On edge-primitive and 2-arc-transitive graphs
Zaiping Lu

TL;DR
This paper investigates the structure of edge-primitive, 2-arc-transitive graphs, proving they have almost simple automorphism groups unless they are cycles or complete bipartite graphs, and provides examples with specific symmetry properties.
Contribution
It establishes a classification result for finite 2-arc-transitive edge-primitive graphs and presents explicit examples with high symmetry.
Findings
Finite 2-arc-transitive edge-primitive graphs have almost simple automorphism groups.
Examples of 3-arc-transitive graphs with faithful vertex-stabilizers are provided.
The classification excludes cycles and complete bipartite graphs.
Abstract
A graph is edge-primitive if its automorphism group acts primitively on the edge set. In this short paper, we prove that a finite 2-arc-transitive edge-primitive graph has almost simple automorphism group if it is neither a cycle nor a complete bipartite graph. We also present two examples of such graphs, which are 3-arc-transitive and have faithful vertex-stabilizers.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsFinite Group Theory Research · Coding theory and cryptography · semigroups and automata theory
On edge-primitive and -arc-transitive graphs
Zai Ping Lu
** Zaiping Lu**
**Center for Combinatorics, LPMC, Nankai University, Tianjin 300071, China **
Abstract.
A graph is edge-primitive if its automorphism group acts primitively on the edge set. In this short paper, we prove that a finite -arc-transitive edge-primitive graph has almost simple automorphism group if it is neither a cycle nor a complete bipartite graph. We also present two examples of such graphs, which are -arc-transitive and have faithful vertex-stabilizers.
Keywords**. Primitive group, almost simple group, edge-primitive graph, -arc-transitive graph.**
2010 Mathematics Subject Classification. 05C25, 20B25
Supported by the National Natural Science Foundation of China (11731002) and the Fundamental Research Funds for the Central Universities.
1. Introduction
All graphs and groups considered in this paper are assumed to be finite.
A graph in this paper is a pair of a nonempty set and a set of -subsets of . The elements in and are called the vertices and edges of , respectively. The number of vertices is called the order of . For , the set is called the neighborhood of in , while is the valency of . We say that has valency or is -regular if its vertices all have equal valency . For an integer , an -arc in is an -tuple of vertices such that and for all possible . A -arc is also called an arc.
Let be a graph. A permutation on is called an automorphism of if for all . Let denote the set of all automorphisms of . Then is a subgroup of the symmetric group , and called the automorphism group of . Note that the group has a natural action on the edge set (and also on the set of -arcs). The graph is called edge-transitive if and for each pair of edges there exists some mapping one of these two edges to the other one. (Similarly, we may define vertex-transitive, arc-transitive or -arc-transitive** graphs.) An edge-transitive graph is called edge-primitive if some (and hence every) edge-stabilizer, the subgroup of its automorphism group which fixes a given edge, is a maximal subgroup of the automorphism group.**
It is well-known that edge-transitive graphs and hence edge-primitive graphs are either bipartite or vertex-transitive. As a subclass of the edge-transitive graphs, edge-primitive graphs posses more restrictions on their symmetries and automorphism groups. For example, a connected edge-primitive graph is necessarily arc-transitive provided that it is not a star graph. In ****[9]****, appealing to the O’Nan-Scott Theorem for (quasi)primitive groups ****[22]****, Giudici and Li investigated the structural properties of edge-primitive graphs, particularly, on their automorphism groups. Let be an arc-transitive and edge-primitive graph which is neither a cycle nor a complete bipartite graph. If is bipartite then let be the subgroup of preserving the bipartition. By ****[9]****, as a primitive group on , only of the eight O’Nan-Scott types for (quasi)primitive groups may occur for , say SD, CD, PA and AS. For the first two types, is bipartite and is quasiprimitive of type CD on each bipartite half. For the last two types, with one exception case, or is quasiprimitive on or on each bipartite half respectively of the same type for on . In this paper, we will work on the types of on and on under the further assumption that is -arc-transitive.
