In this paper, we classify cubic semisymmetric bi-Cayley graphs on nonabelian simple groups, extending a remarkable classification of cubic nonnormal symmetric Cayley graphs on nonabelian simple groups.
Citation: Jiangmin Pan, Yingnan Zhang. On cubic semisymmetric bi-Cayley graphs on nonabelian simple groups[J]. AIMS Mathematics, 2022, 7(7): 12689-12701. doi: 10.3934/math.2022702
[1] | He Song, Longmin Wang, Kainan Xiang, Qingpei Zang . The continuity of biased random walk's spectral radius on free product graphs. AIMS Mathematics, 2024, 9(7): 19529-19545. doi: 10.3934/math.2024952 |
[2] | Tao Cheng, Junchao Mao . A new class of directed strongly regular Cayley graphs over dicyclic groups. AIMS Mathematics, 2024, 9(9): 24184-24192. doi: 10.3934/math.20241176 |
[3] | Nuttawoot Nupo, Sayan Panma . Certain structural properties for Cayley regularity graphs of semigroups and their theoretical applications. AIMS Mathematics, 2023, 8(7): 16228-16239. doi: 10.3934/math.2023830 |
[4] | Yongli Zhang, Jiaxin Shen . Flag-transitive non-symmetric 2-designs with $ \lambda $ prime and exceptional groups of Lie type. AIMS Mathematics, 2024, 9(9): 25636-25645. doi: 10.3934/math.20241252 |
[5] | Wen Teng, Jiulin Jin, Yu Zhang . Cohomology of nonabelian embedding tensors on Hom-Lie algebras. AIMS Mathematics, 2023, 8(9): 21176-21190. doi: 10.3934/math.20231079 |
[6] | Tariq A. Alraqad, Hicham Saber . On the structure of finite groups associated to regular non-centralizer graphs. AIMS Mathematics, 2023, 8(12): 30981-30991. doi: 10.3934/math.20231585 |
[7] | Syafrizal Sy, Rinovia Simanjuntak, Tamaro Nadeak, Kiki Ariyanti Sugeng, Tulus Tulus . Distance antimagic labeling of circulant graphs. AIMS Mathematics, 2024, 9(8): 21177-21188. doi: 10.3934/math.20241028 |
[8] | Yang Yan, Xingguo Zhang, Rize Jin, Limin Zhou . A construction of strongly regular Cayley graphs and their applications to codebooks. AIMS Mathematics, 2024, 9(2): 2672-2683. doi: 10.3934/math.2024132 |
[9] | Chao Qin, Yu Li, Zhongbi Wang, Guiyun Chen . Recognition of the symplectic simple group $ PSp_4(p) $ by the order and degree prime-power graph. AIMS Mathematics, 2024, 9(2): 2808-2823. doi: 10.3934/math.2024139 |
[10] | Fida Moh'd, Mamoon Ahmed . Simple-intersection graphs of rings. AIMS Mathematics, 2023, 8(1): 1040-1054. doi: 10.3934/math.2023051 |
In this paper, we classify cubic semisymmetric bi-Cayley graphs on nonabelian simple groups, extending a remarkable classification of cubic nonnormal symmetric Cayley graphs on nonabelian simple groups.
Graphs considered in this paper are finite, simple and undirected. For a graph Γ, we denote by VΓ, EΓ and AΓ the vertex set, the edge set and the arc set of Γ respectively, and by AutΓ the full automorphism group of Γ. If some G≤AutΓ is transitive on VΓ, EΓ or AΓ, then Γ is called G-vertex-transitive, G-edge-transitive or G-arc-transitive, respectively. An arc-transitive graph is also called symmetric. If G is transitive on EΓ but intransitive on VΓ, then Γ is called G-semisymmetric, and Γ is simply called semisymmetric while Γ is AutΓ-semisymmetric. It is easily known that connected G-semisymmetric graphs are bipartite.
Suppose G is a group acting on a set Ω. The stabilizer of G on a point α∈Ω is the subgroup Gα={g∈G∣αg=α}. Then G is said to be semiregular on Ω if Gα=1 for each point α in Ω, and regular on Ω if G is semiregular and transitive on Ω. A graph Γ is said to be a Cayley graph on a group G if G≤AutΓ is regular on VΓ, and a bi-Cayley graph on a group G if G≤AutΓ is semiregular and has exactly two orbits (with the same length) on VΓ. In particular, Γ is called normal on G if G is normal in AutΓ, and nonnormal on G otherwise. Moreover, follow the notion in [5], a semisymmetric bi-Cayley graph Γ on a group G is called almost-normal if the normalizer NAutΓ(G) is semisymmetric on Γ.
Cayley graphs on nonabelian simple groups have received much attention in the literature, see [8,10,11,16,17,18,19,20,25,26,27,28,29] and references therein. Here a remarkable contribution is the complete classification of connected cubic nonnormal symmetric Cayley graphs on nonabelian simple groups, which was first studied by Li [15] in 1996 and completed by Xu et al. [25,26] in 2007. It finally turns out that such graphs are only two (A48,5)-arc-transitive Cayley graphs on A47. Bi-Cayley graphs can be viewed as natural generalizations of Cayley graphs, and involve many interesting examples that are not Cayley graphs such as the Petersen graph, the Gray graph [2] (which is the smallest cubic semisymmetric graph and a bi-Cayley graph on the extraspecial metacyclic group Z9:Z3 of order 27) and Bouwer graph [3]. Thus the above remarkable classification naturally motivates us to classify connected cubic (normal and nonnormal) semisymmetric bi-Cayley graphs on nonabelian simple groups.
