0% found this document useful (0 votes)
34 views10 pages

Permutations and Orbits

The document discusses permutations and group theory. It defines permutations, symmetric groups, and dihedral groups. It also covers Cayley's theorem stating that every group is isomorphic to a group of permutations. Cycle notation for permutations is introduced along with standard and matrix notations.

Uploaded by

andres aldana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
34 views10 pages

Permutations and Orbits

The document discusses permutations and group theory. It defines permutations, symmetric groups, and dihedral groups. It also covers Cayley's theorem stating that every group is isomorphic to a group of permutations. Cycle notation for permutations is introduced along with standard and matrix notations.

Uploaded by

andres aldana
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

5 Permutations and Orbits

In the earliest conceptions of group theory, all groups were considered permutation groups. Essentially
a group was a collection of ways in which one could rearrange some set or object: for example, each
‘rearrangement’ of an equilateral triangle corresponds to one of its symmetries. It was Arthur Cayley,
of Cayley-table fame, who formulated the group axioms we now use. Importantly, he also proved
what is now known as Cayley’s Theorem: that the old definition and the new are identical.

5.1 The Symmetric Group


So what, first, is a permutation?

Definition 5.1. A permutation of a set A is a bijection ϕ : A → A.

Theorem 5.2. If A is any set, then the set of permutations S A of A forms a group under composition.

Proof. We check the axioms:

Closure If ϕ, ψ are bijective then so is the composition ϕ ◦ ψ. X

Associativity Permutations are functions, the composition of which we know to be associative. X

Identity The identity function id A maps all elements of A to themselves: id A : x 7→ x. This is


certainly bijective. X

Inverse If ϕ is a permutation then it is a bijection, whence the function ϕ−1 exists and is also a
bijection. X

Definition 5.3. The symmetric group on n-letters Sn is the group of permutations of any1 set A of n
elements. Typically we choose A = {1, 2, . . . , n}.

Basic combinatorics should make the following obvious:

Lemma 5.4. Sn has n! elements.2

To describe a group as a permutation group simply means that each element of the group is being
viewed as a permutation of some set. As the following result shows, all groups are permutation
groups, although sadly not in a particularly useful way!

Theorem 5.5 (Cayley’s Theorem). Every group G is isomorphic to a group of permutations.

Proof. For each element a ∈ G, let ρ a : G → G be the function ρ a : g 7→ ag (i.e. left multiplication by
a). We make two claims:

1. Each ρ a is a permutation of G.
2. {ρ a : a ∈ G }, ◦ forms a group isomorphic to G.


1 S is the explicit group of permutations of {1, 2, . . . , n } or the abstract group of permutations of any set with n elements.
n
2 In contrast to C or Z where the subscript is the order of the group.
n n

1
The first claim is straightforward. ρ a−1 is the inverse function to ρ a :
∀ g ∈ G, (ρ a−1 ◦ ρ a )( g) = a−1 ag = g = idG ( g)
whence each ρ a is a bijection.
Now define a map φ : G → {ρ a } by φ( a) = ρ a . We claim this is an isomorphism:
1–1 φ( a) = φ(b) =⇒ ρ a = ρb =⇒ ∀ g ∈ G, ag = bg =⇒ a = b. X
Onto Certainly every permutation ρ a is in the image of φ. X
Homomorphism If a, b ∈ G, then
φ( a) ◦ φ(b) : g 7→ ρ a (ρb ( g)) = abg = ρ ab ( g)
from which φ( ab) = φ( a) ◦ φ(b).
Note that Cayley’s Theorem is not saying that every group is isomorphic to some Sn . It is saying that
every group G is isomorphic to some subgroup of SG .

