CHAPTER 7

Permutations

Very little of this chapter comes from PJE.

Definitions and notation

7.1. Definition. (PJE def. 12.1.6) A permutation of a set A is a bijection ρ : A → A.

7.2. Example. For any set A, the identity map, 1A , defined by 1A (a) = a for all
a ∈ A, is a permutation of A.

7.3. Definition. Recall the finite set Nn = {1, 2, . . . , n}.

The symmetric group on n letters, denoted Sn , is the set of all permutations of Nn .
The identity permutation in Sn is denoted 1n .

7.4. Notation. If ρ ∈ Sn , we can write ρ using two-row notation due to Cauchy:

!
1 2 3 ... n
ρ= .
ρ(1) ρ(2) ρ(3) . . . ρ(n)
In particular, !
1 2 3 ... n
1n = .
1 2 3 ... n
Aside One of the goals of this part of the course is to understand the words
‘symmetric’ and ‘group’.

Example. S3 consists of
! ! !
1 2 3 1 2 3 1 2 3
, , ,
1 2 3 1 3 2 2 1 3
! ! !
1 2 3 1 2 3 1 2 3
, , .
3 2 1 3 1 2 2 3 1

Recall a result from earlier in the course.

Theorem. (PJE theorem 11.1.7)

If A is a finite set then f : A → A is a bijection if, and only if, it is an injection.
65
 Permutations 66

Proof. Covered in the first half of the course. 

7.5. Theorem. If |A| = n ≥ 1 then the number of permutations ρ : A → A is n!. In

particular, |Sn | = n!.
Proof Since A is a finite set, to count bijections it suffices to count injections, the
number of which is, by Proposition 1.3,
n! n! n!
|Inj (A, A)| = = = = n!. 
(n − n)! 0! 1

In particular, |S3 | = 3! = 6. Since we found 6 different permutations above in S3 we, in

fact, found all the permutations in S3 .

Compositions

Recall, if ρ and π are functions A → A, then the composite function is defined by

ρ ◦ π (a) = ρ (π (a))

for all a ∈ A. Further, if ρ and π are bijections then ρ ◦ π is a bijection. Hence the
composition of permutations is a permutation. We record this in the following

7.6. Definition. Let ρ, π ∈ Sn . The composition, or product, of ρ and π is the

permutation
ρ ◦ π ∈ Sn , (ρ ◦ π)(a) = ρ(π(a)), ∀a ∈ Nn .
Alternative notation: we may omit the ◦ sign and write ρπ for the product ρ ◦ π.

7.7. Remark. The product of permutations is, in general, not commutative, mean-
ing that ρπ may be different from πρ.

7.8. Example. Let ρ, π ∈ S5 be given by

! !
1 2 3 4 5 1 2 3 4 5
ρ= and π= .
4 2 1 3 5 2 3 4 5 1
Calculate ρ ◦ π.
Solution To write ρ ◦ π in the same way we have to see first what π does to a given
element of A and then secondly what ρ does to this image. In this example,
ρ π
2 ←− 2 ←− 1 so ρ ◦ π (1) = 2,
ρ π
1 ←− 3 ←− 2 so ρ ◦ π (2) = 1,
ρ π
3 ←− 4 ←− 3 so ρ ◦ π (3) = 3,
ρ π
5 ←− 5 ←− 4 so ρ ◦ π (4) = 5,
ρ π
4 ←− 1 ←− 5 so ρ ◦ π (5) = 4.
 Permutations 67

Thus
! !
1 2 3 4 5 1 2 3 4 5
ρ◦π = ◦
4 2 1 3 5 2 3 4 5 1
!
1 2 3 4 5
= .
2 1 3 5 4

Note we first looked at π then at ρ, so read ρ ◦ π from the right.

Note also that !

1 2 3 4 5
π◦ρ= ,
5 3 2 4 1
so that, for instance, ρ ◦ π (1) = 2 but π ◦ ρ (1) = 5. Thus

ρ ◦ π 6= π ◦ ρ,

which shows that composition of permutations is not commutative.

