18.704 Notes: 1 Introduction To Representation Theory

18.
704 Notes
Seminar in Algebra
Wanlin Li
Fall 2020
1 Introduction to Representation Theory

1.1 September 3
• Study symmetry (group) and structure (action)
• Definition. Representation: F -vector space V equipped with a linear action of
the group G
• Action is a map G × V → V, (g, v) 7→ gv satisfying compatibility properties ev = v
and (gh)v = g(hv), where e is the identity of G
• Linearity comes from g(αv + βw) = α(gv) + β(gw) for α, β ∈ F, v, w ∈ V
• Example. Trivial representation gv = v for all g ∈ G
• Example. Let G = GL2 (C) act on V = C2 by matrix multiplication
• Example. Let symmetric group S2 act on C2 by swapping x and y coordinates;

x
this has the trivial representation as a subrepresentation over W1 = x=y
y
2
The {0} is another subspace of C that is G-stable, as is W−1 =
trivial space
•

x
x = −y
y

2 x
• Every subspace of V = C must be one dimensional Wα = ∈ V x = αy

y
for α ∈ C× ; this shows W1 , W−1 are the only stable nontrivial subspaces
• Stable subspace is equivalent to subrepresentations
• Note V = W1 ⊕ W−1 and these are the only irreducible representations of S2
• Irreducible representations of Sn are tied to Young diagrams of n
1
18.704 Course Notes Wanlin Li
• Example. Compact groups G = S 1 = {z ∈ C : |z| = 1}

ˆ
|f |2 < ∞
2 1 1

• Consider square integrable functions L (S ) := f : S → C
S1
X
• f can be approximated as an e2πinθ and irreducible representations of S 1 are
V = C, e2πiθ ∗ z = e2πinθ z in relation to Fourier analysis
2 Groups and Homomorphisms

2.1 Key Definitions and Properties
• Definition. Group: associativity, identity, invertibility
• Definition. Coset: with H a subgroup of G and x ∈ G, the right coset Hx =
{hx : h ∈ H}
• If G is finite with distinct cosets Hx1 , . . . , Hxr , h → hxi is a bijection and so
|Hxi | = |H| for all i
Lagrange’s Theorem. If G is a finite group and H is a subgroup of G, then |H|

divides |G|.
|G|
• Definition. Index: number of distinct cosets of H in G, written |G : H| = |H|
• Definition. Normal subgroup: N C G if g −1 N g = N for all g ∈ G

• Definition. Factor group: group of cosets G/N when N is a normal subgroup
• Definition. Simple group: group whose only normal subgroups are the trivial
group and itself
2.2 September 8
|G|
• If H is a subgroup of G and G is finite, the index [G : H] = |H| and the right
cosets Hx partition G
• A normal subgroup N C G if N is closed under conjugacy, which allows for the
quotient/factor group G/N to be a group
• A group G is simple if it has no nontrivial proper normal subgroups, similar to
prime factors
• Group homomorphism is a map that preserves group structure
2
• If f : G → H is a group homomorphism, ker f C G and im f ≤ H
Theorem. Suppose f : G → H is a group homomorphism. Then G/ ker f ∼

= im f.
• Construct isomorphism by mapping the coset Kg to f (g)
3 Vector Spaces and Linear Transformations

3.1 Definitions and Properties
• Definition. Vector space: a set V over a field F with scalar multiplication and
abelian vector addition
• Definition. Direct sum: the sum U1 + · · · + Ur = {u1 + · · · + ur : ui ∈ Ui } of
subspaces if every element in the sum can be written uniquely as u1 + · · · + ur
with ui ∈ Ui ; denoted U1 ⊕ · · · ⊕ Ur
• Definition. External direct sum: given F -vector spaces U1 , . . . , Ur , the vector
space V = {(u1 , . . . , ur ) : ui ∈ Ui } with addition and multiplication defined
component-wise
• Definition. Endomorphism: linear transformation from vector space to itself
• Definition. Projection: endomorphism π of vector space V satisfying π 2 = π
3.2 September 8
• Functions from a set S to a vector space form a vector space as well
• Vectors {vi } span an F -vector space V if every element of V is a linear combina-
tion of vi
X
• Vectors are linearly independent if λi vi = 0 means λi = 0 ∀i
• A basis of a vector space is an independent set that spans the entire space
Rank-Nullity Theorem. For a linear transformation f : V → W, then dim V =

dim(ker f ) + dim(im f ).
• For a change of basis matrix T, basis change is given by T −1 BT

• Eigenvalue λ and eigenvector v of linear transformation f satisfy f (v) = λv
where λ is a scalar
• In matrix representation, λ satisfies det(λIn − A) = 0
3
4 Group Representations
• Definition. Representation: homomorphism ρ from group G to GL(n, F ) for
some integer n and field F, where n is the degree of ρ
• Definition. Equivalent representations: two representations ρ : G → GL(m, F ), σ :
G → GL(n, F ) with m = n such that for some invertible n × n matrix T, gσ =
T −1 (gρ)T ∀g
• Definition. Faithful representation: ρ : G → GL(n, F ) with ker ρ = {1}
• A representation is faithful iff the image of the representation is isomorphic to
G
4.2 September 8
• Link group theory and linear algebra by representing group elements as invert-
ible matrices
• Representation must be a homomorphism so that it gives information about the
group structure
• Every group has a representation of every degree by the trivial map gρ = In
• To convert representations into another one, it suffices to convert an invertible
matrix into another invertible matrix while maintaining group structure
• Use change of basis T to get an equivalent representation by sending gρ to
T −1 gρT for all g
• Transforming to an equivalent representation can give more a more intuitive
and understandable representation for the same group
• Trivial representation sends all of G to In
• All representations that are equivalent to a faithful representation are also
faithful
5 FG-Modules
• Definition. FG-module: given group G, an F -vector space V such that multi-
plication vg is defined satisfying group laws and behaves with vector addition
and scalar multiplication
4
• Think of G acting on V
• Notationally let [g] denote the matrix of the endomorphism v 7→ vg of V (relative
to some given basis)
Theorem. If ρ : G → GL(n, F ) is a representation of G over F and V = F n , then

V is an FG-module under the multiplication vg = v(gρ). Additionally there is a
basis B of V such that gρ = [g]B for all g ∈ G.
If V is a general FG-module and B is a basis of V, then g → [g]B is a representation
of g over F.
• FG-module is analogous to a group acting on a vector space

• Definition. Faithful FG-module: FG-module for which vg = v ∀v ∈ V implies g
is the identity of G
Theorem. Suppose that V is an FG-module with basis B and let ρ be the repre-
sentation of G over F given by ρ : g → [g]B . If B 0 is a basis of V, then the represen-
tation φ : g → [g]B0 of G is equivalent to ρ. Conversely if φ is a representation of G
that is equivalent to ρ, then there is a basis B 0 of V such that φ : g → [g]B0 .
5.2 September 10
• Let ρ : G → GL(n, F ) be a representation and V a vector space over F
• V is an F G-module if multiplication by G is defined, distributes over vector ad-
dition, and associates with scalar multiplication
• Given representation ρ and V = F n , setting V g = v(gρ) makes V into an F G-
module and there exists a basis B of V such that gρ = [g]B
• Conversely if V is an F G-module with basis B, g → [g]B defines a representation
• Circumvent representation by starting with group action on a basis and extend-
ing the action to be linear on V
6 FG-submodules and Reducibility

• Definition. FG-submodule: subspace W of an F G-module V such that wg ∈ W
for all w ∈ W, g ∈ G
5
• Definition. Irreducible FG-module: non-zero F G-module V with no F G-submodules

except {0} and V
• Definition. Irreducible representation: ρ : G → GL(n, F ) if the corresponding
F G-module F n with vg = v(gρ) is irreducible
6.2 September 10
• Given an F G-module V, W is an F G-submodule if it is a subspace of W and is
closed under multiplication by G
• Irreducible F G-module has no proper non-trivial submodules
• If V is reducible then there is an F G-submodule
W with dimension less than
Xg 0
dim V and choosing a basis of W gives [g]B = in block form
Yg Zg
7 Group Algebras
7.1 Definition and Properties
• Definition. Group algebra: for a finite group G with elements g1 , . . . , gn and
a field F, a vector space F G over F with natural basis {gi } and natural defini-
tions for addition and scalar multiplication, additionally with multiplication of
vectors defined by
  !
X X X
 λg g  µh h = λg µh (gh)
g∈G h∈G g,h∈G
• In F G, multiplication is associative and distributes over addition, and there is

an identity element
• Algebra axioms imply that it is both a vector space and a ring
• Definition. Regular FG-module: vector space F G spanned by {gi } along with
the natural multiplication vg for v ∈ F G, g ∈ G
• Definition. Regular representation: representation obtained by action of g over
natural basis of F G
• The regular F G-module is faithful
• If V is an F G module, the multiplication vg can be extended to all elements of
the group algebra in the natural way
6
8 FG Homomorphisms
8.1 Definition and Properties
• Definition. FG-homomorphism: a function φ : V → W of F G-modules if φ is
linear and satisfies (vg)φ = (vφ)g for all v ∈ V, g ∈ G; i.e. if φ sends v to w, it
sends vg to wg
Proposition. Let V and W be F G-modules and let φ : V → W be an F G-

homomorphism. Then ker φ is an F G-submodule of V and im φ is an F G-
submodule of W.
• F G-isomorphism is an invertible F G-homomorphism and gives rise to isomor-

phic F G-modules
Proposition. If φ : V → W is an F G-isomorphism, then the inverse φ−1 :

W → V is also an F G-isomorphism.
Theorem. Suppose that V is an F G-module with basis B and W is an F G-module

with basis B 0 . Then V and W are isomorphic iff the representations ρ : g → [g]B
and σ : g → [g]B0 are equivalent.
• [g]B = [g]B0 means that multiplying by g changes B in the same way that it
changes B 0 , e.g. if g swaps the first and second basis elements of B then it also
swaps the first and second elements of B 0
• If V is an F G-module that is a direct sum U1 ⊕ · · · ⊕ Ur of F G-submodules with
bases Bi , combining these bases gives a basis B of V such that [g]B is diagonal
when viewed as a block matrix with [g]Bi along the diagonal
Proposition. Let V be an F G-module and suppose V = U1 ⊕ · · · ⊕ Ur . Each

v ∈ V can be written uniquely as v = u1 +· · ·+ur , so the map πi : V → V, vπi = ui
is well-defined. Each πi is an F G-homomorphism and also a projection of V.
Proposition. If V is an F G-module with V = U1 + · · · + Ur where each Ui is

an irreducible F G-submodule, then V is a direct sum of some subset of these
submodules.
• Choose a subset {W1 , . . . , Ws } such that W1 + · · · + Ws is direct but adding any

other submodule makes it not direct, and let W = W1 + · · · + Ws
• Ui ⊆ W for all i because otherwise W ∩ Ui is a nontrivial F G-submodule of the
irreducible submodule Ui
7
9 Maschke’s Theorem
9.1 Definitions and Key Properties
Maschke’s Theorem. Let G be a finite group, let F be R or C, and let V be an

