Conjugacy

There is another equivalence relation defined on a group, which is very important. The concept of conjugate elements and conjugate classes in a group is basic to the study of normal subgroups.

7.5.1 Definition Let G be a group. We say that elements x, y of G are conjugate (written x ∼ y) if y = gβˆ’1xg for some g ∈ G.  β–‘

Example. Let σ and σ' two permutations of 𝓒n. We say that the permutation σ is conjugate to the permutation σ' if there exists a permutation τ in 𝓒n such that

σ'= τστ−1

Fix any x in a group G, then for *every* g in G, the element gxg-1 is in G, and all of these are "the conjugates" of x. They always exist. Sometimes x itself is the only when, such as when G is abelian.

For example in the permutation group 𝓒3, the element (1, 2) is conjugate to (3, 1) because (3, 1) = (1, 2, 3) (1, 2) (2 , 3)−1.

Conjugacy is an equivalence relation, i.e. Let (G, *) be a group and a,bG. If the element a is conjugate to bG then:

  1. Reflexive, each element is conjugate to itself, since ∀aG, there exists eG such that a = e * a * e−1, which implies, a ~ a.

  2. Symmetric, b ~ a. For a,bG, a = x * b * x−1, for some xG β‡’ b = x−1 * a * x = y * a * y−1 where y = x−1G.

  3. Transitive: For a,b,cG, if a ~ b and b ~ c, then a ~ c.

    a = x * b * x−1  and   b = y * c * y−1
    β‡’ a = x * (y * c * y−1) * x−1
    β‡’ a = (x * y) * c * (y−1 * x−1)
    β‡’ a = (x * y) * c * (x−1 * y)−1 β‡’ a ~ c.

Hence, G is the disjoint union of the equivalence classes, which are called conjugacy classes. Thus the conjugacy class of xG is

{gG | there exists aG with g = axa−1}

Note that the conjugacy class of x is {x} if and only if x commutes with each elements of G. We say that a conjugacy class is trivial if it contains only one member.

To find the elements a, bG which determine the same conjugate of x, notice that axa−1 = bxb−1 if and only if (b−1a) = x(b−1a), it turns out that we need to find the elements that commute with x. We can calculate the size of a conjugacy class in terms of another subgroup of G.

7.5.2 Definition. The centraliser of the element x ∈ G, written CG(x), is the set of all elements of G which commute with x:

CG(x) = {g ∈ G : gx = xg}. β–‘

The centralizer of x contains at least x itself as well as the center Z(G) of G, because the elements in the center commute with every element in G. Also, x ∈ Z(G) iff C(x) = G.

There is a procedure by which conjugates of elements in a permutation group may be computed with ease.

7.5.3 Proposition Let σ a permutation, written as a product of disjoint cycles. Then

  1. Every conjugate permutation σ' = τστ−1 of σ has the same cyclic structure of σ. Furthermore, the integers appearing in the cycles of σ' are obtained applying the permutation τ to the integers that appear in the cycles of σ i.e. it is obtained from that for Οƒ by substituting Ο„(k) for k throughout.

  2. If σ and σ' are two permutation of 𝓒n which have the same cyclic structure, then they are conjugate.

Proof. Let a and b two consecutive numbers in the writing of any of the two cycles of σ (the last element and the first are as well consecutives), i.e.

b = σ(a)

Let’s examine how τστ−1 acts on the different elements. Let τ(a) = s and τ(b) = t, then

τστ−1(s) = τσ(a) = τ(b) = t

that is, b is the consecutive element of a in the cyclic structure of σ, the τ(b) is the consecutive of τ(a) in the cyclic structure of τστ−1. This means that if for example

σ = (a, b, c, d)(e, f, g)(h i)

σ' = (a', b', c', d')(e', f', g')(h', i')

two permutations with the same cyclic structure. We prove now that they are conjugates. Indeed we've σ' = τστ−1, where τ is a permutation such that

τ = |a b c d e f|
|a' b' c' d' e' f'|

that is, having the behavior we described above on the elements of the cyclic structure of σ and on the elements that do not appear in σ in a compatible way with a permutation.

