Solutions to Linear Representations of Finite Groups

Bryn Elesedy∗
Department of Computer Science
University of Oxford

July 19, 2020

Contents
2 Character Theory 2
2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

3 Subgroups, Products and Induced Representations 4

3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

∗ [email protected]

1
2 Character Theory
2.1
Let V and V 0 be the corresponding representations. Then χ+χ0 is the character of the direct sum V ⊕V 0 .
The character of the alternating square is then, for any s in the group,
1
(χ + χ0 )2σ = ((χ(s) + χ0 (s))2 + χ(s2 ) + χ0 (s2 ))
2
1 1
= (χ(s)2 + χ(s2 )) + (χ0 (s)2 + χ0 (s2 )) + χ(s)χ0 (s)
2 2
2
= χ2σ (s) + χ0σ (s) + χ(s)χ0 (s).
We can do the same thing for the symmetric square
1
(χ + χ0 )2α = ((χ(s) + χ0 (s))2 − χ(s2 ) + χ0 (s2 ))
2
1 1
= (χ(s)2 − χ(s2 )) + (χ0 (s)2 − χ0 (s2 )) + χ(s)χ0 (s)
2 2
2
= χ2α (s) + χ0α (s) + χ(s)χ0 (s).

2.2
In the permutation representation, the matrix ρs will be a permutation matrix, with 1 at position (i, i)
if i is fixed by s and 0 on the diagonal otherwise. The character χ(s) is the trace of this matrix, so is the
number of elements fixed by s.

2.3
We may choose bases {ei } for V and {e0j } V 0 such that hei , e0j i = δij . Then for any matrix A we have,
in the given bases, X
hρs x, Ax0 i = xi x0j (ρs )ki Akj .
ij

So the map is invariant if and only if A = (ρ−1 >

We can then define a representation by ρ0s = (ρ−1
s ) .
>
s ) ,
which is clearly a representation. Existence and uniqueness have therefore been established.

2.4
The map ρs is clearly linear and invertible, since the ρi,s are linear and invertible. Further, for any f in
W
ρst f = ρ2,st f ρ−1 −1 −1
1,st = ρ2,s ρ2,t f ρ1,t ρ1,s = ρs ρt f.
We can then calculate the character by expanding in a basis. Let f have components fab in a basis for
W . Then (using Einstein notation)
(ρs f )ij = (ρ2,s )iα (ρ−1
1s )βj fαβ .

We calculate the trace of ρs by contraction of ρs eij with elements e0kl of the dual basis, which are such that
heij , e0kl i = δij δkl . The conclusion then follows from χ(s−1 ) = χ(s)∗ and that ρ1,s is a homomorphism.

2.5
The character of the Punit representation is 1, so the number of times this representation occurs in ρ is
1
therefore (χ|1) = |G| g∈G χ(g).

2.6
a) Let ex be the basis for the permutation representation V . From each of the c orbits, take a
representative arbitrarily and denote them yi i = 1, . . . , c. For each orbit, we can construct an
invariant in V X
νi = ρg eyi .
g∈G

2
 Denote by Wi the space spanned by νi , which is clearly invariant under G and irreducible, Wi is
fixed by G and is hence the unit representation. We can therefore decompose V into the direct sum

U ⊕ W1 ⊕ · · · ⊕ Wc .

We can notice that U cannot contain the unit representation. If it did, then there would have to
be an element of U that is fixed by G, but this element would form n orbit, which contradicts the
hypothesis.
b) In the permutation representation ρs is a permutation matrix, so the character of the representation
χ(s) is the number of elements of X that are fixed by s. Clearly, there must be χ(s)2 elements fixed
by ρs in X × X.
c) • i) ⇐⇒ ii) G is doubly transitive on X × X, so is transitive on X, so the diagonal ∆ is an
orbit. Since G is doubly transitive, ∆c is also an orbit. So these are the only two orbits. These
statements also work the other way around.
• ii) ⇐⇒ iii) is just part a)
• iii) ⇐⇒ iv) is given in the hint. χ = 1 + ψ where ψ is the character of θ. Then since
(χ2 |1) = 2 (two orbits) expanding the sqare tells us that (ψ 2 |1) = 1. However, because ψ
is real valuedP(all of the matrices will be real in the permutation representation) we see that
1 ∗
(ψ 2 |1) = |G| g∈G ψ(s)ψ(s) = (ψ|ψ). We know that if a character η has (η|η) = 1 then it is
the character of an irreducible representation, so θ is irreducible.

2.7
Let χ be a character of G. We know that the regular representation has character rG (s) = |G| 1 {s = e}.
If χ vanishes on all elements apart from the identity then we have that χ(s) = χ(e)1 {s = e}. The number
of times that χ contains the unit representation is the number of distinct orbits in G, which is clearly an
integer. Thus, with 1 denoting the character of the unit representation,

1 X χ(e)
(χ|1) = χ(g) = ∈Z
|G| |G|
g∈G

so χ(e) is an integral multiple of |G| .