F G-module. If U is an F G-submodule of V, then there is an F G-submodule W of
V such that V = U ⊕ W.
• The condition on F is important:

if G is the cyclic group Cp with generator a
1 0
and F = Fp , the map aj → is a representation G → GL(2, F ) but it has
j 1
only one 1D F G-submodule
• For Maschke’s Theorem, pick any subspace W0 ⊂ W0 such that V = U ⊕ W0
• Define φ : V → V to be the projection map from V to U, which has kernel W0
X
•
1
Let θ : V → V be the map vθ = |G| vgφg −1
g∈G
• θ has image in U because the image of φ is in U and U is an F G-submodule and

symbol pushing shows that θ is an F G-homomorphism
1
• If F has finite characteristic, |G| may not be defined and this factor is necessary
for making θ a projection
• Every element of U is fixed under θ, so θ2 = θ is a projection with image U
• Then if W = ker θ, W is an F G-submodule of V with V = U ⊕ W

Xg 0
• In other words, a reducible representation ρ : g → is equivalent to a
Yg Zg

Ag 0
representation g →
0 Bg
• Definition. Completely reducible FG-module: F G-module V such that V =
U1 ⊕ · · · ⊕ Ur where each Ui is an irreducible F G-submodule
Theorem. If G is a finite group and F = R or C, then every nonzero F G-module

is completely reducible.
Proposition. Let V be an F G-module, where F = R or C and G is a fi-

nite group. If U is an F G-submodule of V, then there exists a surjective F G-
homomorphism from V onto U.
• This is just the projection map
8
9.2 September 15
• For Maschke’s Theorem, it is important that F = R or C
• This allows an F G-module to be broken into a direct sum of irreducible F G-
submodules
• Use projection map to create another projection map with an F G-submodule as
the kernel
• Maschke’s Theorem says there exists some submodule that completes the direct
sum, not necessarily a unique one (even though this is true)
• Maschke’s Theorem is stronger than the result of picking a direct sum out of
irreducible submodules
• Maschke’s Theorem holds in characteristic 0 fields or any field in which the
characteristic does not divide |G|
10 Schur’s Lemma
Schur’s Lemma. Let V and W be irreducible CG-modules.
1. If θ : V → W is a CG-homomorphism, then θ is either the trivial map

or an isomorphism.
2. If θ : V → V is a CG-isomorphism, then θ is a scalar multiple of the

identity endomorphism.
• θ must have an eigenvalue λ, so ker(θ − λI) is nonzero and thus is all of V

• This means θ must be a scalar mulitiplication map
Proposition. Let V be a nonzero CG-module and suppose that every CG-

homomorphism from V to V is a scalar multiple of the identity. Then V is
irreducible.
• If V is reducible, it has a projection map to a CG-submodule
Corollary. Let ρ : G → GL(n, C) be a representation of G. Then ρ is irreducible

iff every n × n matrix A satisfying (gρ)A = A(gρ) has the form A = λIn with
λ ∈ C.
9
Proposition. If G is a finite abelian group, then every irreducible CG-module

has dimension 1.
• If G is a finite abelian group and V is an irreducible CG-module, vgx = vxg for

all g ∈ G and multiplication v → vx is a CG-homomorphism which must be λx 1V
• Then every subspace of V is a CG-submodule because it is closed under multi-
plication for any x ∈ G
Theorem. Every finite abelian group is isomorphic to a direct product of cyclic

groups.
• Use this structure result to determine irreducible representations of finite abelian

groups
• With G = Cn1 × · · · × Cnr , let ci be a generator for Ci so that gi = (1, . . . , ci , . . . , 1)
are generators for G
• If ρ is an irreducible representation, for each gi there exists λi ∈ C such that
gi ρ = (λi ) and λi must have order ni
• Thus (g1i1 . . . grir )ρ is multiplication by λi11 . . . λirr and any ni th roots of unity λi
create a representation
• The number of these representations is n1 . . . nr
Theorem. Let G be the abelian group Cn1 ×· · ·×Cnr . The representations ρλ1 ,...,λr
of G are irreducible and have degree 1. These account for all |G| irreducible rep-
resentations of G over C.
Proposition. Let G be a finite group and V a CG-module. If g ∈ G, there is

a basis B of V such that the matrix [g]B is diagonal. If g has order n, then the
entries on the diagonal of [g]B are nth roots of unity.
• This is by decomposing V under the action of the cyclic group hgi

• Definition. Center of group algebra: with G a finite group, the subset Z(CG) of
the group algebra CG satisfying
Z(CG) = {z ∈ CG : zr = rz ∀r ∈ CG}
• Example. If H is a normal subgroup of G, then the sum of the elements of H is

in the center
10
Proposition. Let V be an irreducible CG-module and let z ∈ Z(CG). Then

there exists λ ∈ C such that vz = λv for all v ∈ V.
• This is because elements in the center commute with arbitrary group elements
and so multiplication by z is a CG-homomorphism
• Center Z(G) of G is a normal subgroup of G and a subset of Z(CG)
Proposition. If there exists a faithful irreducible CG-module, then Z(G) is

cyclic.
• Faithful irreducible CG-module gives injective homomorphism from Z(G) to C×
Proposition. Suppose that G is a finite group such that every irreducible

CG-module has dimension 1. Then G is abelian.
10.2 September 15
• Use Schur’s Lemma to characterize representations of finite abelian groups: ev-
ery irreducible CG-module is one dimensional which allows for diagonalization
• Schur’s Lemma is a powerful tool for CG-homomorphisms of irreducible modules
• A finite group is abelian iff all irreducible CG-modules have dimension 1
Summary. A degree n representation of a group G over a field F is a homomorphism

from G to GLn (F ). This connects to F G-modules, which are vector spaces over F on
which G acts. This action must be linear in order to behave with the vector space.
Then representations can also be written (abuse of notation) as homomorphisms from
G to GLF (V ), mapping G to linear maps on V. Breaking the representation into ir-
reducible representations simplifies the matrix of the representation and breaks it
into invariant subspaces. Translating between these interpretations makes it easier
to understand the group structure from different directions.
11 Irreducible Modules and the Group Algebra

• With G a finite group and CG the group algebra (also the regular module), CG
decomposes as U1 ⊕ · · · ⊕ Ur ; claim every irreducible CG module is isomorphic
to one of these
11
Proposition. Let V and W be CG-modules and let ϑ : V → W be a CG-

homomorphism. Then there is a CG-submodule U of V such that V = ker ϑ ⊕ U
and U ∼
= im ϑ.
• This follows from Maschke’s Theorem
Proposition. Let V be a CG-module, and write V = U1 ⊕ · · · ⊕ Us as a direct

sum of irreducible CG-submodules. If U is any irreducible CG-submodule of V,
then U ∼= Ui for some i.
• Proof using Schur’s Lemma

• Set πi the projection map U → Ui , of which at least one map is nontrivial
• However U is not necessarily equal to any Ui
• Definition. Composition factor: an irreducible CG-module U if a CG-module V
has a submodule isomorphic to U
Theorem. Let CG be the regular CG-module and write CG = U1 ⊕ · · · ⊕ Ur as

a direct sum of irreducible submodules. Then every irreducible CG-module is
isomorphic to some Ui .
Proof. Let W be an irreducible CG-module and choose w 6= 0 ∈ W. Then W = {wr :

r ∈ CG}. Define ϑ : CG → W, rϑ = wr so that im ϑ = W. There is a CG-submodule U
of CG such that CG = U ⊕ ker ϑ and U ∼ = im ϑ = W, so U must be irreducible. Since
U is a composition factor of CG, so is W.
• As a result if G is finite, there are only finitely many non-isomorphic irreducible

CG-modules
11.2 September 17
• Group algebra CG is C-vector space with basis G along with multiplication in-
duced by G
• Working over C, Maschke’s Theorem holds for finite groups G
• As a consequence of Maschke’s Theorem, every finite dimensional CG-module
can be decomposed into irreducible CG-submodules
• Goal to determine the irreducible CG-modules up to isomorphism
12
• An example of a nontrivial representation is given by the group algebra CG

• Analog to First Isomorphism Theorem for groups: if V, W are CG-modules and
ϑ : V → W is a CG-homomorphism, there is a CG-submodule U of V such that
V = ker ϑ ⊕ U and U ∼
= im ϑ
• By Maschke, there is a submodule U such that V = ker ϑ ⊕ U
• Claim the map ϑ = ϑU : U → im ϑ is an isomorphism
• Injectivity because ker(ϑ) = ker ϑ ∩ U = {0}
• For surjectivity, take w ∈ im ϑ so w = ϑ(v + u) = ϑ(u) for v ∈ ker ϑ
• Analog to uniqueness of irreducible decomposition: if V = U1 ⊕ · · · ⊕ Us then any
irreducible submodule U ⊂ V is isomorphic to Ui for some i
• Resembles Schur’s Lemma
• Prove using projection homomorphism πi : V → Ui so the restriction of πi to U
is a CG-homomorphism of irreducible modules
X
• Because πi (v) = v, at least one πi is nonzero on U and Schur’s Lemma says
i
U∼
= Ui
• Claim that decomposing regular CG-module gives every irreducible CG-module
• Suppose W is an irreducible CG-module and let w ∈ W, so W = {wr : r ∈ CG}
because this is a CG-submodule of W and W is irreducible
• Define homomorphism φ : CG → W such that rφ = wr and use previous propo-
sition to write CG = ker φ ⊕ U where U is isomorphic to im φ = W
• U is irreducible because W is irreducible, and by the other proposition U ∼
= Ui
for some Ui
• As a consequence there are only finitely many non-isomorphic irreducible CG-
modules for a finite group G and all of these can be found by writing CG as a
direct sum of irreducible submodules
• Example. C3 = hai: write CG = U1 ⊕ U2 ⊕ U3 where U1 = span(1 + a + a2 ),
U2 = span(1 + ωa + ω 2 a2 ), U3 = span(1 + ω 2 a + ωa2 ) where ω 3 = 1
• These are the only irreducible CG-modules up to isomorphism
• U1 corresponds to the trivial submodule
X
• This generalizes to arbitrary finite groups: the span of g forms the trivial
g∈G
submodule
13
12 More on the Group Algebra

• Definition. Hom: HomCG (V, W ) the set of all CG-homomorphisms from V to
W, with addition and scalar multiplication defined naturally
• By Schur’s Lemma, for irreducible V, W the dimension of HomCG (V, W ) is 1 if
the modules are isomorphic and 0 otherwise
Proposition. Let V and W be CG-modules and suppose HomCG (V, W ) 6= {0}.

