MTH 410/514/620: Representation

Theory
Semester 2, 2016-2017

Dr. Prahlad Vaidyanathan

Contents
0.1 Course Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
0.2 Instructor Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1 Character Tables 5
1.1 Linear Characters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Counting Conjugacy Classes . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.1 The symmetric group S3 . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.2 Non-Abelian groups of order 8 . . . . . . . . . . . . . . . . . . . . 10
1.3.3 The Alternating Group A4 . . . . . . . . . . . . . . . . . . . . . . 12

2 Tensor Products of Representations 14

2.1 Tensor Products of Vector Spaces . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Direct Product of Groups . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Inner Tensor Products of Representations . . . . . . . . . . . . . . . . . . 21
2.3.1 Symmetric and Alternating Squares . . . . . . . . . . . . . . . . . 22
2.3.2 Character Table of S5 . . . . . . . . . . . . . . . . . . . . . . . . 24

3 Restriction to a Subgroup 27
3.1 Character Table of A5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.1.1 Conjugacy classes in A5 . . . . . . . . . . . . . . . . . . . . . . . 30
3.1.2 Real Character Values . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1.3 Character Table of A5 . . . . . . . . . . . . . . . . . . . . . . . . 32

4 Induced Representations 35
4.1 Definition and Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.2 Frobenius Character Formula . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.1 A group of order 21 . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.2 A group of order p(p − 1) . . . . . . . . . . . . . . . . . . . . . . 46

0.1 Course Structure

2/1: Section 3.1 of [BS] until 3.1.5

4/1: Until Definition 3.1.14

5/1: Until Definition 3.2.1 (avoiding Definition 3.1.16)

 (End of Week 1)

9/1: Completed Chapter 3.

11/1: Started Chapter 4. Completed until Corollary 4.1.9.

12/1: Until Prop 4.2.3.

(End of Week 2)
b for Zn , Zn × Zm and S3 )
16/1: Until Prop 4.2.10 (including examples of G

18/1: Computed G
b for D4 . Then started Section 4.3, and completed until Theorem 4.3.9

19/1: Completed Section 4.3

(End of Week 3)

23/1: Completed until Theorem 4.4.6. Then defined the Fourier coefficient of a function
w.r.t. a representation as in [BS, Definition 5.5.2]. Then proved [T, Lemma 9.4].

25/1: Completed until Theorem 4.4.12, following [T, Theorem 9.3] for the proof of The-
orem 4.4.7.

26/1: Completed Chapter 4. Discussed the character table of Zn , direct product of two
Abelian groups. Also discussed the group structure on G b = Hom(G, S 1 ), and
Pontrjagin duality for a finite Abelian group.
(End of Week 4)

30/1: Discussed linear characters (see additional notes below)

1/2: Discussed a way of counting conjugacy classes, and then determined the character
table for S3 .

2/2: Calculated the character tables for non-abelian groups of order 8, and for A4 .
(End of Week 5)

No classes. Quiz on 9/2/17.

(End of Week 6)

16/2: Started Chapter 5. Completed until Theorem 5.3.5.

(End of Week 7)

20/2: Skipped Section 5.4, and completed Chapter 5.

22/2: Started Chapter 6. Completed until Remark 6.2.2.

23/2: Completed until Corollary 6.2.5. Included [JL, Examples 22.12(i),(ii)].

24/2: Completed until Theorem 6.3.9.

3
 (End of Week 8)

27/2: Completed Chapter 6.

1/3: Started Chapter 7. Completed until Proposition 7.2.7, skipping parts of Section 7.1

2/3: Completed Section 7.1, and until Theorem 7.2.8.

(End of Week 9)

20/3: Completed Chapter 7.

22/3: Tensor products of vector spaces (see additional notes below for the remainder of
the course)

23/3: Direct product of groups

25/3: Inner tensor product of representations from

(End of Week 10)

27/3: Character table of S5 , and started restriction to a subgroup from

29/3: Continued restriction to a subgroup, and started the Character table of A5

30/3: Completed the character table of A5

(End of Week 11)

3/4: Started Induced representations

5/4: Proved the Frobenius Character formula

6/4: Proved Frobenius reciprocity

(End of Week 12)

10/4: Example of group of order 21

12/4: Example of group of order p(p − 1)

13/4: Review.
(End of Week 13)

0.2 Instructor Notes

Given below are some additional notes meant to supplement the material from the text-
book.

4
1 Character Tables
The goal of these notes is to supplement the discussion at the end of [BS, Chapter 4] by
computing the character tables for some non-abelian groups of small order.

1.1 Linear Characters

Remark. [BS, Exercise 4.6] Let G be a group, H C G, and π : G → G/H be the natural
quotient map. Observe that

1. If ρ : G/H → GL(V ) is a representation, then ρ◦π : G → GL(V ) is a representation.

2. If ρ : G/H → GL(V ) and ψ : G → GL(W ) are two representations, then ρ ∼ ψ iff

ρ ◦ π ∼ ψ ◦ π.

3. ρ is irreducible if and only if ρ ◦ π is irreducible.

Hence, we get a well-defined map

[ →G
µ : G/H b

This is injective by (2) above, but not surjective in general.

Theorem 1.1.1. Let G be a group, H C G, and π : G → G/H be the natural quotient

map. If ϕ : G → GL(V ) is a representation such that H ⊂ ker(ϕ), then ∃ a unique
representation ρ : G/H → GL(V ) such that

ρ◦π =ϕ

Proof. If ϕ : G → GL(V ) such that H ⊂ ker(ϕ), then define

ρ : G/H → GL(V ) by gH 7→ ϕ(g)

1. This is well-defined because if g1 H = g2 H, then g2−1 g1 ∈ H, so g2−1 g1 ∈ ker(ϕ) and

hence
ϕ(g1 ) = ϕ(g2 )

2. ρ is a homomorphism because if g1 H, g2 H ∈ G/H, then

ρ(g1 H · g2 H) = ρ(g1 g2 H) = ϕ(g1 g2 ) = ϕ(g1 )ϕ(g2 ) = ρ(g1 H)ρ(g2 H)

3. It is clear that ρ ◦ π = ϕ by definition

5
 4. As for uniqueness, suppose ψ is another function such that ψ ◦ π = ϕ = ρ ◦ π, then
ψ(gH) = ϕ(g) = ρ(gH) for all g ∈ G.

blin for the

Definition 1.1.2. A linear character is a representation of degree 1. Write G
set of all linear characters of G.
Observe that if ϕ : G → C∗ is a linear character, then

G/ ker(ϕ) ∼
= Image(ϕ) < C∗

so G/ ker(ϕ) is an Abelian group.

Definition 1.1.3. If G is any group, and x, y ∈ G, the commutator of x and y is given
by
[x, y] := xyx−1 y −1
The commutator subgroup of G, denoted by [G, G], is the smallest subgroup of G con-
taining the set
S = {[x, y] : x, y ∈ G}
Equivalently,
[G, G] = {u1k u1k . . . ukk : ui ∈ S, i ∈ {±1}}
In fact, we can refine this further. If u = [x, y], then u−1 = [y, x] ∈ S, so

[G, G] = {u1 u2 . . . uk : ui ∈ S}

In other words, [G, G] is the set of all products of commutators in G.

Theorem 1.1.4. Let G be a group, and [G, G] its commutator subgroup.
1. [G, G] C G

2. If H C G such that G/H is Abelian, then [G, G] ⊂ H

3. In particular, G/[G, G] is Abelian.

4. G is Abelian iff [G, G] = {e}

Proof. 1. Note that if x, y ∈ G and g ∈ G, then

g[x, y]g −1 = [gxg −1 , gyg −1 ]

Hence, gSg −1 ⊂ S, and so [G, G] C G by the description of elements of [G, G] given

above.

2. G/H is abelian if and only if

(xH)(yH) = (yH)(xH) ∀x, y ∈ H ⇔ (xy)H = (yx)H ∀x, y ∈ G

This is equivalent to [x, y] ∈ H for all x, y ∈ H, and so [G, G] ⊂ H

6
 3. Follows from (1) and (2).

4. Trivial.

Theorem 1.1.5. Let G := G/[G, G], and let π : G → G denote the natural quotient
map.

1. If ϕ : G → C∗ is a representation, then ϕ ◦ π is a representation of G

2. If ρ : G → C∗ is a linear character, then ∃ϕ : G → C∗ such that ρ = ϕ ◦ π

3. Consider the injective map

µ:Gb →G b
blin .
as described above. Then Image(µ) = G

Proof. 1. By definition

2. If ρ : G → C∗ is a linear character, then G/ ker(ρ) is abelian as mentioned above.

Hence, [G, G] ⊂ ker(ρ) by the previous theorem. Hence, ∃ unique ρ : G → C∗ such
that ρ = ρ ◦ π.

3. The map G b →G b is well-defined and injective as before. Furthermore, if ϕ ∈ G,

b
then dϕ = 1 since G is abelian, so

ϕ : G → C∗

Hence, ϕ ◦ π : G → C∗ is a degree one representation. Equivalently,

blin
µ(ϕ) ∈ G

blin , then ρ = µ(ρ), where ρ is as in part (2). Hence, ρ ∈

Conversely, if ρ ∈ G
Image(µ).

Corollary 1.1.6. The number of linear characters of G is equal to the index of of [G, G]
in G. In particular, this number divides |G|.

Proof. This follows from the above statement and the fact that G is abelian, and so

|G|
b = |G| = [G : [G, G]]

7
1.2 Counting Conjugacy Classes
Lemma 1.2.1. Let H C G, then H is a disjoint union of conjugacy classes in G.

Lemma 1.2.2. Let H C G and π : G → G/H the quotient map. If D ⊂ G/H is a

conjugacy class, then
π −1 (D)
is a disjoint union of conjugacy classes in G. Furthermore, if

1. If D 6= {π(e)}, then π −1 (D) ∩ H = ∅

2. If D1 and D2 are two disjoint conjugacy classes of G/H, then π −1 (D1 ) ∩ π −1 (D2 ) =
∅.

Proof. We wish to show that, if C is a conjugacy class in G, then either

C ∩ π −1 (D) = ∅ or C ⊂ π −1 (D)

By the previous lemma, if H C G, we may write

H = tki=1 Ci

where Ci are conjugacy classes in G, and suppose

G/H = t`j=1 Dj

where Dj are the conjugacy classes in G/H, then for each 1 ≤ j ≤ `. Suppose D1 =
{π(e)}, we write
π −1 (Dj ) = Bj,1 t Bj,2 t . . . t Bj,sj
where Bj,t are conjugacy classes in G. Hence, we get

Lemma 1.2.3. The collection

F = {C1 , C2 , . . . , Ck , B2,1 , B3,1 , . . . , B`,1 }

are disjoint conjugacy classes in G. Hence,

|Cl(G)| ≥ k + ` − 1

Note: A strict inequality may hold above.

