The Limit Points of the Commuting Probability Function on Finite Groups

Total Page:16

File Type:pdf, Size:1020Kb

The Limit Points of the Commuting Probability Function on Finite Groups The Limit Points of The Commuting Probability Function on Finite Groups Madeleine Whybrow, CID 00640867 Supervisor: Dr John Britnell June 2014 This is my own unaided work unless stated otherwise. Abstract If G is a finite group, then Pr(G) denotes the probability that two elements of G, chosen at random, commute. Joseph [15] has conjectured that if l is a limit point of the function Pr on finite groups then l is rational and is not approached from below. Hegarty [12] showed that this is true if l 2 (2=9; 1]. We discuss the various steps in Hegarty's proof, as well as suggesting ways to extend his work and prove Joseph's conjectures over a larger range. Acknowledgements I would like to thank my supervisor Dr John Britnell for his support and en- couragement. His dedication and expertise has made the writing of this project an enjoyable and educational experience. Contents 1 Introduction 2 2 Preliminaries 5 2.1 Commutators and the Commutator Subgroup . 5 2.2 Restrictions on the Commutator Subgroup . 6 2.2.1 Complete Groups . 6 2.2.2 Dihedral Groups . 8 2.2.3 Restrictions on Z(G)..................... 10 2.3 Nilpotent Groups . 10 2.4 Isoclinic Groups . 11 3 Key Results on the Commuting Probability 15 3.1 Basic Results . 15 3.2 The Commuting Probability and Character Degrees . 18 3.3 Commuting Probability of Semigroups . 20 4 Background 22 4.1 Bounds on The Commuting Probability . 22 4.2 Classification Results . 24 4.2.1 Classification of Groups with Pr(G) > 11=32 . 24 4.2.2 Classification of Groups of Odd Order with Pr(G) > 11=75 29 5 Joseph's Conjectures 33 5.1 Hegarty's Work . 33 5.1.1 Groups with jG0j < 8..................... 33 5.1.2 The Main Proof . 35 5.2 Further Work . 37 5.2.1 Groups with jG0j = 8. 37 5.2.2 A General Formula for the Case jG0j ≤ n. 38 6 Concluding Remarks 41 1 Chapter 1 Introduction Let G be the set of all finite groups then, for G 2 G, we define Pr(G) to be the probability that two elements selected at random (with replacement) from G commute with each other. That is to say jf(x; y) 2 G × G : xy = yxgj Pr(G) = : jGj2 The commuting probability of a group can be thought of as a measure of how commutative a group is. In fact, some sources call Pr(G) the commutativity degree of G to draw attention to this. Many of the results on the probabilty take the approach of linking this numerical measure of commutativity with some structural indicator such as the size of the commutator subgroup or the nilpotency class of the group. During the second half of the twentieth century, there was a spate of work studying the commuting probability. The majority concentrated either on bound- ing Pr for certain groups or on classifying groups with certain values of Pr. We give an overview of the most important results in Chapter 4. In this paper, we will take a slightly different approach, treating Pr as a function from G to (0; 1]. In particular, we will study the limit points1 of the image Pr(G). To our knowledge, this perspective had not been seriously considered until Keith Joseph [15] published a note in the American Mathematical Monthly proposing the following three conjectures: 1. Every limit point of Pr(G) is rational. 2. If l is a limit point of Pr(G), then there exists = l > 0 such that Pr(G) \ (l − l; l) = φ. 3. Pr(G) [ f0g is a closed subset of R. 1 For our purpose, we say that l 2 R is a limit point of a set S ⊆ R if there exists a sequence (sn) of elements of S such that sn ! l. 2 For brevity, we say that a subset S ⊆ R is a good set if, for every limit point l of S, the following hold: 1. l 2 Q 2. there exists = l such that S \ (l − l; l) = φ. Note also that any subset of a finite union of good sets is also good. In his orginal article, Joseph gives little mention of the intuition behind these conjectures. However, the following result gives some indication to the motivation behind the first two conjectures. Lemma 1.1 ([12]). Let n 2 N be fixed then ( n ) X 1 S := : x 2 n x i N i=1 i is a good set. So in order to prove Joseph's conjectues, it would be sufficient to show that G consists of a finite number of families of groups whose commuting probabilities are contained in Sn for some n fixed. There are a number of results (see for example Section 4.2) which show that certain families of groups do indeed have commuting probabilities of this form. Some would have been available at the time when Joseph published his conjectures and we would guess that it is these which motivated his conjectures. Joseph's conjectures were left untreated, as far as we are aware, until recently Hegarty [12] proved that the first two conjectures hold for limit points in the 2 range ( 9 ; 1]. He gives no attention to the third conjecture, save for commenting that it appears much more mysterious than the first two. His method relied on the following result, proved by Guralnick and Robinson [9] which links the commuting probability with the size of the commutator sub- group, G0 (see Section 2.1), and the smallest degree of an irreducible character of G. The proof can be found in Section 3.2. Theorem 1.2. If G is non-abelian, then 1 1 1 Pr(G) ≤ + 1 − d2 d2 jG0j where d is the smallest degree of a non-linear complex irreducible character of G. In fact, when d = 3 and jG0j = 8, 1 1 1 2 + 1 − = : d2 d2 jG0j 9 2 0 So if Pr(G) > 9 then either jG j < 8 or d = 2 (as the character is non-linear). The rest of the proof is then split between these cases. 