The interests for edge-primitive graphs arises partially from the fact that many (almost) simple groups may be represented as the automorphism groups of edge-primitive graphs. Consulting the Atlas ****[3]****, one may get first-hand such examples. For example, the sporadic Higman-Sims group is the automorphism group of a rank graph with order and valency , which is in fact a -arc-transitive and edge-primitive graph; the sporadic Rudvalis group is the automorphism group of a rank graph with order and valency , which is edge-primitive but not -arc-transitive. Besides, the almost groups , , and all have representations on edge-primitive graphs. The reader may refer to ****[11, 12, 18, 21, 26]**** for more examples of edge-primitive graphs which have almost simple automorphism groups. Of course, using the constructions given in ****[9]****, one can easily construct examples of edge-primitive graphs with automorphism groups not almost simple.
We have a strong impression from the known examples for edge-primitive graphs in the literature that a -arc-transitive and edge-primitive graph has almost simple automorphism group unless it is a cycle or a complete bipartite graph. Yet could it be so? Yes, it is true! We shall prove the following result in Section 3.
Theorem 1.1**.**
Let be an edge-primitive -regular graph for some . If is -arc-transitive, then either is a complete bipartite graph, or has almost simple automorphism group.
Remarks on Theorem 1.1.
(1) Li and Zhang ****[18]**** proved that -arc-transitive and edge-primitive graphs have almost automorphism groups. Further, as a sequence of their classification on almost simple primitive groups with soluble point-stabilizers, they give a complete list for -arc-transitive and edge-primitive graphs.
(2) By Theorem 1.1, appealing to the classification of almost simple groups with soluble maximal subgroups, it might be feasible to classify -arc-transitive and edge-primitive graphs with soluble edge-stabilizers.
2. Preliminaries
For the subgroups of (almost) simple groups, we sometimes follow the notation used in the Atlas ****[3]****, while we also use and to denote respectively the cyclic group of order and the elementary abelian group of order .
2.1. Primitive groups
In this subsection, is nonempty finite set, and is a transitive subgroup of the symmetric group . Let be the socle of , that is, is generated by all minimal normal subgroups of .
Consider the point-stabilizer , where . Then
- (1)
** is primitive if is a maximal subgroup of ;** 2. (2)
** is -transitive if is -transitive on , that is, all -orbits on have equal length ;** 3. (3)
** is a Frobenius group if is semiregular on ;** 4. (4)
** is -transitive if is transitive on .**
Note that (4) implies (1) and (2), and (2) implies (1) or (3) (refer to ****[29, Theorem 10.4]****).
Let , a normal subgroup of . Then is -transitive, and , and so is contained in the normalizer of in . Thus, if is maximal then either or . The former case yields , while the latter case gives
[TABLE]
Then we have following simple fact for primitive groups.
Lemma 2.1**.**
If is primitive and then either is regular on or is self-normalized; if is -transitive and then is either regular or -transitive on .
For an almost simple -transitive group , each non-trivial normal subgroup of is primitive, and in fact -transitive except for the case where acting on pionts, refer to ****[1, page 197, Table 7.4]****. Next we consider the normal subgroups of affine -transitive groups. Refer to ****[1, page 195, Table 7.3]**** for a complete list of affine -transitive groups. We consider the affine -transitive groups in their natural actions.
Lemma 2.2**.**
Let be an affine -transitive group and . If is imprimitive on , then is a soluble Frobenius group, is cyclic, and either or , where is not a prime.
Proof. Assume that is imprimitive. Then , and so . Further, by Lemma 2.1 and ****[29**, Theorem 10.4], is a Frobenius group. Let for a prime . We may write , and . Since is imprimitive, is not maximal in , and thus is a normal reducible subgroup of . Then, by ****[13**, Lemma 5.1]****, is cyclic and is a divisor of , where and l\,\big{|}\,k. Finally, the lemma follows from checking all affine -transitive groups one by one. **
If every minimal normal subgroup of is transitive on , then called a quasiprimitive group. Praeger ****[22, 24]**** generalized the O’Nan-Scott Theorem for primitive groups to quasiprimitive groups, which says that a quasiprimitive group has one of the following eight types: HA, HS, HC, TW, AS, SD, CD and PA. In particular, if is quasiprimitive then has at most two minimal normal subgroups, and if two (for HS and HC) then they are isomorphic and regular.
