Buildings and Kneser graphs
Citation for published version (APA):Güven, Ç. (2012). Buildings and Kneser graphs. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR721532
DOI:
10.6100/IR721532
Document status and date: Published: 01/01/2012
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
providing details and we will investigate your claim.
A catalogue record is available from the Eindhoven University of Technology Library. ISBN: 978-90-386-3067-0
The research which leaded to this thesis was financed by Lex Schrijver, who was rewarded with Spinoza Prize of the Netherlands Organization for Scientific Research (NWO) in 2005.
The cover of this thesis is designed by Atike Dicle Pekel Duhbacı. Printed by Print Service at Eindhoven University of Technology.
Buildings and Kneser Graphs
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus, prof.dr.ir. C.J. van Duijn, voor een
commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op woensdag 25 januari 2012 om 16.00 uur
door
Çiçek Güven Özçelebi
P
REFACE
The following questions with their possible generalizations initiated years of mathemati-cal research, including this thesis:
If C is a collection of mutually intersecting k-subsets of a fixed n-set, how big can C be? And in the extreme case, what is the structure of C?
According to Erd˝os [53], this question was answered by Erd˝os, Ko, and Rado, already
in 1938, but they first published this result in 1961 [54]: For 2k≤ n, one must have
|C| ≤ n−1
k−1
, and if equality holds then C is the collection of all k-subsets containing some
fixed element of the given n-set for 2k< n. For 2k = n, this bound can be obtained by
picking one k-set from each complementary pair in any way.
The Kneser graph K(n, k) is the graph with as vertices the k-subsets of a fixed n-set,
where two k-subsets are adjacent when they are disjoint. In this terminology, Erd˝os, Ko,
and Rado found the largest cocliques (independent sets of vertices) in K(n, k).
Many people have studied generalizations and variations of this problem, and that is also what we shall do in this thesis.
The question that initiated this research is the following:
“ Try to describe Erd˝os-Ko-Rado sets with maximal size in the various geometries arising from a spherical diagram by circling any set of nodes.”
Here ‘circling a certain set of nodes’ in a diagram means defining the type of flags of the geometry which will be the vertices of the graph. This problem generalizes the
classical Erd˝os, Ko, and Rado problem, which is about sets, to a problem about buildings
of spherical type.
The natural generalization of being disjoint as sets is being ‘far apart’ as geometrical objects. We have objects and some kind of a distance function, and define a Kneser graph with our objects as vertices, two objects being adjacent when they have maximum distance. The goal is always to find the maximal size of a coclique in such a graph, and to characterize the cases that reach this maximum.
Our objects will usually be flags of some fixed type in a finite building. This thesis consists of three parts and seven chapters.
Part one is the introduction and consists of two chapters.
In Chapter 1, relevant definitions for the content of this thesis are given.
In Chapter 2, we present our first generalization of the Kneser graph, which we call
generalized Kneser graphs, GK(n : k1, . . . , kl). These graphs are defined over inclusive
the conditions under which these graphs are connected are determined. Descriptions for maximal coclique sizes for generalized Kneser graphs in many cases are given, the existence of relations between the maximal coclique sizes for different examples of those graphs are shown.
Part two consists of Chapter 3, Chapter 4, and Chapter 5. In this part, further gen-eralizations of Kneser graphs are introduced, from finite sets to buildings of spherical type, and from being disjoint as sets to being far apart as geometrical objects. Chapter 3 describes this generalization. We introduce Kneser graphs on buildings in this chapter.
While studying generalizations of the Erd˝os-Ko-Rado theorem and the chromatic
number of Kneser-type graphs, one needs information about maximal cocliques of near-maximal size. In Chapter 4 we describe a simple construction that in the most interesting cases produces all such near-maximal cocliques. This chapter treats the results of the paper [22].
In Chapter 5, we work on maximal sizes of cocliques for the point-hyperplane graphs. This chapter is about the results of the paper [11]. Our result gives an upper bound for the coclique size for Kneser graphs on point hyperplane flags, and characterizes the case when equality holds.
Part three consists of Chapter 6 and Chapter 7. In this part, other parameters and eigenvalues of Kneser type graphs are calculated.
In Chapter 6, the Smith Normal Forms (SNF) of some Kneser type graphs are con-jectured and proved. A symmetry relation is generalized which holds in strongly regular graphs with prime power eigenvalues. So once the p-rank is known, the SNF is known in some cases because of this symmetry. We conjecture and come up with partial results for some graphs related to some generalized quadrangles.
In Chapter 7, four geometric objects are taken, and their collinearity graphs are exam-ined. These geometries are parapolar spaces. With the help of this fact, the parameters of the related distribution diagrams are calculated. Knowing those parameters enables us to calculate the eigenvalues of some Kneser type graphs to be powers of q. We make use of Delsarte’s Linear Programming bound and Hoffman bound to come up with bounds on the sizes of some subconfigurations.
C
ONTENTS
Preface vii
List of Figures xiii
List of Tables xv
I
Introduction
1
1 Preliminaries 3
1.1 Graphs . . . 3
1.1.1 Regular partitions . . . 6
1.2 Field with one element and q-analogue . . . . 8
1.3 Groups . . . 8
1.3.1 Coxeter groups . . . 9
1.3.2 Groups with a Tits system . . . 11
1.3.3 Finite simple groups of Lie type . . . 13
1.4 Geometric objects . . . 14
1.5 Buekenhout-Tits geometries . . . 14
1.5.1 Point-line geometries . . . 15
1.5.2 Chamber complexes and foldings . . . 16
1.5.3 Buildings . . . 19
1.5.4 Projective spaces . . . 21
1.5.5 Polar spaces . . . 22
1.6 Association schemes . . . 26
1.6.1 Delsarte’s linear programming bound . . . 29
2 A Generalization of Kneser Graphs on Increasing Sequences of Sets 31 2.1 Introduction . . . 31
2.1.1 Codes . . . 31
2.1.2 (Classical) Kneser graphs . . . 32
2.1.3 Johnson graphs . . . 32
2.1.4 The relation between J (n, k) and K(n, k) . . . . 33
2.1.5 History of the Kneser graphs . . . 33
2.2 Generalized Kneser graphs . . . 34
2.2.1 Connectedness of the generalized Kneser graphs . . . 35