Then V and W have a common composition factor.
• Proof using Maschke’s Theorem to write V = ker ϑ ⊕ U and choosing an irre-

ducible submodule of U
Proposition. Let V, V1 , V2 , W, W1 , W2 be CG-modules. Then
1. dim(HomCG (V, W1 ⊕W2 )) = dim(HomCG (V, W1 ))+dim(HomCG (V, W2 ))
2. dim(HomCG (V1 ⊕V2 , W )) = dim(HomCG (V1 , W ))+dim(HomCG (V2 , W ))
• Proof using projection maps and restrictions

• As a result of these two propositions, if V = U1 ⊕ · · · ⊕ Us as a sum of irreducible
CG-modules and W is any irreducible CG-module, then dim HomCG (V, W ) =
dim HomCG (W, V ) is the number of CG-modules Ui with Ui ∼ =W
Proposition. If U is a CG-module, then dim(HomCG (CG, U )) = dim U.
Theorem. Suppose CG = U1 ⊕ · · · ⊕ Ur as a direct sum of irreducible CG-

submodules. If U is any irreducible CG-module, then the number of CG-modules
Ui isomorphic to U is equal to dim U.
• Definition. Complete set of non-isomorphic irreducible modules: irreducible

CG-modules V1 , . . . , Vk if every irreducible CG-module is isomorphic to some Vi
and no two Vi are isomorphic
Theorem. Let V1 , . . . , Vk form a complete set of non-isomorphic irreducible CG-

modules. Then
Xk
(dim Vi )2 = |G|.
i=1
14
12.2 September 17
• Let HomCG (V, W ) be the set of all CG-homomorphisms from V to W as a vector
space
• If V and W are irreducible, the dimension of HomCG (V, W ) is either 0 or 1 de-
pending on whether these modules are isomorphic
• If HomCG (V, W ) is non-trivial, then V and W have a common composition factor
using Maschke’s Theorem with kernel decomposition
• If V = U1 ⊕ · · · ⊕ Un where each Ui is an irreducible CG-module, then for any
irreducible CG-module W, the dimension of HomCG (V, W ) is equal to the number
of Ui satisfying Ui ∼
= W using direct sum property of the vector space Hom
• If U is a CG-module, then dim(HomCG (CG, U )) = dim U
• Take CG = U1 ⊕ · · · ⊕ Un as a decomposition into irreducible submodules
• For each Vi in a complete set of non-isomorphic irreduciblePCG-modules, P dim Vi =
di is the number of different U j with V i
∼
= Uj , so dim CG = dim Ui = di (dim Vi ) =
2
P
di
• Can use this to characterize finite groups by their irreducible decompositions,
although this does not uniquely determine the group
13 Conjugacy Classes
• Definition. Conjugacy class: the set xG = {g −1 xg : g ∈ G}
Proposition. If x, y ∈ G, then either xG = y G or xG ∩ y G = ∅.
• This means the conjugacy classes partition the group

• If G = xG G
1 ∪ · · · ∪ xl for distinct conjugacy classes, then x1 , . . . , xl are represen-
tatives of the conjugacy classes of G
Proposition. Let x, y ∈ G. If x is conjugate to y in G, then ordx = ordy and xn

is conjugate to y n for every integer n.
• Definition. Centralizer: the set of elements that commute with x:
CG (x) = {g ∈ G : g −1 xg = x}
15
• The centralizer is a subgroup containing x
Theorem. Let x ∈ G. Then the size of the conjugacy class xG is given by
|G|
|xG | = |G : CG (x)| = .
|CG (x)|
In particular |xG | divides |G|.
Proof. If g −1 xg = h−1 xh, then hg −1 ∈ CG (x) and so CG (x)g = CG (x)h. Then the
function f : g −1 xg → CG (x)g is a bijection from xG to the set of right cosets of CG (x),
proving the desired result.
• If |xG | = 1, then g −1 xg = x for all g ∈ G and so x is in the center Z(G)
Class Equation. Let x1 , . . . , xl be representatives of the conjugacy classes of G.

Then X
|G| = |Z(G)| + |xG
i |
xi ∈Z(G)
/
where |xG G
i | = |G : CG (xi )| and both |Z(G)|, |xi | divide |G|.
• In the symmetric group Sn , the conjugacy class xSn consists of all permutations
with the same cycle-shape (cycle lengths) as x
• In the alternating group An , if x commutes with some odd permutation in Sn
then xSn = xAn and otherwise xSn splits into two conjugacy classes of equal size,
represented by x and (12)−1 x(12)
Proposition. Let H be a subgroup of G. Then H C G (H is a normal subgroup)

iff H is a union of conjugacy classes of G.
• Definition. Class sum: letXC1 , . . . Cl Be the distinct conjugacy classes of a fi-

nite group G and let C i = g be an element of the group algebra CG; these
g∈Ci
elements C i are class sums
Proposition. The class sums form a basis of the center of CG.
16
13.2 September 24
• Conjugacy classes xG = {g −1 xg : g ∈ G} partition a group
• Conjugacy classes of the symmetric group are sets of permutations with the
same cycle shape
• Center of group algebra is the set of elements that commute with every element
in the group algebra
• Class sums are the sums of elements in conjugacy classes and these form a basis
for Z(CG)
14 Characters
• Definition. Character: for a given CG-module with basis B, the function χ :
G → C defined by χ(g) = tr[g]B
• The character is independent of the basis, so given a representation ρ : G →
GLn (C) the character can be defined as χ(g) = tr(gρ)
• Irreducible character is the character of an irreducible CG-module
Proposition. Isomorphic CG-modules have the same character. Additionally

if x, y are conjugate elements of the group G, then χ(x) = χ(y) for all characters:
characters are constant on conjugacy classes.
• Definition. Degree of a character: the dimension of the CG-module V for which

χ is a character
Proposition. Let χ be the character of a CG-module V. Suppose that g ∈ G

and ord(g) = m. Then
1. χ(1) = dim V
2. χ(g) is a sum of mth roots of unity
3. χ(g −1 ) = χ(g)
4. χ(g) is real if g and g −1 are conjugate
17
Theorem. Let ρ : G → GLn (C) be a representation of G with character χ. For

g ∈ G,
|χ(g)| = χ(1) ⇔ gρ = λIn
for some λ ∈ C. Additionally, ker ρ = {g ∈ G : χ(g) = χ(1)}.
• Definition. Kernel of a character: ker χ = {g ∈ G : χ(g) = χ(1)}

• If χ is a character of G, then so is χ, and χ is irreducible iff χ is irreducible
• Definition. Regular character: character of the regular CG-module
• If V = U1 ⊕· · ·⊕Ur is a direct sum of irreducible CG-modules, then the character
of V is the sum of the characters of Ui
Theorem. Let V1 , . . . , Vk be a complete set of non-isomorphic irreducible CG-

modules with characters χi and dimensions di . Then χreg = d1 χ1 + · · · + dk χk .
Proposition. If χreg is the regular character, then χreg (1) = |G| and χreg (g) = 0
for g 6= 1.
14.2 September 24
• Trace is the sum of diagonal entries of a matrix and tr(AB) = tr(BA), so trace
is invariant under conjugation
• Character of an element g is the trace of gρ
• Degree of a character is the degree of the representation
• Isomorphic CG-modules have the same character because under appropriate
bases, the matrices of the representation are identical
• Conjugate elements have the same trace and thus characters are constant on
conjugacy classes
• Value of permutation character correspond to the number of fixed points
• χ(1) = dim V and if ord(g) = r, χ(g) is a sum of rth roots of unity because there
exists a basis under which g is diagonal
• If g is conjugate to g −1 , then χ(g) is real because χ(g) = χ(g −1 ) = χ(g)
• As a special case if g has order 2, χ(1) ≡ χ(g) mod 2
18
• The kernel of the representation is the same as the kernel of the character,
where the kernel of the character is defined as the set of elements mapping
to χ(1)
• Definition. Regular character: character of the regular CG-module
• The regular character is zero on any element other than the identity
• Let G be a subgroup of Sn and define ν(g) = π(g) − 1 where π is the regular
character; then ν is a character
• This is because the character contains the trivial character as a component
15 Inner Products of Characters

• Character is a function from G to C, and the set of all such functions forms a
vector space
• Definition. Inner product: given ϑ, φ functions G → C, define the inner product
1 X
hϑ, φi = ϑ(g)φ(g)
|G|
g∈G
Proposition. Suppose G has exactly l conjugacy classes with representatives

g1 , . . . , gl , and let χ, ψ be characters of G. Then
1 X
hχ, ψi = hψ, χi = χ(g)ψ(g −1 ),
|G|
g∈G
and this value is real. This can also be written as

l
X χ(gi )ψ(gi )
hχ, ψi = .
|CG (gi )|
i=1
Theorem. Let U, V be non-isomorphic irreducible CG-modules with characters

χ, ψ. Then hχ, χi = 1 and hχ, ψi = 0.
19
Theorem. Let χ1 , . . . , χk be the irreducible characters of G. If ψ is any character

of G, then ψ = d1 χ1 + · · · + dk χk for non-negative di = hψ, χi i, and additionally
k
X
hψ, ψi = d2i .
i=1
• Definition. Constituent: given a character ψ of G, an irreducible character of