Suppose that has a transitive insoluble minimal normal subgroup . Then for . Write for isomorphic nonabelian simple groups and integer . Then acts transitively on by conjugation. Note that, for and ,
[TABLE]
Then acts transitively on by conjugation. Clearly, ; however, the equality is not necessarily holds even if is quasiprimitive. A sufficient condition for this equality is that is primitive and of type AS or PA, refer to ****[4, Theorem 4.6]**** and its proof. In survey, we have the simple fact as follows.
Lemma 2.3**.**
Assume that has a transitive minimal normal subgroup , where are isomorphic nonabelian simple groups. Let . Then acts transitively on by conjugation. If further is primitive and of type AS or PA, then .
2.2. Locally-primitive graphs
In this subsection, is a connected -regular graph for some , and . Assume further that the graph is -locally primitive, that is, acts primitively on for all .
Fix an edge . Note that induces a primitive permutation group (on ). Let be the kernel of acting on . Then . Set . Then induces a normal subgroup of with the kernel . Writing and in group extensions,
[TABLE]
Assume that is transitive on . Then is a -group for some prime , refer to ****[6]****. Note that is transitive on the arcs of . There is some element in interchanging and . This implies that . Thus we have the following lemma.
Lemma 2.4**.**
Assume that is transitive on , and . Then is a -group, and is isomorphic to a normal subgroup of a point-stabilizer in . In particular, is soluble if and only if is soluble.
The graph is said to be -arc-transitive if has an -arc and acts transitively on the set of -arcs of , where . Note that is -arc-transitive if and only if is transitive on , and is a -transitive group for some (and hence every) . By ****[7, 27, 28]****, we have the following result.
Theorem 2.5**.**
Assume that is -arc-transitive. Then is not -arc-transitive. Further,
- (1)
if then is not -arc-transitive. 2. (2)
if then is a nontrivial -group, , , and , where and is a power of ; in this case, is -arc-transitive if and only if .
3. The proof of Theorem 1.1
In this section, we let be a connected graph of valency , and . Assume that is -edge-primitive, that is, act primitively on . Then, by ****[9, Lemma 3.4]****, acts transitively on the arc set of . Thus, for an edge , and .
Let . Then is transitive on , and so either is transitive on or has two orbits on ; for the latter case, is transitive on . This implies that either , or and . Note that by the maximality of or the transitivity of on . We have
[TABLE]
Then the next lemma follows.
Lemma 3.1**.**
Let . If is transitive on then ; if is intransitive on then . In particular, and .
Let and be the complete bipartite graph and complete graph of valency , respectively.
Corollary 3.2**.**
Let . Then either , or and is self-normalized in , where .
Proof. Assume that . Then, by the O’Nan-Scott Theorem and ****[9**, Lemmas 6.1, 6.2 and Propersition 6.13], has no normal subgroup acting regularly on . Thus , and so is self-normalized in by Lemma 2.1.
Suppose that . Then has order , and so . This implies that , and then is a Sylow -subgroup of . By Burnside’s transfer theorem (refer ****[14**, IV.2.6]****), has normal -Hall subgroup, say . Then this is normal in and regular on , a contradiction. **
By ****[9]****, if then has type SD, CD, AS or PA on ; in particular, has a unique (of course, insoluble) minimal normal subgroup. Thus, if then is insoluble, and so is not abelian by ****[14, IV.7.4]****. If is abelian the following result says that or .
Theorem 3.3**.**
Assume that . Let .
- (1)
If has a normal Sylow subgroup then is also a Sylow subgroup of ; in particular, is not abelain. 2. (2)
If is abelian then is transitive on the arc set of . 3. (3)
If is an abelian -group then and , where is a power of some prime with a power of greater than . 4. (4)
If is an abelian group then and either and , or , and is -arc-transitive, where is a power of some prime.
*Proof. *(1) Assume that is a normal Sylow -subgroup of . Then is a characteristic subgroup of , and so as . Thus , and then by the maximality of . This gives . Choose a Sylow -subgroup of with . Then . This yields , so is a Sylow -subgroup of .