2.3 Maximal cocliques in GK(n : k1, . . . , kl) . . . 42
2.3.1 Foldings . . . 45
II
A Unifying Approach: From Sets to Groups
49
3 Generalizations to Buildings of Spherical Type 51 3.1 Understanding the graphs . . . 513.1.1 q-Kneser graphs . . . 51
3.1.2 Kneser graphs for Coxeter groups . . . 52
3.1.3 Generalization of Kneser graphs over buildings of spherical type . 53 3.1.4 Taking sums over the Weyl group . . . 55
3.2 Cocliques on flags of P G(4, q) . . . . 57
3.2.1 Graphs on point-hyperplane pairs . . . 57
3.2.2 Graphs on point-line flags . . . 57
3.2.3 Graphs on point-plane flags . . . 58
3.2.4 Graphs on line-plane flags . . . 60
4 Unique Coclique Extension Property 61 4.1 Introduction . . . 61
4.1.1 Method . . . 61
4.1.2 Matroids . . . 62
4.1.3 Statement of the main theorem . . . 63
4.2 Subspaces of a projective space . . . 64
4.3 Points in a polar space . . . 65
4.4 Totally singular lines in an orthogonal space . . . 65
4.5 Minuscule weights . . . 66
4.6 Adjoint representation . . . 67
4.7 Nonexamples . . . 68
5 Maximal Cocliques in Point-Hyperplane Graphs 71 5.1 Introduction . . . 71
5.1.1 Rank 1 matrices . . . 72
5.1.2 The thin case . . . 72
5.2 Maximum-size cocliques . . . 73
5.3 Maximum number of points . . . 74
5.4 Classification of cocliques for n≤ 4 . . . . 77
5.4.1 P G(2, q) . . . . 78
5.4.2 P G(3, q) . . . . 78
III
Calculating Some Other Parameters and Eigenvalues
81
6 Smith Normal Forms of Some Kneser Graphs for Buildings 83 6.1 Introduction . . . 836.1.1 Preliminaries . . . 83
6.1.2 Review of the related problems . . . 87
CONTENTS xi
6.3 Non-collinearity graphs of generalized quadrangles . . . 89
6.3.1 Non-collinearity graph of GQ(q, q2) . . . 90
6.3.2 Non-collinearity graph of GQ(q, q) . . . . 90
6.3.3 Non-collinearity graph of GQ(q2, q) . . . . 92
6.3.4 Non-collinearity graph of GQ(q2, q3) . . . 92
6.3.5 Non-collinearity graph of GQ(q3, q2) . . . 92
6.4 Graphs with prime power eigenvalues and SNF . . . 93
6.4.1 Oppositeness graphs . . . 93
6.4.2 Bipartite graph of disjoint point and lines in P G(2, q) . . . . 94
7 From graphs of Lie type to Kneser graphs on Buildings 95 7.1 Introduction . . . 95
7.2 Preliminaries . . . 96
7.2.1 Four diagrams . . . 96
7.2.2 Graphs of Coxeter type . . . 97
7.2.3 Graphs of Lie type . . . 99
7.2.4 Relations between Γ(G, GS, r) and Γ(W, WS, r) . . . 102
7.2.5 Parapolar spaces . . . 103
7.2.6 Reading the diagrams and chain calculus . . . 104
7.3 Four point-line geometries . . . 107
7.3.1 The graph of Coxeter and Lie types E7,1 . . . 107
7.3.2 The graph of Coxeter and Lie types E6,2 . . . 115
7.3.3 The graph of Coxeter and Lie types E8,8 . . . 119
7.3.4 The graph of Coxeter and Lie types F4,1 . . . 124
7.4 Using the Hoffman bound and the DLPB bound . . . 129
7.5 Eigenvalue results for the Kneser graphs on buildings . . . 129
7.5.1 Eigenvalues of K(A2d−1(q),{d}) . . . 130 Abstract 131 Acknowledgements 133 Curriculum Vitae 135 Bibliography 137 Index 144
L
IST OF
F
IGURES
1.1 the Petersen graph with vertices labeled as 2-subsets of the set{1, 2, 3, 4, 5} 4
1.2 the distance distribution diagram of the Petersen graph . . . 7
1.3 the diagrams of the irreducible finite Coxeter systems . . . 10
1.4 for A2, positive roots r, s, and r + s . . . . 11
1.5 a simplicial 3-complex . . . 16
1.6 the diagram of the cube . . . 21
6.1 distribution diagram of the incidence graph of points and lines of P G(2, q) 94 7.1 the diagrams E6,2, E7,1, E8,8, and F4,1 . . . 96
7.2 the diagram of D6where node number 3 is circled . . . 99
7.3 distribution diagram of the E7,1(1) graph . . . 108
7.4 distribution diagram of the graph of Lie Type E7,1(q) . . . 111
7.5 distribution diagram of the E6,2(1) graph . . . 115
7.6 distribution diagram of the E8,8(1) graph . . . 120
L
IST OF
T
ABLES
2.1 diameters of generalized Kneser graphs . . . 37
2.2 maximum coclique sizes . . . 42
5.1 maximal coclique classification for P G(3, q) . . . . 78
7.1 the degrees of the finite irreducible Coxeter systems . . . 99
7.2 parameters for Chevalley groups . . . 102
7.3 the bound given by DLPB and actual YAfor the graph of Coxeter type E7,1 110
7.4 the bound given by DLPB and actual YAfor the graph of Coxeter type E6,2 116
7.5 the bound given by DLPB and actual YAfor the graph of Coxeter type E8,8 121
P
ART
I
I
NTRODUCTION
1
P
RELIMINARIES
In this Chapter, we introduce the general notions that are relevant to the content of this thesis. We give the definitions that are most relevant to a chapter within the chapter.
This book is self contained up to some background in elementary linear algebra, graph theory and group theory.
For graph theory, see [8], [59], [21], [60], and [24].
For standard theory of linear algebraic groups see [72], for the standard theory on groups of Lie type and finite groups of Lie type, see [36] and [37], for classical groups see [34]. For geometry of classical groups, see [92].
We define Coxeter groups, groups with (B, N )-pairs, buildings with related geomet-rical objects, and association schemes here, but for a complete understanding of these objects, see [97], [98], [84], [103], [27] and [1]. For the theory of Coxeter groups and Tits systems, [18] is a well known source.
For coset graphs for parabolic subgroups of groups of Lie type, see [21], Chapter 10. We refer to this chapter quite often in this thesis.
1.1
Graphs
The problems that are in the focus of this thesis can be considered as problems of alge-braic graph theory, which is the study of graphs using algealge-braic methods.
DEFINITION 1.1.0.1. In a graph Γ, a path of length i between two vertices u and v is
a sequence of distinct vertices u = u0, u1, . . . , ui = v such that for any k, 0≤ k ≤ i − 1,
the vertex uk is adjacent to uk+1. A walk is a path in which vertices or edges may be
repeated.
DEFINITION1.1.0.2. A graph is connected when it has one connected component, that is,
(it is non-empty and) for any pair of vertices, there is a path joining them.
All graphs of our interest are finite, without loops, without multiple edges, and undi-rected.
DEFINITION1.1.0.3. A (proper) vertex coloring of a graph Γ = (V, E) is a mapϕ : V → C
from the set V of vertices to a set C of colors, such that for u, v∈ V , if u is adjacent to v,
A proper vertex coloring of a graph Γ gives a partition of the set of vertices. For any
c∈ C, we call ϕ−1(c) a color class.
DEFINITION1.1.0.4. The chromatic number of a graph Γ is the smallest number of colors
needed for a proper vertex coloring of Γ, and is denoted byχ(Γ).
DEFINITION 1.1.0.5. A coclique in a graph is a set of vertices, such that no two vertices
in the set are adjacent. The maximum size of a coclique in a graph Γ is denoted byα(Γ).
Each color class is a coclique.