Theorem 1.1. Let Γ be a connected cubic semisymmetric bi-Cayley graph on a nonabelian simple group T.Then either
(1) Γ is normal on T, and AutΓ=T:Z3 or T:S3; or
(2) Γ is nonnormal on T, and AutΓ=(R×S).O, where S>T, (S,T)=(M24,M23)or (An,An−1) with n∈{12,24,48,96,192,384}, |R||O| divides 384/|S:T|and O≤Out(S).
Applying Theorem 1.1, the connected cubic nonnormal semisymmetric bi-Cayley graphs on nonabelian simple groups may be determined by computation with Magma package [1], see Example 3.5 for a specific example with T=M23.
Normal semisymmetric bi-Cayley graphs are definitely almost-normal, but the converse is not necessarily true. In fact, determining the normality of Cayley or bi-Cayley graphs is difficult in general for finding the full automorphism groups. However, Theorem 1.1 has the following interesting consequence which shows that, for connected cubic semisymmetric bi-Cayley graphs on nonabelian simple groups, the normality and the almost-normality are equivalent.
Corollary 1.2. A connected cubic semisymmetric bi-Cayley graph on a nonabelian simple group Tis normal on T if and only if it is almost-normal on T.
The importance of Corollary 1.2 is that it may lead to an explicit characterization of normal semisymmetric bi-Cayley graphs on nonabelian simple groups. To express the result, we first introduce a general construction of bi-Cayley graphs. Let G be a group, R and L be inverse-closed subsets (may empty) of G∖{1}, and let S be a nonempty subset of G. Define a graph Γ=BiCay(G,R,L,S) with vertex set
{g0∣g∈G}∪{g1∣g∈G}, |
and edge set
{{g0,h0}∣hg−1∈R}∪{{g1,h1}∣hg−1∈L}∪{{g0,h1}∣hg−1∈S}. |
It is easily known that Γ is a bi-Cayley graph on a group isomorphic to G, and each bi-Cayley graph can be constructed in this way (refer to [29]). Also, one may assume 1∈S up to isomorphism.
Theorem 1.3. A graph Γ is a connected cubic normal semisymmetric bi-Cayley graph on a nonabelian simple group Tif and only if that
Γ≅BiCay(T,∅,∅,{1,s,ssg}) |
for some s∈T and g∈Aut(T) satisfying
o(g)=3,ssgsg2=1,⟨s,sg⟩=T, |
and there exists no element h∈Aut(T) satisfying{s,ssg}h={s−1,(ssg)−1}.
Theorem 1.3 helps us finding many specific examples or proving the nonexistence of examples for certain nonabelian simple groups via searching in Magma [1]. For example, computation shows that there is no example for T=A5 and A6, but exist examples for T=A7 and A8, see Examples 4.1 and 4.2.
We remark that a result of Tutte [24,P.59] states that an edge-transitive graph of odd valency is either symmetric or semisymmetric, thus this paper together [9] and [21] completes the characterization of cubic edge-transitive bi-Cayley graphs on nonabelian simple groups.
After making some preparations in Section 2, we prove Theorem 1.1 and Corollary 1.2 in Section 3, and Theorem 1.3 in Section 4.
Bi-coset graph is an important tool for understanding semi-symmetric graphs. Let G be a group and L,R be subgroups of G such that L∩R is core-free in G (namely L∩R does not contain any nontrivial normal subgroup of G). Define a bi-coset graph with vertex set [G:L]∪[G:R], and Lx is adjacent to Ry with x,y∈G if and only if yx−1∈RL. This bi-coset graph is denoted by Cos(G,L,R).
The following lemma is known, refer to [13,Section 3].
Lemma 2.1. Using notation as above, then Cos(G,L,R) is G-semisymmetric.Conversely, each G-semisymmetric graph is isomorphic toCos(G,Gα,Gβ), where α and β are adjacent vertices.
In particular, Cos(G,Gα,Gβ) is connected cubic G-semisymmetric if and only if G,Gα, Gβ satisfy the following conditions: ∙⟨Gα,Gβ⟩=G,
∙|Gα:Gα∩Gβ|=|Gβ:Gα∩Gβ|=3, and
∙Gα∩Gβ is core-free in G.
In 1980, a landscape work of Goldschmidt [14] determined all the triples (Gα,Gβ,Gαβ) satisfying the above conditions (refer to [22,Table 3]), which are called Goldschmidt amalgams after him.
Theorem 2.2 ([14]). Let Γ be a connected cubic G-semisymmetric graph.Then there are exactly fifteenpossible amalgams (Gα,Gβ,Gαβ) with α,β adjacent vertices.In particular, |Gα|=|Gβ|=3⋅2i where 0≤i≤7.
Investigating normal quotients of graphs has been very successful in studying various families of graphs. Let Γ be a G-edge-transitive graph and let N be an intransitive normal subgroup of G. The normal quotient ΓN of Γ with respect to N is defined to be the graph with the set of N-orbits as its vertex set and two N-orbits B1,B2 are adjacent if and only if some vertex in B1 is adjacent in Γ to some vertex in B2. The original graph is said to be a normal cover of ΓN if |Γ(α)∩B2|=1 for each edge {B1,B2} in ΓN and α∈B1.
Lemma 2.3. Let Γ be a connected cubic G-semisymmetric graph with bipartitions Δ1 and Δ2.Let N⊲G and α∈VΓ. Then either
(1) N acts transitively on at least one of Δ1 and Δ2; or
(2) N is semiregular and intransitive on both Δ1 and Δ2, and the following statements are true:
(i) Γ is a normal cove of ΓN.
(ii) GVΓN=G/N, (G/N)αN=GαN/N andΓN is a connected cubic G/N-semisymmetric graph.