Suppose that is abelian. Then , yielding . By Burnside’s transfer theorem, has a normal complement in , that is with and . Note that is a Hall subgroup of . It follows that is characteristic in , and hence . Let runs over the Sylow subgroup of . Then the resulting normal complements intersect at a normal complement of in , which is normal in and regular on . This contradicts Corollary 3.2. Therefore, is nonabelian, and (1) of this theorem follows.
(2) Assume that is abelian. Then by (1), and thus for some . Since is -edge-transitive, is -arc-transitive.
(3) Assume that is an abelian -group. Recall that has a unique minimal normal subgroup, say . Then , and (1) and (2) hold for . Then, since is an abelian -group, is a Sylow -subgroup of , and is not abelian.
Write , where are isomorphic nonabelian simple groups. Recall that is a Sylow -subgroup of . For each , choose a Sylow -subgroup of with . Then . Noting that are all isomorphic, every is nonabelian; otherwise, is abelian, a contradiction. In particular, . Then , and so
[TABLE]
Since is abelian, the only possibility is . Thus is simple.
By ****[10, Corollary 5]****, has cyclic commutator subgroup. Since is nonabelian, by ****[2]****, is isomorphic to one of the Mathieu group , (with divisible by ), (with odd) and (with odd). If , then , and so is maximal in ; however, by the Atlas ****[3]****, a Sylow -subgroup of is not a maximal subgroup, a contradiction. Thus we next let , or .
Since is transitive on , we know that is odd. Thus is an almost simple primitive group (on ) of odd degree. Noting that , by ****[20]****, is known. Notice that the isomorphisms among simple groups (refer to ****[15, Proposition 2.9.1 and Theorem 5.1.1]****). Since is a Sylow -subgroup of , the only possibility is that , and is the stabilizer of some orthogonal decomposition of a natural projective module associated with into -dimensional subspaces. It follows that or , and so or , respectively. Since is transitive on the arcs of , we have . Checking the subgroups of (refer to ****[14, II.8.27]****), we conclude that , , and is -transitive on . Thus .
(4) Assume that is abelian. Let be the unique minimal normal subgroup of . If is a -group, then (4) of this theorem follows from (3).
We next assume that has an odd prime divisor . By (1), the unique Sylow -subgroup of is also a Sylow -subgroup of . Write , where are isomorphic nonabelian simple groups. By (1) of this theorem, is not abelian, so , and then . Thus , and hence acts transitively on by conjugation. Choose, for each , a Sylow -subgroup of such that is the unique Sylow subgroup . Since is abelian, we have for . It follows that for all . The only possibility is that , and so is simple.
Note that is an almost simple group with a soluble maximal subgroup . Then, by ****[18]****, both and are known. Since has an abelian subgroup of index , it follows that either and , or and . Check the subgroups of , refer to ****[25]**** for . The former case yields that and . Assume that and . Then and ; in this case, is -arc-transitive. By ****[5**]****, we have that and is unique up to isomorphism. Thus (4) of this theorem follows. **
Lemma 3.4**.**
Assume that has type PA on . Let . Then for each and ; in particular, every is not semiregular.
*Proof. ***Let . By Lemma 2.3, , and all have equal order. By Theorem 3.3, is nonabelian. Thus is nonabelian for all . Then the lemma follows. **
For the case where is a bipartite graph, we let be the subgroup of preserving the bipartition of . Then , and each bipartite half of is a -orbit on .
Lemma 3.5**.**
Assume that the graph is -arc-transitive, and has type PA on . Then either , or one of the following holds:
- (1)
* is quasiprimitive on ;* 2. (2)
* is bipartite, and is faithful and quasiprimitive on each bipartite half of .*
*Proof. *Since is primitive on , every minimal normal subgroup of is transitive on , and so has at most two orbits on . If is not bipartite then quasiprimitive on .