REMARK1.1.1. For Γ = (V, E),
|V | ≤ α(Γ)χ(Γ). (1.1)
DEFINITION1.1.0.6. A graph of order v is called strongly regular with parameters v, k,λ, µ
whenever it is not complete or edgeless and i . each vertex is adjacent to k vertices,
ii . for each pair of adjacent vertices there areλ vertices adjacent to both,
iii . for each pair of non-adjacent vertices there areµ vertices adjacent to both.
EXAMPLE1.1.0.7. The Petersen graph is strongly regular with parameters (10, 3, 0, 1).
{1,2} {3,4} {3,5} {4,5} {1,5} {2,3} {1,4} {2,4} {2,5} {1,3}
FIGURE1.1: the Petersen graph with vertices labeled as 2-subsets of the set{1, 2, 3, 4, 5}
DEFINITION1.1.0.8. For any pair of vertices u, v in a connected graph Γ, the length of the
shortest path from u to v in Γ is called the distance between them, which we denote by
d(u, v). The diameter of a connected graph Γ is the maximal distance occurring in Γ.
DEFINITION 1.1.0.9. For any pair of vertices u, v in a connected graph Γ, any path of
length d(u, v) is called a geodesic between them. A subset C of Γ is called convex or
geodetically closed if for any pair of vertices in C, all geodesics between them are in C.
1.1 GRAPHS 5
DEFINITION1.1.0.10. [21] A connected graph Γ is called distance-regular if it is regular
of valency k, and if for any two points u, v at distance i = d(u, v), there are precisely ci
neighbors of v in Γi−1(u) and bi neighbors of v in Γi+1(u). The sequence
ι(Γ) :={b0, b1, . . . , bd−1; c1, . . . , cd}
where d is the diameter of Γ, is called the intersection array of Γ; the numbers ci, bi,
and aiwhere
ai:= k− bi− ci (i = 0, . . . , d)
is the number of neighbors of v in Γi(u) for d(u, v) = i, are called the intersection
numbers of Γ. Clearly
b0= k, bd = c0= 0, c1= 1.
The size of Γi(u) is denoted by ki. The valency of Γ is k = k1.
For a distance-regular graph with adjacency matrix A, the matrices Ai, where (Ai)u,v=
1 if d(u, v) = i and (Ai)u,v= 0 otherwise, satisfy the relations [21, p. 127] :
A0= I , A1= A, (1.2)
AAi= bi−1Ai−1+ aiAi+ ci+1Ai+1, (1.3)
A0+ A1+· · · + Ad= J . (1.4)
This means that we can write each Ai as a polynomial in A, and that the minimal
polynomial of A has degree d + 1.
Let us define group actions here since we will define graphs based on them.
DEFINITION1.1.0.11. A group action of a group G on a set X is a map from G× X to X ,
written as g x for g∈ G, x ∈ X , satisfying the following properties:
(i) (g1g2)x = g1(g2x)for all g1, g2in G, x∈ X ,
(ii) 1x = x for all x∈ X , where 1 is the unit element of G.
If a group acts on a set X it automatically also acts on the subsets of X . A group action is transitive if for any a, b in A, there exists a g in G such that ga = b. A group action is
faithful if for any g, h∈ G, g 6= h, there exists an a in A such that ga 6= ha.
DEFINITION 1.1.0.12. A permutation group G is a group of permutations of a set X . It
acts on the elements of the set X naturally. An orbit of G is a set G x ={g x | g ∈ G}. The
stabilizer of x, which is denoted by Gx is the subset of G fixing x, i.e.{g | g x = x}. The
orbits of G on X× X are called orbitals.
DEFINITION1.1.0.13. An automorphism of a (finite) graph Γ = (V, E) is a permutation
σ of the vertex set V , such that for any edge e = (u, v), σ(e) = (σ(u), σ(v)) is also an edge.
The set of automorphisms of a given graph, under the composition operation, forms a group, which is known as the automorphism group of the graph, denoted by Aut(Γ).
DEFINITION 1.1.0.14. A graph Γ is vertex-transitive if for any two vertices v1and v2of
Γ, there existsσ∈ Aut(Γ) such that σ(v1) = v2.
DEFINITION 1.1.0.15. A graph Γ is edge-transitive if for any two edges e1 and e2of Γ,
there existsσ∈ Aut(Γ) such that σ(e1) = e2.
DEFINITION 1.1.0.16. A graph Γ is flag-transitive if for any two edges e1= (v1, u1) and
e2= (v2, u2) of Γ, there existsσ∈ Aut(Γ) such that σ(v1) = u1andσ(v2) = u2.
DEFINITION 1.1.0.17. A graph Γ is distance-transitive if for any two pairs of vertices
u, v and u′, v′ where d(u, v) = d(u′, v′) = i, there existsσ∈ Aut(Γ) such that σ(u) =
u′,σ(v) = v′.
DEFINITION1.1.0.18. Let G be a group, H be a subgroup of it, and S be any subset of G.
One can define a graph Γ(G, H, S) (when HSH = HS−1H) called coset graph of G on H
(with respect to S) on the left cosets of H in G. Two cosets g1Hand g2H will be adjacent
in this graph if and only if H g2−1g1H⊆ HSH.
The group G will act as a group of automorphisms of Γ(G, H, S) by left
multipli-cation: for any edge (g1H, g2H), g(g1H, g2H) = (g g1H, g g2H) is also an edge, since
H(g2−1g−1)(g g1)H = H g2−1g1H ⊆ HSH. This action is transitive: for any g1H, g2H,
(g2g−11 )g1H = g2H. The stabilizer of the vertex H, is the subgroup H.
Later on, we will define such graphs for finite groups of Lie type, where S will be a 1-set, and H will be a parabolic subgroup.
1.1.1
Regular partitions
DEFINITION 1.1.1.1. For a graph Γ = (V, E), we say a partition P of V is regular (or
equitable) if for any Vi, Vjin P, and any vertexν∈ Vi, ei, j, the number of neighbors ofν
in Vjis independent of the choice ofν.
Suppose A is a symmetric real matrix, whose rows and columns are indexed by a
set V = {1, 2, . . . , n}. Let P be a partition of V with parts Vi for 1≤ i ≤ m. Let A be
partitioned according to P into block matrices, that is:
A = A1,1 · · · A1,m .. . . .. ... Am,1 · · · Am,m ,
where Ai, jis the submatrix of A formed by rows in Viand columns in Vj.
The characteristic matrix S is the n× m matrix, whose j’th column is the
character-istic vector of Vj for j = 1, . . . , m. Define ni=|Vi|, K =diag(n1, . . . , nm). Let bi, j denote
the average row sum of Ai, j. Then the matrix B = (bi, j) is called the quotient matrix and
we have
K B = S⊤AS, S⊤S = K.
Let A be the adjacency matrix of a graph Γ with the set V of vertices where|V | = n.
1.1 GRAPHS 7
DEFINITION1.1.1.2. [21] The distribution diagram of Γ with respect to a regular
parti-tion P consists of a number of balloons bVi, one for each Vi ∈ P, and a number of lines
lViVj joining the two balloons, bVi and bVj one for each pair{Vi, Vj} for which ei, j 6= 0.
Lines lViVi are not drawn. In the balloon bVi,|Vi| is written, and at the Vi end of lViVj, ei, j
is written. The number eViVi is written next to bVi.