3 We aim to improve his method, working towards proving Joseph's first two conjectures over a larger range. In particular, ( 1 1 1 17 ≈ 0:2099 if d = 3; jG0j = 9 + 1 − = 81 d2 d2 jG0j 23 0 128 ≈ 0:1797 if d = 4; jG j = 8: So, in order to prove the conjectures over a wider range, we must first prove that the conjectures are true when jG0j < 9. The case jG0j < 8 is treated by Hegarty but only in brief detail. Thus, a large part of our work involves describing explicitly his method. We then discuss possible methods to extend his work. 4 Chapter 2 Preliminaries 2.1 Commutators and the Commutator Subgroup Definition 2.1. For a group G, the commutator of two elements x and y in G is [x; y] = x−1y−1xy: Equally, if X and Y are two subsets of G then [X; Y ] = h[x; y]: x 2 X; y 2 Y i: In particular, the group [G; G] is denoted G0 and is called the commutator or derived subgroup. It is important to note that the product of two commutators is not necessarily a commutator, that is to say that the set of commutators may not be a group. Isaacs [14] constructs a source of examples of groups where this is the case. Coversely, the conjugate of a commutator by any element of G lies in G0 as [x; y]g = [xg; yg]: 0 This clearly shows that G E G. However, a stronger result is that, for φ a group endomorphism of G, we have φ([x; y]) = [φ(x); φ(y)]: That is to say, G0 is invariant under every endomorphism of G, or that it is a fully characteristic subgroup of G. This property is considerably stronger than normality. Another useful property of the commutator subgroup is its relationship to abelian quotient groups: Proposition 2.2. If G is a group and N E G then G=N is abelian if and only if G0 ⊆ N. 5 Proof. Suppose for contradiction that there exists N E G such that G=N is abelian and that there exists x; y 2 G such that [x; y] 2= N. Then, working in G=N we have [xN; yN] = xNyNx−1Ny−1N = [x; y]N 6= N: So [xN; yN] 6= eG=N and xN and yN do not commute. Thus, if G=N is abelian, all commutators, and by extension the commutator subgroup, lie in N. Conversely, if G0 ⊆ N then 8x; y 2 G,[x; y] 2 N. In particular, [xN; yN] = eG=N and so G=N is abelian. An immediate consequence of this result is that G0 is the unique minimum subgroup of G such that its quotient is abelian. As such, G=G0 is called the abelianisation of G. We also note the following result on the centraliser of the commutator sub- group: Lemma 2.3. For any group G 0 0 0 (CG(G )) ⊆ Z(G) ⊆ Z(CG(G )): Proof. This follows from the Hall-Witt identity: [x; [y−1; z]]y[y; [z−1; x]]z[z; [x−1; y]]x = e: 2.2 Restrictions on the Commutator Subgroup For a given group G, it is not always true that there exists a group H such that H0 =∼ G. In certain cases, the commutator subgroup will also limit the groups which Z(G) can be isomorphic to, as shown in Section 2.2.3.
Recommended publications
  • Contrasting the Commuting Probabilities of Groups and Rings
    Contrasting the commuting probabilities of groups and rings STEPHEN M. BUCKLEY AND DESMOND MACHALE Abstract. We contrast the set of commuting probabilities of all finite rings with the set of commuting probabilities of all finite groups. 1. Introduction Suppose F is an algebraic system of finite cardinality, closed with respect to a multiplication operation denoted by juxtaposition. We define the commuting probability of F to be jf(x; y) 2 F × F : xy = yxgj (1) Pr(F ) := jF j2 where jSj denotes cardinality of a set S. Much has been written on Pr(G) when G is a group, and its possible values: see for instance [9], [14], [12], [16], [22], [8], [11], and [7]. Less has been written on the corresponding concepts for rings and semigroups: the only papers on this topic of which we are aware are [17] and [5] for rings, and [18], [10], [1], [21], and [4] for semigroups. Before proceeding, we introduce a little notation. Let G, S, and R be the set of values of Pr(F ) as F ranges over all finite groups, finite semigroups, and (possibly non-unital) finite rings, respectively. For each prime p, we define Gp and Rp similarly, except that we are considering commuting probabilities of p- groups and p-rings, meaning finite groups and rings whose orders are a power of p. Trivially, G, S, and R are all subsets of (0; 1] \ Q. All values in G \ (11=32; 1] were explicitly listed in [22], while all values in R\[11=32; 1] were explicitly listed in [5].
