Resolution of a conjecture about linking ring structures (Q6169102)

The notion of LR-structures was introduced by \textit{P. Potočnik} and \textit{S. E. Wilson} [Ars Math. Contemp. 7, No. 2, 341--352 (2014; Zbl 1306.05251)] as a tool to study tetravalent semisymmetric graphs of girth \(4\). In this paper, the authors use the methods of group amalgams to resolve some problems left open in [loc. cit.] namely Question 1 and Conjecture 8.1. Let \(\Gamma\) be a connected (but not necessarily finite) tetravalent graph and let \(G\) be a vertex-transitive group of automorphisms of \(\Gamma\). A rough classification of these can be obtained by considering the permutation group \(G_v^{\Gamma(v)}\!\) induced by the action of the vertex-stabiliser \(G_v\) on the neighbourhood \(\Gamma(v)\) of a vertex \(v\). If \(G_v^{\Gamma(v)}\) is transitive, then \(G\) acts transitively on the set \(A(\Gamma)\) of arcs (ordered pairs of adjacent vertices) of \(\Gamma\). Arc-transitive graphs have been studied extensively for decades and many aspects of the tetravalent case are now well understood, see for example [\textit{D. Ž. Đoković}, Proc. Am. Math. Soc. 80, 22--26 (1980; Zbl 0441.20015); \textit{A. Gardiner} and \textit{C. E. Praeger}, Eur. J. Comb. 15, No. 4, 383--397 (1994; Zbl 0806.05038); \textit{P. Potočnik}, Eur. J. Comb. 30, No. 5, 1323--1336 (2009; Zbl 1208.05056); \textit{P. Potočnik} et al., J. Symb. Comput. 50, 465--477 (2013; Zbl 1256.05102)]. The next natural case is when the group \(G_v^{\Gamma(v)}\) has two orbits. This case splits into a number of subcases. If one of the orbits of \(G_v^{\Gamma(v)}\) has length \(1\) (and the other has length \(3\)), then one can study the graph \(\Gamma\) through the associated \(6\)-valent \(G\)-arc-transitive graph \(\Gamma'\) obtained by merging pairs \(\{u,w\}\) of vertices of \(\Gamma\) with the property that \(G_v=G_w\). A related approach was used in [Potočnik et al., loc. cit.] to study vertex- but not arc-transitive \(3\)-valent graphs using some theory of \(4\)-valent arc-transitive graphs. This paper deals with a more complex situation where \(G_v^{\Gamma(v)}\) has two orbits of length \(2\). In this case, \(G_v^{\Gamma(v)}\) is permutation isomorphic to either the cyclic group \(C_2 = \langle (1,2)(3,4)\rangle\) or the Klein group \(V_4 = \langle (1,2), (3,4) \rangle\), in their respective intransitive actions on \(4\) points. In these two cases, the group \(G\) has precisely two orbits on the arc-set \(A(\Gamma)\), but can have either one or two orbits on the edge-set \(E(\Gamma)\). If \(G\) has a single orbit on \(E(\Gamma)\), then the graph \(\Gamma\) belongs to the graphs admitting a half-arc-transitive group action, which has been studied extensively, see [\textit{J. A. Al-Bar} et al., Electron. J. Comb. 23, No. 1, Research Paper P1.10, 23 p. (2016; Zbl 1329.05142); \textit{M. D. E. Conder} et al., Ars Math. Contemp. 8, No. 1, 149--162 (2015; Zbl 1317.05190); \textit{Š. Miklavič} et al., Discrete Math. 344, No. 3, Article ID 112263, 22 p. (2021; Zbl 1456.05078); \textit{P. Potočnik} et al., Ars Math. Contemp. 