Čech complexes of hypercube graphs
Stemming from questions on the homotopy type of Vietoris-Rips complexes associated to \(n\)-dimensional hypercube graphs \(\mathbb{I}^n\) [\textit{M. Adamaszek} and \textit{H. Adams}, J. Appl. Comput. Topol. 6, No. 2, 177--192 (2022; Zbl 1490.05279)], this work studies homotopical properties of the Čech complexes associated to hypercubes. The study of simplicial complexes such as Vietoris-Rips complexes or Čech complexes is of fundamental importance in applied topology and topological data analysis, and the case of hypercubes, for example, was motivated by questions in mathematical biology.\N\NDenote by \(\mathcal{N}(G,r)\) the Čech complex associated to a graph \(G\), equipped with the path metric, and with scale parameter \(r\). The Čech complexes of hypercubes, for \(r=0,1\), yield wedges of spheres of dimension 0 and 1, respectively. One of the main results of the paper, Theorem 4.1, deals with the scale parameter \(r=2\), and shows that also the Čech complexes of hypercubes \(\mathcal{N}(\mathbb{I}^n,2)\) are homotopy equivalent to a wedge of (2-dimensional) spheres: \(\mathcal{N}(\mathbb{I}^n,2)\simeq \vee_{2^{n-2}(n^2 -3n + 4)-1}S^2\). The main tools used in the proof, such as the nerve theorem, come from combinatorial topology [\textit{A. Björner}, in: Handbook of combinatorics. Vol. 1--2. Amsterdam: Elsevier (North-Holland); Cambridge, MA: MIT Press. 1819--1872 (1995; Zbl 0851.52016)], and are recollected in Section 2. In a different manner, using simplicial collapsibility techniques, it is then shown in Theorem 5.9 that \(\mathcal{N}(\mathbb{I}^n,3)\) is homotopy equivalent to a simplicial complex of dimension at most 4. Furthermore, by Theorem~5.13, for \(n\geq 4\) the reduced homology \(\tilde{H}_i(\mathcal{N}(\mathbb{I}^n,3),\mathbb{Z})\) is non-trivial if and only if \(i\in\{3,4\}\). The investigation of the homotopy type of Čech complexes of hypercubes is far from being complete, as shown in Table~1, and further related open questions are given in Section 7.\N\NAlthough being closely related, Vietoris-Rips complexes and Čech complexes may show different features, but less is known about the persistent homology of Čech complexes. Section~6 of the paper explores the persistent homology of the filtration \(\mathcal{N}(\mathbb{I}^n,-)\), giving a bound on the length of bars. More precisely, Theorem 6.1 shows that the inclusion \(\mathcal{N}(\mathbb{I}^n,r)\hookrightarrow \mathcal{N}(\mathbb{I}^n,r+2)\) is null-homotopic. This fact implies that in the persistent homology of Čech complexes of hypercubes no bars have length longer than two filtration steps. In the case of Vietoris-Rips complexes of hypercube graphs there are no persistence intervals of length longer than one filtration step; the analogous question whether also the inclusions \(\mathcal{N}(\mathbb{I}^n,r)\hookrightarrow \mathcal{N}(\mathbb{I}^n,r+1)\) are null-homotopic is left open in Question~7.5.
This page was built for publication: Čech complexes of hypercube graphs
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7020288)