Now let be bipartite with bipartition say . Note that for each . Then is locally-primitive on . Suppose that . Then, by ****[23]****, is faithful on both and , and either (2) of this lemma holds, or the unique minimal normal subgroup of is a direct product , where and are normal in and conjugate in , and is intransitive on for . For the latter case, if is intransitive on then is semiregular on by ****[8**, Lemma 5.1]****; if is transitive on then is semiregular on . These two cases all contradict Lemma 3.4. Thus is quasiprimitive on both and . **
As permutation groups on and on , the types of (and ) have been determined in ****[9]****. Then by Lemma 3.5 and combining with the reduction theorems for 2-arc-transitive graphs given by Preager ****[22, 23]****, we get the following result.
Lemma 3.6**.**
Assume that the graph is -arc-transitive. Suppose that . If is not almost simple, then has type PA on and either
- (1)
* is quasiprimitive and of type PA on ; or* 2. (2)
* is bipartite, is faithful and quasiprimitive on each bipartite half of with type PA.*
Now we are ready to give a proof of Theorem 1.1.
Theorem 3.7**.**
Let be a connected -regular graph for some , and let . Assume that is both -edge-primitive and -arc-transitive. Then either , or is almost simple.
*Proof. *Assume that , and let . By the 2-arc-transitivity of on , we know that is a -transitive permutation group of degree .
Let , where are isomorphic nonabelian simple groups. Then , and by Lemma 3.1 or 3.4. Thus is a transitive normal subgroup of .
Assume that is primitive on . Noting that is transitive on , we conclude that is primitive for every . Thus is -locally primitive. Then, by Lemma 3.4 and ****[8, Lemma 5.1]****, we conclude that , and so is almost simple.
Next assume that is imprimitive on .
Note that every non-trivial normal subgroup of an almost simple -transitive group is primitive. Then is an affine -transitive group, and by Lemma 2.2, is a soluble Frobenius group and is cyclic. Set and for a prime and integer with . Then is a divisor of , and .
Assume that . Then , and so is regular on . By ****[17, Lemma 2.3]****, is faithful and hence regular on , and thus , which contradicts Corollary 3.2. Thus .
Note that is a proper divisor of . Neither nor is a prime, in particular, and . Thus has no normal subgroup isomorphic to a projective special linear group of dimension . By Theorem 2.5, , and so .
Let . Then , this implies that and are permutation isomorphic. In particular, . Since , we know that is isomorphic to a subgroup of . Note that and . Then is isomorphic to a subgroup of . In particular, is abelian. Then, by Theorem 3.3, is transitive on the arcs of , and so .
If is a power of then, by Theorem 3.3, , ; however, in this case, is locally primitive on , a contradiction. Thus has odd prime divisors. Let be an odd prime divisor of , and be a Sylow -subgroup of . Then, noting that , we know that is also a Sylow -subgroup of by Theorem 3.3. Thus , where is a Sylow -subgroup of for . Since is isomorphic to a subgroup of , we know that has no subgroup isomorphic to . It follows that .
Now we deduce a contradiction by supposing that .
Let . Since , we have
[TABLE]
It follows that is soluble, and so is soluble as is soluble. Thus is soluble, and is also soluble. Checking the soluble affine -transitive groups, by Lemma 2.2, or . Note that is a reducible subgroup of . Recalling that is not a power of , the latter case does not occur.
Since , we have , and so . Then , and . Recalling that , it follows that acts transitively on by conjugation. Let be the kernel of this action. Then , and each is normalized by . For ,
[TABLE]
This implies that normalizes each . Then is normalized by . Note that is a normal subgroup of , and is a proper divisor of . Let be the Sylow -subgroup of . Then is normalized by , of course, and .
Recalling that , we have . Since is -transitive, is divisible by , and so is divisible by . Note that . Then is not a divisor of for all . Then, by ****[13, Lemma 5.1]****, is irreducible on . It implies that , and thus for . Let run over . It follows that , and hence , . Since is transitive on , by ****[17**, Lemma 2.3]****, we have , which contradicts Lemma 3.4. This completes the proof. **
As consequence of Theorems 3.3 and 3.7, an edge-primitive graph of prime valency is -arc-transitive, and then it has almost simple automorphism group if it is not a complete bipartite graph. See also ****[21]****.
Corollary 3.8**.**
Assume that is a prime and . Then is almost simple, and either with and or is transitive on the -arcs of .