Proof. We only need to prove that (G/N)αN=GαN/N because the other statements are known, see [13,Lemma 5.1] or [19,Lemma 2.3]. For Ng∈G/N, one has the following equivalences:
Ng∈(G/N)αN⟺(αg)N=(aN)Ng=αN⟺αg=αnforsomen∈N⟺g∈NGα⟺Ng∈NGα/N. |
Therefore (G/N)αN=GαN/N. In particular, if Nα=1, then (G/N)αN≅Gα/(N∩Gα)=Gα/Nα=Gα.
For a bi-Cayley graph BiCay(T,∅,∅,S) and τ∈Aut(T), with bipartitions T0:={t0∣t∈T} and T1:={t1∣t∈T}. Define
στ,s:t0→(tτ)0,t1→(stτ)1,foreacht∈T. |
From [29,Theorem 1.1], we have the following assertion which is important for characterizing normal bi-Cayley graphs.
Lemma 2.4. Let Γ=BiCay(T,∅,∅,S) be a normal bi-Cayley graph on a group T.Then
A10=⟨στ,s∣τ∈Aut(T),s∈S,Sτ=s−1S⟩. |
The next lemma is similar to [13,Lemma 6.2], we include a proof here for the convenience of the readers.
Lemma 2.5. Let Γ be a connected cubic G-semisymmetric graph with bipartitions Δ1 and Δ2, where G is faithful on both Δ1 and Δ2.Suppose further that K⊲G is transitive on Δ1 and Δ2, and Kα≠1 for a vertex α of Γ.Then KΓ(v)v≥Z3 is transitive for each vertex v of Γ.
Proof. Assume α∈Δi with i∈{1,2}. Since G is faithful on Δ3−i, there exists β∈Δ3−i such that Kα does not fix β. By the connectivity of Γ, there exists a path (α0,α1,⋯,αt) with α0=α such that Kα fixes α0,…,αt−1 but moves αt. It follows that
1≠KΓ(αt−1)αt−1⊲GΓ(αt−1)αt−1. |
Since Γ is a G-semisymmetric cubic graph, GΓ(αt−1)αt−1 is primitive, hence KΓ(αt−1)αt−1 is transitive.
Without loss of generality, we may assume αt−1∈Δ1 and αt∈Δ2. Let {β1,β2} be an edge of Γ with β1∈Δ1 and β2∈Δ2. Since K is transitive on Δ1, there is x∈K such that βx1=αt−1, hence {β1,β2}x={αt−1,βx2} is an edge of Γ. Now both αt and βx2 are in Γ(αt−1), the transitivity of KΓ(αt−1)αt−1 implies that there is y∈Kαt−1 such that (βx2)y=αt. It follows that y−1x−1∈K such
{αt−1,αt}y−1x−1={β1,β2}, |
namely Γ is K-edge-transitive. Thereby KΓ(v)v≥Z3 is transitive for each vertex v∈VΓ.
Let G=N.H be a group extension. If N is contained in Z(G), the center of G, then the extension is called a central extension, and if further G is perfect, that is, the commutator subgroup G′ equals to G, we call G a covering group of H. Schur [23] proved that every nonabelian simple group T admits a unique maximal covering group M such that each covering group of T is a homomorphic image of M; the center Z(M) is called the Schur multiplier of T, denoted by Mult(T). The next result states that each central extension N.T with T nonabelian simple can derive a covering group of T, refer to [20,Lemma 2.11].
Lemma 2.6. Let G=N.T be a central extension with T a nonabelian simple group.Then G=NG′ and G′=Z(G′).T is a covering group of T, where Z(G′)=N∩G′ is a quotient of Mult(T).
The next lemma can be read out from [11,Lemma 3.3].
Lemma 2.7. Let T<S be nonabelian simple groups.If |S:T|∣384, then either
(1) (S,T)=(An,An−1) with n≥6 dividing 384; or
(2) (S,T)=(M11,PSL2(11)),(M12,M11) or (M24,M23).
With similar arguments as in [8,P. 143], the following lemma is easy to prove by checking the nonabelian simple groups contained in GL(e,2) for 3≤e≤7.
Lemma 2.8. Let L=R:T be a split extension, where |R|∣384 and T is a nonabelian simple group.Then either
(a) L=R×T; or
(b) L≠R×T, T≤GL(e,2) with 2e∣|R|, and e and T satisfy one of the following:
(b.1) e=3 and T=PSL3(2).
(b.2) e=4 and T∈{PSL3(2),PSL4(2),A5,A6,A7}.
(b.3) e=5 and T∈{PSL3(2),PSL4(2),PSL5(2),A5,A6,A7}.
(b.4) e=6 and T∈{PSLk(2)(3≤k≤6),PSL2(8),A5,A6,A7,PSU3(3),PSU4(2), PSp6(2)}.
(b.5) e=7 and T∈{PSLk(2)(3≤k≤7),PSL2(8),A5,A6,A7,PSU3(3),PSU4(2), PSp6(2)}.
The solvable radical of a group is its largest solvable normal subgroup.
Lemma 2.9. Let T be a nonabelian simple group, and let Γ be a connected cubic semisymmetric bi-Cayley graph on T.Suppose R is the solvable radical of AutΓ and 3∣|R|.Then RT=R×T.
Proof. Write L=RT, A=AutΓ and let α∈VΓ. As R∩T is solvable and normal in T, we have R∩T=1 and L=R:T is a split extension. Since Γ is a semisymmetric bi-Cayley graph on T, we obtain |L:Lα|=|VΓ|/2=|T|, so |Lα|=|R| is divisible by 3. By Theorem 2.2, |R|=|Lα| divides 3⋅27=384, and the kernel of Aα acting on Γ(α) is a 2-group, we deduce 3||LΓ(α)α|, and Γ is L-semisymmetric. If 23∤|R|, Lemma 2.8 leads to L=R×T, as required. Thus assume in the following that 23∣|R|. Then 24∣|R| as 3∣|R|.