    [Show full text]
  • On the Probability That a Group Satisfies a Law; a Survey
    ON THE PROBABILITY THAT A GROUP SATISFIES A LAW; A SURVEY M. FARROKHI D. G. Abstract. The aim of this paper is to give a survey of old and new results on probabilities on finite groups arising from words. 1. Introduction The study of groups depends heavily on studying laws (word equations) on groups. The simplest example of this sort are abelian groups which admit the word equation [x; y] = 1, where [x; y] := x−1y−1xy. The next important families of groups defined by means of words are nilpotent and solvable groups, that is, a group is nilpotent or solvable if it satisfies the word equation un = 1 or vn = 1 for some n, respectively. The words un and vn are defined inductively as u2 = v2 := [x1; x2] 0 0 un := [un−1; xn] and vn = [vn−1; vn−1] for all n ≥ 3, where vn−1 and vn−1 denote the same words with disjoint set of variables. We enjoy to remind two further fam- ilies of groups arising from word equations, namely Engel groups which admits a word equation of the form [y; x; : : : ; x] = 1 (x appears n ≥ 1 times) and Burnside groups which admit a word equation of the form xn = 1 for some n ≥ 1. Let G be a group and w = w(x1; : : : ; xn) be a word. Then w is said to be a law for G if w(G) = 1, where w(G) = fw(g1; : : : ; gn): g1; : : : ; gn 2 Gg: The theory of groups is well developed in the past decades resulting in many tools and classification theorems which can be applied to describe finite groups.
    [Show full text]
  • The Probability That a Pair of Elements of a Finite Group Are Conjugate
    THE PROBABILITY THAT A PAIR OF ELEMENTS OF A FINITE GROUP ARE CONJUGATE SIMON R. BLACKBURN, JOHN R. BRITNELL, AND MARK WILDON Abstract. Let G be a finite group, and let κ(G) be the prob- ability that elements g, h 2 G are conjugate, when g and h are chosen independently and uniformly at random. The paper clas- sifies those groups G such that κ(G) ≥ 1=4, and shows that G is abelian whenever κ(G)jGj < 7=4. It is also shown that κ(G)jGj depends only on the isoclinism class of G. Specialising to the symmetric group Sn, the paper shows that 2 κ(Sn) ≤ C=n for an explicitly determined constant C. This bound leads to an elementary proof of a result of Flajolet et al, that 2 κ(Sn) ∼ A=n as n ! 1 for some constant A. The same tech- niques provide analogous results for ρ(Sn), the probability that two elements of the symmetric group have conjugates that commute. 1. Introduction Let G be a finite group. We define κ(G) to be the probability that two elements g, h 2 G are conjugate, when g and h are chosen in- dependently and uniformly at random from G. Let g1; g2; : : : ; gk be a complete set of representatives for the conjugacy classes of G. It is easy to see that k k 1 X X 1 (1) κ(G) = jgGj2 = jGj2 i j C (g )j2 i=1 i=1 G i where CG(g) denotes the centralizer of an element g 2 G.
    [Show full text]
  • Groups with Pr(G)=1/3 1. Introduction We Define the Commuting
    Groups with Pr(G) = 1=3 STEPHEN M. BUCKLEY AND DESMOND MACHALE Abstract. We find all isoclinism families of groups with commuting proba- bility 1=3. 1. Introduction We define the commuting probability of a finite group G to be jf(x; y) 2 G × G : xy = yxgj (1) Pr(G) := ; jGj2 where jSj denotes cardinality of a set S. Much has been written on this concept: see for instance [5], [13], [8], [16], [21], [14], [4], [7], [3], and [11]. In particular, Pr(G) is an isoclinism invariant [14, Lemma 2.4], so to understand which groups have a given commuting probability t, it suffices to find all isoclinism families with commuting probability t. If F is an isoclinism family containing a finite group G, we call F a finite family and define Pr(F) to be Pr(G). Lescot [14] found all families F with Pr(F) ≥ 1=2, and the results of Rusin [21] allow one to determine all F with Pr(F) > 11=32. In this paper, we find all finite families F for which Pr(F) = 1=3. Theorem 1. There are precisely three finite families F with Pr(F) = 1=3. Each has a unique stem group, namely the alternating group A4, the dihedral group D9, and the generalized dihedral group Dih(C3 × C3). We chose the value 1=3 above because it is an interesting cutoff value. Barry, MacHale, and N´ıSh´e[2] showed that a finite group G is supersolvable if Pr(G) > 1=3, and noted that this lower bound is minimal because the alternating group A4 is not supersolvable and satisfies Pr(A4) = 1=3.