Inverses

Recall, a bijection always has an inverse. The inverse of a permutation written in the
two row manner can easily be found by exchanging the rows, and then reordering
the columns so that the entries on the upper row appear in the correct order.

7.9. Definition. The inverse of a permutation ρ ∈ Sn is the bijection inverse to ρ.

This is the permutation ρ−1 ∈ Sn such that ρ ◦ ρ−1 = ρ−1 ◦ ρ = 1n .

Example. In S5 find the inverse of

!
1 2 3 4 5
ρ= .
4 2 1 3 5

Solution ! !
−1 4 2 1 3 5 1 2 3 4 5
ρ = = .
1 2 3 4 5 3 2 4 1 5
You can check that your answer satisfies the definition of inverse:
! !
1 2 3 4 5 1 2 3 4 5
◦
4 2 1 3 5 3 2 4 1 5
! !
1 2 3 4 5 1 2 3 4 5
= ◦ = 15 .
3 2 4 1 5 4 2 1 3 5
 Permutations 68

Cycles

7.10. Definition. If ρ is a permutation in Sn then ρ fixes a ∈ Nn if ρ (a) = a and ρ

moves a if ρ (a) 6= a.

Notation: Fix (ρ) = {a ∈ Nn : ρ(a) = a}, Move(ρ) = Nn \ Fix (ρ).

! !
1 2 3 4 5 1 2 3 4 5
7.11. Example. In S5 , let σ = and τ = .
2 3 1 5 4 5 2 3 4 1
Then Fix (σ) = ∅ and Fix (τ ) = {2, 3, 4}.

Exercise. Is there a permutation τ ∈ S5 such that Fix (τ ) = {1, 2, 3, 4}?

7.12. Definition. Permutations σ, τ ∈ Sn are disjoint if Move(σ) ∩ Move(τ ) = ∅.

7.13. Lemma. Two disjoint permutations σ, τ ∈ Sn commute, i.e., στ = τ σ in Sn .

Proof Assume that permutations σ, τ ∈ Sn are disjoint. We need to show that σ(τ (a)) =
τ (σ(a)) where a ∈ Nn is arbitrary. Since a cannot be both in Move(σ) and in Move(τ ),
one has σ(a) = a or τ (a) = a; without loss of generality, assume σ(a) = a.

We claim that τ (a) is also fixed by σ. This is true if τ (a) = a, since a is fixed by σ.
Otherwise, τ (a) 6= a; the permutation τ is a bijection hence an injection, so, applying
τ to both sides, we obtain τ (τ (a)) 6= τ (a). This shows that τ (a) ∈ Move(τ ), so τ (a) ∈
/
Move(σ), as claimed.

Now, using the assumption a = σ(a), we conclude that σ(τ (a)) = τ (a) = τ (σ(a)). 

7.14. Warning. The converse of the proposition does not hold. If σ, τ commute, i.e.,
στ = τ σ, it does not necessarily mean that σ, τ are disjoint. An easy example is: σ
commutes with σ for all σ ∈ Sn , but σ is not disjoint with σ if σ 6= 1n .

7.15. Definition. Let a1 , a2 , . . . , ar be distinct elements in Nn . If ρ is a permutation

that fixes all the other elements of Nn and if

ρ (a1 ) = a2 , ρ (a2 ) = a3 , ρ (a3 ) = a4 , . . . , ρ (ar−1 ) = ar , ρ (ar ) = a1 ,

i.e.
a1 7→ a2 7→ a3 7→ . . . 7→ ar 7→ a1 ,
then ρ is called a cycle of length r, sometimes called an r-cycle. The r-cycle above
will be denoted by
(a1 , a2 , a3 , . . . , ar ) .
 Permutations 69

Remark. Note that any ai can be taken as the “starting point”, so

(a1 , a2 , a3 , . . . , ar ) = (a2 , a3 , . . . , ar , a1 ) = · · · = (ar , a1 , . . . , ar−2 , ar−1 ) .

Remark. We can take r = 1 in the definition to get a 1-cycle, (a1 ). But such a cycle
fixes all elements of Nn and is thus the identity. Hence all 1-cycles equal the identity,
i.e. (a) = 1n for all a ∈ Nn .