Let R2 be a Sylow 2-subgroup of R and let S={Rr2∣r∈R}, the set of the conjugate subgroups of R2 in R. Since |R| divides 384, |S|=|R:NR(R2)| divides 3, then the nonabelian simplicity of T implies that T acts trivially on S by conjugation, that is, T normalizes R2. Thereby P:=R2T=R2:T≤L, |P:Pα|=|T| and |L:P|=3. Let K be the kernel of L acting on [L:P] by right multiplication. Then K is the largest normal subgroup of L contained in P ([7,Example 1.3.4]) and Z3≤L/K≤S3, implying K≥T. Since |P|=|R2||T|=|T||Pα|, Kα≤Pα is a 2-groups.
If A acts unfaithfully on the bipartitions of Γ, by [13,Lemma 5.2], Γ=K3,3 is arc-transitive, a contradiction. Hence A acts faithfully on the bipartitions of Γ. If Kα≠1, then Lemma 2.5 implies that KΓ(α)α≥Z3, contradicting that Kα is a 2-group. Thus Kα=1. Consequently, T=K⊲L, and so L=R×T.
Denote by soc(G) the socle of a group G, that is, the product of all minimal normal subgroups of G.
Lemma 2.10. Let Γ be a connected cubic G-semisymmetric graph, and let G have a nonabelian simple subgroup Twhich has two orbits of the same length on VΓ.Suppose further G has a trivial solvable radical.Then G is almost simple, and either
(a) soc(G)=T; or
(b) soc(G)>T, soc(G)α≥Z3 for α∈VΓ, and Γ is soc(G)-semisymmetric. Further, one of the following is true:
(b.1) (soc(G),T)=(An,An−1) with n=6 or n≥12 dividing 384.
(b.2) (soc(G),T)=(M24,M23).
Proof. By Theorem 2.2, |Gα|∣384. Since |G:Gα|=|VΓ|/2=|T:Tα|, it follows that |G| divides 384|T|.
Let N be a minimal normal subgroup of G. Since G has a trivial solvable radical, N=S1×S2⋯×Sd=Sd, where each Si≅S is nonabelian simple and d≥1. Since T is nonabelian simple, either T∩N=1 or N≥T. For the former, |N||T| divides 384|T|, so |N| divides 384, hence N is a {2,3}-group and thus solvable, a contradiction. Therefore N≥T. As distinct minimal normal subgroups of G intersect trivially, by the arbitrariness of T, we obtain that N is the unique minimal normal subgroup of G, namely soc(G)=N. If T=N, then soc(G)=T, as in Lemma 2.10(a).
Now suppose T<N. Then |N:Nα|=|T:Tα|, hence |N|=|T||Nα:Tα| divides 384|T|. Assume d≥2. Since S1⊲N and T<N, we have S1∩T⊲T, so S1∩T=1 or T. For the former, then |S1T|=|S1||T| divides |N|, and so divides 384|T|. It follows that S1 is a {2,3}-group and solvable, a contradiction. If the latter occurs, then T≤S1 and |S|d=|N| divides 384|S|, implying |S|d−1∣384, also a contradiction.
The above contradiction shows that d=1, namely soc(G)=S>T. Since |S:T|=|Sα:Tα| divides 384, the pair (S,T) satisfies Lemma 2.7. In particular Out(S)≤Z22. Notice that Gα≥Z3 and
Gα/Sα=Gα/(S∩Gα)≅SGα/S=G/S≤Out(S)≤Z22, |
we conclude that Sα≥Z3 is transitive on Γ(α), thus Γ is S-semisymmetric. Now to complete the proof, we only need to prove (S,T)≠(A8,A7), (M11,PSL2(11)) and (M12,M11).
If (S,T)=(M11,PSL2(11)), by Frattini argument, S=TSα. Notice that |Sα|∣384, by [12,Theorem 1.1], no such factorization exists, a contradiction.
Let α,β be adjacent vertices of Γ. Then |Sα|=|Sβ|, |Tα|=|Tβ|, |Sα:Sα∩Sβ|=3, and ⟨Sα,Sβ⟩=S. Recall |Sα|=|S:T||Tα| divides 384.
Set k=|Tα|, and define
Sk={H∣H≤S,|H|=k|S:T|},T={H∣H≤S,H≅T}, |
Pk={(H1,H2)∣H1,H2∈Sk,|H1:H1∩H2|=3,⟨H1,H2⟩=S}, |
Qk={(|H1∩T|,|H2∩T|)∣T∈T,(H1,H2)∈Pk}. |
Then (Sα,Sβ)∈Pk and (|Tα|,|Tβ|)∈Qk.
Assume (S,T)=(A8,A7). By [6], A8 has no subgroup with order 384, so |Sα|≠384. Then since 3∣|Sα| and |Sα|=|S:T||Tα|=8k divides 384, we conclude k∈{3,6,12,24}. But searching in Magma [1] shows |Pk|=0 for each k∈{3,6,12,24}, a contradiction.
Now assume (S,T)=(M12,M11). Then |Sα|≠384 as M12 has no subgroup with order 384 by [6]. Since |Sα|=|S:T||Tα|=12k divides 384, we deduce k∈{1,2,4,8,16}. However, computation with Magma [1] shows |P4|=|P8|=0, so k≠4 and 8. For k=1,2,16, computation shows Q1={(1,3),(3,1)}, Q2={(2,6),(2,3),(3,2),(6,2)} and Q16={(24,16),(48,16),(16,24),(16,48)}, contradicting |Tα|=|Tβ|.
For convenience, we make the following hypothesis in this section.