2.8
Lmi
a) We can write Vi = j=1 Wi where the Wi are irreducible representations. The map hi : Wi → Vi
can be used to define a linear map hji : Wi → Wi with image the j th copy of Wi by composing
with a linear projection. This will of course still be linear and still commute with ρs because of the
direct sum. By Schur’s lemma we get that each hj is a scalar multiple of the identity. For any hi
we therefore have dim V / dim Wi choices of these scalars, so this is the dimension of H.
b) a(h · w) = (ah) · w = h · (aw) by the definition of tensor product multiplication. Hence F is linear
by linearity of a and the sum given. Recall that ρ acts trivially on h so we have

F (ρs (h · w)) = (ρs h)(ρs w) = ρs h(w) = ρs F (h · w)

because h commutes with ρs . Clearly F is non-zero, otherwise H would be trivial. By Schur’s

lemma we therefore know that H ⊗ Wi ∼ = Vi and that F is a scalar multiple of the identity, hence
provides us with such an isomorphism.
c) I’m
P not totally clear on what this question means. I think ‘in the obvious way’ means h(w) =
α hα wα , in which
P case we can just apply part b) and we are done. Another possible interpretation
is that h(w) = α aα hα (wα ) for some scalars aα , but in this case some of the scalars could be 0
and I think that this would mean that h is not invertible and hence couldn’t be an isomorphism.

3
2.9
Lm
Vi = j=1 Wi and Vi,α is the image projection of the projection pαα . We know from the proof of
proposition 8 that the action of pαα on the basis of Wk {ekα } is pαα ekβ = 1 {β = α} ekα , so Vi,α is the direct
sum
Mm
Vi,α = Span {ejα }.
j=1

We also know from 2.8 that h can be written in terms of a basis hi where each of the hi = λi I is a
scalar multiple of the identity. This means that h is surjective, since each summand of the direct sum
is spanned by one element. It also means that h is injective because the sum is direct, so the elements
of each of the summand are linearly independent. By linearity each h is clearly a homomorphism. The
proof is complete.

2.10
Don’t understand this exercise and can’t finish it. . .
• V (x) is the smallest sub rep of V that contains x. This means that V (x) is the subspace generated
by the orbit of x under G
V (x) = Span {ρs x : s ∈ G}.
Lm Pm
• We can write Vi = j=1 Wi where the j th Wi has basis {ejα }. So x1α = j=1 xj,α ej1 where xj,α are
the components of the j th term in the direct summand that forms x ∈ V in the appropriate basis.

3 Subgroups, Products and Induced Representations

3.1
Let ρ be a representation of G corresponding to the irreducible representation space V . G is Abelian,
therefore for any s, t ∈ G
ρt ρs = ρts = ρst = ρs ρt .
An application of Schur’s lemma then tells us that ρs = λs I is a homothety for any element s of G. V
therefore must have dimension 1, since if it has large dimension then choosing an orthogonal basis for
this space exhibits a direct sum decomposition of V into subspaces stable under this representation of G.

3.2
a) The center C of G is an Abelian subgroup, so we know from the solution to the previous exercise
that ρ is a homothety for any s ∈ C. Write ρs = λs I, then χ(s) = nλs and

|χ(s)| 2 = χ(s)χ(s)∗ = χ(s)χ(s−1 ) = n2 λs λ−1 2

s =n .

Note that λs cannot be zero because ρ must be invertible.

b) X X
|G| = |χ(s)| 2 ≥ |χ(s)| 2 = n2 c
s∈G s∈C

using part a).

c)
Claim 1. ρ faithful =⇒ ρ an injection.
Proof. If ρs = ρg then ρ−1
s ρg = ρs−1 ρg = ρs−1 g = I, which by the hypothesis means that s
−1
g=
e.
Consider now the set of unit modulus complex scalars ωs = n1 χ(s). There are exactly |C| of these
|C|
because ρ is injective. All of these satisfy ωs = 1 because C is a group. These are therefore the
roots of unity, which form a cyclic group under multiplication.

4
3.3
Let’s take the multiplication to mean that χ1 χ2 is the function on G with values χ1 (s)χ2 (s). This
operation is clearly commutative since it returns elements of the underlying field of the representation
space. We are given that Ĝ is closed under multiplication. From the previous exercises we know that the
irreducible characters of G are of the form χ(s) = nλs where λs 6= 0, so these are invertible. This shows
that Ĝ is an Abelian group. Now, since G is Abelian we know it has exactly |G| conjugacy classes and
therefore |G| irreducible characters. Since Ĝ is closed, this means that |Ĝ| = |G| .
Write ψx : χ 7→ χ(x) for x ∈ G. We are told that this is an irreducible character of Ĝ. Define
ˆ
multiplication on Ĝ by ψx ψy χ = χ(x)χ(y). We can verify that ψx ψy χ = χ(x)χ(y) = λx λy = λxy = χ(xy),
ˆ
since the irreducible representations have degree 1 because Ĝ is Abelian. This means that Ĝ is an Abelian
group.
Since Ĝ is an Abelian group of order |G| , we know that it has |G| irreducible characters (with equality
ˆ ˆ
up to isomorphism). So the set Ĝ has order at most |G| . If the map s 7→ ψs is injective then the set Ĝ
has size |G| , and above we checked that this is a homomorphism. An injective map between two finite
sets of the same size is a bijection. Hence, checking injectivity is enough to prove isomorphism.

References
[Serre, 1977] Serre, J.-P. (1977). Linear representations of finite groups. Graduate texts in mathematics;
42. Springer-Verlag, New York.