G such that hψ, χi =
6 0
Theorem. Let V be a CG-module with character ψ. Then V is irreducible iff

hψ, ψi = 1.
Theorem. Suppose that V and W are CG-modules with characters χ, ψ. Then V

and W are isomorphic iff χ = ψ.
Theorem. If χ1 , . . . , χk are the irreducible characters of G, then χi are linearly

independent vectors among all functions from G to C.
Theorem. Let V, W be CG-modules with characters χ, ψ. Then
dim(HomCG (V, W )) = hχ, ψi.
15.2 September 24
• Given characters χ, ψ, hχ, ψi = hψ, χi is an integer
• hχ, ψi = hψ, χi
• Simplify inner product of characters using fact that characters are constant on
conjugacy classes
1 X 1 X
• hχ, ψi = χ(g)ψ(g −1 ) = χ(g −1 )ψ(g) = hψ, χi = hχ, ψi
|G| |G|
g∈G
• To prove the main theorem, write CG = W1 ⊕ W2 where W1 , W2 have no common

composition factor (no irreducible submodules of W1 , W2 are isomorphic)
• Then 1 = e1 + e2 for e1 ∈ W1 , e2 ∈ W2
• For any w1 ∈ W1 , w2 ∈ W2 , claim w2 e1 = w1 e2 = 0, w1 e1 = w1 , w2 e2 = w2
20
• Sending w2 → w1 w2 is a CG-homomorphism from W2 to W1 , so w1 w2 must be 0

• Thus e21 = e1 , e22 = e2 , e1 e2 = e2 e1 = 0
X
• Claim e1 = |G|1
χ(g −1 )g
g∈G
• Look at endomorphism ϑ : w → we1 x−1 for an x ∈ G and compute trϑ

• Trace of w1 → w1 x−1 is χ(x−1 )
• If χ is the character of W1 , then hχ, χi = χ(1)
• Irreducible characters of G are linearly independent in the space of functions
from G to C
• The inner product gives the vector space F un(G, C) a Hermitian form (meaning
hϑ, cφi = chϑ, φi)
• Irreducible characters form a set of orthonormal vectors in F un(G, C) but this
does not form a basis because characters are constant on conjugacy classes
• However, irreducible characters do form a basis for the subspace of functions
that are constant on conjugacy classes
• Character determines a representation: representations have enough struc-
ture that all the information can be recovered from the character
16 September 29
• Definition. Class function: ψ : G → C such that ψ is constant on conjugacy
classes
• Vector space of class functions is equipped with inner product, and the canonical
basis of this space is 1 on one conjugacy class and 0 everywhere else
Theorem. The number of irreducible characters of G is equal to the number of

conjugacy classes of G.
• Regular module CG breaks into W1 ⊕ · · · ⊕ Wk where each Wi is a direct sum of

copies of Vi and Vi form a complete set of non-isomorphic irreducible modules
• If 1 = f1 + · · · + fk , then Z(CG) is in the span of {f1 , . . . , fk }
• By Schur’s Lemma, multiplication by an element z ∈ Z(CG) is multiplication
by a constant
21
• Therefore Z(CG) is a subspace of the space spanned by {f1 , . . . , fk }

• Additionally characters are independent, so characters form a basis for the space
of class functions
• If g, h ∈ G, then g and h are conjugate iff χ(g) = χ(h) for all characters of G
• Definition. Character table: the k × k matrix with ijth entry χi (gj ) for irre-
ducible characters χ1 , . . . , χk of a group G with conjugacy class representatives
g1 , . . . , gk
• The character table is an invertible matrix
Theorem. Let χ1 , . . . , χk be the irreducible characters of G and let g1 , . . . , gk be

representatives of the conjugacy classes. Then the following properties hold for
any r, s ∈ [k]:
• Row orthogonality:
k
X χr (gi )χs (gi )
= δrs
|CG (gi )|
i=1
• Column orthogonality:
k
X
χi (gr )χi (gs ) = δrs |CG (gr )|
i=1
Here δrs is the Kronecker delta.
• The inner product of two different columns will be 0, but the inner product of
one column with itself is the size of its centralizer
Proof. Pick a class function ψs : G → C with ψs (gr ) = δrs . Then ψs (g) = 1 if g is

k
X
conjugate to gs and 0 otherwise. ψs can be written as λi χi where
i=1
1 X
λi = hψs , χi i = ψs (g)χi (g).
|G|
g∈G
Then
1 G χi (gs )
λi = |gs |χi (gs ) =
G |CG (gs )|
22
where CG (gs ) is the centralizer. Therefore

k k
X X χi (gr )χi (gs )
δrs = ψs (gr ) = λi χi (gr ) =
|CG (gs )|
i=1 i=1
as claimed.
• Either set of relations can be used to determine the missing elements of a char-
acter table
• Note that writing CG ∼ = W1 ⊕ · · · ⊕ Wk is canonical where Wi is the sum of dim Vi
copies of Vi , but the breakdown of Wi is not canonical
17 October 1
• Relate characters of groups to avoid doing extensive group theory and algebra
each time
Proposition. Let N C G be a normal subgroup, and let χ̃ be a character of

G/N. Define χ : G → C by χ(g) = χ̃(N g). Then χ is a character of G with the
same degree as χ̃.
• Definition. Lift: for N C G and a character χ̃ of G/N, the character χ of G given

by χ(g) = χ̃(N g) for g ∈G
• Character of quotient group can be used to construct a character of the original
group using cosets
• The lift χ is irreducible iff χ̃ is irreducible, using inner product relation
Theorem. Assume that N C G. There is a bijection between the set of characters

of G/N and the set of characters χ of G containing N in the kernel, given by
associated each character of G/N with its lift to G. Irreducible characters of G/N
correspond to irreducible characters of G containing N in their kernel.
If N C G then there exist irreducible characters χ1 , . . . , χs of G

Proposition. T
such that N = si=1 ker χi .
Proposition. The group G is not simple iff χ(g) = χ(1) for some non-trivial
irreducible character χ of G and some non-identity element g.
23
• In this case g ∈ ker χ and ker χ is a normal subgroup

• Linear character has degree 1
• Definition. Derived subgroup: the subgroup G0 of G generated by all commu-
tators: G0 = {g −1 h−1 gh}
• Claim linear characters of G are lifts to G of irreducible characters of G/G0
Proposition. If χ is a linear character of G, then G0 ≤ ker χ.
Proposition. Assume N C G. Then G0 ≤ N iff G/N is abelian. In particular,

G/G0 is abelian.
• G0 is the smallest normal subgroup of G with an abelian quotient group

• Image of linear character is an abelian group, so G0 (the non-commutative part
of G) must be contained within the kernel of a character
Theorem. The linear characters of G are precisely the lifts to G of the irreducible
characters of G/G0 . In particular, the number of distinct linear characters of G is
equal to |G/G0 | and divides |G|.
Proposition. Suppose that χ is a character of G and λ is a linear character of

G. Then χλ defined as χλ(g) = χ(g)λ(g) is a character of G. Additionally χ is
irreducible iff χλ is irreducible.
• This is because linear characters correspond to representations with image C,

which is abelian
• Want to generalize product of characters and use this to systematically find ir-
reducible characters of groups
• Definition. Tensor product: given C-vector spaces V, W with bases v1 , . . . , vm
and w1 , . . . , wn , the tensor product space V ⊗W is as an mn-dimensional C-vector
space with basis vi ⊗ wj
P P X
• For v = λi vi and w = µj wj , v ⊗ w is defined as λi µj (vi ⊗ wj )
i,j
• Tensor product is bilinear and non-commutative

• Note that not every element of V ⊗ W can be written as v ⊗ w
24
Proposition. If v ∈ V, w ∈ W, and λ ∈ C, then v ⊗ (λw) = (λv) ⊗ w = λ(v ⊗ w).

Additionally if x1 , . . . , xa ∈ V and y1 , . . . , yb ∈ W, then
X X X
xi ⊗ yj = xi ⊗ yj .
i,j
• Any bases of V, W work just as well

• For CG-modules, tensor product is again spanned by vi ⊗ wj and the group mul-
tiplication is defined (vi ⊗ wj )g = vi g ⊗ wj g and extended to general elements as
(v ⊗ w)g = vg ⊗ wg
• However it is not true that (v ⊗ w)r = vr ⊗ wr for general elements r in CG, for
example if r = 2g for g ∈ G
Proposition. Let V and W be CG-modules with characters χ, ψ. Then the

character of V ⊗ W is the product character χψ defined as
χψ(g) = χ(g)ψ(g) ∀g ∈ G.
• Proof by fixing g and choosing basis such that matrix of g is diagonal so that the
trace is easy to compute
1 α1 α12 · · · α1r−1
 
1 α2 α2 · · · αr−1 
• Definition. Vandermonde matrix: invertible matrix  2 2 
. . . ... . 
2
1 αr αr · · · αr r−1
for distinct αi
Theorem. Let χ be a faithful character of G such that χ(g) takes on exactly r

different values over g ∈ G. Then every irreducible character of G is a constituent
of one of the powers χ0 , χ1 , . . . , χr−1 .
• Proof by showing hχj , ψi =

6 0 for an irreducible character ψ and some index j; do
this using Vandermonde matrix
• Suppose V is a CG-module with character χ so V ⊗ V has character χ2
• Define T : V ⊗ V → V ⊗ V by (vi ⊗ vj )T = vj ⊗ vi
• S(V ⊗ V ) = {x ∈ V ⊗ V : xT = x} is the symmetric part of V ⊗ V and A(V ⊗ V ) =
{x ∈ V ⊗ V : xT = −x} is the antisymmetric part
25
Proposition. S(V ⊗V ) and A(V ⊗V ) are CG-submodules of V ⊗V with V ⊗V =

S(V ⊗ V ) ⊕ A(V ⊗ V ).
Proposition. Let v1 , . . . , vn be a basis of V. The vector vi ⊗ vj + vj ⊗ vi form a

basis of S(V ⊗ V ) and the vectors vi ⊗ vj − vj ⊗ vi form a basis of A(V ⊗ V ).
Proposition. If χS is the character of S and χA is the character of A, then

χS (g) = 21 χ2 (g) + χ(g 2 ) and χA (g) = 12 χ2 (g) − χ(g 2 ) .
• This is by choosing a basis {e1 , . . . , en } such that

X ei g = λi ei so g acts on (vi ⊗ vj −
vj ⊗ vi ) by multiplication by λi λj , so χA (g) = λi λj
i<j
• The direct sum gives χ2 = χS + χA

• These propositions give a method for finding irreducible characters: starting
with some irreducible characters and a character χ, find χS and χA and use
inner products to find new irreducible characters and repeat
• Tensor products can also be used to determine the character table of a direct
product G × H
• Choose V a CG-module with basis v1 , . . . , vm and W a CH-module with basis
w1 , . . . , w n
• Then define (vi ⊗ wj )(g, h) = vi g ⊗ wj h and extend linearly
• Then the character of V ⊗ W is χ × ψ defined as (χ × ψ)(g, h) = χ(g)ψ(h)
Theorem. Suppose χ1 , . . . , χa are distinct irreducible characters of G and

ψ1 , . . . , ψb are distinct irreducible characters of H. Then G × H has ab distinct
irreducible characters given by χi × ψj .
18 October 6
• Relate character of G to subgroup of G
• Restrict a CG-module V to subgroup H of G, giving a CH-module denoted V ↓ H
• If V ↓ H is an irreducible CH-module, then it must be an irreducible CG-module
as well
26
• Character χ ↓ H agrees with χ on elements in H and is zero everywhere else

• Inner product on restricted character χ ↓ H restricts sum to just elements of H,
1 P
i.e. hφ1 , φ2 iH = |H| φ
h∈H 1 (h)φ 2 (h)
• If ψi are irreducible characters of H, then χ ↓ H = d1 ψ1 + · · · + dr ψr and ψi is a

constituent of χ ↓ H if its coefficient is nonzero
Proposition. Let H be a subgroup of G and let ψ be a nonzero character of H.