Hypothesis 3.1. Let T be a nonabelian simple group, and let Γ be a connected cubic semisymmetric bi-Cayley graph on T.Denote by Δ1 and Δ2 the bipartitions of Γ.Set A=AutΓ and α∈VΓ.
Lemma 3.2. Under Hypothesis 3.1, then T≠A5.
Proof. If T=A5, then Γ is of order 120. Checking the census of connected semisymmetric cubic graphs on up to 768 vertices ([4]), there is no such graph of order 120 admitting a semiregular automorphism group A5, a contradiction.
Lemma 3.3. Under Hypothesis 3.1, and suppose that A has a trivial solvable radical, and T is not normal in A.Then part (2) of Theorem 1.1 is true with R=1.
Proof. Set S=soc(A). By Lemma 3.2, T≠A5. Then we see from Lemma 2.10 that S>T is a simple group, Sα≥Z3 and either (S,T)=(M24,M23) or (An,An−1) with n≥12 a divisor of 384. If the latter case occurs, by Frattini argument, S=TSα, then the semiregularity of T derives that n=|S:T|=|Sα| is divisible by 3. This together with that n≥12 divides 384 implies n=12,24,48,96,192 or 384, as required.
Lemma 3.4. Under Hypothesis 3.1, and suppose that A has a nontrivial solvable radical R, and T is not normal in A.Then part (2) of Theorem 1.1 is true with R≠1 and n≠384.
Proof. If A acts unfaithfully on Δ1 or Δ2, by [13,Lemma 5.2], Γ=K3,3 is arc-transitive, a contradiction. Thus A acts faithfully on Δ1 and Δ2. Write L=RT. As R∩T is solvable and normal in T, we obtain that R∩T=1 and |Lα|=|R|≠1. By Theorem 2.2, |A|=|T||Aα| divides 384|T|, so |L| divides 384|T|. Hence |R|∣384. If R is transitive on Δ1 or Δ2, then |T| divides |R|, thus T is a {2,3}-group and solvable, a contradiction. Therefore R is not transitive on both Δ1 and Δ2. It then follows from Lemma 2.3 that R is semiregular on both Δ1 and Δ2, A/R≤Aut(ΓR), and the quotient ΓR is a connected cubic A/R-semisymmetric graph. Moreover, notice that A/R has a trivial solvable radical, and L/R≅T has two orbits with equal length on VΓR, hence the triple (ΓR,G/R,L/R) (as (Γ,G,T) there) satisfies Lemma 3.3. Consequently (note T≠A5 by Lemma 3.2), A/R is almost simple, and either
(a) soc(A/R)=L/R; or
(b) soc(A/R)>L/R≅T, soc(A/R)α≥Z3 and ΓR is soc(A/R)-semisymmetric. Further, one of the following is true:
(b.1) (soc(A/R),L/R)=(An,An−1) where n≥12 divides 384.
(b.2) (soc(A/R),L/R)=(M24,M23).
Assume (a) occurs. Then L⊲A. Since A acts faithfully on Δ1 and Δ2, and |Lα|=|R|≠1, by Lemma 2.5, LΓ(α)α≥Z3, so 3∣|R|. It then follows from Lemma 2.9 that L=R×T, hence T⊲A, a contradiction.
Assume (b) occurs. If L≠R×T, as |R|∣384, T satisfies part (b) of Lemma 2.8, which violates that T≅L/R satisfies (b.1) or (b.2) above, a contradiction. Thus L=R×T. Suppose S=M/R=soc(A/R) and C=CM(R). Then T≤C⊲M and C∩R=Z(R)≤Z(C). Observe 1≠C/(C∩R)≅CR/R⊲M/R=S, we derive that M=CR, C∩R=Z(C) and C=(C∩R).S is a central extension. Now Lemma 2.6 implies that C′=(C′∩Z(C)).S is a covering group of S, and C′∩Z(C)=Z(C′) is a quotient of Mult(S). Notice that Mult(M24)=1 and Mult(An)≅Z2 for n≥12.
Suppose Z(C′)>1. Then Z(C′)=Mult(S)=Z2, C′≅Z2.An and (S,T)=(An,An−1) with n≥12 dividing 384. Since Z(C′)⊲M, M/R=S and RT=R×T, we deduce that Z(C′)≤R and C′ has a subgroup TZ(C′)=T×Z(C′)≅T×Z2. However, by [8,Proposition 2.6], the covering group C′=Z2.An with n≥8 has no subgroup isomorphic to Z2×An−1, a contradiction.
Therefore Z(C′)=1, and C′≅S is normal in M, so C′∩R=1 and M=R×C′ as M/R=S. It follows A=(R×C′).O with O≤Out(S). Since |A:Aα|=|M:Mα|=|T|, we deduce |R||O|=|Aα|/|S:T| divides 384/|S:T|. Finally, for the case (S,T)=(An,An−1) with n≥12 dividing 384, as |R|>1, n≠384. Since C′⊲A, by Lemma 2.5, C′α≥Z3, then the semiregularity of T implies that n=|S:T|=|C′:T|=|C′α| is divisible by 3. Now since n≥12 divides 384, we derive that n∈{12,24,28,96,192}, as in part (2) of Theorem 1.1.
Now we are ready to prove Theorem 1.1 and Corollary 1.2.
Proof of Theorem 1.1. Let T be a nonabelian simple group, and let Γ be a connected cubic semisymmetric bi-Cayley graph on T. Set A=AutΓ. If Γ is nonnormal on T, by Lemmas 3.3 and 3.4, part (2) of Theorem 1.1 holds.