Remark. It is possible that two permutations of An which are conjugated in 𝓒n are not conjugated in An, because there is not τ in An such that σ'= τστ−1, that is the τs permutations are in 𝓒n but not in An.

7.5.4 Example. In 𝓒7 let

σ = (1,5) (2,3,4),   τ = (1,4,3)(2,6,7,5)

We have that

τστ−1 = (4,2)(6,1,3)

Since we have that (τ(1), τ(5)) (τ(2), τ(3), τ(4)) = (4,2)(6,1,3).

7.5.5 Example. Given

σ =( 3 , 4 , 5 ) ( 1 , 2 ) ( 7 ,8 ) ( 6 ) ( 9 ) σ' = ( 1 , 6 , 2 ) ( 3 , 5 ) ( 8 , 9 ) ( 4 ) ( 7 )  so  τ = [ 3 4 5 1 2 7 8 6 9 1 6 2 3 5 8 9 4 7 ]

It is well known that an equivalence relation provides a unique way to classify the elements of a set.

7.5.6 Proposition. There are as many conjugate classes in 𝓒n as different cyclic structures.

In 𝓒3, 𝓒4 and 𝓒5 the different cyclic structures

𝓒3   (βˆ’), (βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’)

𝓒4   (βˆ’), (βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’,βˆ’), (βˆ’,βˆ’)(βˆ’,βˆ’)

𝓒5   (βˆ’), (βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’,βˆ’,βˆ’), (βˆ’,βˆ’)(βˆ’,βˆ’), (βˆ’,βˆ’,βˆ’)(βˆ’,βˆ’)

Where (βˆ’) represents the id permutation. Thus in 𝓒3, 𝓒4, 𝓒5 there are respectively 3, 5 and 7 conjugate classes, namely

𝓒3
(βˆ’) = id
(βˆ’,βˆ’) = {(1 2), (1 3), (2 3)}
(βˆ’,βˆ’,βˆ’) = {(1 2 3), (1 3 2)}

𝓒4
(βˆ’) = id
(βˆ’,βˆ’) = {(1 2), (1 3), (1 4), (2 3) (2 4), (3 4)}
(βˆ’,βˆ’,βˆ’) = {(1 2 3), (1 2 4), (1 3 4) (1 3 2), (1 4 2), (1 4 3), (2 3 4), (2 4 3)}
(βˆ’,βˆ’,βˆ’,βˆ’) = {(1 2 3 4), (1 2 4 3), (1 3 2 4) (1 3 4 2), (1 4 2 3), (1 4 3 2)}
(βˆ’,βˆ’)(βˆ’,βˆ’) = {(1 2)(3 4), (1 3)(2 4), (1 4) (2 3)}

Let's count now the number of permutations in each conjugate class. It suffices to count the number of distinct r-cycles, r = 1, ..., n and the cyclic products in 𝓒n.

7.5.7 Proposition. The number of r-cycles in the symmetric group 𝓒n is given, for 1 ≀ r ≀ n, by the following equivalent formula

n!/r(n βˆ’ r)!

Proof. By its cyclic nature, a cycle of length r can be represented in r different equivalent ways by cyclically shuffling its members. For example, the cycle (1, 2, 3) can also be represented as (2, 3, 1) and (3, 2, 1). To obtain the distinct cycles of length r we divide r! by r, that is (r − 1)!. We multiply then, this number by the number of ways we can pick r numbers among n i.e.: Cn, r ways (combinations without repetitions); Thus we have

Cn, r β‹… (r βˆ’ 1)! = n!/r(n βˆ’ r)! β–‘

With a similar reasoning (see also the Exercises) we can derive the number of permutations in 𝓒n whose cycle structure consists of rii cycles of a given length.

7.5.8 Proposition. The number of permutations in 𝓒n whose cycle structure consists of ri cycles of length ki, for i = 1, ..., m is

n!/ (k1r1 β‹…β‹…β‹… kmrm r1! β‹…β‹…β‹… rm!)