Three notations for permutations


Standard notation Suppose that σ ∈ S4 is the following map
1 3
   
2 1
σ:3 7→ 4 , i.e. σ (1) = 3, σ(2) = 1, σ(3) = 4, σ(4) = 2
  

4 2
We could then write
1 2 3 4
 
σ=
3 1 4 2
where you read down columns to find where σ maps an element in the top row. Composition is read
in the usual way for functions, do the right permutation first. Thus if
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
      
ρ= then σρ = = (∗)
1 4 3 2 3 1 4 2 1 4 3 2 3 2 4 1

Matrix Notation σ can also be viewed as acting on the vector (1, 2, 3, 4)T , which suggests a matrix
method for encoding elements of Sn . For example, our permutation σ may be written
0 0 1 0
 
1 0 0 0
σ= 0 0 0 1

0 1 0 0
as a permutation matrix: multiplying on the left by such a matrix permutes the entries of vectors.
Definition 5.6. A permutation matrix is an n × n matrix with exactly one entry of 1 in each row and
column and the remaining entries 0.
Indeed we may conclude:
Theorem 5.7. The set of n × n permutation matrices forms a group under multiplication which is isomorphic
to Sn . By Cayley’s Theorem, every finite group of permutations is isomorphic to a group of matrices.

2
Cycle notation Our example permutation can be more compactly written as σ = (1 3 4 2). We read
from left to right, looping back to 1 at the end, each entry telling us where the previous is mapped to.
Thus (1 3 4 2) maps
1 7→ 3 7→ 4 7→ 2 7→ 1
We have shorter cycles if some of the elements are fixed; for example in our two notations
1 2 3 4
 
(1 3) = ∈ S4
3 2 1 4
Juxtaposition is used for composition:
(1 3 4 2)(2 4) = (1 3 4)
(compare with (∗) above).
Remember that multiplication of cycles is really composition of functions: although each cycle is read
from left to right when determining how it acts on elements of {1, 2, . . . , n}, we multiply cycles by
considering the rightmost cycle first. For example, in S5 ,


 1 7→ 3

2 7 → 3 7 → 5



(1 3 5 4)(2 3 4) : 3 7→ 4 7→ 1 =⇒ (1 3 5 4)(2 3 4) = (1 3)(2 5 4) (†)
4 7→ 2





5 7→ 4

This notation can be used to describe non-symmetric groups: e.g. if we label the corners of a rhombus,
the Klein 4-group V can be written in terms of cycles as a subgroup of S4 .
2

3 1 V = {e, (1 3), (2 4), (1 3)(2 4)}

4
Definition 5.8. Suppose that k ≤ n. A k-cycle in Sn is an element ( a1 a2 · · · ak ).
Cycles ( a1 · · · ak ) and (b1 · · · bl ) are disjoint if no element appears in both cycles: that is, if
{ a1 , . . . , ak } ∩ {b1 , . . . , bl } = ∅
The identity element is the only 0-cycle. It is sometimes written (), if not otherwise denoted by e.
As the example (†) illustrates, when computing the product of several cycles, the result will typically
be a product of disjoint cycles. This will prove very useful in the next section when we discuss orbits.

Subgroup relations between symmetric groups It is easy to see that Sm ≤ Sn ⇐⇒ m ≤ n. For


example, fix the final n − m elements of {1, . . . , n} so that
Sm = {σ ∈ Sn : σ(i ) = i, ∀i > m}.
In fact Sm is a subgroup of Sn in precisely (mn ) different ways: each copy of Sm arises by fixing n − m
elements of the set {1, . . . , n}: there are precisely (n−nm) = (mn ) ways of choosing these fixed elements.

3
5.2 Dihedral groups
Definition 5.9. The dihedral group Dn is the group of symmetries of the regular n-gon (polygon with
n sides).