Assume Γ is normal on T. Since Γ is semisymmetric, Γ is bipartite, and so we may suppose that Γ=BiCay(T,∅,∅,S), with bipartitions T0:={t0∣t∈T} and T1:={t1∣t∈T}, and S={1,a,b}. By Lemma 2.4,
A10=⟨στ,s∣τ∈Aut(T),s∈S,Sτ=s−1S⟩, |
which is transitive on Γ(10)={11,a1,b1} by the semisymmetry of Γ. If στ,s fixes each vertex in Γ(10), then
11=(11)στ,s=(s1τ)1=s1,a1=(a1)στ,s=(saτ)1,b1=(b1)στ,s=(sbτ)1. |
It follows s=1, aτ=a and bτ=b. Since Γ is connected, ⟨a,b⟩=⟨S⟩=T, hence τ=1 and στ,s=σ1,1 is the identity automorphism of Γ, namely A10 acts faithfully on Γ(10). Consequently Z3≤A10≤S3. Now by Frattini argument, we deduce A=T:A10=T:Z3 or T:S3, part (1) of Theorem 1.1 holds.
Proof of Corollary 1.2. We only need to prove the sufficiency.
Suppose on the contrary that T is a nonabelian simple group, and there is a connected cubic semisymmetric bi-Cayley graph Γ on T such that Γ is almost-normal but not normal on T. Set A=AutΓ and G=NA(T). Then Γ is G-semisymmetric, and with the same reason (view G as A there) as in the previous paragraph, we derive that G=T:Z3 or T:S3. Since, by assumption, T is not normal in A, we see from Theorem 1.1(2) that
A=(R×S).O, |
where S>T, (S,T)=(M24,M23) or (An,An−1) with n∈{12,24,48,96,192,384}, |R||O| divides 384/|S:T| and O≤Aut(S)≤Z2. If R=1, as T is maximal in S, we obtain G≤T.Z2, which is impossible as Γ is G-semisymmetric. Thus R≠1. Since R centralizes T, R≤NA(T)=G and so R×T⊲G. This together with T:Z3≤G≤T:S3 implies that R≥Z3. Notice that 3∣|S:T|, we derive that |Aα|=|A:T| is divisible by 32, which is a contradiction by Theorem 2.2.
We end this section with a specific example of cubic nonnormal semisymmetric bi-Cayley graph on M23.
Example 3.5. Let
G=⟨(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,24),(2,16,9,6,8)(3,12,13,18,4)(7,17,10,11,22)(14,19,21,20,15),(1,22)(2,11)(3,15)(4,17)(5,9)(6,19)(7,13)(8,20)(10,16)(12,21)(14,18)(23,24)⟩,L=⟨(1,6,17,11)(2,15,8,4)(3,23,22,5)(7,16,9,21)(10,13,18,24)(12,19,14,20),(1,2)(3,19)(4,6)(5,14)(7,24)(8,17)(9,13)(10,21)(11,15)(12,23)(16,18)(20,22),(1,17)(2,8)(3,22)(4,15)(5,23)(6,11)(7,9)(10,18)(12,14)(13,24)(16,21)(19,20),(1,10,5,17,18,23)(2,21,14,8,16,12)(3,13,6,22,24,11)(4,20,7,15,19,9)⟩,R=⟨(1,2)(2,3)(4,6)(21,10)(16,18)(8,17)(7,24)(12,23)(13,9)(20,22)(5,23)(11,15),(1,6,24)(19,11,14)(4,2,7)(21,13,20)(16,12,17)(8,23,18)(22,9,10)(5,15,3),(1,17)(19,20)(4,15)(21,16)(6,11)(8,2)(18,10)(7,9)(12,14)(13,24)(22,3)(5,23),(1,19)(4,18)(21,6)(16,11)(8,3)(7,5)(12,24)(17,20)(13,14)(22,2)(9,23)(10,15)⟩. |
Let Γ=Cos(G,L,R). Then Γ is a connected cubic M24-semisymmetric bi-Cayley graph on M23.
Proof. A direct computation with Magma [1] shows that L≅Z3:D8, R≅S4, ⟨L,R⟩=G≅M24 and |L:L∩R|=3. By Lemma 2.1, Γ is a connected cubic M24-semisymmetric graph. Notice that both L and R are transitive on the set {1,2,...,24} and |L|=|R|=24, they are regular on {1,2,…,24}. Let T=G1, the stabilizer of G on the point 1. Then T≅M23, G=TL=TR and |G|=|T||L|=|T||R|. It follows that T is regular on the bipartitions of Γ, namely Γ is a bi-Cayley graph on M23. Further, by [9,Lemma 7.1], Γ is not arc-transitive, hence Γ is semisymmetric. Finally as |G|=24|M23|, by Theorem 1.1, Γ is nonnormal on M23.
For a nonabelian simple group T, we may view T as a normal subgroup of Aut(T) up to isomorphism, and then each element in Aut(T) may acts on T by conjugate action.
Proof of Theorem 1.3. (Necessity) By assumption, we may assume Γ=BiCay(T,∅,∅,S) with bipartitions T0:={t0∣t∈T} and T1:={t1∣t∈T}, and 1∈S. Then S1=Γ(10) and |S|=3. By [29,Theorem 1.1],
A10=⟨σθ,s∣θ∈Aut(T),s∈S,Sθ=s−1S⟩ |
acts faithfully on Γ(10)=S1, and A10 contains an element σg,s with order 3 which cyclically permutes all the elements of S1, where g∈Aut(T) and s∈S. It follows S1=1⟨σg,s⟩1={11,s1,(ssg)1} and S={1,s,ssg}. The connectivity of Γ forces ⟨s,sg⟩=⟨S⟩=T. Since o(σg,s)=3, for each t∈T, t0=(t0)σ3g,s=(tg3)0 and t1=(t1)σ3g,s=(ssgsg2tg3)1, or equivalently g3=1 and ssgsg2=1. If g=1, then ⟨S⟩=⟨1,s,s2⟩=⟨s⟩≠T, contradicting the connectivity of Γ. Hence o(g)=3. Moreover, since Γ is not vertex-transitive, by [5,Proposition 3.3(a)], there is no h∈Aut(T) such that Sh=S−1, namely {s,ssg}h={s−1,(ssg)−1}.