8
1.3 Examples
We now construct the character tables for some non-Abelian groups. Given a non-abelian
group G, we will follow these steps:

1. Determine [G, G] by examining normal subgroups H such that G/H is abelian.

2. Determine all linear characters on G by using information from G = G/[G, G]

3. Use the degree formula to enumerate the number and degrees of all irreducible
representations of G.

4. Determine the number of conjugacy classes of G using the previous section, and
also their representatives.

5. Use this to build a partial character table, with some unknown entries.

6. Determine the unknown entries by using the orthogonality relations.

1.3.1 The symmetric group S3

Let G = S3 .

1. Recall that A3 C S3 and S3 /A3 ∼

= Z2 . Hence,

[G, G] ⊂ A3

Since G is non-abelian, [G, G] 6= {e}. Since A3 is cyclic of prime order, we have

[G, G] = A3

2. Since G = G/[G, G] ∼ = Z2 , G has two linear characters obtained by lifting the two
irreducible representations of Z2 .

ρ1 : 1 7→ 1
ρ2 : 1 7→ −1

write ϕi : G → C∗ to be maps, ϕi = ρi ◦ π

3. The degree formula now reads

X
6 = |G| = 2 + n2i
ni >1

Hence, it follows that G has exactly one irreducible representation of degree 2, and
no other representations of higher degree. We denote this representation by ρ.

9
 4. By the previous step, G has 3 conjugacy classes. Notice that H = A3 has is the
union of two conjugacy classes of G.

C1 = {e}, C2 = {(123), (132)}

Also, G/H is abelian, so it has conjugacy classes

D1 = {π(e)}, D2 = {π((12))}

Hence, if F is as in the previous section, then

F = {(e), ((123)), ((12))}

Since |Cl(G)| = 3, it follows that Cl(G) = F.

5. Note that if ρi : Z2 → C∗ is a representation, then

ϕi = ρi ◦ π : G → C∗

is a one-dimensional representation such that

χϕi (g) = χρi (π(g))

So we obtain a partial character table as follows

e (123) (12)
ϕ1 1 1 1
ϕ2 1 1 −1
ρ 2 a b

6. The orthogonality of columns now gives two equations

1 + 1 + 2a = 0 ⇒ a = −1
1 − 1 + 2b = 0 ⇒ b = 0

So the character table of S3 is

e (123) (12)
ϕ1 1 1 1
ϕ2 1 1 −1
χρ 2 −1 0

Note that this agrees with what he had obtained earlier.

1.3.2 Non-Abelian groups of order 8

1. If G is non-Abelian and |G| = 8, then Z(G) 6= {e}, and so |Z(G)| ∈ {2, 4, 8}. Since
G is non-abelian, and
Proposition 1.3.1. If G/Z(G) is cyclic, then G is abelian.

10
 It follows that |Z(G)| = 2 and G/Z(G) ∼ = Z2 × Z2 . In particular, since G/Z(G) is
abelian, it follows that [G, G] ⊂ Z(G). Since [G, G] 6= {e} (since G is non-Abelian),
we have
[G, G] = Z(G)

2. Since G ∼
= Z2 × Z2 , we have 4 irreducible representations of G given by

ρ1 : {(1, 0), (0, 1)} 7→ 1

ρ2 : (1, 0) 7→ 1 and (0, 1) 7→ −1
ρ3 : (1, 0) 7→ −1 and (0, 1) 7→ 1
ρ4 : (1, 0) 7→ −1 and (0, 1) 7→ −1

We write ϕi := ρi ◦ π : G → C∗ .

3. The degree formula gives X

8=4+ n2i
ni >1

Once again, we see that G has exactly one irreducible of representation of degree
> 1. We denote this by ρ, and note that dρ = 2.

4. Since G has 5 irreducible representations, |Cl(G)| = 5. Note that H = Z(G) has 2

conjugacy classes of G, we denote them by

C1 = {e}, C2 = {x}

Since G/H ∼
= Z2 × Z2 , we write

G/H = {π(e), π(g1 ), π(g2 ), π(g3 )}

Each singleton forms a conjugacy class in G/H, so we obtain

F = {{e}, {x}, (g1 ), (g2 ), (g3 )}

Since |Cl(G)| = 5, it follows that Cl(G) = F.

5. Once again, if ϕi = ρi ◦ π, then

χϕi (g) = χρi (π(g))

So we obtain a partial character table as

g 1 x g1 g2 g3
ϕ1 1 1 1 1 1
ϕ2 1 1 -1 1 -1
ϕ3 1 1 1 -1 -1
ϕ4 1 1 -1 -1 1
ρ 2 a b c d

11
 6. Using the orthogonality of columns, we get 4 equations

1 + 1 + 1 + 1 + 2a = 0 ⇒ a = −2
2 − 2 + 2 − 2 + 2b = 0 ⇒ b = 0
2 + 2 − 2 − 2 + 2c = 0 ⇒ c = 0
2 − 2 − 2 + 2 + 2d = 0 ⇒ d = 0

Hence, any two non-Abelian groups of order 8 have the same character table, given
by
g 1 x g1 g2 g3
ϕ1 1 1 1 1 1
ϕ2 1 1 -1 1 -1
ϕ3 1 1 1 -1 -1
ϕ4 1 1 -1 -1 1
χρ 2 -2 0 0 0

In particular, the groups D4 and Q8 are two non-isomorphic groups which have the same
character table.

In fact, more is true: If p is a prime, then any two non-Abelian groups of order p3 have
the same character table. We will prove this later in the course.

1.3.3 The Alternating Group A4

Let G = A4 ,

1. Set H = {e, (12)(34), (13)(24), (14)(23)}. Then H C S4 since it consists of precisely

two conjugacy classes. Hence, H C A4 . Furthermore, G/H is a group of order 4,
and hence is Abelian. By the earlier section,

[G, G] ⊂ H

Since A4 is non-Abelian, [G, G] 6= {e}. However, the non-identity elements in H

form a single conjugacy class in A4 , so since [G, G] C A4 (it must be a union of
conjugacy classes), it follows that [G, G] = H

2. Now G = G/H ∼
= Z3 , so G has 3 linear characters given by

ρi : 1 → ω i−1 , i = 1, 2, 3

where ω = e2πi/3 . Let ϕi = ρi ◦ π

3. Now the degree formula gives 12 = |G| = 3 + di >1 d2i . Hence, G has exactly one
P
more irreducible representation, ρ such that dρ = 3.

12
4. By the previous step, |Cl(G)| = 4. Notice that H is a union of two conjugacy
classes
C1 = {e}, C2 = {(12)(34), (13)(24), (14)(23)}
Also, write
G/H = {π(e), π((123)), π((132)}
then these yield singleton conjugacy classes in G/H. Hence we get

F = {{e}, ((12)(34)), ((123)), ((132))}

Since |Cl(G)| = 4, it follows that Cl(G) = F.

5. As before, the character table now looks like:

g 1 (12)(34) g1 g2
χϕ1 1 1 1 1
χϕ2 1 1 ω ω2
χϕ3 1 1 ω2 ω
χρ 3 a b c
where ω = e2πi/3 .

6. Now the orthogonality of the columns yields

3 + 3a = 0 ⇒ a = −1
1 + ω + ω 2 + 3b = 0 ⇒ b = 0
1 + ω 2 + ω + 3c = 0 ⇒ c = 0

because 1 + ω + ω 2 = 0. This gives the character table of A4 as

g 1 (12)(34) g1 g2
1G 1 1 1 1
χw 1 1 ω ω2
χw 2 1 1 ω2 ω
ρ 3 -1 0 0

13
2 Tensor Products of Representations
Towards the end of the course, we veered away from the textbook completely. I wanted
to cover tensor products, restriction and induction - all topics which, I felt, were covered
poorly in the textbook.

2.1 Tensor Products of Vector Spaces

Let U, V, W, X, etc. denote finite dimensional vector spaces over a field k

Definition 2.1.1. A map f : V × W → X is said to be bilinear if for all αi , βj ∈ k, vi ∈

V, wj ∈ W , we have
!
X X X
f αi vi , βj wj = αi βj f (vi , wj )
i j i,j

Example 2.1.2. 1. If V is an inner product space over R, then the inner product
h·, ·i : V × V → R is bilinear.

2. Cross product R3 × R3 → R3

3. If V is a vector space, and V ∗ its dual, then B : V × V ∗ → k defined by B(v, f ) :=

f (v) is bilinear.

4. ψ : C × Rn → Cn given by (z, v) 7→ (zv1 , zv2 , . . . , zvn )

Definition 2.1.3. 1. Bk (V, W ) is the vector space of all bilinear maps f : V × W → k

2. For v ∈ V, w ∈ W , define v ⊗ w : Bk (V, W ) → k by v ⊗ w(f ) := f (v, w). Notice

that v ⊗ w ∈ Bk (V, W )∗ , the dual space of Bk (V, W )

3. Define V ⊗ W := span{v ⊗ w : v ∈ V, w ∈ W }

Lemma 2.1.4. The map ϕ : V × W → V ⊗ W given by ϕ(v, w) := v ⊗ w is bilinear.

Proof. We prove linearity in the first variable as the other variable is similar. So fix
v1 , v2 ∈ V, w ∈ W, and α ∈ k, and we WTS:

ϕ(αv1 + v2 , w) = αϕ(v1 , w) + ϕ(v2 , w)

14
So fix f ∈ Bk (V, W ), then

ϕ(αv1 + v2 , w)(f ) = f (αv1 + v2 , w)

= αf (v1 , w) + f (v2 , w)
= αϕ(v1 , w)(f ) + ϕ(v2 , w)(f )
= [αϕ(v1 , w) + ϕ(v2 , w)](f )

Theorem 2.1.5. If {vi } and {wj } are bases for V and W respectively, then {vi ⊗ wj } is
a basis for V ⊗ W . In particular, dim(V ⊗ W ) = dim(V ) × dim(W )

Proof. Let S = {vi ⊗ wj : 1 ≤ i ≤ n, 1 ≤ j ≤ m}.