7.16. Definition. A 2-cycle is called a transposition.

7.17. Example. (i) Two permutations seen before were cycles. Namely, ρ, π ∈ S5 ,
!
1 2 3 4 5
ρ= = (1, 4, 3) ,
4 2 1 3 5

and
!
1 2 3 4 5
π= = (1, 2, 3, 4, 5) .
2 3 4 5 1

(ii) In S3 all permutations happen to be cycles, namely

13 , (2, 3) , (1, 2) , (1, 3) , (1, 3, 2) and (1, 2, 3) .

Remark. The inverse of a cycle is obtained simply by writing it in reverse order. So in

S5 ,
!
−1 −1 1 2 3 4 5
ρ = (1, 4, 3) = (3, 4, 1) = ,
3 2 4 1 5

as seen before. And we can compose cycles written in this notation, remembering to
read from the right. So, in S5 ,
!
1 2 3 4 5
ρ ◦ π = (1, 4, 3) ◦ (1, 2, 3, 4, 5) = ,
2 1 3 5 4

as seen before. Again, we did this by noting that π moved 1 to 2 which ρ then fixed.
Next π moved 2 to 3 which ρ moved to 1. Continue.
 Permutations 70

Example. In S3 we can represent all possible products in a table

◦ 13 (2, 3) (1, 2) (1, 3) (1, 3, 2) (1, 2, 3)

13 13 (2, 3) (1, 2) (1, 3) (1, 3, 2) (1, 2, 3)
(2, 3) (2, 3) 13 (1, 3, 2) (1, 2, 3) (1, 2) (1, 3)
(1, 2) (1, 2) (1, 2, 3) 13 (1, 3, 2) (1, 3) (2, 3)
(1, 3) (1, 3) (1, 3, 2) (1, 2, 3) 13 (2, 3) (1, 2)
(1, 3, 2) (1, 3, 2) (1, 3) (2, 3) (1, 2) (1, 2, 3) 13
(1, 2, 3) (1, 2, 3) (1, 2) (1, 3) (2, 3) 13 (1, 3, 2)

Note that because composition of functions is not commutative this table is not sym-
metric about the leading diagonal (which makes it different to earlier tables we have
seen for (Zm , +) , (Zm , ×) and (Z∗m , ×)).

Factoring permutations

Question If we can compose permutations, can we factor them?

Problem with this Question. In the last section we factored integers into prime
numbers. What is the equivalent of prime numbers for permutations?

Algorithm for factorisation into disjoint cycles is best illustrated by an example.

7.18. Example. In S6 factor

!
1 2 3 4 5 6
π= .
5 6 3 1 4 2

Solution

(1) Take the smallest ‘unused’ element in {1, 2, 3, 4, 5, 6} , namely 1. See what π does
to 1 on repeated applications. It sends 1 to 5. Then π sends 5 to 4. Next π sends
4 back to 1. Thus we have a cycle (1, 5, 4).
(2) Next look at the smallest ‘unused’ element, i.e not in the cycles already found. In
this case it is 2. Then we what happens to 2 under repeated applications of π, i.e.
2 7−→ 6 7−→ 2 and so we get another cycle (2, 6).
(3) Repeat by taking the smallest element not in these two cycles. We have only one
such element 3, and we see this is fixed by π, and so we get a 1-cycle (3), which
we know is the identity. When there is at least one non-identity cycle we can omit
the identity (3).
(4) When all elements are ‘used’, i.e. in some cycle, finish.
 Permutations 71

Hence
π = (1, 5, 4) ◦ (2, 6) ◦ (3) = (1, 5, 4) ◦ (2, 6) .

So in this way a permutation is factored into cycles, and thus cycles can be considered
an analogue of prime numbers.

It can be proved that each new cycle contains no elements in any earlier cycle. That
is, the new cycle is disjoint from all the earlier cycles.

The factorisation method above can be formalised as follows.

7.19. Theorem. Every permutation in Sn can be expressed as a product of disjoint

cycles uniquely up to a reordering of the cycles.

Proof. No formal proof given, but you should master the factorisation algorithm above.