(Sufficiency) Suppose Γ=BiCay(T,∅,∅,{1,s,ssg}), where s∈T and g∈Aut(T) satisfy the condition in Theorem 1.3. Clearly Γ is a cubic bi-Cayley graph on T. Set P={1,s,ssg} and G=⟨T,σg,s⟩. By [29,Theorem 1.1], G=T:⟨σg,s⟩≤NAutΓ(T). Since ⟨P⟩=T, Γ is connected. Since P=(11)⟨σg,s⟩, Γ is G-semisymmetric.
We claim that Γ is semisymmetric (note that a G-semisymmetric graph is not necessarily semisymmetric in general). If not, as Γ is of valency 3, Γ is symmetric and G<AutΓ. If T⊲AutΓ, by [5,Proposition 3.3(a)], there is an element h∈Aut(T) such that {1,s,ssg}h=Sh=S−1={1,s−1,(ssg)−1}, which is not possible by the assumption. If T is not normal in AutΓ, by [21,Theorem 1.1(2)], we would have that T=An−1 and AutΓ=(R×An).O, where n∈{24,48,96} and |R||O|∣96n. Notice that An−1 is maximal in An and 96n∣4, we deduce that |NAutΓ(T):T| divides 4, which is also not possible as T:Z3≤G≤NAutΓ(T).
Therefore, Γ is semisymmetric. Hence Γ is almost-normal as G≤NAutΓ(T). Now Corollary 1.2 implies that Γ is normal on T.
We finally address two examples of normal cubic semisymmetric bi-Cayley graphs on A7 and A8, respectively.
Example 4.1. Let s=(1,6,2,5,4,3,7), g=(1,5,4) and Γ=BiCay(A7,∅,∅,{1,s,ssg}). Then Γ is a connected cubic semisymmetric normal bi-Cayley graph on A7 and AutΓ=(Z3×A7):Z2.
Proof. Obviously val(Γ)=|{1,s,ssg}|=3. With Magma [1], one may check that ⟨s,sg⟩=A7, ssgsg2=1, AutΓ=(Z3×A7):Z2 and there is no element h∈S7 such that {s,ssg}h={s−1,(ssg)−1}. By Theorem 1.3, Γ is a connected cubic semisymmetric normal bi-Cayley graph on A7.
Example 4.2. Let s=(1,6,8)(2,7,3,5,4), g=(1,7,8)(4,6,5) and Γ=BiCay(A8,∅,∅, {1,s,ssg}). Then Γ is a connected cubic semisymmetric normal bi-Cayley graph on A8 and AutΓ=(Z3×A8):Z2.
Proof. Clearly val(Γ)=|{1,s,ssg}|=3. A computation with Magma [1] shows that ⟨s,sg⟩=A8, ssgsg2=1, AutΓ=(Z3×A8):Z2 and there is no element h∈S8 such that {s,ssg}h={s−1,(ssg)−1}. By Theorem 1.3, Γ is a connected cubic semisymmetric normal bi-Cayley graph on A8.
This paper classifies cubic semisymmetric bi-Cayley graphs on nonabelian simple groups, which extends a remarkable classification of cubic nonnormal symmetric Cayley graphs on nonabelian simple groups, and together with former known results, completes the characterization of cubic edge-transitive bi-Cayley graphs on nonabelian simple groups.
The authors are very grateful to the referees for the helpful comments. This paper was supported by the National Natural Science Foundation of China (11961076).
The authors declare no conflicts of interest.
[1] |
W. Bosma, J. Cannon, C. Playoust, The MAGMA algebra system I: The user language, J. Symbolic Comput., 24 (1997), 235–265. https://doi.org/10.1006/jsco.1996.0125 doi: 10.1006/jsco.1996.0125
![]() |
[2] |
I. Z. Bouwer, An edge but not vertex transitive cubic graph, Bull. Can. Math. Soc., 11 (1968), 533–535. https://doi.org/10.4153/CMB-1968-063-0 doi: 10.4153/CMB-1968-063-0
![]() |
[3] |
I. Z. Bouwer, Vertex and edge transitive but not 1-transitive graph, Bull. Can. Math. Soc., 13 (1970), 231–237. https://doi.org/10.4153/CMB-1970-047-8 doi: 10.4153/CMB-1970-047-8
![]() |
[4] |
M. Conder, A. Malnič, D. Marušič, P. Potočnik, A census of semisymmetric cubic graphs on up to 768 vertices, J. Algebraic Combin., 23 (2006), 255–294. https://doi.org/10.1007/s10801-006-7397-3 doi: 10.1007/s10801-006-7397-3
![]() |
[5] |
M. Conder, J.-X. Zhou, Y.-Q. Feng, M.-M. Zhang, Edge-transitive bi-Cayley graphs, J. Combin. Theory Ser. B, 145 (2020), 264–306. https://doi.org/10.1016/j.jctb.2020.05.006 doi: 10.1016/j.jctb.2020.05.006
![]() |
[6] | J. H. Conway, R. T. Curtis, S. P. Norton, R. A. Parker, R. A. Wilson, Atlas of Finite Groups, Lonon/New York: Oxford Univ. Press, 1985. |
[7] | J. D. Dixon, B. Mortimer, Permutation Groups, New York: Springer, 1996. https://doi.org/10.1007/978-1-4612-0731-3 |
[8] |
J. L. Du, Y.-Q. Feng, J.-X. Zhou, Pentavalent symmetric graphs admitting a vertex- transitive non-abelian simple groups, Europ. J. Combin., 73 (2017), 134–145. https://doi.org/10.1016/j.ejc.2017.03.007 doi: 10.1016/j.ejc.2017.03.007