    [Show full text]
  • Arxiv:2102.06983V2 [Math.GR] 1 May 2021
    On the commuting probability for subgroups of a finite group Eloisa Detomi and Pavel Shumyatsky Abstract. Let K be a subgroup of a finite group G. The prob- ability that an element of G commutes with an element of K is denoted by P r(K, G). Assume that P r(K, G) ≥ ǫ for some fixed ǫ> 0. We show that there is a normal subgroup T ≤ G and a sub- group B ≤ K such that the indexes [G : T ] and [K : B] and the order of the commutator subgroup [T,B] are ǫ-bounded. This ex- tends the well known theorem, due to P. M. Neumann, that covers the case where K = G. We deduce a number of corollaries of this result. A typical appli- cation is that if K is the generalized Fitting subgroup F ∗(G) then G has a class-2-nilpotent normal subgroup R such that both the index [G : R] and the order of the commutator subgroup [R, R] are ǫ-bounded. In the same spirit we consider the cases where K is a term of the lower central series of G, or a Sylow subgroup, etc. 1. Introduction The probability that two randomly chosen elements of a finite group G commute is given by |{(x, y) ∈ G × G : xy = yx}| arXiv:2102.06983v2 [math.GR] 1 May 2021 Pr(G)= . |G|2 The above number is called the commuting probability (or the commu- tativity degree) of G. This is a well studied concept. In the literature one can find publications dealing with problems on the set of possible values of Pr(G) and the influence of Pr(G) over the structure of G (see [9, 15, 17, 22, 23] and references therein).
    [Show full text]
  • Higher Commutativity and Nilpotency in Finite Groups
    e Bull. London Math. Soc. 44 (2012) 1259–1273 C 2012 London Mathematical Society doi:10.1112/blms/bds055 Higher commutativity and nilpotency in finite groups E. Torres-Giese Abstract Given a finite group and an integer q, we consider the poset of nilpotent subgroups of class less than q and its corresponding coset poset. These posets give rise to a family of finite Dirichlet series parameterized by the nilpotency class of the subgroups, which in turn reflect probabilistic and topological invariants determined by these subgroups. Connections of these series to filtrations of the classifying space of a group are discussed. 1. Introduction In 1936, Hall addressed to some extent the problem of computing the probability that a randomly chosen ordered s-tuple of elements in a finite group G generates the group. Hall proved that this probability, denoted by P (G, s), can be expressed as a finite Dirichlet series. For instance (see [3]), 14 8 21 28 56 84 P (PSL(2, 7),s)=1− − + + + − . 7s 8s 21s 28s 56s 84s Hall’s results allow us to think of this as a function over the complex numbers. Several properties of this series and its connection to the structure of the group have been studied; see [12] and the references contained therein. In [4], this function was related to the coset poset of a finite group. This poset is the set of all proper cosets of a group ordered by inclusion. If we denote this poset by C(G), then Brown (following Bouc) proved that P (G, −1) = 1 − χ(C(G)).
    [Show full text]
  • Commuting Probability and Simultaneous Conjugacy Classes of Commuting Tuples in a Group
    February 5, 2020 COMMUTING PROBABILITY AND SIMULTANEOUS CONJUGACY CLASSES OF COMMUTING TUPLES IN A GROUP UDAY BHASKAR SHARMA AND ANUPAM SINGH Abstract. Let G be a finite group. This expository article explores the subject of commuting probability in the group G and its relation with simultaneous conjugacy classes of commuting tuples in G. We also point out the relevance of this topic in context of topology where similar problems are studied when G is a Lie group. 1. Introduction Let G be a finite group. If G is not commutative then there are pairs of elements in G which do not commute. The property of non-commutativity of elements of G can be measured by commuting probability, which is defined by |{(x,y) ∈ G × G | xy = yx}| cp (G)= 2 |G|2 and more generally by cpn(G) (see Section 2 for definition). Erdös and Turan (see [ET]) took probabilistic approach to group theory in a series of papers titled “Statistical Group k(G) Theory”. Among several results they proved that cp2(G)= |G| where k(G) is the number of conjugacy classes in G. The commuting probabilities (under different names) have been studied for more general groups, for example, compact groups (see [AG, HR1, HR2, RE]). Following the work of Feit and Fine [FF] for general linear groups, this has been further investigated for varieties (see for example, see [BM]). There are various generalisations of commuting probability for groups and we refer to the survey article [DNP] for the same. Fulman and Guralnick studied this problem for Lie algebras over a finite field arXiv:2002.01253v1 [math.GR] 4 Feb 2020 in [FG].
    [Show full text]