*Proof. *Note that is transitive on the arc set of . Let . By Theorem 3.7, it suffices to deal with the case where is not -transitive.
Suppose that is not -transitive. Then with and a divisor of . If then by ****[17**, Lemma 2.3]****, and so , which contradicts Corollary 3.2. Then , and so . By Theorem 2.5, . Then is isomorphic to a subgroup of . Thus is abelian. By Theorem 3.3, , , and . If then is transitive on the -arcs of , which is not the case. Thus , and so by the maximality of . **
4. Examples
Let be a connected -regular graph, where . Let and . Assume that is -arc-transitive. Choose an integer such that is -arc-transitive but not -arc-transitive; in this case, we call a -transitive graph. Then by ****[28]****. If is faithful on then by Theorem 2.5, and yields that and or , see ****[16, Proposition 2.6]****. This leads to the following interesting problem: Do there exist -arc-transitive graphs with faithful stabilizers? We next answer this problem by giving several examples of edge-primitive graphs which are -transitive and have faithful stabilizers.
The first example is the Hoffman-Singleton graph, which has valency , order and automorphism group , where . Let or . For an edge of this graph, or and or , which are maximal subgroups of . Thus the Hoffman-Singleton graph is both -edge-primitive and -arc-transitive. To see the -arc-transitivity, we fix an edge and consider the action of the arc-stabilizer (** or ) on . By the -arc-transitivity of , we have two faithful transitive actions of on and , respectively. Let and . Then , and**
[TABLE]
By the choice of , we know that and are not conjugate in , and so do for and . This implies that the actions of on and are not equivalent. Thus acts on without fixed-points, this yields that is transitive on . It follows that the Hoffman-Singleton graph is -arc-transitive.
In general, combining with ****[16, Proposition 2.6]****, a similar argument as above yields the following result.
Lemma 4.1**.**
Let be a connected -regular graph for , and . If is -arc-transitive and is faithful on , then is -arc-transitive if and only if , and , i.e. , or .
We next give another example.
Example 4.2**.**
By the information given in the Atlas [3] for the O’Nan simple group , there exactly two conjugacy classes and of (maximal) subgroups isomorphic to , which are merged into one class in . Further, there are and involutions such that all are maximal subgroups of with and . Define two bipartite graphs and with vertex set and edge sets
- ;
- .
Then and are both -edge-primitive and -arc-transitive, which have valency and respectively. By Lemma 4.1, only is -arc-transitive.* *
Lemma 4.3**.**
Let be as in Example 4.2. Then .
Proof. Let . Then . By Theorem 1.1, is almost simple, and so . Let be an edge of . Then or , and by Theorem 2.5. Thus, by the group extensions () in Section 2, we conclude that has no prime divisor other than , , and . Since is transitive on the vertices of , we have . It follows that and have the same prime divisors. Using ****[19**, Corollary 5]**, we get , and so . **
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] P.J. Cameron, Permutation Groups , Cambridge University Press, Cambridge, 1999.
- 2[2] P. Chabot, Groups whose Sylow 2-groups have cyclic commutator groups. III, J. Algebra 29 (1974), 455-458.
- 3[3] J.H. Conway, R.T. Curtis, S.P. Noton, R.A. Parker and R.A. Wilson, Atlas of Finite Groups , Clarendon Press, Oxford, 1985.
- 4[4] D.J Dixon and B. Mortimer, Permutation Groups , Springer-Verlag, New York, 1996.
- 5[5] X.G. Fang and C.E. Praeger, Finite two-arc-transitive graphs admitting a Suzuki simple group, Comm. Algebra 27(8)(1999), 3727-3754.
- 6[6] A. Gardiner, Arc transitivity in graphs, Quart. J. Math. Oxford 24 (1973), 399-407.
- 7[7] A. Gardiner, Doubly primitive vertex stabilizers in graphs, Math. Z. 135 (1974), 157-166.
- 8[8] M. Giudici, C. H. Li and C. E. Praeger, Analysing finite locally s 𝑠 s -arc transitive graphs, Trans. Amer. Math. Soc. 365 (2004), 291-317.