Group? Now that we have defined permutation groups it is very easy to see that the dihedral
groups are indeed groups. Observe that a symmetry of an n-gon can be viewed as a permutation of
the corners of the n-gon for which ‘neighborliness’ is preserved. This is similar to how we viewed
the Klein 4-group above.
For example, if we label the corners of the regular hexagon 1 through 6 then 3 2
we see that D6 is the set of σ ∈ S6 such that σ(1) is always next to σ (6) and
σ(2), etc.
Clearly this says that D6 ⊆ S6 . 4 1
To see that D6 is a subgroup of S6 , we need only note that the composition
of two neighbor-preserving transforms must also preserves neighbors, as
does the inverse of such a map. 5 6

Elements of Dn The regular n-gon has 2n distinct symmetries and so | Dn | = 2n. These consist of:
2πj
n rotations For each j = 0, . . . , n − 1, let ρ j be rotation counter-clockwise by n radians.

n reflections Let µ j be reflection across the line making angle with the positive x-axis (make sure
πj
n
you put one of the corners of the n-gon on the x-axis!).3

Remarks Some authors write D2n instead of Dn precisely because | Dn | = 2n: in this course, Dn will
always mean the symmetries of the n-gon.
Every dihedral group Dn is a subgroup of the orthogonal group O2 (R). The correspondence is:

2πj
− sin 2πj 2πj 2πj
          
cos n cos sin
ρj =   n  µ j =
n  n 
2πj 2πj 2πj
− cos 2πj
    
sin n cos n sin n n

It is a good exercise to convince yourself that these matrices really do correspond to the rotations and
reflections claimed. In particular multiply any two of them together and see what you get. . .

Subgroup relations between dihedral groups


Dm ≤ Dn ⇐⇒ m | n. Recall the discussion of geometric proofs earlier where we saw that D3 ≤ D6 .
For instance, we can join every (n/m)th vertex of a regular n-gon to obtain a regular m-gon. Every
symmetry of the m-gon is then a symmetry of the n-gon.
3 For even-sided polygons these are often labelled differently, and split into two subsets of n2 reflections each. The
reflections µi are those which move all the corners of the n-gon, while δi refers to a reflection across a diagonal. We will see
this in our treatment of D4 below. In the abstract, is is simpler not to distinguish between these reflections.

4
Explicit descriptions of D3 and D4
2
D3 is the group of symmetries of an equilateral triangle. If we label the
corners as in the picture, we can easily define the elements of the group.

ρ0 the identity
2π 1
ρ1 rotate counter-clockwise by 3 radians

ρ2 rotate clockwise by 3 radians
µi reflect in the altitude through i
3
Here we write all the elements in permutation notation ad give the Cayley table.

Element Standard notation Cycle notation


1 2 3
 
◦ ρ0 ρ1 ρ2 µ1 µ2 µ3 ρ0 e = ()
1 2 3
Rotations 1 2 3
ρ0 ρ0 ρ1 ρ2 µ1 µ2 µ3
ρ1 (123)
ρ1 ρ1 ρ2 ρ0 µ3 µ1 µ2 2 3 1
1 2 3
ρ2 (132)
3 1 2
ρ2 ρ2 ρ0 ρ1 µ2 µ3 µ1
1 2 3
µ1 µ1 µ2 µ3 ρ0 ρ1 ρ2 µ1 (23)
1 3 2
Reflections

µ2 µ2 µ3 µ1 ρ2 ρ0 ρ1 1 2 3
µ2 (13)
µ3 µ3 µ1 µ2 ρ1 ρ2 ρ0 3 2 1
1 2 3
µ3 (12)
2 1 3

It should be immediately obvious that all the permutations of {1, 2, 3} are elements of D3 , and so
D3 ∼= S3 . Now we consider all the subgroups of D3 and its subgroup diagram.

