A Functorial Link between Quivers and Hypergraphs

Abstract

This paper discusses some issues arising from the category $\mathfrak{H}$ of hypergraphs, the category $\mathfrak{M}$ of (undirected) multigraphs, and the topos $\mathfrak{Q}$ of quivers. First, the natural inclusion of $\mathfrak{M}$ into $\mathfrak{H}$ admits a right adjoint functor by deleting all nontraditional edges. Dually, the operations of taking the underlying multigraph of a quiver and taking the associated digraph of a multigraph form an adjoint pair between $\mathfrak{M}$ and $\mathfrak{Q}$. On the other hand, neither $\mathfrak{H}$ nor $\mathfrak{M}$ is cartesian closed, meaning that neither is a topos like $\mathfrak{Q}$. Moreover, despite $\mathfrak{M}$ being a subcategory of $\mathfrak{H}$, $\mathfrak{H}$ does not have enough projective objects while $\mathfrak{M}$ admits a projective cover for every object.