Characterization of Families of Rank 3 Permutation Groups by the Subdegrees I

Similar documents
A Design and a Code Invariant

The cycle polynomial of a permutation group

Generalised quadrangles with a group of automorphisms acting primitively on points and lines

Permutation Groups. John Bamberg, Michael Giudici and Cheryl Praeger. Centre for the Mathematics of Symmetry and Computation

A Family of One-regular Graphs of Valency 4

Rank 3 Latin square designs

AN EXAMPLE OF USING STAR COMPLEMENTS IN CLASSIFYING STRONGLY REGULAR GRAPHS

ORBITAL DIGRAPHS OF INFINITE PRIMITIVE PERMUTATION GROUPS

Self-Complementary Arc-Transitive Graphs and Their Imposters

Commutative Association Schemes Whose Symmetrizations Have Two Classes*

Imprimitive symmetric graphs with cyclic blocks

Asymptotically optimal induced universal graphs

ON THE ORDER OF ARC-STABILISERS IN ARC-TRANSITIVE GRAPHS, II

arxiv: v1 [math.gr] 7 Jan 2019

Strongly Regular Decompositions of the Complete Graph

RING ELEMENTS AS SUMS OF UNITS

Pentavalent symmetric graphs of order twice a prime power

MTH 309 Supplemental Lecture Notes Based on Robert Messer, Linear Algebra Gateway to Mathematics

DISTINGUISHING PARTITIONS AND ASYMMETRIC UNIFORM HYPERGRAPHS

Intrinsic products and factorizations of matrices

Chordal Coxeter Groups

An approximate version of Hadwiger s conjecture for claw-free graphs

COUNTING INVOLUTIONS. Michael Aschbacher, Ulrich Meierfrankenfeld, and Bernd Stellmacher

EIGENVALUES OF THE ADJACENCY MATRIX OF CUBIC LATTICE GRAPHS

Linear Algebra and its Applications

THE COMPLEXITY OF DISSOCIATION SET PROBLEMS IN GRAPHS. 1. Introduction

Strongly regular graphs constructed from groups

SOME QUESTIONS ARISING FROM THE STUDY OF THE GENERATING GRAPH

Rank 3 permutation groups

8th PhD Summer School in Discrete Mathematics Vertex-transitive graphs and their local actions II

Parity Versions of 2-Connectedness

SOME DESIGNS AND CODES FROM L 2 (q) Communicated by Alireza Abdollahi

Compositional roots of Hénon maps

Triangle-free graphs with no six-vertex induced path

-- \x if A is a nonsquare;

Applying block intersection polynomials to study graphs and designs

Isomorphisms and Normality of Cayley Digraphs of A5

The L 3 (4) near octagon

Square 2-designs/1. 1 Definition

-ARC-TRANSITIVE GRAPHS Dragan Marusic IMFM, Oddelek za matematiko Univerza v Ljubljani Jadranska 19, 1111 Ljubljana Slovenija

Symmetric graphs of order 4p of valency prime

HAMILTON CYCLES IN CAYLEY GRAPHS

Permutation decoding for the binary codes from triangular graphs

Codes from generalized hexagons

Preliminaries and Complexity Theory

Root systems and optimal block designs

Most primitive groups are full automorphism groups of edge-transitive hypergraphs

Near-domination in graphs

Locally primitive normal Cayley graphs of metacyclic groups

On the adjacency matrix of a block graph

May 6, Be sure to write your name on your bluebook. Use a separate page (or pages) for each problem. Show all of your work.

Math 203, Solution Set 4.

arxiv: v1 [cs.ds] 2 Oct 2018

Berge Trigraphs. Maria Chudnovsky 1 Princeton University, Princeton NJ March 15, 2004; revised December 2, Research Fellow.

Independent generating sets and geometries for symmetric groups

Group connectivity of certain graphs

Asymptotically optimal induced universal graphs

Combining the cycle index and the Tutte polynomial?

On non-hamiltonian circulant digraphs of outdegree three

Rao s degree sequence conjecture

arxiv: v2 [math.gr] 20 Apr 2017

Claw-Free Graphs With Strongly Perfect Complements. Fractional and Integral Version.

