Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
gmcninch-tufts
GitHub Repository: gmcninch-tufts/2024-Sp-Math190
Path: blob/main/course-contents/2024-01-31--notes-RT.md
908 views
---
title: Introducing the character table of a finite group date: 2024-01-31
---

\newcommand{\trivial}{\mathbf{1}}

Let GG a finite group; we consider finite dimensional representations of GG on C\mathbb{C}-vector spaces.

Previews!

Let L1,,LmL_1,\cdots,L_m be a complete set of non-isomorphic irreducible representations for the finite group GG, and let χi\chi_i be the character of LiL_i.

Recall that any GG-representation VV can be written as a direct sum of irreducible subrepresentations: G=j=1NSjG = \bigoplus_{j=1}^N S_j where each SjS_j is an irreducible representation. For each 1im1 \le i \le m, we say that the multiplicity of LiL_i in VV -- written [V:Li][V:L_i] -- is the natural number given by [V:Li]=#{jSjLi}[V:L_i] = \#\{j \mid S_j \simeq L_i\}

Thus we have Vi=1mLi[V:Li]V \simeq \bigoplus_{i=1}^m L_i^{\oplus [V:L_i]} where the notation WdW^{\oplus d} means $$W^{\oplus d} = W \oplus W \oplus \cdots \oplus W \quad \text{($dParseError: KaTeX parse error: Expected 'EOF', got '}' at position 9: copies)}̲.$

Next week, we are going to sketch a proof of the following facts

Theorem: : a. The number mm of irreducible representations of GG is equal to the number of conjugacy classes in GG.

b. $\chi_1,\cdots,\chi_m$ are an *orthonormal basis* for the space $\operatorname{Cl}(G)$ of $\mathbb{C}$-value class functions on $G$. c. For any $G$-representation $V$, let $\chi$ be the character of $V$. Then the multiplicity $[V:L_i]$ is given by $$[V:L_i] = \langle \chi,\chi_i\rangle.$$ In particular, write $k_i = [V:L_i]$; then $$\chi = k_1 \chi_1 + k_2 \chi_2 + \cdots + k_m \chi_m.$$ d. The multiplicity with which $L_i$ appears in the *regular representation* $F[G]$ is given by $$[F[G]:L_i] = \dim_F L_i.$$ In particular, if $d_i = \dim_F L_i$, then $$|G| = \sum_{i=1}^m d_i^2.$$

The Hermitian inner product on class functions, again

Enumerate the conjugacy classes of GG, say C1,,CmC_1,\cdots,C_m and choose a representative giCig_i \in C_i for each ii.

Write ci=CG(gi)c_i = |C_G(g_i)| for the number of elements in the centralizer of gig_i, and notice that Cl(gi)=G/CG(gi)=G/ci.|\operatorname{Cl}(g_i)| = |G|/|C_G(g_i)| = |G|/c_i.

Recall that for two class functions f1,f2Cl(G)f_1,f_2 \in \operatorname{Cl}(G) we have defined f1,f2=1GgGf1(g)f2(g).\langle f_1,f_2 \rangle = \dfrac{1}{|G|} \sum_{g \in G} f_1(g) \overline{f_2(g)}.

Lemma: : We have f1,f2=1cii=1mf1(g)f2(g).\langle f_1,f_2 \rangle = \dfrac{1}{c_i} \sum_{i = 1}^m f_1(g) \overline{f_2(g)}.

The character table.

The matrix is known as the character table of the group GG. Consider the m×mm \times m matrix whose rows are indexed by the irreducible characters χ1,,χm\chi_1,\cdots,\chi_m and whose columns are indexed by the conjugacy class representatives g1,,gmg_1,\dots,g_m, and whose entry in the ii-th row and jj-th column is given by χi(gj)\chi_i(g_j). g1g2gmc1c2cmχ1χ1(g1)χ1(g2)χ1(gm)χ2χ2(g1)χ2(g2)χ2(gm)χmχm(g1)χm(g2)χm(gm)\begin{array}{l|llll} & g_1 & g_2 & \cdots & g_m \\ & c_1 & c_2 & \cdots & c_m \\ \hline \chi_1 & \chi_1(g_1) & \chi_1(g_2) & \cdots & \chi_1(g_m) \\ \chi_2 & \chi_2(g_1) & \chi_2(g_2) & \cdots & \chi_2(g_m) \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ \chi_m & \chi_m(g_1) & \chi_m(g_2) & \cdots & \chi_m(g_m) \end{array}

Remark: : a. The fact that the χi\chi_i form an orthonormal basis for the space Cl(G)\operatorname{Cl}(G) is equivalent to the statement that the above matrix is unitary, in the sense that for 1i,jm1 \le i,j \le m we have

$$\sum_{k=1}^m c_k \chi_i(g_k) \overline{\chi_j(g_k)} = \delta_{i,j}.$$ b. Since the transpose of unitary matrix is also unitary, we find for $1 \le i,j \le m$ that $$\sum_{k=1}^m c_i \chi_k(g_i) \overline{\chi_k(g_j)} = \delta_{i,j}.$$

We are now going to compute the character table for a few finite groups

Cyclic groups

Let G=Z/NZG = \mathbb{Z}/N\mathbb{Z} for a natural number NN. If ζ\zeta is a primitive NN-th root of unity, we have for 0i<N0 \le i < N a homomorphism ρi:GF×\rho_i:G \to F^\times determined by the equation ρi(1+NZ)=ζi.\rho_i(1 + N\mathbb{Z}) = \zeta^i.