1. S is linearly independent: If αi,j ∈ k such that

X
αi,j vi ⊗ wj = 0 (∗)
i,j

Fix i, j and let fi,j : V × W → k be given by

fi,j (vk , w` ) = δi,k δk,`

extended to a bilinear map on V × W . Then fi,j ∈ Bk (V, W ), and

(vk ⊗ w` )(fi,j ) = fi,j (vk , w` ) = δi,k δk,`

Hence, applying (∗) to fi,j gives

αi,j = 0
This is true for all 1 ≤ i ≤ n, 1 ≤ j ≤ m, so S is linearly independent.

2. S spans V × W : By definition,

V ⊗ W := span{v ⊗ w : v ∈ V, w ∈ W }

so it suffices to show that v ⊗ w ∈ span(S) for any v ∈ V, w ∈ W . So fix v ∈ V, w ∈

W , then write X X
v= αi vi and w = βj wj
i j

Then since the map (v, w) 7→ v ⊗ w is bilinear, we get

X
v⊗w = αi βj vi ⊗ wj ∈ span(S)
i,j

15
Proposition 2.1.6 (Universal Property - I). If X is a finite dimensional vector space,
and g : V × W → X is a bilinear map, then ∃!T : V ⊗ W → X linear such that T ◦ ϕ = g.
In other words, there is an isomorphism

BX (V, W ) ∼
= Homk (V ⊗ W, X)

Proof. If g : V × W → X is bilinear, define

T : V ⊗ W → X given by T (vi ⊗ wj ) = g(vi , wj )

extended linearly to V ⊗ W . This is well-defined by the previous theorem. Furthermore,

T is linear and
T ◦ ϕ(vi , wj ) = g(vi , wj )
Since both sides are bilinear, they must agree on V × W .

For uniqueness, note that if S : V ⊗ W → X is a linear map such that

S◦ϕ=g

Then
S(vi ⊗ wj ) = g(vi , wj ) = T (vi ⊗ wj ) ∀i, j
Since S and T are linear, it follows that S = T by the previous theorem.

Theorem 2.1.7 (Universal Property - II). Let U be a finite dimensional vector space
and ψ : V × W → U is a bilinear map such that, for any bilinear map h : V × W → X,
∃!S : U → X such that S ◦ ψ = h, then there is an isomorphism µ : U → V ⊗ W such
that µ ◦ ψ = ϕ

Proof. Let (U, ψ) be a pair as above. By the previous theorem (V ⊗ W, ϕ) is another pair
that satisfies the same property. By the previous theorem, ∃T : V ⊗ W → U such that

T ◦ϕ=ψ

Similarly, ∃S : U → V ⊗ W such that

S◦ψ =ϕ

Hence,
S◦T ◦ϕ=ϕ and T ◦S◦ψ =ψ
By the uniqueness, it follows that S ◦ T = idV ⊗W . Similarly,

T ◦ S = idU

and hence S is the required isomorphism.

Example 2.1.8. 1. C ⊗ Rn ∼
= Cn

16
 Proof. Define ψ : C × Rn → Cn by
(z, v) = (zv1 , zv2 , . . . , zvn )
This is a bilinear map. Hence, ∃T : C ⊗ Rn → Cn such that
T (z ⊗ v) = ψ(z, v)
Now note that
ψ(1, ei ) = ei
so ψ is surjective. Hence, T is surjective. However,
dim(Cn ) = 2n = dim(C) × dim(Rn ) = dim(C ⊗ Rn )
and so T must be injective and hence an isomorphism.
2. Cn ⊗ Cm ∼
= Cnm
Proof. Define ψ : Cn × Cm → Cnm by
ψ(x, y) = (x1 y1 , x1 y2 , . . . , x1 ym , x2 y1 , . . . , x2 ym , . . . , xn ym )
Then follow the argument as above.
3. V ⊗ V ∗ ∼
= Endk (V )
Proof. Define ψ : V × V ∗ → Endk (V ) by
ψ(v, f )(w) = f (w)v
Then ψ is bilinear, so follow a similar argument as above.
Definition 2.1.9. Let T : V1 → V2 and S : W1 → W2 be two linear maps. Then define
ψ : V1 × W1 → V2 ⊗ W2 by ψ(v, w) = T (v) ⊗ S(w)
Then ψ is clearly bilinear. So ∃!R : V1 ⊗ W1 → V2 ⊗ W2 such that
R(v ⊗ w) = T (v) ⊗ S(w) ∀v ∈ V1 , w ∈ W1
We write R = T ⊗ S

2.2 Direct Product of Groups

Theorem 2.2.1. Let ρ : G → GL(V ) and π : H → GL(W ) be two representations. Then
∃ a unique representations
ψ : G × H → GL(V ⊗ W )
such that
ψ(g, h)(v ⊗ w) = ρg (v) ⊗ πh (w)
This is called the outer tensor product of ρ and π and we write ψ = ρ⊗π
b

17
Proof. 1. For each (g, h) ∈ G × H fixed, define

ϕ : V × W → V ⊗ W given by ϕ(v, w) = ρg (v) ⊗ πh (w)

This map is clearly bilinear, so ∃ a unique linear map

R(g,h) : V ⊗ W → V ⊗ W such that R(g,h) (v ⊗ w) = ρg (v) ⊗ πh (w)

So we define ψ(g, h) := R(g,h)

2. We first check that ψ is well-defined: To see this, note that

R(g−1 ,h−1 ) (v ⊗ w) = ρg−1 (v) ⊗ πh−1 (w)

Hence, for any v ∈ V, w ∈ W , we have

R(g,h) ◦ R(g−1 ,h−1 ) (v ⊗ w) = v ⊗ w = R(g−1 ,h−1 ) ◦ R(g,h) (v ⊗ w)

But V ⊗ W = span{v ⊗ w : v ∈ V, w ∈ W }, so since both sides are linear maps, we

see that
R(g,h) ◦ R(g−1 ,h−1 ) = I = R(g−1 ,h−1 ) ◦ R(g,h)
Hence, R(g,h) ∈ GL(V ⊗ W )

3. Now we check that ψ is a homomorphism: As above, it suffices to show that

R(g1 ,h1 ) ◦ R(g2 ,h2 ) (v ⊗ w) = R(g1 g2 ,h1 h2 ) (v ⊗ w) ∀v ∈ V, w ∈ W

This follows from the definition and the fact that ρ and π are representations.

4. Uniqueness follows from the uniqueness of the previous definition.

Theorem 2.2.2. With the notation as above,

χρ⊗π
b (g, h) = χρ (g)χπ (h)

Proof. Fix (g, h) ∈ G × H. Since ρg is diagonalizable, ∃ a basis S = {v1 , v2 , . . . , vn } of V

such that
ρg (vi ) = λi vi ∀1 ≤ i ≤ n
Similarly, ∃ a basis T = {w1 , w2 , . . . , wm } of W such that

πh (wj ) = µj wj ∀1 ≤ j ≤ m

Let B = {vi ⊗ wj : 1 ≤ i ≤ n, 1 ≤ j ≤ m}, then B is a basis for V ⊗ W . Furthermore, if

ψ = ρ ⊗ π, then
ψ(g,h) (vi ⊗ wj ) = λi µj (vi ⊗ wj )

18
Taking a trace, we get
X
χψ (g, h) = λi µj
i,j
! !
X X
= λi µj
i j

= χρ (g)χπ (h)

Theorem 2.2.3. Let ρi : G → GL(Vi ) and πi : H → GL(Wi ) for i = 1, 2. If ψi = ρi ⊗π

b i,
then
hχψ1 , χψ2 iL(G×H) = hχρ1 , χρ2 iL(G) hχπ1 , χπ2 iL(H)
Proof. We compute
1 X
hχψ1 , χψ2 iL(G×H) = χψ1 (g, h)χψ2 (g, h)
|G × H|
(g,h)∈G×H
1 X
= χρ (g)χπ1 (h)χρ2 (g)χπ2 (h)
|G||H| g∈G,h∈H 1
! !
1 X 1 X
= χρ (g)χρ2 (g) χπ (h)χπ2 (h)
|G| g∈G 1 |H| h∈H 1
= hχρ1 , χρ2 iL(G) hχπ1 , χπ2 iL(H)

Corollary 2.2.4. 1. Let ρ : G → GL(V ) and π : H → GL(W ). Then ρ⊗π

b is
irreducible if and only if both ρ and π are irreducible.
2. Let ρi : G → GL(Vi ) and πi : H → GL(Wi ) be irreducible. Then ρ1 ∼ ρ2 and
π1 ∼ π2 if and only if
ρ1 ⊗π
b 1 ∼ ρ2 ⊗π
b 2
Proof. 1. Recall that if ϕ is any representation of a group, then
hχϕ , χϕ i ≥ 1
and equality holds if and only if ϕ is irreducible. Now simply apply the previous
theorem.
2. Note that for any (g, h) ∈ G × H
χψi ((g, h)) = χρi (g)χπi (h)
Hence, if ρ1 ∼ ρ2 and π1 ∼ π2 , it follows that
χψ1 = χψ2
and so ψ1 ∼ ψ2 .

19
 3. Conversely, if ψ1 ∼ ψ2 , then by part (1)

hχψ1 , χψ2 i = 1

From this it follows that

hχρ1 , χρ2 i = hχπ1 , χπ2 i = 1

By Schur orthogonality, it follows that ρ1 ∼ ρ2 and π1 ∼ π2 .

Theorem 2.2.5. The map

b×H
α:G b → G\
× H given by ([ρ], [π]) 7→ [ρ⊗π]
b

is a well-defined bijection.