Order of a permutation

7.20. Definition.

• The positive powers ρn of a permutation are defined inductively by setting ρ1 = ρ

and ρk+1 = ρ ◦ ρk for all k ∈ Z+ .

n
• The negative powers of a permutation are defined by ρ−n = ρ−1 for all n ∈ Z+ ,
i.e. taking positive powers (just defined) of the inverse of ρ.
• Finally, we set ρ0 = 1.

It can be shown by induction that powers satisfy the expected properties of exponents:

7.21. Claim. ρm+n = ρm ◦ ρn for all m, n ∈ Z. 

7.22. Corollary. Powers of a permutation ρ commute: ρm ρn = ρn ρm for all m, n ∈ Z.

Proof By Claim 7.21, both sides are equal to ρm+n . 

Now, the method described above of factorising a permutation started by taking an

element of A, repeatedly applying ρ until you returned to a when you then have a cycle.
This italicized sentence is an assumption: we have to show that repeatedly applying ρ
to a does, in fact, gets us back to a. This follows, for example, from the next Lemma.

7.23. Lemma. Let ρ be a permutation in Sn . There exists m ≥ 1 for which ρm = 1n .

 Permutations 72

Proof. Consider the set ρj : j ≥ 0 ⊆ Sn . The set Sn of all permutations of Nn is



finite (the number of all permutations is n!), hence ρj : j ≥ 0 is a finite set. Therefore


we must have repetition, i.e. ∃` > k ≥ 0 for which ρ` = ρk . Pre-multiplying both sides
by ρ−k , we obtain

ρ`−k = ρ` ◦ ρ−k by Claim 7.21

= ρk ◦ ρ−k since ρ` = ρk
= ρk−k again by 7.21
= ρ0 = 1n by definition.

Thus we have found an m = ` − k ≥ 1 for which ρm = 1n . 

This result motivates the following

7.24. Definition. The order of a permutation ρ ∈ Sn is the least positive integer d

such that ρd = 1n .

Remark. The order of a permutation exists, because by Lemma 7.23, it is the least
element of a non-empty set of positive integers.

Remark. Hence given a permutation in Sn , given a ∈ Nn , there will be an m such that

ρm (a) = a, so repeated application of ρ on a leads back to a thus giving a cycle. This
is as required for our method of factorisation into disjoint cycles.

7.25. Example. (Powers of a permutation via disjoint cycles)

Let !
1 2 3 4 5 6
π= .
5 6 3 1 4 2
Write π as a product of disjoint cycles. Hence find π 2 , π 3 , π 4 , π 5 , π 6 .

Solution Using the algorithm above to factorise π into disjoint cyces, we obtain

π = (1, 5, 4)(2, 6).

Now, since disjoint cycles commute, we can swap (1, 5, 4) and (2, 6) in the following
calculation:

π 2 = ππ = (1, 5, 4)(2, 6)(1, 5, 4)(2, 6) = (1, 5, 4)(1, 5, 4)(2, 6)(2, 6) = (1, 5, 4)2 (2, 6)2 .

We observe that the permutation (1, 5, 4)2 maps the elements 1, 5 and 4 as follows:

1 7→ 5 7→ 4, 5 7→ 4 7→ 1, 4 7→ 1 7→ 5
 Permutations 73

where each arrow is one application of the cycle (1, 5, 4); elements other than 1, 5 and
4 are fixed by (1, 5, 4)2 . We conclude that

(1, 5, 4)2 = (1, 4, 5).

We can similarly find the square of the transposition (2, 6):

(2, 6)2 = 1.

We arrive at
π 2 = (1, 4, 5).
Arguing in the same way, we find

π 3 = (1, 5, 4)3 (2, 6)3 = 1(2, 6) = (2, 6),

π 4 = (1, 5, 4)4 (2, 6)4 = (1, 5, 4)1 = (1, 5, 4),
π 5 = (1, 5, 4)5 (2, 6)5 = (1, 4, 5)(2, 6),
π 6 = (1, 5, 4)6 (2, 6)6 = 11 = 1.

Note, writing π as disjoint cycles makes it easier to compute π k . We found that the
order of π is 6.