EXAMPLE1.1.1.3. When a graph is distance regular, the distribution diagram where the
partition P is based on the distance to a fixed vertex, is called a distance distribution
diagram.
0 2 1 2 1 1 3 6
FIGURE1.2: the distance distribution diagram of the Petersen graph
When the partition is regular, the row sum of the matrix Ai, jis constant, and is equal
to ei, jfor all i, j∈ {1, . . . , m}. The matrix (ei, j)Vi,Vj∈Pis the quotient matrix of Γ.
Then, Ai, j1 = ei, j1 for i, j = 0, . . . , d, where 1 is the all 1’s vector. Thus,
AS = SB which leads to the following well known result:
LEMMA1.1.2. If for a regular partition P, the vector v is an eigenvector of B, for an
eigen-valueλ, then S v is an eigenvector of A for the same eigenvalue λ.
Proof. Bv =λv implies AS v = SBv = λS v. For a distribution diagram, one can consider the balloons as vertices and lines as edges. The graph obtained is called a distribution graph.
For more details, see [61], [24].
DEFINITION1.1.1.4. Given a coset graph Γ = (G, H, r), the associated double coset graph
∆ = DC(G, H, r) is the distribution graph of Γ with respect to the partition{H gH|g ∈ G}
of the vertex set of Γ.
The graph ∆ has a regular partition. The vertex set of ∆ is H\G/H = {H gH | g ∈ G}
and two vertices H g1H and H g2Hare adjacent if and only if H g1H⊆ H g2H r H.
The vertex H r H is the unique neighbor of H.
PROPOSITION 1.1.3. [21] Let Γ = (G, H, r) be a coset graph of G on H, and let ∆ =
DC(G, H, r) be the associated double coset graph of G on H. Then, for arbitrary g1, g2
in G:
1. The cosets g1H and g2H are at distance i in Γ if and only if H and H g−12 g1H have distance i in ∆.
2. The graph Γ is connected if and only if ∆ is connected; in this case, Γ and ∆ have the same diameter.
3. G in its action by left multiplication acts distance-transitively on Γ if and only if ∆ is a walk, possibly with loops added.
1.2
Field with one element and q-analogue
Let q be a prime power pt for some prime p. We denote the finite field of order q by
Fq. A field with one element does not exist, since in a field, the identity of multiplication
and addition should be distinct. However, the suggestive name field with one element is used for a mathematical object, which behaves like a field, and has characteristic 1. It
is denoted byF1. In 1956, Jacques Tits suggested studying the mathematics ofF1[95].
DEFINITION 1.2.0.5. A q-analogue of a mathematical expression E is an expression Eq
which is parameterized by a variable q, and whose limit as q approaches to 1 isE.
EXAMPLE1.2.0.6. Let [n]qbe1−q n 1−q. Since lim q→1 1− qn 1− q = n,
[n]qcan be regarded as a q-analogue of the number n. For an n-dimensional vector space
overFq, the number of projective points (one-dimensional subspaces) in it is 1−q
n
1−q. We
consider it as a q-analogue of the set of size n.
EXAMPLE1.2.0.7. There is a q-analogue of n! (n-factorial) which we will call [n]|q. Since
limq→1[1]q[2]q[3]q. . . [n]q= n! we can define this q-analogue of n! as follows:
[n]|q:= [1]q[2]q[3]q. . . [n]q.
Conversely, one can consider the combinatorial case of q = 1 as a limit of q-analogs
as q→ 1.
1.3
Groups
A group G is called a simple group if its only normal subgroups are the trivial group
and the group itself, where a subgroup H of G is a normal subgroup if for any g∈ G,
its left coset gN and right coset N g are equal as sets. Finite simple groups are the building blocks of all finite groups. By Jordan-Hölder Theorem, any finite group G can be broken down into uniquely determined simple components which are the factors of a composition series for G [70], [73], [74].
Since, the groups of our interest for this thesis are examples of those, we will give the following theorem, which gives the complete classification of finite simple groups.
THEOREM1.3.1. [4] Every finite simple group is one of 26 sporadic simple groups or is an
example (up to isomorphism) of at least one of the following four categories: • cyclic groups of prime order,
• alternating groups of degree at least 5,
• simple groups of Lie type, including both the classical Lie groups, namely the groups of projective special linear, unitary, symplectic, or orthogonal transformations over a finite field, and the exceptional and twisted groups of Lie type (including the Tits group).
In this thesis we will mostly be concerned with the simple groups of Lie type - see also Section 1.3.3.
1.3 GROUPS 9
1.3.1
Coxeter groups
DEFINITION1.3.1.1. A Coxeter group is a group W together with a set R of generators
of W, such that a presentation of W is given by:
〈r ∈ R | r2= 1, (rs)mrs= 1 for all r, s∈ R 〉
where mrs is the order of rs in W . The matrix M :={mi j}1≤i, j≤nis called the Coxeter
matrix. The pair (W, R) is also called a Coxeter system.
DEFINITION 1.3.1.2. The Coxeter-Dynkin diagram (Coxeter diagram) of a Coxeter
sys-tem (W, R) is a labeled graph with vertex set R, where two vertices r, s are joined by an
edge labeled mrs. Edges labeled 2 are omitted. Labels 3 are omitted, and when mi j= 4,
there is a double edge.
This diagram determines W uniquely.
A Coxeter group is of spherical type if the group is finite. The possible connected components of the Coxeter diagrams of Coxeter groups of spherical type are given in Fig-ure 1.3. A Coxeter system is irreducible if its diagram is connected. The Coxeter group of a disconnected diagram is the direct product of the Coxeter groups of its components.
EXAMPLE 1.3.1.3. The Dihedral group D2m is an example of the Coxeter groups. Its
diagram is called I2(m). Its Coxeter group presentation is the following:
D2m:=〈s, t|s2= t2= (st)m= 1〉.
DEFINITION1.3.1.4. For a Coxeter system (W, R) for any element w∈ W, the length of w,
denoted by l(w) is the length of the shortest expression of w as a product of factors in R. In case W is finite, which is the case of our interest, it has a unique longest element
w0. This element satisfies l(w0w) = l(ww0) = l(w0)− l(w) for w ∈ W ([18], Chapter IV,
Exercise 22). It follows that w0has order (at most) 2 and that conjugation by w0induces
a graph automorphism on the Coxeter diagram.
The following theorem classifies all finite Coxeter groups.
THEOREM 1.3.2. [48] Finite Coxeter groups are those whose diagrams are given by the
disjoint unions of the diagrams of An, Cn, Dn, E6, E7, E8, F4, H3, H4, I (m)
2 .
Root systems
Let the Euclidean inner product be denoted by (·, ·).
DEFINITION 1.3.1.5. A reduced root system is a finite collection Φ of non-zero vectors
spanningRlfor some l≥ 1 such that:
i. ifα∈ Φ then Rα ∩ Φ = {α, −α}, ii. ifα, β∈ Φ then 2(β,α) (α,α) ∈ Z, and iii. ifα, β∈ Φ then wα(β ) := β− 2(β,α) (α,α)α∈ Φ.
FIGURE1.3: the diagrams of the irreducible finite Coxeter systems
Here, wα’s are the reflections in the plane perpendicular toα.