Proof. 1. α is well-defined by the previous Corollary

2. To see that α is injective by the previous corollary, part 2.

3. To see that α is surjective, we show that

|Cl(G)||Cl(H)| = |Cl(G × H)|

If (g, h), (x, y) ∈ G × H, then

(x, y)−1 (g, h)(x, y) = (x−1 gx, y −1 hy)

Hence, (g, h) ∼ (g 0 , h0 ) if and only if g ∼ g 0 and h ∼ h0 . Hence the map

α : Cl(G) × Cl(H) → Cl(G × H) given by ([g], [h]) 7→ [(g, h)]

is a well-defined bijection.

Example 2.2.6. We determine the character table of S3 × Z2 . We have the character

table of G = S3 as
e (12) (123)
χ1 1 1 1
χ2 1 -1 1
χ3 2 0 -1

and that of Z2 is given by

0 1
χ1 1 1
χ2 1 -1

20
Now the representatives of the conjugacy classes of S3 × Z2 are

{(e, 0), (e, 1), ((12), 0), ((12), 1), ((123), 0), ((123), 1)}

We multiply characters to get the character table of S3 × Z2 to be

(e, 0) (e, 1) ((12), 0) ((12), 1) ((123), 0) ((123), 1))
χ1 × χ1 1 1 1 1 1 1
χ1 × χ2 1 -1 1 -1 1 -1
χ2 × χ1 1 1 -1 -1 1 1
χ2 × χ2 1 -1 -1 1 1 -1
χ3 × χ1 2 2 0 0 -1 -1
χ3 × χ2 2 -2 0 0 -1 1

Compare this with the discussion in [BS, Section 4.5]. This is, in fact, the tensor product
of two square matrices representing the character tables of S3 and Z2 .

2.3 Inner Tensor Products of Representations

Theorem 2.3.1. Let ρ : G → GL(V ) and π : G → GL(W ) be two representations of a
group G. Then ∃! representation ϕ : G → GL(V ⊗ W ) such that

ϕg (v ⊗ w) = ρg (v) ⊗ πg (w)

This is called the inner tensor product of ρ and π and is denote by by ρ ⊗ π.

Proof. Consider the outer tensor product

ρ⊗π
b : G × G → GL(V ⊗ W )

and the diagonal homomorphism ∆ : G → G × G given by g 7→ (g, g). Then define

ϕ = (ρ⊗π)
b ◦∆

Then ϕ satisfies the required condition. Uniqueness also holds as before.

Theorem 2.3.2. If ρ, π as above, then

χρ⊗π (g) = χρ (g)χπ (g) ∀g ∈ G

In particular, the product of two characters is a character.

Proof. By the earlier theorem,

χρ⊗π (g) = χψ (g, g) = χρ (g)χπ (g)

21
Example 2.3.3. The character table of S4 described in [BS, Example 7.2.13] is given
below. Let π denote the augmentation representation of S4 and ρ the irreducible repre-
sentation of degree 2.
1 (12) (123) (1234) (12)(34)
χ1 1 1 1 1 1
χ2 1 -1 1 -1 1
χ3 = χπ 3 1 0 -1 -1
χ4 = χ2 χ3 3 -1 0 1 -1
χ5 = χρ 2 0 -1 0 2
χρ χπ 6 0 0 0 -2
χ2 χρ 2 0 -1 0 2
Hence,

1. If η = ρ ⊗ π, then η has degree 6. In particular, η is not irreducible, so the inner

tensor product of irreducible representations need not be irreducible.

2. Also, if µ = χ2 ⊗ ρ, then

χµ (g) = sgn(g)χρ (g) = χρ (g)

since χρ (g) = 0 for all g ∈

/ A4 . Hence, µ ∼ ρ. In particular,

χ2 ⊗ ρ ∼ χ1 ⊗ ρ

but χ2 is not equivalent to χ1 .

Compare these examples with Corollary 2.2.4.

2.3.1 Symmetric and Alternating Squares

Definition 2.3.4. Let V be a vector space, then ∃! linear map T : V ⊗ V → V ⊗ V such
that
T (v ⊗ w) = w ⊗ v
Write
S 2 (V ) = {x ∈ V ⊗ V : T x = x}
A2 (V ) = {x ∈ V ⊗ V : T x = −x}

Lemma 2.3.5. 1. V ⊗ V = S 2 (V ) ⊕ A2 (V )

2. Let ρ : G → GL(V ) be a representation. Write ϕ = ρ ⊗ ρ. If T as above, then

T ϕg = ϕg T ∀g ∈ G

Proof. HW

22
Definition 2.3.6. Let ρ : G → GL(V ) be a representation of G. Then by the previous
two lemmas, we may define

ρS = (ρ ⊗ ρ)|S 2 (V ) and ρA = (ρ ⊗ ρ)|A2 (V )

Then
ρ ⊗ ρ ∼ ρS ⊕ ρA
These are called the symmetric square and alternating square of ρ respectively.

Lemma 2.3.7. Let {v1 , v2 , . . . , vn } be a basis for V . THen

1. {vi ⊗ vj + vj ⊗ vi : 1 ≤ i ≤ j ≤ n} is a basis for S 2 (V )

2. dim(S 2 (V )) = n(n + 1)/2

3. {vi ⊗ vj − vj ⊗ vi : 1 ≤ i ≤ n} is a basis for A2 (V )

4. dim(A2 (V )) = n(n − 1)/2

Proof. Let S = {vi ⊗ vj + vj ⊗ vi : 1 ≤ i ≤ j ≤ n}, then S ⊂ S 2 (V ). Similarly, if

T = {vi ⊗ vj − vj ⊗ vi : 1 ≤ i < j ≤ n}, then T ⊂ A2 (V ). Furthermore, S and T are
linearly independent since the set {vi ⊗ vj : 1 ≤ i, j ≤ n} is linearly independent. Hence,

dim(S 2 (V )) ≥ n(n + 1)/2 and dim(A2 (V )) ≥ n(n − 1)/2

However,
dim(S 2 (V )) + dim(A2 (V )) = dim(V ⊗ V ) = n2
So both the above inequalities are equalities and the results follow.

Proposition 2.3.8. Let ρ : G → GL(V ) be a representation with character χ. Suppose

χS and χA denote the characters of ρS and ρA respectively, then
1
χS (g) = (χ2 (g) + χ(g 2 )) ∀g ∈ G
2
1
χA (g) = (χ2 (g) − χ(g 2 )) ∀g ∈ G
2
Proof. Fix g ∈ G, then ρg is diagonalizable. So choose a basis {v1 , v2 , . . . , vn } of V such
that
ρg (vi ) = λi vi ∀1 ≤ i ≤ n
Hence,
n
X n
X
2
χ(g) = λi and χ(g ) = λ2i
i=1 i=1

If wi,j = vi ⊗ vj + vj ⊗ vi , then

ρS (g)(wi,j ) = ρg (vi ) ⊗ ρg (vj ) + ρg (vj ) ⊗ ρg (vi ) = λi λj wi,j

23
Similarly, taking ti,j = vi ⊗ vj − vj ⊗ vi , then

ρA (g)(ti,j ) = λi λj ti,j

Hence,
X
χS (g) = λi λj
1≤i≤j≤n
X
χA (g) = λi λj
1≤i<j≤n
Xn
⇒ χS (g) = λ2i + χA (g)
i=1
2
= χ(g ) + χA (g)
2
⇒ χ(g ) = χS (g) − χA (g) (∗)

Also,
n
!2 n
X X X
2
χ(g) = λi = λ2i + 2 λi λj
i=1 i=1 i<j
2
= χ(g ) + 2χA (g)
2
⇒ χ(g) = χS (g) + χA (g) (∗∗)

Solving (∗) and (∗∗) gives the required result.

2.3.2 Character Table of S5

We now determine the character table of S5 . Let G = S5

1. As done for S4 , we see that [G, G] = A5 . Hence, G has two linear characters

χ1 and χ2 = sgn

2. The augmentation representation ρ is a degree 4 irreducible representation with

character
χ3 (g) = |Fix(g)| − 1

3. Let ϕg = χ2 (g)ρg is another irreducible degree 4 representation with character

χ4 (g) = sgn(g)(|Fix(g)| − 1)

4. The conjugacy classes of S5 are given as

e (12) (123) (12)(34) (1234) (123)(45) (12345)
1 10 20 15 30 20 24

24
 Hence, S5 has 7 irreducible representations. We have determined 4 so far, so we
have a partial character table as below
e (12) (123) (12)(34) (1234) (123)(45) (12345)
χ1 1 1 1 1 1 1 1
χ2 1 -1 1 1 -1 -1 1
χ3 4 2 1 0 0 -1 -1
χ4 4 -2 1 0 0 1 -1

5. Let ρ be as above, then if χS and χA are the characters of the symmetric and
alternating squares of ρ, then we can obtain their values by the previous theorem.
For instance,
1 1 1
χS ((123)) = (χ((123))2 + χ((123)2 )) = (12 + χ((132))) = (1 + 1) = 1
2 2 2
Similarly, we obtain the values of χS and χA as below
e (12) (123) (12)(34) (1234) (123)(45) (12345)
χS 10 4 1 2 0 1 0
χA 6 0 0 -2 0 0 1
6. Now,
1
hχA , χA i = [(1 · 36) + (20 · 0) + (15 · 4) + (30 · 0) + (20 · 0) + (24 · 1)] = 1
120
So χA is the character of an irreducible representation. This must necessarily be
different from the ones already obtained since it has degree 6. We write χ5 = χA .

7. Now,
hχS , χS i = 3
so it does not correspond to an irreducible representations, but calculating inner
products gives

hχS , χi i = 1 i ∈ {1, 3}, and

hχS , χj i = 0 j ∈ {2, 4, 5}

Hence, ∃ a sixth irreducible representation ψ such that

χS = χ1 + χ3 + χψ

We write χ6 = χψ and note that

χ6 = χS − χ1 − χ3 (∗)

In particular, χ6 (1) = 10 − 1 − 4 = 5.