Remark. Computing all the successive powers of π until we obtain identity is inefficient
if we only want to find the order of π. What if we had a permutation from S100 ? How
many powers would we have to compute? Can the order be anywhere near 100! (greater
than the number of electrons in observable universe)?

Question. Is there a better way to find the order from disjoint cycles?

We first answer the question for permutations for which the order is easy to find: cycles.

7.26. Lemma. A cyce of length r has order r.

Sketch of proof. To simplify the notation, instead of a cycle (a1 , a2 , . . . , ar ) we will use
the cycle π = (1, 2, . . . , r). Note that π(1) = 2, π 2 (1) = π(2) = 3, etc; using induction,
one shows that π k (1) = 1 + k if k < r. This means that π k 6= 1 if k < r.

On the other hand we have π r (1) = π(π r−1 (1)) = π(r) = 1. Moreover, for all i,
1 ≤ i ≤ r, we have π r (i) = π r (π i−1 (1)) which equals π i−1 (π r (1)) = π i−1 (1) = i. This
shows that π r = 1. Hence r is the least positive integer with this property, that is, the
order of π. 

7.27. Remark. One should be aware that powers of a cycle may not be cycles: e.g.,
(1, 2, 3, 4)2 = (1, 3)(2, 4).
 Permutations 74

7.28. Proposition. If the order of ρ is d then, for any m ∈ Z, ρm = 1 if, and only if,
d | m.

Proof. By the Division Theorem, write the integer m can be written as m = dq + r

where q ∈ Z and 0 ≤ r < d. We do this because we know that any product of dth
powers of ρ gives the identity. Indeed, Then

ρm = ρdq ρr = (ρd )q ρr = 1q ρr = ρr .

If ρm = 1, then ρr = 1 which is impossible if 0 < r < d by definition of d as the least

positive integer with this property. Hence r = 0, meaning d | m.

Vice versa, if d | m, then r = 0 and ρm = ρ0 = 1. 

We are almost ready to state the result about the order of an arbitrary permutation. It
depends on the following

7.29. Definition. The lowest common multiple of positive integers m1 , m2 , ..., mt ,

denoted by lcm (m1 , m2 , ..., mt ) is the least positive integer f such that m1 | f , m2 |
f, . . . , mt | f .

Remark. Properties of lcm. The lcm of positive integers m1 , m2 , ..., mt exists: it is

the least element of the set of all common multiples of m1 , . . . , mt which is not empty, for
example because it contains the product m1 . . . mt which is clearly a common multiple.
(Note, lcm may not be equal to the product, for example lcm(4, 6) = 12 not 24).

Moreover, if m1 | k, m2 | k, . . . , mt | k then f = lcm(m1 , m2 , ..., mt ) | k. That is, the

least common mupltiple divides any other common multiple.

To see this, write k = f q + r where 0 ≤ r < f , them each mi divides k and f hence
divides their linear combination r = k − f q. So r is a common multiple, but f is the
least positive common multiple, yet r < f . Hence r cannot be positive. The only
remaining possibility is r = 0 so that f | k.

Remark. Compare the definition of lcm to that of gcd.

7.30. Theorem. Suppose that

π = π1 ◦ π2 ◦ .... ◦ πm

is a decomposition into a product of disjoint permutations. Then

order(π) = lcm (order(π1 ), . . . , order(πm )) .

 Permutations 75

Proof. Consider the kth power

π k = (π1 ◦ π2 ◦ .... ◦ πm )k .

Since the permutations on the right hand side are disjoint the compositions commute,
so they can be moved around to give

π k = π1k ◦ π2k ◦ .... ◦ πm

k
.

Assume now that π k = 1, so π k moves no elements. Because the permutations are

disjoint each of π1k , π2k , ...., πm
k move different elements and so π k moves no elements if,

k moves no elements, i.e. π k = 1 for all 1 ≤ i ≤ m.

and only if, each of π1k , π2k , ...., πm i
Let di be the order of πi for each 1 ≤ i ≤ m, then by Proposition 7.28 πik = 1 for all
1 ≤ i ≤ m iff di | k for all 1 ≤ i ≤ m. Finally, in searching for the order of π we want
the least k divisible by all the di . By definition, this is exactly lcm(d1 , . . . , dm ). 