DEFINITION1.3.1.6. A subcollection ∆ of a root system Φ is called a fundamental system
of roots when it has the properties:
i . the collection ∆ is a basis ofRl,
ii . each root when written as a linear combination of vectors in ∆, has either only nonnegative or nonpositive coefficients.
Each root system has a fundamental system of roots I . Given a root system Φ one can
choose a set of positive roots. This is a subset Φ+of Φ such that:
i . for each rootα∈ Φ exactly one of the roots α, −α is contained in Φ+,
ii . for any two distinct rootsα, β∈ Φ+such thatα + β is a root, α + β is a root in Φ+.
1.3 GROUPS 11
a fundamental system I , one has Φ+, the set of roots with positive coefficients, given Φ+,
one has I , the set whose elements cannot be written as the sum of two elements of Φ+.
Let R ={wα|α ∈ Φ+}. The group W := 〈wα|α ∈ Φ〉 is called the Weyl group of the root
system and (W, R) is a Coxeter system [21] (p. 310). For a subset J of a fundamental
system of roots, letRJ be the space spanned by J , ΦJ= Φ∩ RJ, and WJ be the subgroup
of W generated by the fundamental reflections wr for r∈ J.
PROPOSITION 1.3.3. [36] The set ΦJ is a system of roots inRJ, J is a fundamental system
in ΦJ. The Weyl group of Φ is WJ.
DEFINITION1.3.1.7. The subgroups WJ and their conjugates are called the parabolic
sub-groups of W .
Most of the finite Coxeter systems can be described in terms of their root systems.
The exceptions are the groups called H3and H4, and the family I2(m) for m6= 2, 3, 4, 6
[21], (p. 310).
EXAMPLE1.3.1.8. For the diagram A2, we can name the positive roots as r, s, r + s and
the negative roots as−r, −s, −rs.
The corresponding Weyl group is: {1, wr, ws, wrws, wswr, wrwswr = wswrws} =
Sym(3), 1 = 1, wr = (12), ws = (23), wrws = (132), wswr, = (123), wrwswr =
(12)(23)(12) = (13).
In Figure 1.4, the generators are the reflections that are perpendicular to r and s.
wr
ws
FIGURE1.4: for A2, positive roots r, s, and r + s
In Chapter 7, we will define root system graphs and work on some examples of those.
1.3.2
Groups with a Tits system
DEFINITION1.3.2.1. A group G is called a group with a Tits system (B, N , W, R) , if
i . the Group G has fixed subgroups B and N that generate it,
ii . the group H := B∩ N is normal in N, W := N/H,
iv . if ni ∈ N maps to wi under natural homomorphism of N into W , and if n is any element of N , then
BniB.BnB⊆ BninB∪ BnB,
v . if niis as above, then niBni6= B.
The group W is called the Coxeter group of the Tits system.
The Coxeter diagram of G is the Coxeter diagram of the Coxeter group (W, R). If for
a Tits system, the related Coxeter system has diagram Xn, the Tits system is said to be of
type Xn. The rank of G is the size of R.
All Chevalley groups are groups with Tits systems (see [36] Proposition 8.2.1, Theo-rem 13.5.4.), and every finite group with a Tits system of rank at least 3 is a Chevalley group [97].
The subgroup B is sometimes called the Borel subgroup, H is sometimes called the
Cartan subgroup, and W is called the Weyl group.
The following proposition describes the subgroups of G.
PROPOSITION 1.3.4. [36] Let G be a group with a (B, N )-pair. Then for each subset J of I ,
let NJ be the subgroup of N that is mapped to WJ under the natural homomorphism. Then,
BNJB is a subgroup of G.
The following proposition describes the relation between the subgroups of G and the elements of W .
PROPOSITION 1.3.5. [36] Let G be a group with a (B, N )-pair. Let n, n′ be elements of N .
Then BnB = Bn′B if and only if n, n′ are mapped to the same element of W under natural homomorphism from N into W . Thus, there is a one to one correspondence between the double cosets of B in G and the elements of W .
From now on, we use the notation BwB for any double coset BnB where n∈ N, when
wis the image of n under the homomorphism described above.
THEOREM1.3.6. [21] [36] Let G be a group with Tits system (B, N , W, R) and let I , J⊆ R.
Then:
i . the pair (W, R) is a Coxeter system, ii . G = BN B,
iii . the set GJ = BWJB is a subgroup of G. Conversely, any subgroup of G containing B is of this form,
iv . each subgroup GJ is equal to its normalizer, hence, distinct subgroups GI, GJ can not be conjugate in G,
v . the subgroup GJ for distinct subsets J of I are all distinct, furthermore, GI∩GJ= GI∩J. Thus, the subgroups GJ form a lattice isomorphic to the lattice of all subsets of I , vi . the map WIwWJ 7→ GIwGJ is a bijection from WI\W/WJonto GI\G/GJ, in particular,
if for w, w′∈ W, we have BwB = Bw′B, then w = w′,
1.3 GROUPS 13
viii . if g B g−1⊆ GI, then g∈ GI (g∈ G), ix . GIwGJ = BWIwWJB (w∈ W).
REMARK1.3.7. [21] W = N/(B∩ N), and R is uniquely determined by R = {r ∈ W\{1} |
B∪ BrB is a group} so that, instead of giving the Tits system (B, N, W, R), it suffices to
give the (B, N )-pair (B, N ). The groups with Tits systems are also called groups with (B, N )-pairs.
Let G be a group with a (B, N )-pair and Weyl group W . The Bruhat decomposition of G:
G =[· BwB
is a decomposition of G as a disjoint union double cosets of the form BwB, where w is in
W (see [18], Chapter IV).
A subgroup that contains B is called a standard parabolic subgroup, that is, it is of
the form GI for some I⊂ R. There is a bijection between the set of subsets of R, and the
set of standard parabolic subgroups of G. Hence, for a Chevalley group with diagram Xn,
there are 2nsuch standard parabolic subgroups. Any conjugate of a standard parabolic
subgroup is called a parabolic subgroup.
1.3.3
Finite simple groups of Lie type
Finite simple groups of Lie type are subgroups of the group of invertible matrices over a
finite field namely G L(n, q), hence G≤ GL(n, q) for some n ≥ 2 and prime power q.
They are classified as the classical groups (projective special linear groups,
or-thogonal groups, symplectic groups, unitary groups), and exceptional groups and
twisted groups. The non-twisted groups are also called Chevalley groups.
Projective special linear groups have diagram An, orthogonal groups O2n+(q) have
di-agram Dn, O2n+1(q) have diagram Bn, symplectic groups ) S pn(q) have diagram Cn, and
the exceptional groups E6, E7, E8, F4, G2are named after their diagrams.
The groups2An,2Dn,2E6,3D4, are called Steinberg groups. The groups2B2(22n+1),
2F
4(22n+1), and2G2(32n+1) are known as Suzuki-Ree groups. Together, these are called
twisted Chevalley groups. The group2F4(2) is not simple. Its derived group2F4(2)′is
known as the Tits group.
History
For more details about the history of the classification of finite simple groups, see[90]. Our aim here, is to have a short look at the story behind the construction of the mathe-matical objects of our interest, namely, buildings. Groups with (B, N )-pairs are groups of automorphisms of these geometries.
