S. D. V. Stephens

notes / Studies in Algebra and Group Theory / chapter 9: Representation Theory / From my book

B13.4 Characters

For we decomposed representations by looking at the eigenvalues of two group elements. For a general group this is a lot of information to organize, and it turns out that much less suffices: the traces of the group elements.

Definition 13.4.1 (Character). The character of a representation of is the function given by , the trace (Definition 5.3.11) of the operator by which acts on .

Definition 13.4.2 (Class function). A class function on is a function with for all , that is, a function that is constant on each conjugacy class of (Definition 3.2.12).

Since , we have : every character is a class function. Isomorphic representations have the same character, because an isomorphism of representations satisfies .

Proposition 13.4.3 (Properties of characters). Let and be representations of the finite group , and let .

  1. .
  2. and .
  3. and .
  4. .
  5. If is the permutation representation of an action of on a finite set , then , the number of points of fixed by .
  6. and .

Proof. Since has finite order, Lemma 13.1.5 gives a basis of with , where the are roots of unity; then . Similarly, let be a basis of with .

(i) The identity element acts as , whose trace is .

(ii) In the basis of , the matrix of is block diagonal, so the traces add. The vectors form a basis of , and , so .

(iii) We have , and because ; so . On , acts by the dual map , whose matrix in the dual basis is the transpose of the matrix of . A matrix and its transpose have the same trace, so .

(iv) By Proposition 13.1.10, as representations, so this follows from (ii) and (iii).

(v) The matrix of in the basis is a permutation matrix, and its diagonal entry in position is if and otherwise.

(vi) By Proposition 9.10.19, has a basis and a basis , and multiplies both and by . Since has eigenvalues , Hence , and . ∎

Note. For a -dimensional representation , the character is itself, so it is multiplicative: . For higher-dimensional representations this fails; already differs from when . Characters turn the operations and on representations into sums and products of functions, but they do not respect the group law.

Definition 13.4.4 (Character table). The character table of a finite group lists the value of the character of each irreducible representation of on each conjugacy class. Its rows are indexed by the irreducible representations, up to isomorphism, and its columns by the conjugacy classes, each labelled by a representative and the number of elements of the class.

Example 13.4.5 (The character table of ). By Proposition 13.3.5, the irreducible representations of are , and . By Theorem 3.4.6, the conjugacy classes of are , the three transpositions and the two -cycles.

size132
111
11
20

The first column consists of the dimensions, and the rows of and are the constant function and . For , we can use the eigenvalues from Lemma 13.3.4 ( and for ; and , with , for ). Alternatively, is the permutation representation , whose character counts fixed points and takes the values on the three classes; subtracting gives .

Characters give a fast way to decompose representations. The character of is , with values . If , then , and evaluating on the three classes gives , and . The only solution is , as in Example 13.3.6.

13.4.1 Why Traces Suffice: Symmetric Polynomials

Here is a heuristic reason why the character should capture the eigenvalues. An unordered list of complex numbers , possibly with repetitions, can be recorded by the coefficients of the polynomial , whose roots they are; by the fundamental theorem of algebra, every monic polynomial of degree arises from exactly one such list. The coefficient of in is , where are the elementary symmetric polynomials. The group acts on by permuting the variables, and a symmetric polynomial is a polynomial fixed by every element of .

Theorem 13.4.6 (Fundamental theorem on symmetric polynomials). The subring of symmetric polynomials is the polynomial ring : every symmetric polynomial is a polynomial in , in exactly one way.

We will not use this theorem, and we only indicate why it holds. For , where and , the first few symmetric polynomials are In general, the nonconstant monomials of a symmetric polynomial that are not divisible by are powers of and of , and they occur in pairs, so . Since is divisible by for , this can be rewritten as , where is symmetric of lower degree than ; induction on the degree finishes the argument.

The theorem can also be viewed through representation theory. The homogeneous polynomials of degree form the space , on which acts by its permutation representation , with invariant part the trivial summand spanned by . The homogeneous polynomials of degree form of this space (Theorem 9.10.15), and the symmetric ones among them are its invariant vectors, which make up the trivial summands in a decomposition of into irreducibles. Carrying out a proof along these lines would require more about the representations of than we develop.

Another family of symmetric polynomials is given by the power sums ; for instance, and . One can show in the same spirit that also generate: . So an unordered list is determined by its power sums , .

Now let be the eigenvalues of acting on . Their power sums are , and since is a group, each is again an element of . So the single function determines the eigenvalues of every element of on . (There is also a direct argument: if has order , the multiplicity of a root of unity as an eigenvalue of is , because is if and otherwise.) We will see in the next section that the character determines much more: the representation itself, up to isomorphism.