Remark. Although the Theorem is true for any set of disjoint permutations, in practice,
given a permutation π we decompose it into a product of disjoint cycles.

Corollary. Suppose that

π = σ1 ◦ σ2 ◦ .... ◦ σm
is a decomposition into a product of disjoint cycles, then the order of π is the least
common multiple of the lengths of the cycles σ1 , σ2 , . . . , σm .

7.31. Example. In S12 consider

!
1 2 3 4 5 6 7 8 9 10 11 12
π =
6 3 5 10 2 1 4 9 7 8 12 11

= (4, 10, 8, 9, 7) ◦ (2, 3, 5) ◦ (1, 6) ◦ (11, 12) .

The order equals lcm (5, 3, 2, 2) = 30.

7.32. Example. What is the largest order of all permutations in S12 ?

Solution We need to find positive integers a, b, c, ... that sum to 12 but for which
lcm (a, b, c, . . . ) is as large as possible. Just search to find 12 = 3 + 4 + 5, when
lcm (3, 4, 5) = 60. So, for example

(1, 2, 3) ◦ (4, 5, 6, 7) ◦ (8, 9, 10, 11, 12)

!
1 2 3 4 5 6 7 8 9 10 11 12
=
2 3 1 5 6 7 4 9 10 11 12 8
has order 60. 
 Permutations 76

7.33. Example. S8 . What is the order of

π = (1, 2, 4, 6, 8) ◦ (2, 3, 6) ◦ (6, 7)?

Solution CAREFUL, the cycles are not disjoint! We have to write this as a product
of disjoint cycles. The way we do this is similar to factorising a permutation given in
two-row notation. We start with 1 and note that an element is not moved by a cycle or
a product of cycles where it does not appear. So, for example,
(2,3,6)◦(6,7) (1,2,4,6,8)
1 −−−−−−−→ 1 −−−−−−→ 2

giving π(1) = 2. Continue:

(1,2,4,6,8)◦(2,3,6)◦(6,7)
1 −−−−−−−−−−−−−−−−−−→ 2
(1,2,4,6,8)◦(2,3,6)◦(6,7)
−−−−−−−−−−−−−−−−−−→ 3
(1,2,4,6,8)◦(2,3,6)◦(6,7)
−−−−−−−−−−−−−−−−−−→ 8
(1,2,4,6,8)◦(2,3,6)◦(6,7)
−−−−−−−−−−−−−−−−−−→ 1,

completing the first cycle (1, 2, 3, 8). Now start with 4:

(1,2,4,6,8)◦(2,3,6)◦(6,7)
4 −−−−−−−−−−−−−−−−−−→ 6
(1,2,4,6,8)◦(2,3,6)◦(6,7)
−−−−−−−−−−−−−−−−−−→ 7
(1,2,4,6,8)◦(2,3,6)◦(6,7)
−−−−−−−−−−−−−−−−−−→ 4.

Finally, 5 is fixed by π so will not appear in the decomposition into disjoint cycles. We
obtain
π = (1, 2, 3, 8) ◦ (4, 6, 7),
now a composition of disjoint cycles. The order is lcm (4, 3) = 12. 

Binary Operations

Question, why, earlier in the course did we call (Sn , ◦), the set of permutations on n
elements under composition, the Symmetric Group on n elements?

7.34. Definition. A binary operation on a set S is a function from the set S × S of

ordered pairs to S. We will denote it in general as ∗, so for each (a, b) ∈ S the function
sends (a, b) → a ∗ b, a value in S. Thus

∀a, b ∈ S, a ∗ b ∈ S.
 Permutations 77

If C ⊆ S we say that C is closed under ∗ iff

∀c, d ∈ C, c ∗ d ∈ C.