Since finite simple groups are the buildings blocks of finite groups, understanding and classifying them is a good start for understanding all finite groups.
Let us have a brief look at the time line. In 1832 Galois introduced the concept of
a normal subgroup and found the simple groups An(n = 5) and PS L2(Fp), (p≥ 5). In
1861 Mathieu found the first two Mathieu groups M11, M12, the first sporadic simple
groups. In 1870, Jordan began to build a database for finite simple groups. He listed the alternating and projective special groups as simple groups. In 1873 Mathieu found three
more Mathieu groups M22, M23, M24. In 1892, Hölder underlined the problem of giving
an overview of all finite simple groups. In the same year, Cole determined all finite simple groups up to order 500 except a few cases. In 1905, Dickson introduced the simple groups
of type G2 over finite fields. Chevalley introduced Chevalley groups in 1955. Steinberg
in 1959, Suzuki in 1960, Ree in 1961 introduced, respectively, Steinberg, Suzuki and Ree groups. Tits introduced (B, N )-pairs for groups of Lie type and found the Tits group in
1964. Janko constructed the Janko group J1in 1965, which is a sporadic group. In 1968,
Higman and Sims constructed the Higman-Sims group. Same year, Conway discovered
the three Conway groups. In 1969, the Suzuki sporadic group, the Janko group J2, the
Janko group J3, the McLaughlin group, and the Held group were found. Fischer found the
three Fischer groups in 1971. In 1972, Lyons found the Lyons group and in 1973 Rudvalis found the Rudvalis group. In 1973, Fisher discovered the baby monster group and used this with Griess to discover the Monster group. As a result, Thompson discovered the Thompson sporadic group and Norton discovered the Harada-Norton group. In 1974, Tits showed that groups with (B, N )-pairs of rank at least 3 are groups of Lie type. In 1976, O’Nan introduced the O’Nan group, and Janko introduced the last sporadic group,
the Janko group J4. Frobenius, Dedekind, Burnside, Moore, Disckson, Killing, Cartan,
Hall, Schur, Zassenhaus, Miller, Fitting, Grün, Wielandt, Suzuki, Chevalley, Steinberg, Ree, Gorenstein, Lyons, Aschbacher, Smith, Solomon are some of the many people who contributed to the process of classification and construction of finite simple groups.
The classification process mostly took place in between 1955 and 1983. The proof is believed to be finished as of 2004 [3], and from then on, people focused more on understanding and improving the proof.
1.4
Geometric objects
In this thesis, we deal with some graphs related to finite simple groups of Lie type. To understand these groups better, Tits [97] came up with some geometrical objects called buildings. Buildings will be defined in this section. We will define some other geomet-rical objects in this section to be able to talk about buildings. Groups with (B, N )-pairs are groups of automorphisms of buildings. Similarly, Weyl groups are groups of auto-morphisms of subcomplexes called apartments. See [97], or [36](Chapter 15), for more detail. The content of the following three sections is based on this chapter.
For finite geometries in general, we refer to [33].
1.5
Buekenhout-Tits geometries
Most of the definitions you will see in this section will be used in Chapter 7. Much more about incidence geometries can be found at [30].
DEFINITION1.5.0.1. A Buekenhout-Tits geometry (or an incidence geometry) Γ(X ,∗, t)
with type set I is a set X of objects together with a map t : X → I that assigns a type
to each object, and a symmetric relation∗ called incidence such that two objects of the
same type are never incident. A geometry can be considered as a|I|-partite graph Γ over
X, where the incident pairs are adjacent.
1.5 BUEKENHOUT-TITS GEOMETRIES 15
DEFINITION1.5.0.2. A flag is a subset of X where any pair of elements are incident. For a
flag F , the image J of F under t is the type of F . A chamber is a maximal flag of type I . Γ is thick if every non-maximal flag is contained in at least 3 chambers of Γ. The
rank of a geometry is the size of the type set I .
For any i∈ I, when the number of chambers in Γ containing a given flag of type I\{i}
is 2, the geometry is thin.
A morphism from Γ to Γ′ is a map α from X to X′, for which incidence of a pair
of elements implies the incidence of their images, moreover when a pair of elements are of the same type, so are their images. An isomorphism of two Buekenhout-Tits geometries is a bijection on their sets of objects, which preserves incidence and types. An
automorphism is an isomorphism on the geometry.
The rank (corank) of a flag F is|F| (|I\t(F)|).
DEFINITION1.5.0.3. If F is a flag of Γ, then the residue ΓF of F in Γ is the geometry over
I\t(F), whose elements are all elements which are incident to all elements of the flag F.
In ΓF, the incidence relation and the type map are inherited from the geometry Γ(X ,∗, t).
The incidence graph of Γ(X ,∗, t) is an I-partite graph on X , where two vertices are
adjacent when they are incident.
A geometry is connected when its incidence graph is connected. A geometry is
residually connected when the residue of every flag of co-rank at least 2 is connected,
and residue of every flag of corank 1 is non-empty.
For an element x of X , the set consisting of x and all the elements incident with x is
denoted by x∗. For a subset Y of X , Y∗=∩x∈Yx∗. Using this notation, the set of objects
of ΓF is F∗\F.
DEFINITION1.5.0.4. If F is a flag of Γ(X ,∗, t), and J is a subset of I, the J-shadow ShJ(F )
(on P) is t−1(J )∩ F∗, i.e. the set of all J -flags that are incident with F . The J -space is
the set of all J -flags equipped with all J -shadows of all possible flags in Γ.
Point-line geometries and chamber complexes are examples of Buekenhout-tits ge-ometries.
1.5.1
Point-line geometries
DEFINITION1.5.1.1. A point-line geometry (P, L,∗) is an incidence system (or geometry)
of rank two, consisting of a set of points denoted by P, a set of lines denoted by L and an
incidence relation∗ ⊆ (P × L)∪(L×P) which is symmetric, (so if a point is on a line, then
the line is passing through this point and vice versa) and moreover, any pair of points in
Plie together on at most one line of the geometry. An incidence system is linear if any
pair of distinct points in the system are on exactly one line.
If two points are on the same line, they are collinear, if two lines have a point in common, they intersect, if not, they are skew.
If two points are on the same line, they are collinear, if two lines have a point in common, they intersect, if not, they are skew.
By definition, the length of shortest cycle (girth) of the incidence graph of a linear point-line geometry can not be 4. So it is at least 6.
DEFINITION 1.5.1.2. A geometry of rank two with types called points and lines is called
a generalized m-gon for m ≥ 2 if the incidence graph has diameter m and girth 2m.
A generalized polygon is called regular of order (s, t) for s, t cardinal numbers finite or infinite if for any point, there are t + 1 lines passing through it and for any line, there are
s +1 points on it.
THEOREM1.5.1. [55] For a regular generalized m-gon, for s, t > 1, both finite, m∈ {2, 3,
4, 6, 8}.
Usually, lines will have at least two points, and in such a situation the lines can be identified with the set of points that are incident with them, and considered as subsets of
P. A line is called thick if there are at least three points on it, a line is called thin if there
are exactly two points on it.