25
8. Now if ϕg = χ2 (g)ψg , then ϕ is an irreducible representation of degree 5 such that

χϕ (g) = χ2 (g)χψ (g) (∗∗)

In this case, using equation (∗), we see that

χψ ((12)) = χS ((12)) − χ1 ((12)) − χ3 ((12)) = 4 − 1 − 2 = 1 6= χϕ ((12))

Hence, ϕ is not equivalent to ψ. We write χ7 = χϕ , so equations (∗) and (∗∗) allow

us to complete the character table of S5 .
e (12) (123) (12)(34) (1234) (123)(45) (12345)
χ1 1 1 1 1 1 1 1
χ2 1 -1 1 1 -1 -1 1
χ3 4 2 1 0 0 -1 -1
χ4 4 -2 1 0 0 1 -1
χ5 6 0 0 -2 0 0 1
χ6 5 1 -1 1 -1 1 0
χ7 5 -1 -1 1 1 -1 0

26
3 Restriction to a Subgroup
Definition 3.0.1. Let G be a group, H < G and ρ : G → GL(V ) be a representation.
We may restrict ρ to obtain a representation
ρ|H : H → GL(V )
This is called the restriction of ρ to H.
Note that even if ρ is irreducible, ρ|H may not be.
Proposition 3.0.2. Suppose ∃H < G such that H is Abelian, then
dρ ≤ [G : H] ∀ρ ∈ G
b

Proof. Let ρ : G → GL(V ) be irreducible and d = dρ , then ρ|H : H → GL(V ) is a

representation. Hence, ∃ one dimensional representations ϕ1 , ϕ2 , . . . , ϕd : H → C∗ such
that
ρ ∼ ϕ1 ⊕ ϕ2 ⊕ . . . ⊕ ϕd
In particular, ∃W < V such that dim(W ) = 1, which is invariant under ρ|H . Write
W = span{v}, and set
W 0 = span{ρg (v) : g ∈ G}
Then, W 0 < V is ρ(G)-invariant. Since ρ is irreducible,
V = W 0 = span{ρg (v) : g ∈ G}
Now suppose g ∈ G, h ∈ H, then
ρ(gh)(v) = ρ(g)ρ(h)(v) = λρ(g)(v)
and so ρ(gh)(v) ∈ span{ρ(g)(v)}. Hence, if G/H = {g1 H, g2 H, . . . , g` H}, with ` = [G :
H], then
V = span{ρ(gi )v : 1 ≤ i ≤ `}
In particular, d = dim(V ) ≤ ` = [G : H].
Example 3.0.3. Let G = Dn be the dihedral group of order 2n. Then any irreducible
representation of G has degree 1 or 2. If n = p, prime, we describe all the irreducible
representations of G.
1. Write G = Dp = ha, b : ap = b2 = 1, bab = ap−1 i, and H = hai. Then H C G and
G/H ∼= Z2 , so [G, G] ⊂ H. However, |H| = p and G is non-abelian, so [G, G] = H.
Hence, G has exactly two linear characters given by
χ1 : a 7→ 1, b 7→ 1
χ2 : a 7→ 1, b 7→ −1

27
 2. Since every other irreducible representation has degree 2, the degree formula gives

2p = 2 + 4k ⇒ k = (p − 1)/2

and so G has exactly (p − 1)/2 irreducible representations of degree 2.

3. For 1 ≤ j ≤ (p − 1)/2, define ψj : G → GL2 (C) by

 j   
ζ 0 0 1
a 7→ and b 7→
0 ζ −j 1 0

As in the earlier HW, ψj is irreducible. Furthermore, if ζ j = ζ ±i , then

p | (j ± i)

This is impossible if 1 ≤ i, j ≤ (p − 1)/2, and so for such i, j, we have that ψj (a)

and ψi (a) have different eigen-values. In particular, ψj is not equivalent to ψi .

4. Thus, the irreducible representations of G are

b = {χ1 , χ2 , ψj : 1 ≤ j ≤ (p − 1)/2}
G

Definition 3.0.4. As observed above, even if ρ : G → GL(V ) is irreducible, its restriction

ρ|H : H → GL(V ) may not be irreducible. Write H b = {ψ1 , ψ2 , . . . , ψr }, and set

si = hχρ|H , χψi iL(H)

Then si are the multiplicities of ψi in ρ|H . We say that ψi is a constituent of ρ|H if si 6= 0.

Note that r
X
χρ|H = si χψi
i=1

Theorem 3.0.5. Let H < G, and let ψ : H → GL(W ) be a non-zero representation of

H. Then ∃ an irreducible representation ρ : G → GL(V ) such that

hχρ|H , χψ iL(H) 6= 0

In particular, every irreducible representation of H occurs as a constituent of an irre-

ducible representation of G.
b = {ϕ(1) , ϕ(2) , . . . , ϕ(s) }, di = dϕ(i) , and let χi = χϕ(1) . Let L : G →
Proof. Write G
GL(L(G)) denote the left regular representation, then
s
X
χL = di χi
i=1

28
Let ψ : H → GL(W ) as above, then
s
X
di hχϕ(i) |H , χψ i = hχL |H , χψ iL(H)
i=1
1 X
= χL (h)χψ (h)
|H| h∈H
1 |G|
= χL (e)χψ (e) = dψ 6= 0
|H| |H|
Hence, ∃1 ≤ i ≤ s such that
hχϕ(i) |H , χψ iL(H) 6= 0

Proposition 3.0.6. Let H < G, ρ : G → GL(V ) be an irreducible representation of G.

b = {ψ1 , ψ2 , . . . , ψr }, and write
Let H

si = hχρ|H , χψi iL(H)

Then r
X
s2i ≤ [G : H]
i=1

and equality holds if and only if

χρ (g) = 0 ∀g ∈ G \ H

Proof. We know that

r
X 1 X
s2i = hχρ|H , χρ|H iL(H) = χρ (h)χρ (h)
i=1
|H| h∈H

Since ρ is irreducible on G, we have

1 X
1 = hχρ , χρ iL(G) = χρ (g)χρ (g)
|G| g∈G
1 X 1 X
= χρ (h)χρ (h) + χρ (g)χρ (g)
|G| h∈H |G|
g∈G\H
r
|H| X
= s2i + K
|G| i=1

Hence,
r
X (1 − K)|G|
s2i =
i=1
|H|
Note that K ≥ 0 and K = 0 if and only if χρ (g) = 0 for all g ∈ G \ H. This gives the
result.

29
Corollary 3.0.7. Let H < G be a subgroup of index 2, and let ρ : G → GL(V ) be an
irreducible representation of G. Then one of the following happens:

1. ρ|H is an irreducible representation of H.

2. ∃ψ1 , ψ2 ∈ H
b such that ρ|H = ψ1 ⊕ ψ2 .

Furthermore, part (2) occurs if and only if χρ (g) = 0 for all g ∈ G \ H.

3.1 Character Table of A5

3.1.1 Conjugacy classes in A5
Definition 3.1.1. Let G be a group and x ∈ G.

1. The conjugacy class of x in G is denoted by xG = {yxy −1 : y ∈ G}

2. The centralizer of x in G is

CG (x) = {y ∈ G : yx = xy} = {y ∈ G : yxy −1 = x}

Note that if we let G act on itself by conjugation, then the conjugacy class of G is the
orbit of x, while the centralizer of x is the stabilizer of x. So by the orbit-stabilizer
theorem,
|xG | = [G : CG (x)]
Now, for any σ ∈ An , write
σ Sn and σ An
to denote the conjugacy classes of σ in Sn and An respectively. Clearly,

σ An ⊂ σ Sn

Note that since An C Sn , we have σ Sn ⊂ An

Proposition 3.1.2. For σ ∈ An with n > 1, we have

1. If σ commutes with an odd permutation, then σ An = σ Sn

2. If σ does not commute with some odd permutation, then

σ Sn = σ An t ((12)σ(12))An

and
|σ Sn |
|σ An | = |((12)σ(12))An | =
2

30
Proof. 1. Suppose τ ∈ Sn is an odd permutation which commutes with σ, then we
WTS: σ Sn ⊂ σ An . So fix η ∈ σ Sn and δ ∈ Sn such that

η = δσδ −1

If δ ∈ An , then η ∈ σ An . If not, then δ 0 = δτ ∈ An and

δ 0 σδ 0−1 = δσδ −1 = η ⇒ η ∈ σ An

2. Suppose σ does not commute with any odd permutation. Then, by definition,

CSn (σ) = CAn (σ)

Hence,
An |An | |Sn | |σ Sn |
|σ | = [An : CAn (σ)] = = =
|CAn (σ)| 2|CSn (σ)| 2
Now observe that

σ Sn = {δσδ −1 : δ ∈ An } t {δσδ −1 : δ ∈ Sn \ An }

Now δ is odd if and only if η = δ(12) ∈ An . Hence,

{δσδ −1 : δ ∈ Sn \ An } = {η(12)σ(12)η −1 : η ∈ An } = ((12)σ(12))An

The theorem now follows.

We now examine the conjugacy classes in S5

e (12) (123) (12)(34) (1234) (123)(45) (12345)

1 10 20 15 30 20 24

Of these, (12) ∈
/ A5 , (1234) ∈
/ A5 , (123)(45) ∈
/ A5 . Also,

(45)(123)(45) = (123) ⇒ (123)S5 = (123)A5

(12)(12)(34)(12) = (12)(34) ⇒ ((12)(34))S5 = ((12)(34))A5

and
120
CSn ((12345)) = =5
24
is not divisible by two. Hence, (12345)An 6= (12345)Sn . Hence,

(12)(12345)(12) = (13452)

is another representative of a conjugacy class in A5 . So we get the conjugacy classes in

A5 are
e (123) (12)(34) (12345) (13452)
1 20 15 12 12

31
3.1.2 Real Character Values
Lemma 3.1.3. If ρ : G → GL(V ) is a unitary representation and g ∈ G, then

χρ (g −1 ) = χρ (g)

Proof. Since ρg is diagonalizable, write

[ρg ]B = diag(λ1 , λ2 , . . . , λn )

where each λi ∈ S 1 . Hence,

[ρ−1 −1 −1
g ]B = diag(λ1 , . . . , λn )

But λ−1
i = λ1 , so the result follows by taking traces.

Theorem 3.1.4. Let G be a group and g ∈ G. If g is conjugate to g −1 if and only if

χρ (g) ∈ R for all ρ ∈ G
b

Proof. By the previous lemma

χρ (g −1 ) = χρ (g) ⇔ χρ (g) ∈ R

So the corollary follows from Mid-Sem Exam, Problem 2.