Then there exists an irreducible character χ of G such that hχ ↓ H, ψiH 6= 0.
Proposition. Let H be a subgroup of G, let χ be an irreducible character of

G, and let ψ1 , . . . , ψr be the irreducible characters of H. Then χ ↓ H = d1 ψ1 +
r
X
· · · + dr ψr , where the non-negative integers d1 , . . . , dr satisfy d2i ≤ |G : H|,
i=1
with equality iff χ(g) = 0 for all elements g ∈ G\H.
r
X 1 X
• This follows from d2i = hχ ↓ H, χ ↓ HiH = χ(h)χ(h) and hχ, χiG = 1
|H|
i=1 h∈H
Proposition. Suppose that H C G. Let V be an irreducible CG-module and U

an irreducible CH-submodule of V ↓ H. For every g ∈ G, let U g = {ug : u ∈ U }.
Then
1. The set U g is an irreducible CH-submodule of V ↓ H and has the

same dimension as U.
2. Interpreted as a CH-module, V is a direct sum of some of the CH-

modules U g.
3. If g1 , g2 , g ∈ G and U g1 , U g2 are isomorphic as CH-modules, then U g1 g

and U g2 g are isomorphic as well.
Clifford’s Theorem. Suppose that H C G and that χ is an irreducible character

of G. Then all the constituents of χ ↓ H have the same degree, and if ψ1 , . . . , ψm
are the constituents of χ ↓ H then χ ↓ H = e(ψ1 + · · · + ψm ) for some positive
integer e.
• Previous proposition implies that all constituents of χ ↓ H have the same degree
27
Proposition. Suppose that H is a normal subgroup of index 2 in G, and let χ

be an irreducible character of G. Then either χ ↓ H is irreducible, or it is the
sum of two distinct irreducible characters of H of the same degree.
X
• Constituents must satisfy d2i ≤ [G : H] = 2
• From G/H ∼ = C2 , the non-trivial linear character of G/H lifts to the character
λ(g) = 1 if g ∈ H, −1 otherwise
Proposition. Let H be a normal subgroup with index 2 in G and let χ be an

irreducible character of G. Then the following are equivalent:
• χ ↓ H is irreducible
• χ(g) 6= 0 for some g ∈ G, g ∈
/H
• χ and χλ are not equal in G, where λ is the nontrivial character of C2
• If H C G with |G : H| = 2 and χ is an irreducible character of G such that χ ↓ H

is irreducible, then if φ is irreducible in G and φ ↓ H = χ ↓ H then φ = χ or
φ = χλ
• In the other case that χ is irreducible in G but χ ↓ H = ψ1 + ψ2 in H, then any
irreducible character φ of G such that φ ↓ H has ψ1 or ψ2 as a constituent must
be equal to χ
• Restriction constructs CH-module from CG-module; induction attempts to con-
struct CG-module from CH-module
Proposition. Let H ≤ G be a subgroup and let U be a CH-submodule of CH.

If ϑ is a CH-homomorphism from U to CG, then there exists r ∈ CG such that
uϑ = ru for all u ∈ U.
• Such homomorphisms take a simple form

• Construction writes CH = U ⊕ W with φ : CH → CG given by u + w 7→ uϑ
• Then if r = 1φ, uϑ = ru
• As a result if U, V are CG-submodules of CG, U ∩ V = {0} there exists r ∈ CG
such that ru = u and rv = 0 for all u ∈ U, v ∈ V
• Definition. Induction: given a subgroup H ≤ G and a CH-submodule U of CH,
the CG-module U ↑ G = U (CG) = span {ug : u ∈ U, g ∈ CG}
28
• Note that U is a CH-module, but U ↑ G is a CG-submodule of CG

• If U is a CH-submodule of CH with U = U1 ⊕ · · · ⊕ Um , then U ↑ G = (U1 ↑
G) ⊕ · · · ⊕ (Um ↑ G)
• Additionally if U, V are isomorphic CH-submodules of CH, then U ↑ G ∼
=V ↑G
as CG-modules
• This allows the definition of U ↑ G for U not a submodule of CH by decomposing
U as a direct sum of CH-submodules of CH
Theorem. Suppose that H ≤ K ≤ G. If U is a CH-module, then (U ↑ K) ↑ G ∼

=
U ↑ G.
• Definition. Induced character: given ψ the character of a CH-module U, the

character ψ ↑ G of the induced CG-module U ↑ G
Proposition. Suppose H ≤ G and U is a CH-submodule of CH, with V a CG-

submodule of CG. Then the vector spaces HomCG (U ↑ G, V ) and HomCH (U, V ↓
H) have equal dimensions.
• Let ϑ ∈ HomCG (U ↑ G, V ) and define ϑ̃ : U → CG to be the restriction of ϑ to U,

which is a map in HomCH (U, V ↓ H)
Frobenius Reciprocity Theorem. Suppose H ≤ G. Let χ be a character of G

and ψ a character of H. Then hψ ↑ G, χiG = hψ, χ ↓ HiH .
• Recall hψ ↑ G, χiG = dim(HomCG (U ↑ G, V )) and hψ, χ ↓ HiH = dim(HomCH (U, V ↓

H)) which are equal when χ, ψ are irreducible (since this depends on being a
submodule of CH, CG)
• As a corollary if f is a class function on G and ψ is a character of H, then hψ ↑
G, f iG = hψ, f ↓ HiH
(
ψ(g) g ∈ H
Proposition. Given a character ψ of H ≤ G, define ψ̇(g) =
0 g∈/ H.
Then the values of the induced character are given by
1 X
(ψ ↑ G)(g) = ψ̇(y −1 gy).
|H|
y∈G
29
• Proof by showing this function is a class function whose inner products with
irreducible characters χ of G agree with inner products of ψ ↑ G with χ
|G|
• Then the degree (ψ ↑ G)(1) = |H| ψ(1)
(
G 1 y ∈ xG
• For x ∈ G, define fx (y) = as a class function, where xG is the
0 y∈ / xG
conjugacy class of x
χ(x)
• Not difficult to see hχ, fxG iG = |CG (x)|
• If xG ∩ H = ∅, then (fxG ↓ H)(x) = 0
Proposition. Let ψ be a character of the subgroup H of G and suppose x ∈ G.

If no element of xG lies in H, then (ψ ↑ G)(x) = 0. Otherwise H ∩ xG 6= ∅, then

ψ(x1 ) ψ(xm )
(ψ ↑ G)(x) = |CG (x)| + ··· +
|CH (x1 )| |CH (xm )|
where xi are elements of H and fxG ↓ H = fxH1 + · · · + fxHm is a decomposition of

H ∩ xG .
19 October 8
• Definition. Algebraic integer: complex root of monic integer polynomial; equiv-
alently, an eigenvalue of some integer matrix
• Set of algebraic integers forms a ring
• Clean proof using tensor products
• Since χ(g) is a sum of roots of unity, χ(g) is an algebraic integer; thus if χ(g) is
rational, then it is an integer
X
• If C is conjugacy class of G, then C = x ∈ CG
x∈C
Lemma. Suppose g ∈ G and that C is the conjugacy class of G containing

g. If U is an irreducible CG-module with character χ, then uC = λu where
λ = |C|G| χ(g)
G (g)| χ(1)
.
• Because C is in the center of CG, there exists λ ∈ C such that uC = λu for all
u∈U
30
X
• Thus [x] = λI and taking the trace gives |C|χ(g) = λχ(1)
x∈C
X
Lemma. Let r = αg g ∈ CG, where each αg is an integer. If u is a nonzero
g∈G
element of CG such that ur = λu, then λ is an algebraic integer.
• This statement is equivalent to u being an eigenvector of a matrix in the αg

|G| χ(g)
• Thus if χ is irreducible and g ∈ G, λ = |CG (g)| χ(1) is an algebraic integer
Theorem. If χ is an irreducible character of G, then χ(1) divides |G|.
|G| χ(gi )
Proof. If g1 , . . . , gk are representatives for the conjugacy classes of G, then |CG (gi )| χ(1)
k
X |G| χ(gi )χ(gi )
and χ(gi ) are both algebraic integers. Then is an algebraic
|CG (gi )| χ(1)
i=1
|G|
integer which is equal to χ(1) by the inner product on χ. This value is clearly rational,
so it must be an integer and χ(1) divides |G|.
• Example. If |G| = p2 , χ(1) < p because the sum of squares of degrees of irre-
ducible characters is |G| and so every irreducible character has degree 1, which
implies G is abelian
• No simple group can have an irreducible character of degree 2, by considering
representation ρ : G → GL2 (C) with character χ
• Kernel of ρ is {1} because G is simple, so G is non-abelian and has no nontrivial
linear characters since its group of commutators must be all of G
• Since g → det(gρ) is a linear character, det(gρ) = 1 for all g and additionally G
has even order, thus containing an element of order 2
• The subgroup generated by this element is normal, contradicting G being simple
Corollary. Suppose that p is prime and the degree of every irreducible charac-
ter of G is a power of p. Then G has an abelian normal p-complement, meaning
there exists a normal abelian subgroup N of G with |G : N | a power of p and |N |
coprime to p. In particular this means G cannot be simple unless it has prime
order.
31
n
X
• ω i is an integer by inducting on n
i=1
gcd(i,n)=1
Theorem. Let g be an element of order n in G and suppose g is conjugate to g i

for all gcd(i, n) = 1. Then χ(g) is an integer for all characters χ of G.
Proof. Let V be a CG-module with character χ of degree m. By Schur’s Lemma, there

m
X
is some basis B such that [g]B is diagonal. Then χ(g) = ωi is a sum of nth roots
i=1
m
X n
X
of unity and χ(g i ) = ωji , so χ(g i ) ∈ Z, and by the conjugacy result this
j=1 i=1
gcd(i,n)=1
gives φ(n)χ(g) ∈ Z. Then χ(g) is rational, but it is an algebraic integer, so χ(g) is an
integer.
• Using Galois theory, the converse holds true as well