Subgroup Isomorph Generating sets D3


{ ρ0 } C1 { ρ0 }
{ ρ0 , µ1 } C2 { µ1 }
{ ρ0 , µ2 } C2 { µ2 }
C3 C2 C2 C2
{ ρ0 , µ3 } C2 { µ3 }
{ ρ0 , ρ1 , ρ2 } C3 {ρ1 }, or {ρ2 }
any pair {ρi , µ j } where
D3 S3
i = 1, 2 and j = 1, 2, 3 C1

How can we be certain that there are no other subgroups of D3 ? A careful consideration of generating
sets should convince you. For example, suppose that a subgroup contains two reflections: WLOG
suppose these are µ1 , µ2 . We compute the subgroup generated by {µ1 , µ2 }. It must include

µ1 µ2 = ρ1 , ρ21 = ρ2 µ1 ρ1 = µ3

and thus the entire group.

5
D4 is the group of symmetries of the square. It consists of four rotations
2
and four reflections: the notation δj for reflection across a diagonal is
used here, rather than labelling all reflections µ j .

ρ0 the identity
ρ1 rotate counter-clockwise by π
2 radians 3 1
ρ2 rotate counter-clockwise by π radians

ρ3 rotate counter-clockwise by 2 radians
µi reflect across midpoints of sides
δi reflect across diagonals 4
Element Standard notation Cycle notation
1 2 3 4
 
◦ ρ0 ρ1 ρ2 ρ3 µ1 µ2 δ1 δ2 ρ0 e = ()
1 2 3 4
ρ0 ρ0 ρ1 ρ2 ρ3 µ1 µ2 δ1 δ2 1 2 3 4
(1234)

Rotations
ρ1
ρ1 ρ1 ρ2 ρ3 ρ0 δ2 δ1 µ1 µ2 2 3 4 1
1 2 3 4
ρ2 (13)(24)
3 4 1 2
ρ2 ρ2 ρ3 ρ0 ρ1 µ2 µ1 δ2 δ1
1 2 3 4
ρ3 ρ3 ρ0 ρ1 ρ2 δ1 δ2 µ2 µ1 ρ3 (1432)
4 1 2 3
µ1 µ1 δ1 µ2 δ2 ρ0 ρ2 ρ1 ρ3 1 2 3 4
µ1 (12)(34)
µ2 µ2 δ2 µ1 δ1 ρ2 ρ0 ρ3 ρ1 2 1 4 3
1 2 3 4
Reflections

µ2 (14)(23)
δ1 δ1 µ2 δ2 µ1 ρ3 ρ1 ρ0 ρ2 4 3 2 1
1 2 3 4
δ2 δ2 µ1 δ1 µ2 ρ1 ρ3 ρ2 ρ0 δ1 (24)
1 4 3 2
1 2 3 4
δ2 (13)
3 2 1 4
All the subgroups are summarised in the following table. In particular, note that D4  S4 : the latter
has many more elements!
D4
Subgroup Isomorph Generating sets
{ ρ0 } C1 { ρ0 }
{ ρ0 , µ i } C2 {µi } for each i
{ρ0 , δi } C2 {δi } for each i V C4 V
{ ρ0 , ρ2 } C2 { ρ2 }
{ ρ0 , ρ1 , ρ2 , ρ3 } C4 {ρ1 } or {ρ3 }
{ ρ0 , µ1 , µ2 , ρ2 } V {µ1 , µ2 }, {µ1 , ρ2 } or {µ2 , ρ2 }
{ρ0 , δ1 , δ2 , ρ2 } V {δ1 , δ2 }, {δ1 , ρ2 } or {δ2 , ρ2 } C2 C2 C2 C2 C2
any pair {ρi , µ j } or {ρi , δj }
D4 D4 where i = 1, 3 and j = 1, 2 or
any pair {µk , δl } where k, l = 1, 2
C1
In the subgroup diagram, the middle C2 is {ρ0 , ρ2 } while the two copies on each side contain either
the reflections δi or µi .

6
5.3 Orbits
In this section we continue the idea of a group being a set of permutations. In particular, we will see
how any element σ ∈ Sn partitions the set {1, 2, . . . , n}. This concept will be generalized later when
we consider group actions.