Since GG is abelian, its conjugacy classes are singletons; thus the number of irreducible representations is N=GN = |G|.

Each ρi\rho_i determines a 1-dimensional irreducible representation, and ij    ρi≄ρji \ne j \implies \rho_i \not \simeq \rho_j.

Orthonormality of characters implies that ρi,ρj=δi,j\langle \rho_i, \rho_j \rangle = \delta_{i,j}

This equality can actually be deduce in a more elementary way. For example, write ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 1: \̲t̲r̲i̲v̲i̲a̲l̲ ̲= \rho_0.

Note that ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 16: \langle \rho_1,\̲t̲r̲i̲v̲i̲a̲l̲ ̲\rangle = \dfra… since ζ\zeta is a root of TN1T1=TN+TN1++T+1\dfrac{T^N-1}{T-1} = T^N + T^{N-1} + \cdots + T + 1.

Consider a function fF[Z/NZ]f \in F[\mathbb{Z}/N\mathbb{Z}]. Using orthonormality of characters we deduce f=i=0N1f,ρiρi.f = \sum_{i=0}^{N-1} \langle f, \rho_i \rangle \rho_i.

The assigment if,ρii \mapsto \langle f,\rho_i \rangle is usual known as the discrete Fourier transform of ff.

The group D6=S3D_6 = S_3.

Let G=S3G = S_3 be the symmetric group on 3 letters, so that G=6|G| = 6.

Note that the subgroup H=(123)GH=\langle (123) \rangle \subset G has index 2 and is thus a normal subgroup. (In fact, HH is the alternating group A3A_3).

Notice that the centralizer of (123)(123) coincides with HH, so that (123)(123) has exactly [G:H]=2[G:H] = 2 conjugates (namely, (123)(123) and (132)(132)). On the other hand, the centralizer of (12)(12) is (12)\langle (12) \rangle so that (12)(12) has 33 conjugates; namely (12)(12), (13)(13) and (23)(23).

Thus 11, (12)(12), and (123)(123) is a full set of representatives for the conjugacy classes of GG.

In particular, we expect to find 3 irreducible representations of GG.

There are exactly two homomorphisms GF×G \to F^\times which contain HH in their kernel; one is the trivial mapping ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 1: \̲t̲r̲i̲v̲i̲a̲l̲, and the other is the sign homomorphism sgn:GF×\operatorname{sgn}:G \to F^\times; it is the unique mapping for which HkersgnH \subset \ker \operatorname{sgn} and sgn((12))=1\operatorname{sgn}((12)) = -1.

Thus ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 1: \̲t̲r̲i̲v̲i̲a̲l̲ and sgn\operatorname{sgn} are 1-dimensional irreducible representations of GG.

It remains to find one more irreducible representation.

Let Ω={1,2,3}\Omega = \{1,2,3\} and consider the standard action of G=S3G = S_3 on Ω\Omega. Write χ=χΩ\chi = \chi_\Omega for the character of this representation.

You will see for homework that χ(σ)\chi(\sigma) is equal to the number of fixed points of σ\sigma on Ω\Omega.

We have seen that the trivial representation ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 1: \̲t̲r̲i̲v̲i̲a̲l̲ appears in the representation F[Ω]F[\Omega]; thus ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 22: …ga] = W \oplus \̲t̲r̲i̲v̲i̲a̲l̲.

You will argue in homework that the character of WW is given by ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 15: \psi = \chi - \̲t̲r̲i̲v̲i̲a̲l̲.

Thus χ\chi and ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 8: \chi - \̲t̲r̲i̲v̲i̲a̲l̲ are given by:

χ(σ)ψ(σ)132(12)10(123)01\begin{array}{l|ll} & \chi(\sigma) & \psi(\sigma) \\ \hline 1 & 3 & 2\\ (12) & 1 & 0 \\ (123) & 0 & -1 \end{array}

Now, we compute ψ,ψ=1/622+1/200+1/311=2/3+1/3=1\langle \psi, \psi \rangle = 1/6 \cdot 2 \cdot 2 + 1/2 \cdot 0 \cdot 0 + 1/3 \cdot -1 \cdot -1 = 2/3 + 1/3 = 1 This shows that ψ\psi is an irreducible representation. Thus the character table of GG is given by

ParseError: KaTeX parse error: Undefined control sequence: \trivial at position 69: …& 3 \\ \hline \̲t̲r̲i̲v̲i̲a̲l̲ ̲& 1 & 1 & 1 \\ …