• As a corollary, all character values of symmetric groups are integers
• Note that all character values of S4 are integers, but this is not true for A4
Lemma. Let p be a prime and let g ∈ G. There exist x, y ∈ G such that g = xy =

yx, ord(x) is a power of p, and p - ord(y). Furthermore these x, y are unique.
ν
• If order of g is upν , then au + bpν = 1 and setting x = g au , y = g bp works
• The element y is the p0 -part of g
• With ζ a primitive nth root of unity, pZ[ζ] = {pr : r ∈ Z[ζ]} which is a principal
ideal of Z[ζ]
• Because there are only finitely many ideals of Z[ζ] containing pZ[ζ], there is a
maximal ideal containing pZ[ζ]
• This maximal ideal P is prime, and P ∩ Z = pZ
Theorem. Let g ∈ G and let y be the p0 -part of g. If χ is any character of G, then

χ(g) − χ(y) ∈ P where P is the maximal ideal of Z[ζ] containing pZ[ζ].
ν
• If ord(g) = m = upν then y = g bp with order u and both of these divide |G| = n
32
X
• Both χ(g), χ(y) can be written as ωi for ωin = 1
ν ν 2ν
• If s is an mth root of unity, then s = sau+bp and sp = sbp ; by binomial expan-
ν ν ν
sion (s − sbp )p ∈ pZ[ζ] and s − sbp ∈ pZ[ζ], so applying this term by term shows
χ(g) − χ(y) is in this ideal
• As a corollary if χ is a character such that χ(g), χ(y) are both integers, then
χ(g) ≡ χ(y) mod p
• If the order of g is a power of p and χ(g) ∈ Z, then χ(g) ≡ χ(1) mod p because the
p0 -part of g is the identity
20 October 15
• Recall that U ↑ G is the CG-module obtained by multiplying elements of U by
elements of the group algebra
Alternatively, let IndG
L
H (W ) = s∈R sW where R is a set of representatives of
•
the cosets of H
• Definition. Double coset: if K, H are subgroups of G, the double coset KgH =
{kgh : K ∈ K, h ∈ H} and the set of all double cosets K\G/H partition G
• With H, K ≤ G, choose ρ : H → GL(W ) a linear representation of H and let
V = IndG
H (W )
• For s ∈ S, let Hs = sHs−1 ∩ K be a subgroup of K and set ρs (x) = ρ(s−1 xs) for
x ∈ Hs
• Induce this representation to K: IndK
Hs (Ws )
Proposition. The restriction ResK V = ResK IndG H (W ) is isomorphic to the

K
direct sum of the representations IndHs (Ws ) for each s ∈ S.
L
• By definition V = x∈G/H xW as an induced module
• If V (s) = span {xW, x ∈ KsH} is a subspace of V for s ∈ S, then V is a direct
sum of the V (s) because double cosets partition G
• V (s) is stable under K
Mackey’s Irreducibility Criterion. Let H be a subgroup of G and let W be a

CG-module corresponding to a representation ρ. Then V = IndG H W is irreducible
s
iff both W is irreducible and for each s ∈ G − H, ρ and Ress (ρ) of Hs are disjoint
where Hs = sHs−1 ∩ H, ρs = ρ(s−1 xs), and Ress (ρ) is the restriction from H to Hs .
33
• As a corollary, if H is normal in G then Hs = H and Ress (ρ) = ρ, so IndG H (ρ)

is irreducible iff ρ is irreducible and ρ is not isomorphic to any conjugate ρs for
s∈/H
• Definition. Isotypic representation: direct sum of isomorphic irreducible rep-
resentations
Proposition. Let A be a normal subgroup of G and let ρ : G → GL(V ) be an

irreducible representation of G. Then either there exists a subgroup A < H < G
and an irreducible representation σ of H such that ρ is induced by σ, or the
restriction of ρ to A is isotypic.
• If A is abelian and the restriction of ρ to A is isotypic, then ρ(a) is a homothety
Corollary. If A is an abelian normal subgroup of G, the degree of each irre-

ducible representation ρ of G divides the index [G : A].
• Definition. Semidirect product: the product A · H where A ∩ H = 1, so every

element of the product can be uniquely written as ah for a ∈ A, h ∈ H
21 October 20
• G is a finite group and V is a finite dimensional complex vector space
• Representation of G is group homomorphism ρ : G → GL(V ), where GL(V ) is
isomorphic to GLn (C) for some chosen basis
• CG-module is a vector space V equipped with a linear action of G on vectors,
which carries the same information as a representation
• Degree of representation corresponds to dimension of module and equivalence
of representations corresponds to isomorphism of modules
• CG-submodule is a subspace that is stable under G; this corresponds to a change
of basis such that the representation matrices are all upper/lower block-triangular
• Maschke’s Theorem goes further to say that such matrices can be simultane-
ously block-diagonalizable, i.e. if U ⊆ V then there is another CG-submodule W
such that V = U ⊕ W
• The proof first picks any subspace W 0 with U ⊕ W 0 = V, where W 0 is not neces-
sarily G-stable
• Instead, construct a projection from V to U with kernel that is G-stable
34
• This projection works by averaging a linear map p0 from V to U over G: v 7→

1 X
ρ(g)(p0 (ρ(g −1 )v))
|G|
g∈G
• Thus if v is in the kernel of p, p(v) = 0 and since ρ(g)p(v) = 0, p(ρ(g)v) = 0 and

the kernel is G-stable
• Schur’s Lemma says that for irreducible CG-modules V, W HomCG (V, W ) is C if
V ∼
= W and 0 otherwise
• As a result, every irreducible representation of G is 1-dimensional iff G is abelian
• The regular representation is the |G|-dimensional vector space CG with natural
basis and group action acting as a permutation matrix on the basis (elements
of G)
• This views the group algebra as a module over itself
• The importance of the regular representation is that it contains every irre-
ducible representation of G
• If V1 , . . . , Vk with dimensions ni is a complete
Llist of all irreducible representa-
tions of G, CG is canonically the direct sum ki=1 ni Vi
• ni Vi is the direct sum of ni copies of Vi , but this is not canonical and depends on
a choice of basis
k
X
• Counting dimensions, |G| = n2i where ni divides |G| and furthermore ni di-
i=1
|G|
vides |A| where A ⊂ G is any normal abelian subgroup
• The character χV of a representation (V, ρ) is a homomorphism sending g to the

trace of ρ(g)
X
1
• For any two characters χV1 , χV2 , the inner product is defined as |G| χV1 (g)χV2 (g)
g∈G
which is equal to dim HomCG (V1 , V2 )
• Thus representations are determined by their characters and irreducible repre-
sentations have orthogonal characters
• The character of the regular representation is |G| on the identity and 0 on ev-
k
X
erything else, so χreg = n i χ Vi
i=1
35
• The representation theroy of G is captured by a character table (which does not

uniquely determine the group in general) where rows are distinct irreducible
representations and columns are conjugacy classes
• If H is a subgroup of G and V, ρ is a representation of G, it can also be viewed
as a representation of H which is the restriction to H
Ln
• If (W, π) is a representation of H, extending to G gives IndG H (W ) = i=1 gi W
where gi are coset representatives of H in G
• The regular representation is the induction of the trivial representation from
the trivial group to G
• By Frobenius reciprocity, hIndG
H (W ), V iG = hW, ResH (V )iH
• As a special case if IndG

H (W ) is irreducible, then W is irreducible and it appears
exactly once in ResH IndG H (W )
• Mackey’s criteria say that this is equivalent to W being irreducible and W, W s

non-isomorphic as representations of H ∩sHs−1 , where W s is the representation
given by conjugating by s
  
 1 a b 
• Heisenberg group H(Fq ) = 0 1 c  : a, b, c ∈ Fq has characters that are
0 0 1
 
either linear or determined by the central character θ̃ that sends a matrix to a
linear homomorphism θ(b)
• This is part of a short exact sequence 0 → Fq → H(Fq ) → F⊕2
q →0
22 November 12: Applications of Representation Theory in

Computer Vision
• Goal of computer vision is to gain high-level understanding from digital images
to aid in medicine or robotics, e.g. identifying and predicting cancer from images
• Definition. Sample complexity: number of training samples needed for learn-
ing algorithm to learn given function
• Collecting sufficient data is difficult for algorithms with high sample complexity,
especially when input is corrupted by noise that causes algorithm to output an
incorrect answer
• Models tend to be not particularly robust, e.g. simply rotating an image causes
wildly inaccurate classification
36
• Use representation theory to address sample complexity and robustness