DEFINITION1.5.1.3. For a point-line geometry, the collinearity graph is the graph whose
vertices are the points of the geometry. In this graph any two vertices are adjacent if and only if the corresponding pair of points is collinear.
1.5.2
Chamber complexes and foldings
We take a set Θ with a partial order⊆ on it.
DEFINITION1.5.2.1. The set Θ is a simplex if it is isomorphic to the set of all subsets of
some set, partially ordered by inclusion.
For n (0 ≤ n ≤ ∞), take a (n + 1)-set S. Call the collection of all its subsets S .
Elements of S are the vertices, ordered pairs are edges, and subsets of S are the faces
of the n-dimensional simplex (or n-simplex )S . Any set that is isomorphic to a power
set of an (n + 1)-set with a partial order⊆ defined on them is called an n-dimensional
simplex.
FIGURE1.5: a simplicial 3-complex
DEFINITION1.5.2.2. The set Θ is called a (simplicial) complex if for any A∈ Θ, the set of
elements B such that B⊆ A forms a simplex, and for any A, B in Θ, they have a greatest
lower bound A∩ B.
1.5 BUEKENHOUT-TITS GEOMETRIES 17
DEFINITION 1.5.2.3. For any A∈ Θ, its rank is the number of elements B minimal with
respect to properties B⊆ A, B 6= 0.
Thus, the set of elements B ⊆ A is isomorphic to the set of subsets of a set with
cardinality rank A. The rank of A is denoted by rk A.
For a simplex A in Θ, the set of simplices containing A is called star of A, S t(A). Star of A is a complex itself with the order induced from the original complex, but not a subcomplex of it, unless A is 0. For any element B in S t(A), its rank there is the
codimension of A in B, which is denoted by cod imBA.
DEFINITION1.5.2.4. A complex is called a chamber complex if every element is contained
in an element of maximal dimension called a chamber, and if given two chambers C, C′,
there exists a sequence of chambers C = C0, C1, . . . , Cm= C such that
cod imCi−1(Ci−1∩ Ci) = cod imCi(Ci−1∩ Ci)≤ 1 for all i = 1, 2, . . . , m.
Such a sequence is called a gallery. Two chambers C, C′ are adjacent if cod im(C∩
C′) = 1. So, a gallery is a sequence of chambers where any pair of consecutive chambers
are either adjacent or identical.
DEFINITION1.5.2.5. In a chamber complex, a set L of chambers is called convex , if every
minimal gallery whose extremities belong to L has all its terms in L.
DEFINITION1.5.2.6. A chamber complex is called thin when each simplex of codimension
1 in any chamber is contained in exactly two chambers. A chamber complex is called
thick when each simplex of codimension 1 in any chamber is contained in at least three
chambers.
Let Θ′be a chamber complex.
DEFINITION1.5.2.7. A mapα : Θ→ Θ′ is a morphism of chamber complexes ifα(C) is
a chamber of Θ′for any chamber of C of Θ, and for each chamber C of Θ,α induces an
isomorphism on the simplices defined by C,α(C).
LEMMA1.5.2. [36] Let Θ, Θ′be two chamber complexes in which each element of
codimen-sion 1 is contained in at most 2 chambers, and letα, β be two morphisms of Θ into Θ′ in-jective on the set of chambers. Suppose there exists a chamber C∈ Θ such that α(A) = β(A) for all A⊆ C. Then α = β.
DEFINITION1.5.2.8. For a thin chamber complex Θ, an endomorphismα on it is called a
folding if it satisfies:
i . α2=α,
ii . Each chamber inα(Θ) is the image of exactly two chambers of Θ.
DEFINITION 1.5.2.9. A set of chambers in a chamber complex is called convex if every
PROPOSITION 1.5.3. [36] Letα be a folding of Θ and C, C′be a pair of adjacent chambers of Θ such thatα(C′) = C. Suppose there is a foldingβ such that β (C) = C′. Then,β has the same property for any other adjacent chambers of this type, if D, D′ are adjacent and
α(D′) = D, thenβ (D) = D′.
The folding β of 1.5.3, which, if exists, is determined uniquely by α, is called the
opposite folding ofα.
PROPOSITION 1.5.4. [36] Let α, β be opposite foldings of Θ. Then there exists an
auto-morphism ρ of Θ which coincides with β on α(Θ) and with α on β (Θ). Also ρ2 is the
identity.
Abstract Coxeter complexes
DEFINITION 1.5.2.10. An abstract Coxeter complex is a thin chamber complex Σ such
that given any pair C, C′of adjacent chambers, there exists a foldingα of Σ with α(C′) =
C.
A retraction will be called a retraction.
PROPOSITION 1.5.5. [36] Let Σ be an abstract Coxeter complex, C a chamber in Σ, and
S(C) the simplex of all faces of C. Then there exists a unique idempotent morphism of Σ onto S(C).
Let us call this mapρC, the retraction of Σ onto the simplex of faces of C.
The following relations is an equivalence relation: given A, B∈ Σ, we write A s B if
ρC(A) =ρC(B).
LEMMA1.5.6. [36] The equivalence relation defined on Σ is independent of the chamber.
The equivalent elements of Σ have the same type. Each chamber has one face of each type.
LEMMA1.5.7. [36] Let Σ be an abstract Coxeter complex and let C be a chamber of Σ. Let
γ be an endomorphism of Σ leaving invariant the type of each face of C. Then, γ preserves the type of each element of Σ.
Endomorphisms and automorphisms of the kind discussed in 1.5.7 are called
type-preserving.
LEMMA1.5.8. [36] The automorphismsρ defined in 1.5.4 are type preserving.
Let W (Σ) be the group generated by all the automorphisms of Σ of the kind described in 1.5.4.
PROPOSITION1.5.9. [36] Let Σ be an abstract Coxeter complex. Then, W (Σ) is the group of
all type-preserving automorphisms of Σ.
THEOREM1.5.10. [36] Let Σ be an abstract Coxeter complex and let C be a chamber of Σ.
Then the reflections in the faces of codimension 1 in C generate the group W (Σ). Moreover, W (Σ) is a Coxeter group with respect to these generators.
1.5 BUEKENHOUT-TITS GEOMETRIES 19
1.5.3
Buildings
Jacques Tits introduced the geometries called buildings as simplicial complexes with a family of subcomplexes.
DEFINITION1.5.3.1. A building is a pair Ω,A where Ω is a chamber complex, and A is a
set of subcomplexes, called apartments, satisfying the following axioms: i . The chamber complex Ω is thick.
ii . The apartments of Ω are thin chamber complexes.
iii . Given any two chambers C, C′ in Ω, there exists an apartment Σ∈ A such that
C∈ Σ and C′∈ Σ.
iv . If A, A′ are elements of Ω which are contained in each of the apartments Σ, Σ′ ∈
A , there exists an isomorphism between Σ, Σ′leaving invariant A, A′and all their
faces.
It follows that any two apartments of a building Ω are isomorphic. Let G be a group with a (B, N )-pair as described in Section 1.3.2.
EXAMPLE1.5.3.2. [36] Let Ω be the set of left cosets gGJ for all g∈ G, and all subsets J
of I . Let Σ0be the subset of Ω consisting of all elements nGJ for all n∈ N and all J ⊂ I.