7.35. Example.

• +, −, × are binary operations on the set Z of integers. / (divide) is not a binary

operation on Z.
• Furthermore, if m ≥ 1, then +, −, × are binary operations on the finite set Zm
of congruence classes mod m. The subset Z∗m of Zm , which consists of invertible
congruence classes, is closed under × but not under +.
• If A is a set, then ∪, ∩, \ are binary operations on the power set P(A).
• If n ≥ 1, the set Fun(Nn , Nn ) of all functions from Nn to itself has binary operation
◦ (composition). The set Sn of all permutations of Nn is closed under ◦.

7.36. Example. Z20 is closed under ×20 . But {[4]20 , [8]20 , [12]20 , [16]20 } ⊆ Z20 is also
closed, we can draw up a table

× [4]20 [8]20 [12]20 [16]20

[4]20 [16]20 [12]20 [8]20 [4]20
[8]20 [12]20 [4]20 [16]20 [8]20
[12]20 [8]20 [16]20 [4]20 [12]20
[16]20 [4]20 [8]20 [12]20 [16]20

A binary operation may (or may not) satisfy the following important properties.

7.37. Definition.

(i) A binary operation ∗ on S is commutative if,

∀a, b ∈ S, a ∗ b = b ∗ a.

(ii) A binary operation ∗ on S is associative if,

∀a, b, c ∈ S, (a ∗ b) ∗ c = a ∗ (b ∗ c) .

(iii) A binary operation ∗ on S has an identity e ∈ S if

∀a ∈ S, e ∗ a = a and a ∗ e = a.

Remark. In the definition of an identity, we have to check both e ∗ a and a ∗ e since we

are not assuming that ∗ is commutative.
 Permutations 78

7.38. Example. On the set Z:

• +, × are commutative and associative;

• + has identity 0: ∀a ∈ Z 0 + a = a + 0 = a;
• × has identity 1: ∀a ∈ Z 1 × a = a × 1 = a;
• the binary operation − on Z is not commutative, not associative and has no identity.

On the set Sn of permutations:

• ◦ is associative; ◦ is not commutative for n ≥ 3; ◦ has identity 1n .

7.39. Example. {[4]20 , [8]20 , [12]20 , [16]20 , ×}. This binary operation is commutative
and associative. Looking back at the table above we see that the identity is [16]20 .

This last example is important, it shows that we get identities different to 1 and 0!

Note the use of the word “an” in the definition. But

7.40. Lemma. Suppose that ∗ is a binary operation on a set S and that (S, ∗) has an
identity. The identity is unique.

Proof . Suppose that e and f are identities on S. Then

e = e ∗ f since f is an identity (used here on the right),

= f since e is an identity (used here on the left). 

So we can now talk about “the” identity.

If, in the multiplication table for (S, ∗) , we can find an element whose row (and whose
column) is identical to the heading row (respectively heading column), then we have
found the identity.

7.41. Definition. Let S be a set with a binary operation ∗ and identity e ∈ S. We

say that an element a ∈ S is invertible if there exists b ∈ S such that

a ∗ b = e and b ∗ a = e.

We say that b is the inverse of a, and normally write b as a−1 .

 Permutations 79

7.42. Example. In (Z6 , ×) the element [2]6 has no inverse.

Solution If [2]6 had an inverse, i.e. [b]6 then

[2]6 [b]6 = [1]6 .

Multiply both sides by [3]6 to get

[6]6 [b]6 = [3]6 , i.e. [0]6 = [3]6 ,

since [6]6 = [0]6 , a contradiction.

The problem here is that 6 = 2 × 3 is composite. We have got round this in two ways in
this course. First we can look at (Zp , ×) with p prime, when every non-zero element has
an inverse. The second way it to look at (Z∗m , ×) where we have simply thrown away all
the elements that don’t have an inverse!

If, for an i ∈ S we can look in its row in the multiplication table and find the identity
in column j, say, and find in row j the identity in column i then i and j are inverse to
each other. If we can do this for every i ∈ S then every element will have an inverse.

