Endomorphism and automorphism graphs of finite groups
Midhuna V Ajith, Peter J Cameron, Mainak Ghosh, Aparna Lakshmanan S
TL;DR
The paper defines and analyzes endomorphism and automorphism graphs associated with finite groups, embedding them in the framework of transformation monoids. It develops compressed and identity-deleted variants, studies their structural properties (planarity, girth, completeness, diconnectivity) and behavior under direct products, and provides detailed computations for cyclic groups, abelian $p$-groups, dihedral/dicyclic/symmetric/metacyclic groups. It also presents counterexamples to natural questions about isomorphisms of graphs and digraphs and discusses the relationship between endomorphism and power graphs. The results yield a nuanced portrait of how endomorphism structure encodes group-theoretic information, with complete classifications in several families and several open questions for nonabelian cases. The work contributes methodologically by leveraging the transformation-monoid viewpoint and conceptually by highlighting the limits of endomorphism data in distinguishing groups.
Abstract
Let $G$ be a group. The directed endomorphism graph, $\dend(G)$ of $G$ is a directed graph with vertex set $G$ and there is a directed edge from the vertex $a$ to the vertex $b$ if $a \neq b$ and there exists an endomorphism on $G$ mapping $a$ to $b$. The endomorphism graph, $\uend(G)$ is the corresponding undirected simple graph. The automorphism graph of $G$ is similarly defined for automorphisms: it is a disjoint union of complete graphs on the orbits of $\Aut(G)$. The endomorphism digraph is a special case of a digraph associated with a transformation monoid, and we begin by introducing this. We have explored graph theoretic properties like size, planarity, girth etc. and tried finding out for which types of groups these graphs are complete, diconnected, trees, bipartite and so on, as well as computing these graphs for some special groups. We conclude with examples showing that things are not always simple.