K 4 -free graphs with no odd holes

Self-complementary circulant graphs

Energy of Graphs. Sivaram K. Narayan Central Michigan University. Presented at CMU on October 10, 2015

Divisible Design Graphs

Topics in Graph Theory

arxiv: v1 [math.co] 19 Nov 2016

Simple connectedness of the 3-local geometry of the Monster. A.A. Ivanov. U. Meierfrankenfeld

Overgroups of Intersections of Maximal Subgroups of the. Symmetric Group. Jeffrey Kuan

Distinguishing Chromatic Number and NG-Graphs

Permutation Groups and Transformation Semigroups Lecture 4: Idempotent generation

Graph Classes and Ramsey Numbers

Graphs with prescribed star complement for the. eigenvalue.

An algebraic proof of the Erdős-Ko-Rado theorem for intersecting families of perfect matchings

5 Flows and cuts in digraphs

MAXIMAL SUBALGEBRAS AND CHIEF FACTORS OF LIE ALGEBRAS DAVID A. TOWERS

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.

Math 3121, A Summary of Sections 0,1,2,4,5,6,7,8,9

TRIPLE FACTORIZATION OF NON-ABELIAN GROUPS BY TWO MAXIMAL SUBGROUPS

Strongly regular graphs and the Higman-Sims group. Padraig Ó Catháin National University of Ireland, Galway. June 14, 2012

Michael Giudici. on joint work with variously John Bamberg, Martin Liebeck, Cheryl Praeger, Jan Saxl and Pham Huu Tiep

Parity Dominating Sets in Grid Graphs

A note on cyclic semiregular subgroups of some 2-transitive permutation groups

Two Variable Pfaffian Identities and Symmetric Functions*

1 Introduction We adopt the terminology of [1]. Let D be a digraph, consisting of a set V (D) of vertices and a set E(D) V (D) V (D) of edges. For a n

Quivers of Period 2. Mariya Sardarli Max Wimberley Heyi Zhu. November 26, 2014

Arc-transitive pentavalent graphs of order 4pq

Intriguing sets of vertices of regular graphs

Strongly Regular Graphs, part 1

1.3 Vertex Degrees. Vertex Degree for Undirected Graphs: Let G be an undirected. Vertex Degree for Digraphs: Let D be a digraph and y V (D).

12.1 The Achromatic Number of a Graph

SHUHONG GAO AND BETH NOVICK

Vertex opposition in spherical buildings

k-degenerate Graphs Allan Bickle Date Western Michigan University

The doubly negative matrix completion problem

Connectivity of Cayley Graphs: A Special Family

A hypergraph Turán theorem via lagrangians of intersecting families

Codes, Designs and Graphs from the Janko Groups J 1 and J 2

Automorphism groups of wreath product digraphs

Transcription:

Vol. XXI, 1970 151 Characterization of Families of Rank 3 Permutation Groups by the Subdegrees I By D.G. HIGMAN 1. Introduction. The terminology and notation of [5] for rank 3 permutation groups are used throughout. We consider two cases of the problem of determining the rank 3 permutation groups for which the degree and subdegrees are specified in terms of a parameter. The results are as follows. Theorem I. I/G is a rank 3 permutation group o/degree m 2 with subdsgrees 1, 2 (m -- 1) and (m -- 1)2, m > 2, then G is isomorphic with a subgroup o/the wreath product Zm ~ Z2 o/the symmetric groups of degrees m and 2, in its usual action on m 2 letters. Theorem H. I/ G is a rank3 permutation group o/ degree (2)with 8ubdegre~8 1, 2(m -- 2) and (m- 2) 2, m > 5, then G is isomorphic with a 4-/old transitive subgroup o/ m in its action on the 2-element subsets, unless one o/the/ollowing holds: (a) G ~ PILL2 (8), (b) # ~ 6 and m ~ 9,17, 27 or 57, (c) /~=7andm=51, or (d) ~u ---- 8 and m ---- 28, 36, 325, 903 or 8,128. Concerning the exceptional cases in Theorem II, PILL2 (8) as a subgroup of Z~ is the only example of a subgroup of Zm, m ~ 4, which is not 4-fold transitive but has rank 3 on the 2-element subsets (see Lemma 5). The case /~----6, m--" 9 is realized by the group G2 (2) and is known [4] to be the only such group. The remaining cases are undecided. 2. Rank 3 groups and strongly regular graphs. The proofs of Theorems I and II rest on the connection between rank 3 permutation groups of even order and strongly regular graphs [5, 10], and the known characterizations of the graphs of L2-type [9] and triangular type [2, 3] and [6, 7]. The graphs considered in this paper are finite, undirected, and without loops. A graph ~ with n vertices is strongly regular [1] ff there exist integers k, l, ~,/x such that (1) each vertex is adjacent to exactly k vertices and non-adjacent to exactly 1 other vertices, k and l positive, and