![]() |
[9] |
J. L. Du, M. Conder, Y, -Q. Feng, Cubic core-free symmetric m-Cayley graphs, J. Algebraic Combin., 50 (2019), 143–163. https://doi.org/10.1007/s10801-018-0847-x doi: 10.1007/s10801-018-0847-x
![]() |
[10] | S. F. Du, M. Y. Xu, A classifiction of semisymmetric graphs of order 2pq, Comm. Algebra, 28 (2000), 2685–2715. |
[11] |
X. G. Fang, X. S. Ma, J. Wang, On locally primitive Cayley graphs of finite simple groups, J. Combin. Theory Ser. A, 118 (2011), 1039–1051. https://doi.org/10.1016/j.jcta.2010.10.008 doi: 10.1016/j.jcta.2010.10.008
![]() |
[12] |
M. Giudici, Factorisations of sporadic simple groups, J. Algebra, 304 (2006), 311–323. https://doi.org/10.1016/j.jalgebra.2006.04.019 doi: 10.1016/j.jalgebra.2006.04.019
![]() |
[13] | M. Giudici, C. H. Li, C. E. Praeger, Analysing finite locally s-arc-transitive graphs, Trans. Amer. Math. Soc., 350 (2003), 291–317. |
[14] |
D. M. Goldschmidt, Automorphisms of trivalent graphs, Ann. Math., 111 (1980), 377–406. https://doi.org/10.2307/1971203 doi: 10.2307/1971203
![]() |
[15] | C. H. Li, Isomorphisms of finite Cayley graphs (Ph.D. thesis), The University of Weastern Australia, 1996. |
[16] | B. Ling, A two-transitive pentavalent nonnormal Cayley graph on the alternating group A119 (in Chinese), Acta Sci. Natur. Univ. Sunyatseni, 57 (2018), 85–88. |
[17] |
B. Ling, A note on tetravalent s-arc-regular Cayley graphs of finite simple groups, Ars Combin., 144 (2019), 49–54. https://doi.org/10.1097/01.BMSAS.0000554724.29762.5b doi: 10.1097/01.BMSAS.0000554724.29762.5b
![]() |
[18] |
B. Ling, B. G. Lou, On arc-transitive pentavalent Cayley graphs on finite nonabelian simple groups, Graph Combin., 33 (2017), 1297–1306. https://doi.org/10.1007/s00373-017-1845-9 doi: 10.1007/s00373-017-1845-9
![]() |
[19] | G. X. Liu, Z. P. Lu, On edge-transitive cubic graphs of square-free order, Europ. J. Combin., 45 (2015), 41–46. |
[20] |
J. M. Pan, Y. Liu, Z. H. Huang, C. L. Liu, Tetravalent edge-transitive graphs of order p2q, Sci. China Math. Ser. A, 57 (2014), 293–302. https://doi.org/10.1007/s11425-013-4708-8 doi: 10.1007/s11425-013-4708-8
![]() |
[21] | J. M. Pan, Y. N. Zhang, An explicit characterization of cubic symmetric bi-Cayley graphs on nonabelian simple groups, Discrete Math., in press. |
[22] |
C. W. Parker, Semisymmetric cubic graphs of twice odd order, Europ. J. Combin., 28 (2007), 572–591. https://doi.org/10.1016/j.ejc.2005.06.007 doi: 10.1016/j.ejc.2005.06.007
![]() |
[23] |
I. Schur, Untersuchen über die Darstellung der endlichen Gruppen durch gebrochenen linearen Substitutionen, J. Reine Angew. Math., 132 (1907), 85–137. https://doi.org/10.1515/crll.1907.132.85 doi: 10.1515/crll.1907.132.85
![]() |
[24] | W. T. Tutte, Connectivity in graphs, Toronto: Toronto Univ. Press, 1966. https://doi.org/10.3138/9781487584863 |
[25] |
S. J. Xu, X. G. Fang, J. Wang, M. Y. Xu, On cubic s-arc-transitive Cayley graphs on finite simple groups, Europ. J. Combin., 26 (2005), 133–143. https://doi.org/10.1016/j.ejc.2003.10.015 doi: 10.1016/j.ejc.2003.10.015
![]() |
[26] |
S. J. Xu, X. G. Fang, J. Wang, M. Y. Xu, 5-arc-transitive cubic graphs on finite simple groups, Europ. J. Combin., 28 (2007), 1023–1036. https://doi.org/10.1016/j.ejc.2005.07.020 doi: 10.1016/j.ejc.2005.07.020
![]() |
[27] |
F. G. Yin, Y. Q. Feng, Symmetric graphs of valency 4 having a quasi-semiregular automorphism, Applied Math. Comput., 399 (2021), 126014. https://doi.org/10.1016/j.amc.2021.126014 doi: 10.1016/j.amc.2021.126014
![]() |
[28] |
F. G. Yin, Y. Q. Feng, J. X. Zhou, S. S. Chen, Arc-transitive Cayley graphs on nonabelian simple groups with prime valency, J. Combin. Theory Ser. A, 177 (2021), 105303. https://doi.org/10.1016/j.jcta.2020.105303 doi: 10.1016/j.jcta.2020.105303
![]() |
[29] | J. X. Zhou, Y. Q. Feng, The automorphisms of bi-Cayley graphs, J. Combin. Theory Ser. B, 116 (2016), 504–532. |