Definition 5.10. The orbit of σ ∈ Sn containing j ∈ {1, 2, . . . , n} is the set

orb j (σ) = {σk ( j) : k ∈ Z} ⊆ {1, 2, . . . , n}

Observe that orbσk ( j) (σ) = orb j (σ) for any k ∈ Z.


Be careful: each orbit is a subset of the set {1, 2, . . . , n}, not of the group Sn .

Examples If σ ∈ Sn is written in cycle notation (recall Definition 5.8) using disjoint cycles, then the
cycles are the orbits! For example, in S5 ,

Orbits of (134) are {1, 3, 4}, {2}, {5}


Orbits of (12)(45) are {1, 2}, {3}, {4, 5}

The same does not hold if the cycles are not disjoint. For example, σ = (13)(234) ∈ S4 maps

1 7→ 3 7→ 4 7→ 2 7→ 1

so there is only one orbit: orb j (σ) = {1, 2, 3, 4} for any j. In fact, σ = (1234), from which the orbit is
obvious.

Given that disjoint cycle notation is so useful for reading orbits, it is a natural question to ask if any
permutation σ can be written as a product of disjoint cycles. The answer, of course, is yes, with the
disjoint cycles turning out to be precisely the orbits of σ!

Theorem 5.11. The orbits of any σ ∈ Sn partition X = {1, 2, . . . , n}.

Proof. Define ∼ on X = {1, 2, . . . , n} by

x ∼ y ⇐⇒ y ∈ orbx (σ)

We claim that ∼ is an equivalence relation.

Reflexivity x ∼ x since x = σ0 ( x ). X

Symmetry x ∼ y =⇒ y = σk ( x ) for some k ∈ Z. But then x = σ−k (y) =⇒ y ∼ x. X

Transitivity Suppose that x ∼ y and y ∼ z. Then y = σk ( x ) and z = σl (y) for some k, l ∈ Z. But
then z = σk+l ( x ) and so x ∼ z. X

The equivalence classes of ∼ are clearly the orbits of σ, which therefore partition X.

Corollary 5.12. Every permutation can be written as a product of disjoint cycles.

7
Proof. Write out each of the orbits of σ ∈ Sn in order, placing each orbit in parentheses ( ). Since the
orbits of σ partition X = {1, 2 . . . , n} the cycles obtained are disjoint.
More concretely,

orb1 (σ) = {1, σ(1), σ2 (1), . . .}

If this orbit is the entirity of X, then we are finished. Otherwise, let

x1 = min{ x : x 6∈ orb1 (σ)}

and construct its orbit:

orbx1 (σ) = { x1 , σ( x1 ), σ2 ( x1 ), . . .}

This orbit must be disjoint with orb1 (σ). Now repeat. It is immediate from the construction that
   
σ = 1 σ (1) σ 2 (1) · · · x1 σ ( x1 ) σ 2 ( x1 ) · · · · · · · · ·

Examples If you mechanically follow the algorithm for multiplying cycles (see the previous section)
you will automatically end up with a product of disjoint cycles:

1. (13)(234)(1432) = (123)

2. (13)(24)(12)(34) = (14)(23)

Note that disjoint cycles commute! E.g. (14)(23) = (23)(14).

Now that we are able to write any permutation as a product of disjoint cycles, we are able to compute
much more easily. For example:

Theorem 5.13. The order of a permutation σ is the least common multiple of the lengths of its disjoint cycles.

Proof. Write σ as a product of disjoint cycles σ = σ1 · · · σm . Since disjoint cycles commute, it it imme-
diate that

σn = σ1n · · · σmn

Moreover, since the terms σjn permute disjoint sets, it follows that

σn = e ⇐⇒ ∀ j, σjn = e

A k-cycle clearly has order k (the least positive integer l such that ( a1 · · · ak )l = e). If the orbits of σ
have lengths α j ∈ N respectively, it follows that

σjn = e ⇐⇒ α j | n

Thus n must be a multiple of α j for all j. The least such n is clearly lcm{α j }.