Proof. We write down a template consisting of ri cycles of length ki, whit − in place of the elements {1,..., n}. For example, if the cycle lengths are 3,3,2,1, then the template is

(_,_,_)(_,_,_)(_,_)(_,_)(_)

We can fill the blanks in n! different ways. However, there are two ways in which the number n! is an overcount of the desiered number. For each r-cycle, you divide by ki as only the cyclic order within a cycle plays a role, not which element we start with. Second, for each cycle length ki, the ri! arrangements of the ri cycles of length ki are counted separately, but they give the same permutations, so we must divide also by r1! β‹…β‹…β‹… rm!.  β–‘

𝓒4
Cyclic structure Number of distinct conjugates
(βˆ’) 1
(βˆ’,βˆ’) C4,2 = 6 (To construct a transposition (a,b) we choose 2 of 4 elements, and this can be done in 6 ways.
(βˆ’,βˆ’,βˆ’) C4,3 β‹… 2! = 8 (To construct a 3-cycle, we choose 3 of 4 elements, but then each choice can be arranged in two different ways).
(βˆ’,βˆ’,βˆ’,βˆ’) C4,4 β‹… (4 - 1)! = 6
(βˆ’,βˆ’)(βˆ’,βˆ’) [C4,2 β‹… (2-1)! β‹… C2,2 β‹… (2 - 1)!]/ 2! = 3 or equivalently (4!)/(4 β‹… 2!) = 3

If you try to count permutations of type (-,-)(-,-)(-,-) in 𝓒6 you select the pairs forming the 2-cycles in three stages. Because you don't care about the ordering of the stages, the answer is (C6,2 β‹… C4,2 β‹… C2,2)/3! = 15.

Based on the two previous examples we can state the following proposition.

7.5.9 Definition. Let n an integer, we say that the sequence of positive integers n1,n2 β‹…β‹…β‹… nt with n1 > n2 > n3 β‹…β‹…β‹… > nt is a partition of the integer n.

n = n1 + n2 + ... + nt

For example the partitions of the integer 4 are:

4 = 4
4 = 3 + 1
4 = 2 + 2
4 = 2 + 1 + 1 + 1
4 = 1 + 1 + 1 + 1

you will notice that each parittion of the integer 4, corresponds to one of the five cyclic structure of 𝓒4, (the last corresponds to the identity, the first to a 4-cycle).

7.5.10 Proposition. The number of Conjugacy class 𝓒n is the same as the partitions of the integer n.

If we want to calculate the Conjugacy classes of An we must be careful: permutations which are conjugate in Sn are not necessarily conjugate in An. e.g in S3, (1 2 3) is conjugate to (1 3 2) because (2 3)−1 (1 2 3) (2 3) = (1 3 2); but (2 3) is not in A3. This tells us that (1 2 3) and (1 3 2) are in separate conjugacy classes.

Let's examine another example.

Example 7.5.11. The elements of the alternating group A4 of S4 are given below:

(id),   (124)   (142)   (12)(34)
(123)   (143)   (234)   (13)(24)
(132)   (134)   (243)   (14)(23)

To go from (123) to (243) requires you change 1 into 2 and 2 into 4. So clearly (12)(24)(123) ((12)(24))−1 = (243). It involves two substitutions. And (12)(24) is in A4. You can get the others in the class the same pattern.

In the full symmetric group S4 all 3-cycles are conjugate, e.g. going from (123) to (132) takes one substitution, 2 into 3:

(23)(123)(23)−1 = (132)

But (2 3) does not lie in A4, thus we split the 3-cycle of S4 in two conjugacy classes so that (123) and (132) are in separate classes of A4. The same observation applies to other pairs of 3-cycles that involve the same set of objects. On the other hand (12)(34), (13)(24) and (14)(23) remain conjugate in A4, for example

(234)−1 (12)(34) (234) = (13)(24)

Thus is found that hte conjugacy classes of A4 are as follows

C1 = {id}
C2 = {(123), (142), (134), (243)}
C3 = {(132), (124), (143), (234)}
C4 = {(12)(34), (13)(24), (13)(24), (14)(23)}.   β– 

«The transformation group Index Exercises on Conjugacy classese »