152 D.G. HIGMAN ARCH. MATH. (2) two adjacent vertices are both joined to exactly 4 other vertices and two nonadjacent vertices are both adjacent to exactly # vertices. Assume that ~r is strongly regular. Then (3) ~Z=k(k--4--2), (4) the minimum polynomial of the adjacency matrix A of fr is (x - 2) (x~ - (4 -/~) x - (2 -/~)), (5) A has k as eigenvalue with multiplicity 2, and ~he multiplicities/, g of the roots r, s of x 2 -- (4 -- #) x -- (k -- #) as eigenvalues of A are respectively ]_ (k + l)s + k and g-- (k + l)r + k 8--F r~8 with ] + g =- k + l, and (6) one of the following holds (a) 2=1,#=4-~-1=2/2 and ]=9=2, or (b) d ----(4 -- #)2 _~ 4(2 -- #) is a square. The strongly regular graph ~r is connected ff and only if/z > 0, and its comple- ment ~ is connected if and only if # < 2. We say that ~ is primitive if ~ and ~ are connected, so that (7) ~r is primitive if and only if 0 < ft < k. If # ----- 0, then (identity) w (adjacency) is an equivalence relation on the set of vertices, hence (8) If/~ = 0, then 2 -J- 2 [ n, and ff tt = 2, then 1 -J- 1 [n. If G is a rank 3 permutation group of even order on a finite set X, ] X[ ---- n, and if lj and T' are the nontrivial orbits of G in X X, then the graphs ~ = (X, LJ) and = (X, F) are a complementary pair of strongly regular graphs, each admitting G as a rank 3 automorphism ~oup, the parameters k, l and 4,/x being respectively the subdegrees (other than 1) and the intersection numbers of G [5, 10]. Primitivity of the permutation group G is equivalent ~o primitivity of the graph ~. 3. Proof of Theorem I. The wreath product Zm ~ Z2 of the symmetric group Zm of degree m > 2 and the symmetric gtoup Z2 of order 2 is constructed as the split extension of the group N = Zm Zm by X2 ---- <~>, where ~ acts on iv according to (~l, zt2) ~ = (~2, ~1), gi ~Zm. This group acts as a rank 3 permutation group on X X, X -~ {1, 2... m} according to (i,?')(... ) = (i"~, j=') and (i, ])~ ----(], i), with 2 = 2(m -- 2), 1 = (m -- 2) 9" and # = 2. The action is imprimitive for rn = 2 and primitive for m >= 3. The graph ~fm afforded by the suborbit of len~h 2 (m -- 2), in which vertices (a, b) and (c, d) are adjacent if and only ff they are distinct and a = c or b = d, is isomorphic wit h the line graph of the complete bipartite graph on 2m vertices, i.e. the strongly regular graph of L2-type on 2m vertices. Lemma 1. Xm ~ Z2 is the ]ull automorphism group o/~m, m > 2.