8
Example The order of σ = (145)(3627)(89) ∈ S9 is lcm(3, 4, 2) = 12.
We can easily calculate σ3465 for the above σ. Since 3465 = 12 · 288 + 9 we have

σ3465 = (σ12 )288 σ9 = σ9 = (145)9 (3627)9 (89)9 = (3627)(89)

since (145), (3627) and (89) have orders 3, 4 and 2 respectively.

5.4 Transpositions
Instead of breaking a permutation σ into disjoint cycles, we can consider a permutation as being
constructed from only the simplest bijections.

Definition 5.14. A 2-cycle ( a1 a2 ) is also known as a transposition, since it swaps two elements of
{1, 2, . . . , n} and leaves the rest untouched.
Theorem 5.15. Every σ ∈ Sn is the product of transpositions.

Proof. There are many, many ways to write out a single permutation as a product of transpositions.
One method is to first write σ as a product of disjoint cycles, then write each cycle as follows:

( a1 · · · ak ) = ( a1 ak )( a1 ak−1 ) · · · ( a1 a2 )

Just look carefully to see that this works!

Example (17645) = (15)(14)(16)(17)


Definition 5.16. A permutation σ ∈ Sn is even/odd if it can be written as the product of an even/odd
number of transpositions.

Theorem 5.17. The concepts of even/odd are well-defined: every permutation is either even or odd, and not
both.

Proof. Recall that any permutation σ ∈ Sn can be written as an n × n permutation matrix (Definition
5.6). A 2-cycle is a permutation matrix which swaps two rows: it therefore differs from the n × n
identity matrix only in that two of its rows are swapped. For example

1 0 0 0
 
0 0 0 1
(24) = 
0
 ∈ S4
0 1 0
0 1 0 0

From linear algebra we have that swapping two rows of a matrix changes the sign of its determinant.
Hence det(2-cycle) = −1. It follows that
(
1 if σ is the product of an even number of 2-cycles,
det(σ) =
−1 if σ is the product of an odd number of 2-cycles.

In particular σ cannot be both odd and even.

9
5.5 The Alternating Groups
The concept of even permutations leads us to the definition of a new collection of groups.
Definition 5.18. The alternating group An (n ≥ 2) is the group of even permutations in Sn .
n!
Theorem 5.19. An has exactly half the elements of Sn : that is | An | = 2.
Proof. Since n ≥ 2, we have (12) ∈ Sn . Define φ : An → {odd permutations} by φ(σ) = (12)σ. We
claim that this is a bijection
1–1 φ(σ) = φ(τ ) =⇒ (12)σ = (12)τ =⇒ στ. X
Onto If ρ is an odd permutation, then (12)ρ is even and so in An . Therefore ρ = φ((12)ρ). X
Since φ is a bijection, it follows that there are exactly the same number of even and odd permutations
in Sn . Exactly half of them are therefore even.

Examples: small alternating groups


1. A2 = {e} is extremely boring!
2. A3 = {e, (13)(12), (12)(13)} = {e, (123), (132)} is simply the cyclic group of order 3.
3. A4 = {e, (123), (132), (124), (142), (134), (143), (234), (243), (12)(34), (13)(24), (14)(23)} is the
first genuinely new group in the alternating family. It has order 12 and is non-Abelian: for
example
(123)(124) = (13)(24) 6= (14)(23) = (124)(123)
We already know of one non-Abelian group of order 12: the dihedral group D6 . But we can
quickly see that A4 is non-isomorphic to D6 : all elements of A4 have order 1, 2 or 3, while D6 ,
being the symmetries of a hexagon, contains a rotation of order 6.
A concrete appearance of A4 can be seen as the group of rotations of a tetrahedron. Either label
the corners of a tetrahedron, or the faces, with the numbers 1, 2, 3, 4. Can you visualize how
each element of A4 transforms each tetrahedron?

10

You might also like