• Definition. Homogeneous space: given a group G acting on set X, X is a homo-
geneous space if for any x, y ∈ X, there exists g ∈ G such that y = gx
• In computer vision, X may be a set of images and G may be some group of image
transformations (e.g. rotations, dilations, reflections, etc.)
• Choose x0 ∈ X and associate x0 with its stabilizer Gx0 = {g ∈ G : gx0 = x0 } to
create a bijection associating X with G/Gx0
• Definition. Lifted function: given f : X → C, f ↑G : G → C sends g to f ([g]X )
where X is a quotient group of G and [g]X is the coset containing g
• Definition. Convolution: with G countable and X , Y quotient groups of G with
f : X → C, g : Y → C, the convolution f ∗ g is defined as
X
(f ∗ g)(u) = f ↑G (uv −1 g) ↑G (v)
v∈G
for u ∈ G
X
• Normal convolution over an image is defined as (f ∗K)(x) = f (x−u)K(u)
u∈[w]×[w]
for a filter/kernel K of size w × w
• Given set X and vector space V, denote LV (X) as the space of functions f : X →
V where elements of X are neurons and f (x) is the activation of x
• Definition. Multi-layer feed-forward NN: let X0 , . . . , Xl be a sequence of index
sets with vector spaces V0 , . . . , Vl and linear maps φ1 , . . . , φl where φi : LVi−1 (Xi−1 ) →
LVi (Xi ) and σi : Vi → Vi are non-linear functions; then the multi-layer feed-
forward NN is a sequence of maps f0 7→ f1 7→ · · · 7→ fl with fi (x) = σi (φi (fi−1 (x)))
• Each Xi is a layer of neurons and each map fi sends one layer to the next by
combining linear transformation with pointwise non-linearity
• Feed-forward component comes from input signals always moving forward to
deeper layers and never going backwards
• Definition. G-convolutional NN: with N a multi-layer feed-forward network in
which the ith index set is G/Hi for Hi ≤ G, N is a G-CNN if φi is a generalized
convolution of the form φi (fi−1 ) = fi−1 ∗ Ki for some filter Ki
• CNN has fewer parameters than fully connected feed-forward network, and since
it applies the same filter everywhere in the image, it can identify the same fea-
ture under translations in the image
37
• Definition. Intertwiner: given G-sets X, Y, f : X → Y is equivariant if f (g · x) =

g · f (x) and an intertwiner is an equivariant linear map between F G-modules
(F G-homomorphism)
• Definition. G-equivariant feed-forward NN: multi-layer feed-forward NN N if
transforming the input by f0 7→ g · f0 (for any g ∈ G) transforms the activations
of the other layers by fi 7→ g · fi for the same value of g
Theorem. If G is a compact group and N is an L + 1-layer feed-forward NN, then

N is G-equivariant iff it is a G-CNN.
• A CNN feature map takes an image as input and passes it through several layers
of convolutions, feature maps, subsampling, etc. to produce some output
• Convolution can be unrolled to be represented as matrix multiplication (although
it is inefficient), but it is easier to think about
• CNNs are already equivariant to translation, but they do not perform well under
rotation
• Definition. Wallpaper group: discrete group of isometries of the Euclidean
plane containing two linearly independent translations, with point group the
subgroup generated by elements fixing the origin (excludes translations)
• Rather than making each filter equivariant, make the entire set of kernels equiv-
ariant
• Apply the group element to each kernel in the original set of filters to produce
a larger set of filters and apply each filter to the image to produce separate
channels, so that applying G to the image will just cyclically permute the order
of results in the channels
• Represent image as f : Z2 → Rk (0 outside the image pixels) and let FS be the
set of all images with domain pixels S
• Definition. Steerable CNN: define a CNN Φ : FS → FF with a filter bank
Ψ : FB → Rk where the size |B| of the filter is much smaller than the image size;
make the filter bank H-invariant to make φπ (g)f = π 0 (g)φf for all g ∈ G and
f ∈ FS
• Due to convolution, this setup causes each layer to become G-equivariant using
the induced representation of H to G
• Steerable CNN uses Z2 o H, but other G-CNNs may use other groups, e.g.
SO(3) o H (spherical CNN)
38
• Overall strategy is to make the kernel invariant under the G-stabilizer of some
particular point
• Fast Fourier Transform exists for abelian groups and Sn but not in general
23 November 17: The McKay Correspondence

• McKay correspondence as a bijection between finite subgroups of SU (2) (special
unitary group) and ADE diagrams
• Definition. ADE diagrams: five graphs defined as An a path of length n, Dn a
path of length n − 2 with two branches of length 1 at the end, and E6 , E7 , E8 are
fixed graphs resembling a path with an extra leaf
– An :
– Dn :
– E6 :
– E7 :
– E8 :
• These graphs correspond to integral solutions of p1 + 1q + 1
r > 1 where branches
of length p − 1, q − 1, r − 1 emanate from a central node
1 1 1
• There are many mathematical structures relating to solutions of p + q + r >1
and thus ADE diagrams
• Symmetries of platonic solids must satisfy this relation
• Given a platonic solid, barycentrically subdivide the faces and project the re-
sulting triangles onto a sphere
• There are 2p, 2q, 2r triangles around each edge, vertex, face respectively, so the
internal angles of the spherical triangles are πp , πq , πr

• Surface area of each triangle is R2 πp + πq + πr − π , so p1 + 1q + 1r > 1
• Platonic solids usually described in terms of symmetries, which are finite sub-
groups of SO3
Theorem. Every finite subgroup of SO(3) is isomorphic to one of

Cycn , D2n , T, O, I where T is the tetrahedral group, O is the octahedral group,
and I is the icosahedral group.
39
Orbit-Stabilizer. Let G be a group acting on a set X. For all x ∈ X,
|OrbG (x)||StabG (x)| = |G|.
• If G is a finite subgroup of SO(3), each non-identity element has an axis of ro-

tation, each with two poles
• If P is the set of all poles, then P is divided into m orbits under the action of G
• Count the size of the set I = {(g, p) : g 6= 1, g(p) = p} which is the total number
of times any pole is mapped to itself by a non-identity element of G
• Computing this number in two ways restricts the possible stabilizer sizes of each
pole and also recovers the ADE problem
• Definition. Unitary matrix: complex square matrix with U U ∗ = I
• There is a standard double cover π : SU (2) → SO(3) which allows recovery of
the finite subgroups of SU (2) from SO(3)
• Definition. Spectral radius: given a graph G, the spectral radius ||G|| = max{|λ|}
over eigenvalues λ of the (undirected) adjacency matrix of G
• Spectral radius is some measure of graph complexity
Lemma. If G, H are disjoint graphs, then ||G ∪ H|| = max{||G|| , ||H||}.
Perron-Frobenius. Let G be a connected graph. If the adjacency matrix AG

has an λ-eigenvector with positive real entries, then ||G|| = λ. For any subgraph
H ⊆ G, ||H|| ≤ ||G|| .
• Proof is purely linear algebra

• If G contains any edge, then ||G|| ≥ 1 using the subgraph property
Theorem. If G is a connected graph with ||G|| < 2, then G is an ADE diagram.
Proof. In general, we use a proof technique of assigning weights to each vertex of a

subgraph H. If the sum of all the weights of the neighbors of each vertex of H is a
constant λ times the weight of the vertex, then H has spectral radius at least λ.
If C is a cycle of G, the adjacency matrix of the subgraph C has eigenvalue 2, so

G must be a tree. We can show G has no vertex with degree ≥ 4 by taking a star
subgraph with central weight 2 and leaf weights of 1, and similarly we can show G
40
has at most one vertex of degree 3. Therefore G must consist of a central vertex of
degree ≤ 3 with at most 3 lines emanating from it, so G is represented by some triple
(p, q, r) like in the ADE problem. The goal from here is to constrain p, q, r and show
1 1 1
p + q + r > 1.
This is done by casework showing that otherwise G must contain as a subgraph at
least one of (3, 3, 3), (2, 4, 4), (2, 3, 6).
• Definition. Cartan matrix: given a finite (directed) graph G indexed by {1, . . . , r},
the matrix with entries (2δij − nij ) where δ is the Kronecker delta and nij is the
number of edges connecting i and j
• Definition. Positive definite matrix: matrix M for which there exists a vector
r
X
(xj ) such that cij xj > 0
j=1
Lemma. If G is a connected graph whose Cartan matrix is positive definite,

then G is an ADE diagram.
• Consider n-dimensional representation W of some finite group G over C, and let

{V1 , . . . , Vr } be a complete set of non-isomorphic irreducible representations of
G
• For each i, j, let nij be the multiplicity of Vi in W ⊗Vj so nij = dim Hom(Vi , W ⊗Vj )
• Definition. McKay diagram: given a group G, the McKay diagram of G with
respect to W is the graph on {1, . . . , r} where i, j are connected by nij edges
• Reduced McKay diagram has the trivial representation removed as a vertex
• If W is self-dual then the McKay diagram is undirected with nij = nji
• If W is faithful, then the McKay diagram is connected
• The vector (dim Vj ) is an eigenvector of (nij ) with eigenvalue dim W
McKay Correspondence. If G is a subgroup of SU (2) and W is its natural two-

dimenisonal representation, then the reduced McKay diagram of G with respect
to W is an ADE diagram.
Proof. The natural two-dimensional representation W of G is faithful and self-dual,

so the McKay diagram is undirected and connected. The Cartan matrix has an eigen-
vector (dim Vj ) with eigenvalue 0 because
(2δij − nij )(dim Vj ) = (2 − dim W )(dim Vj ) = 0.
41
However, removing the trivial representation makes this product strictly positive.
Then the Cartan matrix is positive definite, and the reduced McKay diagram is an
ADE diagram.
24 November 19: Applications of Representation Theory to

Physics
• Definition. Lie group: group which is also a finite-dimensional smooth mani-
fold (locally similar to Euclidean space)
• Definition. Lie algebra: tangent space g to Lie group G; specifically, the tangent
hyperplane at the identity of an embedding of G in Rn
• The tangent to any path through the identity is in the Lie algebra
• Definition. Commutator: [X, Y ] = XY − Y X
• Lie algebra is closed under commutator operation by constructing a path etY Xe−tY
with derivative at 0 given by [X, Y ]
• Definition. Lie bracket: any binary operator g × g → g, e.g. the commutator
• Definition. Structure constant: given generators T aP
, T b , T c of the Lie algebra g,
constants f abc for which [T , T ] = if T as a sum c f abc T c over C
a b abc c
• Definition. Lie algebra homomorphism: map π 0 : g1 → g2 that is linear and

preserves the Lie bracket
• Definition. Dirac notation: column vector (ket) denoted |ψi and row vector (bra)
denoted hψ| = (|ψi)∗ using the conjugate transpose
• These represent possible states of measurable quantities
• Inner product on state (bra-ket) denoted hφ|ψi = hψ|φi is linear and orthogonal
over the standard basis
• In quantum physics, measurable quantities take discrete pre-determined val-
ues, which are eigenvalues of an operator associated with the quantity (must be
real)
• Result of measuring an observable Ô (which is a Hermitian operator) is given
by the inner product hψ|Ô|ψi
• Hamiltonian is the observable associated with energy with solution states that
form a basis for the vector space of wavefunctions
42
Spectral Theorem. Let H be a Hermitian operator on a vector space V. There

is an orthonormal basis of V consisting of eigenvalues of H.
• Definition. Spin: (quantized) intrinsic component of angular momentum