8, No. 1, 133--148 (2015; Zbl 1317.05191); \textit{A. Ramos Rivera} and \textit{P. Šparl}, J. Comb. Theory, Ser. B 135, 256--278 (2019; Zbl 1404.05081); \textit{P. Spiga}, Graphs Comb. 32, No. 5, 2135--2144 (2016; Zbl 1349.05168); \textit{P. Spiga} and \textit{B. Xia}, J. Comb. Theory, Ser. A 180, Article ID 105406, 17 p. (2021; Zbl 1512.05405); \textit{B. Xia}, J. Comb. Theory, Ser. B 147, 159--182 (2021; Zbl 1458.05266)]. On the other hand, the case where \(G\) has two orbits on \(E(\Gamma)\) has received much less attention so far. In this case, the analysis can again be split into two subcases, depending on whether \(G_v^{\Gamma(v)}\) is isomorphic to \(C_2\) or to \(V_4\). In the first subcase, one can easily see that the vertex-stabiliser \(G_v\) is itself isomorphic to \(C_2\), which forces the group \(G\) to be relatively small (indeed \(|G| = 2|V(\Gamma)|\)), which allows the use of a number of standard group-theoretical approaches. In this paper, the authors are interested in the remaining subcase, where \(G_v^{\Gamma(v)} \cong V_4\) and \(G\) is intransitive on \(E(\Gamma)\). This relates to the notion of LR-groups which was introduced in [Potočnik and Wilson, loc. cit.]: Let \(G\) be a vertex-transitive group of automorphisms of a connected tetravalent graph \(\Gamma\) such that \(G_v^{\Gamma(v)}\) is permutation isomorphic to the Klein group \(V_4\) in its faithful intransitive action on four points. If \(G\) has \textit{two} orbits on \(E(\Gamma)\), then the group \(G\) is called an \textit{LR-group} of automorphisms of \(\Gamma\). Let \(\Gamma\) be a connected tetravalent graph, let \(\mathcal{C}\) be a partition of \(E(\Gamma)\) into cycles and let \(\{\mathcal{L},\mathcal{R}\}\) be a partition of \(\mathcal{C}\) such that every vertex of \(\Gamma\) is incident to one cycle in \(\mathcal{L}\) and one cycle in \(\mathcal{R}\). Define \(\Aut(\Gamma,\mathcal{C}) = \{ g\in \Aut(\Gamma) \mid \mathcal{C}^g = \mathcal{C} \}\) and \(\Aut^+(\Gamma,\mathcal{C}) = \{g\in \Aut(\Gamma,\mathcal{C}) \mid \mathcal{L}^g = \mathcal{L} \text{ and } \mathcal{R}^g=\mathcal{R}\}\). Then the pair \((\Gamma,\mathcal{C})\) is called an \textit{LR-structure}, and \(\mathcal{C}\) is called an \textit{LR-decomposition} of \(\Gamma\) if the group \(\Aut^+(\Gamma,\mathcal{C})\) acts transitively on \(V(\Gamma)\), and for every \(v\in V(\Gamma)\) and for every cycle \(C\in \mathcal{C}\) passing through \(v\), some \(g\in \Aut^+(\Gamma,\mathcal{C})_v\) acts as a reflection on \(C\) and fixes every vertex of the other cycle in \(\mathcal{C}\) passing through \(v\). An LR-structure \((\Gamma,\mathcal{C})\) is called \textit{self-dual} provided that \(\Aut^+(\Gamma,\mathcal{C})\) is a proper subgroup of \(\Aut(\Gamma,\mathcal{C})\) and is \textit{non-self-dual} if \(\Aut^+(\Gamma,\mathcal{C})= \Aut(\Gamma,\mathcal{C})\). In this paper, the authors give an affirmative answer to the question below asked in [Potočnik and Wilson, loc. cit.]. They also prove that the conjecture below posed in [Potočnik and Wilson, loc. cit.] is correct. {Question} (Question 1 in [Potočnik and Wilson, loc. cit.]) If \(\mathcal{C}\) and \(\mathcal{C}'\) are two distinct LR-decompositions of a finite tetravalent graph \(\Gamma\), is it true that there exists \(g\in \Aut(\Gamma)\) such that \(\mathcal{C}^g = \mathcal{C}'\)? {Conjecture} (Conjecture 8.1 in [Potočnik and Wilson, loc. cit.]) If \((\Gamma,\mathcal{C})\) is a finite LR-structure for which \(\Aut^+(\Gamma, \mathcal{C})\) is a proper subgroup of \(\Aut(\Gamma)\), then \((\Gamma,\mathcal{C})\) is self-dual.