Vol. XXI, 1970 Characterization of Families of Permutation Groups I 153 Proof. Each vertex (a, b) is contained in exactly two maximal cliques, with vertices {(a, x)[x = 1, 2..., m} and {(y, b)[y -~ 1, 2,..., m}, respectively. If a is an automorphism of ~m, we have two possibilities. Case 1. (1, 1) ~ = (al, bl) and (1, 2) a = (al, b2). In this case (i, j)a ~ (al,?'~) and (i, 1) ~ ~ (i% bi) with ~, fle Zm such that 1 ~ = ai and 1~ = bl. If i and ] are both =~ 1, then (i, ~), as the unique vertex =~ (1, 1) adjacent to (i, 1) and (i, ~), must be mapped by a onto the unique vertex ~= (ai, bi) adjacent to (i% bi) and (al, ]~), i.e., (i, 2")~ = (i% ~). Hence a = (~, fl) e h r. Case 2. (1, 1)a = (ai, bl) and (1, 2) a = (a2, bi). In this case (1, i) a = (]~, bi) and (2, 1) a -= (al, 2 ~) with (~, fl) ezra, so (i, ])a ---- (i~, i a) ---- (i, ])(~'~)*. Hence a~- (cr ~ ez~ ~Z2. Lemma 2. Let ~ be a strongly regular graph such that k ~ 2(m -- 1) and 1 = (m- 1) 2 ]or some m ~ 2. Then ~ is isomorphic with ~q~m unless m = 4, in which case there is, up to isomorphism, exactly one exceptional graph. Proof. In view of [9], it suffices to prove that/~ = 2. By (8) of Section 2 we have # > 0. By (3) of Section 2 we have #(m-- I) = 2(2m-- 3-- 2). Assume first that /~ is even, /i ~ 2#0. Then /~o(m -- i) ~- 2m -- 3 -- 2, so that (#o -- 2)m ----- ~uo -- 3 -- 2. We have 2>=0, and since 0</~k=2(m-- 1), 0</zo~m--l. Hence (#o -- 2)m --< /zo -- 3 --< m -- 4, sothat(/io--3)m~--4 and hence #o ----< 2. If/~o = 2, then ~ = -- 1 which is impossible, hence #0 ----- 1 and /l ---- 2 as claimed. Now assume that/~ is odd, so that 2 [ m -- 1 and we can write m = 2 t + 1, t _>-- 1. Then/~t = 4t -- 1 -- ~, that is, ~ = (4 --/z)t -- 1, so that/z ~ 3. By (6) of Section 2 we have that 9t 2 -+- 4t or t 2 ~- 8t + 4 must be a square according as # = 1 or/~ ---- 3. But the non-negative integral }olutions of the equations y2 _-- 9t 2 + 4t and y2 = t 2.~_ 8t -}- 4 are respectively y ---- 0, t ---- 0 and y ---- 2, t ~ 0, contrary to t _--> 1. Since it is known [4] that the exceptional graph in case m = 4 of Lemma 2 does not admit a rank 3 automorphism group, we have Theorem I as an immediate Corollary to Lemmas 1 and 2. If H ~ Zm, m --_> 2, is doubly transitive, then H ~27u is a rank 3 subgroup of Zm ~Z2 on the m 2 letters. There are rank 3 subgroups of Zm l Z2 not of this type and we make no attempt here to classify them. Of course it should be noted that Lemma 2 is purely a result about graphs with no reference to groups. The same remark applies to Lemma 4 in the next section. 4. Proof of Theorem li. The symmetric group 2:m on X ---- {1, 2... m}, m --> 3, acts as a rank 3 group: on ~the set of ( 2 ) 2-element subsets of X, with subdegrees k ~- 2(m -- 2), 5----(m-2 2), and ~u -- 4. The action is imprimitive if m ---- 4 and g primitive ff m _-- 5. The graph Jm afforded by the snborbit of length 2 (m -- 2), in