0 1
• Spin operators S1 , S2 , S3 correspond to spin along axes x, y, z: σ1 = , σ2 =
1 0

0 −i 1 0
, σ3 =
i 0 0 −1
• These are usually written in the eigenbasis of the third operator, and this or-
thonormal basis can be used to measure spin along any axis
• Pauli matrices obey commutation relation where [σi , σj ] = 2iεijk σk , where εijk is
sgn(ijk) if the indices are a permutation of 1, 2, 3 and 0 otherwise
• This reflects the symmetry of the special unitary group SU (2)
• Group of 1 × 1 unitary matrices is given by {eiθ |θ ∈ R}
• Irreducible representations of U (1) biject to the integers by πn (eiθ ) = einθ for
n∈Z
• Given the spin operators Sj = 21 σj , the raising operator S+ = S1 + iS2 and
lowering operator S− = S1 − iS2 raise and lower the eigenvalues of S3 by discrete
amounts respectively
• This is useful in determining the irreducible representations of SU (2)
• Systems of multiple particles can result in spin that is not ± 21 by addition of
angular momentum
1
• Quantum computation uses spin 2 particles as qubits, which allows for gates
beyond classical gates
25 December 1: Monstrous Moonshine

Theorem. Every finite simple group either belongs to one of several infinite
classes:
• Cyclic groups Z/pZ of prime order
• Alternating groups An for n ≥ 5
• Groups of Lie type
or it is one of 26 exceptions, known as the sporadic groups.
43
• Monster group is the largest of the sporadic groups

• Definition. Modular
function: holomorphic function that is invariant under
az+b
SL2 (Z), i.e. f cz+d = f (z)
• j-invariant is modular function arising from elliptic curves

• Definition. Coxeter group: with G a graph on r1 , . . . , rn , the group generated
by r1 , . . . , rn subject to ri2 = 1, (ri rj )2 = 1 for all ri , rj not joined by an edge, and
(ri rj )3 = 1 for all ri , rj joined by an edge
• This is a reflection group; deflating all generated translations makes it finite
• Instead if every translation τ has τ 2 = 1, a biflation is obtained
• Construct bi-monster group using Coxeter group of indicidence graph of P3 (fi-
nite projective plane of order 3) and deflating all free 12-gons
• This gives the Cartesian product of the monster group with itself, along with an
extra involution
• Representations of monster group related to q-expansion of normalized j-function
• Definition. Composition group: finite length series 1 = H0 C H1 C · · · C Hn = G
of a group G such that each Hi is a maximal strict normal subgroup of Hi+1
Theorem. Jordan-Hölder Theorem If a group has at least one composition se-

ries, then all such composition series have the same composition length, and the
composition factors are the same up to permutation and isomorphism.
• Definition. Elliptic function: meromorphic function that is periodic with re-

spect to a lattice L
• Definition. Eisenstein series: given lattice L = [1, τ ] := Z + τ Z with τ in the
upper half plane H, the weight k Eisenstein series for L is given by
X 1 X 1
Gk (L) = k
=
(m + nτ ) ∗
ωk
m,n∈Z w∈L
(m,n)6=(0,0)
• Elliptic curves over C can be identified with complex tori C/L
44
26 December 8
• Glauberman correspondence as way of relating characters of different groups
• Group S acts on group G by (g, s) 7→ g s and on functions f : G → G by f s (g) =
f (g s )
• Denote S-invariant character by IrrS (G) = {χ ∈ Irr(G)|χs = χ ∀s ∈ S}
• Example. If G acts on itself by conjugation, then IrrG (G) = Irr(G)
Glauberman. When S is solvable and gcd(|S|, |G|) = 1, there is a canonical

bijection between IrrS (G) and Irr(CG (S)).
• Here CG (S) are fixed points of G under S
Fixed Point Subset Lemma. Let S be a p-group acting on a set A and F a

field with characteristic p. If f : A → F is an S-invariant function, then
X X
f (a) = f (a).
a∈A a∈CA (S)
• Proof by partitioning A into orbits

• If f is the identity, then |A| ≡ |CA (S)| mod p
• Assume S is a p-group acting on G with gcd(|G|, p) = 1
• Claim the map K 7→ K ∩ C from the set of S-invariant conjugacy classes of G
into conjugacy classes of C is a bijection
• Let R be the ring of algebraic integers and m a maximal ideal containing pR, so
F = R/m is a field with characteristic p
• Since p - |G|, Maschke’s Theorem still holds
• Definition. Central character: for χ ∈ Irr(G), the character λχ : Z(FG) → F
defined from an irreducible FG-module V with character χ, with λχ (z) = λ with
vz = λv for all v ∈ V
• Class sums form a basis for Z(FG)
χ(x)|K|
Lemma. If K is a conjugacy class of G and x ∈ K, then λχ (K̂) = χ(1) .
45
• Glauberman map from IrrS (G) → Irr(CG (S)) constructed using map from irre-
ducible character χ to the central character λχ , then a restriction map to CG (S),
and finally back to G
• Canonical map Z(FC) → Z(FG) given by K

\ ∩ C → K̂ for K̂ an S-invariant
conjugacy class of G
def
• Then restriction map δχ (K
\ ∩ C) = λχ (K̂)
• The resulting restriction δχ is a central character of C
• Example. Let p = 2 and take G = H ×H, with S = Z/2Z acting on G by swapping
• If χ ∈ IrrS (G), then χ = θ1 × θ2 for θi ∈ Irr(H)
• Fixed point subset of G is isomorphic to H
• Using S-invariance, θ1 and θ2 are equal and a conjugacy class Hi of H corre-
sponds to the conjugacy class Hi0 = {(x, x)|x ∈ Hi }
Tilde-moving Lemma. For any χ ∈ IrrS (G) and any x ∈ C, χ̃(1)χ(x) =

χ(1)χ̃(x) in F.
• χ̃ is the result of the Glauberman map (composition of central character bijec-

tion, central character restriction, and central character bijection), but it is also
the unique irreducible constituent ψ of χ ↓ C whose multiplicity hψ, χ ↓ Ci isn’t
divisible by p
• Glauberman correspondence creates a natural bijection between S-invariant ir-
reducible characters of G and irreducible characters of CG (S)
46

18.704 Notes: 1 Introduction To Representation Theory

Uploaded by

Copyright:

Available Formats

18.704 Notes: 1 Introduction To Representation Theory

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

18.704 Notes: 1 Introduction To Representation Theory

Uploaded by

Copyright:

Available Formats

18.

1 Introduction to Representation Theory

• Example. Compact groups G = S 1 = {z ∈ C : |z| = 1}

2 Groups and Homomorphisms

Lagrange’s Theorem. If G is a finite group and H is a subgroup of G, then |H|

• Definition. Normal subgroup: N C G if g −1 N g = N for all g ∈ G

• If f : G → H is a group homomorphism, ker f C G and im f ≤ H

Theorem. Suppose f : G → H is a group homomorphism. Then G/ ker f ∼

• Construct isomorphism by mapping the coset Kg to f (g)

3 Vector Spaces and Linear Transformations

Rank-Nullity Theorem. For a linear transformation f : V → W, then dim V =

• For a change of basis matrix T, basis change is given by T −1 BT

Theorem. If ρ : G → GL(n, F ) is a representation of G over F and V = F n , then

• FG-module is analogous to a group acting on a vector space

6 FG-submodules and Reducibility

• Definition. Irreducible FG-module: non-zero F G-module V with no F G-submodules

• In F G, multiplication is associative and distributes over addition, and there is

Proposition. Let V and W be F G-modules and let φ : V → W be an F G-

• F G-isomorphism is an invertible F G-homomorphism and gives rise to isomor-

Proposition. If φ : V → W is an F G-isomorphism, then the inverse φ−1 :

Theorem. Suppose that V is an F G-module with basis B and W is an F G-module

Proposition. Let V be an F G-module and suppose V = U1 ⊕ · · · ⊕ Ur . Each

Proposition. If V is an F G-module with V = U1 + · · · + Ur where each Ui is

• Choose a subset {W1 , . . . , Ws } such that W1 + · · · + Ws is direct but adding any

Maschke’s Theorem. Let G be a finite group, let F be R or C, and let V be an

• The condition on F is important:

• θ has image in U because the image of φ is in U and U is an F G-submodule and

Theorem. If G is a finite group and F = R or C, then every nonzero F G-module

Proposition. Let V be an F G-module, where F = R or C and G is a fi-

• This is just the projection map

Schur’s Lemma. Let V and W be irreducible CG-modules.

1. If θ : V → W is a CG-homomorphism, then θ is either the trivial map

2. If θ : V → V is a CG-isomorphism, then θ is a scalar multiple of the

• θ must have an eigenvalue λ, so ker(θ − λI) is nonzero and thus is all of V

Proposition. Let V be a nonzero CG-module and suppose that every CG-

• If V is reducible, it has a projection map to a CG-submodule

Corollary. Let ρ : G → GL(n, C) be a representation of G. Then ρ is irreducible

Proposition. If G is a finite abelian group, then every irreducible CG-module

• If G is a finite abelian group and V is an irreducible CG-module, vgx = vxg for

Theorem. Every finite abelian group is isomorphic to a direct product of cyclic

• Use this structure result to determine irreducible representations of finite abelian

Proposition. Let G be a finite group and V a CG-module. If g ∈ G, there is

• This is by decomposing V under the action of the cyclic group hgi

• Example. If H is a normal subgroup of G, then the sum of the elements of H is

Proposition. Let V be an irreducible CG-module and let z ∈ Z(CG). Then

Proposition. If there exists a faithful irreducible CG-module, then Z(G) is

• Faithful irreducible CG-module gives injective homomorphism from Z(G) to C×

Proposition. Suppose that G is a finite group such that every irreducible

Summary. A degree n representation of a group G over a field F is a homomorphism

11 Irreducible Modules and the Group Algebra

Proposition. Let V and W be CG-modules and let ϑ : V → W be a CG-

• This follows from Maschke’s Theorem

Proposition. Let V be a CG-module, and write V = U1 ⊕ · · · ⊕ Us as a direct

• Proof using Schur’s Lemma

Theorem. Let CG be the regular CG-module and write CG = U1 ⊕ · · · ⊕ Ur as

Proof. Let W be an irreducible CG-module and choose w 6= 0 ∈ W. Then W = {wr :

• As a result if G is finite, there are only finitely many non-isomorphic irreducible

• An example of a nontrivial representation is given by the group algebra CG

12 More on the Group Algebra