7.43. Example. {[4]20 , [8]20 , [12]20 , [16]20 , ×}

× [4]20 [8]20 [12]20 [16]20

[4]20 [16]20 [12]20 [8]20 [4]20
[8]20 [12]20 [4]20 [16]20 [8]20
[12]20 [8]20 [16]20 [4]20 [12]20
[16]20 [4]20 [8]20 [12]20 [16]20

Since the identity is [16]20 we note

[4]20 × [4]20 = [16]20 so [4]−1

20 = [4]20 ,

[8]20 × [12]20 = [16]20 so [8]−1

20 = [12]20 ,

[12]20 × [8]20 = [16]20 so [12]−1

20 = [8]20 .

(The inverse of the identity is always itself!)

7.44. Lemma. Assume that the binary operation ∗ on S is associative. Assume that
(S, ∗) has an identity e and a ∈ S has an inverse. Then the inverse is unique.
 Permutations 80

Proof. If an element a has two inverses, b, c ∈ S say, then

a∗b = e and b∗a=e

a∗c = e and c ∗ a = e.

From these we keep b ∗ a = e and a ∗ c = e. These pieces of information are combined in

the following way,

b = b ∗ e = b ∗ (a ∗ c) since c is an inverse of a,
= (b ∗ a) ∗ c by associativity,
= e∗c since b is an inverse of a,
= c.

Thus b = c and the inverse is unique. 

So we can now talk about “the” inverse of an (invertible) element.

Groups

7.45. Definition. Given a set G and binary operation ∗ on G we say that (G, ∗) is a
group if, and only if,

G1. G is closed under ∗, [this is part of the definition of binary operation]

G2. ∗ is associative on G,
G3. (G, ∗) has an identity element, i.e.

∃e ∈ G : ∀a ∈ G, e ∗ a = a ∗ e = a,

G4. every element of (G, ∗) has an inverse, i.e.

∀a ∈ G, ∃a−1 ∈ G : a ∗ a−1 = a−1 ∗ a = e.

We say that (G, ∗) is a commutative or abelian group (after Niels Abel) if, and only
if, it is a group and ∗ is commutative.

Recall that in the course we showed that Z∗n is closed under multiplication. This was
done by taking [a]n , [b]n ∈ Z∗n and showing that

(4) ([a]n [b]n )−1 = [b]−1 −1

n [a]n .

What is important here is not the value of the inverse but that the product [a]n [b]n has
an inverse. For this implies [a]n [b]n ∈ Z∗n as required for closure.

But it can be shown that (4) holds in any group.

 Permutations 81

7.46. Proposition. Assume that (G, ∗) is a group. If x, y ∈ G then

(x ∗ y)−1 = y −1 ∗ x−1 .

Notice how the order has changed.

Proof . First note that (x ∗ y)−1 is, by definition, an inverse of x ∗ y.

Next note that

(x ∗ y) ∗ y −1 ∗ x−1 (x ∗ y) ∗ y −1 ∗ x−1



=
using ∗ is associative
x ∗ y ∗ y −1 ∗ x−1



=
again using ∗ is associative
= (x ∗ e) ∗ x−1
= x ∗ x−1
= e.

So (x ∗ y)∗ y −1 ∗ x
−1 = e. It is similarly shown that y −1 ∗ x−1 ∗(x ∗ y) = e. Together

these mean that y −1 ∗ x−1 is an inverse of x ∗ y.

Yet the inverse in a group is unique so the two inverses we have here must be equal, i.e.
(x ∗ y)−1 = y −1 ∗ x−1 . 

Question But why do we call (Sn , ◦) the symmetric group?

Consider, as an example, n = 4. Think of a square in the plane, center at the origin,
with vertices at (1, 1) , (−1, 1) , (−1, −1) and (1, −1) , labelled clockwise, 1,2,3 and 4.
What symmetries does the square have? It has rotational symmetries about the origin.
If we rotate by π/2 in the clockwise direction we see that corners map 1 → 2, 2 → 3,
3 → 4 and 4 → 1. So this rotation can be represented by the cycle (1, 2, 3, 4).
In the other direction what would (1, 2) ◦ (3, 4) represent? It would be a reflection in a
line through the origin.
For Student: What are the permutations that represent the other symmetries of the
square?
In this way we see that S4 contains the symmetries of the square. Hence the use of the
word “symmetry” in the name of (Sn , ◦).