154 D.G. HIGMAN ARCH. MATH. which two 2-element subsets S and T are adjacent ff and only ff IS n T[ ---- 1, is isomorphic with the line graph of the complete graph on m vertices, i.e., the gtaph of oo (;) vertices. Lemma 3. Zm is the [ull automorphism group o/jm, m > 5, while 4 has index 2 in the automorphism group o/j4. Proof. The maximal cliques of Jm containing a vertex {a, b} are of two types. [a]---~ {{a,x}lxex- {a}}, (I) [b] = {{y, b) l Y e Z -- {b)) and (II) {a,b}, {b,z} and {a,z}, z~ {a,b}. If a is an automorphism of Jm and {a, b} a = {c, d}, then, if m > 5, [a]a = [c] or [d]. Here we define a n = c if [a] a = [c]. Then it is easy to verify that zr e m and ~z induces a. In case m ---- 4 there are involutions in the automorphism group of ]4 interchanging the two types (I) and (II) of maximal cliques, e.g., ~ mapping {1, 3} onto {2, 4} and fixing all other vertices. We see that <Z4, cr is the full automorphism group of f14. Lemma 4. Let ~ be a strongly regular graph with k = 2(m -- 2) and l =(m;2), " " m > 4. Then ~t -~ 4 unless /x=6 and m=7,9,17,27 or 57, #=7 and m=51, or /.t = 8 and m=28,36,325,903 or 8,128. Proof. If ~u = 0, then m ---- 4 by (8) Section 2 and hence ), ---- 3 by (3), which is impossible by (6). Hence # > 0. By (3) we have (*) #(m-- 3) ---- 4(2m-- 5 -- 4). We consider two cases according as/~ is even or odd. Case 1. #----2/xo. In this case /xo(m--3)=2(2m--5--~), = (4 --/x0)m + 3/~0 -- 10. Hence, since /~o < (m -- 2), we have so that 0_--<2)~ (~o-- 4)m < 3//o-- 10 ~ 3(m-- 2) -- 10 < 3 m-- 16. 5-m Hence (/~o -- 7) m =< --16 and ~uo < 7. If #0:5or6, then 2-- ~ or4--mso that m = 5 or 4 respectively. In the first case, /x0 = 5 > m -- 2 = 3 and in the second/xo = 6 > m -- 2 = 2, so both are ruled out. The remaining possibilities are: 2 3m--7 9 m 2 -- 34 m -f- 25 2 3 4 6 m--1 m--2 2 (m -- 2) ~- By (6) of Section 2, the entries in the last row are squares. 8m+l

u XXI, 1970 Characterization of Families of Permutation Groups I 155 For /x0 ---- 1 we have the equation 4y 2 = 9m 2 -- 34m -~ 25, for which the (nonnegative integtal) solutions are y ---- 1, m ---- 3 and y ---- 0, m = 1, both of which are impossible. m--5 If/x0 ---- 3 we have by (4) and (5) of Section 2 that r -- ----if-- and s = -- 4, so that ] _ 4 mm - 2). Since (m + 3, m) 13 and (m -~ 3, m -- 2) 15, we have that m ~- 3 ] 4-3- 5 giving m ---- 7, 9, 12, 17, 27, or 57. If m = 12, then 2 ----- 1312, so this case is impossible. If #0 ---- 4, we have the equation y2 ---_ 8ra ~- 1. Writing y ---- 2w + 1, we have (w+,) m----, k=w 2~-w-4 and l-----~(w 2-~w-6)k, and since 2=1 and /~ ~-- 8, we get r = w -- 3 and s = -- w -- 4 by (4) of Section 2. Hence by (5) of Section 2, f = w(w + 2) (w + 3) (w2 w - 4) S(2w + 1) and since w _= 3 and (2w + 1, w + 2) 13, (2w 1, w + 3) 15 and (2w + 1, w 2 + w-- 4)]17, we must have 2w ~- 113 9 5 9 17, giving i.e., w = 7, 8, 25, 42 or 127, m---- 28, 36, 325, 903 or 8,128. Case 2. # -= 1 (2). In this case (*) implies that 41m--3, and we write m=4t-~3. Then by (*), ~ ---- (8 --/~)t ~ 1 and ~u ~ 9. The possibilities are: ] 1 3 5 7 9 7t-~l 5t~l 3t~-i t~-i 0 d 49 t2 -~ 32t-~- 4 25t2~12t 9t2 ~ 8t-}- 4 t2 -}- 20t ~ 16 85 Again, the entries in the last row are squares by (6) of Section 2, so/~ :~ 9. The nonnegative integral solutions of the equations y2 _-- 49t 2 -b 32t ~- 4, y2 _-- 25t 2 ~ 12t and y2 = 9t2 ~_ 8t + 4 are y = 2, t = 0, y---- t = 0 and y = 2, t = 0 respectively, so these cases are ruled out. Finally, the equation y2 -- t2 ~_ 20t ~- 16 has just the solutions y ---- 4, t = 0, which is ruled out, and y = 20, t = 12, giving m ~ 5t. Lemma 5. I/ G ~ m, m ~ 4, has rank 3 on the 2-element subsets, then G is 4-/old transitive unless m = 9 and G ~ PI~L2 (8). nkl Proof. The cases m ---- 4 and 5 follow from the fact that (b-~ divides I G]. Assume that m > 5 and that G is not 4-fold transitive. Clearly G is transitive on the set of 4-element subsets, i.e., G is 4-homogeneous, and ff S is a 2-element subset, then Gs is transitive on the 4-element sets containing S. Hence ff A is a 4-element set, GA: GA, S ~-- 6. Hence GAlA has order 12, and we have by KA~TO~'S result [8] that possibilities are m G 6 PSL2(5), PFL2(5) 9 PSL2 (8), PFL2 (8) 33 PI'L2 (32). nkl The divisibility ~ [G[ rules out au but the case m = 9, G ---- PFL2 (8).