Corollary 3.1.5. For very representation ρ of A5 , χρ (g) ∈ R
Proof. It suffices to show that every element in {e, (123), (12)(34), (12345), (13452)} is
conjugate to its own inverse. This is evident for elements in {e, (123), (12)(34)}. For the
other two, check that

(12345)−1 = (54321) = (15)(24)(12345)(15(24)

and
(13452)−1 = (25431) = (12)(35)(13452)(12)(35)

3.1.3 Character Table of A5

Now consider the character table of S5 obtained in the previous section.

e (12) (123) (12)(34) (1234) (123)(45) (12345)

χ1 1 1 1 1 1 1 1
χ2 1 -1 1 1 -1 -1 1
χ3 = χρ 4 2 1 0 0 -1 -1
χ4 = χχ2 ⊗ρ 4 -2 1 0 0 1 -1
χ5 = χρA 6 0 0 -2 0 0 1
χ6 = χψ 5 1 -1 1 -1 1 0
χ7 = χχ2 ⊗ψ 5 -1 -1 1 1 -1 0

32
Restricting to H = A5 , we see that

1. χ1 ((12)) 6= 0, so χ1 |H is irreducible.

χ2 ((12)) 6= 0, so χ2 |H is irreducible. However, χ2 |H = χ1 |H

2. χ3 ((12)) 6= 0, so ρ|H is irreducible.

χ4 ((12)) 6= 0, so (χ2 ⊗ ρ)|H is irreducible. However,

χ4 (g) = χ3 (g) ∀g ∈ A5

so ρ|H ∼ (χ2 ⊗ ρ)|H

3. χ5 (g) = 0 for all g ∈ S5 \ A5 , so ρA = ψ1 ⊕ ψ2 for two irreducible representations

ψ1 and ψ2 of A5

4. χ6 ((12)) 6= 0, so ψ|H is irreducible.

As above, ψ|H ∼ (χ2 ⊗ ψ)|H .

So we obtain a partial character table

e (123) (12)(34) (12345) (13452)

1 20 15 12 12
ϕ1 = χ1 |H 1 1 1 1 1
ϕ2 = χ3 |H 4 1 0 -1 -1
ϕ3 = χ6 |H 5 -1 1 0 0
ϕ4 = χψ1 n1 a1 a2 a3 a4
ϕ5 = χψ2 n2 b1 b2 b3 b4

Note that if ni = dψi , i = 1, 2, then

1 + 16 + 25 + n21 + n22 = 60 ⇒ n21 + n22 = 18 ⇒ n1 = n2 = 3

Furthermore,
χψ1 + χψ2 = χρA
Hence, we get

a1 + b1 = χρA ((123)) = 0 ⇒ b1 = −a1

a2 + b2 = χρA ((12)(34)) = −2 ⇒ b2 = −2 − a2
a3 + b3 = χρA ((12345)) = 1 ⇒ b3 = 1 − a3
⇒ b4 = 1 − a4

So we get an incomplete table as

33
 e (123) (12)(34) (12345) (13452)
1 20 15 12 12
ϕ1 = χ1 |H 1 1 1 1 1
ϕ2 = χ3 |H 4 1 0 -1 -1
ϕ3 = χ6 |H 5 -1 1 0 0
ϕ4 = χψ1 3 a1 a2 a3 a4
ϕ5 = χψ2 3 −a1 −2 − a2 1 − a3 1 − a4

Orthonormality of columns gives

20
[1 + 1 + 1 + a21 + a21 ] = 1 ⇒ a1 = 0
60
15
[1 + 1 + a22 + b22 ] = 1 ⇒ a22 + b22 = 2
60
12
[1 + 1 + a23 + b23 ] = 1 ⇒ a23 + b23 = 3
60
12
[1 + 1 + a24 + b24 ] = 1 ⇒ a24 + b24 = 3
60
Since b2 = −2 − a2 and a22 + b22 = 2, it follows that

a2 = b2 = −1

Now since b3 = (1 − a3 ), we see that a3 and a4 are both solutions to the equation
√
2 1± 5
x −x−1=0⇒x=
2
Since ϕ4 6= ϕ5 , the character table of A5 is
e (123) (12)(34) (12345) (13452)
1 20 15 12 12
ϕ1 = χ1 |H 1 1 1 1 1
ϕ2 = χ3 |H 4 1 0 -1 -1
ϕ3 = χ6 |H 5 -1 1 0 0
ϕ4 = χψ1 3 0 −1 x y
ϕ5 = χψ2 3 0 −1 y x
√ √
1+ 5 1− 5
where x = 2
and y = 2

34
4 Induced Representations
4.1 Definition and Examples
Definition 4.1.1. Let G be a group and H < G. Let ρ : H → GL(W ) be a representa-
tion.
1. Define X = {f : G → W }. Note that X is a vector space under the pointwise
operations. Define
I(W ) := {f ∈ X : f (gh) = ρh−1 (f (g)) ∀g ∈ G, h ∈ H}
Note that I(W ) is a vector subspace of X.
2. For g ∈ G, define
Tg : I(W ) → I(W ) given by Tg (f )(x) := f (g −1 x)
Then Tg is well-defined
Proof. If f ∈ I(W ), then for any h ∈ H, x ∈ G,
Tg (f )(xh) = f (g −1 xh) = ρh−1 (f (g −1 x)) = ρh−1 Tg (f )(x)
Hence, Tg (f ) ∈ I(W )
3. Moreover, Tg ∈ GL(I(W ))
Proof. Simply check that
Tg ◦ Tg−1 (f )(x) = Tg−1 (f )(g −1 x) = f (gg −1 x) = f (x) ∀x ∈ G, f ∈ V
Hence, Tg ◦ Tg−1 = idI(W ) . Similarly, Tg−1 ◦ Tg = idI(W )
4. Finally, the map ϕ : G → GL(I(W )) given by
ϕ(g) = Tg
is a representation of G.
Proof. For g1 , g2 ∈ G, f ∈ I(W ), and x ∈ G, we have
(Tg1 ◦ Tg2 )(f )(x) = Tg2 (f )(g1−1 x)
= f (g2−1 g1−1 x)
= f ((g1 g2 )−1 x)
= Tg1 g2 (f )(x)

35
The representation ϕ : G → GL(I(W )) is called the induced representation of ρ : H →
GL(W ), and is denoted by ϕ = IndG
H (ρ).

Proposition 4.1.2. dim(I(W )) = dim(W )[G : H]

Proof. Write G/H = {x1 H, x2 H, . . . , x` H}, so that ` = [G : H]. Define a map

T : I(W ) → ⊕`i=1 W given by f 7→ (f (x1 ), f (x2 ), . . . , f (x` ))

T is clearly linear. We claim that T is bijective, which proves the theorem.

1. T is injective: Suppose T (f ) = 0, then f (xi ) = 0 for all 1 ≤ i ≤ `. Then if

g ∈ G, ∃1 ≤ i ≤ ` and h ∈ H such that g = xi h. Hence,

f (g) = f (xi h) = ρh−1 f (xi ) = 0

Hence, f = 0

2. T is surjective: Given (w1 , w2 , . . . , w` ) ∈ ⊕`i=1 W , define f : G → W such that

f (xi h) = ρh−1 (wi ) ∀h ∈ H, 1 ≤ i ≤ `

This is well-defined since G = t`i=1 xi H. Furthermore, for any g ∈ G, h ∈ H, write

g = xi h0 , so that h0 h ∈ H, and

f (gh) = f (xi h0 h) = ρ(h0 h)−1 (wi ) = ρh−1 ρ(h0 )−1 (wi ) = ρh−1 f (g)

Hence, f ∈ I(W ). Now clearly, T (f ) = (w1 , w2 , . . . , w` ) holds.

Example 4.1.3. 1. Let H = {e} < G and χ1 : H → C∗ be the trivial representation.

Then, by the above definition, W = C,

X = {f : G → C} = L(G) and I(W ) = X = L(G)

Finally,
Tg (f )(x) = f (g −1 x)
Hence, IndG
H (χ1 ) is the left regular representation of G.

2. Let H = G and ρ : G → GL(W ) be any representation. Then

X = {f : G → W } and I(W ) = {f ∈ X : f (xg) = ρg−1 (f (x)) ∀g, x ∈ G}

and let ρb = IndG

H (ρ). Define T : I(W ) → W by f 7→ f (e). Then T is well-defined,
and linear. Also, if S : W → I(W ) given by

S(w)(x) := ρx−1 (w) ∀x ∈ X

36
 Then, for any f ∈ I(W ), and x ∈ X
(S ◦ T )(f )(x) = S(f (e))(x) = ρx−1 (f (e)) = f (ex) = f (x)
Hence, S ◦ T = idI(W ) . Also,
(T ◦ S)(w) = T (S(w)) = S(w)(e) = ρe−1 (w) = w
and so T ◦ S = idW . Hence, T is an isomorphism. Furthermore, for any g ∈ G, f ∈
I(W ),
ρg (f )) = Lg (f )(e) = f (g −1 e) = f (eg −1 ) = ρg (f (e)) = ρg (T (f ))
T (b
Hence, T ◦ ρbg = ρg ◦ T . Hence, T ∈ HomG (b
ρ, ρ). Hence,
IndG
H (ρ) ∼ ρ

3. Let G = D2n = ha, b : an = b2 = 1, bab = an−1 i and let H = hai. Let ρ ∈ H b be an

irreducible representation of H, then H ∼
= Zn , so ∃k ∈ {0, 1, . . . , n − 1} such that
ρ(a) = ζ k
where ζ = e2πi/n . Here, W = C, so X = {f : G → C} = L(G). Also, I(W ) < X is
a space of dimension
dim(C)[G : H] = 2
By the above proposition, we have an isomorphism
I(W ) → C2 given by f 7→ (f (e), f (b))
Let B = {f1 , f2 } ⊂ I(W ) be functions such that
f1 (e) = 1, f1 (b) = 0 and f2 (e) = 0, f2 (b) = 1
Write ρb = IndG
H (ρ). Then,

ρba (f1 )(e) = f1 (a−1 ) = f1 (ea−1 ) = ρa−1 (f1 (e)) = ρa−1 (1) = ζ −k
ρba (f1 )(b) = f1 (a−1 b) = f1 (ba) = ρa (f1 (b)) = ρa (0) = 0
ρba (f2 )(e) = f2 (a−1 ) = ρa (f2 (e)) = ρa (0) = 0
ρba (f2 )(b) = f2 (a−1 b) = f2 (ba) = ρa (f2 (b)) = ζ k
Hence,  −k 
ζ 0
[b
ρ a ]B =
0 ζk
Also,
ρbb (f1 )(e) = f1 (b−1 e) = f1 (b) = 0
ρbb (f1 )(b) = f1 (b−1 b) = f1 (e) = 1
ρbb (f2 )(e) = f2 (b−1 e) = f2 (b) = 1
ρbb (f2 )(b) = f2 (b−1 b) = f2 (e) = 0
and so  
0 1
[b
ρb ]B =
1 0

37
4.2 Frobenius Character Formula
We fix some notation for this section:

1. Let ρ : H → GL(W ) be a representation and ρb = IndG H (ρ). We wish to determine

the character of the induced representation. We write

χ = χρ and IndG
H (χ) = χρb

To do this, we assume that W has an inner product h·, ·i and that ρ is a unitary
representation of H.