Then gΣ0 is the set of all cosets gnGJ for all n∈ N and all J. Let A be the family of
subsets gΣ0of Ω for all g∈ G. Then, Ω is a building and, A is a set of apartments in Ω.
The building constructed in this way will be called Ω(G; B, N ).
We pick a building, an apartment of it, and a chamber of that apartment. We call
them respectively Ω, Σ, and C. For each element A∈ Ω there exists an apartment Σ′
containing A and C, by axiom (iii) in Definition 1.5.3.1. By (i v), there is an
isomor-phism Σ′→ Σ which leaves invariant all faces of C. By Lemma 1.5.2, there is only one
such isomorphism. The image of A under this isomorphism is an element of Σ, which is
independent of the choice of Σ′, by (i v). This image will be called retΣ,C(A).
LEMMA1.5.11. [36] The map
Ω→ Σ,
A→ retrσ,C(A)
is a retraction from Ω onto Σ.
THEOREM1.5.12. [36] The apartments of a building are abstract simplicial complexes.
DEFINITION1.5.3.3. Let A, A′be two elements of a building Ω. Then A, A′are said to have
the same type in Ω if they have the same type in any apartment containing A, A′.
The following theorems are converses of each other, relating Ω(G; B, N ) and G.
THEOREM1.5.13. [36] In the building Ω(G; B, N ) the group G operates by left
multiplica-tion as a group of type preserving automorphisms, which is transitive on the pairs (C, Σ) where C is a chamber and Σ is an apartment containing C.
THEOREM1.5.14. [36] Let (Ω,A ) be a building and G a group of type preserving automor-phisms of Ω which is transitive on the pairs (C, Σ), with C∈ Σ, where C is a chamber and
Σ is an apartment of Ω. Let C0, Σ0be a fixed chamber and apartment with C0∈ Σ0, let B
be the stabilizer of C0 in G and N be the stabilizer of Σ0. Then the subgroups B, N form
a (B, N )-pair in G. Moreover, W = N/(B∩ N) is isomorphic to the group W(Σ) of type
preserving automorphisms of each apartment Σ of Ω.
A building is called spherical if its apartments are finite. Buildings of spherical type are the ones of our interest.
The type of the building is that of the Coxeter group. The types of spherical buildings
are the types of finite irreducible Coxeter groups. The thick buildings of type Anare the
projective spaces. The buildings of type Bn, Cnand Dnare polar spaces.
Tits proved that a finite building whose associated Coxeter group is an indecompos-able Weyl group of rank at least 3 must be a building Ω(G; B, N ) where G is a finite Chevalley group of twisted group, and deduced that the only finite simple groups with a (B, N )-pair of rank at least 3 are the finite Chevalley groups and twisted groups which have this property [97].
In his famous lecture notes, Tits [97] phrased these under two main problems: “(A) Determination of the buildings of rank ≥ 3 and irreducible, spherical type, other than H3and H4. Roughly speaking, those buildings all turn out to be associated to simple algebraic or classical groups. An easy application provides the enumeration of all finite groups with (B, N )-pairs of irreducible type and rank≥ 3 up to normal subgroups contained in B.
(B) Determination of all isomorphisms between buildings of rank≥ 3 and spherical type,
associated with algebraic or classical simple groups, and in particular, the determination of the full automorphism group of these buildings."
More information about buildings can be found at, [97], [33], [36], [27], [86], and [103].
Buekenhout-Tits diagrams
DEFINITION1.5.3.4. A diagram for a geometry Γ is a labeled complete graphD on the
type set, where each label Di j is a class of rank 2 geometries. The graphD is called a
Buekenhout-Tits diagram Xnof the geometry Γ (with rank n) when for any flag of type
I\{i, j}, the residue ΓF belongs to the class of geometriesDi j. Sometimes, Γ is called a
geometry of type Xn.
Our classes of geometries will be self-dual, so that we need not worry about directing
the edges of D. Indeed, each class Di j will be the class of all generalized m-gons, for
some fixed m = mi j. Now the edge i j will be labeled by the number m.
Conventionally, edges labeled 2 are omitted. In that case, every i object is incident with every j object. The label 3 is omitted. Here, the i and j objects form the points and lines of a projective plane. Instead of a label 4 one draws a double edge. Here, the i and
jobjects form the points and lines of a generalized quadrangle.
EXAMPLE1.5.3.5. The geometry of 8 corners, 12 edges and 6 faces of a cube satisfies the
1.5 BUEKENHOUT-TITS GEOMETRIES 21
1 2 3
FIGURE1.6: the diagram of the cube
Diagram Xn,i is diagram Xn with type i circled, i.e., selected as the point type, and
here n indicates the rank, i.e. the number of nodes.
For each choice of a type i0in I , a geometry Γ(X ,∗, t) with type set I gives a point-line
geometry where P = t−1(i0) is the set of objects of type i0, and L is the set of i0-shadows
of flags which are of cotype i0in Γ. When we say l is a line, we consider it as the set of
points that it is incident with it. For instance, consider the diagram of the cube (Figure
1.6 ). Let the set of points be t−1(1). The lines of the geometry are the 1-shadows of
{2, 3}-flags. Here a line is a set of two points.
DEFINITION 1.5.3.6. The point-line geometry obtained from a geometry of type Xn by
choosing type i as the point type described as above is called the incidence system of
type Xn,i, and this choice is indicated in the diagram by circling the node representing
the point type. The incidence system of type Xn,i, associated with a thick building of type
Xnis called a Lie incidence system of type Xn,i (or Xn,i(K) where K is the underlying
field).
1.5.4
Projective spaces
Here we introduce projective geometries. They are examples of incidence geometries. More detailed information about these objects can be found in [32] and in [68].
Let V be an (n + 1)-dimensional left vector space over a fieldF.
DEFINITION1.5.4.1. A projective space P G(n,F) is an incidence geometry (X ,∗, t) with
type set I of rank n. The set X consist of subspaces of V except the trivial and full
subspace,∗ is symmetrized strict inclusion, I = {1, 2, . . . , n} and t maps every element of
X to its vector space dimension.
We are interested in finite projective spaces, overFq. For any finite prime power q,
for the unique fieldFqwe denote the corresponding projective space by P G(n, q).
Rank 1 or vector space dimension 1 subspaces are called points, and vector space dimension 2 subspaces are called lines. Projective dimension is one less than vector space dimension.
The vectors of V (n + 1,F)\{0} are called equivalent if and only if x = k y for some
k ∈ F\{0}. So the point set of PG(n, F) is the set of all equivalence classes under this
relation. Let P(x) be the equivalence class of a point x. Then, P(x1), . . . , P(xk) are
linearly independent in the projective space if the corresponding vectors are linearly
independent in the vector space. A subspace of P G(n,F) of dimension r is a set of all
points, whose corresponding vectors form a (r + 1)-dimensional subspace of V (n + 1, q) together with 0.
We will give an axiomatic definition of projective space and the projective planes, which are projective spaces of dimension 2, here.
A projective space is a point-line geometry S = (P, L,∗) that satisfies the following