156 D.G. HIGMAN ARCH. MATH. Note. The result stated by KA~TO~ in [8] is proved there only under the additional assumption that GA[ A has order ~: 8, which is quite sufficient for our purposes. According to [2, 3] and [6, 7] if ~ is a strongly regular graph with ]c ---- 2 (m -- 2), 2 and/z ~- 4, then ~ ~ Jm unless m ---- 8, in which case there are exactly three exceptions. Since none of these exceptional graphs admits a rank 3 automorphism group [4], and since the complement of ~r in the case/z ---- 6, m ---- 7 has the parameters of J7, Theorem II follows by Lemmas 3, 4 and 5. We remark that for m ~ 4, any rank 3 permutation group with subdegrees 1, 4, 1 is either isomorphic with Z4 or the full automorphism group of J4. The exceptional case ~u ----- 6, m ~ 9 is realized by the group G2 (2). It is in fact known that there is just one strongly regular graph with these parameters which admits a rank 3 automorphism group, and that G~ (2) is its full automorphism group, no proper subgroup of which has rank 3 [4]. The question of the existence (even of graphs) for the remaining exceptional eases is undecided. References [1] R. C. BosE, Strongly regular gtaphs, partial geometries and partially balanced designs. Pacific g. Math. 13, 389--419 (1963). [2] L. C. Cm~G, The uniqueness and nonuniqueness of the triangular association scheme. Sci. Record 3, 604--613 (1959). [3] L. C. C~A:SG, Association schemes of partially balanced block desigus with parameters v ---- 28, nl = 12, n2 = 15, P~I ---- 4. Sci. Record 4, 12--18 (1960). [4] M. HEST~S, D. G. HIG~A~ and C. C. Sn~s, Rank 3 permutation groups of small degree (in preparation). [5] D. G. HIG~A~, Finite permutation groups of rank 3. Math. Z. 86, 145--156 (1964). [6] A. J. HO~'FV~Ze, On the uniqueness of the triangular association scheme. Ann. Math. Star. 31, 492--497 (1960). [7] A. J. HOFFMA~, On the exceptional case in a characterization of the arcs of a complete graph. IBM J. Res. Develop. 4, 487--496 (1961). [8] W. M. KANTOR, 4-homogeneous groups. Math. Z. 103, 67--68 (1968). [9] S. S. Sm~IKEA~DE, The uniqueness of the L2-association scheme. Ann. Math. Star. 30, 781--798 (1959). [10] C. C. Sn~s, On graphs with rank 3 automorphism groups. J. Comb. Theory (to appear). Anschrift des Autors: D. G. Higman Department of Mathematics University of Michigan Ann Arbor, Mich. 48104, USA Eingegangen am 18. 7. 1969