2. A set T = {x1 , x2 , . . . , x` } ⊂ G is called a transversal of H in G if

`
G
G= xi H
i=1

3. If I(W ) as above, we define an inner product on I(W ) as

`
X
hf1 , f2 i = hf1 (xk ), f2 (xk )i
k=1

Note that this defines an inner product on I(W ) by the proof of Proposition 4.1.2.

4. Choose an orthonormal basis {e1 , e2 , . . . , en } of W and . For 1 ≤ i ≤ `, 1 ≤ j ≤ n,

let fi,j ∈ I(W ) such that
fi,j (xk ) = δi,k ej
Then {fi,j : 1 ≤ i ≤ `, 1 ≤ j ≤ n} forms an orthonormal basis for I(W ) (using the
isomorphism from Proposition 4.1.2)

Theorem 4.2.1 (Frobenius Character Formula). Let ρ : H → GL(W ) be a representation

with character χ, and let χG denoted the character of the induced representation IndG
H (ρ).
If T = {x1 , x2 , . . . , x` } denotes a transversal of H in G, then
X
IndGH (χ)(g) = χ(x−1
i gxi )
x−1
i gxi ∈H

Proof. Let fi,j be the ONB of I(W ) as defined above, then we wish to determine
n
X
hb
ρg (fi,j ), fi,j i
j=1

1. Consider each term, then by definition

`
X
hb
ρg (fi,j ), fi,j i = hb
ρg (fi,j )(xk ), fi,j (xk )i = hb
ρg (fi,j )(xi ), ej i
k=1

38
 Now,
ρbg (fi,j )(xi ) = fi,j (g −1 xi )
Since g −1 xi ∈ G = t`m=1 xm H, ∃ unique 1 ≤ m ≤ ` such that

g −1 xi ∈ xm H

and so ∃ unique h ∈ H such that g −1 xi = xm h. Hence,

ρbg (fi,j )(xi ) = fi,j (xm h) = ρh−1 (fi,j (xm ))

(
0 : i 6= m
= ρh−1 (δi,m ej ) =
ρh−1 (ej ) : i = m

Now,
i = m ⇔ g −1 xi ∈ xi H ⇔ x−1 −1 −1
i g xi ∈ H ⇐ xi gxi ∈ H

and in this case, h = x−1 −1

i g xi , so h
−1
= x−1
i gxi . Hence,
(
0 : x−1
i gxi ∈ H
ρbg (fi,j )(xi ) =
ρx−1
i gxi
(ej ) : otherwise

Hence,
X
IndG
H (g) = hb
ρg (fi,j ), fi,j i
i,j
X n
X
= hρx−1
i gxi
(ej ), ej i
x−1
i gxi ∈H
j=1
X
= χ(x−1
i gxi )
x−1
i gxi ∈H

Example 4.2.2. Let G = D2n = ha, b : an = b2 = 1, bab = an−1 i, H = hai and ρ : H →

C∗ be the map
a 7→ ζ k
where ζ = e2πi/n and 0 ≤ k ≤ n − 1. Then [G : H] = 2, and a transversal of H in G is
{e, b}. Also,

eae = a ∈ H, and bab = an−1 ∈ H

ebe ∈
/ H, and bbb = b ∈
/H
⇒ IndG
H (χ)(a) = χ(a) + χ(a
n−1
) = ζ k + ζ (n−1)k = ζ k + ζ −k
IndG
H (χ)(b) = 0

This agrees with the calculation in the example at the end of the previous section.

39
For a function f : H → C, we write
(
f (g) : g ∈ H
f˙(g) :=
0 : otherwise

Then the Frobenius Character formula gives

`
X
IndG
H (χ)(g) = χ̇(x−1
i gxi )
i=1

Proposition 4.2.3. For any g ∈ G,

1 X
IndG
H (χ)(g) = χ̇(x−1 gx)
|H| x∈G

Proof. For any x ∈ G, ∃ unique 1 ≤ i ≤ `, h ∈ H such that x = xi h. Then

χ̇(x−1 gx) = χ̇(xi gxi )

Hence,
X `
X X `
X
−1
χ̇(x gx) = −1
χ̇(x gx) = |H|χ̇(x−1
i gxi )
x∈G i=1 x∈xi H i=1

Definition 4.2.4. Let H < G, and Z(L(H)), Z(L(G)) denote the spaces of class func-
tions on H and G respectively.
1. Define ResG
H : Z(L(G)) → Z(L(H)) by

a 7→ a|H

Note that if a is a class function, then so is a|H .

2. Define IndG
H : Z(L(H)) → Z(L(G)) by

1 X
IndG
H (b)(g) 7→ ḃ(x−1 gx)
|H| x∈G

Then this map is well-defined

Proof. Let y ∈ G, we wish to show that
−1
IndG G
H (b)(ygy ) = IndH (b)(g)

To see this, note that

−1 1 X 1 X −1
IndG
H (b)(ygy ) = ḃ(x−1 ygy −1 x) = ḃ(z gz)
|H| x∈G |H| z∈G

since the map x 7→ y −1 x is a bijection on G.

40
Proposition 4.2.5. If ρi : H → GL(Wi ), i = 1, 2 are two representations of H, then
IndG G G
H (ρ1 ⊕ ρ2 ) ∼ IndH (ρ1 ) ⊕ IndH (ρ2 )

Proof. Let χ, ϕ, and ψ denote the characters of IndG G G

H (ρ1 ⊕ ρ2 ), IndH (ρ1 ), and IndH (ρ2 )
respectively. Then by the Frobenius character formula and the fact that IndG H is additive,
we get

χ = IndG
H (χρ1 ⊕ρ2 )

= IndG
H (χρ1 + χρ2 )

= IndG G
H (χρ1 ) + IndH (χρ2 )
=ϕ+ψ

The result now follows from the fact that two representations of G with the same character
must be equivalent.
G
Note that both ResGH and IndH are linear maps. Now recall that both Z(L(G)) and
Z(L(H)) are inner product spaces.

Theorem 4.2.6 (Frobenius Reciprocity). For any a ∈ Z(L(G)), b ∈ Z(L(H))

G
hResG
H (a), biL(H) = ha, IndH (b)iL(G)

Proof.
1 X
ha, IndG
H (b)iL(G) = a(g)IndGH (b)(g)
|G| g∈G
1 X 1 X
= a(g) ḃ(x−1 gx)
|G| g∈G |H| x∈G

Now, x−1 gx ∈ H ⇔ ∃h ∈ H such that g = xhx−1 . So rearranging, we get

1 XX
ha, IndG
H (b)iL(G) = a(xhx−1 )b(h)
|G||H| x∈G h∈H
1 XX
= a(h)b(h)
|G||H| x∈G h∈H
1 X
= hResG
H (a), biL(H)
|G| x∈G
= hResG
H (a), biL(H)

Definition 4.2.7. Let V, W be inner product spaces and T : V → W, S : W → V . We

say that S is an adjoint of T if

hT v, wiW = hv, SwiV ∀v ∈ V, w ∈ W

41
 G
Hence, Frobenius Reciprocity states that ResG H and IndH are adjoint to each other.
Remark. Let V, W be inner product spaces with ONB’s B1 = {e1 , e2 , . . . , en } and B2 =
{f1 , f2 , . . . , fm } respectively. If T : V → W and S : W → V are adjoints of each other,
then
hT (ej ), fi i = hej , S(fi )i = hS(fi ), ej i
Hence, the matrix of S is the conjugate transpose of the matrix of T .
Example 4.2.8. Let G = S5 , H = A4 , let B1 = {χ1 , χ2 , . . . , χ7 } denote the charac-
ters of irreducible representations of G, and let B2 = {ψ1 , ψ2 , . . . , ψ4 } be the irreducible
characters of H. Recall the character table of G
e (12) (123) (12)(34) (1234) (123)(45) (12345)
χ1 1 1 1 1 1 1 1
χ2 1 -1 1 1 -1 -1 1
χ3 4 2 1 0 0 -1 -1
χ4 4 -2 1 0 0 1 -1
χ5 6 0 0 -2 0 0 1
χ6 5 1 -1 1 -1 1 0
χ7 5 -1 -1 1 1 -1 0
and the character table of H
g e (12)(34) (123) (132)
ψ1 1 1 1 1
ψ2 1 1 ω ω2
ψ3 1 1 ω2 ω
ψ4 3 -1 0 0
Restriction gives
g e (12)(34) (123) (132)
χ1 |H 1 1 1 1
χ2 |H 1 1 1 1
χ3 |H 4 0 1 1
χ4 |H 4 0 1 1
and so on. Hence, taking inner products, we get
χ1 |H = ψ1
χ2 |H = ψ1
χ3 |H = ψ1 + ψ4
and so on. Hence, the matrix of ResG
H with respect to these bases B1 and B2 can be
computed to be  
1 0 0 0
1 0 0 0
 
1 0 0 1
 
1 0 0 1
 
0 0 0 2
 
0 1 1 1
0 1 1 1

42
Hence, the matrix of IndG
H is the transpose conjugate of this matrix. In particular, we
can determine
IndG
H (ψ2 ) = ψ6 + ψ7

and other such identities.

