{R}^m)\), for sufficiently large \(m\), where \(\mathcal{G}_d(\mathbb{R}^m)\) is the Grassmannian of \(d\)-dimensional hyperplanes in \(\mathbb{R}^m\). Let \(X\subset \mathbb{R}^n\) be a finite subset, and let \(p\colon X\to \mathcal{G}_d(\mathbb{R}^m)\) be a vector bundle on \(X\). Consider the bundle as a subset of \(\mathbb{R}^n\times \mathcal{G}_d(\mathbb{R}^m)\) by defining \[ \check{X}= \{(x, p(x))\;|\; x\in X\}\subseteq \mathbb{R}^n\times M_m(\mathbb{R}), \] where \(M_m(\mathbb{R})\) is the space of \(m\times m\) real valued matrices that naturally contains the Grassmannian. One now defines the Čech filtration \((\check{X}^t)_{t\geq 0}\) associated to \(\check{X}\), where \(\check{X}^t\) is the \(t\)-thickening of \(\check{X}\) in the ambient space \(\mathbb{R}^n\times M_m(\mathbb{R})\). This setup allows the author to define a map \(p^t\colon \check{X}^t\to \mathcal{G}_d(\mathbb{R}^m)\), for each \(t\ge 0\), and hence the Stiefel-Whitney classes \[ w_i^t(X) = (p^t)^*(w_i), \] where \(w_i\) is the \(i\)-th Stiefel-Whitney class in \(H^i(\mathcal{G}_d(\mathbb{R}^m),\mathbb{Z}_2)\). The information thus obtained for varying dimensions \(i\) and values of \(t\) is then encoded in a diagram, similar in nature to the standard barcode used in TDA. The main theoretical result in this paper is Theorem 1, which is a stability statement of persistent Stiefel-Whitney classes, followed by a number of nice corollaries. The author also provides an algorithm for computing persistent Stiefel-Whitney classes and a publically available implementation. The methods are clearly generalisable to other types of characteristic classes, such as Chern classes for complex vector bundles, etc.











This page was built for publication: Computing persistent Stiefel-Whitney classes of line bundles

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2125405)