Dice Question Streamline Icon: https://streamlinehq.com

Directed-graph torsion in magnitude (co)homology beyond the image of the bidirection-forgetful functor

Establish whether there exist directed graphs whose magnitude (co)homology exhibits torsion not arising from any undirected graph via the functor ρ: Graph → Quiver that makes each undirected edge bidirectional; specifically, construct a directed graph D such that torsion in MH_*^*(D) does not coincide with torsion from MH_*^*(ρ(G)) for any undirected graph G, or prove that all torsion in magnitude (co)homology of digraphs lies in the essential image of ρ.

Information Square Streamline Icon: https://streamlinehq.com

Background

Magnitude homology for quivers extends the undirected-graph theory. Prior works have exhibited torsion in magnitude (co)homology of graphs, and this paper proves finite generation results for quiver magnitude cohomology in bounded-genus settings.

The authors explicitly state that it is unknown whether torsion observed for quivers always comes from undirected graphs via the functor ρ that forgets orientation by making edges bidirectional, and pose a concrete question about constructing genuinely directed torsion phenomena.

References

However, we do not know if such torsion always comes from undirected graphs via the functor $\rho$, or not. Is it possible to extend the constructions of and so to get torsion of directed graphs which does not come from the essential image of $\rho\colon Graph\to Quiver$ of Remark~\ref{rem:adjunctiongraphs}?

The weak categorical quiver minor theorem and its applications: matchings, multipaths, and magnitude cohomology (2401.01248 - Caputi et al., 2 Jan 2024) in Section 6 (Applications to magnitude homology)