More generally if H < G, write G b = {ρ1 , ρ2 , . . . , ρn } and H

b = {ϕ1 , ϕ2 , . . . , ϕm }. Then
restriction of irreducible representations gives
m
X
ResG
H (ρi ) ∼ ri,j ϕj
j=1

for some non-negative integers ri,j ∈ Z. Induction gives

n
X
IndG
H (ϕj ) ∼ sj,i ρi
i=1

Frobenius Reciprocity states that ri,j = sj,i for all i, j.

Corollary 4.2.9 (Induction in stages). Suppose H < K < G and ρ : H → GL(W ) is a

representation. Then
IndG K G
K (IndH (ρ)) ∼ IndH (ρ)

Proof. Let ρb = IndK

H (ρ), then by definition

IndK
H (χρ ) = χρb

for any class function b ∈ Z(L(G))

hIndG G
K (χρb), biL(G) = hχρb, ResK (b)iL(K)

Furthermore,
hχρb, ResG K G
K (b)iL(K) = hχρ , ResH (ResK (b))iL(H)

But ResK G G
H (ResK (b)) = ResH (b). Hence,

hIndG G G
K (χρb), biL(G) = hχρ , ResH (b)iL(H) = hIndH (χρ ), biL(G)

This is true for every b ∈ Z(L(G)), so

χIndGK (IndK
H (ρ))
= IndG K G G
K (IndH (χρ )) = IndK (χρb) = IndH (χρ ) = χIndG
H (ρ)

Hence the result.

43
4.3 Examples
4.3.1 A group of order 21
In S7 , define
a = (1234567), b = (235)(476) and G := ha, bi
Then a7 = b3 = 1, b−1 ab = a2 , hence

G = {ai bj : 0 ≤ i ≤ 6, 0 ≤ j ≤ 2} ⇒ |G| = 21

1. Let H = hai, then |H| = 7 and b−1 ab ∈ H, so H C G. Finally, G/H ∼

= Z3 is
abelian, so
[G, G] ⊂ H
Since |H| = 7 and [G, G] 6= {e}, we have [G, G] = H. Hence, G has 3 non-trivial
characters, we denote by {χ1 , χ2 , χ3 }.

2. Now we determine conjugacy classes in G: Recall that if x ∈ G, then xG denotes

the conjugacy class of x, CG (x) the centralizer of x in G, and

|G|
|xG | =
|CG (x)|
by the orbit-stabilizer theorem.
a) Note that eG = {e} = C1 .
b) If x = a, then a ∈ CG (a), so H ⊂ CG (a), so 7 | |CG (a)|. Since b ∈
/ CG (a), it
follows that |CG (a)| < 21. Since |CG (a)| | 21, it follows that

|CG (a)| = 7 ⇒ CG (a) = H

Hence, |aG | = 3. The relation b−1 ab = a2 implies that a2 ∈ aG . Hence,

a4 ∈ aG . Thus it follows that

C2 = aG = {a, a2 , a4 }

c) Similarly, |CG (a3 )| = H, and so |(a3 )G | = 3, and as above

C3 = (a3 )G = {a3 , a5 , a6 }

d) As done for a above, |bG | = 7. Check that

C4 = (b)G = {ai b : 0 ≤ i ≤ 6}

e) Similarly,
C5 = (b2 )G = {ai b2 : 0 ≤ i ≤ 6}
These are all the conjugacy classes of G.

44
3. We have a partial character table given by
x e a a3 b b2
|xG | 1 3 3 7 7
χ1 1 1 1 1 1
χ2 1 1 1 ω ω2
χ3 1 1 1 ω2 ω

4. Now we induce representations from H. Let ζ = e2πi/7 , and define

ρ : H → C∗ given by a 7→ ζ

and let ψ = χIndGH (ρ) . By the Frobenius Character formula,

`
X
ψ(g) = ρ̇(x−1
i gxi )
i=1

where {x1 , x2 , . . . , x` } are a set of representatives for G/H. Now |H| = 7, so

|G/H| = 3, and we take
x1 = e, x2 = b, x3 = b2
Since H is normal, x−1i gxi ∈ H for all g ∈ H. Furthermore, if g ∈
/ H, then
−1
xi gxi ∈
/ H for all i. Hence,

ψ(g) = 0 ∀g ∈
/H

Also,

ψ(e) = ρ(e) + ρ(b−1 eb) + ρ(b−2 eb2 ) = 3ρ(e) = 3

ψ(a) = ρ(eae) + ρ(b−1 ab) + ρ(b−2 ab2 ) = ρ(a) + ρ(a2 ) + ρ(a4 ) = ζ + ζ 2 + ζ 4
ψ(a3 ) = ζ 3 + ζ 5 + ζ 6

So this gives us values in the table as

x e a a3 b b2
G
|x | 1 3 3 7 7
ψ 3 ζ + ζ2 + ζ4 ζ3 + ζ5 + ζ6 0 0
Now calculate
1
hψ, ψi = [3 + 3|ζ + ζ 2 + ζ 4 |2 + 3|ζ 3 + ζ 5 + ζ 6 |2 ]
21
and check that
|ζ + ζ 2 + ζ 4 |2 = (ζ + ζ 2 + ζ 4 )(ζ −1 + ζ −2 + ζ −4 )
= 1 + ζ −1 + ζ −3 + ζ + 1 + ζ −2 + ζ 3 + ζ 2 + 1
= 3 + ζ + ζ2 + ζ3 + ζ4 + ζ5 + ζ6
= 3 + (−1) = 2

45
 Similarly for the third term, so we get
1
hψ, ψi = [3 + 6 + 6] = 1
21
Hence, ψ is irreducible.
5. Now let ϕ : H → C∗ be given by
ρ(a) = ζ 2
Then if η = IndG
H (ρ), we get, by a similar calculation

x e a a3 b b2
G
|x | 1 3 3 7 7
η 3 ζ3 + ζ5 + ζ6 ζ + ζ2 + ζ4 0 0
Hence,
hη, ηi = 1
so η is also irreducible. This gives the character table of G as
x e a a3 b b2
|xG | 1 3 3 7 7
χ1 1 1 1 1 1
χ2 1 1 1 ω ω2
χ3 1 1 1 ω2 ω
ψ 3 ζ + ζ2 + ζ4 ζ3 + ζ5 + ζ6 0 0
η 3 ζ3 + ζ5 + ζ6 ζ + ζ2 + ζ4 0 0

4.3.2 A group of order p(p − 1)

Let p ∈ N prime, and let G be the group of matrices given by
  
a b ∗
G= : a ∈ Zp , b ∈ Zp
0 1
Then G is a non-abelian group with |G| = p(p − 1). Let H be the subgroup
  
1 b
H= : b ∈ Zp
0 1
Note that H C G and
G/H ∼
= Z∗p
which is cyclic (and hence abelian). Hence, [G, G] ⊂ H. Since G is non-Abelian,
[G, G] 6= {e}
Since |H| = p, it follows that [G, G] = H. Hence, G has precisely (p−1) linear characters,
denoted by {χ1 , χ2 , . . . , χp−1 }.

G has conjugacy classes given by

46
  
1 0
1. C1 =
0 1
 
1 1
2. Let x = , then we have
0 1
   −1   
a b 1 1 a b a a+b 1/a −b/a
=
0 1 0 1 0 1 0 1 0 1
 
1 a
=
0 1

Hence,  
a b
∈ CG (x) ⇔ a = 1
0 1
  
1 b
CG (x) = : b ∈ Zp
0 1
In particular, |xG | = |G|/|CG (x)| = (p − 1).

3. Now consider an element of the form

 
x 0
z := , x ∈ Z∗p , x 6= 1
0 1

Then
   −1   
a b x 0 a b ax b 1/a −b/a
=
0 1 0 1 0 1 0 1 0 1
 
x −bx + b
=
0 1

Hence,  
a b
∈ CG (z) ⇔ −bx + b = 0 ⇔ b = 0
0 1
Hence,   
a 0 ∗
CG (z) = : a ∈ Zp
0 1
and so |z G | = |G|/|CG (z)| = p

4. Finally, if π : G → G/H denotes the quotient map, then if x1 6= x2 , then

 −1       
x1 0 x2 0 1/x1 0 x2 0 x2 /x1 0
= = ∈
/H
0 1 0 1 0 1 0 1 0 1

and so    
x1 0 x2 0
π 6= π
0 1 0 1

47
 if x1 6= x2 . Since G/H is abelian, this implies
   
x1 0 x2 0
π ∼π ⇔ x1 = x2
0 1 0 1

and hence    
x1 0 x2 0
∼ ⇔ x1 = x2
0 1 0 1
Hence, by part (3), we get (p − 1) conjugacy classes
 G
x 0
Cx = for x ∈ Z∗p , x 6= 1
0 1

each of which have cardinality p.

5. Now calculating cardinalities, we get

X
1 + (p − 1) + p = p + (p − 2)p = p(p − 1)
x∈Z∗p ,x6=1

and so these are all the conjugacy classes in G. In particular, G has p conjugacy
classes.

Hence, G has exactly one more irreducible representation ψ. The degree formula reads

p(p − 1) = p − 1 + d2ψ ⇒ dψ = p − 1

Let ϕ : H → C∗ be the map  

1 b
7→ e2πi/p
0 1
and let η = IndGH (ϕ). We claim that η = ψ is the required irreducible representation.
Note that dη = p − 1. Furthermore, by Frobenius reciprocity

hχη , χi i = hχϕ , ResG

H (χi )i

Now, ResG H (χi ) is the trivial representation for all 1 ≤ i ≤ n, and ϕ is a non-trivial
irreducible representation. So by Schur Orthogonality,

hχϕ , ResG
H (χi )i = 0

Hence, the Maschke decomposition of η has the form

χη = mχψ

However, dη = dψ , so m = 1 and η is irreducible.

48
Bibliography
[A] Additional Notes http://home.iiserb.ac.in/~prahlad/current_course/
additional_notes.pdf

[B] A. Baker, http://www.maths.gla.ac.uk/~ajb/dvi-ps/groupreps.pdf

[BS] B. Steinberg, Representation Theory of Finite Groups, Springer 2012

[JL] G. James, M. Liebeck, Representations and Characters of Groups (2nd Ed), Cam-
bridge Univ Press (2001)

[T] Teleman, https://math.berkeley.edu/~teleman/math/RepThry.pdf

49